- Email: [email protected]
Aluminium chloride as a solvo acid and friedel-crafts catalyst
J. Inorg. Nucl. Chem., 1957,Vol. 4, pp. 30 to 39. PergamonPress Ltd., London
A L U M I N I U M C H L O R I D E AS A SOLVO A C I D A N D FRIEDEL-CRAFTS CATALYST* JOHN L. HUSTONand CHARLESE. LAN61Chemistry Department, Loyala University, Chicago
(Received21 September 1956) Abstraet--A detailed kinetic study has been made of the rate of chloride exchange between liquid phosgene and solute aluminium chloride. At low concentrations the rate is first order in aluminium chloride, but it changes over to a rate intermediate between 1 and 2 at higher concentrations. The cross-over point occurs at high concentration when the temperature is low and low concentration when the temperature is high. The energy of activation is 14.1 kilocalories in the first order region and 18"3 kilocalories in the region of higher order. It is suggested that phosgene and aluminium chloride undergo a weak acid-base interaction leading to chloride exchange. It is also suggested that the cross-over to higher order is due to the appearance of ions in the solution, rather than to a specific chemical effect. In conformity to this picture, it is found that chloro-aluminate ions, for example, A1C14-, undergo much slower exchange of chloride with solvent phosgene than does aluminium chloride. It is also found that the presence of ions in this solution can have a drastic effect toward raising the rate of exchange between phosgene and solute aluminium chloride. Hydrogen chloride suppresses the rate of exchange of chloride between phosgene and solute aluminium chloride, possibly by means of such a reaction as A1Cla + HC1 + COCla ~ HCOCla + + AIC14It appears that while there is no compound formation between hydrogen chloride and aluminium chloride by themselves, yet aluminium chloride can force hydrogen chloride to donate a proton to a proton acceptor, even one as weak as phosgene. INTRODUCTION
A RATHER extensive study of the properties of phosgene solutions of aluminium chloride by GERMANN and co-workers ~1) led to the first formulation of what has come to be called the solvent systems theory of acids and bases. ~2) GERMANNobserved that a solution of aluminium chloride in phosgene conducts much better than the pure solvent, evolves carbon monoxide on electrolysis or on reaction with alkali metals, and is capable of reacting with ionic chlorides to give conducting solutions. On the basis of these observations he constructed the following formulation: A parent solvent is a substance from which a system of solvo acids, solvo bases, and solvo salts is derived. A solvo acid is any electrolyte which, in a given parent solvent, yields cations identical with the cations of the self-ionization of the parent solvent, and in addition possesses atoms characteristic of the anion of the parent solvent. Thus sulphuric acid is a solvo acid in water, while hydrochloric acid is not. He considered aluminium chloride to be a solvo acid in phosgene by means of the process AlzC1G ~
COC12~
Co ++ ~
A12C1 s -
* Presented at the Dallas Meeting of the American Chemical Society, April, 1956. t Taken in major part from the Master of Science thesis of Mr. Lang. cl) A. F. O. GERMANN and D. M. BmOSEL J. Phys. Chem. 29, 1469, et ante (1925). 12) T. MOELLER Inorganic Chemistry, Wiley, N e w York, p. 322ff. (1952). 30
Aluminium chloride as a sotvo acid and Friedel-Crafts catalyst
31
A solvo base was similarly detined. A solvo salt can then be made by the reaction o f a solvo acid with a metal or with a solvo base. CO ~' - - C a - - C a
~ ! ~ CO
CO ~- !.CaClz-:~Ca~i
+ COCIz
As part of a recent investigation of the validity of the solvent system postulates
AND
DISCUSSION
