EXCHANGE BOBI'IZWIW 'A "M ' THE JOHNS HOPKINS UNIVERSITY The Absorption Coefficient of Solutions of Cobalt Chloride in Water and Various Alcohols for Mono- chromatic Radiation DISSERTATION SUBMITTED TO THE BOARD OF UNIVERSITY STUDIES OF THE JOHNS HOPKINS UNIVERSITY IN CONFORMITY WITH THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY BY JOHN FOSTER HUTCHINSON June 4, 1916 EASTON, PA.: ESCHENBACH PRINTING Co. 1916 THE JOHNS HOPKINS UNIVERSITY The Absorption Coefficient of Solutions of Cobalt Chloride in Water and Various Alcohols for Mono- chromatic Radiation DISSERTATION SUBMITTED TO THE BOARD OF UNIVERSITY STUDIES OF THE JOHNS HOPKINS UNIVERSITY IN CONFORMITY WITH THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY BY JOHN FOSTER HUTCHINSON June 4, 1916 EASTON, PA.: ESCHENBACH PRINTING Co. 1916 The Absorption Coefficient of Solutions of Cobalt Chloride in Water and Various Alcohols for Monochromatic Radiation 1 Experiments have shown that in the case of certain solutions the absorption of monochromatic radiation may be represented by the formula I = I X 10' (i) where I is the original intensity of the radiation. I is the intensity of the radiation after passing through a layer of solution of thickness t millimeters, and a is a constant for the solution in question called the absorption coefficient of the solu- tion for the specified frequency of radiation. Experiments have also shown that different values of a are obtained if there is any change in : (a) the nature of the solvent or of the dissolved substance, (6) the concentration of the solution, (c) the temperature, (d) the wave-length of the radiation, etc. To solve the problem of light absorption in solutions it is necessary to determine the explicit form of the relation be- tween the absorption coefficient a, and the quantities of which it is a function. At the present time our knowledge is far too meagre to indicate more than a qualitative idea of the nature of this relation. In the present investigation a has been measured in those regions of the spectrum where the pure solvents possess appreciable absorption. It is assumed that the total ab- sorption of the solution is the sum of two parts, the first being the absorption due to the presence of the salt, the second being the absorption due to the pure solvent. In calculating this 1 This is a report bn part of an investigation carried out with the aid of a grant from the Carnegie Institution 'of Washington. 371461 second part, it is assumed that the absorption due to the solvent is the same as it would be if there were no dissolved salt present. We, therefore, write a = Ac + a where a is the absorption coefficient for the pure solvent, c is the concentration in gram-molecules of salt per liter of solution, and A is called the molecular absorption coefficient of the salt in the solution. From this relation we get -.(2) The present investigation consisted of a systematic and thorough study of the absorption coefficient a. This quantity has been measured at intervals of 2o/zju to 4Ojuju throughout the region of the spectrum from 600 juju to 1300/1^ for many solutions. The work has been restricted to a study of in- organic salts in aqueous and alcoholic solution. All the measurements have been carried out with solutions at room temperature. The values of a, when plotted as ordinates against the corresponding wave-lengths as abscissas, form the absorption curve. For each salt a series of solutions varying in concentration from saturation to moderate dilution were prepared, and the absorption curve has been drawn for each solution. From the measured values of a and a , and from the known value of c, A has been calculated for each wave- length by means of formula (2). The values of A for a given wave-length have been plotted as ordinates against the corre- sponding values of c as abscissas. The curves thus formed will be referred to as the A-c curves. If experiments with any solution show that the relation between a and c is a linear one, and therefore that A in formula (2) is a constant, the K-c curves for such a solution are straight lines parallel to the axis of abscissas. Previous researches 1 2 have shown that in general the K-c curves are not straight lines parallel to the axis of abscissas, or, in other 1 Sammlung chemischer und chemisch-technischer Vortrage, 9, i (1904). 2 Proc. Roy. Soc. Edinburgh, 33, 156 (1912-13). words, that A is not a constant. 