JUNE, 1908. Vol. XXXIII., No. 387. STUDIES IN STEAM DISTILLATION. BY H. DROOP RICHMOND, F.I.C. (Read at the Meeting, A p d 1, 1908.) PART THEORETICAL. THE vapour given off by a saturated aqueous solution at the boiling-point must be saturated--that is, the ratio of the number of molecules of the substance to those of steam will be the s&me as the ratios of the vapour pressures at the temperature of ebullition measured in the same molecular state of association. The ratio df the molecular composition of the vapour to that of the solution will be the rate of distillation of the substance relative to that of water. Unfortunately, as both water and many other substances are associated, it is but rarely that the rate of distillation of any substance, even in dilute solution, can be deduced from the vapour pressuresTHE ANALYST.and maximum solubility. I t is clear, however, that of two substances having vapour pressures of the same order, the less soluble will distil the faster. If the conditions be such that the molecular state of association does not change, the rate of distillation will remain constant ; the rate of distillation of a substance in dilute solution appreciably, though not absolutely, conforms to this law. Expressed mathematically, calling the number of molecules of substance y, and of water x7 the ratio of the molecules of substance to those of water in the vapour at any moment d?l will be dz, and the relation of this ratio to the ratio in solution will be expressed by & Y the equation = a - -. dx x If the volume is kept constaut, as is practically the case in an ordinary steam distillation, x becomes a constant, and may be taken as 1 ; as we must in this case consider the relation of the quantity of water removed to the quantity of substance left behind, the equation must then be written - "' =ay.Integrating, we get the equation -- log y = ax + C ; or, as the integration constant disappears on evaluating for zero, y = e-'lX (9 = amount of substance left in solution, original amount being taken as 1, when x=aiiiount of water distilled, the original amount of water being taken as 1). If, however, we do not keep the volume constant, but make no addition what- ever to the liquid, as is the case when a solution is distilled, the equation is dx k a Y . and integrating we get log y = a log x, or y = x ( ~ (?j = amount of substance dx x' left in solution, and x = the amount of water remaining in the solution, the original amounts being taken as 1).For dilute solutions .c may be taken as the total amount of solution remaining without appreciable error. As in an ordinary distillation condensation always occurs, it will be well to consider the eflfect of this. Let us suppose that a fraction b of the aqueous vapour condenses, and the fraction of the vapour of the substance that condenses will be n x b ; the equation for The above equations suppose that no condensation takes place. dy - abdg = c1 ,y -+ nbdy d.x - Dtl.ra .?:+ bdx ' but as &j and ( 7 . ~ are equilibrium at any moment will be 1 - h y infinitesimally small in relation to y arid L, the equation becomes t7!1=.a x (In: 1 - ab ..I' -that is, the efl'ect of condensation will be to increase the rate of distillation if cd is greater than 1, and to decrease it if less. If the amount of condensation is known, or is kept constant, the apparent rate of distillation can be determined ; but it is rare that either of these conditions obtains. In many methods depending on differences of the rate of distillation, attempts are made to obtain constancy of condensation by prescribing dimensions of various kinds ; but the important points, such as external temperature and conductivity of the still head, are usually ignored. I t is quite tt simple matter to eliminate condensation in the still by providing it with a steam- jacket. If a, greater quantity of substance is present in the still than can dissolve in theTHE ANALYST, 211 water, the composition of the vapour will remain constant provided that the rate of solution is as rapid as the rate of removal by distillation.The condition which tends to ensure this is that the substance shall be intimately distributed throughout the liquid, which can be attained by agitation. If, as is often the case, a non-volatile substance is present in which the substance is soluble, the distillation will follow a simple law provided that the rate of diffusion of the substance from the non-volatile solvent to the water is as rapid as the rate of removal. For equilibrium between the water and the solvent we have the equation , (y’= the amount of substance in the water, and s the volume of the solvent); then y’= y.Substituting this in the equation 2 = a $, we get %!= ac -!’ and integrating the equation becomes log y = a log (s +cx)+ C. dx s + CJ-’ Evaluating the constant for y = 1, .I: = 1, we get log y = n log (s + cx) - a log (s + c), or ; thiEi equation shows that it is impossible to distil all the substance If the volume of water is kept constant, and the differential equation This condition requires intimate admixture. _ _ ?