AICI:~COCI,2 exchange Data from these experiments are summarized in Table 1 and in Fig. 1,<~ a log-log plot of rate of exchange of chloride between aluminium chloride and solvent phosgene. The small numbers to the right of the various straight lines are the slopes of these lines and, therefore, the order of exchange rate with respect to aluminium chloride. The most striking feature of the plots is, of course, the abrupt change in the order fi-om about one to higher values. It will be observed that this change occurs at lowest concentration when the temperature is high and highest concentration when the temperature is low. lndeed, if the experinaents had been performed at sufficiently low or sufficiently high temperatures, no such change in order could have been observed. The exchange process would have been either entirely first order or entirely of higlner order. This observed change in order seems to be unique in the literature of exchange reactions. Probably the most thoroughly investigated exchange between a solute and a solvent of low dielectric constant is the chloride-catalysed exchange of thionyl chloride and solvent sulphur dioxide./a~) In this case the exchange order is one with respect to thionyl chloride. And it appears, on the basis of more meagre data, that the order is also one with respect to acetyl chloride in acetic anhydride, acetic acid in acetic anhydride, and acetic anhydride in acetic acid. lay) It does appear that the exchange between thionyl bromide and solvent sulphur dioxide is zero order with respect to the solute. (a<') But a change in order was observed in none of these cases. a,,~ R. E. Jolt',soy, T. H. NORRIS, and J. L. H t STON J. Amer. Clwm. &,c. 73, 3052 (1951). :~'~ E. A. FVA',S, J. L. HUSTON, and T. H. NORRIS d. A m e r . Chem. Soc. 74, 4985 (1952). :~'~ B. J. M,XSIERS and T. H. NORRis J. A m e r . Chem. Soc. 77, [346 (1955). ~) J. 1,. l t u s / o N J. k~ork,. Nucl. Chem. 2, 128 11956), ~:" Equivalent fraciion of a l u m i n i u m chloride : (cquivaIents a l u m i n i u m chloride) -; (equivalents aluminium chloride : equivalents phosgene). Equivalent fraction is used rather than mole fraction in order that concentration should be expressed in terms of the exchanging species, i.e. chloride. See G. FRIEDLANDER alld J. W. KENNI'DY (1955), Nuclear altd Radiockemi.strr, Wiley, New York, p, 315 If. 1i)r quantitative theory of exchange reactions. Rate of exchal~ge ~ a s calculated from the cxpression R~
0-693'a)(b) T.
whereaandharelheconcentrationsofaluminium chloride and phosgene. Values ol" T, w e r e o b t a i n e J f'rom plots of log (/-i'raction exchanged) versus time.
32
JOHN L. HUSTON and CHARLES E. LANG"
10 -~
10 -2
10 -3
10 - 4 .o ° 10- s Ju
iU
10- 2
Equivalent
fraction
10-~
AI CI3
Fro. 1.
l o - ~ L _~=
10
~0
~ ~_=
1() 4 =
1(~5
3'30
3'50
3'70
I/T x t03 FiG. 2.
3'90
33
Aluminium chloride as a solvo acid and Friedel-Crafts catalyst TABLE I.--PHOSGENE-ALUMINIUMCHLORIDEEXCHANGE
T~
R /, I 0
Slope
0-00423 0.00865 0.0396 0.0663 0.136 0-194
405 hr 460 hr 450 hr 420 hr 285 hr 185 hr
0.072 I 0.129 (I.586 1.20 ::.86 5.85
0'96 0"96 0"96 0"96 l '99
0"000503 0"00213 0-00419 0"00921 0-0192 0.0402 00694 0.132 0"173 0.300
48,0 hr 59,0 hr 58.0 hr 60,5 hr 52.0 hr 36.0 hr 25.0 hr 14.5 hr 10.5 hr 7.5 hr
0.0726 0.250 0.499 1.045 2.5 l 7.42 17.9 54.8 94-4 194
0'93 0"93 0'93 0'93 1 "60
0.000098 0.OOO479 0-00109 0'00236 0.00532 0.00570 0.0265 0.0739 0.211 0.417
5.20 hr 5"40 hr 4.70 hr 5.08 hr 3.70 hr 3.47 hr 2.13 hr 1.33 hr 0.57 hr 0.26 hr
0-131 0-616 1.61 3.20 9.90 11.3 84.1 358 2036 6546
1'03 1.03 1.03 1.03 1.34 1.34 1.34 1.68 1.68 1.68
Equivalent fraction
....21"1'
ff
= 25.0'
1 "99
1 '60 l "60 1 "60 1 "60 1 "60
Fig. 2 shows three plots for determination of activation energy, made at concentrations indicated in Fig. 1. It can be seen that good straight lines are obtained at regions 1 and 3 but n o t at region 2 which was deliberately chosen to include both first order and higher order plots. The fact that region 3 does give a good straight line indicates that a single exchange process is taking place in this region, and that the existence of orders intermediate between one and two is n o t to be ascribed to c o n c u r r e n t u n i m o l e c u l a r a n d bimolecular processes involving a l u m i n i u m chloride. It seems that the first order region corresponds at least approximately with the m i n i m u m of GERMANN'S conductivity curves) 61 a n d thus the rising order at higher concentrations may be due to a greater p o p u l a t i o n of ions in these regions, rather than to any specifically chemical effect. Perhaps the nature of the solutions in these regions of first order a n d higher order can be represented as 2A1CI a ~ - [A1C12+, A1Cla-] and (l~} k . F. O. GERMANN
