ll 3 O B*< FIRE ASSAYING McGraw-Hill BookCompany Electrical World The Engineering and Mining Journal Engineering Record Engineering News Railway Age Gazette* American Machinist Signal Engin.... C, C'. c. g... Y. ... x. . . . FIG. 39. DIAGRAM OF ASSAY BALANCE . the central knife-edge. . the outer knife-edges. .adjustment for center of gravity of the balance system. .adjustments for equal moment of arms. . center of gravity of the balance system. . pointer-arm. . distance of deflection of center of gravity, or the gravity lever-arm. . lever-arm of small weight TO' in pan. .small weight. . mass of balance system. 44 A MANUAL OF FIRE ASSAYING From this it follows that if M increases, that is, if the mass of the balance system becomes greater, and other conditions remain constant, the weight m' must increase to cause the same deflection; i.e., the sensibility of the balance will be lessened, the sensibility being the amount of deflection caused by a given mass. Assay balances must show a sensibility of at least one- half division of the scale with a weight of 0.01 mg. or even 0.005 mg. This shows the necessity for an extremely light construc- tion, or a small mass of the balance system. It is evident from the equation that the sensibility might be preserved by increas- ing x f i.e., by lengthening the arms of the balance; in practice, however, this would also very materially increase M, so that the gain is more apparent than real. Long arm balances are also very slow of vibration. Formerly, long arm balances were common, but in modern assay balances the arm rarely exceeds 2.5 in. From the equation it also follows that, if the center of gravity of the balance system is placed at A, the knife-edge, x becomes zero and we have M Xo = m'x'; or m'x f = o; or m' approaches o, for x' is practically constant; in other words, the balance will become extremely sensitive, an infinitely small weight in the pan causing rotation. While the balance would be very sensitive, it would also be very unstable and " cranky." In designing balances it is hence very important to preserve a mean between high sensibility and stability, and as this latter is obtained mainly by lowering the center of gravity, which lessens sensibility, in part by increasing friction on the knife edges, these must be constructed with extreme care, and must be as nearly true as possible. Most assay balances are provided with fall away pan rests operated by a thumb screw on the outside of the case. When this screw is turned to the left, the rests drop away from the pan, and on further turning the balance beam is lowered and set free so that the central knife edge rests on its bearing, and the balance is free to act. The screw-ball D is provided to adjust the center of gravity, which should be somewhat below the knife-edge A. The center of gravity is adjusted so that a weight of 0.01 mg. in the pan or on the beam will cause a deflection of from one-half to one divi- BALANCES AND WEIGHTS 45 sion of the pointer. The lower the center of gravity of the balance system, the more rapid the oscillation of the balance. The higher or the nearer the point of suspension, the slower the oscil- lations and the greater the sensibility. Weighing. Before weighing, the balance is always thoroughly cleaned in every part from dust by a soft camel's-hair brush, made perfectly level by adjusting the leveling screws, and the pointer standardized to o by the little thumb-screws C, C'. To do this, the balance is set in motion until the pointer swings to from 5 to 8 divisions on the scale each side of the zero mark. If the balance-arms are equal in moment, the pointer will swing practically an equal number of divisions on each side, losing, however, a trifle on each swing, thus: +8, 7.75, +7.5, 7.25, +7, 6.75, etc., the loss being due to friction and to a gradual settling back into equilibrium. If the swings are not as outlined, the adjustment is made until they become so. The balance is then tested for sensibility as described; and the adj ust- ment made for it, if necessary, by moving the center of gravity. If the balance-arms are suspected of being unequal in length (though this is rare in good balances), weighing by "substi- tution," or double-weighing is adopted. In this method, the object to be weighed is placed first in one pan and weighed, and then in the other, the true weight being the square root of the product of the two weights found. When the sensibility of the balance is accurately known, no adjustment for equal moment of arms need be made, but weighing may be done by deflection, after the true zero or equilibrium point is found. This is found as follows: Start the balance swinging and count swings to the left as minus and to the right as plus. Suppose the swings are as follows: 8, +3, 7.5. The zero-point then is +3 = 4.75 (divisions). =2.375 (divisions). or the true zero, or true "point of rest" is 2.375 divisions to the left of the zero mark on the scale. Then place the particle to be weighed on the right-hand pan and weigh again to determine the point of rest under these con- ditions. The swings are as follows: 10, +2, 9.5. The sensi- 46 A MANUAL OF FIRE ASSAYING bility of the balance being 0.5 division deflection for each 0.01 mg. the new zero-point is ) +2 =-7.75 (divisions) -7.75 - = 3.875 (divisions), new point of rest, and the weight of the particle is the difference in deflection between the two points of rest, (3.875 2.375 = 1.5 division) divided by 0.5, or 0.03 mg. In practice, in place of two readings on one side and one on the other of the zero, it is better to make three and two readings respectively. This method, however, is not generally to be recommended; the "rider" should be used for the determination of the fractional parts of the milligram. The balance should also be adjusted for equal moment of arms, as described, before weighing. In order to detect inequality in the length of the arms, stand- ardize the balance to the true zero, place a 1-gram weight on the right pan, and an old or worn 1-gram weight on the left pan, and bring the balance into approximate equilibrium by adding minute quantities of old rider wire to the short weight. Let the gram weight in the right pan be called A. Let the counterpoise in the left pan be called B. Let R be the right lever-arm and L the left lever-arm. Determine the zero-point of the balance in the manner de- scribed. If this zero-point differs from that of the unloaded balance, bring the balance to the old zero-point by moving the rider on the left or right arm, as required. Let the weight indicated by the rider be called +m or m, as it may act with or against B to bring the balance system back to the original zero-point. Now shift the weight A to the left pan and B to the right pan; remove the rider and again determine the zero-point, and then manipulate the rider to bring the balance system to the zero-point of the unloaded balance and call the weight indicated by the rider n, as it may act with or against A. The following equations will then result: 1. AR = (Bm)L A = (Bm) L R 2. BR = (An}L BALANCES AND WEIGHTS 47 L L A+B 3. A+B = (B + Amn) --, or R (B + Amri) L mn L mn 4. =1 , or, approximately, = 1 . R A+Bmri R 2A If m = - n , or the reversal of the masses shifts the zero- point exactly as much to one side as it was before on the other L L of the actual o, the balance has equal arms; i.e., -- = 1, should R R not exceed 1 0.000003. Some assayers weigh by "no deflection." They adjust the balance to the true zero, place the bead to be weighed in the right-hand pan, and then by the addition of weights and the moving of the rider by repeated trials, balance the bead, so that finally, when the balance is lowered gently on its knife-edge, no deflection of the pointer takes place. This method, however, is not recommended, as it disregards friction and inertia, and for small weights gives inaccurate results. PRACTICAL NOTES ON THE ASSAY BALANCE. - in those labor- atories where the balance cannot be supported on stone piers trouble may be experienced from jarring of the balance. This can largely be eliminated by supporting the levelling screws on truncated pyramids cut out of rubber packing, making the lower base of the support 2 in. square and the upper one 1 in. square, with a thickness of about 1 in. A small square of ground glass may be cemented to the top of each support to take the thrust of the levelling screw. Another method of avoiding the jar is to bore four holes \ in. deep into the balance table top, and insert No. '5 rubber stoppers on top of which a small piece of heavy sheet lead is placed, about f in. thick. The level screws should be sunk into the lead about -fa in. deep, for the best effect 1 . One source of trouble with delicate assay balances is their tendency to become magnetized or charged with static electricity which will cause them to act in a very erratic manner during weighing. Balance beams constructed of material subject to magnetization should be avoided. When a balance of this kind is in use it may become necessary to change its position to avoid in part the magnetizing forces. For instance the balance beam should not be parallel to a north and south line. Balances constructed of non-magnetic material may be subject to similar i D. M. Liddell, Eng. and Min. Jour., LXXXIX, 305. 48 A MANUAL OF FIRE ASSAYING trouble due to charges of static electricity. This is particularly true in hot dry climates, and may be accentuated by insulating the balance from its surroundings by glass or rubber supports. When trouble of this kind occurs it may be desirable to ground the balance by a copper wire. 1 Balances which have pans that are blackened on one side and are bright on the other, seem to be subject to peculiar disturbance at certain times. In this case both sides of the pan should be blackened. FIQ. 40. PULP BALANCE. It is necessary to have an even temperature in the balance- room preferably about 60 F. Sunlight should be excluded if possible. The balance must not be exposed to a source of heat which will radiate unsymmetrically, otherwise unequal expansion of balance-arms will cause incorrect weights. In weighing, the balance-door should always be closed to avoid the disturbing effect of slight air currents. The true weight of a mass can be determined only by correcting for the buoyant effect of air. The error, however, is so small that it may ordinarily be neglected. 2 Pulp and reagent balances are shown in Fig. 40. The ordinary type of assay button balance is illustrated in Fig. 41. Fig. 42 1 A. Austin and Swift Hunter, "Balances," M. and Sci. Press, XCVII, 224. 2 Ostwald, "Physico-Chemical Measurements," 1894, p. 38. Ames and Bliss, "A Manual of Experiments in Physics," p. 151. BALANCES AND WEIGHTS 49 shows the "non-column" type of button balance. The very short column of these balances by decreasing the length of the pan hangers, tends to concentrate the movable mass near its central axis thus giving great stability of poise, while preserving sensitiveness; the pointer extends upward, and the scale is above the beam. In some forms the pointer is horizontal and the scale vertical, placed to the side of the beam. FIG. 41. ASSAY BUTTON BALANCE. WEIGHTS. The weights used in weighing beads are milligram weights, usually from 1 mg. up to 1000 mg., the units being as follows: 1, 2, 5, 10, 20, 50, 100, 200, 500 and 1000 mg. They are best made of platinum, as the material must be not readily corroded, so that the weight will remain constant. Riders are used to determine weights up to 1 mg. the balance-beams being divided into 100 equal spaces, each space being equivalent to 0.01 mg. with a 1-mg. rider. Riders are made of fine platinum wire, and for assay balances usually come as 0.5- and 1-mg. riders. One-milligram riders are commonly used. Where the balance can readily be made sensitive to 0.005 mg., 0.5-mg. riders can be used with profit; otherwise 1-mg. riders are preferable, as 50 A MANUAL OF FIRE ASSAYING they are not so readily injured by handling. Riders are fre- quently sold which are not of true weight, and it is essential to check them before using. The same is true of weights. It is desirable for every assay office to have a set of standardized FIG. 42. NON-COLUMN TYPE OF ASSAY BALANCE. (Keller.) FIG. 43. PLATINUM ASSAY WEIGHTS. weights for comparison. These standardized weights can be purchased from the balance firms; or a set may be corrected by the Government Bureau of Standards. 1 1 Consult Circular No. 3, U. S. Bureau of Standards, Dept. of Commerce and Labor, Washington, D. C. BALANCES AND WEIGHTS 51 The Assay-ton System. Gram and assay-ton weights are used to weigh pulp and fluxes. The assay-ton system was de- vised by Professor Charles F. Chandler, of Columbia University, New York, and reconciles the difficulties arising from the fact that all ores, etc., are weighed by the avoirdupois system, while precious metals are weighed by the troy system. The basis of FIG. 44. ASSAY TON WEIGHTS. FIG. 45. GRAM WEIGHTS. the assay ton is the number of troy ounces in 1 ton (2000 Ib.) avoirdupois. 1 ton =20001bs.; 1 Ib. (avoirdupois) =7000 troy grains; therefore, 1 ton = 14,000,000 troy grains. 1 oz. (troy) = 480 grains; therefore, ~ = 29, 166 oz. (troy). 52 A MANUAL OF FIRE ASSAYING Then, taking 1 mg. as the unit, 1 assay ton = 29,166 mg., or 29.166 grams, and 1 mg. bears the same relation to 1 assay ton as 1 oz. troy bears to 1 ton of 2000 Ib. avoirdupois. From this it follows that if 1 assay ton of ore is taken, and the silver and gold from this is weighed in milligrams, this weight will represent ounces troy per ton of ore. Fig. 43 shows a set of platinum assay weights; Figs. 44 and 45 show a set of assay-ton and gram weights, respectively. CHAPTER V REDUCTION AND OXIDATION REACTIONS REDUCTION. A reduction reaction, as particularly denned for assaying, is one in which a metal is reduced from its com- pounds by some reducing agent. The chemical definition is also applicable in that, in assaying, we frequently reduce a compound from a state of higher oxidation to a lower state of oxidation by means of a reducing agent. An oxidation reaction is "one in which a metal or a com- pound is changed to a compound of a higher state of oxidation; for example, Pb to PbO, S to SO 2 , or PbO to PbO 2 . Reduction and oxidation reactions frequently occur in assaying, and it is essential that the assayer be thoroughly familiar with the theory and facts. In speaking of reducing agents and reduction with special reference to assaying, we have chiefly in mind such re- agents as reduce metallic lead from litharge in the crucible. The chief of these are: (1) argol, (2) charcoal or coke or coal dust, (3) flour or sugar. These are added to the charge in sufficient quan- tity to produce the proper size of lead button in the crucible assay. It often happens that an ore will contain reducing agents, chiefly sulphides, so that it becomes unnecessary to add an ex- traneous agent. In fact, it may contain an excess of reducing agent, requiring an oxidizing agent to destroy the excess. The reduction of lead by argol is expressed by the following equation: 10 PbO + 2KHC 4 H 4 O 6 = lOPb + 5H 2 O + K 2 O + 8CO 2 376 2070 One gram of argol will reduce 5.50 grams of lead from 5.93 or more grams of PbO. The above formula for argol is that of pure bitartrate of potassium. Argol contains 4 as impurity a ' certain amount of carbonaceous matter, so that its reducing power will be increased. It will be found that the actual reducing power of 1 gram of argol varies between 7 and 9.5 grams of lead, depend- ent on the argol used. The reduction of lead by charcoal is expressed by the following reactions: 12 414 53 54 A MANUAL OP FIRE ASSAYING One gram of carbon will reduce 34.5 grams of Pb. As char- coal, coal or coke dust will contain more or less inert ash which has no reducing effect, the actual amount of lead reduced will be materially less. It will usually be found to range between 20 and 30 grams per gram of carbonaceous reducing agent used. Flour will reduce from 9 to 12 grams of lead per gram, de- pending on the nature of the flour. The common sulphides most frequently found in ores, and which give the ores containing them reducing powers, are : Pyrite (FeS 2 ), pyrrhotite (Fe 7 S 8 ), arsenopyrite (FeAsS), chalcopyrite (CuFeS 3 ), chalcocite (Cu 2 S), stibnite (Sb 2 S 3 ), galena (PbS), and sphalerite (ZnS). The amount of lead reduced per gram of the respective sul- phides varies according to the combination of conditions, which will be fully discussed. Taking pyrite as an example, the following equation expresses the reaction which takes place when it is fused w r ith soda and litharge: (a) 2FeS 2 + 15PbO = Fe 2 O 3 + 4SO 3 + 15Pb 240 3105 One gram of pure pyrite reduces 12.9 grams of lead. The result can readily be obtained by the following charge: Pyrite .......................... ......... 3 grams Na 2 CO 3 .................................. 10 grams PbO ..................................... 100 grams The result could not be obtained were the pyrite to be fused with litharge alone, as the presence of soda, a strongly alkaline base, induces the formation of sulphuric anhydride (SO 3 ), which combines with soda to form sodium sulphate (Na 2 SO 4 ). This sodium sulphate will float on top of the slag and is not decom- posed by the temperature usually attained in the muffle. It separates out on cooling as a fused white mass. Its melting- point is 885 C. 1 When the oxidizing action in the above charge is diminished by decreasing the litharge 2 to below 70 grams, the iron is only partially oxidized to the ferric condition and the two following equations express the reactions: 3 1 W. P. White. Am. Jour. Set., XXVIII, 471. 2 E. H. Miller, "The Reduction of Lead from Litherage," in Trans. A. I. M. E., XXXIV, 395. * It must be borne in mind that while we speak of a "reducing" or an "oxidizing" reaction, the reaction is really of both natures, for while litharge is "reduced," the iron pyrite is "oxidized." REDUCTION AND OXIDATION REACTIONS 55 FeS 2 + 7PbO = FeO + 2SO 3 + 7Pb 2FeS 2 + 15PbO=Fe 2 3 + 4SO 3 + 15Pb The first equation will give 12 grams of Pb per gram of pyrite, and the second will give 12.9 grams. The accompanying table gives the reducing powers of the various substances as deter- mined by the litharge-soda charge given for pyrite. TABLE II. REDUCING POWERS OF AGENTS Name of reducing agent Quantity of lead in grams reduced by 1 gram of reducing agent Argol 9 61 Flour . . 10.53 Sugar 11.78 26 Sulphur 18 11 (See Table III) Pyrite Pyrrhotite Stibnite 12.24 8.71 7.17 Chalcocite 4.38 Sphalerite 8 16 When no soda is present to induce the formation of alkaline sulphates, the following reaction takes place, sulphur dioxide (SO 2 ) being formed: 120 1035 or 1 gram of pyrite reduces 8.6 grams of lead. In the assay, as ordinarily performed, the foregoing conditions are modified by the presence of other substances, in the main by silica. Lead oxide readily forms silicates with silica, and the mono-, bi-, and tri-silicates are easily fusible, while those of a higher degree are fusible with difficulty. When a reducing agent (argol, sulphides, etc.) is fused with a silicate of lead, or with a charge containing litharge and silica, only a little lead is reduced when the silica is present in amounts to form a trisilicate or above, and only somewhat more when the silica is present in amounts to form a mono- or bisilicate. The reason for this is that the silicates of lead are not reduced by sulphides or carbonaceous reducing agents at temperatures below about 1000 C. 1 Above 1 Consult MetaUurgie, IV, 647. 56 A MANUAL OF FIRE ASSAYING that temperature reduction takes place more readily. The higher the silicate degree the more difficult is the reduction. If, however, certain other bases, such as ferrous oxide (FeO), soda (Na 2 O), or lime (CaO), are present (as is the case with most ores), reduction of lead from the silicate occurs, with ferrous oxide or soda, at a comparatively low temperature; but with lime alone, only at a high temperature. The following equation expresses this condition: Pb 2 SiO 4 + 2FeO + C = Fe 2 Si0 4 + C0 2 + 2Pb No difficulty is encountered in reducing lead from the borates of lead and soda, by the ordinary reducing agents, at 1100 C. While soda influences the amount of lead reduced from litharge by the sulphides present, it has not that influence on carbonace- ous reducing agents, except in so far as it may reduce the acidity of the charge and thus favor reduction. The following charge gave results as tabulated below: 1 Reducing agent 1 gram Sodium carbonate ... 10 grams Litharge 45 grams Silica 7 grams Pyrite, in this table, shows a reduction of 9.30 grams of lead per gram, a figure to be expected when its sulphur goes off partly as SO 2 and partly as SO 3 . If the soda in the preceding charge is increased, the lead button will approach the maximum re- ducible by pyrite. TABLE III. REDUCING POWER OF AGENTS Name of reducing agent Quantity of lead reduced by 1 gram of reducing agent Argol. . . 9.6 Flour.. 10 92 Sugar. .. 11 74 Charcoal 26 08 Pyrite 9 30 Sulphur 18 IP NOTE. Compare Table II with this. 1 "The Reduction of Lead from Litharge," Trans. A. I. M. E. XXXIV, 395- 2 Due to the ready distillation of sulphur, this figure is difficult to obtain; 1 gram of sulphur will usually reduce 6 or 8 grams of lead. REDUCTION AND OXIDATION REACTIONS 57 When carbonaceous reducing agents are used to obtain the required lead button, the nature of the charge, as regards acidity (due to SiO 2 or borax), has little influence on the size of button, provided sufficient bases, outside of PbO, are present to decom- pose lead silicates formed, and the silicate degree does not exceed a monosilicate. The amount of litharge present has some in- fluence. The quantity of carbonaceous reducing agent remaining constant, the size of button will increase somewhat with increas- ing amounts of PbO in the charge. When the reducing agent is a sulphide (often a natural constituent of the ore), the acidity of the charge influences, to a certain extent, the size of button obtainable. It is, however, the amount of alkaline base present (K 2 O,Na 2 O) that exerts the most powerful influence, its presence inducing the formation of SO 3 and, consequently, sulphates, thus reducing larger amounts of lead than when no alkaline bases are present, the sulphur going off as SO 2 . OXIDATION. Oxidation of impurities in ores is frequently necessary in order to obtain good results in the assay. When ores contain an excess of sulphides, arsenides, etc. (by an excess is meant a quantity above that which will give the required size of lead button), an oxidizing agent is required to oxidize this excess, enabling it to be volatilized or slagged. Oxidation of impurities is accomplished in one of two ways. 1. By the addition of potassium nitrate (KNO 3 ) to the charge (or other oxidizing agents). 2. By roasting the ore, thus using the oxygen of the air for the oxidation of impurities. When niter is added to an assay, it reacts with the most easily oxidizable compound in the charge, which is usually the reducing agent, i.e., the sulphide present. Extraneous reducing agents, such as argol, flour, or charcoal, are present simulta- neously with niter only when it is desired to determine the oxi- dizing power of niter against these reagents. For the sake of convenience, the oxidizing power of niter is expressed in terms of lead. If finely divided lead is fused with niter, the fusion reaching a temperature of 1000 C. after one-half hour, the fol- lowing reaction takes place, approximately: 7Pb+6KN0 3 = 7PbO + 3K 2 + 3N 2 + 40 2 ; or 1 gram of niter oxidizes 2.39 grams of lead. The actual num- ber of grams of lead oxidized, determined by a considerable num- ber of experiments, has been found to be 2.37. The analysis of 58 A MANUAL OF FIRE ASSAYING the gas caught from the fusion showed 10.75 per cent, oxygen, the balance being nitrogen. Oxides of nitrogen were absent. This indicates that when niter is used in the crucible fusion, oxygen is evolved which, under certain conditions, may escape from the charge without reaction. As already stated, the niter will react with the reducing agent; expressing its oxidizing power in terms of lead is merely for convenience. In certain types of charges, i.e., those containing litharge, niter, and reducing agent, or litharge, soda, niter, and reducing agent, practically theoretical results may be obtained; e.g., the oxidizing power of niter as compared to charcoal is expressed by the following equation: or 1 gram of niter oxidizes 0.15 gram of carbon. Taking the reducing power of pure carbon as 34.5 grams of lead, the oxidizing power of niter against carbon, expressed in terms of lead, is 0.15X34.5, or 5.17 grams. Ten fusions of a charge composed of 85 grams PbO, 1 gram charcoal, 3 grams KNO 3 , with 5 grams PbO as a cover, gave very concordant results, and showed the oxidizing power of niter to be 5.10. The reducing power of the charcoal was determined by five fusions with the same charge, omitting the KNO^ 1 These results, of course, can also be obtained by an impure charcoal, for, taking one which has a reducing power of 26 grams of lead 9A n (this was used in the above fusions), it then contains or 34.5 0.765 gram pure carbon. If 3 grams of niter have been added to the charge, the available carbon for reduction will be 0.765 (3X0.15) or 0.315 gram, which will reduce 34.5X0.315, or 10.75, grams of lead. The oxidizing power of niter expressed in lead, then, is 26-10.75 - - , or 5.12 grams. o Considering a sulphide and niter, and it is in this connection that niter is almost invariably used, the following reaction takes place in the litharge-soda charge already mentioned: 6KNO 3 + 2FeS 2 -Fe 2 O 3 +SO 3 +3K 2 SO 4 +3N 2 SO 3 + Na 2 CO 3 =Na 2 SO 4 +CO 2 or 1 gram of niter oxidizes 0.39 gram of pyrite. In the litharge- soda charge, 1 gram of pyrite reduces 12.22 grams of lead; there- > Thia finding confirms that of E. H. Miller, in Trans. A. I. M. E., XXXIV, 395. REDUCTION AND OXIDATION REACTIONS 59 fore, 1 gram of niter in this instance would oxidize 12.22x0.39, or 4.76, grams of lead. The accompanying table 1 shows actual results obtained for the oxidizing power of niter against different reducing agents. TABLE IV. OXIDIZING POWER OF NITER Reducing agent Oxidizing power of niter in terms of lead Pyrite ! 4.73 grams Charcoal ] 5.15 grams Flour I 5 . 09 grams Argol ! 4.76 grams It follows, therefore, that the oxidizing power of niter varies with the reducing agent used. When the assay charge contains silica and borax glass, the above figures no longer hold, for in their presence oxygen is evolved by the niter, which escapes from the charge, as in the case of the oxidation of metallic lead by niter. The amount of oxygen lost (thus reducing the oxidizing power of niter) is prob- ably a function of the rate of rise of temperature, but evidence also points to the fact that silica reacts with the niter, setting free oxygen, at a temperature very close to that at which niter reacts with charcoal, or at which oxygen will react with carbon. Niter fuses at 339 C., but does not give off oxygen when fused alone until 530 C. is reached. Charcoal ignites at temperatures 2 ranging from 340 C. to 700 C., depending upon the temperature at which it was burnt, while silica begins to react with niter at very nearly 450 C., probably according to the following reaction: 2KN0 3 + SiO 2 = K,SiO 3 + 5O + N 2 Thus, during the period in which the temperature in the crucible gradually rises to a yellow heat (that of the muffle), oxygen escapes during the range from 400 C. to 500 C., etc., this last being taken as an average temperature at which charcoal will begin actively to oxidize. 3 l lbid. 2 From a number of experiments by the author, willow charcoal was found to begin reaction with niter at very close to 440 C. * This is offered tentatively, as an explanation of what occurs. 60 A MANUAL OF FIRE ASSAYING Niter will begin to react with argol and pyrite at practically its melting-point. The oxidizing power of niter against charcoal in charges containing silica will frequently vary between 3.7 and 4.2 grams of lead, averaging about 4 grams. This is 1.1 grams lower than in the litharge*-soda charge. The oxidizing power of niter against sulphides is but little lowered by the presence of silica or borax glass. When the oxidizing power of niter against pyrite (sulphides) is considered, and expressed in terms of lead, the varying reducing power of suphides in different charges has to be taken into account. Taking as an example a charge containing considerable silica, sf that a large part of the soda (alkaline base) is absorbed as a silicate, leaving but little to form sulphate from the oxidation of the pyrite, it is found that the reducing power of pyrite is 9 grams of lead, as already noted. In this charge, niter will react with pyrite as follows: or 1 gram of niter oxidizes 0.475 gram pyrite. The oxidizing power of niter expressed in lead is then 9 X 0.475, or 4.275 grams. Actually, it will be very little lower than this, as but little oxygen escapes without action. The actual figure obtained by experi- ment is very close to 4.20. It is evident from this that the oxidizing power of niter varies with the type of charge used. It ranges, for pyrite, from about 4 grams in acid charges to 4.76 in basic charges (containing no silica). It varies still more with other sulphides. It has been the practice of assayers in making the niter fusion to run a pre- liminary assay in a comparatively basic charge (approximately the litharge-soda type), and use the figure obtained for the re- ducing power of the ore in this charge in calculating the amount of niter for the final fusion, usually made quite acid. In this way discordant results are obtained, for both the reducing power of the ore and the oxidizing power of niter vary in the different charges. Supposing that the preliminary assay showed the reducing power of a nearly pure pyrite to be 12 grams of lead per gram of ore. Using a 0.5 assay ton in the final fusion, on this basis the amount of lead reduced would be 12X15, or 180 grams. Subtracting the weight of the lead button, 20, from this leaves the equivalent of 160 grams of lead to be oxidized. Taking it REDUCTION AND OXIDATION REACTIONS 61 as the oxidizing power of niter in the final charge, 40 grams of niter would be added. But in the final charge, owing to its acidity, the reducing power of the pyrite is but 10 grams of lead per 1 gram of ore, and the total reducing power of 0.5 assay ton is 150 grams. It therefore follows that the final result will show no button. The oxidizing power for niter which should have been used is -^X4, or 5.3, and 31 grams of niter added. This, then, would give approximately the proper sized button. As the range of reducing power for pyrite is from about 9 to 12.2 grams of lead, according to whether the charge is acid and contains little soda, or is of the litharge-soda type, the most satisfactory way to determine the amount of niter to add^s to have the nature of the preliminary charge the same as that of the final charge, and then use the figure 4 to 4.2 as the oxidizing power of niter. 1 The following charges are recommended to determine oxidizing and reducing powers: PRELIMINARY ASSAY, No. 1 PRELIMINARY ASSAY, No. 2 5 grams of pyritous ore 5 grams of pyritous ore 8 grams of SiO 2 8 grams of SiO 2 100 grams of PbO 100 grams of PbO 12 grams of Na 2 CO 3 12 grams of Na 2 CO 3 Borax glass cover 3 grams of KNO 3 Borax glass cover The difference in weight of the lead buttons of preliminary assays Nos. 1 and 2, divided by 3, will give the oxidizing power of niter in the type of charge used. The weight of the button of preliminary assay No. 1, divided by 5, gives the reducing power of the ore. PRELIMINARY ASSAY, No. 3 5 grams of pyritous ore 12 grams of Na 2 CO 3 100 grams of PbO Salt cover It will be noted that the reducing power of the ore is greater than that obtained in preliminary assay No. 1. In order to determine the reducing power of argol and charcoal, make up the following charges in duplicate: PRELIMINARY ASSAY, No. 4 PRELIMINARY ASSAY, No. 5 5 grams SiO 2 5 grams SiO 2 . . 60 grams PbO 60 grams PbO 10 grams Na 2 CO 3 * 10 grams Na 2 CO 3 2 grams argol 1 gram charcoal or coke or coal Borax glass cover dust Borax glass cover 1 This has reference to pure dry KNOs. 62 A MANUAL OF FIRE ASSAYING In order .to determine the oxidizing power of niter as com- pared to charcoal, make up the following charge in duplicate: PRELIMINARY ASSAY, No. 6 5 grams SiO 2 1 gram charcoal, etc. 60 grams PbO 3 grams KNO, 10 grams Na 2 C0 3 Borax glass cover Calculate results as directed for niter in pyritous ores. Certain basic ores will have an appreciable oxidizing power, so that when the usual amount of reducing agent is added to the charge to obtain a 20-gram lead button, it is found that, due to the oxidizing power of the ore, the button is deficient in size. The oxidizing ingredients of an ore are generally hematite (Fe 2 O 3 ) ; magnetite (Fe 3 O 4 ), and manganese oxides; e.g., MnO 2 . The reac- tion which takes place is as follows: 2Fe 2 O 3 + C = 4FeO + CO 2 One gram of Fe 2 O 3 requires 0.037 gram of carbon to reduce it to FeO. In order to determine the oxidizing power of an ore, make up the following charge, if the ore consists mostly of base. When considerable silica is present in the ore, decrease the silica in the charge: 1 assay ton of ore 15 grams SiO 2 20 grams Na 2 CO 3 1 . 5 grams coal 90 grams PbO Borax glass cover CHAPTER VI THE CRUCIBLE ASSAY; ASSAY SLAGS In almost every instance, when a crucible assay is to be made, the ore and the fluxes added are thoroughly incorporated by mixing, so that, theoretically at least, every particle of the ore is in contact with a particle or particles of fluxes and reducing agent, the most favorable condition to produce a thorough reac- tion among them. The separation of the precious metals is dependent upon their affinity for metallic lead, forming an alloy of lead, gold and silver, in which lead greatly preponderates, and which readily settles by gravity from the balance of the ore and fluxes which have united to form a slag. The ore to be assayed must in all instances be in a finely crushed condition, varying in American practice, from 80-mesh up to 200-mesh material. What takes place within the crucible depends upon some or all of the following factors: 1. The fineness of crushing. Are all the particles of gold and silver or their alloy present, entirely set free from the in- closing gangue? In some ores this takes place with much coarser crushing than in others. In other ores the metals are so finely disseminated that all are not set free within the limits of crushing as carried out. 2. The mode of occurrence of the gold and silver. Is it in the free state, as is most generally the case with gold, or are the precious metals in the form of a more or less complex mineral compound (tellurides, argentite, etc.), which must be decomposed before the gold and silver will alloy with the lead? 3. The physical properties of the slag produced; e.g., its formation point, its fluidity at temperatures somewhat above its formation point, and its fluidity after superheating. 4. The chemical nature of the slag, its acidity or basicity, the nature of the bases present, more particularly copper, zinc, antimony, manganese, iron, etc. If a crucible be broken open and its contents examined shortly after fusion has commenced, these will be found to consist of a 63 64 A MANUAL OF FIRE ASSAYING heterogeneous mass through which are scattered innumerable particles of lead, both microscopic and macroscopic. The larger particles have been formed by the coalescence of the smaller par- ticles gradually settling through the charge toward the bottom of the crucible to form the final lead button as the temperature rises and the charge becomes more fluid and less resistant. It is evident that the completeness of the collection of the precious metals "depends upon the main factors already outlined. The temperature at which carbon begins to react with PbO to form Pb 1 is 530 to 555 C,, well below 884 C., the melting-point of PbO. The formation point of a borate silicate, PbO, Na 2 O, 4SiO 2 , 2B 2 O 3 (Seger Cone No. 0.022) the constituents of which are contained in nearly all assay charges, is 590 C. In the fusion of a mixture containing silica, various bases and borax glass, that silicate-borate having .the lowest formation point will form, and then as the temperature rises absorb either silica or base or both, as these are in excess of the ratio required to form the lowest formation-point compound. If the temper- ature does not rise high enough to cause this absorption, the excess of silica or base or both will remain in suspension in the formed silicate-borate, practically in an unaltered condition. If the formed silicate, etc., constitutes the greater part of the mass, there will be an imperfect non-homogeneous slag; if the excess of silica or base forms the greater part of the material, there will be a slightly fritted mass. Taking the simplest case, and also the most uncommon, that of an ore containing free gold completely liberated by crushing, the particle of lead, 2 formed at a comparatively low temperature, can unite at once, as soon as formed, with the gold particle not in- closed in gangue and commence settling to the bottom to form the lead button. It is evident that in this instance the homogeneous fusion and chemical decomposition of the ore are immaterial. Taking, however, the far more common case, in which the metals are not completely liberated by crushing, it is evident that the particle of gold still inclosed within the gangue cannot be reached by the lead already reduced, and it becomes practically essential to hold the lead in place until the ore particle containing the gold 1 Doeltz und Graumann, Metatturgie, IV, 420. According to Roscoe and Schorlemmer, Treatise on Chemistry, II, 865 (1907), CO reacts with PbO to form Pb, at 100 C. H reacts with PbO to form Pb at 310 C. Mostowitsch, Metallurgie, IV, 648. 2 There will probably be many particles of lead for each gold particle present, so that no gold will escape for lack of lead. THE CRUCIBLE ASSAY 65 is thoroughly broken up chemically and liquefied, so that the lead can absorb the gold. If the lead settles through the charge before this decomposition takes place, gold will remain in the slag. The only way to control this condition is: (a) By fine crushing, liberating the metals as completely as possible. (6) By the choice of a slag having the proper physical proper- ties, i.e., a low formation point and a viscous nature near the formation point. (c) By a comparatively slow fusion during the early stages of the assay, to prevent as much as possible the rapid settling away of the lead particles through the still existing interstices of the charge. Where compounds of the precious metals are in the ore, such as argentite (Ag a S), tellurides, calaverite and sylvanite, (AuAgTe 4 ), etc., these are readily decomposed by the litharge as follows: Ag 2 S + 2PbO = 2Pb Ag + S0 2 The tellurides will be especially considered in Chapter X, on " Special Methods of Assay." ASSAY SLAGS. An assay slag from the crucible assay con- sists in most instances of silicates and borates of metallic bases. While usually of a homogeneous nature, a slag is rarely a chem- ical compound. It is to be considered in most cases as a com- plex "solid solution," this term as applied here including both the crystalline isomorphous mixtures, or "mixed crystals," and the amorphous glasses. As an example: Litharge with silica forms certain silicates which are chemical compounds, but which have not been definitely determined, though very likely Pb 2 SiO 4 is one of them, judging by cooling curves which have been taken. 