/I - c Y - ! / I z CX s+cx s + c.c ( I I/ = (S + C ) when all the water has passed over. - = &?J’ is used, this becomes - - ?/, and integrating we get d,?’ - nc (1J tlx s+a This equation is much simpler than the former, and lends itself more readily to experimental verification; it shows that it is an advantage to keep the volume constant by passing steam into the still, and there is the further advantage that, agitation is thereby secured.If we have two or more substances in solution, so long as the quantity of each is suficiently small not to affect the molecular state of association of the others or of the water, the ratios (ijtf : ayii ~ ~ 1 1 ‘ . . . . ( I . ~ ( I l ! j ’ I L L ~ ~ y ~ t : c ~ l l p l l ~ . . . . : will hold; in other words, each substance in dilute solution retains its rate of dis- tillation in the presence o€ others. If we have one substance in greater amount than will dissolve in the water, and another substance which is soluble in both water and in the first subetance, we get the equation as above for the second substance Y’ ,I‘ - -’ - ,s ” ; if the substance not wholly dissolved can rapidly go into solution, and replace that which has distilled, we get the equation -.const., or s = kx ; and substituting this we get ”= ak * ‘- which will hold until the whole of the substance has passed into solution. The agitation of the solution must be very thorough for the experimental verification of this equation. ax: - dx k + c 2 ’212 THE ANALYST, PART II.--THE RECOVERY OF AMYL ALCOHOL FROM THE ACID LIQUORS OHTAINED IN THE GERBER METHOD. The recovery of amyl alcohol entails so little labour and expense that a distinct economy is effected in laboratories where many determinations of the fat in milk by the Gerber method are made. The working out of the process has proved an interesting study in distillation, and the results obtained have incidentally proved that no appreciable amount of amyl alcohol is contained in the layer of fat which is separated and read off for the determination of the percentage of fat in milk.Early in 1905, J. A. Goodson, working in my laboratory, observed that by a simple steam distillation of the acid liquors a large proportion (upwards of 70 per cent.) of the amyl alcohol came over when about 50 C.C. of water had distilled per litre of liquor ; considerable blackening occurred, and a good deal of sulphurous acid passed over. Repetition of Goodson's experiments showed that if the acid liquors were allowed to cool to room temperature before being placed in the distillation vessel, and heated solely by steam, the dilution caused by the condensation of the steam was sufficient to prevent almost entirely the blackening, and to suppress completely the evolution of sulphurous acid, while the yield of recovered amyl alcohol W;LS increased. A number of quantitative estimations were made, which it is not necessary t o give in full.The following is the result of one which is typical: The acid liquors from 306 fat determinations were placed in a stoneware jar, and 500 C.C. of the water condensed in a previous distillation added; steam was passed in till the vapours passed over, and at this point it was estimated that the total volume would have been 8,060 C.C. if measured at 15" C. Fractions were collected as under : 1 ... 2 ... 3 ... 4 ... 5 ... 6 ... 7 ... 8 ... C'twdc Ariiyl Alcoliol.c. v. ... 78 ... 59 ... 48 ... 32 ... 45 . . . 0 ... 21 ... 7 Aqueous Layer. 50 47 50 51 100 5 0 152 153 c. c. Ae was to be expected, the ainyl alcohol contained water, and the water amyl alcohol, both probably being saturated, or nearly so. By distilling 500 C.C. of crude amyl alcohol till no more water passed over, separating the amyl alcohol which had distilled, fractionating this, and repeating this process three times, 37 C.C. of water (saturated with amyl alcohol) were separated. This, corrected for the amyl alcohol in the water, corresponds with 7.12 per cent. by volume.THE ANALYST. 213 Five hundred C.C. of condensed water were distilled, and the following fractions collected : Airiyl Alcoliol Total. c. c. 1 ... ... 8.0 2 . . . ._. 10.5 3 ... .. . 12.5 4 ... ... 14.0 5 ... ... 15.5 G ... . . . 16.5 Water TO tit I . (’. C . 7-0 9.5 12.5 16.0 19.5 93.5 Amy1 Alcohol Total. c. c. 7 .. . . . 17.0 8 . . ... 17.0 9 ... ... 17.2 10 ... ... 17.2 11 ... ... 