1 It was the purpose of the present investigation to determine the form of the A-c curves. The chlorides, nitrates and sulphates of cobalt, nickel, and chromium, and a few other salts, have been studied in this investigation. This paper is a report on the results of the work on a single salt, cobalt chloride. Apparatus The apparatus used for determining the light absorption coefficient has been developed by previous workers in this laboratory. 2 The arrangement of the ap- paratus is shown in Fig. i. The light from a Nernst glower, g, run at no volts on 0.8 am- pere direct current from a con- stant potential storage battery, s was rendered parallel by a ;::[-- 3 lens, fc, 3.8 cm indiameter and with a focal length of 20 cm. The light after passing through cell k' was focussed on the slit A of the spectrograph by a second lens, fc, 3.8 cm in diam- eter and with a focal length of 20 cm. A shutter, s, was placed between the glower g and lens fc, by means of which the light could be turned on and off. The optical system 1, A Fig. i thus far described, consisting of the glower, the two lenses, k and fc, and the cells, was held by a solid metal frame work, and was perpendicular to the plane of the drawing in Fig. i . The light after passing through the lens fc was reflected on to slit 1 Beer has stated (Pogg. Ann., 86, 78-88 (1882)) as a result of his ex- periments on aqueous solutions of inorganic salts that A is a constant with respect to c. This "law" of Beer has been shown to be the exception rather than the rule, and therefore in this paper no reference has been made to "Beer's Law." 2 Carnegie Inst. of Wash., Pub. 190 and 230. A by a right-angle glass prism (not shown in Fig. i) close to slit A. The spectrograph consisted of the Littrow mounting of a plane grating. The grating had a ruled area 6 cm by 7.5 cm and was ruled 15000 lines to the inch. The cone of light from slit A was reflected by a right-angle glass prism through the large achromatic lens / 3 , 10 cm in diameter and with a focal length of 75 cm. The spectrum was brought to a focus at slit B. The grating possessed a bright first order, and this first order spectrum was used throughout the present work. The dispersion was such that with slit B one millimeter wide a "monochromatic beam" of light containing a wave-length range of 20 A, or 2ju/z, passed through. In this work both slit A and slit B were always one millimeter in width. The correction for finite slit-width was negligible. The grat- ing was mounted on a turn-table which was rotated from the outside by a worm-screw, thus causing various wave- lengths to pass through slit B. The monochromatic beam of light from slit B was focussed by a lens, / 4 , 35 cm in diameter and with a focal length of 6 cm on the junction of the radio- micrometer. The solution for which a was to be determined was placed in two glass cells exactly alike in all respects, except that one cell was thin and the other thick. The energy, I, of the mono- chromatic beam of light after passing through the thin cell containing a thickness, h, of the solution, and the energy, I', after passing through the thick cell containing a thickness, h', of solution, were measured in arbitrary units, i. e., deflections of the radiomicrometer. If the initial intensity, I , of the light falling on the cell was the same in each case I = I X io~ ah ah> I- d , , or a = ylogj, (3) where d and d' are the deflections produced by I and I', respectively, and / is the difference in thickness in millimeters of the two cells. This method eliminated all corrections for reflections from the glass surfaces, and thus gave a directly. Results Cobalt Chloride in Water (Figs. 2, 2 a) Twenty- three solu- tions were prepared varying in concentration from c = 3.23 to 600 700 .800 900 1000 1100 Fig. 2< c o.i. The more concentrated solutions were quite stable and showed no signs of decomposition even after standing in the bottles for several days. In the more dilute solutions, however, there appeared a flocculent precipitate which in- creased their absorption materially. On this account a second set of solutions whose concentrations varied from c = i.o to c = o.i were prepared, and the results of the measurements of these appear in the figure. The absorption curves include the long- wave side of the yellow-green cobalt absorption band and the short-wave side of the infra-red band, and show the region of transmission