1 This silicate is capable of dissolving either PbO or SiO 2 and forming homogeneous "solid solution" within certain limits, the solid solutions in cases when the silica contents are above 11.94 per cent. rcorresponding to Pb 2 Si0 4 being glasses. In a similar way all the common bases, Na 2 O, K 2 0, FeO, CaO, MgO, A1 2 O 3 , ZnO and MnO form silicates which are sol- uble in each other when molten, and when frozen will form either complex isomorphous mixtures or amorphous glassy "solid solutions." An assay slag is therefore usually a complex 1 Wl. Mostowitsch, Metallurgie, IV, 651. S. Hilpert, Metallurgie, V, 535. 5 66 A MANUAL OF FIRE ASSAYING "solid solution." Boric acid and alkaline borates act similarly to silica, and if borax is used in the fusion the final slag will be a complex "solid solution" of silicates and borates of PbO, Na 2 O, FeO, CaO, etc., dependent upon the bases in the ore and the fluxes used. Silicates are defined in degree by the ratio of oxygen in the base to that in the acid. The chemical classification is as follows: TABLE V. SILICATE DEGREES Name Oxygen Ratio, Base to Acid ; Example Orthosilicate Metasilicate Sesquisilicate Bisilicate 1 to 1 Ito2 1 to 3 1 to4 MgO.FeO.SiO 2 MgO.CaO.2SiO 2 K 2 O.Al 2 O 3 .6SiO a Ca0.2SiO 2 The metallurgical classification is made on the same basis, i.e., oxygen in the base to that in the acid, but is somewhat different. It is the one adopted in these notes. TABLE VI. SILICATE DEGREES Formula, RO (base) Name Formula, R 2 O 3 (base) 4RO. SiO 2 2RO. SiO 2 4RO. 3SiO 2 Subsilicate Monosilicate Sesquisilicate 4R 2 O 3 . 3SiO 2 2R 2 O 3 . 3Si0 2 4R 2 O 3 . 9SiO 2 RO. SiO 2 R O 3 3SiO 2RO. 3SiO 2 Trisilicate 2R O 3 9SiO Borates may be classified in a somewhat similar manner. In general, it may be stated that the higher the silicate degree, the more infusible is the mixture, and that a polybasic mixture, one of many bases, is more easily fusible than one of few. These general statements are not without exceptions, for certain bi- silicates and trisilicates have a lower fusing point than the corre- sponding monosilicate, etc. It also depends greatly upon the base what the fusibility of the silicates will be. PbO, Na 2 O, and THE CRUCIBLE ASSAY 67 K 2 give easily fusible silicates; FeO and MnO give compara- tively readily fusible silicates; A1 2 O 3 , CaO, and MgO give diffi- cultly fusible silicates. When, however, silicates of all these vari- ous bases are mixed and go into solution as a homogeneous mass, the effect of this mixture on the melting-point of the mass is often to lower it. In fact, the silicate mixtures are to be looked upon from the same point of view as metallic alloys; there may be eutectic mixtures, i.e., mixtures of two or'more constituents which have a lower melting-point than either of the constituents, as is illustrated in the accompanying diagram. 1 ' Rhodonite Hypersthene 2 pln.0) 2 (Sl.Oj) 2 (Fe O) 2 (81 O 8 j 1100C 1000C 985C Hypersthene 80 40 60 80 IQfr - 1100C 1050 1000 C 80 60 40 20 Rhodonite FIG. 46. FREEZING-POINT CURVE; RHODONITE-HYPERSTHENE. The eutectic mixture, or the composition of lowest melting- point in the series occurs at 20 per cent, hypersthene (the bisilicate of iron) and 80 per cent, rhodonite (the bisilicate of manganese). The melting-point of this mixture is 985 C., which is considerably lower than that of either constituent alone. In the series CaSiO 3 Na 2 SiO 3 a minimum occurs in the freezing-point curve at a composition of 80 per cent. Na 2 SiO 3 and 20 per cent. CaSiO 3 , the freezing temperature being 920 C. while the freezing point of Na 2 Si0 3 is about 1010 C. and that of CaSiO 3 is 1505 C. 2 Typical Assay Slags. A slag of low formation temperature and considerable viscosity at that temperature corresponds to Seger Cone No. 0.022 -Na 2 O.PbO.4SiO 2 .2B 2 O 3 , 590 C. This may be written: PbO.4SiO 2 .Na 2 B 4 O 7 . 1 J. H. L. Vogt, D e Silikatschmdzldsungen, II, Christiana. 2 R. C. Wallace, Zeit. Anorg. Chem.. LXIII, 2 68 A MANUAL OF FIRE ASSAYING By calculation from the atomic weights the following charge will yield this slag: PbO 33.3 grams SiO 2 36.2 grams Na 2 B 4 O 7 30.4 grams The slag, corresponding to Seger Cone 0.017 and melting at 740 C., may be desirable for aluminous ores: (Na 2 O.PbO.Al 2 3 .6SiO 2 .2B 2 O 3 ), which may be written (Na 2 B 4 O 7 . PbO. A1 2 O 3 . 6SiO 2 ). The following charge will yield this slag: Na 2 B 4 O 7 22.9 grams A1 2 O 3 . 11. 5 grams PbO 24.9 grams Si6 2 40.7 grams TABLE VII. ASSAY SLAGS 1 Formula Silicate degree Approximate temperature (Centigrade) at which fluid Remarks 1. 2Na s O.SiO 2 ! Monosilicate. . . 1070 Vitreous, colorless, trans . parent. 2. NazO.SiOj Bisilicate 1090 Stony, white, crystal- line. 3. 2PbO.Si0 2 Monosilicate. . . 1030 Vitreous, light yellow, transparent. 4. PbO.Si0 2 Bisilicate 1050 Vitreous, light yellow. transparent. 5. NaaO.FeO.SiO. Monosilicate. . . 1070 Very fluid, stony black. 6. Na 2 O.FeO.2SiO Bisilicate 1070 Vitreous, black. 7. PbO.FeO.SiOj Monosilicate. . . 1100 Resinous, black. 8. Na 2 O.PbO.SiO 2 Monosilicate. . . 1020 Vitreous, yellow-green. 9. Na:O.PbO.2SiO 2 Bisilicate j 1030 Vitreous, yellow-green. 10. 2(PbO.FeO.CaO)3SiO 2 Monosilicate. . . 1110 Vitreous, black. 11. Na 2 O.PbO.FeO.CaO.2SiO- . . Monosilicate. . . 1030 Vitreous, black, con- tains sq. crystals. 12. Na 2 O.PbO.FeO.Ca0.4Si02 . . i Bisilicate 1100 Vitreous, black. 13. 2(Na 2 O.PbO.CaO)3Si0 2 . . . . Monosilicate. . . 1090 Stony, light yellow. 14. 2(Na20.FeO.CaO)3Si0 2 Monosilicate. . . 1150 Viscous, stony, gray- brown. 15. 2(N a2 O.PbO.FeO)3Si0 5 . . . . Monosilicate. . . 1030 Vitreous, black. A partial replacement of the silica by borax glass in the foregoing slags will appreciably lower the formation points. Bases such as FeO, CaO, MgO, MnO, BaO, and A1 2 O 3 are present in greater or lesser quantity in almost all ores, and SiO, is present in practically every ore, so that such slags as those 1 Elmer E. West, Laboratory, S. D. School of Mines, 1904 Stony slags indicate incomplete solution of some of the ingredients. THE CRUCIBLE ASSAY 69 outlined must necessarily be made. The easily fusible bases PbO and Na 2 O serve to lower the formation point of the slag. If it is accepted that the composition of the slag in the assay is practically the constant factor, it is evident that when the ap- proximate composition of the ore is known, we will add either basic or acid fluxes, in such proportions as to produce the proper slag decided upon. The most desirable constitution for an assay slag in general, is that of a monosilicate or a sesquisilicate, some- times, but more rarely, a bisilicate. If the ore is basic a bisilicate may be approached, if acid a monosilicate, or even a sub-silicate, in order to insure complete decomposition of the ore. The accompanying table will simplify slag calculations: TABLE VIII. THE CALCULATION OF SLAGS 1 UNIT MOLECULAR BASE RATIO; E.G., PsO: NA 2 O: FEO, ETC. = !: 1: 1 One part of base by weight Parts of other bases necessary Parts of SiO 2 necessary for monosilicate Na 2 PbO CaO A1 2 3 FeO ZnO Na 2 1.000 0.279 0.862 1.108 0.608 0.780 0.763 3.590 1.000 3.095 3.976 2.181 2.801 2.738 0.903 0.252 0.779 1.000 0.549 0.704 0.689 1.646 0.459 1.419 1.823 1.000 1.284 1.255 1.160 0.323 1.000 1.284 0.705 0.905 0.885 1.311 0.486 0.365 0.136 1.130 0.419 1.452 0.539 0.7971 0.886 1.023 0.379 1.000 0.371 | PbO FeO CaO A1.0, CuO ZnO One part by weight of Na 2 C SiOj requires to form 2.07 the monosilicate parts PbO 7.36 ! parts CaO 1.86 parts A1 2 1.14 parts FeO ZnO 2.40 2.70 parts parts CuO 2.63 parts When a bisilicate is to be calculated, the silica required for a monosilicate is determined and then multiplied by two. Vice versa, when the bases for the monosilicate have been calculated and a bisilicate is to be formed, the bases must be divided by two. The same reasoning applies to other silicate degrees. 1 Based on Balling's table. 70 A MANUAL OF FIRE ASSAYING Example of the Calculation of an Assay Slag. The problem is to calculate a charge to produce the following monosilicate: Na 2 O.PbO.FeO.CaO.2SiO 2 . Taking as the unit 10 grams of Na 2 O, it follows from the preceding table that the weights of the substances required are: Na 2 10X1 =10.0 grams PbO 10X3.59=35.9 grams FeO 10X1.16 =11.6 grams CaO 10X0.903= 9.03 grams The silica required will be: for the Na 2 10 XO. 486 = 4. 86 grams PbO 35.90X0.136 = 4.86 grams FeO 11.60X0.419=4.86 grams CaO 9. 03X0. 539=4. 86 grams Total 19 . 44 grams SiO a The silica may be determined by calculating it for one base and multiplying that figure by the number of oxygen molecules in the bases present, after having reduced the slag formula to its lowest possible terms. Before making up the charge, it is essential to remember that the Na 2 in this instance is furnished in the form of NaHCO 3 , which contains approximately 40 per cent, of Na 2 O, and that the FeO is furnished by an iron ore of the following approximate composition: Fe 2 O 3 , 80 per cent.; SiO 2 , 17 per cent. The lime is furnished by limestone, CaCO 3 , practically pure. It is also necessary to provide a lead button; so extra litharge must be furnished. To reduce the lead, coal dust is added. Some of the coal will be used up to reduce the Fe 2 O 3 to FeO. Hence the following calculations are to be made: 10 grams Na 2 O 10 are required; therefore - X 100 = 25 grams of NaHCO 3 must be added. PbO contains 92 per cent, of Pb; therefore, in order orv v/ -I f\f\ to obtain a 20-gram lead button, = 22 grams of PbO must be added, in addition to the 35.9 grams for the silicate a total of 57.9 grams of PbO. Eleven and six-tenths grams of FeO are required. Fe 2 O 3 consists of 90 per cent, of FeO and 10 per cent, of O 2 ; and as the ore is 80 per cent, of Fe 2 O 3 , ~ THE CRUCIBLE ASSAY 71 = 16.1 grams of ore will be required. The limestone contains 54 per cent. CaO; therefore, - : -- - =16.7 grams of limestone will be required. The coal in use has a reducing power of 20 grams of lead per gram of coal. The following reaction takes place between carbon and the Fe 2 3 . 2Fe 2 3 + C = 4FeO + CO 2 . 12 One gram of Fe 2 O 3 requires - = 0.037 gram of charcoal. 320 20 X 100 But as the coal used is only ~T~T~ = 58 per cent, as strong as charcoal, the following quantity will have to be added to the 16.1 grams of Fe 2 3 to reduce it: 0.037X16.1X80 = 0.82 gram coal. 0.58 To this must be added 1 gram for the reduction of the 20-gram lead button, giving 1.82 grams, of coal to be added. Since the iron ore contains silica, this is to be deducted from the silica calculated. The amount of Si0 2 in the ore is 16.1 X 17 per cent. =2.74 grams. The correct charge then is : 25 grams ...... NaHC0 3 16.7 grams ..... limestone 57. 9 grams ...... PbO 16.7 grams ..... silica (19.44-2.74) 16. 1 grams ...... Fe 2 O 3 (iron ore) 1 .82 grams ..... coal Salt cover Following is the calculation of the same slag, but for a quartz ore containing 95 per cent. SiO 2 . The formula for the slag is: Na 2 O.PbO.FeO.Ca0.2SiO 2 . Taking as the unit 1 assay ton of ore, or, in round numbers, 30 grams, this will contain 28.50 grams of SiO 2 . These 28.5 grams are to be divided into 4 equal parts to satisfy the 4 bases present. Therefore, 7.1 grams of SiO 2 will go to such an amount of each base as will form a monosilicate. 7 . 1 grams SiO 2 require 7 . 1 X 2 . 07 = 14 . 7 grams Na 2 O 7 . 1 grams SiO 2 require 7 . 1 X 7 . 36 = 52 . 25 grams PbO 7 . 1 grams SiO 2 require 7. 1X2. 40 = 17. 04 grams FeO 7 . 1 grams SiO 2 require 7.1X1. 86 = 13. 20 grams CaO 72 A MANUAL OF FIRE ASSAYING The bicarbonate of soda required is = 37 grams. The PbO required is 52.25 + 22 = 74.25 grams including the lead button. The FeC0 3 (siderite) required is = 27 grams. 62 13.20X100 The limestone required is = 24.4 grams. t)TC The complete charge is: 1 assay ton ore 27 grams FeCO 3 37 grams NaHCO 3 24.5 grams CaCO 3 '74 grams PbO 1 gram coal Salt cover In one case the ore is of a basic nature hematite and lime- stone (17 grams of each), and in the other case it is of an acid nature quartz; yet the slag produced is the same in both cases. This brings out the fact that the slag is the constant and that fluxes are added of such nature and in such quantity, deter- mined by the ore, as to produce a slag of fairly constant composi- tion. It is to be noted that the slag made in the two assays contains four bases, PbO, Na 2 O, FeO, CaO, and that these are present in unit molecular base ratio. As a matter of fact, the assayer rarely adds CaO or FeO as fluxes, but when these are present in the slag, they are derived from the ore. The bases added as fluxes are practically limited to three, PbO, Na 2 O and, at times, K 2 O, so that when an ore consisting chiefly of Si0 2 is to be assayed, the slag made will approximate a monosilicate and borate of lead oxide and soda. The table of assay slags given mentions only those in which the bases are present in the unit molecular ratio. It is evident that where an ore is considered in which numerous bases are present, these are not contained in the unit molecular ratio, so that the formula of the slag made will rather have this general form: (xPbO, yNa 2 0, zFeO, tMgO) vSiO 2 , in which, for a monosilicate, considering the letters as oxygen coefficients, x + y + z + t = 2v. In order to get a slag of low for- mation point, the coefficients of the more infusible bases, such as CaO, MgO, A1 2 O 3 , will have to be materially smaller than those of the more fusible bases, PbO, Na 2 O, and FeO. THE CRUCIBLE ASSAY 73 In assay practice, it is neither possible nor desirable to make analyses of ore before assaying for gold and silver. The assayer, however, is supposed to have a good working knowledge of lithology and mineralogy, which will enable him to form a correct judgment of the contents of his ore within fair limits. It will be comparatively easy for him to tell at once whether he has lime- stone or dolomite, or an ore containing much limonite or hema- tite or the iron sulphides; or whether magnesia, baryta or other bases are present, and in what general proportions. Following are analyses of silicious and lead-antimonial ores: TABLE IX. SILICIOUS ORES No. 1 No. 2 No. 3 No. 4 No. 5 No. 6 Gold 0.63 oz. 0.85oz. 3.35oz. 2.00oz. 0.78oz. 0.90oz. Silver 2.00oz. 6.08 oz. 1.75oz. 0.62 oz. l.OOoz. per cent. per cent. per cent. per cent. per cent. per cent. Silica 65.38 80.00 80.90 84.80 77.38 93.72 Iron 13.40 7.50 9.94 7.50 3.54 2.67 Sulphur 11.40 4.40 4.53 0.75 4.42 0.69 Arsenic 0.90 2.00 0.29 0.00 0.55 0.02 Antimony trace trace ' trace trace trace 0.089 Tellurium 0.003 trace 0.007 trace Zinc Copper 0.02 0.004 0.013 0.008 trace Manganese trace 0.54 trace 0.96 0.082 Alumina 5.43 1.79 1.70 1.02 2.80 3.53 Lime 2.10 1.70 0.50 0.90 0.56 Magnesia 0.20 trace trace TABLE X. LEAD-ANTIMONIAL ORES No. 1 No. 2 No. 3 Silica Ferrous oxide Alumina 60.1 per cent. 5.2 per cent. 9 5 per cent . 57.65 per cent. 0.70 per cent. 1 40 per cent 59.50 per cent. 4.60 per cent. 9 00 per cent Magnesia Lime , Lead Antimony Sulphur 2.68 per cent. ' trace 10.6 per cent. 4.4 per cent. . 5 per cent. 2.09 per cent, trace 16.86 per cent. 11.84 per cent. 3.00 per cent, trace 10.1 per ceat. 7.55 per cent. 44 per cent. Water . 74 A MANUAL OF FIRE ASSAYING TABLE XI. HEMATITE TABLE XII. LIMESTONE Analysis of a Hematite Analysis of a Limestone Silica Ferrous oxide 14 . 20 per cen 73 . 68 per cen 5.03 per cen Silica Alumina and ferric oxide. . . . 1.94 per cent. 0.68 per cent. 18 per cent 53 61 per cent Manganous oxide. . . . 0.19 per cen 0. 101 per cen Carbonic acid Water 43.81 percent. These analyses are given to show what the chief base constit- uents may be, and how ores will range from acid types to basic ones. Whenever sulphides are present, it is to be noted that the oxidation of these leaves basic oxides to be fluxed. At times, instead of silicate and borate slags, it is desirable to make oxide slags in the crucible assay. This, of course, can only be done when silica is absent from the ores, or when a very large excess of litharge is used in the fusion. Litharge, which melts at 884 C., possesses the property of dissolving or holding in suspension certain quantities of other metallic oxides. These slags are discussed in the chapter " Assay of Impure Ores. " The charge for the monosilicate of lead and soda is (using the unit molecular base ratio) : . 5 assay ton silica or quartz ore 39 grams NaHCO 3 55 grams Pbo Borax glass cover For the bisilicate it is : . 5 assay ton silica or quartz ore 20 grams NaHCO 3 28 grams PbO Borax glass cover Allowing for a 20-gram lead button, the charges are : No. 1, Monosilicate ore (quartz), . 5 assay ton No. 2, sesquisilicate (approxi- mate) ore (quartz), . 5 assay ton No. 3, bisilicate ore (quartz), . 5 assay ton Na 2 CO 3 26 grams Na 2 CO 3 20 grams NaCO 3 14 grams. PbO 77 grams PbO 60 grams PbO 50 grams. Coal 1 gram | Coal 1 gram i Coal 1 gram Borax glass cover | Borax glass cover Borax glass cover THE CRUCIBLE ASSAY 75 All of the above charges will yield satisfactory slags in an ore assay if the ore is of the nature described. No. 3 is the cheapest in point of cost; No. 2 is the one most frequently made. Color of Slags. Most slags from ore assays will be from light to very dark green in color or almost black, this color being due to various proportions of ferrous silicate. When iron is absent, the color of lead silicates (yellow) may predominate, or white and gray or colorless slags, due to silicates of CaO,MgO,ZnO, etc., be produced. Copper produces red slags, due to cuprous silicate. Cobalt gives blue slags. When much lime is present in an ore, this is best calculated to a bisilicate or even higher, while the other bases can be calculated to the monosilicate. CHAPTER VII CUPELLATION Cupellation has for its object the oxidation of the lead in the gold, silver, etc., alloy to PbO, which in part (98.5 per cent.) is absorbed by the cupel, and in part (1.5 per cent.) volatilized. The silver and gold of the alloy are left as a metallic bead. The process is carried out in cupels. Cupels are shallow porous dishes, made generally of bone-ash, or magnesia, produced by calcining magnesite. Portland cement may be used as a cupel material. Leached wood-ashes (particularly from beech-wood) and lime and magnesia have also been used for cupels. A mixture of bone-ash and leached wood-ashes, in the proportion of 1 to 2 and 2 to 1 respectively, has been used, and is said to give a much smaller absorption of the precious metals than bone-ash cupels. 1 Bone Ash Cupels. The bone which yields the bone-ash on calcining has the following composition. 2 Sheep bones Cattle bones Ca 3 (POJ 2 ! 62.70 per cent. CaCO 3 ' 7 . 00 per cent. Mg 3 (PO 4 ) 2 | 1 .59 per cent. CaF 2 2 . 17 per cent. 58.30 per cent. 7.00 per cent. 2.09 per cent. 1.96 per cent. Organic matter ! 26.54 per cent. ' 30.58 per cent These bones will produce bone-ash of the following composition : No. 1 No. 2 84 . 39 per cent. 83 . 07 per cent. CaCO, ........................ ; 9 .42 per cent. 10 . 00 per cent. CaF 2 ......................... , 4.05 per cent. 3.88 per cent. | 2 . 15 per cent. 2 . 98 per cent. 1 Kerl, Probir Kunst, 1886, p. 91. * Hemtz, Erdmari&Jour. fur P. Chem., XLVIII, 24. 76 CUPELLATION 77 The bone-ash used for cupels must be specially treated by washing with an aqueous solution of ammonium chloride (this salt to the extent ol 2 per cent, of the weight of the bone-ash to be treated). 1 This reacts with CaCO 3 and any CaO present, con- verting them into CaCl 2 , which is removed by washing with water. The presence of CaC0 3 is very undesirable in bone-ash for cupels, as it begins to give off CO 2 at 800 C., about the temperature of the beginning of cupellation, causing a serious spitting of the lead button, which entails a loss of the precious metals. Cupels should not be kept where the nitrous fumes from parting can be absorbed by them, as these will form Ca(N0 3 ) 2 with any CaO that may be present, which also is decomposed about the temperature of cupellation. Bone-ash melts at about 1450 C. (Hempel). The physical nature of the cupel, especially as regards porosity, is very important. For this reason there should be a careful ad- justment of the relative amounts of different sized particles present. Practically, only the fraction of 1 per cent, of the bone-ash should remain on a 30-mesh screen. If there is an insufficiency of fine particles in the bone-ash, the cupel will be too porous and cause a relatively heavy absorption of gold and silver. If the bone-ash is too fine, the cupels made from it will be too dense, prolonging the cupellation and causing losses, mainly by increased volatilization. The following is a screen analysis of the bone-ash commonly purchased, but which is rather coarse : Through a 20-mesh screen, 100 per cent. On a 30-mesh screen, 2 . 90 per cent. On a 40-mesh screen, 6 . 40 per cent. On a 60-mesh screen, 10 . 04 per cent. On a 80-mesh screen, 2 . 00 per cent. On a 100-mesh screen, 11 .20 per cent. Through a 100-mesh screen, 68 . 88 per cent. Cupels should be as uniform as possible as regards density, and for this reason are best made by machine, in which a constant pressure may be obtained, rather than by hand molds. Fig. 47 shows a good type of cupel machine. Considerable pressure may be used, and the cupels made quite firm. It is not possible to specify the proper condition in definite terms, but a batch of cupels, after being made up and carefully dried for at least three weeks or a month, should be tested by cupeling a weighed quan- 1 W. Bettel. Proc. Chem. and Met. Soc. of S. A., II, 599. 78 A MANUAL OF FIRE ASSAYING tity (200 mgs.) of c. p. silver with 20 grams of lead at the proper temperature, 850 C., and the loss noted. It should not exceed from 1.5 to 1.8 per cent. The bone-ash to be made into cupels is mixed with from 8 to 12 per cent, of water, in which is dissolved a little K 2 C0 3 , or to which has been added a little molasses or stale beer. After making, the cupels should be carefully and slowly dried. If Fia. 47. CUPEL MACHINE. possible, cupels should be several months old before using. In the Royal British Mint no cupels less than two years old are used for bullion assays. If cupels are too rapidly dried, or have been made up too wet, they crack and check when placed in the furnace and make the assays conducted in them unreliable. The importance of good cupels cannot be overestimated. Very frequently, inaccuracies in the assays are due chiefly to the cupel. The shape of the cupel has some influence on the loss of precious metals by absorption. If the cupel is very flat and shallow, so that the molten lead covers a large area and has little depth, the time of cupellation is decreased as the surface CUPELLATION 79 exposed to oxidation is increased, but as the absorption of precious metals is probably a function of the area exposed, it will be large in shallow cupels. 1 Magnesia Cupels. Of recent years the so-called "patent" cupels have come into wide use especially in England and South Africa and to a lesser extent in the United States. These cupels are made almost invariably of a magnesia base. This magnesia is produced by calcining crude Austrian, Californian or Turkish magnesite, and is used largely in the steel industry for basic refractory brick. The composition is about 90 per cent MgO, and 10 per cent, of impurities, chiefly CaO, Fe 2 O 3 , A1 2 O 3 and SiO 2 . The cupels are in- variably very hard and firm, of a brown color and are formed under high pressure. The exact composition of the cupels is generally a trade secret. Magnesia cupels cannot very readily be made in the laboratory like bone-ash cupels, and in almost all instances their cost is higher. A number of brands are on the market; as the Morganite cupel, madfe by the Morgan Crucible Co., Battersea Works, London, those made by Deleuil, Paris, and the Mabor, Scalite, Velterite, Star, etc., brands. Morganite cupels, 31.5 m.m. top diameter (about 1.25 in.) the common size, cost $3.35 per 100 in St. Xouis. The properties of various types of cupels are discussed in a following section. Portland Cement Cupels. Satisfactory cupels may be made of ordinary Portland cement provided the amount of mixing water is carefully adjusted. 2 The amount of water should be 8 per cent, of the weight of the cement. If less than 5 per cent, water is used the cupels are too fragile, if 20 per cent, is used they will not readily pass the cupel machine. Upon heating, cupels with less than 5 per cent, and with more than 15 per cent, water cracked about the edges. Cupels made of one-half cement and one-half bone-ash give good results. Cement cupels are very cheap as compared to bone-ash. Cement will cost from 35 cents to $1.00 per 100 lb., while bone- ash costs from $5.00 to $8.00 per 100 lb. Cement cupels should be thoroughly dried before use, otherwise they will develop cracks during heating. 1 H. K. Edmands, Eng. and Min. Jour., LXXX, 245. 2 T. P. Holt and N. C. Christensen, "Experiments with Portland Cement Cupels," Eng. and Min. Jour., XC, 560. J. W. Merritt, "Cement vs. Bone-Ash Cupels," Min. and Sci. Press. C. 649. 80 A MANUAL OP FIRE ASSAYING CUPELLATION. 1 When ready to cupel lead buttons, the cupels are' placed, empty, in the red-hot muffle and allowed to remain there for about 10 minutes in order to expel any moisture, or organic matter present (if molasses water has been used in making them up). If the buttons were placed into the cold cupel, the lead would melt before all the remaining moisture is expelled, which would then pass up violently through the molten lead, causing what is termed "spitting," i.e., the projection of small lead particles, carrying gold and silver from the cupel. Some cupels, made from bone-ash containing CaC0 3 , will com- mence to spit after the cupellation has proceeded for some time and the temperature has risen to above 800 C. This can be stopped by pulling the cupel to the cooler (front) part of the muffle, although the cupellation, after spitting, is to be consid- ered unreliable. When a piece of wood or coal is placed in the muffle to "open up" lead buttons, the cupels absorb gases at times, wjiich later on, when the temperature rises, are again ex- pelled, with a spitting of the lead. When the lead button is put into the hot cupel, the lead melts (326 C.) and is covered by a gray-black scum. If the lead button is practically pure, as it should be, this black scum dis- appears when the lead reaches a temperature of 850 C. This is called the "opening up" or "uncovering" of the lead button. The molten lead then appears bright, begins to "drive," and active and rapid oxidation commences. Lead buttons should uncover as soon as possible in the muffle. If other and more difficultly fusible metals, such as Cu, Fe, etc., are present, the temperature of uncovering is higher and the temperature re- quired for cupellation is higher. These foreign metals should, however, as a general rule, be absent. Little flakes of PbO form on the surface of the molten lead and slide down the convex surface of the button, and are absorbed by the porous mass of the cupel. The process of cupellation is dependent upon the relation of the surface of the cupel to that of the molten lead alloy and the litharge which is formed by oxidation. There is a great difference between the surface tension of molten lead and litharge and while litharge can " wet " the bone ash surface and hence be absorbed, molten lead cannot do so, or only to a very slight extent and hence is not absorbed. In the same manner metallic silver and gold, left on the cupel by 'The description which follows refers in the main to bone-ash cupels. CUPELLATION 81 the oxidation of the lead will not be absorbed by the cupel. As will be noted further on there is always a loss of precious metal during cupellation, the greater part of which is caused by absorption by the cupel. Whether this absorption is due to some small part of the lead alloy passing into the cupel, or to an oxidation of some silver with consequent absorption has never been definitely determined. It is true that the different cupel materials and the physical condition of the cupel as regards porosity influence absorption, the greatest factor, however, is temperature of cupellation, a comparatively slight increase of temperature causing a marked increase in absorption. Whether this increased absorption is due to an increased oxidation of the precious metal, or a decrease in the surface tension of the lead alloy is open to question. This subject is again referred to on page 133. The temperature of cupellation is the most important single factor in the operation. Three distinct temperatures must be considered, (1) the temperature of the cupelling lead; (2) the temperature of the muffle, by which is meant the temperature of the interior of a blank cupel, directly adjoining the one con- taining the lead, and (3) the temperature of the air in the muffle, near the cupel. The vital temperature is that of the cupel- ling lead, but as this is difficult to measure except by special appa- ratus, the "muffle temperature," which always bears a distinct relation to the temperature of the cupelling lead is used hereafter in designating the "temperature of cupellation." The temperature of cupellation for pure lead buttons should be 850 C. to "uncover" the button, this may be lowered to about 770 C. during the major part of the cupellation, but must be raised again to about 830 C. near the end to finish the opera- tion. This applies to bone-ash cupels. The temperature of the lead itself during cupellation is higher than that indicated by the blank cupel near it, owing to the rapid oxidation of the lead. This is shown by the brighter color of the lead. Any foreign metals, as Cu, Sb, Fe, Zn, etc., which are present are oxidized (some by the PbO formed), and absorbed by the cupel, if not present in too large amounts. Zn + PbO=Pb + ZnO. Such elements as Sb, As, and Zn, when present in the button, are in part volatilized as oxides, and in part absorbed. When 6 82 A MANUAL OF FIRE ASSAYING cupellation for silver is carried on, the temperature should not be above 820 C., in which case crystals of litharge (feathers) form on the side of the cupel toward the muffle mouth. If the temperature is too low for the cupel to successfully absorb practically all of the PbO, these feathers form low down in the. cupel. When the temperature is about right, they form near the upper rim of the cupel. It is, however, to be noted that the draft through the muffle influences the formation of feather litharge; i.e., if the draft is strong, feathers will form, although the temperature is somewhat above 820 C. During cupellation, the door of the muffle should never be left wide open, but should be set slightly ajar, so that the cold air will not strike directly upon the cupels. When silver and gold are cupelled for, owing to the higher melting-point of the silver-gold alloy, the finishing temperature will have to be 860 C. at least. As the cupellation proceeds, the percentage of lead in the alloy decreases and that of Ag and Au increases. The litharge thrown off from the center of the button is in larger specks, and brilliant, and the button assumes a more rounded form. When this phenomenon appears, the cupel should be pushed back into the hotter part of the furnace or the temperature of the furnace raised somewhat. When the last of the Pb goes off, large buttons are covered with a brilliant film of colors (interference colors) and the button appears to revolve axially. The colors then disappear, the bead becomes dull, and then again takes on a silvery tinge. If now the temperature of the muffle is below that of the melting-point of silver (962 C.), or below that of the gold-silver alloy constituting the bead, or if the cupel be withdrawn from the furnace, the "blick" or "brightening" or "flash" of the bead takes place; i.e., the bead suddenly becomes very bright, at the moment of solidification, owing to the release of the latent heat of fusion, which raises the temperature of the bead very much for a short time. The bead has been in a state of surfusion, i.e., in a state of fusion below its true freezing-point, toward the last of the cupelling operation; and if it be lightly jarred or the temperature allowed to drop still lower (by taking it out of the muffle), it suddenly congeals and assumes a state normal (solid) to the temperature existing. The release of the latent heat, raising the temperature of the bead, causes the brightening. The "brightening" of very small beads is rarely noticeable. CUPELLATION 83 Silver and gold beads still containing small amounts of Pb or Cu do not brighten so noticeably. If even minute quantities of rhodium, iridium, ruthenium, osmium, or osmium-iridium, are present, buttons will not flash. Platinum and palladium are excepted. Silver beads after cupellation, and at the moment of solidi- fication, also "sprout." According to Gay-Lussac molten silver dissolves 22 times its volume of oxygen, at the freezing-point. Later researches 1 prove this practically correct. At 1020 C. molten silver will hold 19.5 volumes of oxygen (at 760 mm. and O C) and at the melting-point somewhat more. For any given temperature the oxygen dissolved is proportional to the square root of the oxygen pressure. In air at 760 mm. pressure the oxygen has a partial pressure of 150 mm. and the volume of oxygen dissolved by molten silver under assay conditions is 9.65 volumes at the freezing-point of silver. The oxygen is dissolved either as monatomic oxygen or as silver oxide (Ag 2 O), in dilute solution. It is probable that this silver oxide, not being soluble in solid silver is dissociated with explosive violence, with the liberation of oxygen, when the silver solidifies. This oxygen, suddenly expelled when the bead solidifies, causes a cauliflower-like growth on the bead. Small particles of silver may even be projected from it and cause a serious loss. When gold is present in the silver bead to the extent of 33 per cent, or more, sprouting does not take place. Silver beads containing small quantities of Pb, Cu, Zn, Bi, etc., will not sprout, so that if a button does sprout it is a sign of purity. Buttons below 5 mgs. in weight do not sprout readily; large buttons, however, do. Sprouting can be prevented by slow cooling in the muffle, or by having ready a hot cupel which can be set, inverted, over the one holding the bead, and withdrawing both from the muffle, thus cooling the bead slowly. Sprouted beads are to be rejected as an assay. When cupelling for silver alone, or for silver and gold, it is necessary to watch the end of the cupellation carefully, and to promptly remove the cupel about 30 seconds to 1 minute after the bead has become dull. A heavy loss of silver commences if the silver buttons are kept beyond that time in the furnace. If silver is not to be determined, but gold only, the buttons may 1 Donnan and Shaw, Jour. Soc. Chem. Ind., XXIX, 987. Sieverts und Hagenacker, Zeit Phys. Chem., LXVIII, 115. 84 A MANUAL OF FIRE ASSAYING be left in for 5 to 10 minutes without loss of gold. Gold beads will retain minute amounts of lead which cannot be removed by permitting the bead to stay in the muffle. It is to be noted, however, that silver lead alloys containing between 80 and 90 per cent, of silver also show the phenomenon of sprouting or developing a cauliflower-like growth on solidification. * The bead, when cold, is taken from the cupel with a pair of pliers, and cleaned of bone-ash by flattening somewhat with a hammer. It should be examined with a glass to make sure that no bone-ash adheres to it. The bead should be either white or yellow, depending on the amount of gold present, round and not flat (the latter indicating the presence of foreign metals), and should possess a crystalline surface where it adhered to the bone-ash. It should be firmly attached to the bone-ash of the cupel. If it is not, this fact indicates that lead is still present. It should also have no rootlets extending into the cupel. The cupel, after cupellation, should be smooth and firm, not fissured and cracked, and of a light yellow color when cold. Other colors indicate the presence of foreign metals. The freezing-point curve of lead-silver (Fig. 48) will give some idea of the proper temperature of cupellation. A lead button is to be considered as an alloy of lead and silver (or gold) which in the process of cupellation undergoes the change from practically pure lead to that of pure silver (or gold). A 20-gram button containing 200 mgs. of silver contains 1 per cent, of Ag. An alloy of lead and silver containing 4 per cent, of Ag is of "eutectic composition" and melts at 303 C., the melting-point of pure lead being 327 C. Most assay buttons will contain very much less than 1 per cent, of silver and will melt practically at the melting-point of lead. Leaving out of consideration for the moment that lead "uncovers" at 850 C. in an oxidizing atmosphere, and the proper temperature required to cause a ready absorption of PbO by the cupel, it is evident that for a lead button weighing 20 grams and containing 20 mgs. of silver (0.1 per cent.), the temperature required to keep the button molten ranges from 327 C. to 303 C., until the button has decreased |f in weight by the loss of Pb, practically the entire time of cupellation. When the button has reached V of i ts original weight, the K. Friedrich, Metallurgie, III, 398. CUPELLATION 85 temperature required to keep it molten will rapidly increase, according to the curve, as more lead is oxidized, until, in order to prevent freezing and get pure silver, a temperature of 910 C. 1000 AgO 10 20 30 40 50 60 70 80 PblOO 9080 70005040 3030 Fro. 48. FREEZING-POINT CURVE, LEAD-SILVER 100 0' 1000 900 700 1062 CuO PblOO 30 40 50 60 70 80 70 60 50 40 30 30 49. FREEZING-POINT CURVE, LEAD-COPPER 100 and slightly above must finally be reached. 1 In order, however, to cause a rapid formation of PbO and its ready absorption by the cupel, and not have heavy losses of Au and Ag, it is found 1 While the melting point of silver is 962 C., this temperature is not necessary as surfusion takes place. 