17.3 Water Total. c. c. 28.0 33.0 37.8 42.8 50.0 By extrapolation of the results after correction for the water in the amyl alcohol, it was estimated that the dry amyl alcohol in the water was 3.63 per cent. by volume. The experiments aff‘ord data for deducing the rate of distillation of amyl alcohol relative to that of water both from aqueous and acid solutions. There are certain facts which slightly vitiate the deductions, but not to any serious extent. First it was observed, as recorded by Balbiano (BCY., 1876, 9, 1437), that the aqueous solution of amyl alcohol, though quite clear when cold, became turbid on warming, and a slight separation of globules on the surface took place when near the boiling- point ; this means that the strength of the amyl alcohol solution at the boiling-point was not accurately known.The second point is that the continued small con- densation of steam in the jar from which the acid liquors were distilled prevents the volume from being accurately known at any moment ; it was assumed to be constant fhroughout, and the error thus introduced is not very important, For the deductions it was assumed that the ratio between the amyl alcohol and water in the vapour at any moment bore a constant relation to the ratio between amyl alcohol and water in the liquid; this is practically Henry’s law.Calling the amyl alcohol y in the distilling vessel, and the water x, this relation is expressed mathematically as - t7‘’- cLy or, as the water is assumed to be constant in the dx- x ’ experiment with acid liquors, and its value 1, -- ‘ly = 09. (The niinus sign is due to the fact that the water measured is removed from the liquid.) Integrating this gives the simple formula: log y = -ax (:c being the quantity of water distilled, and y the quantity of amyl alcohol remaining). I n the distillation of the aqueous portion, the volume of water is constantly decreasing; it is convenient to let y and x represent the quantities of smyl alcohol and water remaining in the flask. rlx Integrating the first expression, the formula in this case is log y = a log x. Correcting in each case the amyl alcohol for dissolved water, and water for dissolved arnyl alcohol, also in the first experiment making a, correction of 2.5 per cent.of amyl alcohol used in the 306 determinations for mechanical loss (a near214 THE ANALYST. approximation), and adding on the amyl alcohol dissolved in the 500 C.C. of water added, we get the following values : y (total=l). 0.764 0.590 0.442 0.342 0.198 0.165 0.085 0.048 - 0.2G9 - 0.528 - 0.816 - 1.073 -- 1.620 - 1.802 - 2.465 - 3.037 and for the aqueous solution : y (total=l). 0.576 0.450 0.340 0.257 0.174 0.108 0.080 0 so69 0.047 0.030 0.020 log y, - 0.552 - 0.798 - 1.079 - 1.359 - 1.749 - 2.226 - 2.523 - 2.674 - 3.058 - 3.507 - 3.912 log 5 0.0069 0.0133 0-0200 0.0266 0.0395 0.0458 0.0650 0.9841 - 0.0152 - 0.0209 - 0.0269 - 0.0347 - 0.0424 - 0.0511 - 0.0605 - 0.0712 - 8.0810 - 0.0921 - 0.1069 0.0151 0,0207 0.0265 0.0341 0.0415 0.0498 0.0587 0.OG87 0-0782 0.0880 0.1014 39.0 Mean, 40.4 39.3, 39.3 36.1 1 2 (total= 1).i4j 39.2 i;:; From curve 43.1. 41.7 i 37% 38.0 36-8 The values were plotted out as curves, and the mean value of a read from the curves. I n the case of the distillation of water saturated with ainyl alcohol, it was evident both from the results and the curve that there was a disturbing influence a t the commencement of distillation, which was alniost certainly the separation of a. portion of the amyl alcohol as oily drops before mentioned; it wag therefore thought advisable to read a from the curve. I t is probable that this result is low, as it is seen that the value drops towards the end of the experiment, and a very slight alteration in the extrapolated value for the total dissolved amyl alcohol would not only correct this, but would slightly raise all the values.There is another method for deducing the value of a ; from Henry's law i t follows that a saturated solution should be in equilibrium with a saturated vapour. Assuming that the amyl alcohol is pure iso-amyl alcohol, which is of course not the case, but which assumption will not lead to great error, and that a saturated solutionTHE ANALYST. 215 at 100" C. is equal to one that contains at 15" C. 3.5 per cent. by volume of amyl alcohol, we find that the molecular ratio of amyl alcohol to water in a saturated vapour at 100' C. will be the vapour-pressure of water at 100" C.divided by the vapour-pressure of amyl alcohol at 100" C. = - - , the molecular ratio of a saturated solution will be = - which is n figure of the same order as that deduced from the distillation values. (mean), 3S.2 (from curve), and 40.2 (mean) and 40.1 (from curve). are the values of a, and the percentage by volume of sulphuric acid : 230 760 3.5 x 0.82 i 88 96.5 x 1+18 230 3-5 x 0.82 i 88 , and the value of a will equal 760t -gs15 xi 18 - = 49 8, Other distillation experiments from acid solutions gave the values of a 38.2 The three acid solutions had not quite the same concentration; the following Percc~ntage by Volume of' a Sulphiiric Acid. 38.2 ... ... ... ... ... 39.4 39.3 ... ... ... ... ... ... 37.9 40.1 ... ... ... ... ...... 36.3 Now, by the law of mass action, C A ~ H S ~ ~ = KCAm~H.CH2s04 (C = molesular con- centration) and the total molecular concentration of the amyl alcohol is proportional to y = C t i r n ~ ~ S ~ , + C A r l l ~ ~ ~ = (1 + KC/II,SO~) CA~~OEI, and we may write our fundamental (a'= the relation of the ratio of amyl differential equation as - alcohol to water in the vapour to the ratio of amyl alcohol as AmOH to water in the solution). d?