between the two bands. The minimum of absorption is at 7 6 4 MM- The A-c curves for wave-lengths 605 ju JJL to 7641*1*, in- clusive, which lie on the edge of the yellow-green band, show that A decreases in a marked manner with dilution, reaching a minimum value at about c = i.o. Below c = i.o A ex- periences a slight increase. The A-c curves for those wave-lengths in the region of transparency, from 842/4/4 to 979 MM> are straight lines parallel to the axis of abscissas, showing that in this region A is con- stant for all concentrations. For wave-lengths greater than 979 MM> which lie on the edge of the infra-red band, A is a TABLE I A for Cobalt Chloride in Water TTT_. T 1^i*i rt~4-l- c = 0.65 c = 3.10 w ave-iengtn Houston From this work Houston From this work 645 MM 0.041 0.0340 684 . 024 0.0232 O.200 720 0.031 0.0123 O.04I 0.0330 750 0.028 0.0090 0.037 0.0150 794 0.028 0.0109 0.016 0.0138 850 0.028 0.0147 0.018 0.0165 910 0.028 0.0175 0.029 0.0198 980 0.040 0.0275 0.038 1070 0.070 0.0762 0.074 constant within the error of experiment. The two band edges in question are thus seen to behave quite differently as dilution proceeds. Houston 1 has drawn the absorption curves for two solutions of cobalt chloride in water; and Table I shows the comparison between Houston's values and the values found in this work. The agreement between Houston's values and the values of A found in the present investigation is far from satis- factory. However, both sets indicate similar changes in A with c. Cobalt Chloride in Methyl Alcohol. (Figs, j, Seven solutions were prepared varying in concentration from c = 0.7 to c = o.i. The solutions appeared to keep very well, and no such precipitate was formed as was noticed in the water solutions. The absorption curves show that the character of the absorption of the alcohol solutions was quite different from that of the water solutions, the absorption curve for the alcohol solution being shifted toward the red, 1 Proc. Roy. Soc. Edinburgh, 31, 521 (1910-11). 8 .1400 .1500 200 .1100 A .1000 .0900 m w 744 .0700 .0600 7 1056 0500 .0400 .0500 .oeoo .0100 1018 so that the minimum of absorption was now found at the shift thus amounting to about 80 MM- The shift toward the red of the edge of the band in the green was suffi- cient to make this band ab- sorb nearly all of the visible red light. (Instead of speak- ing of the ' 'shift of a band," some have preferred to speak of the bands in the different solvents as entirely different bands.) As a consequence QSOO the more concentrated solu- tions appeared a deeppurple, becoming more and more pink as the dilution in- creased. The A-c curve for 744 MM shows that A decreases by a large amount with dilution, dropping from 0.128 for c 0.7 to 0.080 for e = o.i. This is the only A-c curve which has been plotted for a wave-length lying on the edge of the red-yellow absorp- tion band, for this edge is tremendously sharp compared to the edge of the analogous band of the water solu- tion. The A-c curves for the region of transmission, 764 MM to 920 MM, and for the edge of the infra-red band 920 MM to 1 1 34 MM, show that A for these regions of the spec- trum remains approximately constant for all concentrations. Cobalt Chloride in Ethyl Alcohol. (Fig. 4) Four solutions were prepared varying in concentration from c = 0.4 to c = o. i . A month later a second series of more dilute solutions for which c was 0.08, 0.06, 0.05, 0.04, 0.03, 0.02, o.oi, 0.005, were prepared, and their absorption curves drawn only in the regions of moderate absorption, from 1056 ju/i to 1134/zju, and for 724/4 ju. In the other regions they either absorbed too much, or too little, so that no con- fidence could be placed in the values of A. 800 900 1000 1100 1200 1300/iu Fig. 4 The absorption curves for the ethyl alcohol solutions are similar in then- general character to those for methyl alcohol. The minimum of absorption occurs in the same place, at 842/1/1, and the steepness of the edge of the bands is much the same. The ethyl alcohol solutions were of a pure deep blue in the higher concentrations, becoming greenish blue as dilu- tion increased. 10 The A-c curves for 724 MM and 744 MM show that A de- creases with dilution, and the decrease in this case is far greater than in the case of methyl alcohol. For wave-lengths 764 MM to 979 MM m the region of transmission, A is fairly constant. For the region on the edge of the infra-red band, 1018 to 1134, the A.