86 A MANUAL OP FIRE ASSAYING that a temperature of about 850 C. is best for the main part of the cupellation. It is evident, however, that in order to finish the cupellation, the heat must be raised toward the end, other- wise the alloy of lead and silver, as it increases in silver percent- age, will tend to freeze, i.e., to solidify. It is also to be noted, however, that this tendency, with most lead buttons of ordinary silver contents, is not reached until very near the end of the cupellation. It is an old saying amongst assayers that "a cool drive and a hot blick" are essential to a good cupellation. In the cupellation for silver it would seem at first sight that a final temperature of 962 C. is necessary in order to prevent freezing and to obtain a silver bead free from lead. However, the phe- nomenon of the "surfusion" of the silver, i.e., silver in a molten state below its true melting-point, due probably to its formation from its lead alloy by the oxidation of the lead, appears to indicate that this temperature is not necessary. It is true, nevertheless, that the finishing temperature, depending somewhat upon the amount of silver present, may not fall much below 910 C. It is plain that buttons may be cupelled at temperatures much above those stated, but the loss of silver and gold, both by ab- sorption and volatilization, is very much increased with the higher temperatures. The reasoning outlined for silver applies also to gold, except that, owing to the somewhat higher melting-point of gold (1063 C.), the finishing temperature should be a little higher. It is of interest at this point to more fully discuss the question of temperature of cupellation. This term has been used in a vague manner by writers on the subject and has been used to signify generally the temperature of the air of the muffle, either at the side or just above the cupel, or that of the interior of the cupel. Due to the heat of combustion of the lead neither of these temperatures is the true temperature of cupellation. The actual temperature of cupellation has only recently been deter- mined, 1 due probably to the fact that this determination involves experimental difficulties, since the protective tube of the thermo couple in almost any form is rapidly destroyed by the corrosive action of the litharge. As already stated three temperatures may be considered during cupellation. (1) The temperature of the cupelling lead; (2) the temperature of 1 C. H. Fulton and O. A Anderson and I. E. Goodner, West. Chem. and Met., IV, 31. which consult for methods of temperature determinations of cupellation. CUPELLATION 87 the muffle as determined by that of a blank cupel, adjoining the one containing the lead; and (3) the temperature of the air immediately surrounding the cupel. This is invariably lower than the first two temperatures, which accounts for the low FIG. 50. TEMPERATURE-CURVE SHOWING DIFFERENCE BETWEEN AIR IN MUFFLE AND CUPELLING LEAD. temperature figures that have been assigned to the cupellation process. It is to be noted that there is an air draft through the muffle during cupellation, cold air constantly entering at the mouth of the muffle, so that the air in the muffle does not attain the temperature of the muffle walls. The actual air temperature 88 A MANUAL OF FIRE ASSAYING is also probably somewhat lower than the thermo couple junction shows, since this absorbs heat* radiated from the muffle walls more rapidly than the air. Fig. 50 shows the two temperature curves, one, the actual temperature of the cupelling lead and the other that of the air in the muffle, close by and at a level with the top of the cupel. The general form of the curve is due to fluctuations of temperature in the muffle, caused by firing and attempts to regulate the temperature, by draft and otherwise. It will be noted that the temperature of the cupel rises rapidly after oxidation has commenced, attaining a maximum of 940 C. and then falling as the muffle cooled. The interesting data is the difference between the temperatures of the air in the muffle and the cupel, which is greatest during the period of active oxidation. The maximum difference is 145 C. The lead "froze" or was covered over with a coating of PbO, preventing further cupella- tion at 802 C, the air in the muffle being then at 675 C. The actual minimum temperature of cupellation in this case was therefore 802 C., 127 higher than the air temperature. 1 Experiment. To determine the temperature of the "opening or uncovering" of the button; i.e., the beginning of cupellation, and the "freezing" of the button; i.e., where cupellation is stopped by the formation of PbO which is not absorbed. In this experiment, 134 grams of lead were used. The pres- ence of gold or silver has no influence on these critical tempera- tures, as the melting-point of the alloys is usually far below the "uncovering" temperatures and the precious metals form no oxides which would complicate matters. The influence of such metals as copper will be referred to further on. The set was run with a blank, at the same temperature as the cupel before the lead was added. Fig. 51 gives the curves plotted as before. The results show that the button begins to uncover at 800 C. and 804 C. and begins to "freeze" at 804 and 788 C. These are the actual cupel temperatures. A repetition of the experiment in the same cupel shows "uncovering" at 832 C., 829 C., and 834 C. and a freezing at 850 C. Other results show the begin- ning of "uncovering" at 797 C. and completely open at 805 C. Another shows an opening to occur at 811 C. Another shows 1 In order to definitely prove the difference in temperature to be due to the oxidation of the lead, a set was run in which the lead in the cupel was covered by a clay dish luted on, practically preventing oxidation. In this instance the muffle and cupel were at nearly the same temperature for the space of an hour, first one being a little higher and then the other CUPELLATION 89 S S H 5 H a 8 = 3 2 si S a !l 3 g SSSSSSSSS8 90 A MANUAL OF FIRE ASSAYING an "uncovering" at 826 C.; another experiment, at 809 C. In general, the opening temperatures and freezing temperatures are near each other, as is to be expected in so far as the two are, in the absence of any silica, practically the result of the same process. The freezing temperature, however, may be somewhat higher or lower for a number of reasons developed below. These critical temperatures are of importance in so far as they mark the minimum possible temperature of the beginning of cupellation. From the curves it will be noted that as soon as the button uncovers, there is a sharp rise in the temperature of the lead whether the muffle temperature rises or not, due to the oxidation of the lead. In the curves where the muffle blank and the cupel were at the same temperature before the dropping in of the lead, the oxidation raises the cupel temperature from 20 to 150 C. above that of the muffle dependent upon the rate of oxidation, i.e., the air supply. Since it has been well established that the chief cause determining the loss of precious metal by absorption and volatilization is the temperature, it is at once apparent that for careful work the air supply of the muffle is just as important as a regulation of the temperature of the muffle itself. In Fig. 51 the cupel temperature does not rise greatly above the muffle temperature. This is due to the fact that, just as soon as the lead had opened, the furnace was again cooled in order to get a determination of the temperature of the "freezing," thus pre- venting the attainment of maximum oxidation. What determines the "uncovering" and "freezing" of the buttons? It would appear at first sight that the critical tem- perature of "uncovering" and "freezing" is the melting-point of litharge, in so far as the melting of the cover of oxide and its absorption by the cupel would naturally mark the "opening." Recent and accurate determinations of the melting-point of pure litharge, give 906 C. 1 and 884 C., 2 with the latter probably the figure to be preferred. In these researches it is noted that be- fore the melting-point is reached there is a decided soft and pasty stage, which is ascribed to the marked volatilization of PbO from the solid state. This volatilization begins just below 800 C. 3 and is a function of the area exposed and the tempera- ture. As in the case under consideration, the film of PbO on 1 O. Doeltz and Mostowitsch, M etallurgie, IV, 290. 2 Mostowitech, M etallurgie, IV, 468. 8 O. Doeltz and C. A. Graumann, Metallurgie, III, 408, 'CUPELLATION 91 the button gives probably the greatest area in relation to volume possible, this volatilization is an important factor in the uncover- ing of the button, in the case of the absence of silica or borax, and the practical ceasing of this volatilization marks the "freez- ing" of the button. It will be noted that all the temperature determinations of the "opening" and "freezing" are well be- low the melting-point of litharge. It is here self-evident that when a button is put into a cupel whose temperature is 900 C. and above, that the temperature at which "opening" is observed has no special significance. Lead buttons from crucible assays practically always have adhering to them small amounts of silicious slag, and the bone- ash at times contains minute quantities of SiO 2 . When the lead button melts in the cupel, this slag and fine, loose bone-ash go to the surface into the litharge film. According to recent research on the lead-silicates, 1 the silicates 6PbO.Si0 2 , 5PbO.SiO 2 and 4PbO.SiO 2 are thinly fluid at 794 C., 796 C. and 726 C. respec- tively. The percentage composition of these silicates is as follows: Silicate Litharge Silica 6 PbO.SiO 2 95.68% 4.32% 5 PbO.Si0 2 94.86% 5.14% 4 PbO.SiO 2 93.66% 6.34% It is evident that when the very small amount of litharge which forms the film is considered, that minute quantities of sil- ica only are necessary to materially lower the "opening" tem- perature of the button. From these facts it follows that the "opening" or "uncovering" temperature is not a fixed temper- ature, but will depend upon the following factors: 1. The presence of silica in a condition to combine with lead (very probably also of borax). Where this silica comes from has already been mentioned. 2. The vaporization of solid litharge. As the rate of vapor- ization depends upon the temperature, and the relation of area exposed to volume present, a button with a thick covering will not open at as low a temperature as one with a thin cover- ing. This can be demonstrated by placing a button in the cupel at a temperature below 700 C. and permitting it to form a heavy film of PbO, then raising the temperature to the usual "uncov- 1 Wl. Mostowitsch, Melallurgie, IV, 64.7. 92 A MANUAL OF FIRE ASSAYING ering" point, and placing another button into a second heated cupel. The last button will uncover first, as its thinner cover of litharge will vaporize in less time. 3. The presence of foreign metals in the lead, such as cop- per, iron, etc., will raise the "uncovering" temperature. This is a frequently observed fact, and the reasons for it are practically obvious. If the temperature of the cupel at the moment of un- covering could remain fixed, the increase in the oxidation of the lead would very soon balance the vaporization at that tem- perature and the button would again freeze, but it has already been noted that a very sharp rise in temperature at once occurs automatically; i.e., independent of the muffle, due to the rapid oxidation of the lead ; this effects a marked increase in vaporiza- tion, keeping the button open and soon in most instances the temperature of the button itself passes to and beyond the melt- ing point of litharge (884) and cupellation proceeds rapidly. Cupellation, however, can seemingly be carried on below the melting-point of litharge, as Figs. 51 and 52 will show. The particles of litharge formed on the surface of the button, though solid, are pasty and capable of being asborbed by the cupel, or the surface of the cupelling lead being the area of the most active oxidation is at or above the temperature of melting litharge, which the thermo-j unction at the bottom of the lead does not indicate. 1 In one experiment, containing considerable silver, it was noted that very near the end of the cupellation when the amount of silver was large and that of lead small, the button was cupel- ling at an indicated cupel temperature of 750 C., the button then solidified and proved to be a lead-silver alloy. The temperature of 750 evidently did not represent the surface temperature of the button as was indicated by the brightness of the PbO specks formed; i.e., the amount of heat liberated by the small amount of lead oxidized was insufficient to make any material impression on the thermo-j unction. The "Freezing" of the Button. When the temperature of the muffle falls so that the heat of oxidation of lead is no longer 1 In an experiment to shed light on this point, 70 grams of pure PbO were placed in a cupel and heated to 815 C. for the time of 20 minutes. The litharge showed vaporization , but none was absorbed by the cupel. In a duplicate experiment the temperature was raised to 883 C. just below the melting-point, and while the litharge did not melt, all of it was rapidly absorbed by the cupel. In the first case the mass of litharge was sintered. Absorption thus probably occurs in the "pasty" stage mentioned. CUPELLATION 93 enough to keep its temperature such that the rate of vaporization is in excess of the rate of oxidation, the molten button will be- come covered by a film of litharge and cupellation ceases. "Feathers" are crystals of solid litharge sublimed from the vapor and deposited on the cupel walls. That the cupel walls are invariably cooler than the cupelling lead is self-evident. Feathers will therefore form when the temperature of the cupel wall is below or near that of " uncovering. " They will not form above about 820 C. The cooler wall and that on which feathers most usually form is that toward the muffle mouth, due to the direct impingement of cooler air currents. These feathers form the best guide to the temperature of cupellation ordinarily available. During their formation the actual temperature of the cupelling lead is usually from 840 to 900 C., although it may be appreciably higher if the oxidation be rapid. The rapidity of the oxidation depends largely on the air supply, and this heavy air current striking the cupel may cool the walls sufficiently to cause a heavy sublimation of feathers, although the true temper- ature of cupellation, i.e., that of the button may be unduly high. Resume. From the foregoing, it appears that the "uncov- ering" of the button occurs at from 800 to 840 C. dependent on several factors, and that the actual minimum temperature of cupellation, may be placed at about 850 C., but usually rises above this, i.e., independent of the muffle, frequently to 930 and 940, unless the muffle temperature is lowered after uncover- ing. The necessary finishing temperature is, however, higher than 850 C. Experiment. To determine the phenomena incident to the "finishing" of a cupellation containing silver, i.e., that of "surfusion," "sprouting," freedom of the silver bead from lead, temperature necessary to finish, etc. It has frequently been noted that a cupellation containing silver and gold, or both, could seemingly be "finished," i.e., all the lead eliminated therefrom when the temperature of the muffle was well below that of the melting-point of silver; i.e., 962 C., or that of the gold-silver alloy. From the foregoing, it is evident that the temperature of the muffle is not by any means the same as that of the cupellation. Roberts-Austen 1 quotes Dr. Van Riemsdijk, stating that "he observed that a globule of 1 An Introduction to the Study of Metallurgy, 5th Ed., p. 50, citing Ann. de Chim. et Phys. t. XX (1880>, 66. 94 A MANUAL OF FIRE ASSAYING gold or silver in a fused state will pass below its solidifying point without actually solidifying, but the slightest touch with a metallic point will cause the metal to solidify and the consequent release of its latent heat of fusion is sufficient to raise the globule to the melting-point again, as is indicated by the brilliant glow which the button emits." Rose 1 also quotes the same author, and it is evident that the gold and silver globules mentioned are derived from cupellation. Six sets of experiments were carried on in this connection, some of which are plotted in Figs. 52, 53, and 54. It is evident from these Figs, that surfusion unquestionably occurs, and in a most marked manner, the greatest degree of surfusion noted being 77 C. All of the buttons " sprouted, " i.e., showed cauli- flower-like growths of silver on final solidification. This sprout- ing has always been considered a sign of purity of the silver, 2 particularly pointing to the absence of lead. In order to test this point, some of the silver buttons from the experiments were very carefully examined for lead in quantities of a gram, and showed but traces of it, quantities not determinable. Some showed minute quantities of copper. In effect they were all "fine silver." The surfusion is therefore very real. In the authorities cited on surfusion, the statement is made that on solidification from surfusion, the "flash" of the button occurs, showing the raising of the temperature to the melting-point of the silver. H. M. Howe 3 states: "Once freezing sets in (in the surfused metal or alloy) the heat which it evolves raises the temperature toward, and more often quite to, the true freezing- point, where it remains during the remainder of the freezing." In experiments carried on with the following quantities of silver, 10, 14, 18, 30.4 and 30 grams, the "flash" was not observable, neither by the eye nor by any actual rising deflection of the galvanometer pointer, although a repeated and careful search was made for this. In order to determine whether the size of the button had any influence on the "flashing," various amounts of silver, beginning with 350 mgs. and varying by 50 mgs. up to 850 mgs., were cupelled so as to finish with surfusion. It was found that the beads up to and including 650 mgs. flashed markedly, that of 700 mgs. faintly only, and those above showed no " flash. " 1 Metallurgy of Gold, 4th Ed., p. 598. 2 Rose, Metallurgy of Gold, p. 477. Collins, Metallurgy of Silver, 1900, p. 2. Schnabel, Metall-Huettenkunde, 1901, p. 605, 2nd Ed. 3 Iron, Steel and Other Alloys, 1903, p. 20. CUPELLATION 95 If the differences in temperature between the cupelling alloy and the muffle blank at any time interval be plotted as ordinates from a basal line, it is readily shown by the different curves, that the greatest difference occurs at the close of the cupellation; in some instances, just as the last of the lead oxidizes (play of colors). The differences noted show the marked evolution of heat at the "finishing" of the cupellation, and are due to the release of the latent heat of fusion. In the case of the large Difference Carve FIG. 52. CURVE SHOWING TEMPERATURE DURING CUPELLATION OF Pb = Ag. buttons, however, this does not seem to be sufficient to cause an actual rise of temperature in the cupel, when the muffle tempera- ture is actively sinking, as was the case in experiments shown by Figs. 52, 53 and 54. As already stated, however, no "flash" was observable to the eye in the larger silver buttons, nor did the galvanometer indicate it, as surely might be expected. The "lag" or time interval between the occurrence of a temperature and its recording by the galvanometer, is not great when an iron protective tube is employed. This is shown very plainly by 96 A MANUAL OF FIRE ASSAYING Note:- Constant difference betwi Cupel and Muffle Blan!s=20 < 18 Grams, Ag. 1 85 - Pb. o d eJ 8 1 I 3 Difference Curve Basal Line FIQ. 53. CunvE SHOWING TEMPERATURE DURING CUPELLATION OF Pb = Ag. Note:- Shortly before addition of Alloy, Temperatures of Cupel and Muffle uk were ideuttc 10 Grams, Ag. Pb. 0248 8101214101820222426283332313838404244484850525458580062646868707274767880 Miuute* Fia. 54. CURVE SHOWING TEMPERATURE DURING CUPELLATION OF Pb = Ag. CUPELLATION 97 the marked "jogs" in the cupel curves when the lead alloy is added to the cupels (Figs. 53 and 54). The lowest "finishing" temperature found showed surfusion extending to about 885 C. or 77 C. below the melting-point of silver. The temperature indicated by the "muffle blank," which at the start was 10 be.low the cupel, was 845 C. The last of the lead went off from the cupel at 910. This represents about the minimum "finish- ing temperature," judging by the general appearance of the cupellation. It is to be noted, however, that this "finishing temperature " is reached automatically in the cases where the muffle temperature is such as to afford "uncovering" of the but- ton and prevention of freezing; i.e., approximately 830 to 840 C. on the average, in the case .of pure lead buttons. One experi- ment was carried out in which the cupelling lead alloy showed a temperature of 750 01 not far from the end of cupellation, but at this temperature the button solidified into a lead-silver alloy. The approximate composition of the silver-lead alloy freezing at this temperature is 70 per cent. Ag, 30 per cent Pb. Resume. It appears from the foregoing that: 1. In the case of the lead buttons not containing any ap- preciable amount of copper or iron, etc., a muffle temperature of at least 800 C. and, better, one of 850 is required to "uncover" or start cupellation. 2. That this temperature may be lowered to about 770 C. during the oxidation of the greater part of the lead. 3. That toward the end of the cupellation or the "finishing," in case of silver, it must again be raised to about 830 C. in order to get a pure silver button. 4. That the actual temperature of the cupelling lead is always appreciably higher than the muffle temperature. 5. That the actual finishing temperature of the cupellation cannot safely be carried below about 910 C. 6. That the greatest observed surfusion of silver was 77 C. and that this is probably very near the maximum. 7. That silver beads finishing with surfusion are free from lead. 8. That "feathers" or crystals of sublimed litharge on the cupel are an indication of the proper cupellation temperature, provided the air draft is not excessive. 9. That it is just as essential to regulate the air draft of the muffle as its temperature. For an explanation of this seemingly low cupellation temperature see p. 92. 7 98 A MANUAL OF FIRE ASSAYING AVhere very accurate cupellation work is required, such as in bullion assaying and where the amount of work justifies it, a furnace designed for close temperature and air control is practically essential. In view of the recent improvement in electrically heated furnaces, in which temperatures can be rapidly and ac- curately controlled, and the muffle heated uniformly, practically eliminating the thermal gradient, a furnace of this type would seem best adapted for the work. INFLUENCE OF BASE METAL IMPURITIES. When the lead buttons are contaminated with base metals, such as copper, the temperature of cupellation must be higher in order to pre- vent freezing. The reason for this is readily apparent when the freezing-point curve (see Fig. 49) of the lead-copper series of alloys is inspected. The freezing-point of an alloy contain- ing 10 per cent. Cu and 90 per cent. Pb is 900 C. While the original copper percentage in the lead button may be quite small, the copper does not oxidize as readily as the lead, and tends to concentrate in the button, rapidly raising the melting-point of the alloy. For the removal of copper in cupellation the ratio of Pb to Cu should be at least 200 to 1 or more. Even then Cu will be re- tained by the silver and gold in small amounts. If it is less than this considerable copper is very apt to be retained with the silver and gold. In order to cupel at all, the ratio of Pb to Cu must be at least 20 to 1. In general, buttons to be cupelled should be free from base metal impurities. If they are unavoid- ably present in the button from the crucible assay, the base metals should be removed by scorification before cupellation. Impurities in lead buttons are detected by the behavior of the button. Zn, As, Sb, and S tend to make the button brittle when hammered; iron and copper, etc., tend to make it hard. PbO in the lead button makes it brittle. PbO is often found in lead buttons that have been produced at too low a temperature. Where the gold and silver contents of the lead button approach 30 per cent, of its weight, it is brittle. However, impurities in the lead button will not always be indicated by brittleness or hardness; without these characteristics, impurities may still be present in sufficient amount to cause loss. All impurities do not cause like amounts of loss in cupellation. The loss due to the presence of impurities is chiefly in absorption by the cupel, and comparatively small by volatilization. CUPELLATION The accompanying table 1 shows the influence of impurities. Twenty-five-gram lead buttons were cupelled, containing 1 gram of the impurity specified, 4 mgs. of Ag, and 1 mg. of Au. The temperature of cupellation was 1000 C., in order to prevent freezing as a result of impurity. The high losses are due in part to the high temperature em- ployed. The table really gives the relative influence of the im- purities. Bismuth has been used in place of lead for cupellation. While in the table bismuth is stated to be the cause of a very heavy absorption, this is not substantiated by other researches. 2 When it is present in the lead button it tends to concentrate dur- ing the cupellation, and is removed by oxidation toward the last of the operation. Some of it is very apt to be retained by the pre- cious metal bead. Cupellation may be carried on with bismuth, but the absorption is much higher. 3 The presence of Bi in the cold cupel may be recognized by the fact that the place which the silver button occupies is brown and surrounded by concentric rings of a yellow and blackish-green color. Copper colors the cupel from a dirty green to a black, dependent on the amount of copper. TABLE XIII .--INFLUENCE OF IMPURITIES Impurity Loss of Gold Loss of Silver Remarks None Tin 1 . 2 per cent. 2 . per cent. 11 .8 per cent. 13 . 9 per cent. Arsenic Antimony. Zinc Cadmium Iron Manganese Molybdenum Vanadium Copper 3 . 9 per cent. 5.3 per cent. 9 . 3 per cent. 3 . 5 per cent. 4.0 per cent. 13.6 per cent. 11.0 per cent. 7 . 7 per cent. 10 per cent 16.3 per cent. 13.3 per cent. 17.6 per cent. 13.1 per cent. 16.6 per cent. 24 . 3 per cent. 26.2 per cent. 21 . 7 per cent. 32 6 per cent. Most of this loss, even with Te and Se, is cupel absorp- tion. Bismuth 4 21 8 per cent. 27 9 per cent. Thallium Tellurium . . . .' Selenium 23 . 1 per cent. 55.8 per cent. 54 . 1 per cent. 34 . 4 per cent. 67 . 9 per cent. 64 . 5 per cent. 1 T. K. Rose, in Jour. Chem. Met. and Min. Soc. of S. A., Jan , 1905. 2 K Sander, Berg- und Huettenmaennische Zeitung, 1903, p. 81. See also Min. Ind., XII, 244. 3 Smith, in Jour. Chem. Soc., 1894, 863. 4 Doubtful. 100 A MANUAL OF FIRE ASSAYING Tin, arsenic, zinc, cadmium, iron, and manganese cause scoria to form on the cupel, due to the formation of oxides which are not readily absorbed. Iron causes a dark coloration of the cupel. Antimony in considerable quantity causes the cupel to check and crack. The same may be said of copper. Copper. This metal is oxidized with more difficulty than lead, the Cu 2 O forming by aid of the action of PbO; however, Cu 2 O, again coming into contact with metallic lead, is reduced to Cu, and in this way is persistent tow r ard the end of the cupellation, although a large excess of Pb over Cu is present, and finally some remains with the Au and Ag. The loss of silver during the cupel- lation is due mainly to absorption, in large part as oxide. This oxidation of the silver in the presence of much lead is not to be ascribed to the action of atmospheric oxygen, but rather to "oxygen carriers," such as PbO, Cu 2 O, etc. It is very probable that Cu 2 O acts peculiarly in this manner, and the high absorption noticed when Cu is present is due to this fact. It is to be noted that losses in silver occur toward the end of the cupellation, and occur in great part just before finishing; the small black-green rings, surrounding the place where the silver bead rests, locates most of the silver. It is the concentration of the copper, silver, and gold that causes the high absorption. Lodge 1 shows the influence of small amounts of copper on the cupellation of silver and gold. TABLE XIV. COPPER IN CUPELLATION OF SILVER AND GOLD Silver milli- grams Lead ; Copper grams grams Percentage Temperature Percent- of copper degrees cen- age of in lead \ tigrade 2 loss Ratio Pb toCu 202 10 '0.0101 0.1 775 1.05 1000 to 1 203 10 0.0202 0.2 775 1.08 500 to 1 202 IP 0.0303 0.3 775 1.29 333 to 1 202 10 0.0404 0.4 775 1.45 250 to 1 204 10 0.0500 0.5 775 Cu re- 200 to 1 tained 1 "Notes on Assaying," p. 143 et seg. 3 Temperature of air in muffle. CUPELLATIOX 101 Gold milli- grams T , Percentage Temperature of copper degrees cen- in Pb tigrade 1 Percentage of loss Ratio Pb toCu 202 10 no. 775 0.155 202 10 0.1 775 0. 19 2 All contained ' 1000 to 1 201 10 0.2 775 0.20 copper on 500 to 1 200 10 0.3 775 0.13 finishing 333 to 1 201 10 0.4 775 0.165 250 to 1 202 10 0.5 775 0.250 200 to 1 Gold is more retentive of copper than silver. It is to be noted that even with a ratio of 200 Pb to 1 Cu, it is not possible to remove all copper, and beads obtained from mattes and heavy copper ores should be examined for copper; otherwise silver results may be high. Retained copper in these silver beads will compensate for loss of silver, but the amount retained is so vari- able that this error cannot be considered to compensate the loss. Tellurium. Tellurium has a great affinity for gold and silver, and if present in an ore in any appreciable amount, some of it will go into the lead button with the gold and silver, and thus have its influence on the cupellation. It tends to concentrate during the cupellation and is with difficulty removed by oxidation. When there is present in the lead button more than 15 per cent, of the gold and silver weight in tellurium, the beads resulting from cupellation have a dull and frosted appearance. Larger amounts than this cause the beads to divide and split up in the cupel. F. C. Smith 3 shows the influence of tellurium on the cupellation as follows, these results being confirmed by J. C. Bailar 4 and others. 1 Temperate of air in muffle. 2 Actual losses; copper retained, 0.16 per cent. Gold about the same weight as before cupellation. 3 "The Occurrence and Behavior of Tellurium in Gold Ores," etc., in Trans. A. I. M. E, XXVI, 495. 4 "West. Chem. and Met," I, 119. 102 A MANUAL OF FIRE ASSAYING TABLE XV. TELLURIUM IN CUPELLATION OF GOLD AND SILVER Mgs. of bullion Containing Mgs.Te added Loss by absorption Loss by volatili- zation Au Ag Au Ag Au Ag 29.8 24.76 5.04 5.0 per cent. 13.44 per cent. 27.08 per cent. 5.65 per cent. 0.69 28.45 23.64 4.81 15.0 34.22 35.78 5.28 1.75 22.17 18.42 3.75 15. 1 29.85 32.01 11.92 17.95 CUPELLED WITH 12 GRAMS OP LEAD Note the similar effect of selenium. Antimony. The presence of antimony causes increased losses by absorption, although its effect is not as pronounced as that of copper or tellurium. During the cupellation litharge and antimony combine to form antimoniate of lead, which, if present in considerable amount, may cause the formation of scoria on the cupel. Small amounts of antimony tend to remain with the gold and silver, as with copper and tellurium. As a guide in cupellation, the following scale of color temper- atures is given. 2 Degrees Centigrade Lowest red visible in the dark 470 Dark blood-red or black-red 532 Dark red, blood-red, low red 566 Dark cherry-red 635 Cherry-red, full red 746 Light cherry, light red 843 Orange 900 Light orange 941 Yellow 1000 Light yellow 1080 White 1205 CUPELLATION IN CUPELS OF DIFFERENT MATERIAL. The cupel material has a decided influence on the progress of a cupella- tion. What has preceded refers more particularly to bone-ash 1 Selenium instead of tellurium. 2 White and Taylor, in Trans. Am. Soc. Mch. Eng., XXI, 628. H. M. Howe, in Eng. and Min. Jour., LXIX, 75. CTJPELLATION 103 cupels. In cupels with a magnesia base the process as regards tem- perature differs somewhat, due to the different thermal properties of the two types of material. The following difference in thermal properties may be noted. 1 Bone-ash cupel, mean specific heat between 15 and 100 C. is 0.185. Magnesia cupel, mean specific heat for same temperatures is 0.215. A bone-ash and magnesia cupel of identical volumes weigh respectively 22 and 29 grams. The heat conductivity of magnesia cupels is very much greater than that of bone-ash cupels. When the two types of cupels are heated to 90 C. in a steam bath, at the end of 14 minutes the magnesia cupels are at 90 C. and the bone-ash cupels at only 60 C. During cupellation of lead at the end of 6 minutes from the addition of the button the magnesia cupel showed practically the same temperature in the cupelling lead as in the bottom of the cupel, viz. 920 C., while the bone-ash cupel in the same muffle showed a temperature of 990 C. for the cupelling lead, and only, 932 C. in the bottom. The total heat capacity of a magnesia cupel is more than 50 per cent, greater than that of a bone-ash cupel of the same volume, so that on cooling the two types of cupel the magnesia cupel retains a higher temperature somewhat longer than the bone-ash cupel in spite of its greater diffusivity of heat. From this data the reason of the behavior of magnesia and bone-ash cupels during cupellation is apparent. It will be noted: (1) That in magnesia cupels the lead is less bright and hence at a lower temperature than in bone-ash cupels, although the muffle temperature is the same. This is due to the fact that the extra heat generated by the combustion of the lead is dif- fused as rapidly as generated by the superior diffusivity of the magnesia cupel and hence cannot serve to raise the temperature of the lead, as is the case in the bone-ash cupel. Hence for the same "muffle temperature" the actual cupellation tempera- ture of the lead in the magnesia cupels is 50 to 60 C. lower than in the bone-ash cupels. To this fact is due the lower losses of precious metal in magnesia than in bone-ash cupels. From the discussion under " cupellation temperature" it will have been noted that with bone-ash cupels, if once the muffle has attained a temperature sufficiently high to cause the uncovering of the button, the rise in temperature of the lead due to its oxidation, is sufficient to carry the cupellation to a finish pro- vided the muffle temperature is not lowered at the end of the 1 Bannister and Stanley, "Thermal Properties of Cupels." Bui. 56. I. XL M. (1909). 104 A MANUAL OF FIRE ASSAYING operation. This is not the case with magnesia cupels for now obvious reasons* and it will be necessary to raise the muffle tem- perature toward the end of the operation or what amounts to the same thing, push the cupel to the hotter part of the muffle. Assayers who are used to bone-ash cupels, therefore, have some difficulty at first due to "freezing" of buttons when using magnesia cupels. 2. Magnesia cupels retain a higher temperature longer than bone-ash cupels when withdrawn from the furnace or moved to the cool part of the muffle, and hence silver buttons show a lesser tendency to sprout, due to the slow cooling they undergo. The lead in magnesia cupels seems to open somewhat more readily and cupels slightly faster than in bone-ash cupels. The accompanying tables give data of results obtained by bone- ash and magnesia cupels on pure silver and on a copper matte. 1 TABLE XVI. COMPARISON OF BONE-ASH AND MAGNESIA CUPELS ON C. P. SILVER CUPELLED WITH 10 GRAMS SHEET LEAD Amount of silver taken, mgs. Bone-ash cupels. Silver bead, weight, mgs. Magnesia cupels (Morganite) . Silver bead, weight, mgs. 5 4.85 4.80 10 10.00 10.00 15 14.72 20 19.30 25 24.41 15 14.36 14.50 20 ' 18.92 19.52 25 23.84 5 4.94 4.89 10 9.68 9.86 15 14.70 14.80 20 19.98 19.68 25 24.60 24.84 The sheet lead used contained a little silver. Cupellation in most cases was carried out with feathers. It is to be noted that when low finishing temperatures are employed, as is apt to be the case with magnesia cupels, the beads may retain small 1 By O. A. Anderson and C. H. Fulton, S. D. School of Mines, Laboratory. CUPELLATION 105 amounts of other metals notably, lead, 1 to which may be due in some cases the higher results obtained. TABLE XVII. COMPARISON OF BONE-ASH AND MAGNESIA CUPELS ON A COPPER MATTE No. Weight lead button grams Matte taken a. t. Bone-ash cupels Magnesia cupels(Morg.) Au + Ag Ag Au Au+Ag Ag Au 1 2 3 4 5 6 7 8 9 10 11 18 24 28 30 14 18 25 10 14 31 24 0.05 0.10 0.15 0.25 0.05 0.10 0.20 0.05 0.10 0.20 0.25 11.0 21.12 31.30 56.10 11.0 21.8 44.5 10.9 21.3 42.6 53.0 10.38 19.80 29.22 52.82 10.36 19.48 41.90 10.30 20.10 40.04 49.75 0.62 1.32 2.08 3.28 0.64 1.32 2.60 0.60 1.20 2.56 3.25 13.0 23.3 27.2 12.37 22.02 25.17 0.63 1.28 2.03 12.50 27.0 47.0 11.2 22.0 43.1 53.6 11.82 25.67 44.37 10.58 20.70 40.52 50.28 0.68 1.33 2.63 0.62 1.30 2.58 3.32 The assays given in the table were made by the excess litharge method. The average result stated in ounce per ton is as follows: for bone-ash cupels, gold 12.86 oz., silver 202.67 oz.; for magnesia cupels, gold 13.16 oz., silver 222.76 oz. These results are un- corrected assays, viz., do not include the slag or cupel absorption. In practice it was found necessary to make these corrections to obtain concordant results. It will be noted that the magnesia cupels give higher results on gold and very much higher results on silver. This last is without question due in large part to the retention of copper by the beads, and calls for caution in the use of magnesia cupels on this type of material. Portland Cement Cupels. During cupellation Portland cement cupels act very similarly to bone-ash cupels. The loss is some- what higher than in bone-ash cupels. The accompanying table gives losses in Portland cement cupels and bone-ash cupels and those made of one-half of each material. The temperatures are average temperatures during cupellation, from the opening of the button to the "blick." One hundred mgs. of silver were D. M. Liddell, Eng. and Min. Jour., LXXXIX. 254. 106 A MANUAL OF FIRE ASSAYING cupelled with about 20 grams of lead. 1 The temperatures were measured by inserting a thermocouple into a hole bored beneath the bowl of the cupel. They hence represent a temperature which is a mean between that of the cupelling lead and a muffle "blank" cupel. TABLE XVIII. CUPELLATION LOSSES WITH DIFFERENT TYPES- OF CUPELS. Average U. S. Portland R. D. Portland One-half cement, Bone temp. cement, cement, one-half bone-ash, ash, loss deg. C. loss per cent. { loss per cent. loss per cent. per cent. 915 1.30 1.34 1.21 1.26 925 1.81 1.72 1.54 1.70 945 2.53 2.56 2.42 2.42 965 3.37 3.42 3.05 2.96 Another test to determine the relative absorption of bone-ash and cement cupels 2 gave the following results: On 10 mgs. silver with 15 grams lead, at an orange heat (very high) cement cupels showed 6.64 per cent, absorption and bone-ash cupels, 6.38 per cent. At a light cherry heat, cement cupels showed 4.91 per cente and bone-ash 4.62 per cent, absorption. It is to be noted that the percentage absorption other factors being equal is dependent on the amount of precious metal cupelled (see p. 163). In using cement cupels, the beads must be carefully cleaned otherwise when parting in nitric acid insoluble silica is apt to remain which will be weighed as gold. The bead on cement cupels is likely to be more flat than on bone-ash cupels. 1 Holt and Christensen, Eng. and Min. Jour., XC, 560. "Experiments with Portland Cement Cupels." 