/ a', Y d.c x( 1 + KCHI,SO,) a' 1 -I- KCrrS;.l' Therefore, n = The accuracy of the figures is not suflicient to render the calculation of K (the dissociation co-efficient) of any real value, but it is seen that the values of a vary inversely as the concentration of the sulphuric acid. I t is also seen that the dissociation of AmEISO, is large.From the values of cc given above, it is calculated that to distil 95 per cent. of the amyl alcohol it is necessary to distil- P.(*. of \Vatu per 10 0.c. Sulphuric Acid. Pc~r Cent. of Rater. 7.77 ... ... ... ... ... ... 2.00 7.64 ... ... ... ... ... .. 2.02 7.49 ... ... ... ... ... ... 2-06 Dilution of the acid liquid is seen to be such a slight disadvantage with regard to the quantity distilled that it practically is immaterial; as, however, approxi- mately 2% of the amyl alcohol distilled is in solution in the water, it is a real advantage to save this, and add it to the next distillation. I t was found from the mean of a large number of distillations that the volume of crude axnyl alcohol recovered was equal to 97.8 per cent. of the amyl alcohol used.The recovered product had always a slight brownish colour, and a smell which differed from that of the original.216 THE ANALYST. Trials were made to ascertain whether the crude amyl alcohol could be used for the Gerber process, but it was found that the results for fat were always high, averaging 0.12 per cent. in excess. A fractionation was carried out, and it was found that boiling commenced at 34" C., and the boiling-point rose steadily to about 85" C. before water began to come over ; the temperature then rose slowly to 92' C., at which it remained for a long time, this being the temperature which previous experiments with purer alcohol showed was the boiling-point of amyl alcohol saturated with water ; the temperature then rose slowly to 123" C., when all the water appeared to have distilled.As noted previously, the portion of the amyl alcohol distilling below 123' C. was three times refractionated, and 7.12 per cent. of water separated; 8 per cent. boiled below 123" C., a fraction boiling between 123" and 131" C. was distilled amounting to HO per cent., and a residue of 4 per cent. remained, which was very dark in colour. The fractions were used for testing milk, and gave the results below : Below 123" C. reading showed ... ... 0.8 excess 123" to 131' C. reading showed (about) ... . . , 0.05 ,, Residue reading showed ... ... v . . ... 0.8 ,, An attempt was made to dehydrate the crude product by adding 30 grams of caustio soda to 400 c.c., but although an aqueous layer separated, the alcohol still contained 1.5 per cent.of water, and so much soda dissolved that the distillation could not be carried very far, only 66 per cent. of the fraction 123" to 131" C. being obtained. The addition of 10 per cent. of anhydrous sodium sulphate and 10 per cent. of lime respectively only had a slight effect in dehydrating, and caused 10 and 16 per cent. of mechanical loss respectively, the fractions obtained being 65 per cent. and 68 per cent. respectively. All these also gave an excess reading of about 0.05 per cent. when used for fat estimation. These were refractionated, and small portions hoiling below 126' C. and above 131" C. were rejected, which both gave readings about 0.2 per cent. in excess. The mixed fractions, 126" to 131" C., were then refractionated, and the following fractions obtained : Fraction. 92" to 126" C. 126" to 128" C. 128" to 128.5" C. 128.5' to 129' C. 129" to 130' C. 130" to 130.3' C. 130.3' to 131" C. Residue Percentage. 5.3 (trace of water) 13.8 12.6 6.3 21.2 527.6 9.9 3.2 - 6.9 25.2 12.6 21.2 92.0 14.1 -THE ANALYST. 21 7 Except the residue, which gave an excow reading of 0.2 per cent., all the frac- tions gave excess readings of about 0.05 per cent. These results indicate that the chief constituents are a substance boiling a little above 128" C. (active amyl alcohol) and a substance boiling a little above 130" C. (iso- amyl alcohol), and that it is impossible to separate any substance that gives the excess reading of 0.05 by fractionation. I t wag deduced from the results of distillation, and from the excess readings in the Gerber tests, that the crude aniyl alcohol recovered by steam distillation had the following composition : Water . . . ... ... ... ... ... ... 7.1 per cent. Substance of high boiling-point (above 131" C.) ... 0.4 ,, &4myl alcohol (giving excess readings of 0.05 per cent.) ,, Substance of low boiling-point (34" C. ?) . . . ... 0.9 y j 91.7 By fractional distillation of several litres of the recovered amyl alcohol (after the addition of small quantities of lime to neutralise the acid) 87-5 per cent. of amyl alcohol was recovered from the crude product.