6A1CI a ~
[AlaCI~]+ -~ [AlaCllo]--
J. Ph)'s. Chem. 29, 1148 (1925).
34
JOHN L. HUSTONand CHARLESE. LANG
It m a y be noted that GERMANN'S v a p o u r pressure data indicate aluminium chloride to be partially dimerized in phosgene solution and these equations could as well have been formulated in terms of A12C16.Iv) Phosgeno-aluminates
It has already been suggested 14) that solute aluminiurn chloride behaves toward solvent phosgene as a strong Lewis acid toward a slightly weaker one, the initial "neutralization" to form an aluminium chloride-phosgene complex being rapid, but the actual transfer of a chloride ion involving an appreciable activation energy. The kinetic data here reported are clearly compatible with this picture. In order to obtain collateral supporting evidence we have made a brief investigation of the exchange of chloride between solvent phosgene and chloro-aluminate ions, for example A1Cla-, AI.~Cls=, etc. These ions m a y be regarded as aluminium chloride partially neutralized, and would be expected to exchange at a slower rate with the solvent than does aluminium chloride itself. Table 2 summarizes data on the rate o f exchange of solvent phosgene with chloroaluminate and with mixtures of aluminium chloride and chloro-aluminate, the latter TABLE 2 Phosgeno-aluminates Reactants
1
2 3 4 5 6 7 8
9 10
A1C13= ÷ CaC12= A1Cla ÷ CaC12A1Cla= + CaCI2 A1Cla + CaC12A1CI3= + CaC12 A1Cla ÷ CaC12= A1Cla= + NaCI= AIC13= + NaCI= A1CI3= + NaC1 A1CI3 + NaCI =
Concentration A1C13 A1CI~0 0 0.0113 0.0104 0.0160 0.0193 0.0 0.0 0.0167 0.0169
T~. hr Observed
0.019 0.0398 0.0063 0.0059 0.0199 0.0209 0.0094 0-0352 0.0201 0.0201
180 190 59 62 73 72 1201~ 1520 39 41
A1CI3 alone ~al 55 37 54 54 39 37 59 40 37 37
~) That is, T½would have been this had no ionic chloride been added. ~b~LOWbecause of incomplete reaction. data taken f r o m the previous paper/41 The outstanding feature of these data is the great difference between the half-times reported in each case in the last two columns for runs 1, 2, 7 and 8. Clearly the chloro-aluminate ion, whatever its precise formula m a y be, is not nearly as active towards exchange with solvent phosgene as is aluminium chloride. A n d it is remarkable that there is so little difference between the last two columns for runs 2 and 4 and 9 and 10. In each o f these cases approximately one-half o f the aluminium chloride has been converted to the m u c h less kinetically active chloro-aluminate ion, yet the half-time has been only slightly effected. A n d even in the cases o f runs 5 and 6 the increase in half-time has been much less than one 17) A. F. O. GERMANNand G. H. MCINTYRE J. Phys. Chem. 29, 102 (1925).