2 J. W. Merritt, "Cement vs. Bone-ash Cupels," Min. and Set. Press., C, 649. CHAPTER VIII PARTING Parting is the separation of gold from silver by means of acid. In assaying, nitric acid is almost exclusively used, although sulphuric acid may be employed. In order to separate silver from gold by means of acid, it is essential that there be present at least twice as much silver as gold. When less silver is present, it is impossible to separate all of the silver from gold by means of acid (see assay of gold bullion, in Chapter XII). When the above-stated amount is present, it requires acid of not less than 1.26 specific gravity, boiling for at least 20 or 30 minutes, to separate the silver from gold. The ratio of 2 and 2.5 to 1 is used practically only in the bullion assay. In parting beads from ore assays, it is considered necessary to have at least five times as much silver as gold present. The addition of silver to gold or to the gold-silver alloy in order to prepare for parting is termed "inquartation," from the fact that at least 3 parts of silver to 1 part of gold were formerly con- sidered necessary. The nitric acid usted for parting must be free from hydrochloric acid and chlorine in order not to have a solvent action on the gold. 1 Nitric acid should be examined for chlorides before being used for parting. In order to part silver from gold successfully, the following points must receive careful considera- tion: (1) The strength of the acid used; (2) the temperature of the acid; (3) the ratio of gold to silver in the bead to be parted. 1. The proper strength of acid is of great importance. For- merly, most authorities recommended that acids of 1.16 and 1.26 sp. gr. respectively 2 parts water to 1 of acid (1.42 sp. gr.) and 1 of water to 1 of acid be used, first the weak acid and then the stronger acid. T. K. Rose recommends 4 parts acid to 3 parts water, which strength, if the acid be heated, will not break up the gold in the bead into fine particles, even if 50 parts of silver are present to 1 part of gold. Gold is less apt to break up when it is less than 0.10 mg. in weight. Keller 2 recommends 1 Consult the caption "Solution of Gold by HNO 3 ," in Chapter XI. 2 Keller, Trans. A. I. M. E., XXXVI, 3. 107 108 A MANUAL OP FIRE ASSAYING acid of the following strength: 1 part acid (sp. gr. 1.42) to 9 parts distilled water. In this strength of acid the gold almost invari- ably remains in a coherent mass, even when the silver is 500 times as much as the gold. This is the strength of acid recom- mended for ordinary assay purposes. The beads should be boiled in the acid for at least 10 to 15 minutes in order to insure parting. 2. It is essential to have the acid at the boiling-point before dropping in it the bead to be parted. Putting the bead into cold acid and heating up gradually is almost certain to leave the gold, especially where the ratio of silver to gold is high, in a powdered, fine condition, very apt to cause losses in washing and subsequent handling of the gold. Cold acid should not be used. 3. While the best ratio of silver to gold, for parting ordinary beads, is 5 to 1, this ratio is not always under control, since the assayer must be content in many cases with the ratio that the ore furnishes him, when this is more than 5 to 1. If less than 5 to 1, silver should be added in order to bring it up to this ratio. The silver may be added directly to the crucible or scorification fusion, or to the lead button during cupellation if it is not essential to determine the silver in the ore. If it is essential to determine the silver, and inquartation is necessary, the bead from the cupellation is first weighed, the requisite amount of silver is added to the bead, both wrapped up in about 2 grams of sheet lead, and then it is recupelled and parted. Beads which need inquartation may also be fused with silver, on a piece of charcoal, by means of the blowpipe; but this method is not to be recommended, as it frequently occasions loss. Many assayers, if they suspect an ore to be deficient in silver for parting, add silver to the crucible, not determining the silver in this assay, but running a separate scorification assay for this purpose. Another way 1 is to add to the charge a desired number of cubic centimeters of AgN0 3 solution of such strength that 1 c.c. contains 1 mg. Ag. The proper deduction can then be made from the weight of the bead, but some allowance must be made for the silver absorbed by the cupel. After parting, the acid is poured from the parting cup or flask in which the operation has been conducted, and the gold residue is washed, at least three times, with warm distilled water 1 F. G. Hawley, Eng. and Min. Jour., XC, 649. PARTING 109 in order to remove all trace of silver nitrate. The black stain occurring in parting cups after heating for the annealing of the gold is due to metallic silver reduced from silver nitrate by the FIG. 56. PARTING FLASKS. heat, showing insufficient washing. Parting may be carried on in small porcelain crucibles called " parting cups, " or in test- tubes, or in flasks similar to copper-assay flasks. In order to 110 A MANUAL OF FIRE ASSAYING part in flasks or test-tubes, it is essential to have the gold stay as a coherent mass, so as to prevent loss in transference. When parting cups are used, after washing, the gold is carefully dried and the gold annealed at a dull-red heat, either in the muffle or by means of the blowpipe. After acid treatment, the gold is left as a soft black mass, probably an allotropic condition of the gold; but upon heating this is changed to the normal yellow metallic state in which it is weighed. Fig. 55 shows a convenient parting bath with test-tubes; Fig. 56 shows parting flasks commonly in CHAPTER IX THE ASSAY OF ORES CONTAINING IMPURITIES Impurities, from the assayer's point of view, are such sub- stances, contained in ores, furnace products, or other material, as necessitate some particular method of assay or treatment, or the observing of special precautions not included in the ordinary crucible assay as already outlined. Common impurities are sulphur, arsenic, tellurium, antimony, zinc, copper, etc. Of these sulphur is by far the most common. In performing an assay it is usually the aim of the assayer, whenever this is possible, to produce by direct fusion, either by the crucible or scorification method, a pure lead button weighing approximately 20 grams. If the button is smaller than this, there is danger of not collecting the values; if larger, cupellation is too prolonged and losses are increased. In the assay of low- grade gold ores it may be desirable to produce lead buttons of 25 to 30 grams in order to obtain the best results. The impurities mentioned affect either the size of the button, or the purity of the button, or both. To show the effect of sulphur the following definite example is taken. Given an ore containing pyrite, which, in a charge yielding the ordinary type of monosilicate slag, gives a reducing power of 5 grams of lead per gram of ore. If the following charge, 15 grams of ore 70 grams of PbO 30 grams of Na 2 CO, 8 grams of SiO 2 Borax glass cover be made up and fused, a 60-gram button (approximately) will be produced, on top of which will be a small quantity of " matte, " i.e., an artificial sulphide of the metals, in this case iron and lead. This matte is brittle and may contain some silver and a little gold. On hammering the button, it is lost. In general, it is an undesir- able product to make. A small amount of matte is produced in this case, since the ore has the power to reduce 75 grams of lead from PbO, while only 70 grams of PbO are present, so that the 111 112 A MANUAL OF FIRE ASSAYING excess sulphide of the ore not acted upon by the PbO remains in the charge, uniting with some of the lead to forma sulphide of iron and lead. The button is also much too large to cupel. If in the charge the PbO is materially increased, the ore will react to the extent of its full reducing power, a lead button of 75 grams will be produced, no matte will be found, and the slag will be improved, owing to the addition to it of the fusible base PbO. If the PbO in the charge be materially reduced, the lead button will be much smaller (owing to the dearth of PbO available for reduction), considerable matte will be formed, and the slag will be poor. If the silica be increased, so that sufficient be present to form the higher silicates with all the bases present, practically no lead will be reduced, for the sulphide has not the power to reduce miich Pb from lead soda silicates unless a free base be present, e.g., (PbO.Na 2 O)2SiO 2 +FeS 2 (no action), or possibly (PbO.Na 2 0)2Si0 2 +FeS 2 = (FeO.Na 2 O)2SlO 2 +PbS + S. In this way the sulphur remains in the charge in the form of sulphide sulphur. Soda will cause the formation of SO 3 , if PbO is present to furnish the oxygen, and if it can act as a free base, i.e., if it is not combined with silica (see Chapter V, on Reduction and Oxidation Reactions) . An increase of soda without an increase of PbO or SiO 2 will lessen the amount of matte, as sulphur will tend to combine to some extent with the Na 2 to form, w r ith the FeS, a double sulphide of iron and soda,. etc., which will be dis- solved in the slag. The above outlines the effect of such impur- ities as sulphur and arsenic, and shows the necessity of special methods of assay directed toward the getting rid of impurities. The impurities mentioned may be divided into two classes: (a) Those which can be volatilized by oxidation or otherwise, e.g., sulphur, arsenic, and antimony! (6) Those which cannot be volatilized, e.g., copper, zinc, etc. Some of these may be partly volatilized, as antimony and zinc. For the removal of all of them, however, whether by volatilization or by slagging, oxidation is essential. In one method employed on light sulphide or arsenic ores, the iron-nail method, sulphur and arsenic are carried into the slag as a double sulphide or arsenide of soda and iron, etc. Crucible Fusions. ASSAY OF ORES CONTAINING IMPURITIES 113 The following methods are standard methods for the assay of impure ores, and are discussed in detail: 1. The roasting method. 2. The niter method. (a) The common niter method. (6) Miller's oxide slag method, (c) Perkins' excess-litharge method. 3. The iron-nail method. (a) The niter-iron method. 4. The cyanide method (rarely used). *5. .The scorification method. 6. The combination wet-and-dry method (removal of impurities by solution) . THE ROASTING METHOD. It is usual to carefully weigh out 0.5 or 1 assay ton of the ore to be assayed, and place it in a roast- ing dish of sufficient size to permit of stirring without loss by spilling. The dish is placed in the muffle, the temperature of which is not above a " black red " and the firing of which is under good control, so that the temperature will not rise too rapidly. In the case of an ordinary sulphide ore, such as a pyrite, or, for example, a chalcopyrite and quartz, the following reactions take place, if the roasting is carried on slowly at a low heat: 1 3CuFeS 2 + ISO + heat = Cu 2 S + 3FeSO 4 + CuSO 4 + SO 2 At 590 C. the ferrous sulphate decomposes spontaneously, sulphatizing the balance of the copper: Cu 2 S + 2FeSO 4 +6O = 2CuSO 4 +Fe 2 O 3 + SO 3 At 655 C. the copper sulphate decomposes 'into basic sulphate and SO 3 , and at 700 C. into CuO and SO 3 , as follows: 2CuSO 4 = CuO.CuSO 4 + S0 3 , CuO.CuS0 4 = 2CuO+S0 3 ; so that the final products of the roast, when carried to above 700 C., are ferric and cupric oxide, with a complete removal of the sulphur. If the temperature is not carried above 700 C., sulphur remains in the charge as sulphate, which may again be reduced in the crucible to sulphides: 2CuSO 4 + 3C = Cu 2 S + S0 2 + 3CO 2 If, for any reason, it is not desirable to carry the temperature as high as 700 C., the ore, after roasting until no further smell 1 R. H. Bradford, Trans. A. I. M. E., XXXIII, 68. 114 A MANUAL OF FIRE ASSAYING of SO 2 is discernible, is cooled and mixed with 5 to 10 grams of powdered (NH 4 ) 2 CO 3 , and reroasted at a low heat, the sulphuric anhydride (SO 3 ) being eliminated as volatile ammonium.sulphate, (NH 4 ) 2 S0 4 : CuS0 4 + (NH 4 ) 2 C0 3 = CuO + (NH 4 ) 2 SO 4 + C0 2 Any silver in the ore that has been roasted will be in the form of Ag 2 SO 4 , or if arsenic and antimony are present, partly in the form of arseniates and antimoniates. If the roasting temperature is carried to 870 C. and above, the silver sulphate will be de- composed, leaving the silver in the form of metallic silver. In order to avoid loss of silver it is best not to carry the temperature above 700 C. In roasting simple pyrite ores, the reactions are similar, but simpler, and the temperature need not be carried above 600 C. During roasting, the ore should be stirred frequently in order to expose fresh surfaces to oxidation. When ores contain arsenic and antimony, the roasting opera- tion is more difficult and complex, and considerable care and skill are required to eliminate the greater part of these two volatile elements. The reason for this is that the arsenic and antimony pass by roasting first to the state of the lower oxides As 2 3 , Sb 2 O 3 , which are volatile, and then to the state of the higher oxides As 2 O 5 , Sb 2 O 5 , forming arseniates and antimoniates of cer- tain metals present in the ore, some of which are stable even at high temperatures, thus fixing the arsenic and antimony in the roasted ore, and not eliminating it. The arseniates (or anti- moniates) which ordinarily form are those of copper, iron and silver. The best conditions for the elimination of arsenic and antimony are alternate oxidation and reduction at a low heat. The presence of sulphur tends to aid the elimination of arsenic and antimony by the formation of the volatile sulphides of these elements. The reducing action necessary for the elimination of arsenic and antimony is best obtained by mixing w T ith the ore equal volumes of coal dust or charcoal, and roasting at a dark red heat until the coal is burnt off, then cooling, adding more coal dust, and reroasting. In this way the greater part of the arsenic and antimony can be readily volatilized, except in very rich silver ores. When galena ores are to be roasted, the ore is best mixed with an equal volume of silica and roasted at a very low heat. In this roast PbSO 4 is formed to a considerable extent, ASSAY OF ORES CONTAINING IMPURITIES 115 which at a higher heat is decomposed by the SiO 3 present, as follows: PbSO 4 + SiO 2 = PbSiO 3 + SO 3 Care must be taken with this roast as, at the formation point of lead silicate, silver losses are apt to occur. A successful roast will be indicated by a yellow color (lead silicate), and an unsuc- cessful one by a black or gray color (fused, undecomposed sul- phides). In general, heavy sulphide ores that contain their chief value in gold may be roasted, when this is carefully done, without loss of gold; but silver ores, especially when of high grade, are apt to give low results. In making up the charge for the roasted ore, it is to be noted that from a sulphide ore (pyrite, etc.) the product is frequently of an oxidizing nature and basic, which must be taken into account in adding the fluxes. In galena ores, when silica has been added, this must be accounted for. The roasting method is frequently used for heavy sulphide ores, especially when they have a low value in gold and silver, as it permits of a large amount of ore being taken (1 assay ton and more), which after roasting presents no difficulty in making the proper fusion. THE NITER METHOD. The first step in the niter method is the making of a preliminary assay according to the directions already given. The precautions concerning the reducing power of the sulphides in different types of charges must be carefully noted; it is best to have the preliminary charge of the same composition as the final assay charge. Or else the reducing power may be determined by the soda-litharge charge and this cut down by 25 per cent., 20 grams deducted for the lead button, and the remainder divided by 4 to get the amount of niter to add, in grams, if the monosilicate slag is to be made in the assay. The amount of ore taken for the niter assay varies according to the grade of the ore in gold and silver and according to the amount of impurity present. It is rarely desirable to add more than 20 grams of niter to the charge, as larger amounts cause difficulty through the evolution of too much gas. One-half assay ton is the amount of ore most frequently taken. Sometimes, with ores containing much impurity, 0.10 to 0.25 assay ton is used. Twenty-gram crucibles (170 c.c. capacity) are used for amounts of 0.5 assay ton of ore and less, and 30-gram crucibles (240 c.c. capacity) for 1 assay ton of ore. 116 A MANUAL OF FIRE ASSAYING MILLER'S OXIDE-SLAG METHOD. This method is a modified niter method applicable to such ores as contain practically no silica; i.e., heavy sulphide ores, such as pyrites, arsenopyrite, mattes, etc. It is based on the fact that PbO has the power to hold in solution and in suspension oxides of such metals as copper, iron, etc. (see p. 122, where " scorification " is discussed), in certain amounts. Niter is added to oxidize the sulphides, etc., and Na 2 CO 3 to aid in the complete oxidation of the sulphur by the formation of sulphates, in the manner already discussed. The first step, as in the ordinary niter method, is the preliminary assay, according to the following charge: Ore 3 grams PbO 50 grams Na 2 CO 3 8 grams The final charge is as follows: Ore 0.5 assay ton PbO 70.0 grams Na 2 CO 3 12.0 grams KNO 3 (calcuated for a 20-gram button) Quick fires, 1100 C., 30 minutes, are found to be best. The slags are usually dull black and pour readily, and the button separates easily from the slag. (In slags high in silica or con- taining much borax, the lead buttons are apt to adhere closely to the slag.) With the oxide-slag method, trouble is sometimes experienced through the lead refusing to collect and remaining shotted through the slag. The difficulty is usually due to too much soda (especially if considerable niter is used) although too low a temperature of fusion is also a factor. The method gives reliable results on gold and silver, compar- ing well with the other standard methods. 1 PERKINS' EXCESS-LITHARGE METHOD. 2 This method is based on the fact that PbO will dissolve oxides of other metals and, if present in great excess, will prevent, to a large extent, the reduc- tion of other metals, such as Cu and Sb. The presence of so much PbO also insures a strongly oxidizing tendency in the crucible, preventing impurities entering into the button. It is desirable to add or have present SiO 2 in such an amount as will form a monosilicate with the bases present, including 1 Miller, Hall and Falk, "The Reduction of Lead from Litharge," etc., in Trans. A. I. M. E., XXXIV, 398. 399. * W. G. Perkins, "The Litharge Process," ibid., XXXI, 913. ASSAY OF ORES CONTAINING IMPURITIES 117 some litharge, but leaving much litharge uncombined in the "charge. The following table shows the proportion of PbO required to form fusible compounds with the principal metallic oxides: 1 TABLE XIX. PbO REQUIRED WITH METALLIC OXIDES One part of Cu 2 O CuO ZnO Fe 3 O FesO* MnO SnO 2 SbzOi As 2 Os Requires parts of PbO.. 1.5 1.8 84 10 10 13 5 1 In order to carry out the excess-litharge method intelligently, it is necessary to know the approximate composition of the ore, so as to provide the proper amount of PbO and SiO 2 . The best fusion exhibits, in a section of the cone of the slag after breaking, silicates of lead, iron, etc., on the outer surface, gradually passing to crystalline litharge toward the center. The temperature of fusion should not exceed 1050 to 1100 C. It must be above 884 C. (melting-point of PbO) . The first step is the making of a preliminary assay in order to determine the amount of niter to be added. 2 The final charge most frequently used is: Ore. 0.25 to 0.5 assay ton Na 2 CO 3 12 grams PbO 8 to 10 assay tons SiO 2 10 grams Niter to obtain 20-gram button The button is generally clean, and separates easily from the slag. The excess litharge method will give somewhat low results on silver, especially on high grade ores but will give good results on gold. In ores of the following analysis, SiO 2 40 to 60%; Fe, 5%; CaO, 2%; Pb, 15 to 40%; Zn, 2%; Ag, 20 to 80 oz.; S, 1%; and a trace of copper, the results in the accompanying table were obtained by the use of charges A, B and C. 3 Charge A Charge B Charge C PbO Borax glass. . . Flour 25 grams 4 grams 2 . 25 grams 50 grams 4 grams 2 . 25 grams 75 grams 4 grams 2 . 25 grams NaHCO 3 K,CO Ore. . . 25 . grams 25 . grams 0.5 a.t. 25 . grams 25 . grams 0.5 a.t. 25 . grams 25 . grams 0.5 a.t. 1 Hofman, "Metallurgy of Lead," p. 7. 2 In place of niter, it may be necessary, in this method or in Miller's method, to add argol, if ore is not reducing. 3 Kenneth Williams, Jour. Ind. and Eng. Chem., II, 406. 118 A MANUAL OF FIRE ASSAYING TABLE XX. AVERAGE RESULTS SHOWING EFFECT OF AN INCREASE OF PuO ON SILVER RESULTS Ore No. Charge A, ounces Ag per ton Charge B, ounces Ag per ton Charge C, ounces Ag per ton 1 51.05 50.86 50.62 2 42.36 42.20 42.02 3 27.82 27.70 27.55 An ore from Cobalt, Canada, 1 containing 5.06 per cent. Ni and 9.12 Co, chiefly as niccolite and smaltite, and some free silver was assayed by the following charge: Ore NaHCO 3 Borax glass Argol 0.05 a. t. 10 grams 10 grams 1 . 5 grams Litharge as given in table. TABLE XXL AVERAGE RESULTS SHOWING THE EFFECT OF AN INCREASED AMOUNT OF PnO ON SILVER RESULTS Litharge, grams Lead button, grams Silver, ounces per ton Silver in slag, ounces per ton Silver in cupel, ounces per ton 30 19 2051.4 9.6 34.0 40 21 2056.0 40 21 2050 80 30 1968.6 80 22 1944.6 135.2 35.0 80 21 1984.8 70.2 34.6 80 21 1914.8 THE IRON-NAIL METHOD. This method does not attempt to oxidize impurities, but aims to carry sulphur, etc., into the slag. The ore is decomposed by the iron nails added to the charge and by the PbO present. As iron reduces PbO to Pb, the amount of litharge added to the charge is limited to 25 to 30 grams. The amount of soda needed is large, as this flux is depended upon to carry the sulphur into the slag. The slag should be below a monosilicate in degree, and high in soda, as basic alkaline slags have a high solvent power for sulphides. 1 R. W. Lodge, "The Effect of High Litharge in the Crucible Assay for Silver," Trans. A. I. M. E., XXXVIII. 638. ASSAY OF OKES CONTAINING IMPURITIES 119 A typical charge on an ore that has a reducing power of about 4 grams of Pb per gram of ore is : l Ore 0.5 assay ton SiO 2 2 grams NaHCO 3 30 grams borax 8 grams PbO 30 grams nails 17 grams Salt cover The soda should usually be twice the amount of ore in the charge. The reactions that take place are approximately as follows: 7PbO + FeS 2 + 4NaHCO 3 = 7Pb + 2Na 2 SO 4 + FeO + 4CO 2 + 2H 2 O Part of the ore is decomposed by the PbO, and part of the S may go off as SO 2 , as discussed in previous pages. The iron nails decompose the balance of the sulphides: PbS+Fe = Pb+FeS (if galena is present or lead sulphide forms). The iron sulphide (FeS) is dissolved by the alkaline slag, forming probably double sulphides of soda and iron. To show the nature of the iron-nail fusion, the following results of two fusions on a pyrite ore containing 39.5 per cent. S a reducing power equal to about 8 are given: 2 Charge 1 Charge 2 1 assay ton ore 0.5 assay ton 30 grams NaHCO 3 30 grams 30 grams PbO 30 grams 4 grams SiO 2 4 grams 4 nails 4.... 10 grams borax glass cover. 10 grams The following results were obtained: No. 1 No. 2 Slag 60 grams 65 grams Matte 23.5 grams none Lead 24 . 5 grams 26 . 5 grams Crucible and charge before fusion 685 grams 662 grams Crucible and charge after fusion 665 grams 642 grams Loss in weight 20 grams 20 grams Nails before fusion 64 grams 63 grams Nails after fusion 43 grams 49 grams Loss of iron 21 grams 14 grams Per cent, of S in slag 6.73 7.63 Sin slag 4.03 grams 4. 96 grams S in ore 1 1 . 85 grams 5 . 92 grams S passed off as S0 2 0.95 grams 0.96 grams S in matte 6 . 87 grams none 1 Lodge, "Notes on Assaying," p. 99. 3 Lodge, iiid., p. 101. 120 A MANUAL OF FIRE ASSAYING It will be noted that the charges are identical as far as the fluxes are concerned, but that the amount of ore differs. It is desirable in heavy sulphide ores to keep the ore down to 0.5 assay ton and lower if necessary. Care must be taken not to have the slag above a monosilicate in degree, for if higher in SiO 2 there will be particular danger in this charge of not having the sulphides oxidized by the PbO, more sulphide being retained in the charge than it can dissolve, and forming a matte, even with small amount of ore. THE NITER-IRON METHOD. This method is in principle the same as the iron-nail method. An amount of niter is added at random, sufficient to oxidize but a portion of the sulphides, the Balance being decomposed by the nails. THE CYANIDE METHOD. Sometimes, when no other fluxes are at hand, or when a rapid assay is to be made in which accuracy is not essential, a fusion of ore with cyanide may be made, and the resultant button cupelled for silver and gold. The method is a rapid one and gives good malleable buttons, but is apt to be low in gold and silver, especially in silver. The cyanide used should be pure, free from carbonates or other impurities, and the fusion should be made at a low temperature. The following charge is used: Ore 0.5tol assay ton PbO 25 grams KCN 3 assay tons When the ore contains copper and other base-metal impuri- ties, these are reduced and enter the lead button. Sulphur is taken up by the slag as potassium-sulpho-cyanate (KCNS). In general, it is a method not to be recommended. The following results show the loss in silver which takes place in this method. 1 TABLE XXII. LOSS OF SILVER IN CYANIDE METHOD Niter method Cyanide method Silver, by uncorrected assay Silver in slag Silver from cupel 563 . 73 mgs. 4.10 mgs. 7.81 mgs. 525 . 5 mgs. 36 . 8 mgs. 6 . 56 mgs. E. H. Miller. "Corrected Assays." in Seh. Mines Quart., XIX, November. 1897. ASSAY OF ORES CONTAINING IMPURITIES 121 The results are averages of duplicate assays. The loss of gold in the slag by cyanide fusion is not nearly so marked as that of silver. A COMPARISON OF THE DIFFERENT CRUCIBLE METHODS OF ASSAY FOR IMPURE ORES. In very impure ores, containing large amounts of sulphur, arsenic, etc., the roasting method is applicable when gold only is to be determined, or when silver results need not be very accurate. The roasting method gives uniformly lower silver results than most of the other methods, although to a large extent this is due to roasting at too high a temperature. The roasting method has the advantage that when ores are low grade large quantities of ore can be taken, which is not always possible with the other methods. Roasting, however, must be skillfully conducted in order to be successful. The niter method is a desirable and clean method of assay giving accurate results. Where large quantities of niter are em- ployed, the oxidizing action in the crucible is greatly increased, and it is probable that thereby losses in silver are apt to occur by the slagging of the silver. There is no accumulated evidence on this subject, but many assayers hold this opinion. 1 The niter method is desirable for such ores as do not contain amounts of sulphur requiring extra- ordinary amounts of niter. Usually, the limit of niter in a charge is placed at about 20 grams; if the ore should require more than this, it is generally considered advisable to reduce the quan- tity of ore taken for the assay. This has the disadvantage of multiplying the error of the assay, when finding the value per ton. The modified niter methods discussed offer advantages in the slagging of base-metal impurities. This is particularly true of copper and zinc. It is- very much easier to cause copper to enter the slag when an oxide slag is made than when a silicate is made. This is partly due to the oxidizing nature of the high litharge charges. The best method for the slagging of base-metal impur- ities is the excess-litharge method. The iron-nail method is a standard method, which can be successfully applied to most sulphide ores and, with care, to arsenical ores. It is not applicable to ores containing base- metal impurities, such as copper, for, being essentially reducing in its nature, practically all of the base-metal impurities will be found in the lead button. When used with arsenical ores, 1 E. C. Woodward, Minn, and Sri. Press, CII, 301, gives data which rather tends to show that niter does not have this effect. 122 A MANUAL OF FIRE ASSAYING the temperature employed should be low, not above 1050 C.; otherwise *speiss (an artificial arsenide of iron) is apt to form, which may carry values. It also has the objection, in the case of very impure ores, that small quantities must be taken for assay, involving serious risk of multiplying an error of assay. SCORIFICATION. This is the oxidizing fusion of ore with metallic lead in the muffle-furnace, producing, in the main, a litharge slag, i.e., an oxide slag. It is a method of assay which requires no previous preparation of the ore or preliminary assay, and as practically only one flux is employed, it is both a cheap and a rapid method. It is also a thoroughly reliable method, when proper precautions are taken and when it is employed on ma- terial suitable for the purpose. The operation is performed in shallow fire-clay dishes, called scorifiers. The sizes commonly used are: 1 . 5-in. scorifiers; cubic contents 15 c.c. 2.0-in. scorifiers; cubic contents 25 c.c. 2. 5-in. scorifiers; cubic contents 37 c.c. 3 . 5-in. scorifiers; cubic contents. . . . '. 100 c.c. The dimensions referred to are outside dimensions. The size most commonly employed is the 2.5-in. one. Before these dishes are used it is usual to line the inside with ferric oxide. This is done by preparing crushed iron ore or ochre, mixing with water, and painting the inside of the dishes. This gives them a basic lining, and to some extent prevents the oxide slag from attacking the silica in the clay. Some scorifier slags, especially if they contain copper, are very corrosive. The amount of ore taken for scorification varies from 0.10 assay ton to 0.25 assay ton; but 0.10 assay ton is the amount most frequently taken. The larger amounts are rarely used, unless the ore contains practically no bases. Sometimes, for very impure material, as little as 0.05 assay ton is taken. The amount of test lead varies according to the nature of the ore. The more impure the ore the larger will be the ratio of lead to ore. With 0. 10 assay ton the test lead will vary from 40 to 100 grams. A common charge is 40 to 50 grams of test lead for ordinary ores. As already pointed out, certain quantities of litharge are required in order to make fusible compounds with the metallic oxides. If the ore con- tains small amounts of the metallic oxide, the test lead will be ASSAY OF ORES CONTAINING IMPURITIES 123 small in amount; if it contains appreciable quantities of ferric oxide (Fe 2 3 ) or Cu, etc., large amounts of test lead will be re- quired. It is best to add a small amount of borax glass to the charge, from 1 to 1.5 grams, scattering it over the surface of the lead. This aids in the solution of the bases present. When the ore contains the basic oxides mentioned, borax glass up to 3 and 4 grams will materially aid in forming good slags, without infusible scoria. This infusible scoria often appears in ores containing large amounts of bases, and is very apt to give low results by entangling unfused portions of ore within itself. It is best to mix the weighed-out portion of ore with one-half of the test lead to be used, and then cover over with the balance. The scorification may be divided into the following distinct steps: 1. Melting. In this stage the lead melts, and the ore, being of a lesser gravity, rises to the surface of the molten lead and floats there. 2. Roasting. The ore on the surface of the lead is attacked by the oxygen of the air and roasts in the same way as de- scribed under "Roasting of Ores." 3. Scorification Proper. The lead commences to oxidize, forming litharge. A small percentage (3) volatilizes and the balance forms a fusible slag. This now absorbs the oxides formed by the roasting, dissolving them and forming an igneous solution. The silver and gold, liberated, are absorbed by the remaining metallic lead. The slag, as it forms, drops to the side, forming a slag ring, with the center of the lead bath open to the atmosphere. The reason for this is that the meniscus of molten lead is convex, thus causing the collecting of the slag on the rim of the scorifier. The scorification continues until the whole of the lead is covered over with slag. It is then con- sidered finished and the assay is poured. Should the assay be left in the muffle, the lead will still continue to oxidize, although none is exposed to the air, the interchange of oxygen taking place by means of the litharge and other oxides present. The size of the lead button desired from this assay ranges from 15 to 20 grams. If the scorification is continued to produce smaller buttons, losses are apt to occur by oxidation of the silver, especially if this is present in considerable amounts, thus forming rich slags. 124 A MANUAL OF FIRE ASSAYING The temperature of scorification ranges from 1000 C. to 1100 C., although with pure ores higer temperatures may be employed. When impure ores containing much base metal are scorified, the buttons from the scorification are very apt to be contami- nated with base metal, especially copper, and will then have to be rescorified with more test lead, in order to get a pure button for cupellation. All metals are to some extent oxidized simultaneously, but a mixture of metals may be roughly separated by successive oxidation, each metal in turn partially protecting the metal next in order, while the latter may act as an oxygen carrier to the former. 1 The order of oxidation is as follows: Fe to Fe 2 O 3 Cu to Cu 2 O Zn to ZnO Pt to Pb to PbO Ag to Ag 2 O Ni to Ni 2 O 3 Au to AuO The order of oxidation of the following elements is not so certain: Sb to Sb 2 O s Bi to Bi 2 O 3 As to As 2 O 3 Te to TeO 2 C to CO, S toSO a The order given in the table shows the difficulty encountered in the removal of copper by scorification, as lead stands ahead of it in the order of removal, and it is very difficult and requires a number of re-scorifications, if the amount of copper is large, to reduce it to such an amount as to prevent loss in cupellation. Iron and zinc are very readily removed by scorification (oxida- tion). Certain elements, like Te and Se, are difficult to remove from the lead button, and may tend to concentrate with the Au and Ag in the final cupellation. The slag from the scorification assay should be homogeneous and glassy. If it has an earthy appearance, it is an indication of too low a temperature having been used, and the button is apt to be brittle, due to contained PbO. White patches of sulphate of lead on the slag after pouring also indicate rather too low a temperatuie of scorification, as this sulphate forms at a low temperature under slow oxidation. The scorification method is a reliable one on most materials, with the exceptions enumerated below. As the usual quantity 1 T. K. Rose. "Refining Gold Bullion, etc., with Oxygen Gas," in Trans. I. M. M., April, 1905. ASSAY OF ORES CONTAINING IMPURITIES 125 taken for assay is 0. 1 to 0.2 assay ton, it is evidently not a suitable method for low-grade ores, especially low-grade gold ores, where at least 0.5 to 1.0 assay ton must be taken in order to get accurate results, and avoid the multiplication of the error of weighing. It is practically impossible to get reliable results on $5 to $10 gold ores by ordinary scorification. If, however, 10 assays of 0.1 assay ton are made, the buttons from these combined and re-scorified into one button, which is then cupelled, the results are reliable, but not so good as from the crucible assay on the same total amount, on account of the multiplicity of weighing and other operations, which occasion errors and losses. The method in this instance would also be more costly of time and materials. For ordinary and rich silver ores, and very rich gold ores or furnace products, such as bullions, mattes, etc., the method is a desirable one. It requires no preliminary operations and thus saves valuable time. The slag loss is frequently somewhat higher than in the crucible assay. It is, as ordinarily performed (in duplicate), a cheap method as regards fluxes, etc. It does not give good results on very basic ores, i.e., those containing hema- tite, manganese oxides, etc., as in this case, unless a great deal of lead is used, scoria are apt to form in the slag, which may entangle lead and undecomposed ore. Neither does it give good results on telluride ores, cyanide precipitates, or ores that contain chloride of silver. When basic material is to be scorified, small additions of SiC>2, up to 1 gram, may prove advantageous. In general, however, the addition of fluxes, except test lead, is not to be recommended. Scorification may be modified by the addition of considerable amounts of borax glass, litharge, silica, when it approaches the crucible assay in character with none of its advantages. THE COMBINATION METHOD. The trouble arising from the presence of considerable amounts of base metals, such as copper and zinc, has been fully discussed in previous pages, as well as the difficulty of their removal by fusion methods. For this reason the combination wet- and dry-method has been developed, to remove the objectionable impurities by solution. The method is used chiefly on copper-bearing material, such as heavy copper ores, copper mattes, blister copper, and to a lesser extent on zinc ores, and on cyanide precipitates produced by zinc, and has been advocated for telluride ores. Van Liew's Method for Blister Copper. This is a standard 126 A MANUAL OP FIRE ASSAYING method for copper material. Weigh out duplicate samples of 1 assay ton each of copper borings, add 350 c.c. cold water and 100 c.c. HNO 3 (sp. gr. 1.42)', and set in a cool place for 20 hours, stirring from time to time. Then, if the copper is not dissolved, add from 5 to 30 c.c. more of concentrated acid. At the end of 26 to 28 hours the solution of the copper is complete. Do not apply heat in order to minimize as much as possible the solution of small quantities of gold, by whatever action this may take place. The oxides of nitrogen in the solution are removed by blowing air into it for 20 to 30 minutes. Salt solution (containing 0.54207 grams of NaCl per 1000 c.c.) is added in sufficient quantity to precipitate the Ag present as chloride. 1 c.c. of this solution will precipitate 1 mg. of Ag, and an excess of 4 to 8 c.c. above that required for the Ag should be added. If the amount of Ag in the copper is small, add 10 c.c. of a saturated solution of lead acetate and 2 c.c. of concentrated H 2 SO 4 in order to form PbSO 4 , to aid in settling the silver chloride. Let this stand for about 12 hours and filter the precipitate into the proper sized filter, and wash it well into the point of the filter paper. Dry the filter carefully in the air bath, and when dry, add 8 grams of test lead on top of the precipitate, and carefully transfer to a scorifier containing 2 grams of lead. This is placed in the muffle, heated just to incipient redness, and the filter papers burnt off, but only until the flame disappears, and not into ash. This takes only a minute or so, the precaution being taken to prevent loss of silver by volatilization as AgCl, the lead and carbon present reducing the AgCl to Ag. Then add 3 to 4 grams of PbO, and the same amount of borax glass, raise the heat until well molten, and pour. No scorification is necessary, as no impurities are present. The lead button will weigh 5 to 8 grams and is cupelled with feather litharge. The results should check within 0.2 to 0.3 oz. for Ag and very closely for gold. 1 Sulphuric Acid Method for Blister Copper. 