-C curves show that A increases with dilution. These last mentioned curves illustrate the magnitude of the error in the determination of A in the case of very dilute solutions. Cobalt Chloride in Propyl Alcohol. (Fig. 5) Eight solutions were prepared varying in concentration from c = 0.434 to c = o.io. The character of the absorption II curves is the same as that of the ethyl alcohol solutions, the minimum of absorption occurring again at 842ju/z, and the steepness of the edges of the bands being similar. The propyl alcohol solutions were also deep blue, becoming a greenish blue upon dilution. The absorption curve for c = 0.434 has been drawn in greater detail, readings having been taken at every iojuju. The A-c curve for 744 juju, lying on the edge of the yellow- red absorption band, shows A to decrease greatly with dilution. This curve (and the A-c curves for io56/*ju and 1095^/4) have been plotted on a scale of ordinates ten times as small as the other A-c curves. For wave-lengths in the region of low absorption, 764 jit/* to 842juju, A is approximately con- stant, although in this region the values of a are so small that the values of A are liable to considerable inaccuracy. The A-c curves for wave-lengths 920 ju /* to 109 5 ju/i, on the edge of the infra-red band, show that A increases rapidly with dilution. Cobalt Chloride in Iso-Butyl Alcohol. (Fig- 6) Four solutions were prepared varying in concentration from c = 0.194 to c = 0.05. The absorption curves have the same character as those for the ethyl alcohol solutions, and the color of the solutions in the bottles was the same, being a deep blue which changed to a greenish blue upon dilution. In preparing the solutions the usual precedure was followed, namely, to make the dilutions by addition of the pure alcohol to the saturated mother solution. It was found that a pre- cipitate appeared immediately upon dilution. The solu- tions were then filtered, and the concentrations measured by a determination of the density. The value of the concentration determined in this way was found to agree within the error of experiment with the concentration calculated from the known amount of dilution. This showed that the loss by precipitation was either negligible, or that the precipitate contained nearly equal parts of cobalt chloride and iso-butyl alcohol. The filtered solutions appeared quite free from any visible particles. 12 In the cells they had a somewhat cloudy appearance, sugges- tive of a colloid condition. They showed slightly a Tyndall cone in blue light. An examination of these freshly filtered iso-butyl alcohol (and also the iso-amyl alcohol) solutions with the ultramicroscope showed that they were not colloidal in nature, but that they contained a number of particles. Whether these particles were newly formed precipitate, or some impurity, is unknown. 1500/ujA The A-c curves for 73 4 MM and 744 MM, wave-lengths lying on the edge of the yellow-red absorption band, show again that A decreases rapidly with dilution. For the wave-lengths 754 MM and 764 MM in the region of transmission A is a con- stant. The behavior of the edge of the infra-red band is similar to the case of the propyl alcohol solutions, for A in- creases with dilution, as shown by the rise in the K-c curves for wave-lengths IOISMM to 1133 MM- Cobalt Chloride in Iso-amyl Alcohol. (Fig. 7) Six solutions were prepared varying in concentration from c = 0.064 to c = o.oio. The solutions in the bottles were of a deep blue color in the higher concentrations, which changed to a greenish blue upon dilution. The general character of the absorption curves is the same as that of the ethyl alcohol solutions. The iso-amyl alcohol solutions exhibited the same phenomenon of precipitation upon dilution as has been de- scribed in the case of the iso-butyl alcohol solutions. They also had the same appearance in the cells, and under the ultra- microscope. A study was made of the precipitate which was thrown down in these solutions, for the deposit in the case of the iso-amyl alcohol solutions was more abundant than in the case of the deposits in the other cobalt-chloride solutions. The solution was allowed to stand for two weeks, and then the precipitate was filtered off. This precipitate consisted of 14 blue needle crystals mixed with a flocculent scale-like residue. Analysis showed that in this flocculent residue there was present 54 percent