2 To 80 c.c. of cone. H 2 SO 4 add 25 c.c. of a solution of CuSO 4 (160 grams per 1000 c.c.) using a low wide No. 5 beaker. Heat to such a tem- perature that on the addition of the copper borings action com- mences immediately; add 1 a. t. borings, spreading them over the bottom of the beaker. Heat until all dissolving action has ceased, 1 R. W. Van Liew, in Eng. and Min. Jour., LXIX, 498 et seq. 2 F. F. Hunt, "Determination of Gold in Copper Bullion," Eng. and Min. Jour., LXXXVII, 465. ASSAY OF ORES CONTAINING IMPURITIES 127 usually from 1 to 1J hours; then cool and add 400 c.c. of dis- tilled water, stirring to prevent caking of the crystals. Bring to just a boil, filter, and wash the beaker thoroughly, using a rubber- tipped glass rod as a stirrer. Place the filter-paper with the residue in a 2.5-in. scorifier, dry and burn off the paper; add 35 grams test lead and 1 gram silica, scorify to a button of about 9 grams, cupel, and part as usual. Silver may be determined by adding salt solution, as in Van Liew's method, and 10 c.c. of a 10 per cent, solution of lead acetate, stirring well and letting stand over night. Then filter with the usual precaution, and add the paper and precipitate to the same scorifier containing the gold, and proceed as in the case of gold only. In place of cupric sulphate, mercuric nitrate or mercuric sulphate 1 may be used, the equivalent of about 100 mgs. of mercury for an assay ton of borings. The mercury salt is best added to the copper borings, stirring a little and then adding the 80 c.c. sulphuric acid and boiling on a hot plate for three-quarters of an hour. Then proceed as already described. When mercury salt is used in the above quantity on low-grade bullions containing from 10 to 50 oz. Ag per ton all the silver is thrown down with the gold. If more silver is present salt solution should be added in sufficient quantity to precipitate the silver and any mercury that has passed into solution. The object of the addition of cupric sulphate or the mercuric salt is to prevent the formation of copper sulphides, which will remain in the residue and make necessary more than one scorifi- cation to remove the copper before cupellation. The sulphuric-acid method is stated to give results equal to the " all fire " method (p. 139) on gold. Combination Assay for Matte. Van Liew's method of treating in the cold is rarely suitable for mattes, as heat is usually essential in order to insure a decomposition of the matte in a reasonable length of time. Take 2 duplicates of I assay ton each and treat in large beakers, provided with watch-glass covers, with 100 c.c. of distilled water and 50 c.c. HNO 3 (sp. gr. 1.42). After the violent chemical action subsides, add 50 c.c. more of concentrated acid, and warm the beakers on a hot plate until everything soluble is dissolved: usually the residue is white or grayish. 1 F. B. Flinn, Eng. and Min. Jour., LXXXVII, 569; also M in. and Sci. Press, CI, 148. 2 "Assay of Copper and Copper Matte," in Trans. A. I. M. E.. XXV. 258. 128 A MANUAL OF FIRE ASSAYING Next evaporate a considerable part of the acid by boiling, ex- pelling all of the nitrous fumes, dilute to 500 c.c., add 3 c.c. of concentrated H 2 SO 4 , 10 e.g. of saturated lead acetate solution, and enough salt solution of the strength mentioned for blister copper to precipitate the silver; then stir briskly and let them stand over night. Next morning warm the solutions on a steam bath and filter through rather thick filter-paper. Filtrates must be perfectly clear and free from suspended PbSO 4 . Wash beakers and residue thoroughly with hot water, dry the filters in an air bath, and then wrap them up in about 8 grams of sheet lead and scorify with 40 grams of test lead and 1 gram of borax glass. Cupel the buttons with feather litharge. Re-assay the slag from the scorification and the cupel and add the resultant gold and silver to the assay. When heavy copper ores are to be assayed by this method, which are apt to leave large amounts of silicious residue, the general method for mattes is followed, except that the residues after filtering and drying are treated as follows: Take a 20-gram crucible and place in it i assay ton of PbO; then put the filter-paper containing the residue on top of this, place the crucible in the mouth of the muffle at a low heat, burn off the filter-paper until the flame subsides, remove from the muffle, put a cover on the crucible, and allow to cool. When cold add 0.5 assay ton PbO, 15 grams of Na 2 CO 3 , 2 grams of argol, mix well with a spatula, and put on a cover of borax glass. Then proceed as in the ordinary assay. General Precautions to be Observed in the Combination Assay. The combination methods on copper material agree well with the standard scorification methods for the same material when cor- rection of cupel loss is made for the latter method. The scori- fication methods will often seem to give higher results, but this is in most cases due to the fact that 'the silver beads frequently contain from 2.5 to 4 per cent, copper. The combination method gives in most cases (Van Liew's method possibly excepted) uniformly lower results in gold (4 per cent.) than the standard corrected scorification method. This is generally ascribed to the formation of nitrous acid (HNO 2 ) during solution, which, in connection with nitric acid, is said to have a solvent action on gold; but such authorities as W. F. Hillebrand 1 dispute this. 1 W. F. Hillebrand and E. T. Allen, "Comparison of a Wet and Crucible-Fire Methods for Gold Telluride Ores," Bull. 253, U. S. G. Survey. ASSAY OF ORES CONTAINING IMPURITIES 129 The solution may be due to the formation of H 2 SO 4 during solu- tion, as the mixture of this acid and HNO 3 has a solvent action, or to the presence of impurities like chlorides or HC1, etc., or possibly to the presence of nitrates, particularly those of iron or copper. It has been demonstrated that gold is soluble in hydrochloric acid solutions of iron alum, and of cupric chloride, but not in pure HC1. 1 The fact that the combination method on copper-bearing material gives low results on gold is, however, well established. Owing to the number of manipulations in the combination assay, it is often apt to give low results in the hands of inexpe- rienced chemists, mainly due to the mechanical losses in handling. The directions given should be carefully followed, especially those regarding amount of solution, strength of acid, temperature, time, etc. Neatness is indispensable. The HN0 3 must be pure. The directions regarding the burning off of the filter-paper must be closely followed. The amount and strength of the salt solu- tion must be carefully adhered to and it must be added at the proper time. Some assayers, instead of adding salt solution at the same time as H 2 SO 4 and Pb(C 2 H 3 O 2 ) 2 , filter off the residue containing the gold and make a separate precipitation for the silver, believing that the addition of a salt solution may cause a slight redissolving of the gold. At this point of the assay that is, however, hardly probable. A large amount of NaCl is to be avoided, as AgCl is very appreciably soluble in brine. C. White- head recommends NaBr or KBr instead of NaCl for this reason. Combination Method for Precipitates from the Cyanide Process. 2 Where the troublesome base-metal impurity is zinc instead of copper, as in this case, sulphuric acid can be substituted with advantage for HNO 3 . The method is as follows: Of the precipitates 0.10 assay ton is taken, placed in a beaker, and 20 c.c. of sulphuric acid (concentrated) and 60 c.c. of water are added. This is heated on a hot plate for about one hour, or until zinc and zinc oxide are in complete solution. Add salt solution of the strength already mentioned in the paragraph on Van Liew's method for blister copper, in slight excess, to pre- cipitate the silver present, remembering that 1 c.c. will precipitate 1 W. J. McCaughey, Jour. Am. Chem. Soc., XXXI, 1261. 2 Fulton and Crawford, "Notes on Assay of Zinc Precipitates Obtained in the Cyanide Process," in School of Mines Quart., XXII, 153. 130 A MANUAL OF FIRE ASSAYING 1 mg. of silver. Stir briskly with glass rod to agglomerate the silver-chloride formed. The residues are then filtered through the proper sized filter, carefully washed with hot water into the point of the filter-paper, and dried in the air bath at a low heat. After drying, transfer to a 20-gram crucible containing 1 assay ton of litharge, and burn the filter-paper off in the manner already described. Then add 15 grams of soda and 2 grams of argol, mix thoroughly, and cover with a heavy cover of borax glass. Fuse and cupel the resultant lead button. Weigh the gold and silver bead, and from a preliminary assay determine the proper amount of silver necessary in order to inquart the bead. The amount of silver should be just about 2.5 times the amount of gold. Roll out the bead, after flattening with a hammer, until, after repeated rollings, the fillet will have about the thickness of a visiting card. It is best to anneal the bead at a red heat between the various rollings, in order to prevent cracking on the edges. Then part in a parting flask in hot nitric acid having a specific gravity of 1.26. Boil twice for at least 20 minutes each time, in order to insure the complete remova of the silver. This method of parting leaves the gold in one coherent mass, termed a " cornet," and is identical with the method practised in the gold bullion assay. CHAPTER X SPECIAL METHODS OF ASSAY TELLURIDE ORES. Gold ores containing the precious metals in the form of tellurides of gold and silver, mainly calaverite and sylvanite, are more difficult of assay than ordinary gold ores, and special methods are essential in order to get good results. The scorification assay is not reliable for telluride ores, giving almost uniformly low results. It is not used by assayers and chemists of the great telluride ore district in Colorado Cripple Creek. It seems that in scorification the main cause of loss is volatilization, for while the slag loss is higher than for ordinary ores, slag and cupel corrections still leave the results from this assay far below those of the crucible assay when properly performed. Of recent years selenium gold ores have been found 1 and in general the precautions necessary for the assay of telluride ores apply also to selenium gold and silver ores. Tellurium has a great affinity for gold and silver that for silver being greater than that for gold; and if a high-grade telluride ore be assayed, even by special method, the beads from the cupellation will frequently still contain tellurium. 2 In the crucible assay the losses, which are somewhat greater than in ordinary ores, occur in the slag, and from the presence of the Te in the lead button, causing absorption of precious metals by the cupel. The aim in the crucible assay is to remove the tellurium from the gold and silver and slag it. This is best accomplished by the presence of considerable litharge as an oxi- dizing agent, and otherwise properly balancing the flux. The flux recommended quite generally by Cripple Creek assayers is made up as follows: Potassium carbonate 7 parts Flour 1.0 parts Sodium carbonate 6 parts Litharge 30 . parts Borax glass 5.5 parts 1 "Selenium Gold Ore," Eng. and Min. Jour., XC, 418; Min. and Sci. Press, C, 224. 2 E. C. Woodward, "Cupel Losses in Telluride Ores," in West. Chem. and Met., I, 120. 131 132 A MANUAL OF FIRE ASSAYING This is for the ordinary silicious Cripple Creek ores. About 75 grams of this flux is used with 0.5 assay ton of ore. This gives the following charge: Ore 0.5 assay ton Borax glass 8.5 grams PbO 45. 5 grams Na 2 CO 3 9.0 grams Flour 1.5 grams K 2 CO 3 10 . 5 grams The heat recommended is such that a temperature of 1063 C., the melting-point of gold, is reached at the mouth of the muffle. Some assayers recommend a soniewhat greater temperature to insure the decomposition of the tellurides. The time of fusion should be about 45 to 50 minutes. In most telluride ores the silver contents are not great enough to permit of the parting of the bead obtained from cupellation. It is therefore necessary to add silver at some stage before part- ing and in this instance it is best done during the crucible assay, since by doing this there is apt to be less absorption of gold dur- ing cupellation on account of the presence of silver in the lead button. It is essential to recognize that the flux recommended above for tellurides does not make what can be strictly termed an "excess-litharge charge." Hillebrand and Allen 1 recommend the following charge for Cripple Creek ores: Ore 1 assay ton Borax glass 10 grams NaHCO 3 1 assay ton Reducing agent (if necessary) PbO 6 assay tons Salt cover This approaches more nearly the excess-litharge charge. The salt as a cover may with advantage be replaced by litharge. The fusion should be conducted slowly and at a temperature not exceeding 950 to 1000 C. It is essential in telluride ores to have the sample crushed to 120- or, better, to 150-mesh. The reason for this is that, owing to the irregular distribution of values in these ores, fine crushing is required to get a true sample, and also because the low melting- point of the charge usually employed makes this essential. The precise behavior of tellurium in the crucible assay, and dur- ing scorification or cupellation, has not as yet been investigated 1 "A Comparison of a Wet and Crucible- Fire Methods for the Assay of Gold Telluride Ores," Bull. No. 253, U. S. G. Survey. SPECIAL METHODS OF ASSAY 133 with scientific thoroughness though some preliminary work has been done. 1 The following facts are reasonably well established: 1. The great affinity of tellurium for gold and silver, resulting in tellurium passing to the lead button with the precious metals, unless a charge be used that is essentially oxidizing in its char- acter, and effecting the slagging of the tellurium. 2. During the oxidation of the lead button by cupellation or scorification the tellurium tends to concentrate in the remaining lead-gold- silver alloy although in different degree in the two operations, the concentration being more pronounced in scorification. 3. The effect of tellurium on the lead-gold-silver alloy is to very greatly decrease its surface tension, so much in fact that if proportion- ately sufficient tellurium be present the surface tension is changed sufficiently to cause the alloy to "wet" the cupel and be absorbed as alloy, thus causing heavy losses of precious metals. If the proportion of lead to tellurium to gold should attain the concen- tration of 10:1:1, complete absorption may take place leaving no gold bead. Such a result will, however, occur only in excep- tional cases, as in the direct cupellation of telluride mineral, etc. The accompanying data due to S. W. Smith, shows the relative elimination of tellurium during cupellation and scorification. Lead buttons of 20 grams containing 0.05 gram each of gold and tellurium were submitted to cupellation and scorification, and the process interrupted at intervals for the determination of tellurium. TABLE XXIII. SHOWING ELIMINATION OF TELLURIUM DUR- ING CUPELLATION AND SCORIFICATION Cupellation Scorification Cupelled Per cent, of Per cent, of I Scorified | Per cent, of down to wt. original lead tellurium in ( I down to wt. original lead J tellurium in laining , the button remaining 20.00 100.00 2.48 20.00 100.00 2.48 11.86 59.00 3.25 6.145 30.7 5.94 4.25 21.00 2.49 2.085 10.4 12.90 1.635 8.15 1.80 1 S. W. Smith, Trans. I. M. M.. Buls. 44, 45 and 47 (1908). Holloway and Pearse, Trans. I. M. M., Buls. 39. 40, and 45 (1907 and 1908). 134 A MANUAL OF FIRE ASSAYING It will be noted that during cupellation the tellurium at first concentrates in the lead and then begins to be eliminated, the percentage decreasing. In the case of scorification there is a very decided concentration of tellurium in the lead toward the end of the operation. The difference is probably due in large part to the fact that in cupellation the tellurium is in part absorbed by the cupel as a lead tellurium alloy, which action cannot take place in scorification. The removal of tellurium by oxidation from lead, thus passing into cupel or slag, is evidently a difficult process. In an alloy of lead, gold and silver and tellurium, the tellurium can form compounds with both the precious metals and lead. It will be distributed between the two according to the relative masses present and the relative chemical affinities. If then in the alloy relatively much lead be present (100 to 200 parts Pb per part Au and Te) by far the larger part of the tellurium will be eliminated by absorption as lead telluride, and only a little will stay with the precious metals. This small amount will not be sufficient to materially lessen the surface tension of the bead at the end of the cupellation and hence absorption will be small. If, however, the amount of lead be small so that the relative amount of tellurium and gold be increased the absorption of the latter may be very heavy. 1 It follows therefore that in lead buttons obtained in the crucible assay, which may contain tellurium, it is better to cupel directly in order to avoid heavy absorption in the cupel. Scorification might be resorted to in this case for two reasons: 1. To reduce a large lead button; 2. in the mistaken idea of eliminating tellurium. From what has gone before it is evident that a large button is not disadvantageous as it really tends to decrease the absorp- tion of precious metal when tellurium is present. The preliminary scorification of lead buttons from crucible assays of telluride ores has been shown to give low results. 2 The cause for the almost universally low results on telluride ores by the scorification method is also to be attributed in part to the above reasons. 4. Silver seems to exert a protective action on the gold and lessen the absorption of the latter, due probably to the greater affinity of silver for tellurium, thus forming silver telluride to the exclusion of the formation of much gold telluride, consequently i T. K. Rose, Trans. I. M. M., Bui. 40 (1908). 3 C. H. Fulton. School of Mines Quart., XIX, 419, and S. W. Smith, ibid. SPECIAL METHODS OF ASSAY 135 lessening the absorption of gold. It is therefore desirable to perform the assay in the presence of considerable silver, which will have to be added anyway at a later stage to give a bead that will part. In the assay of telluride ores the general object therefore will be to remove as much tellurium from the gold and silver before cupellation as is possible. This is best done by the performance of a crucible assay with an oxidizing charge. The oxidizing properties of the charge are obtained by the use of an excess of litharge. If we consider the ordinary telluride ore as composed of a silicious or shaly gangue containing the precious metals as tellurides and containing also certain amounts of sulphides, then when this is subjected to fusion with litharge (a large excess) the telluride minerals and sulphides are oxidized, the tellurium probably forming tellurate of lead or, in the presence of soda, tellurate of soda. (See behavior of sulphur, Chapter V.) If, however, an insufficient amount of PbO is present so that it forms lead silicates only with the silica of the ore, the oxidizing effect will be much diminished, since lead silicates form at a low temperature and do not readily give up oxygen. It is therefore desirable to form a slag which has the characteristics of an excess litharge charge, viz., is not glassy, but of an earthy dull appearance. Considerable soda should be present to aid the oxidation of the impurities. The borax glass should not exceed 5 to 10 grams, and the lead button made should be large, 25 to 30 grams. The fusion should be made slowly, particularly at first, and the temperature not exceed about 1000 C., since there is a possibility of dissociating the tellurium compounds in the slag and sending the tellurium into the lead button. It is probably impossible to remove all the tellurium from gold and silver by such an oxidizing fusion for the reason that the re- duction of lead from some of the litharge at a certain stage of the assay for the collection of the gold and silver, also again reduces some of the tellurium which has been oxidized. It is desirable to obtain the full oxidizing effect of the litharge before the reduction of lead takes place and for this reason charcoal is to be recom- mended as the reducing agent, when this needs to be employed, instead of argol or flour, since the CO evolved by the two latter begins to reduce Pb from PbO at about 300 C. less than solid carbon, which acts at about 550 C. (page 64). 136 A MANUAL OF FIRE ASSAYING For ordinary silicious telluride ores of only slight reducing power the following charge is recommended: Ore, PbO, Na 2 CO 3 , Borax glass, Charcoal, Silver foil, PbO cover, 0.5 100.0 30.0 6.0 1.1 a. t. gr. g r - gr- gr. 10 to 20 mgs. 10 gr. The fusion should be conducted slowly at first, the final tem- perature not much exceeding 1000 C. If the button from the fusion is thought to contain tellurium, as is probably the case in the assay of a high-grade ore, it will be desirable to place it in a 20-gram crucible, cover with 30 grams PbO, mixed with 2 grams borax glass and bring to fusion, then pour and proceed as usual. This treatment will eliminate con- siderable tellurium from the lead. (S. W. Smith.) It is stated 1 that in the oxidizing roasting of Cripple Creek telluride ores, in their preparation for chlorination or cyanida- tion, the greater part of the tellurium in the raw ore is found in the roasted ore as a tellurite of iron. Some assayers add an iron nail to the assay, not so much to desulphurize as to provide an excess of iron for the purpose of combining the tellurium with it, as in the case of sulphur. For the quantity of tellurium present, its influence on the assay is certainly profound. The following table gives an idea of the quantity present: TABLE XXIV. QUANTITY OF TELLURIUM IN ORES Element Cripple Creek Ore Cripple Creek Ore Black Hills Cambrian Black Hills Cambrian Tellurium Gold Silver 0.0742 per cent.. 0.0506 per cent.. 0.0075 per cent.. 0.092 percent.. 0.060 percent.. 0.0103 per cent. 0.0033 per cent.. 0.0026 per cent.. 0.010 percent. 0.003 percent. As already stated, tellurium is with difficulty separated from gold and silver, and in spite of an oxidizing charge is frequently carried down in the lead button. The loss then takes place in the cupel, tellurium causing a heavy absorption. Some loss, however, takes place by volatilization. There is also a somewhat 1 Trans. I. M. M., Ill, 49, 50. SPECIAL METHODS OF ASSAY 137 higher slag loss in the telluride assay than in the assay of ordinary ores. 1 Hillebrand and Allen, already quoted, assayed telluride ores by the combination wet-and-dry assay, getting the gold and silver free from tellurium, but found that the crucible assay as ordinarily performed for telluride ores gave just as satisfactory, if not better, results. A STUDY OF THE ASSAY OF BLACK HILLS CAMBRIAN ORES. These ores are probably complex tellurides. The ores were oxidized and of the following average composition: Si0 2 = 71.5 per cent.; Fe 2 O 3 = 16.3 per cent.; A1 2 O 3 =4.8 per cent.; CaO = l.o per cent.; Gold = 0.79 oz.; Ag = 0.10 oz. Samples of this type of ore, representing controls on car-load lots, were assayed by assayers A and B in the same laboratory, with the same kind of cupels, and great regard to temperature of cupellation. Assayer A made fusions on one assay ton lots, in triplicate, with the following stock flux: Na 2 eO 3 3. 25 parts Borax glass ... 5. 00 parts K 2 CO 3 2. 25 parts Argol 1.00 parts PbO 18. 00 parts 29 . 50 parts The amount of flux used was 4 assay tons per assay ton of ore, with quite a heavy borax glass cover. Fusions made at 1100 C., approximately. The stock flux is equivalent to the following charge: Ore 1 assay ton PbO 73 . 2 grams Na 2 CO 3 13. 2 grams Borax glass 20. 3 grams K 2 COj 9.1 grams Argol 2 4.0 grams On account of the negligible quantity of Ag present, every assay was salted with Ag. The beads were parted in acid 1 to 9, and were in each case required to check against each other in weight. The beads were then weighed together and the resultant weight divided by 3 to obtain the amount of gold. Assayer B made assays on the same pulp samples with the following stock flux: Na 2 CO 3 3.25 parts Borax glass 2.00 parts K 2 CO 3 2. 25 parts Argol 0.75 to 1 .00 part PbO 22. 00 parts 30. 25 parts 1 C. H. Fulton, School of Mines Quart., XIX. F. C. Smith Trans. I. M. M..IX. 344- Min. Rep., LI, 163. Hillebrand and Allen, Bull. No. 253, U. S. G. Survey, 12, 14. 2 This amount of argol required because ores are oxidizing. The button produced usually 22 to 25 grams. 138 A MANUAL OF FIRE ASSAYING Three assay tons of flux were used to each 0.5 assay ton of ore, with a soda cover one-quarter inch thick. Assays were made in quadruple, all fusions being salted with Ag, parted in 1 to 4 acid, and the beads required to check against each other in weight and then weighed together, and the sum divided by 2 to get the value per ton. The stock flux is equivalent to the following charge: , Ore 0.5 assay ton PbO 67 grams Na 2 CO 3 9 grams Borax glass 6 grama K 2 CO 3 6 grams Argol 2.5 grams The results of these series of assays were as follows: LOT No. ASSAYER A ASSAYER B Oz. Au per ton Oz. Au per ton 88870 0.74 0.78 88832 0.80 0.83 88874 0.71 0.80 88823 0.82 0.98 88721 0.80 0.88 88851 0.85 0.89 88818 0.85 0.91 88940 0.79 0.84 3669 0.82 0.91 88853 0.85 0.91 88890 0.81 0.83 71957 0.79 0.82 88826 0.77 0.82 3843 0.77 0.81 88780 0.80 0.88 22522 0.66 0.73 98509 0.69 0.79 22050 0.50 0.58 Assayers A and B then exchanged fluxes, and as they checked each other's previous results closely, it became evident that the flux of assayer A was ill-balanced and would not give good results. Slag and cupel corrections were made by Assayer A on assays made with his flux, but even these corrections added failed to bring his results up to those of assayer B. The question arises as to what is the specific trouble with flux A. On examination, it will be found to contain an excessive amount of borax glass, especially when the cover is considered. It is very probable that the acidity of the charge (although a good fluid slag is obtained) is so great, taking into account both the silica of the ore and the borax glass, that the ore is not com- SPECIAL METHODS OF ASSAY 139 pletely decomposed by the basic ingredients of the charge; i.e., the soda and litharge become saturated with borax and then do not completely decompose the silicious ore. The fact that re- assays of the slag do not bring the results up to the figures ob- tained by assayer B does not necessarily imply that the slag does not contain these values, as the charge used to flux the slags and cupels again contains much borax glass, so that practically the same conditions obtained as before. THE ASSAY OF COPPER-BEARING MATERIAL. Copper-bear- ing material includes ores containing copper and furnace products, chiefly mattes, blister copper, etc. Copper, which in the assay has a strong tendency to go into the lead button, causes, when present in sufficient quantity, serious losses by cupel absorption. Therefore all methods of assays for this class of material endeavor to eliminate copper from the lead button to be cupelled. A standard method for the assay of material high in copper, espe- cially for Ag, is the combination assay for blister copper and mattes, described in Chapter IX. Another standard method, especially for gold, and one that is carried out frequently as a check to the above, is the scorifica- tion or "all fire" method. This is performed as follows: Ten samples, of 0.10 assay ton each, are taken and placed in 3-inch scorifiers with 50 grams of test lead (the silver content of which is accurately known) ; 25 grams of the lead are mixed with the matte, or borings, etc., and the other 25 grams used as a cover. On top of the charge is placed 1 gram each of silica and borax glass. The scorification is carried on at a moderate tem- perature until the assays are just about to slag over, which takes usually about 25 minutes, and then they are poured. The result- ant button will weigh about 15 to 16 grams and be quite hard with copper. The buttons, cleaned from slag, are scorified, test lead being added to make the total lead up to 40 grams. The second scorification will take about 30 minutes and the resultant buttons will weigh from 10 to 12 grams. These are cupelled in 10 separate cupels, placed so as to be subject to uniform tem- perature, i.e., in one horizontal row across the muffle. Cupella- tion should be conducted at as low a temperature as is feasible. The beads are weighed separately and then together. They are then grouped in two lots of 5 each, which are parted in acid, strength 1 to 9, the beads being kept in this acid at nearly boiling temperature for 20 minutes and finished for 5 minutes with 140 A MANUAL OF FIRE ASSAYING 1.42 sp. gr. acid (full strength). The ten cupels are taken in lots of two each (only the litharge-stained part is taken), crushed to pass 100-mesh and assayed by the following charge: 100 grams PbO 45 grams borax glass 20 grams Na 2 CO 3 3 grams argol Soda cover The lead buttons are cupelled, and the silver and gold obtained added to the first weights. The scorification slags may also be reassayed and this correction added, but in practice the cupel correction is the only one usually allowed. Sometimes no correc- tion is allowed. It is to be noted that, even with a rescorifica- tion of the first button of the assay, the final silver beads, from 55 per cent. Cu matte containing 180 oz. Ag per ton and 2.31 oz. gold, will contain from 2.5 to 4 per cent. _ copper, which must be deducted in order to get correct silver results. (For a further discussion of scorification slag losses and cupel absorption in assaying copper-bearing material, see Chapter XI.) The scorification method is generally employed for the deter- mination of gold in mattes, and the combination method for the determination of silver. Of recent years, special crucible methods for copper mattes and copper-bearing material have been devel- oped with considerable success. 1 A satisfactory method on copper mattes, up to 20 per cent, copper and high in gold and silver, was practised by the Standard Smelting Company, at Rapid City, S. Dak. The matte sample is put through a 120-mesh screen, and for controls 4 assays of 0.25 assay ton each are made, with the following stock flux: Silica 11 parts Sodium carbonate 25 parts Litharge 70 parts Niter 5 parts An 0.25 assay ton matte is run with a 3.5 assay ton flux and a thin borax glass cover. The flux figured to the charge is as follows: 0.25 assay ton matte 24.0 grams Na 2 CO 3 10.5 grams SiO 8 5.0 grams KNO 3 67.0 grams PbO The heat used is high and the fusion short, giving a clean fluid slag and a bright button of approximately 20 grams. These buttons are cupelled directly for gold and silver. One cupel and "An All-fire Method for the Assay of Gold and Silver in Blister Copper," in Trans. A. I. M. E., XXXIII 670. Perkins. "The Litharge Process for the Assay of Copper- bearing Ores," ibid., XXXI, 913. SPECIAL METHODS OF ASSAY 141 one slag are then re-run in the same crucible that the original fusion was made in, and the result of the four corrections added to the sum of the original buttons. No scorification is made before cupellation. The average correction, on the usual grade of matte (5 oz. Au, 40 oz. Ag), is 2.5 per cent, gold and 5.5 per cent, silver. Below is a comparison of this method with the standard scorification assay, including cupel and slag correc- tion. The copper content of this matte was 19.98 per cent. TABLE XXV. COMPARISON OF METHODS IN ASSAY * Crucible method Scorification method Gold (oz.) Silver (oz.) Gold (oz.) Silver (oz.) 4.10 0.10 36.24 1.00 3.90 0.25 35.07 1.95 Correction 4.20 37.24 4.15 37.02 The returns on this pulp by the refiner were: gold, 4.19 oz.; silver, 36.71 oz. Matte No. 1545; copper content, 17.6 per cent. TABLE XXVI. COMPARISON OF METHODS IN ASSAY Crucible method Scorification method Gold (oz.) Silver (oz.) Gold (oz.) Silver (oz.) Original assay Correction 3.42 0.10 31.94 1.85 3.40 0.11 31.86 1.93 3.52 33.79 3.51 33.79 The following table shows results by this method with correc- tion and refiners' results (by same method without correction) : 142 A MANUAL OF FIRE ASSAYING TABLE XXVII. CORRECTED AND UNCORRECTED ASSAYS ON COPPER MATTE Lot No. Crucible method: Standard Smelting Company Crucible method: Refiner Gold Silver Gold Silver Copper oz. per ton oz. per ton oz. per ton oz. per ton % 17.73 75.4 17.67 74.13 7.2 1404 17.75 73.5 17.625 72.42 9.5 1412 11.35 45.7 11.145 42.95 10.3 1435 10.02 39.9 10.065 39.08 12.9 1450 9.10 44.6 9.02 43.01 13.2 1457 6.89 48.75 6.815 46.28 15.02 1458 9.95 58.37 9.935 52.31 19.9 1470 8.235 50.86 8.24 52.31 20.5 1471 4.845 40.52 4.87 39.48 18.4 1484 7.34 45.04 7.265 44.68 18.27 1489 6.45 47.98 6.83 46.67 19.4 1500 5.78 45.34 5.84 45.06 20.5 1503 4.61 27.11 4.58 26.51 18.8 1513 3.815 32.75 3.80 31.55 11.8 1525 3.12 39.62 3.205 33.84 12.4 1529 3.10 33.00 3.36 31.14 18.7 1533 4.20 37.24 4.19 36.71 20.0 .4.10 oz. per ton Silica . . . . 3.3 per cent. 31.55 oz. per ton Lime . . . . 0.5 per cent. 17.4 per cent. Sulphur. . . . . . . . 29.1 per cent. 45 . 9 per cent. Lead . . . . trace 2 . 5 per cent. A typical sample of matte on which these assays were made analyzes as follows: Gold Silver Copper... Iron . . The crucible charge employed can readily be modified to apply to mattes higher in copper or greater in reducing power. Perkins' excess-litharge method has already been described. He states that for low-grade copper-bearing material (2 to 4 per cent.), 5 assay tons of PbO to 0.5 assay ton of ore will remove most of the copper, if the balance of the fluxes is properly proportioned, i.e., if there is ample free PbO to dissolve copper oxides. For high-grade mattes, etc. 48 to 60 per cent, copper SPECIAL METHODS OF ASSAY 143 8 assay tons of PbO to 0. 1 assay ton of matte will remove most of the copper. Perkins also developed a crucible method for metallic copper, as follows: Weigh out 0.25 assay ton of copper borings, divide it into 3 approximately equal parts, and place in 20-gram crucibles. In this way weigh out 4 sets, getting 12 assays. Into each crucible put 800 mgs. of powdered sulphur, mix thoroughly with the copper, and then on top of this put the following flux, being careful not to mix the flux with the copper: 0.25 assay ton PbO ............ 8.0 assay ton K 2 CO 3 ............. 0.25 assay ton SiO 2 ............ 0.5 assay ton Salt cover Place the crucibles into a dark-red muffle and gradually raise the temperature for 45 minutes to a yellow heat. The tempera- ture regulation is important, and it is necessary to produce a neutral or reducing atmosphere in the muffle by the presence of coal or coke. The buttons, weighing about 18 grams each, are put together in lots of three, representing 0.25 assay ton, and scorified at a low heat. The resultant buttons should weigh 5 to 6 grams. Each of these buttons is now rescorified with 25 grams of lead at a low heat, until 6-gram buttons are obtained. These are cupelled with feathers. This method is stated to give results on gold equal to the all-scorification method, and on silver equal to the combination method. THE ASSAY OF ZINCIFEROUS ORES AND METALLURGIC PRODUCTS CONTAINING ZINC. Zinc most frequently occurs in ores as the sulphide, sphalerite, and, in certain metallurgical products, as the metal (zinc cyanide precipitates) . Zinc boils at 940 C., and rapidly volatilizes. Zinc oxide volatilizes slowly at 1180, and rapidly at 1400. Zinc silicates alone are difficultly fusible, but are readily so when mixed with borax or boricacid or ferrous silicate. 1 The presence of zinc in material to be assayed calls for certain precautions, and in general the assay is difficult. Metallic zinc has a great affinity for gold and silver, greater than lead, as is shown by the Parkes process for the desilverization of lead bullion. Under oxidizing influences 2 the formation of zinc oxide and its volatilization causes losses of gold and silver. That this loss is mechanical does not make it less serious. The boil- 1 Rose. "Refilling Gold Bullion, etc., in Oxygen Gas," in Trans. I. M. M., April, 1905. 2 Williams, in Jour. Chem. Met. and Min. Soc. of South Africa, III, 132. 144 A MANUAL OF FIRE ASSAYING ing-point of zinc occurs at a temperature somewhat below the normal for ordinary scorification, and it is this fact, coupled with the fact that the zinc oxide formed is with difficulty soluble in litharge, that make accurate assay-results hard to obtain, especially in scorification. Zinc containing gold and silver may be distilled off and volatilized with very little loss of gold and silver, if the conditions are reducing. 1 Scorification is frequently employed for zinciferous ores, although it is not generally satisfactory. When used, it is best carried out in a way similar to that adopted for copper-bearing material, using from 0.05 to 0.10 assay ton of ore with from 50 to 80 grams of test lead, 2 grams of borax glass, and 1 gram of silica, the last being essential to flux the zinc oxide formed. Otherwise insoluble scoria and crusts form on the scorifier. Slag and cupel corrections are generally necessary and from 5 to 10 assays are made, the results being averaged. As zinc is readily oxidized, lead buttons contaminated with zinc are not to be feared and rescorification is rarely necessary. Among the most im- portant zinciferous material presented for assay are the zinc-gold precipitates from the cyanide process. Scorification is not de- sirable for these. 2 They are best assayed by the crucible method or by one of the combination methods already described. Crucible Method. The crucible method best suited for unox- idized zinc ores is the niter method, with sufficient silica present to form at least the monosilicate with zinc. Borax glass and much litharge is also desirable. On a practically pure sphalerite the following charge will give good fusions at temperatures of about 1100 C.: Ore 0.5 assay ton SiO 2 8 grams Na 2 CO 3 15 grams KNO, 22 grams PbO 150 grams Heavy borax glass cover. 3 This charge can be modified, as regards niter and silica, to suit any sphalerite ore. A good crucible charge for cyanide precipitates, containing up to 50 per cent, zinc, is: Precipitates 0.1 assay ton SiO 2 .' 5 grams Na 2 CO 8 5 grams Na 2 B 4 O T 2 grams PbO 70 grams Flour 1 gram Light borax glass cover 1 Rose, ibid., and references. 2 "Notes on the Assay of Zinc Precipitates, etc.," in School of Mines Quart., XXII, 153. 3 A similar charge is recommended by Lay, for complex zinc-lead concentrate; see Min. Ind., XIII. 287. SPECIAL METHODS OF ASSAY 145 The following method 1 is used on cyanide precipitates containing 12,000 to 22,000 oz. Ag, and 300 oz. Au per ton at the mill of the N. Y. and Honduras Rosaria Min. Co. in Honduras, C. A. In a 20 gram crucible mix 27 grams test lead, 2.5 grams borax glass, 0.5 gram silica with 0.1 a. t. of the precipitates, tapping the crucible to make certain that no material adheres to the sides. In another crucible mix 33 grams of PbO, 25 grams of Na 2 C0 3 , 4.5 grams borax glass, 1.5 grams of silica, and 0.15 grams charcoal. After mixing transfer this second charge on top of the contents of the first crucible, and make the fusion as usual. The button separates very cleanly from the slag. The slag and cupel are reassayed, and one weighing made on the three beads recovered, from the cupellation. The slag corrections are small, amounting to about 25 oz. Ag per ton of precipitates containing 14,000 oz. The gold loss in the slag is very small. The assays are run in triplicate. In the assay of such high- grade material weighing of precipitates must be done on anal- ytical balances. Assay of Plumbago Crucibles for Gold and Silver. Graphite or plumbago crucibles are extensively used in the smelting of cyanide-zinc precipitates, and the old discarded ones are usually sold in lots to some smelter; they often contain considerable gold and silver. These pots present difficulty in assaying, chiefly on account of the graphite and zinc contained. From a given weight of sample, the metallics and scales are separated by passing the material through a 150-mesh screen, and a regular scale assay is made as outlined at the end of this chapter. The pulp is assayed as follows: 2 From 0.05 to 0.10 assay ton is taken and mixed with a little more than one-half its weight of niter and 30 grams of litharge, placed in a 2.5 in. scorifier, covered with 30 grams of litharge and afterward with a thin cover of borax glass, placed in a muffle, and fused finally at a yellow heat. The buttons are cupelled, weighed, and parted as usual. Crucible assays may also be made on this material by the niter excess-litharge fusion, with a charge as follows: 0.1 assay ton graphite 5 grams Na 2 CO 3 70 grams PbO 5 o 11 grams KNO 3 (according to 5 grams SiO 2 carbon contents of pulp) Borax glass cover 1 Private communication, E. Van L. Smith Name of the originator of the method not known to the author. 2 A modification of T. L. Carter's method; see Eng. and Min. Jour. LXVIII, 155. 10 146 A MANUAL OF FIRE ASSAYING In both methods it is essential that the amount of pulp, usually, should not exceed 0.1 assay ton, the carbon giving difficulties with greater amounts than this. The Assay of Residues from Zinc Distillation (containing con- siderable carbon) for Silver and Gold. 1 From 0.10 to 0.5 assay ton of the powdered residue is mixed with 35 grams of niter and 10 grams of Na 2 O 2 (sodium peroxide), and dropped, in lots of 5 grams each, into a red-hot crucible which can be readily covered, and the oxidation reactions permitted to complete themselves. The flux then added consists of 70 grams of litharge, 10 grams of borax glass, 10 grams silica, 2 grams argol and a light borax glass cover. The fusion is carried out at a yellow heat and the buttons cupelled as usual. THE ASSAY OF ANTIMONIAL AND ARSENICAL ORES FOR GOLD AND SILVER. Gold- and silver-bearing antimonial ores, such as stibnite, jamesonite, etc., are usually assayed by the niter method, 2 in the presence of considerable soda and niter, to induce the formation of the antimoniate of soda. A preliminary assay to determine the amount of niter is essential. The follow- ing charge is recommended for nearly pure stibnite: 3 Ore 0.5 assay ton KNO 3 18 grams PbO 120 grams Borax glass 6 grams Na 2 CO 3 10 grams SiO 2 10 grams Salt cover The fusion should be conducted slowly and at a low tempera- ture. The button will usually contain very little antimony, the cupel not showing scoria or cracks. If it does contain enough to cause losses in cupellation, the buttons should be scorified. Smith 4 gives the following charge for ore containing ap- proximately 75 per cent, stibnite. The niter, etc., can be varied for the ore as the gangue increases: Ore 1 assay ton Borax glass 8 grams PbO 75 grams KNO 3 20 to 25 grams Na 2 CO 3 25 grams Salt cover. Another method, practically as good as the niter method, is the roasting with charcoal or coke-dust. 5 The sample of ore, 1 K. Sander, in Eng. and Min. Jour., LXXIII, 380. 2 William Kitto. "The Assay of Antimonial Gold Ores," in Trans. I. M. M., 1906, Nov. 8 and Dec. 13. 8 Smith, "The Assay of Complex Gold Ores." in Trans. I. M. M., IX, 332. 4 Smith, ibid. 6 Sulman, Trans. I. M. M., IX, 340. SPECIAL METHODS OF ASSAY 147 usually i assay ton, is mixed with approximately its own volume of coke-dust or coal-dust, placed in a 5-in. roasting dish, covered with another dish, and roasted in a muffle with closed door, at a temperature not exceeding a dark cherry-red (635 C.), for about 35 to 40 minutes. This will cause the volatilization of 95 to 96 per cent, of the antimony as sulphide without appreciable loss of gold. The roast should have a yellow appearance when finished, and can be fused with the following charge: Roasted ore SiO 2 7 grams PbO 70 grams Argol 2 grams Na 2 CO 3 20 grams Borax glass cover This method gives good results on jamesonite ores. Arsenical ores are assayed by the same methods as the anti- monial ores; also by the iron-nail method, although this last is not generally to be recommended. The subject of the best method of assay of antimonial and arsenical ores still lacks thorough investigation. The chief points may be outlined as follows: 1. In the roasting, unless great care is taken as regards temperature, mechanical loss of gold and silver takes place, owing to the rapid disengagement of the arsenic and antimony oxides, or sulphides of these metals. Unless the roast is con- ducted at a low heat and in the presence of considerable carbon, arseniates and antimoniates of base metals or silver may form, holding values which later on are not completely decomposed in the crucible, owing to their stability at a high temperature, the result being appreciable slag losses. 2. In the niter method, the presence of much niter, with its powerful oxidizing effect, may also induce the formation of arseniates and antimoniates, containing silver and possibly gold, which will remain in the slag. 3. In the iron-nail method, unless the 'fluxes are carefully adjusted and the temperature kept below 1100 C., speiss carrying values is very apt to form above the lead button, and thus neces- sitate a re-assay, or a treatment of this speiss. The Assay of Arsenical Nickel-cobalt Silver Ore. 1 Two types of ores may be considered. 1. Those high in Ag and also high in Ni and Co contents, and 2. those low in Ag, but high in Ni i D. K. Bullens, Eng. and Min. Jour.* XC, 809. Lodge, Trans. A. I. M. E., XXX VIII, 148 A MANUAL OF FIRE ASSAYING and Co contents. It is essential to flux the Ni and Co in the slag since these elements seriously interfere with cupellation causing low results. Ni present in the lead button to the amount of 0.5 per cent, causes a scum of NiO to be left on the cupel. More than this causes the "freezing" of the button. The effect of cobalt is not so pronounced as that of nickel. For the ores high in silver the scorification assay is to be recommended with the following charge: Ore 0.05 to 0.10 a. t. Lead 65 to 75 grams Borax glass 3 to 5 grams Silica 1 to 3 grams Slag and cupel corrections should be made. It is desirable at times to check results by wet analysis for silver. For ores low in silver the crucible assay with high litharge gives better results than .the scorification assay. Small amounts of ore, 0.10-0.2 a. t., should be used, for the nickel, cobalt and arsenic in the ores are apt to form a speiss in the assay. For ores containing metallic silver in any amount the "scale assay" should first be made. THE ASSAY OF SULPHIDES, MAINLY PYRITE, BUT CON- TAINING SMALL AMOUNTS OF COPPER, ZINC SULPHIDES, ETC. Where gold only has to be determined in ores of this charac- ter, the roasting method is satisfactory. This, however, proves unreliable for silver, and in many cases (as at Leadville) the silver contents of these sulphides are the most important. The best method, after many trials, was found to be the niter fusion on comparatively small lots of ore. The ore has the following analysis: Iron 33 to 44 per cent. Zinc 4 to 8 per cent. Sulphur 38 to 45 per cent. Copper . 5 to 3 . 5 per cent. Insoluble 4 to 20 per cent. Lead to . 4 per cent. Four assays are made on 0.25 assay ton each, with 3 to 4 assay tons of the following flux, the amount depending on the reducing power; i.e., on the amount of sulphides present: * PbO 8 parts SiO 2 1.5 parts KNO 3 1.5 parts Borax glass 1.5 parts Na 2 CO 3 3.0 parts SPECIAL METHODS OF ASSAY 149 Either a salt or a soda cover is used. The temperature of fusion is brought up gradually to a yellow heat. With 0.25 assay ton this gives the following charge: 1 Ore 0.25 assay ton Na 2 CO 3 24 grams PbO 62 grams Borax glass 11 grams KNO 3 12 grams SiO 2 11 grams The buttons are usually clean, and separate well from the slag. Another method which may be used on this type of ore is the niter-iron method. This has the advantage that no preliminary assay is necessary to determine the amount of niter for the proper size button, but that only sufficient niter is added to partially oxidize the sulphides, the iron nails being relied upon to decom- pose the balance of the ore. On ores of the class shown by the analysis, the following charge is successful: Ore 0.5 assay ton SiO 2 8 grams Na 2 CO 3 25 grams Borax glass .... 8 grams PbO 30 grams Iron nails 2 to 3 tenpenny KNO 3 15 grams Thin borax glass cover If the ore has a lesser reducing power than shown by the analysis given, niter and silica should be decreased in the charge. Rapid Methods for Sulphide Ores. 2 The approximate per- centage of sulphides in ores may be quickly determined by vanning with sufficient accuracy for the addition of the proper amount of niter. An ordinary color or spotplate, used in volu- metric chemical analysis is best used for the vanning. Small quantities of ore are placed in the four outside depressions and carefully vanned in a basin of water until only the sulphides are left. The quantity of these is estimated in per cent, of the total amount of ore taken for the vanning test. In gaining experience with this method it may be desirable for the assayer to make comparisons with ores of known sulphide contents. If pyrite be taken as the sulphide of unit reducing power, then chalcopyrite, blende, pyrrhotite, and arsenopyrite will have a reducing power of , stibnite of , and galena and chalcocite of that of pyrite. In the complex sulphide ores the relative amounts of the different sulphides are estimated and the amounts converted into terms of pyrite. In the ordinary excess litharge charge with a fair amount of borax and soda and with 0.5 a.t. of ore, 15 per cent. 1 See also W. G. Vail, "Niter Assay for Sulphide Ores," in West. Chem. and Met. ,11, 14. 2 F. G. Hawley, Eng. and Min. Jour., LXXXIX, 1221, XC, 647; also Min. and Set. Press, CI, 147, and E. T Hall, M in. and Sci. Press, CI, 345. 150 A MANUAL OF FIRE ASSAYING of pyrite will reduce a 22 gram button. For every 5 per cent, of pyrite present above 15 per cent., 2.1 grams of niter are neces- sary to destroy the excess reducing power. Two stock fluxes are used in the assay of ores. 1. The reducing flux, designed to give a 22 gram button with a neutral ore on a charge of 0.5 a.t. ore and a measure or scoop of flux (84 grams). This flux is made as follows: PbO, 15 parts; Na 2 C0 3 , 4 parts; borax, 2 parts; flour 0.44 parts. When 84 grams of flux are used this gives the following charge: Ore 0.5 assay tons PbO 60 grams Na 2 CO 3 16 grams Borax 8 grams Flour 1.75 grams 2. Non-reducing flux to be used in connection with niter for sulphide ores which will give a button larger than 22 grams. This flux is made as follows: PbO 15 parts; Na 2 CO 3 , 3.5 parts; borax 2.5 parts; silica 0.5 parts. When 84 grams of this flux are used it gives the following charge: Ore 0.5 assay ton PbO 60.0 grams Na 2 CO 3 14.0 grams Borax 10.0 grams Silica 2.0 grams Niter As necessary When sulphide ores are assayed which do not contain sufficient sulphides for a 22 gram button, the reducing and non-reducing fluxes are mixed in such proportion as to obtain the correct result. Thus suppose an ore contains 10 per cent, pyrite, its reducing power would be ||X22 = 14.6 grams lead on the basis of 0.5 a.t. The deficiency in lead is therefore 22 14.6 = 7.4 grams. In order to obtain 7.4 grams lead the following amount of re- ducing 'flux is required: f f X 7.4 = 28.27 grams. The balance of the charge of 84 grams will be made up of non-reducing flux, and the whole charge will be: Ore 0.5 assay ton Reducing flux 28.27 grams Non-reducing flux 55 . 73 grams The fluxes and niter are measured by volume in properly designed scoops or measures. SPECIAL METHODS OF ASSAY 151 For high sulphide ores when very accurate results are required a preliminary assay is made as follows: Ore 3 . 64 grams Non-reducing flux 50.0 grams This is run in a 10 gram crucible. This charge will give a lead button weighing as much as the niter necessary to oxidize all the sulphides in 0.5 a.t. of the ore. Place the lead button obtained in one scale pan of the pulp scale and from the hook above the other pan suspend by fine wire a weight so that with the wire it amounts to 6 grams. Then add niter to the pan having the 6 gram weight until the scale is in balance. This amount of niter is the proper amount necessary to produce a 22 gram button with the ore and the non-reducing flux if 0.5 a.t. of ore is taken for assay. (Consult Chapter V.) For important assays it is desirable to make 4 assays, combine the buttons from 2, and scorify into one button each. Make the two cupellations, weigh the beads separately for Ag, combine them for parting and make one weighing on gold. If the ores assayed contain more than 12 per cent, copper, it is desirable to take the lead buttons from the assay and place them into crucibles with 50 grams of litharge and 2 grams SiO 2 , place in the muffle and leave there four or five minutes after the PbO has melted. Then withdraw the crucible and with the tongs give the contents a rapid swirling motion for a few minutes and then pour. This treatment eliminates most of the copper re- maining in the button. Then cupel and part as usual. It is to be noted that the methods described may have to be modified to suit particular conditions. THE ASSAY OF MATERIAL CONTAINING METALLIC SCALES. Ores of this kind are difficult to assay and obtain correct re- sults from, as the metallic particles (usually gold or silver) are so unevenly distributed as to make it practically impossible to ob- tain an accurate sample. Two methods of assay are available: (a) Approximately 500 grams of ore (or less, if deemed advis- able) are weighed out, crushed, and put through a 150- or 200- mesh screen, care being taken to separate out the scales as closely as possible. Screening and crushing should frequently succeed each other. When all the scales have been separated out, they are transferred to a parting cup and dissolved in 3 to 5 c.c. of nitro-hydrochloric acid, if gold, or in nitric acid if silver or copper. The pulp is then heaped up into a cone in a large porcelain dish, 152 A MANUAL OF FIRE ASSAYING the gold, etc., solution poured on the apex of the cone, and the parting cup washed out thoroughly with warm distilled water, using no more than is necessary to completely wash it out. The bed of pulp should be thick enough to readily absorb all of the solution, and not permit it to penetrate to the dish. The pulp is then dried in an air bath at 120 C., thoroughly mixed on glazed paper, and put through the screen repeatedly. It is then assayed by the method suitable to it, like any other ore. (6) From 200 to 500 grams of ore are weighed out, crushed and screened, and the scales separated, as described above. The scales and the pulp are then weighed and the loss in dusting noted. The scales are assayed by scorification; the lead button is cupelled, and the bead weighed and parted. Then 15 grams of the ore is weighed out in duplicate, fused with the proper charge, the lead buttons from these fusions cupelled, and the beads weighed and parted. From the results obtained, the total amount of gold and silver in the original ore is calculated, con- sidering both pulp and scales. The gold and silver, respectively, found is multiplied by 29.166 and divided by the original weight of ore, taken in grams; this gives the value in ounces per ton. THE ASSAY OF ORES CONTAINING THEIR CHIEF VALUE IN FREE GOLD. As already pointed out, these ores are difficult to get correct results from. Even though the free gold particles are very fine, it is impossible to distribute them uniformly throughout the bulk of the sample. The proper way to assay material of this kind is to take from 1000 to 1500 grams of the ore, crushed through a 100-mesh screen, place in a large Mason jar with a tight screw cover, mix to a rather thick pulp with water and then add 3 to 4 c.c. of mercury from a burette. It is essential that the mercury should be free from gold and silver, or its contents of precious metals known. Most mercury as purchased contains some gold. The jar and its contents are then agitated for two hours, best in some mechanical agitator. Then carefully separate the mercury from the ore by panning in a gold pan, saving all the pulp, in another pan of somewhat larger size. None of the fine slimes of the ore must be permitted to escape. It may be necessary to add a little more mercury and a very little sodium amalgam during panning to collect any floured and sickened mercury. The pulp is allowed to settle in the pan, the surplus water carefully poured off, and the pan then set on a hot plate to dry. When dry it SPECIAL METHODS OF ASSAY 153 is mixed on a cloth, and 1 a.t. samples taken and assayed by a proper method. The mercury is carefully transferred from the pan to a porcelain dish, washed with water to free from sands, dried with filter- or blotting-paper and then transferred to a 20 gram crucible in which 20 grams of lead have been placed. To the crucible is then added a charge consisting of 30 grams PbO, 10 grams Na 2 C0 3 , 5 grams borax glass and 0.5 gram argol and silver foil enough to part the gold. The fusion is made by raising the heat very gradually; it is best to use a muffle that has not yet become red, and has a good draft through it, to prevent the escape of mercury fumes into the room. The button from the fusion is cupelled in the usual manner. The gold is weighed in mgs. and the weight divided by the grams of ore taken and multiplied by 29.166} gives the oz. gold per ton present as "free" gold. This figure added to the assay results from the pulp gives the total contents of the ore in oz. per ton. Another method 1 is carried out as follows: Take 6 a.t. of the sample crushed to pass 80 mesh, add sufficient water and a small amount of sulphuric acid to make a thin paste in an 8 in. porcelain mortar, add 8 grams of redistilled mercury and grind thoroughly for 30 minutes. Then separate the tailings from the mercury by washing them off with a stream of water obtained, say by attaching a hose to a hydrant. During the washing the mortar should be given a rotary motion. Collect the overflow from the mortar in a large gold pan and treat the pannings as described in the method above. When the mercury in the mortar is quite clean from sands give it a final wash with water, then dry it with filter-paper and transfer to a crucible containing enough litharge with reducing agent to giye a 20 grams button. Add enough silver to part the gold and start the fusion at a very low heat in the furnace. Cupel the lead button and weigh. Divide the weight of the gold by the number of assay tons of ore taken and add this to the figure obtained from the assay of the pulp in order to get the total value of the ore in ounces per ton. When the ore to be assayed contains arsenopyrite and graphite some of these will adhere to the mercury. In order to overcome this difficulty add to the washed mercury in the porcelain mortar 5 c.c. of cone. HNO 3 and enough silica to make a thin paste. > A. T. Roos, Mining World, XXXII, 319. 154 A MANUAL OF FIRE ASSAYING Grind for a few minutes and then wash the silica and acid off with water and proceed as before. AMALGAMATION TEST TO DETERMINE THE AMOUNT OF "FREE" GOLD PRESENT. 1 One hundred grams of crushed ore are weighed out into a citrate of magnesia bottle, 150 c.c. of water added and then 2 c.c. of pure mercury from a burette. The stopper is clamped, the bottle rolled in a piece of cloth and placed in a moving shaker for two hours. It is then removed, opened, covered with the thumb, shaken, and inverted over a 3 in. porcelain dish, and as much clean mercury as possible allowed to run out. A little more water is added and more mercury allowed to run out into another dish and so on as long as any comes out. If the mercury is not floured nearly all is removed in two operations. All the clean mercury is then put into a 250 c.c. beaker. The bottle is then shaken well and again inverted to let a little sand run out into a dish. This sand is then panned into an enameled dish, usually only a few globules of mercury being obtained. If much is found from this third inversion the whole charge must be panned and if the mercury is floured a small glob- ule of liquid sodium amalgam should be added to the pan. In ordinary routine work the tailings from the panning are discarded. For special purposes as when the tailings are to be tested by cyaniding, concentration, etc., they are allowed to settle completely, decanted and if necessary dried for further tests or for assay. To the mercury after its collection in the 250 c.c. beaker 0.5 gram of pure silver is added (if this has been carefully prepared and cleaned by treating in cyanide solution or weak nitric acid or by slightly amalgamating the surface it may be used to pick up the small globules of mercury collected in panning) about 150 c.c. of HN0 3 , sp. gr. 1.14, previously warmed to about 70 C. is now poured into the beaker which is set into an enameled pan on a hot plate and left there till all the mercury has dissolved. If not too hot it is unnecessary to cover the beaker. As soon as the mercury disappears the liquid is filtered on a 12.5 cm. paper previously wetted. The residue is rinsed on to the paper, washed once or twice with very dil. HNO 3 (not over 5 per cent.) and once with water. Test lead is then sprinkled on the paper, it is folded, placed on test lead in a scorifier and enough lead added to make a 20 gram button. 1 Method used at the Homestake Mine, S. D. Communicated by Wm. J. Sharwood. SPECIAL METHODS OF ASSAY 155 Silver is added to insure parting and also a few grams of borax glass. The scorifier is then charged into the muffle, the paper burned, the charge scorified for a few minutes, poured, the button cupelled and the bead parted and the gold weighed. From a 100 gram ore sample each mg. of gold represents 0.29166 oz. or $6.03 free gold per ton. Notes on the Carrying Out of the Amalgamation Test. The amal- gamation test is carried out for the purpose of determining the amount of precious metals that can be recovered from the ore in milling operations by means of amalgamation. The size of the crushed ore will influence the results; therefore, in different tests the degree of fineness must be nearly constant. The exact fine- ness used depends upon conditions. Temperature has its influ- FlO. 57. HOMESTAKE AalTATOB FOR AMALGAMATION AND CYANIDE TESTS. ence; if the tests are carried out at temperatures higher than the normal daily temperature results will be higher. If the tempera- ture be low results will be lower. The addition of silver to the mercury reduces the time required for its solution by about one- half. Extreme care must be taken to get mercury and silver practically free from gold. It is desirable to run a blank assay on say 20 c.c. of mercury and 5 grams of silver (representing ten times the quantity of each used in the assay). The mercury and silver are dissolved in acid as stated in the amalgamation test and the residues treated as described. The best mercury obtained after testing a number of new flasks and when used with commer- cial proof silver required a correction of 0.02 mg. of gold for 2 c.c. mercury and 0.5 gram silver. The silver contained nearly half of the gold thus found. If the amount of silver in the ore recoverable by amalgamation is to be determined the parting of the mercury by nitric acid must be replaced by the crucible 156 A MANUAL OF FIRE ASSAYING fusion of the mercury as described in method 1 for the assay of ores containing free gold. Equipment Required for Routine Amalgamation Tests. 1. Shaking box with 24 compartments each 3 in. square and 6 in. deep as shown in Fig. 57. The following are the details of construction. Sides and bottom made of 1.75 in. lumber; partitions of 0.5 in. lumber; sills 4X4 in. lumber; connecting rod If Xf in. oak lumber, 4 ft. long; supports of light steel l.SX-j^- in. and 2 ft. long; shaft 1 in. 250 r. p. m.; throw of eccentric, 2 in. A frue vanner eccentric rod and supports may be used in the construction of the agitator. A pad should be placed at the bottom of each pocket. Pieces of canton flannel, 12 to 15 in. square, are used to wrap each magnesia bottle. 2. Twenty-four citrate of magnesia bottles with spring clamps and rubber washers. These have a capacity of 350-370 c.c. 3. Twenty-four beakers, capacity about 250 c.c. 4. Two enameled iron pans to hold 12 beakers each. 5. Six porcelain dishes, 3 in. in diameter and 2 enameled iron pans, 8 in. in diameter and 2 in. deep. 6. Filtering rack for 12-2 in. funnels. Twelve extra beakers. 7. Copper weighing scoop and copper funnel with steep sides for charging bottles. 8. Cylinder graduated to 100, 150, and 200 c.c. 9. Glass stopper burette standing in enameled iron pan. 10. Supplies as mentioned in the assay. THE ASSAY OF CYANIDE SOLUTIONS. Method I. 1 Measure out any convenient volume into a beaker (preferably 10 or 20 a.t using beakers of 500 to 700 c.c. capacity). Add 10 to 20 c.c. of lead acetate solution containing 10 to 20 per cent, of the salt, then introduce 3 to 4 grams of zinc dust in the form of an emulsion or suspension in water and stand on a hot plate. When moderately heated but before boiling, acidify with about 20 c.c. strong hydrochloric acid, either c.p. or of the best commercial grade. Boil until action nearly ceases, and the reduced lead has collected into a spongy mass. Filter on a "quick" paper, and wash pre- cipitate twice with hydrant water. Remove the filter-paper and precipitate and squeeze out as much water as possible. Place in a 2 in. scorifier with 10 to 15 grams of test lead and 3 to 5 grams of borax glass. Place at once in the muffle, burn the paper, scorify for only'a few minutes, pour, cupel lead button, part and ' Due to Mr. Allan J. Clark, Homeetake Mining Co. SPECIAL METHODS OF ASSAY 157 weigh. Unless silver is to be determined, silver foil should be added to the scorifier for inquartation, or a measured volume of dilute AgN0 3 solution may be added to the beakers from a burette. Notes on the Method. About 100 grams zinc dust are usually mixed with 300 c.c. of water in a bottle with an -in. glass tube passing through the cork. This mixture is shaken into a capsule of the proper size used as a measure. The other reagents must be roughly measured. Their proportions should be varied slightly until conditions are found which yield a " sponge " of lead quickly with the particular solutions regularly assayed. Impure hydro- chloric acid does not give good results, nor do other acids. It is essential that nearly all the zinc be dissolved before filtering. Comparatively cheap filter-papers answer well. If a 300 c.c. flask be regraduated to deliver 301.45 c.c. every mg. of gold obtained from this volume represents $2.00 gold value per ton. If copper is present in solutions a somewhat longer scorification than above stated may be desirable. The method was suggested by Chiddey's method 1 in which zinc shavings and lead salt are used to produce a lead sponge. In this original method the lead sponge is recovered by hand and not filtered and then cupelled direct without a preliminary scorification. Clark's method gives somewhat better results on low grade solutions than evaporation methods with litharge or litharge bearing flux. It has the particular advantage of being an ex- ceedingly rapid method as compared to the tedious evaporation methods. Method 2. Evaporation Method. Measure out 5 to 10 a. t. or more of solution by means of a properly graduated flask and transfer to either porcelain or agate ware evaporating dishes, of 300 to 500 c.c. capacity. To the solution add 50 to 60 grams of litharge and place the dishes in a sand bath on a hot plate and carefully evaporate to almost complete dryness. If agate w r are dishes are used it is essential that the agate lining be unbroken, otherwise precious metals will precipitate on the iron surface and adhere to the same, giving low results in the assay. Tin dishes should not be used. When practically dry transfer contents by 1 A. Chiddey, Eng. and Min. Jour., LXXV, 473. Consult also W. H. Barton. West. Chem. and Met., IV, 67, and A. Whitby, Jour. Chem. Met. and Min. Soc. S. A., X, 134, 211, 288. 158 A MANUAL OF FIRE ASSAYING means of a spatula to a glazed paper, and remove any adhering litharge from the dish by means of a moist piece of filter-paper, thoroughly wiping out the dish. If the evaporation has not been carried too far this can readily be done. Then mix in a 20 grams crucible, 25 grams litharge, 15 grams Na 2 CO 3 , 2 grams argol, 2 grams SiO 2 and 5 grams borax glass, and transfer the litharge from the evaporation and filter-paper to the crucible and again mix with a spatula. Fuse the charge and proceed as usual. Unless silver is to be determined add silver foil to the crucible before fusion. Evaporation methods conducted in dishes made of lead foil have the disadvantage of permitting the use of comparatively small quantities of solution only and very frequently give low results. THE ASSAY OF SLAGS AND CUPELS FOR THE CORRECTION ASSAY. (a) Slags: The charge for these depends upon whether they are acid or basic. Particular care must be taken to get a charge that will completely decompose the original slag. If this is acid, the charge should aim to make a new slag more basic, and vice versa. The lead button should be from 25 to 30 grams in weight. Many assayers frequently add simply litharge and reducing agent to the slag in making the fusion. This is not always desirable, for if the slag already has much litharge in it, soda, etc., may with profit be added as the extra base in place of litharge. (6) Cupels: The bone- ash of the cupel will not unite with fluxes to form slags, but remains suspended in the fusion. For this reason the cupel should be put through a 150- to 200-mesh screen before assaying, the litharge-stained portion only being taken. For one large cupel, or two small ones, the charge is as follows: * Cupel Borax glass 45 grams PbO 60 grams Argol 2.5 grams Na 2 CO, 25 grams Soda cover Fluorspar is not desirable in the assay of cupels, as it merely adds another ingredient in suspension. Magnesia cupels may be fluxed with the following charge: Cupel Borax glass 20 grams PbO 40 grams Silica 10 grams Na a CO a 20 grams Argol 2.5 grams Borax cover SPECIAL METHODS OF ASSAY 159 Cement cupels are more easily fluxed and an ordinary crucible charge for a somewhat basic ore will answer very well. THE ASSAY OF MATERIAL CONTAINING METALLIC IRON. 1 Material of this kind will be obtained in the clean up of mortar boxes of stamp mills, the iron being present as pellets, and much larger pieces mixed with sand, pebbles, etc. It cannot be crushed and is assayed in the state received. Its correct sampling is practically impossible. Crucible fusions are made in the pres- ence of bisulphate of soda and niter. The charge is as follows: Material to be assayed 1 assay ton Bisulphate of soda 8 to 24 grams Na 2 CO 3 25 grams SiO 2 10 grams Borax glass 25 grams Litharge 35 grams Niter 1 to 4 grams The fusion should be conducted at a high heat for about 45 minutes. Then add to the crucible 15 grams of PbO mixed with 2 grams argol and continue fusion for 20 min. more until quiet. The action of the bisulphate is probably as follows. It breaks up on heating. 2NaHS0 4 = Na 2 S0 4 + H 2 + SO 3 The metallic iron is converted into FeSO 4 by the SO 3 in the early stage of the fusion, and is then converted into ferrous silicate as the temperature rises. The litharge and niter aid in the oxidation of the iron. Practically all of the PbO is reduced by the metallic iron. Interaction also takes place between the NaHSO 4 and the Na 2 CO 3 dependent on the quantities present. Na 2 CO 3 may with advantage be replaced by lime for this reason. The charge may have to be modified considerably in quantities of the reagents present to suit the material to be assayed. i "Modification of Method of H. R. Jolly," Jour. Chem. Met. and Min. Soc. S. A., VIII. 343. CHAPTER XI ERRORS IN THE ASSAY FOR GOLD AND SILVER LOSSES IN THE CUPELLATION OF PURE GOLD AND SILVER. These losses may be divided into (1) losses by absorption, (2) losses by volatilization. The losses of gold and silver in the cupellation are functions of (a) the temperature of cupellation; (b) the amount of lead with which the gold and silver is cupelled; (c) the physical nature of the cupel; (d) the nature and amount of impurities present; (e) the influence which silver has on the gold loss, and vice versa. There is considerable literature extant upon losses in cupella- tion of the two precious metals, but in the older researches the temperature influence is but vaguely defined, owing to the lack of means for ready and satisfactory temperature measurements, a deficiency which is now supplied by the LeChatelier platinum- rhodium pyrometer. Losses are also expressed as percentages of the total amount of metal cupelled, and then the average per- centage losses are indicated. That this is very deceptive is made evident by reference to the curve of losses accompanying this chapter. It is for this reason that the statement of results given by Mason and Bowman, 1 that the average loss in cupellation of pure silver under normal conditions is 1.99 per cent, and for gold 0.296 per cent., does not convey any very definite idea, unless the amount of metal cupelled is accurately specified, as well as the temperature. This fact has been noted by other observers, 2 but no effort has been made to express results coordinately. The following data show the losses which occur: 1 Jour. Am. Chem. Soc., XVI, 505. 2 Kaufman, in Eng. and Min. Jour., LXXIII, 829. Miller and Fulton, in "School of Mines Quart.," XVII, 169 ERRORS IN THE ASSAY FOR GOLD AND SILVER 161 TABLE XXVIII. CUPELLATION OF PURE SILVER (J. EAGER AND W. WELCH x ) Amt. of silver milligrams Amt. of lead grams Temperature deg. Cent. 2 Total losses per cent. 204.62 10 700 1 . 02 (average) 205 10 775 1.28 203 10 850 1.73 203 10 925 3.65 203 10 1000 4.87 TABLE XXIX. CUPELLATION OF PURE SILVER (L. D. GODSHALL 3 ) Amt. of silver . Amt. of lead Mgs. grams Approximate Temp. deg. Cent, of air in muffle Total loss in per cent. 2 7.5 750 3.66 2 15.0 750 4.40 2 22.5 750 5.52 2 30.0 750 5.96 5 7.5 750 3.29 5 15.0 750 2.63 5 22.5 750 3.83 5 30.0 750 4.31 10 7.5 750 3.73 10 15.0 750 2.89 10 22.5 750 4.47 10 30.0 750 4.26 20 7.5 750 3.42 20 15.0 750. 2.34 20 22.5 750 3.59 20 30.0 750 3.10 50 7.5 750 2.14 50 15.0 750 2.46 50 22 . 5 750 2.33 50 30.0 750 2.89 100 7.5 750 2.11 100 15.0 750 2.40 100 22.5 750 2.10 100 30.0 750 2.28 200 7.5 750 1.71 200 15.0 750 1.64 200 22.5 750 1.62 200 30.0 750 2.07 1 Lodge, "Notes on Assaying," p. 59. 2 Of air in muffle, directly above cupel. 3 Trans. A. I. M. E., XXVI, 473. 11 162 A MANUAL OF FIRE ASSAYING TABLE XXX. CUPELLATION OF PURE SILVER. (KAUFMAN 1 ) Amt. of silver Mgs. Amt. of lead grams Approximate | Temp. deg. Cent. I of air in muffle ; Total loss in per cent. 25 5 750 2.14 25 10 750 2.63 (2.38, 2.43) 25 15 750 2.69 25 25 750 2.09 (2.48, 2.44) 50 5 750 1.43 50 10 750 2.23 (2.10, 1.96) 50 15 750 2.14 50 25 750 .86 (2.25, 2.37) 100 5 750 .30 100 10 750 .61 (1.82, 1.42) 100 15 750 .68 100 25 750 .12 (1.93, 2.12) 200 5 750 .86 200 10 750 .24 (1.29, 1.17) 200 15 750 .40 200 25 750 .74 (1.46, 1.76) Parentheses indicate different types of cupels, viz., bone-ash made up respectively with pearl-ash and stale beer. The main figures were obtained by bone-ash oupels made up with water. The results, viewed as a whole, indicate that all three types have equal merit. Godshall (Table XXIX), experimented with different types of standard bone-ash cupels (some made at the mint), with the same result. The agreement amongst the different writers is very close, when the fact is taken into consideration that in the last two cases no precise statement concerning temperature is made, and that the amounts of lead differ somewhat. 1 ! Authority Amt. of silver mgs. Amt. of lead Grams Temperature Deg. Cent. Air in muffle. Total loss per cent. Eager and Welch. 205 10 775 1.28 Godshall 200 7.5 ?(feathers) 1.71 Kaufman 200 10 ? (feathers) 1.24 Liddell 2 102 20 ?(feathers) 1.70 ' Eng. and Min. Jour., LXXIII, 829. 2 Eng. and Min. Jour., LXXXIX, 1264. ERRORS IN THE ASSAY FOR GOLD AND SILVER 163 1+ ' i- : + 1 ' i! 1 | ! Ij j / I JS ' + cj : i e ! + a : I S i | / ' s i i |j 1 11 i j ' ' i | i | . / ! |t sf ! )/ ::: 1 1 / Jj 1 g -1 -Hi.*.-! "i! I I' : J /, // i / i / I / ,- /] f/j (/ i- /// / / / / /I / i / i ' ! \ 8- ///I 6601 JO 164 A MANUAL OF FIRE ASSAYING The accompanying curves are constructed from figures in Mr. GodshalPs paper. The general averages are taken, and while his losses are perhaps a trifle higher than the best work calls for at the present day (owing to a better recognition of the precise temperature required), they form the best and most complete data for the construction of curves showing the relation between amounts of silver cupelled and the percentage loss. I refrain from a mathematical discussion, but an equation covering the case is tentatively offered. 1 The influence of the size of lead button is clearly discernible by the ordinates of the curves. The temperature variations will show in the same way. The literature of gold losses is considerably less than that for silver. Rose 2 discusses them in the gold bullion assay. He gives the total loss on bullion 916.6 fine, under normal tempera- ture conditions, as from 0.4 to 0.8 per 1000, of which 82 per cent, is cupel absorption, 10 per cent, volatilization (probably), and 8 per cent, solution in acid. This, calculated to percentage on actual gold, is equivalent to 0.0803 per cent, for the highest loss. (This is cupel loss only, not including solution loss.) Hillebrand and Allen's results contain interesting data re- garding the relative losses by absorption and volatilization, to which reference will be made again. CUPELLATION OF GOLD-SILVER ALLOYS. The loss of gold and silver in cupellation is somewhat different when both gold and silver are present from the loss when either metal alone is present. TABLE XXXI. CUPELLATON OF GOLD (EAGER AND WELCH 3 ) Amt. of gold Amt. of lead i Temp. C. 4 , Total loss per cent. 201 10 775 0.155 201 10 850 0.430 204 10 925 0.460 201 10 . 