by weight of cobalt chloride. If this precipitate was a compound of the cobalt chloride and the alcohol, this percentage of the chloride would indicate that the compound contained two molecules of the chloride to three of the alcohol. The A-c curves for the edge of the yellow-red absorption band, at 7 14 MM and 724 MM> show that A decreases with dilution. In the region of transparency between the two bands A is constant, as shown by the A-c curves for 734 MM and 744 MM- The A-c curves for the edge of the infra-red band show that A increases very rapidly with dilution. Discussion of Results with Cobalt Chloride This study of cobalt chloride in water and alcoholic solu- tion brings out the following facts : In the region of wave-lengths lying on the long wave- length edge of the yellow-red absorption band the A-c curves show that A decreases with dilution. The decrease in A observed in the case of the water solution is considerable, and in the case of the alcoholic solutions this decrease becomes more and more marked as the molecular complexity of the alcohol increases. Jones and Anderson 1 studied solutions of cobalt chloride in water, methyl alcohol, and ethyl alcohol. Plates 2, 4, and 5 of their paper showed that for wave-lengths on the red edge of the yellow-red absorption band A de- creased with dilution, and also showed that this decrease was much more marked in the case of the alcoholic solutions than in the water solution. This is in accord with the facts brought out by the measurements discussed in the preceding para- graphs. In the region of low absorption between the two bands it is concluded that A is constant. As has been mentioned already, in the section concerning cobalt chloride in propyl alcohol, the values of a. for the region between the two bands 1 Carnegie Inst. of Wash., Pub. no. 15 are so small that the values of A are in many cases worthless. In the region of wave-lengths lying on the edge of the infra-red band A experiences deviations from a constant value, and again these deviations show a certain regularity concomitant with the increasing molecular complexity of the solvent. In this region A is nearly constant for the water solutions, but increases with dilution for the alcohol solu- tions, the increase becoming greater as the molecular weight of the alcohol increases. Conclusion The relation between A, the molecular light absorption coefficient of the solution, defined by equation (2), and c, the concentration of the solution in gram-molecules of salt per liter of solution, has been determined. It has been found that in general A is not a constant. In certain cases A de- creases with dilution, in other cases A increases with dilution, and in still other cases as dilution proceeds A decreases to a minimum, and then increases again. Another possible com- bination, namely, that A should increase to a maximum and then decrease, was not met with. At present there is no adequate theory to explain the facts which have been recorded here. The fact that A varies with c has been probably correctly attributed by Jones and Anderson 1 and others to the formation of complexes, which were considered to be loose chemical compounds of molecules of the salt with molecules of the solvent. Undoubtedly the changes in A with c observed in this investigation may be explained in a qualitative manner by the hypothesis of com- plexes, or "solvates" as they have been called. But before it can be useful for the interpretation of quantitative data, the solvate hypothesis must be couched in mathematical terms. The Johns Hopkins University May 10, iQi6 1 Loc. cit. BIOGRAPHICAL NOTE. John Foster Hutchinson, son of Edwin James Hutchinson, and Jennie (Lilly) Hutchinson, was born in Tonawanda, N. Y., on June 6, 1893. His early training was received in Ferguson, S. C. In 1909 he entered The Military College of South Carolina, and in 1913 was admitted to the degree of Bachelor of Science. In 1914 he received the degree of Master of Arts from the College of Charleston. He entered the Johns Hopkins University in 1914 as a graduate student, making Chemistry the principal subject, Physical Chemistry the first subordinate, and Mineralogy the second subordinate. He attended grad- uate courses given by Professors Jones, Morse, Reid, Love- lace, Frazer and Swartz. He was the holder of Hopkins Scholarships during the years 1914-1915 and 1915-1916. 371461 UNIVERSITY OF CALIFORNIA LIBRARY