1000 1.430 201 10 1075 3.000 1 1 am indebted to Prof C. C. Van Nuys, ! " Metallurgy of Gold," 1902, p. 506. 3 Lodge, "Notes on Assaying," p. 142. * Of air in muffle, directly above cupel. I. A., for the curves and the equations. ERRORS IN THE ASSAY FOR GOLD AND SILVER 165 TABLE XXXII. CUPELLATION OF GOLD (HlLLEBRAND AND ALLEN 1 ) Amt. of gold mgs. Amt. of lead grams Approximate Temp. deg. Cent, of air in muffle Total loss per cent. Total loss cupel absorption per cent. Total loss volatilized per cent. 30 58 25 750 feathers 36 30.32 30.63 30.45 30.16 30.66 10 34 25 25 25 25 25 25 increased increased increased increased back of muffle 750 front of 1.19 1.76 3.78 4.17 4.43 0.29 80 77 93 92 20 23 7 8 10.25 25 muffle increased 4.68 10 29 25 1 36 10 27 25 increased 10 42 10.17 25 back of muffle 16.43 TABLE XXXIII. CUPELLATION OF GOLD (ROSE 2 ) Temp, of cup- Amt. of gold Amt. of silver Amt. of Pb ellation deg. Total loss gold mgs. mgs. grams Cent. per cent. air in muffle 1 4 25 900 1.2 1 6 25 900 1.05 1 8 10 900 0.90 1 10 25 900 0.80 1 6 25 700 0.45 1 10 25 700 0.39 500 1250 10 900 0.055 1 Bull. No. 253, U. S. G. Survey. *Eng. and Min. Jour., LXXIX, 708. 166 A MANUAL OF FIRE ASSAYING TABLE XXXIV. CUPELLATION OF GOLD-SILVER ALLOTS (HlLLEBRAND AND ALLEN) ALL CUPELLATIONS MADE WITH 25 GRAMS OF LEAD Amount gold mgs. Amount silver mgs. Temp. 6 C. Total loss Ab 1uS by ! v "" Gold per cent. SUver per cent. Gold per cent. Silver per cent. Gold per cent. Silver per cent. 30.06 30.40 30.60 30.07 30.61 15.56 15.14 15.15 15.44 15.52 15.39 10.67 10.57 10.53 10.63 10.60 10.21 90.51 90.19 90.74 90.67 90.75 45.06 45.19 45.41 45.30 45.59 45.05 30.33 30.64 30.42 30.52 30.38 30.44 750 (air in muffle) 0.50 1.22 2.32 3.76 3.89 0.19 0.40 1.52 2.07 2.59 2.40 0.47 1.61 5.13 10.63 12.46 12.53 1.70 3.73 5.51 7.66 7.98 1.91 3.30 4.14 5.78 6.55 6.61 2.17 5.68 10.19 15.99 18.34 18.69 33 17 48 33 33 27 19 22 37 38 30 29 increased back of muffle. . . . front, 750 increased increased increased increased back of muffle. . . . front, 750 67 83 52 67 67 73 81 78 63 62 70 71 increased increased increased back of muffle. . . . Rose shows (Table XXXIII) the protective action that silver exercises over gold, the total loss of gold decreasing as the amount of silver present increases. Hillebrand and Allen show how the total loss is distributed between absorption by the cupel and volatilization. It is evident that while the total loss of gold is decreased by the presence of silver, the volatilization loss of gold is increased by the presence of silver (compare Tables XXXII and XXXIV) . When gold and silver are present in the ratio of 1 to 2, the averages are as follows: Of the total gold loss, 68 per cent, is absorbed, 32 per cent, is volatilized. Of the total silver loss, 71 per cent, is absorbed, 29 per cent, is volatilized. However, as the total loss is determined by the difference in weight between the proof gold and silver and the weights of the cupelled bead and parted gold, and the volatilization item by the difference between the total loss and the amount recovered by the re-assay of the cupel, it is evident that certain errors obtain which apparently make the volatilization loss appear greater ERRORS IN THE ASSAY FOR GOLD AND SILVER 167 than it really v is. The error, however, cannot be very great. The data are inconclusive regarding the influence of the tempera- ture on the relative losses by absorption and volatilization, but it seems indicated that the volatilization loss is proportionately greater with an increase of the temperature s of cupellation. LOSSES IN THE ASSAY OF ORES. Table XXXV, etc., show losses of gold and silver in the assay of ores, during fusion and cupellation, as influenced by the presence of certain impurities. TABLE XXXV. TELLURIDE ORES Amount of gold in weight of ore taken for assay Weight of lead button Method of fusion. Slag loss Cupel absorption Milligrams Grams Per cent. Per cent. ( l )493.83 20 Crucible 7 0.51 1.56 95.57 20 Crucible 0.38 0.23( 5 ) 1.54 20 Crucible 1.30 0.40 1.95 20 Crucible 0.50 0.50 1.19 20 Crucible 0.40 1.17 20 Crucible 0.40 3.60 20 Crucible 2.14 0.80 6.20 20 Crucible 0.64 0.32 6.23 20 Crucible 0.50 0.64 1.38 20 Crucible 0.80 1.00 ( 2 ) 18.18 25 Crucible 0.49 5.85 25 Crucible 1.03 0.12() ( 3 ) 34.0 27 Crucible 0.21 0.23 34.0 27 Crucible 0.56 34.0 25 Crucible 0.15 0.41 68.0 25 Crucible 0.13 0.07 68.0 25 Crucible 0.16 0.22 (*) 15.5 Crucible 8 0.25 0.19 15.49 Crucible 0.13 0.38 19.54 Crucible 0.20 0.23 19.63 ' Crucible 0.10 0.25 Woodward, in "West. Chem. and Mel.," I, 12. Fulton, in "School of Mines Quart.," XIX, 419. Lodge, in "Tech. Quart." 1899, XII, 171 (averages). Bull. No. 253, U. S. G. Survey (averages; Hillebrand and Allen). Average of 34 fusions, tellurium in all beads. Average of 10 fusions. Cripple Creek flux. Excess-litharge charge. 168 A MANUAL OF FIRE ASSAYING TABLE XXXVI. ZINCIFEROUS MATERIAL, ETC. Amount of Au and Ag in weight of ore taken for as- Weight of lead button Slag loss Cupel absorption say. Method of assay Remarks Au Ag Au Ag Au Ag mgs. mgs. Grams per cent. per cent. per cent. per cent. "232.0 287.0 18 Scorification after 0.06 0.40 0.11 1.30 Zn. ppt. containing acid treatment 42.3 per cent. Zn 232.0 284.0 18 Scorification after 0.04 0.34 0.08 1.10 Figures represent acid treatment averages. 232.0 284.0 18 Direct crucible 1.04 1.10 0.16 1.50 fusion 2 233.0 197.0 20 Crucible fusion 0.06 0.51 0.18 1.18 Zn. ppt. containing after acid treat- 14.3 per cent. Zn, ment 9.1 per cent. Cu. 233.0 202.0 20 Direct crucible 0.15 2.73 0.16 1.29 Figures represent fusion averages 3 561.0 20 Crucible fusion 0.75 1.38 Galena niter method 567.0 20 Crucible fusion 0.65 1.42 Galena niter method 3 175.0 20 Crucible fusion 0.23 1.90 Silicious ore con- niter method taining some copper 174.0 20 0.37 1.66 Silicious ore con- taining some copper 1 Fulton and Crawford, in School of Mines Quart., XXII, 153 2 Lodge, in Trans. A. I. M. E., XXXIV, 432. Miller, in School of Mines Quart, XIX, 43 ERRORS IN THE ASSAY FOR GOLD AND SILVER 169 TABLE XXX VII (>). HIGH-GRADE CARBONATE AND SULPHIDE SILVER ORES 2 Amt. of Au and Ag in weight of ore taken for Weight of lead button Method Slag loss Cupel absorption Second correction assay of from fusion of slags assay anu cupels 01 nrst Au Ag Grams Au per Ag per Au per Ag per correction per cent. mgs. cent. cent cent. cent. 1130 24 Crucible fusion 0.76 1.25 0.248 1130 28 Crucible fusion 0.361 0.956 Crucible fusion 1130 28 Double amt. of 0.511 0.88 0.15 fluxes Crucible fusion 1130 32 Double amt. of 0.736 1.02 0.15 fluxes 226 15 Scorification 2.8 1.40 0.559 24 1719 25 Crucible fusion 0.05 0.143 0.07 0.711 12 860 20 Scorification 0.12 0.72 0.080.86 TABLE XXXVIII. CUPRIFEROUS MATERIAL Amt. of Au and Agin weight of ore taken for Weight of lead button Method Total loss recovered including slag and cupel assay of assay Remarks Au Ag Grams Au Ag mgs. mgs. per cent. per cent. i 11.46 88.0 20 Crucible fusion 0.96 2.90 Mattes containing about 20 per cent. Cu. 5.36 46.8 20 Crucible fusion 2.98 4.30 4.20 37.24 20 Crucible fusion 2.38 2.70 3.52 33.79 20 Crucible fusion 2.84 5.47 4.20 37.24 20 Scorification 5.95 5.26 3.52 33.79 20 Scorification 3.12 5.71 The foregoing tables represent for the most part averages, and in every case the losses for the normal assay; i.e., in the case of the fusion, the charge known to yield the best results, and the proper temperature for cupellation. The losses are therefore 1 First five results on lead carbonate ore, last two on silver sulphides. All results repre- sent averages. 1 Miller and Fulton, ibid, XVII, 160. 170 A MANUAL OF FIRE ASSAYING to be ascribed to the nature of the material assayed, chiefly to the influence of certain elements present. In considering the percentage of loss, it must be recalled that this varies inversely with the amount of precious metal in the charge, i.e., with the size of the gold-silver bead. The sum of the cupel absorption and the slag loss (which can, in part, be recovered) is not the total loss, as it does not include that by volatilization, which is small in most cases, but in some cases, again, may be quite appre- ciable, as in the case of telluride ores. What the loss is in slag, when no element like tellurium, copper, zinc, etc., is present, may be seen by reference to Table XXXVI, to those assays fused after acid treatment, and to TableXXXVII, showing crucible fusions on lead carbonate ore. The slag loss in gold and silver for these ores is very small. In cases where the impurity present and causing loss is nearly all eliminated in the fusion, e.g., zinc, antimony, etc., the cupel absorption is practically that for pure silver and gold under the same circumstances. Where the impurity is tellurium, or selenium, or copper, the cupel absorption is decidedly increased. One fact is to be noted, the fact that the slag losses present no regularity, even for the same material. This is prob- ably due partly to differences of slag composition among different experimenters, and partly to difference of temperature of fusion; and also to the method of refusion of slag. The high loss in scorification slags shown in Table XXXVII for lead carbonate ores containing silver is due to the general un- suitability of the ore for scorification, although scorification slags show higher losses than crucible slags. That, in spite of this, scorification assays, on silver-bearing material show equally good and better results in many cases than the crucible assay, is due to the fact that the silver beads retain small quantities of lead and copper (see further on), and to the fact that in the multi- plication of the weight of the silver bead by 5 or 10, or whatever the assay-ton factor may be, this error is multiplied, giving an apparently better result. The amount of slag has comparatively little influence on the amount of precious metals retained, provided the amount of collecting lead is ample. Buttons of less than 18 to 20 grams should not be made, and if the amount of slag is great or the quantity of silver and gold in the qharge is more than 500 mgs., 25- and 30-gram buttons are essential. In the case of large buttons which contain no impurity, it is also best to cupel direct, ERRORS IN THE ASSAY FOR GOLD AND SILVER 171 if possible, rather than rescorify to smaller size, as the rescorifi- cation causes greater loss than the direct cupellations. During scorification there is also an appreciable loss of the precious metals by volatilization, which is absent in the crucible assay. This, in the case of telluride or zinciferous ores, may become so great as to put scorification out of the question. OTHER ERRORS. Retention of Lead in Cupelled Beads. Small quantities of lead are almost invariably retained in the gold and silver beads with ordinary temperatures of cupellation. Hille- brand and Allen, 1 in two careful experiments on sets of three beads, approximately together 90 mgs. gold, found that 0.30 per cent, and 0.37 per cent., respectively, of lead were retained. This retention of lead cannot be corrected by leaving the bead in the muffle for some length of time after the blick, as this is, of course, prohibitive in the case of silver, and in the case of gold seems to actually cause an increase of weight. It has already been stated that copper and tellurium are very apt to be present in the final bead, when in the ore in any appreciable quantity. The retention of base metal by the bead causes a plus error in silver, but will not effect the result on gold unless the parting is by H 2 SO 4 ; and where the weight of the bead is multiplied by a factor to get results per ton, the final error in silver may be very appreciable. The presence of copper in the final bead practically insures the complete removal of the lead. In order to show what is usually termed "fine silver" the follow- ing analysis of Government fine silver is appended. Ag, 99.929%; Cu, 0.056%; Pb, 0.003%; Au, 0.007%; As, 0.001%; Sb, 0.002%; Fe, 0.001%; Zn, trace. 2 Retention of Silver by the Parted Gold. Ordinary parted gold, after the proper treatment with weak and strong acid, retains from 0.05 to 0.10 per cent, of silver. In the assay of gold bullion after the first acid treatment of the quartation alloy, the gold on the average retains 0.25 per cent, silver. After the second acid treatment, the final silver retention is from 0.06 to 0.09 per cent., depending on the time of boiling. If the amount of silver to gold in the quartation alloy is less than 2.5 to 1, some- what more than the above amount of silver will be retained. 3 Silver can, practically, be completely extracted by more than Bull. No. 253, U. S. G. Survey. * Min. Ind., XV, 545. 3 Rose, "Metallurgy of Gold," p. 453. 172 A MANUAL OF FIRE ASSAYING two treatments with acids, according to Hillebrand and Allen. 1 In the ordinary assay for ores as usually carried out, it is safe to assume that some silver is invariably retained by the gold, and frequently much more than is supposed; however, with low- grade ores, this retention is negligible. Solution of Gold by Acid. It is essential that the nitric acid used for parting be free from impurities, especially from hydro- chloric acid and chlorine; otherwise solution of gold is sure to follow. Gold is quite soluble in mixtures of hot sulphuric and nitric acid, 2 and is again precipitated by dilution. According to Hillebrand and Allen, 3 nitrous acid (HNO 2 ) and mixtures of HNO 3 and HNO 2 do not dissolve gold, though there is much earlier literature to the contrary. Nitrous acid has frequently been considered in this connection, as it is formed to some extent by the action of HNO 3 on silver. According to Rose, 4 some gold is dissolved by nitric acid on continued boiling to constant gravity of acid. This solution is placed in the bullion assay at 0.05 per cent, or 0.5 parts per 1000. Hillebrand and Allen state that the loss of gold by solution is very small and irregular. It may be disregarded in the ore assay. The solubility of gold in HNO 3 is readily demonstrated when large quantities of gold are used. F. P. Dewey 5 in careful experiments showed the solution of gold to the extent of 660 mgs. per liter of of cone, acid, on boiling about 25 grams gold .for two hours. He states that the temperature (120 C.) required to boil cone, acid has as decided an influence as the strength of the acid. Occluded Gases. Parted gold beads and "cornets" retain about twice their volume in occluded gases after annealing. The principal gas is stated to be carbon monoxide. Two volumes amount to 0.02 per cent, by weight, which is already allowed for in the silver retention. Errors in Weighing. The best scales are accurate to 0.01 mg., and scales can be obtained weighing to 0.005 mg. This last is used in assay offices, where great accuracy is required, on such material as bullions, rich mattes, etc. It is usually an unnecessary refinement in the ordinary ore assay, for the reason that the probable error in the assay is greater than this. 1 Bull. 253, U. S. G. Survey. 2 Lenher, in "Jour. Am. Chem, Soc.," XXVI, 552 3 Ibid. 4 Ibid., p. 507. 6 Jour. Am. Chem. Soc., XXXII, 318. ERRORS IN THE ASSAY FOR GOLD AND SILVER 173 The errors in the assay for gold and silver may be summarized as follows: 1. Losses by absorption in the slag of the fusion. 2. Losses by volatilization during fusion. 3. Losses by absorption during cupellation. 4. Losses by volatilization during cupellation. 5. Errors by gain in weight of bead, due to retention of foreign elements. This affects results on silver chiefly. 6. Errors in weight of gold after parting by the retention of silver and occluded gases. 7. Losses of gold by solution in nitric acid. 8. Errors in weighing. The chief losses are Nos. 1 and 3, which can be recovered by "corrected assay," i.e., by re-assay of slag and cupel, to the ex- tent of about 80 to 85 per cent. Wherever considerable accuracy is required, corrected assays should always be made. The losses by volatilization are usually slight, although from the foregoing data these are sometimes seen to be considerable. The retention of foreign metals by the bead is a plus error in favor of silver, and the retention of silver in the parted gold is a plus error in favor of gold. Silver losses are considerably greater in magnitude than gold losses. The total amount of precious metal recovered by the assay varies with the nature of the material. Designating the total amount of gold and silver in an ore or produet as 100, the corrected assay will show from 99 to 99.8 per cent, of the gold, and from 98 to 100+ per cent, of the silver, the high silver result in some cases being due to retention of foreign metal. In the bullion assay for gold, the algebraic sum of the errors outlined, the losses being designated minus and the gains plus, is called the "surcharge." In the gold bullion assay this will vary from +0.025 per cent, in very pure gold bullion, to 0.25 per cent, in base bullion, passing to zero for a bullion about 800 fine. CHAPTER XII THE ASSAY OF BULLION GENERAL. Bullion is classified as follows: 1. Lead bullion, usually the product of the lead blast-furnace; 95 per cent, and more lead, containing some copper, antimony, etc., silver and gold. 2. Base bullion, containing from 100 to 925 parts of silver per 1000, gold in varying amounts, and a large percentage of base metals, chiefly copper, zinc, lead, etc. Produced most frequently by cyanide mills. 3. Dore bullion, containing 925 to 990 parts of silver per 1000, some gold, and base metals, mostly copper, but also lead, antimony, zinc, etc. 4. Fine silver bullion, free from gold, containing 990 and more parts silver per 1000, but some base metals, usually copper. 5. Silver bullion, containing little base metal and less than half its weight in gold. 6. Gold bullion, containing little base metal and more than half its weight in gold. 7. Fine gold bullion, free from silver, containing from 990 to 1000 parts gold per 1000. Silver and gold in all bullions but lead bullion are estimated in parts per thousand, and bullion is said to be so many parts fine. Thus, if 1 gram (1000 mgs.) of bullion is taken for assay and it contains 925 mgs. gold, it is said to be 925 fine. In the assay of gold bullion the "millieme" system of assay weights is used, a millieme being 0. 5 mg., and the assay is re- ported in parts of 10,000, or the fineness with one decimal added. Thus the above bullion would be reported as 925.0 fine. In this system the 500-mg. weight is stamped 1000, the 250-mg. weight 500, etc. The scales used must therefore be sensitive to 0.05 mg., or 0.1 millieme. This presents no difficulty, as ordinary assay balances are sensitive to 0.01 mg. with a load of 0.5 gram. Lead bullion is recorded in oz. per ton, in the same way as for ores. 174 THE ASSAY OF BULLION 175 THE ASSAY OF LEAD BULLION. The sample of bullion may be melted under charcoal and granulated in cold water, or it may be rolled out into a strip in the rolls, and the pieces cut at inter- vals from this for the sample. If lead bullion is free from copper, antimony, zinc, sulphur and arsenic, etc., it may be cupelled directly for gold and silver. In this case, 4 portions of 0.5 assay ton each are wrapped in about 7 grams of sheet lead, placed in the hot cupels, and cupelled with feathers. The cupels are fused with the following charge: Stained part of cupel 45 grams borax glass 80 grams PbO 2 grams argol 15 grams Na 2 CO 3 Thin litharge cover The buttons from this fusion are cupelled and the weight of the gold and silver added to that obtained from the first cupella- tion. If the bullion contains base metals which will influence the results of the cupellation, 4 portions of either 0.5 or 1.0 assay ton are weighed out and mixed with 30 to 50 grams of test lead; 1.5 grams of borax glass and 0.5 gram of silica are put on top of the lead and the charge scorified. The resultant buttons, which should weigh about 15 grams, are then cupelled. The scorifier slag and cupel are re-assayed by the above charge and the correc- tion added. THE ASSAY OF SILVER BULLION 1 (also applicable to Base Bul- lion, Dore Bullion, etc.). CUPELLATION METHOD. This method is used as an approximation for bullions in which silver is to be determined accurately, serving as a preliminary assay for the salt titration, mint, or Gay-Lussac method. (a) Preliminary Assay. Exactly 500 mgs. of bullion are weighed out on an assay balance in order to save calculation, wrapped in 10 grams of sheet lead, and cupelled at 850. C., or with ample feathers of litharge. The silver bead is cleaned, weighed and parted in 1 to 9 HNO 3 for at least 20 minutes; then, if any gold shows, heated for 5 minutes more in concentrated acid, washed, and the gold dried, annealed and weighed. The amount of gold found, subtracted from the weight of the bead, gives the approximate silver, and the weight of the bead, sub- tracted from the amount of bullion taken (500 mgs.), gives the 1 For sampling of silver bullion, see "The Assay of Gold Bullion," later in this Chapter. 176 A MAXUAL OF FIRE ASSAYING base metal. This base metal is usually copper, and its presence may be detected by the coloring of the cupel. (&) Making the Check Assay. As the loss of silver and gold is a question of temperature, amount of precious metal present, amount of lead of cupellation, and amount and kind of base metal present, it is desirable to have the regular cupellation, accompanied by a check assay, made up as nearly as possible to the composition of the bullion to be assayed, and cupelled under the same conditions. The check assay is therefore made up from data obtained in the preliminary assay. As the silver determined in this preliminary assay is low, due to absorption and volatiliza- tion, a correction of 1.2 per cent, is added as an approximation or, rather, the amount of Ag found is considered as 98.8 per cent, of that present, and this amount of proof silver weighed out. To this is added, in proof gold, the amount of gold found in the preliminary assay. The difference between the sum of the corrected silver and the gold, and 500, is the amount of base metal to be weighed out for the check. As already stated, the base metal is usually copper, and in making up the check c.p. sheet copper is used. The check thus weighs 500 mgs. and approx- imates very closely the composition of the bullion. Duplicates of 500 mgs. of bullion are now weighed out, and these and the check each wrapped in the proper amount of sheet lead, as de- termined from the table below: TABLE XXXIX. LEAD RATIO IN CUPELLATION Fineness in Amount of Amount of lead silver copper present for cupellation Ratio of lead Milliemes Milliemes Grams 1000 3 900 100 7 140 to 1 800 200 12 120 to 1 500 500 18 72 to 1 300 700 .21 60to 1 (c) The Assay. Three cupels are placed in a row across the muffle, so as to be exposed as nearly as possible to the same THE ASSAY OF BULLION 177 temperature, and three more cupels are placed near them to act as covers for the cupellation when finished, in order to prevent sprouting. When the cupels have had all volatile matter expelled the assays are dropped into them, the check in the center one, and the cupellations carried on in the usual way, with feathers. After the blick, the cupels are drawn to the front of the muffle and covered with extra cupels. Sprouted buttons must be re- jected. The beads are now cleaned, weighed, and rolled out, parted in flasks, with the acids as described for the preliminary assay, and the gold weighed. The difference between the silver actually used in the check and that found by assay is the correction to be added to the mean silver result of the two bullion assays made, which should not differ by more than a millieme (0.5 point fineness). This correction may be plus or minus, according to the amount of copper in the bullion; for with much copper, some of this may be retained by the silver and give rise to a minus correction. The gold is corrected in the same way as the silver. The sub- traction from 500 of the sum of the corrected silver and gold gives the amount of base metal. The individual results obtained, express the assay results in fineness. When metals of the platinum group are present, the method "must be modified as outlined, in Chapter XIII, for the assay of platinum, etc. WET METHODS: GAY-LUSSAC OR MINT METHOD. This method is a. most accurate one and is based on the complete precipitation of Ag as AgCl in a nitric acid solution by means of sodium chloride. The reaction is as follows: AgN0 3 + NaCl = AgCl + NaN0 3 1 part Ag = 0.54207 NaCl The standard solution of NaCl usually employed is of such strength that 100 c.c. precipitate 1 gram of Ag, so that 5.4207 grams of c.p. NaCl are dissolved per liter of distilled water to give the standard solution. This solution can also be made up by using a saturated salt solution at 60 F., and then adding 2.07 parts of this to 97.93 parts of distilled water. The last method of obtaining the solution is not as good as the first, owing to the difficulty of obtaining the precise temperature of 60 F. and keeping it there. Aside from the standard solution men- tioned, there is required another of one-tenth its strength (obtained 178 A MANUAL OF FIRE ASSAYING by taking 1 part of the standard NaCl solution and adding to it 9 parts of distilled water), and an acidulated solution of AgN0 3 , obtained by dissolving 1 gram of proof silver in 15 c.c. of HN0 3 , 1.26 sp. gr., and diluting with distilled water to 1000 c.c. It follows from the above that 1 c.c. of the one-tenth solution will just precipitate the Ag in 1 c.c. of the acidulated silver nitrate solution. The standard NaCl solution is termed the "normal salt" solution in the assay, although not properly so; the weak solution is termed the "decimal salt solution," and the silver nitrate solution the "decimal silver" solution. Standardizing Solutions. The apparatus required is: 1. A large bottle or carboy, containing the normal salt solution placed on an elevated shelf so that the solution may be si- phoned by means of glass tubing and rubber hose to the main 100-c.c. pipette. 2. Liter bottles containing respectively the decimal salt and the decimal silver solutions. 3. An accurate 100-c.c. pipette, clamped to a suitable stand, and provided at the top with a glass overflow-cup containing a moistened sponge to catch the overflow of the normal salt solution. 4. Two small graduated 10-c.c. pipettes, one for the decimal salt and one for the decimal silver solution. Burettes may be used in place of these. 5. A number of strong 8- to 12-oz. bottles, similar to reagent bottles, provided with rubber corks. The standardizing of solutions is carried out as follows: Two portions of exactly 1002 mgs. proof silver are dissolved in 15 c.c. of 1.26 sp. gr. HNO 3 , the nitrous fumes are removed by boiling, the solution is transferred to the titration bottles and water added to bring up the amount of solution to 125 c.c. The 100-c.c. pipette is then rilled with normal salt solution to the mark, after washing out with salt solution to prevent dilution. The filling is done by fastening the siphon hose to the bottom of the pipette, opening the clamp on the hose, and letting the pipette fill, with a little overflow. The solution is then shut off by clamping the hose, a finger placed on the top opening of the pipette to prevent the solution running out, and the hose removed. The pipette is then permitted to drain to the 100-c.c. mark, and the solution held there by closing the top of the pipette with the finger. The THE ASSAY OF BULLION 179 bottle containing the dissolved proof silver is then placed under the pipette and the normal salt solution permitted to completely drain into it. The bottle is then violently shaken for three or four minutes, either by hand or a mechanical agitator, and the AgCl allowed to settle, leaving the supernatant liquid clear. If the normal solution is made up correctly, it will have precipitated just 1000 mgs. of silver, leaving 2 mgs. unprecipitated. One c.c. of decimal salt solution is now added to the bottle by means of one of the 10-c.c. pipettes or a burette, which, if the solution still contains Ag unprecipitated, gives rise to a white cloud of AgCl. The bottle is again shaken, the precipitate allowed to settle, and another c.c. of decimal salt solution added. If this fails to give a precipitate, then 100.1 c.c. of normal salt solution are equivalent to 1002 mgs. of silver (1 c.c. of decimal salt solution = 0.1 c.c. normal salt solution). If the second addition of decimal salt solution gives a precipitate, the shaking and settling are repeated, and a third and fourth, etc., addition made, until no further cloud appears. The assayer soon learns to judge by the density of the cloud whether only part of the c.c. has been used up. In this way he should be able to judge to the fourth of a c.c. or the half of a millieme. If the first addition of decimal salt solution fails to give a precipitate, the normal solution con- tains an excess of salt, and 2 c.c. of decimal silver solution are now added, one of which neutralizes or precipitates the 1 c.c. of decimal salt solution added, the other acting on the excess of salt in the solution. The decimal silver solution is added until no further cloud appears, in the same way as described for the decimal salt solution. In this way the exact strength of the normal salt solution is determined in duplicate. If it is incorrect to the extent of more than 2 points fineness either way (i.e., either strong or weak), it is corrected by the addition of either water or salt, and restandardized, and, when correct, a new decimal salt solution made up from it. Its strength is finally recorded on the bottle as follows: 100 c.c. = 1000 mgs. Ag, or whatever it may actually be. The Assay. It is evident from the preceding that the amount of bullion to be taken for assay must contain as nearly as possible 1000 mgs. Ag in order to make the titration with solution as short as possible, and avoid undue additions of the decimal solutions. For this reason the bullion on which the silver deter- mination is to be made is first assayed by the cupellation method, 180 A MANUAL OF FIRE ASSAYING or at least a preliminary assay, described under this method, is made, and from these data the amount of bullion containing 1000 mgs. of silver calculated. For instance, suppose the cupella- tion method shows the bullion to be 900 fine in silver, then 900 : ' 1000 :: 1000 : x. fineness : amt. of bullion :: silver : amt. of bullion, or 1111.11 mgs. bullion contains 1000 mgs. Ag. This amount of bullion is then weighed out in duplicate and dissolved in acid, placed in titration bottles, as described above, under " Standardi- zation of Solutions, " and titrated. The calculation for fineness is as follows: Suppose the strength of the normal solution is 100 c.c. = 1001 mgs. Ag, and that 99.8 c.c. of normal solution were used in the titration (100 c.c. normal salt, and 2 C.c. decimal silver) ; then 100 : 1001 :: 99.8 : x the x, or amount of silver in bullion, equaling 998.99 mgs.; and the fineness is 1111.11 : 998.99 :: 1000 : y the y, or fineness, equaling 899.1. The only metal interfering with the salt titration is mercury, which will be precipitated by the NaCl as Hg 2 Cl 2 ; the addition of 20 c.c. sodium acetate and a little free acetic acid to the assay will prevent the precipitation of the mercury. Mercury can be detected in the titration if the AgCl has not turned dark as the result of exposure to sunlight. Mercury will be found sometimes in mill bullions which have been retorted at too low a temperature. The assay and standardization of the solution should be carried out where there is no sun, and where light is not too strong. THE ASSAY OF GOLD BULLION FOR SILVER BY A WET METHOD. The accurate estimation of silver in bullions contain- ing a large proportion of gold is not all that can be desired by the ordinary fire method. The Gay-Lussac method is generally not applicable on account of the large amount of bullion that must be taken for a sample in order to get 1 gram of silver. The foUowing wet method 1 will yield good results. Take 0.5 gram of the bullion, fuse with 1.5 gm. of pure cadmium under a cover of potassium cyanide in a porcelain crucible in the flame of a blast lamp. Enough cyanide must be used to 1 E. H. Taylor, Australian Mining Standard, August 26, 1908, 235. Consult also J. E. Clennel. Eng. and Min. Jour., LXXXIII, 1099. THE ASSAY OF BULLION 181 cover the cadmium. Five minutes is sufficient to insure fusion. Allow to cool, place in stream of running water which will rapidly dissolve the cyanide and leave the alloy. Transfer this to a flask with 20 c.c. of water, add 40 c.c. of HNO 3 in installments of 10 c.c. each while boiling for one hour. Dilute to 150 c.c. and add 10 c.c. of ferric alum indicator and titrate with the standard solution of NH 4 CNS. This solution is made as follows: 1.6 grams of pure NH 4 CNS are dissolved in 1000 c.c. of distilled water. This is standardized against pure sil- ver foil dissolved in HNO 3 and diluted to 150 c.c. 1 c.c. of the solution equals approximately 4.483 parts of Ag per 1000 under the conditions described above. The indicator is a sat- urated solution of ferric alum. The appearance of the red color marks the end point. Copper in amounts of 100 parts per 1000 in the bullion does not interfere with the delicacy of the end point. In case the bullion is very high in gold the cadmium must be increased. The parted gold is recovered from the residues in the flasks. THE ASSAY OF GOLD BULLION. 1. Sampling. Bullion bars and retort sponge, as shipped to the United States assay offices and mints, is remelted into bars to make the deposit uniform. These are sampled by taking chips from diagonally opposite corners, each of which is rolled into a fillet and assayed by different assayers, who are required to check with each other within narrow- limits; if they do not, the bar is remelted, stirred thoroughly, and recast; then sampled again and assayed. If base bullion, or one which liquates seriously on cooling, is to be assayed, dip-samples are taken from the molten bullion by means of a small graphite ladle, and the sample granulated in warm water. Silver bullion is sampled in the same manner. 2. Preliminary Assay. This is made in the way described for silver bullion, except that in the assay of gold bullion no determination of silver is made by cupellation; but if this is to be determined, the mint wet method is used. Experienced assayers can judge the approximate fineness of gold bullion by the color, and add the proper amount of silver necessary to insure parting. In the San Francisco mint, 2 parts of Ag to 1 of Au are used. 1 The British royal mint formerly used 2.75 parts of Ag. to 1 of Au, 2 but now uses 2 to 1. More than 3 parts Ag 1 John W. Pack, "Assaying of Gold and Silver in U. S. Mint", in M in. and Sci. Press, LXXXVII, 317. 2 Rose, in Eng. and Min. Jour., LXXX, 492. 182 A MANUAL OF FIRE ASSAYING to 1 of Au should not be used, otherwise the "cornet" of gold is apt to break up. With less than 2 parts of Ag ; too much Ag is retained, although with continued boiling 1.75 parts Ag will part Au from Ag. 1 For the preliminary assay, 500 mgs. (1000 milliemes) are weighed out, silver added according to judg- FIG. 59. JEWELERS' ROLLS. ment to bring the ratio of silver to gold to 2 or 2.5 (allowing for silver in the alloy), and the bullion and silver wrapped in 10 grams sheet lead and cupelled at 850 C. The resultant bead is cleaned, weighed, flattened and rolled out in jeweler's rolls to a fillet of the approximate thickness of a visiting card. If some copper is present in the bullion, enough 1 Rose, "Metallurgy of Gold," p. 493. THE ASSAY OF BULLION 183 is retained by the gold bead to toughen it, and it can be easily rolled without cracking, if, between reductions by the rolls, the fillet is annealed at a dull-red heat. The presence of copper in the button aids in the total removal of lead during the cupellation. 1 The fillet is then again annealed and rolled into a spiral, called a "cornet," and parted in a parting flask. This is filled with 30 c.c. of HNO 3 sp. gr. 1.20, free from Cl, H 2 SO 4 , H 2 SO 3 , or any sulphide, and heated to boiling (or at least 90 C.) for 20 minutes. The acid is then decanted off, and the cornet washed carefully several times with hot distilled water by decantation. Then 30 c.c. of boiling nitric acid, sp. gr. 1.30, are added to the flask, and the cornet boiled again for 20 minutes, after which the acid is decanted, and the washing with hot water repeated. During the boiling, a parched pea added to the flask prevents bumping. The flask is now filled to the very top with cold distilled water, a suitably sized porcelain parting-cup placed over the mouth, fitting reasonably tight, and the flask inverted. The cornet will settle into the parting-cup, and the flask is then gently tipped to permit the water to escape, the water is decanted from the parting-cup, and the cornet gently dried. When dry, the cornet is transferred to a clay annealing cup, the cover is put on, and the cup is placed in the muffle, and the cornet annealed at a full-red heat. It is then weighed. The weight of the gold plus that of the added silver, subtracted from the weight of the cupelled bead, gives the approximate amount of silver in the assay. This added to the weight of the gold and subtracted from 500 mgs. (the weight of bullion taken) gives the approximate amount of base metal. If the amount of silver added to part the gold has raised the ratio of Ag to Au over 3 to 1, the gold will probably have broken up, or at least parts will have broken from the edges of the cornet; care must, in this case, be taken to collect all of it in the washing. If the results show that the ratio of Ag to Au has been less than 2 to 1, the cornet must be recupelled with 2.5 parts Ag and parted as described. The Assay. The final assay is made up from data obtained in the preliminary assay. Duplicates on 1000 milliemes are run, with a check assay made up in composition as near to that of the bullion as possible, as described for the cupellation assay of silver. In making up the check, proof gold and proof silver are Rose, "Refining Gold Bullion," in Trans. I. M. M., April 13, 1905. 184 A MANUAL OF FIRE ASSAYING used, and c.p. copper foil. The United States mints use various proof alloys in the making up of check assays. For the assay of fine gold bars (990 fineness and above), a proof alloy of 1000 gold, 2000 silver, and 30 parts copper is used. For coin riletal (900 parts fine), a proof alloy of gold 900 parts, silver 1800 parts, copper 100 parts is used. For the determination of base metal (the difference between the gold and silver, and the 500 mgs. taken for assay), a proof alloy of gold 900 parts, silver 90 parts, copper 10 parts is used. 1 In this last the gold need not be proof gold, but may be remelted cornets. It is to be noted that these proof alloys are made up on the assumption that 2 parts of Ag to 1 of gold are used in parting. The British mint uses a proof alloy, or trial plate, 916.6 fine in gold. For the assay of crude gold bullion, i.e., mill bullion, the proof alloy for fine gold bars is generally used. The amount of lead used in the cupellation is as follows: 2 TABLE XL. LEAD RATIO IN CUPELLATION Amount of gold per Amount of lead 1000 parts Ratio of lead to copper (base metal present) Milligrams Grams 916.6 8 96 to 1 866 9.15 68 to 1 770 14.75 64 to 1 666 16.00 48 to 1 546 17.50 38 to 1 333 18.00 27 to 1 To the duplicates of the 1000 milliemes of bullion, the proper amount of Ag is added, to bring the ratio of Ag to Au to 2 to 1, and then they are wrapped in the proper amount of c.p. sheet lead. The check is made up as indicated by the preliminary assay, and the three assays cupelled as described for the assay of silver bullion. The three beads are then treated and parted, as described for the preliminary assay. The two bullion assays should not differ by more than 0.25 part of a millieme. The 1 John Pack, ibid. 2 Rose. "Metallurgy of Gold." 1902, p. 494. THE ASSAY OF BULLION 185 correction as indicated by the check should then be applied, whether this be plus or minus. The difference between the fine . gold in the check and that obtained by the assay of the check is the surcharge, which is more definitely defined in Chapter XI, on " Errors in the Assay for Gold and Silver." This surcharge will usually amount to about for a bullion of about 700 to 800 fine; above that there will be a "plus surcharge," and below that a " minus surcharge." The plus surcharge will be subtracted and the minus surcharge added. THE PREPARATION OF PROOF GOLD. This is prepared by dissolving practically pure gold (cornets) in nitro-hydrochloric acid, permitting the solution, after some dilution, to stand for four days to allow AgCl to settle out. It is then decanted very care- fully by siphoning. The gold chloride solution is then evaporated almost to dryness, taken up with plenty of distilled water, a few c.c. of NaBr or KBr solution added, allowed to stand for some days, and again decanted by siphoning, after which operation it is slowly dropped from a burette into a beaker containing c.p. aluminium foil. When precipitation is complete, HC1 is added to dissolve the excess of Al, and the residual gold is washed thor- oughly with water by decantation, and then dried and melted into a bead in a fresh cupel (but not cupelled with Pb) . The gold is then rolled into a thin strip for use. 1 Proof silver is prepared by dissolving c.p. silver foil in HNO 3 , and then precipitating with HC1 after filtering. The AgCl is thoroughly washed with diluted HC1 and converted into metallic silver by Al in the presence of HC1, all Al being dissolved out. The washed silver is then fused in a porcelain crucible, and rolled into strips. 2 1 Consult also Rose, "Metallurgy of Gold," p. 10, and Pack, ibid. 2 John Pack, ibid. CHAPTER XIII THE ASSAY OF ORES AND ALLOYS CONTAINING PLAT- INUM, IRIDIUM, GOLD, SILVER, ETC. Materials containing some of the above elements are pre- sented to the assayer for determination in the shape of sands containing chiefly platinum, alloys and jewelers' sweeps, and, more rarely ores containing platinum in the form of the mineral sperrylite, etc. The assay for platinum and associated metals is a difficult one, due to the fact that in the parting of the precious metal beads, by acids, complex reactions take place, by which platinum, palladium, silver, etc., both go into solution and are retained in the residue, unless certain well established ratios of metals present are observed and the parting operation repeated several times. The alloys of platinum and silver have been most thor- oughly investigated in this connection. 1 When the alloy is more complex, i.e., contains also gold, palladium, iridium, rhodium, etc., the difficulties of the assay are increased; the data at present available are meager. Platinum nuggets from the Urals contain: 2 Pt, 60 to 86.5 per cent.; Fe, up to 19.5 per cent.; Ir, up to 5 per cent.; Rh, up to 4 per cent.; Pd, up to 2 per cent.; also Os, Ru, Cu, Au, and iri- dosmium. When material containing Au, Ag, Pt, Pd, Ir, Rh, Ru, Os, and IrOs is fused by the crucible assay or melted with lead, the Au, Ag, Pt, Pd, Ir, Rh, IrOs are collected by the lead and the Ru, and Os only partially so. If the resultant lead button is cupelled, the final bead will contain the Au, Ag, 3 Pt, Pd, Ir, Rh, IrOs, and a comparatively small portion of the Os and Ru, the most of these two metals being lost by oxidation. The presence of any considerable amounts of Os and Ru in the lead button, 1 Thompson and Miller, in Jour. Am. Chem. Soc., XXVIII, 1115. See this paper for other references. 2 Kemp, in Eng. and Min. Jour., LXXIII, 513 (Notes on Platinum and Associated Metals). 3 Exclusive of losses by absorption and volatilization. 186 THE ASSAY OF ORES AND ALLOYS 187 owing to the fact that they will not alloy readily, causes them to appear as a black scum or as spots on the bead, near the end of the cupellation. The presence of the platinum group of metals, raising the melting-point of the gold-silver alloy, renders neces- sary a high temperature of cupellation in order to remove lead. Even then, when the ratio of Ag to Pt, etc., is less than 5 to 1, lead will be retained in varying proportions at the cupellation temperature of gold bullion. To get rid of the lead, the propor- tion should be 10 to I. 1 The following points on the first cupella- tion of the lead buttons, resulting from the assay of material containing Pt, etc., will give the assayer an idea of what is present. When Pt alone, or with very little silver is present, the bead from the cupellation (at a comparatively high temperature) is rough, dull gray, flat, and contains lead. If more silver is present, but less than 2 parts of Ag to 1 of Pt, the beads are rough, flat, and have a crystalline surface. If more than 2 parts of Ag are present and not more than 15, the bead approaches more nearly the appearance of a normal silver bead, but has a more steely appearance and is flatter in proportion to the Pt, etc., contained. Beads containing more platinum than 1 in 16 will not blick or flash. 2 The effect on the appearance of the bead of Pd, Rh, Ir is similar to that of Pt, but not identical. Owing to the difficulty in alloying iridium, this, when present, is apt to be found at the bottom of the bead, in the shape of fine black crystalline particles. 3 THE ACTION OF ACID ON THE ALLOY BEADS. A great deal of literature exists on this point; but most of it is very conflict- ing; some facts, however, have been definitely established. Nitric Acid. In an alloy of Pt and Ag treated by HNO 3 , platinum goes into solution in various proportions, depending on the ratio of Ag to Pt, and probably to some extent on the strength of acid. It has been stated that when the ratio of Ag to Pt is 12 or 15 to 1, this solution of Pt is complete in one treatment, but this has been disproved by later investigation. 4 In order to accomplish the solution of Pt, the acid treatment 1 Sharwood, "Cupellation on Platinum Alloys, containing Ag and Au," in Jour. Soc. Chem. Ind., XXIII, No. 8. 2 Schiffner, in Min. Ind., VIII, 397. 3 Rose, "Metallurgy of Gold," p. 514. 4 Thompson and Miller, in Jour. Am. Chem. Soc., XXVIII. 115. 188 A MANUAL OF FIRE ASSAYING must be repeated at least once or twice, with a possible recupel- lation of the residue with silver before the second treatment. It is even then doubtful if all of the Pt can be dissolved. The Pt goes into solution in the nitric acid in colloidal form, giving a brown to blackish color to the solution. When gold is present in the silver-platinum alloy, the solubility of the Pt seems to be decreased, 1 unless the ratio of Pt to Au to Ag is 1:2:15, 2 when most, but not all, of the Pt and all the Ag go into solu- tion. Palladium goes into solution with nitric acid when at least 3 parts of Ag to 1 of Pd are present, 3 yielding an orange-colored solution; but double parting is necessary to insure complete so- lution. (This point is not sufficiently established. 4 ) The orange- colored solution indicates colloidal palladium. Iridium and Rhodium. Iridium present in the beads is unacted upon by HNO 3 and remains with the gold. 5 Rhodium is slightly dissolved, but most of it remains with the gold. Iridos- mium is not dissolved. Osmium is dissolved. Ruthenium is not dissolved. Sulphuric Acid. Platinum, alloyed with silver and gold, can be separated from the silver and remains with the gold, if concentrated sulphuric acid is used in parting. In order to insure thorough parting, at least 10 parts of silver to 1 part Pt and gold should be present, and double parting resorted to, otherwise silver will remain with the residue. 6 The parting in H 2 SO 4 leaves the Pt and gold in a very fine state of division (but not as a colloid), some of which is very apt to be lost in decanting, so that it is best to separate by filtering through an ashless filter. It is also to be noted that lead may be present in consequence of too low a cupellation temperature, in which case the residue should be treated with ammonium acetate, to remove lead sulphate. Palladium. In parting with H 2 SO 4 this goes into solution with the silver, giving an orange-colored solution. Whether this solution is complete, has not as yet been demonstrated. 7 Ir, IrOs, Rh, and Os and Ru in the bead are not dissolved. Nitro-Hydrochloric Acid. From the residue of the sulphuric acid parting, the Pt, Au, and any Pd left is dissolved by dilute 1 Sharwood, ibid. 2 Lodge, "Notes on Assaying," p. 215. 3 Rose, "Metallurgy of Gold," p. 514. 4 Lodge, "Notes on Assaying," pp. 218, 219. s Rose, ibid. 6 Thompson and Miller, ibid. 7 Lodge holds the contrary, p. 219. THE ASSAY OF ORES AND ALLOYS 189 aqua regia, 1 to 5, leaving Ir, IrOs, and Rh, and some Ru and Os, if present. This last residue, treated with strong aqua regia, removes Ir, leaving iridosmium and rhodium as a final residue. METHODS OF ASSAY. 1. Ores. Rich ores, carrying Ft, etc., in grains, present difficulty in sampling, inherent to any ore containing " metallics. " It is best to take from 30 to 50 grams of the sample and fuse it with 6 times its weight of lead in a crucible, fluxing the gangue. The lead is poured, and after cooling the slag is detached carefully, the lead platinum alloy being brittle, weighed and remelted under charcoal in order to insure a uniform alloy, and then granulated as fine as possible by pouring into a large volume of cold water from a considerable height. The resultant sample is then dried and is ready for assay. An amount containing approximately 200 mgs. Pt is weighed out and scorified with 50 grams Pb into a 20-gram button. If, in the low-grade ores, the Pt, etc., is present as grains, a weighed quantity is concentrated by panning and the concen- trates scorified with 20 to 25 times their weight of test lead, and the button treated according to method No. 1 or 2, as below. If the ore contains the rare metal in other form, crucible fusions are made on 1 assay ton, as with gold and silver ores, and if very low grade, the buttons from 4 to 5 fusions are scorified into one button, final duplicates being made as usual. The lead buttons are treated as below. 2. Alloys. An amount of drillings or filings (representing a true sample of the alloy),. containing, if possible, not to exceed 200 mgs. of Pt, etc., is weighed out and scorified with 80 grams of test lead, to a button of about 18 to 20 grams. The lead buttons are treated as outlined below. First Method. The lead button obtained by any of the foregoing methods is cupelled at a temperature of at least 900 C., or, better, 950 C., and the resultant bead examined. If, from the foregoing description of the appearances of a bead, it is thought that the ratio Ag to Pt, Au, etc., is less than 10 to 1, the button is removed, the necessary silver added to bring it up to the above ratio, recupelled with 5 to 8 grams of lead at a tem- perature of 900 C., and weighed. The bead is then flattened and rolled out into a cornet, if large and not too brittle, and parted with 15 c.c. H 2 S0 4 concentrated, boiling for 15 to 20 minutes. The acid is then decanted into a beaker and saved, 190 A MANUAL OF FIRE ASSAYING the residue re-treated with 5 c.c. more of acid for 10 minutes, and the residue and acid washed into the beaker containing the first acid. The acid is then diluted and the residue separated by nitration through a small ashless filter, and thor- oughly washed with hot water to insure removal of Ag 2 SO 4 . The filter-paper is dried and carefully transferred to a porcelain parting-cup or an annealing cup, and the carbon burnt off in the muffle. The annealed residue is brushed out on the scale pan of the bead balance and weighed. It consists of gold, plati- num, indium, iridosmium, rhodium, and possibly osmium and Ru (if any escaped oxidation during the cupellation) , and perhaps some palladium. Its color will be gray or black, if the rare metals are present to any extent. If not, the characteristic gold color will show. The palladium is largely in the filtrate. (It is questionable how complete this solution is. 1 ) If it has been unnecessary to add Ag to the cupellation to get the 10 to 1 ratio, the difference in weight between the original bead and the weight of the residue represents the Ag. If silver had to be added and the bead recupelled, the weight of the added silver plus that of the residue, subtracted from the weight of the recupelled bead, gives the silver. Allowance must, however, be made for con- siderable loss of silver as a result of high cupellation temperature. If accurate silver results are required, a duplicate assay on the material must be run, and the silver requisite to bring the ratio up to 10 to 1 is added at once to the lead button, one cupellation only being made. At the same time this is run, a check assay is run beside it, made up of the same weight of lead, and the proper weight of silver, i.e., the amount added to the first cupel- lation plus the amount approximately known to be in the assay. The loss in this will give the correction to be added to the assay for Ag. It may be desirable to determine Ag in the wet way. (See " The Assay of Silver Bullion. ") The residue is now wrapped in 8 to 10 grams of lead foil with at least 20 times its weight in silver and cupelled again at a high temperature. The bead, if large, is rolled out and heated to boiling in a mattrass or flask for 20 minutes with HNO 3 , sp. gr. 1.20, after which the acid is decanted into a beaker, and the treatment repeated with HNO 3 of 1.26 sp. gr. The residue, if finely divided, should now be filtered through an ashless filter and washed as already described. If not, the filtrate can be 1 Ricketts and Miller, in "Notes on Assaying," state that the Pd dissolves with the Ag. THE ASSAY OF ORES AND ALLOYS 191 decanted and the residue washed. The residue consists of Au, Ir and iridosmium, and some Rh and Ru. If there is a sus- picion that -any platinum, etc., remains, the residue must be re-treated with acid until of constant weight. The platinum is in the filtrate, which will be colored brown or black. The difference between the weights of the first and second residues is platinum, the result possibly being somewhat high if palladium is present in the material assayed. The second residue is now warmed in a mattrass with dilute aqua regia 1 (1 to 5) for 15 minutes. This dissolves the gold, some of the Ru and very little Rh, leaving the Ir, iridosmium and Rh, with some Ru. The residue is either filtered or decanted, as necessary, dried, annealed, and weighed. The difference in weight between the second and third residues represents gold, somewhat high, if the Ru has partly escaped oxidation and volatilization during cupellation. The gold can be recovered by precipitation with oxalic acid, as described in the second method. If the third residue is treated with strong aqua regia, and boiled, it dissolves out the iridium, leaving as a residue the iridosmium and most of the Rh. This is dried, annealed, and weighed, the difference in weight between the third and fourth residues representing iridium, and the weight of the fourth residue representing iridosmium and Rh. The method determines Ag, Pt, Au, Ir, and iridosmium plus Rh. The probable errors in the determination have been pointed out. Palladium can be satisfactorily determined only by wet analysis. Second Method. 2 Take the lead button from the ore or alloy assay, and scorify at a high heat, with additional test lead, if necessary, to a weight of 8 to 10 grams. It should contain less than 5 per cent. Pt. etc., in order to be malleable. Roll out the button into a long thin fillet and place in a large beaker with 200 c.c. of HNO 3 , sp. gr. 1.08, 3 and heat until all action ceases. Filter through a small ashless filter and wash the residue with hot water. Dry the residue and paper, transfer to a large-size parting-cup and ignite in the muffle, to burn off the carbon, and oxidize any Pb not dissolved. Then heat to boiling in the cup with HN0 3 , 1.08 sp. gr., decant, wash thoroughly with hot water, dry, anneal, and weigh the residue. This consists of Au, Pt, Ir, iridosmium, and most of the Rh, as well as the Ru and Os which 1 Concentrated aqua regia is 1 part HNOj, sp. gr. 1.42, and 3 parts HC1, sp. gr. 1.20. 3 E. H. Miller, in School of Mines Quart., XVII, 26. 3 81 parts distilled H 2 O to 19 parts HNOj cone. (sp. gr. 1.42). 192 A MANUAL OF FIRE ASSAYING escaped oxidation and volatilization during the scorification. The filtrate contains the Ag and Pd and a little of the Rh. Replace the residue in the capsule and warm (not boil) with dilute aqua regia (1 to 5) for 10 minutes. This dissolves the Au and Pt. Decant the solution into a small beaker, wash the residue, dry, anneal, and weigh. The second residue consists of Ir, IrOs, Rh, and a little Os and Ru. This residue is boiled with strong aqua regia, which dissolves the Ir and some Os and Ru, and leaves in the third residue the IrOs and Rh, with a little Os and Ru. This is washed, decanted, and weighed as before. The nitrate from the treatment of the first residue, which con- tains the gold, is evaporated just to dryness, but not baked, so as to prevent reduction of the gold chloride, taken up with dis- tilled water and a drop of HC1, and the gold in it precipitated by warming with crystals of oxalic acid for a half hour, filtering, and drying the yellow coherent precipitate of gold. This is transferred, filter-paper and all, to a piece of sheet lead, silver added to the weight of 3 times the gold present, approximately, and cupelled, the bead being parted in HNO 3 as usual and the gold annealed and weighed. The weight of the gold, subtracted from the difference in weight between the first and second residues is the platinum. This last may also be estimated by destroying the oxalic acid in the filtrate from the separation of gold, and pre- cipitating as (NH 4 ) 2 PtCle. 1 It is to be noted that, by the assay as outlined, neither osmium nor ruthenium can be determined, owing to their volatility during part of the operation; that palladium cannot be readily deter- mined, owing to its varying solubility; and that when rhodium or the above metals are present in any appreciable quantity, some of the results obtained are liable to error. Rhodium, osmium, and ruthenium are among the rarer of the group, and are frequently absent. The methods outlined will serve to determine reasonably well platinum, gold, silver, iridium, and iridosmium plus rhodium. When the other elements of the group are present, wet methods, not within the scope of this book, must be resorted to. In the ordinary assay, as carried out for gold and silver, platinum and palladium may escape the assayer if present in only small quantities, for obvious reasons. Parting in sulphuric acid is therefore necessary to determine whether they are present. 2 1 Crookes, "Select Methods." 3 An orange-colored solution indicates palladium. CHAPTER XIV THE ASSAY OF TIN, MERCURY, LEAD, BISMUTH AND ANTIMONY The assay of ores for base metal by fusion is still carried out in practice, especially for lead and tin. The fire assay gives, not the correct metal content, but the yield obtainable in smelting, although in metallurgic operations the yield may be greater or less. The smelter, therefore, purchases lead, tin, and copper ores on the basis of the "dry" or fire assay. The fire assay of copper is practically no longer in use, except in part of the Lake Superior district, on metallic copper concentrates, and in pur- chasing copper ores the assay is made by the standard electro- lytic method, or a volumetric method, and a percentage of from 1 to 1.5 deducted to indicate dry assay. The usual deduc- tion is 1.3 per cent. Thus the dry assay of copper on an ore is equivalent to the percentage obtained by the electrolytic method less 1.3 per cent. While wet methods, with a deduction, will in all probability be employed eventually for all lead ores, as it is now for impure lead ores, pure lead ores are still assayed by the fire method. Tin ores are almost invariably assayed by the fire method, as the wet analysis of tin is long and tedious. THE ASSAY OF TIN ORES. The fire assay of tin ores is appli- cable only to those ores in which tin exists as cassiterite, the oxide (Sn0 2 ). The chief reasons for inaccuracies in the fire assay of tin are: 1. Some of the tin, reduced in the assay from the oxide, is apt to be volatilized at the temperatures necessarily employed. 2. Metallic tin may be slagged by alkaline carbonates used in some of the methods of assay, forming stannates. 3. Foreign metals present in the ore are apt to be reduced and enter the button. 4. Sulphides present carry tin into the slag. If sulphates are present, they are reduced to sulphides. 5. Silica and silicates, always present in the ore, even after very careful concentration, carry tin into the slag, as silicate, 13 193 194 A MANUAL OF FIRE ASSAYING while the SnO 2 passes through the lower stage of oxidation in being reduced to metallic tin, 6. The cassiterite, before reduction, is apt to combine with basic fluxes present in the assay, and be carried into the slag as stannates. From this, therefore, it is evident that the fire assay for tin is only an approximation, although in many cases a very close one. If the result on a tin ore by the fire method checks that of the standard wet method (the modified Rose method 1 ), it is to be ascribed to a balancing of errors, due to the presence of other metals in the ore, which have been reduced into the tin button. Preparation of the Ore for Assay. It is essential to remove all the gangue of the ore and have for the assay nothing but the cassiterite, as far as this is possible. The ore is roughly crushed on a buck board and put through a 40-mesh screen, crushings and screenings succeeding each other at frequent intervals in order to avoid the "sliming" of the cassiterite. If the ore is low-grade, i.e., below 2 per cent. Sn, 1000 grams of the crushed ore is weighed out and carefully panned in a gold pan, the first pannings being saved for repanning. The ore is concentrated just as much as possible without incurring loss of cassiterite. The concentrates from the repanning of the tailings of the first treatment are added to the main lot of con- centrates. Some or all of these will, unless the ore is very pure, contain probably garnets, feldspar, tourmaline, magnetite, zircons, wolframite, columbite, sulphides, quartz, etc. The concentrates are carefully transferred to a porcelain dish, dried, and roasted at a bright-red heat in order to decompose sulphides and sulphates. While the concentrates are still red-hot, they are transferred into a beaker containing water in order to make garnet and other silicates soluble (all except uvarovite), and after decanting water, treated with nitro-hydrochloric acid to remove most of the contaminating minerals, except quartz, wolframite, and some garnet. The concentrates are then fil- tered off and dried. If quartz is present, this can be removed by transferring the filtered concentrates to a platinum dish and treating with HF. This, however, will rarely be necessary. The concentrates are then crushed in an agate mortar to pass a 100- mesh screen and treated as described below. The Assay. The two best methods for assay are the cyanide Hofman, "The Dry Assay of Tin Ores," in Trans. A. I. M. E., XVIII, 1. THE ASSAY OF TIN, MERCURY, LEAD, ETC. 195 fusion and the German method, with black flux substitute. Of these two, the cyanide fusion is generally to be preferred, as any minerals still left in the cassiterite have less influence on the assay, and the loss of tin by volatilization is reduced to a mini- mum, on account of the low temperature employed. The Cyanide Fusion. 1 It is essential to use only the purest cyanide obtainable the best sodium or potassium cyanide on the market for use in the cyanide process. Such impurities as K 2 CO 3 , sulphates, and sulphides in cyanide cause serious losses in the assay. The best alkaline cyanide to use is sodium cyanide, which may readily be procured at the present time. Some of the ordinary commercial cyanide known as "potassium cyanide" fuses at such a low temperature that the concentrates sink to the bottom of the crucible before reduction, and when reduction finally takes place the little globules of tin are found to be very difficult to collect. In order that the fusion may be successful, it is essential to follow directions closely. It is best to use 10 grams of concentrates, or an amount near that; usually the amount of concentrates obtained from the concentration of the ore approximates this if the proper amount of ore is chosen for concentration. Two grams of powdered cyanide are firmly tamped into a 20-gram crucible, the concentrates are mixed with 30 grams more of cyanide, placed in the crucible, and covered with 5 grams more. The crucibles are placed in the muffle at a full-red heat (750 C.), and are kept at this temper- ature for about 15 to 20 minutes. The charge will become very liquid, and will be a brown-red. The temperature should not be so high as to cause the cyanide to boil and evolve heavy fumes. It may, however, be kept too low, in which case the chemical reactions will not complete themselves and the tin will fail to collect into a button. If the concentrates still con- tain some foreign minerals, the fusion takes longer than 20 minutes. The crucibles are then withdrawn, cooled, and the button recovered by breaking the crucible. There will be two distinct slags, the lower one, surrounding the button, usually light green, amorphous and subtranslucent, and the upper one, or fused cyanide, opaque, milk-white and coarsely granular, soluble in water. The tin button should be white and soft; if not, it contains foreign metals. The German Method. The German method is based on the i Hofman, ibid. 196 * A MANUAL OF FIRE ASSAYING fusion of the cassiterite concentrates with charcoal and black flux substitute, which has the composition, 2 parts K 2 CO 3 , 1 part flour. Five grams of the concentrates are intimately mixed with 1 gram of pure wood charcoal and put into a No. D lead crucible or an ordinary 20-gram crucible. On top of this are placed 15 grams of black flux substitute, with which 1.25 grams borax glass have been mixed. Finally a pure salt cover is added, and a piece of charcoal, the crucible covered with a clay cover, placed in the muffle, and heated at a moderate heat until boiling of the charge has ceased, and then for one-half to three-quarters of an hour more at a white heat. The crucible is then removed from the muffle, allowed to cool, and broken for the tin button. This should be white and soft, as in the cyanide fusion. During the fusion, as the temperature rises, the charcoal reduces the stannic oxide to metallic tin, while any ferric oxide is reduced to ferrous oxide, if the heating is gradual, and is taken up by the slag. As the temperature rises, the flour in the black flux substitute partially decomposes, liberating carbon through- out the charge, which, as fusion takes place, prevents any stannic oxide not as yet reduced from uniting with the alkali of the flux. The slag, after cooling, should be crushed and panned for any prills of tin which have not entered the button. These are weighed and added to the weight of the button. Results Obtainable. Black Hills cassiterite concentrates, roasted, quenched, and treated with nitro-hydrochloric acid. 1 Wet method of Rose-Chauvenet with K 2 CO 3 =67.84 per cent. Sn German method = 67 . 58 per cent. Sn Cyanide method =67.49 per cent. Sn Stream tin from Durango, Mexico, 2 Wet method (Rose) =65.62 per cent. Sn German method =63 . 92 per cent. Sn Cyanide method = 65 . 19 per cent. Sn It is to be noted that while the dry methods approach very closely to the wet analysis, which gives the actual tin in the ore, the dry assay results are due more or less to a balancing of errors. Frequently dry assays will give higher results than the analysis; this is due usually to reduced iron. 1 Hofman, ibid. 2 E. H. Miller. "The Assay of Tin Ores," in "School of Mines Quart.," XIII, No. 4. THE ASSAY OF TIN, MERCURY, LEAD, ETC. 197 Of the influence of foreign minerals left in the cassiterite concentrates, quartz has the worst, causing heavy losses. Feld- spar and tourmaline have similar effect, but not to so marked a degree. Mica and garnet give high results, due to the reduction of iron, although tin is lost in the slag. Columbite acts in a similar manner. With the German method the result is much more seriously affected by these impurities than with the cyanide fusion. 1 THE ASSAY OF MERCURY. Mercury occurs in ores chiefly as cinnabar (HgS), and may with accuracy be determined by Chism's method. 2 For low-grade ores, the method is especially satisfactory, and has the advantage of being rapid and short. It is based on the fact that mercury is distilled from HgS, etc., in the presence of iron filings, and can be caught on silver-foil. The difference in weight between the mercury-impregnated silver- foil and the foil before the assay gives the mercury. The appa- ratus required is as follows: 1. A small ring-stand. 2. A fire-clay annealing cup (No. B or C). 3. A piece of . carefully annealed silver-foil 1.5 in. square, which is fitted and bent down to make a reasonably tight cover for the annealing cup. 4. A flat silver or copper dish, holding 20 to 25 c.c. of water. A silver crucible may be used in place of this. 5. A piece of asbestos board, 4 in. square and about 0.20 in. thick, in the center of which a circular hole has been carefully cut, into which the annealing cup will fit so as to project about 0.5 in. below the bottom of the board. 6. A small alcohol lamp, of about 60 c.c. capacity. 7. A wash-bottle with cold water, and a glass tube for a siphon. The silver-foil is carefully fitted over the top of the annealing cup, the edges being bent down so as to make a close- fitting cover and prevent the escape of mercurial vapor. The silver dish should be polished on the bottom, and be in close contact with the foil, so that the cooling effect of the water will be fully transmitted. The Assay. For low-grade ores from 0.5 to 1 gram is taken and mixed with from 30 to 50 parts of iron filings. These filings 1 Hofman, ibid. 2 R. E. Chism, in Trans. A. I. M. E., XXVIII, 444. Consult also, G. A. James, Eng. and Min. Jour., XC, 800 and W. W. Whitton, Calif. Tech. Jour., Sept., 1904; Min. Ind., XVII, 751. 198 A MANUAL OF FIRE ASSAYING should all pass a 40-mesh screen. A select lot of filings are best digested with alcohol for some time to remove oil and grease, then heated in a muffle to a dull-red heat for 10 minutes, cooled, and stored in a tight bottle. It is essential to have the filings free from oil and grease, else this will be deposited on the silver- foil with the mercury. The amount of mercury in the ore should not be so great as to cause too heavy a coat on the silver-foil. For high-grade ores, not more than 0.1 to 0.2 gram should be used. Very small amounts of mercury can be detected by this method. The ore, mixed with filings, is placed in the annealing cup, which is set into the asbestos board on the ring-stand, the silver- foil weighed accurately, after igniting, to within 0.1 mg., and fitted to the cup, and the silver dish, filled with cold water, placed on the foil. The alcohol flame is then allowed to play just on the bottom of the cup, but not to spread around the sides. The flame should be about 1.25 in. high and is best shielded by a screen to steady it. The bottom of the crucible should not become more than a dull red, otherwise mercury will escape condensation. The time of heating should be from 10 to 15 minutes. It is best to heat for about 10 minutes, then cool, and reheat for 3 to 5 minutes. Longer heating than this causes loss of mercury. The degree and time of heat are very important. During the heating the water in the dish should be replaced once or twice. It can easily be removed by a bent tube that has been filled with water, acting as a siphon. While the warm water is being removed, cold water is added from a wash-bottle. After the proper heating, the alcohol lamp is removed, the assay allowed to cool somewhat, the silver dish removed, and the silver-foil with the mercury transferred by forceps to a desiccator and then weighed. The difference in the weight of the foil after and before the assay is the weight of the mercury, from which the percentage is calculated. The foil can be used again after driving off the Hg at a red heat in the muffle, or with a Bunsen burner. A piece of foil can be used about six times. It should be weighed before each assay. The method also serves as a very sensitive and easily applied qualitative test on ores. The following figures will serve to show the accuracy of the method :* 1 G. N. Bachelder, in "School of Mines Quart.," XXIII, 98. THE ASSAY OF TIN, MERCURY, LEAD ; ETC. 199 BY ELECTROLYSIS FROM CYANIDE SOLUTION Ore No. 1 12.37 per cent. Ore No. 2 67.26 per cent. BY CHISM'S METHOD 12.44 per cent. 67 . 23 per cent. The accompanying illustration (Fig. 60) shows the apparatus employed. THE ASSAY OF LEAD ORES. The fire assay of lead ores will probably pass out of use in time, just as the fire assay of copper has done. At the present time it is still largely used, although for complex ores containing much copper or bismuth, or antimony with the lead, it is not in vogue. It is, however, still the criterion FIG. 60. APPARATUS REQUIRED FOR THE MERCURY ASSAY. in the purchase of pure sulphide and oxidized lead ores, and also such complex ores as furnished by the Leadville, Colorado, district. Unoxidized ores of this type contain pyrite, blende, galena, some little chalcopyrite and gangue. Oxidized ores contain cerrusite, anglesite, calamine, limonite, etc., and gangue. The object of the assay is to bring the lead of these ores down into a button, free from other base metals, such as Cu, Zn, Bi,. Sb, Fe, and free also from S and As. The loss of lead by volatili- zation and slagging and the reduction of base metals should be kept to a minimum. As already stated, this is a difficult thing 200 A MAXUAL OF FIRE ASSAYING to do; so that pure ores will invariably give low results, and impure ones high. There are three methods of assay, differing in the flux used; (1) the lead flux method; (2) the soda-argol method; (3) the cyanide fusion. Of these, the lead flux method is chiefly used throughout the West. The soda-argol method is a good one on ores not basic. The cyanide method is only applicable to pure ores. With impure ores it tends to reduce other base metals, due to its powerful reducing action. Various mixtures of lead flux are used, of which three are made up as follows: No. 1 No. 2 No. 3 4 parts NaHCO s 2 parts NaHCO 3 6.5 parts NaHCO 3 4 parts K 2 CO 3 2 parts K 2 CO 3 5 parts K 2 CO 3 2 parts flour 1 part flour 2 . 5 parts flour 1 part borax glass 1 part borax glass 2 . 5 parts borax glass Flux No. 3 is probably the best for most purposes, as deter- mined on a series of ores, the results with it being slightly higher. 1 For assay, 10 grams of ore (100-mesh fine) are mixed with 30 grams of flux, placed in a No. 6 or D crucible, or in a 20-gram crucible, covered with 8 grams more of flux, and put into the muffle at a low heat, \vhich is then raised to a light yellow (1080 C.). The fusion should take about 30 to 35 minutes. Nails are added to the charge., two tenpenny nails for heavy sulphides, one for light sulphides or oxidized ores. When the charge is taken from the muffle, the nails are removed from the crucible by a pair of short hand tongs, care being taken to wash off all adhering lead globules. The crucible is then shaken and tapped thoroughly, and poured. The lead buttons are cleaned by hammering and then weighed. The percentage is obtained by multiplying by 10. The reactions in the crucible are as follows : 7PbS + 4K 2 C0 3 - 4Pb + 3(K 2 S, PbS) + K 2 SO 4 + 4CO 2 K,S, PbS + Fe = Pb + K 2 S + FeS The carbon liberated in finely divided particles from the flour on heating reduces any lead oxides or carbonates in the ore, while the iron reduces lead from its sulphides and sulphates. The assay should check (in triplicate) within 0.5 per cent. >McElvenny and Izett, in "The Chemical and Fire Methods of Determining Lead Ores," Min Rep., XL VIII, 26. THE ASSAY OF TIN, MERCURY, LEAD, ETC. 201 The soda-argol method uses the following flux: NaHCO 3 6 parts Argol 1 part For 10 grams of ore, 35 grams of flux are taken, with a light flux cover. The fusion is performed as described for the lead flux method. The method is good on ores containing some silica, but not on basic ores or pure galenas, as all acid is lacking in the flux. A borax glass cover is best where the method is employed on basic ores. In the cyanide method, pure cyanide should be used, and the temperature should be kept much lower than for the other two methods. For the regulation of temperature, reference is made to the assay of tin by the cyanide fusion. For the fusion, 10 grams of ore are mixed with 35 grams cyanide, and a light cyanide cover used. Concerning the accuracy of the method the following figures are appended: 1 Fire assay (lead flux) Gravimetric (PbSOJ Per cent. Per cent. 1. Galena 76 78.68 2. Galena 3. Cerrusite 37 9 37.40 10.60 4. Pyrite, Sphalerite, Galena 5. Galena and Stibnite 6. Cerrusite 24.7 28.7 37.8 18.46 27.25 38.60 THE ASSAY OF ANTIMONY AND BISMUTH ORES. For accurate and satisfactory determinations on these ores, wet methods must be resorted to. Antimony occurs chiefly as the sulphide, stibnite, although the oxides and some native metal are found as ore. Bismuth as an ore occurs chiefly as the native metal, but is found also in combination with oxygen, sulphur, etc. For the assay, the following charge is best: Ore 10 grams Cyanide 40 to 50 grams Cover of cyanide. Fuse at a full red heat, as given for tin, for 30 minutes. The resultant buttons are brittle and cannot be hammered. 1 Determination of Lead in Ores, I. T. Bull, School of Mines Quart., XXII, 348. APPENDIX 1 sit I HI & i & I s s 3 (N 0) US -* rt N 00 N "* S? I- ^ c^ 1 ! E S 1 O CO h- M< ** CO O - o o o o 2 | | | 1 ?! 1 111 > < . ^ c-i |, 1 S s 1 8 11 iH CO rt S l ' 1 CO ^ I 1 11 1 i I 1 10 I - 2 s. ; S f 1 1 8 CO sis 1 * a i a -l CO ?, S S S