A l u m i n i u m c h l o r i d e as a s o l v o acid a n d Friec2el-Crafts c a t a l y s t
35
might expect, considering that the aluminium chloride concentration is here in the region where the half-time is concentration-dependent (the order is greater than 1). It would seem that unreacted aluminium chloride is undergoing exchange more rapidly with the solvent in these solutions than it could if no ions had been present. Of course it is not surprising that in a solvent of such low dielectric constant as phosgene "ionic strength effects" should be very large. In any case, this may be taken as supporting evidence for the hypothesis previously offered to explain the change of order. It may be noted that in the case of run 7 there was apparently incomplete reaction between sodium chloride and aluminium chloride. In this run (which had only a very small excess of sodium chloride over the stoichiometric quantity) the plot of log (1-fraction exchange) versus time underwent a sharp break at about 40 ~ exchange and became almost flat. Runs 1, 2 and 8 gave good straight lines, and it may be assumed that there was substantially complete reaction in these three cases from the very beginning. It is of interest that the unreacted calcium chloride and sodium chloride (solids) of runs 1 and 8 underwent at least appreciable heterogeneous exchange with the solution, the specific activity diminishing in each case by about 11 ~/o. No doubt this exchange was by chloride ions immediately adjacent to the surface of the solid. Since these solids were prepared in just the same way as the solids used in the previous work, ca) it is clear that the failure of solid ionic chlorides to undergo heterogeneous exchange with pure liquid phosgene (containing no solute) was not due to any lack of development of their surface area, but rather to the absence of ionic species in pure phosgene. The great difference in rate between runs 2 and 8 may be due to a difference in solubility of calcium chloride and sodium chloride in phosgene. It is not unreasonable that solid sodium chloride should permit the higher concentration of chloride ions in phosgene and thus permit the equilibrium C1- + A1C13~ A1C1C to proceed further to the right. It may be asked whether we have any positive evidence that the formula of the chloro-aluminate ion is indeed A1C14- and not, for example AlzCl7-. In three runs the solutions were separated from unreacted solid ionic chlorides by decantation, and after removal of solvent phosgene the previously dissolved material was taken up in water for analysis. In all three cases the amount of chloride found was much closer to that expected for A1CI~- than AI~C17- ; this does not of course preclude the existence of such ions as AlzC1 s" or other polymers of A1CI4-Is)
AlCl.~HCl It has been clearly shown that aluminium chloride and hydrogen chloride form no compound (e.g. HA1CI~) of any substantial stability, lu/ At temperatures as low as 120 ° no appreciable amount of hydrogen chloride is taken up by aluminium chloride, the conductivity of a saturated solution of aluminium chloride in liquid hydrogen cs) See Experimental part for details. (9,) H. C. BROWN and H. PEARSALL J. A m e r . Chem. Soc. 73, 4681 (1951). (~") R. L. RICHARDSON and S. W. BENSON J. A m e r . Chem. Soc. 73, 5096 (1951). ,101 M. BLAU, W. T. CARNALL, and J. E. WILLARD J. A m e r . Chem, Soc. 74, 5762 (1952).
36
JOHN L. HUSTON and CHARLES E. LANG
chloride is about the same as that of hydrogen chloride alone, and at higher temperatures the mixed gases show negligible deviation from ideality. However, such exchange experiments as have been done on this system, (10) have left room for feeble compound formation between the two substances. There was substantial heterogenous chloride exchange at 25° and even at --80 °. Probably this heterogeneous exchange should be compared to the heterogeneous exchange between aluminium chloride and carbon tetrachloride,(m there being of course not the slightest indication of any compound formation between these two substances. Yet it is well known that hydrogen halides have a promoting action on aluminium halide operating as a Friedel-Craft catalyst. It has been four~d that aluminium chloride, which has no appreciable solubility in toluene, goes into solution in the presence of hydrogen chloride, apparently by means of the reversible reactions: (9~) CHaC6H5 + HC1 @ ½A12C16~-~ [CHaC6Ha][A1C14-] CHaC6H 5 -~- HC1 -}- AI~CI6~+- [CHaC6Ha+][A12C1c] (Higher complexes presumably can also form). BROWN and PEARSALLhave suggested that it is such complexes which promote the solubility of aluminium halide catalyst and that these complexes play an important role in most Friedel-Craft reactions by furnishing a highly polar medium in which ionic intermediates may form and react. In other words, it might be stated that aluminium chloride can react with hydrogen chloride provided there is some place for the proton to go; some Br~nsted base to accept it. The phosgene-aluminium chloride system is formally similar to the system used by BROWNand PEARSALL, but phosgene certainly is a much weaker Br~Snstedbase than toluene, and by using it we have a more drastic test of the ability of aluminium chloride and hydrogen chloride to react in the presence of a proton acceptor. Accordingly, we have done a few experiments to ascertain the effect of hydrogen chloride on the rate of exchange of chloride between aluminium chloride and solvent phosgene. In Fig. 3 are plotted the ratio T½ (HC1 not present) -- ~ (HC1 present) as a function of the mole ratio of hydrogen chloride to aluminium chloride, concentrations of aluminium chloride being maintained as nearly constant as possible (equivalent fraction _~_ 0.038). The data have no precise numerical significance since we do not have the Henry's Law constants needed to calculate precisely the concentration of hydrogen chloride in the liquid phase, t12) but it is clear that hydrogen chloride suppresses the rate of exchange. Since aluminium chloride is, as we have seen, much more active kinetically toward exchange than chloroaluminate ion(s), it is probable that the mechanism of this suppression is a reversible equilibrium such as A1C1a -? HC1 + COCI~ ~-HCOC12+ q- A1C14Although definite compounds are formed in the toluene experiments of BROWN (111 C. H. WALLACEand J. E. WILLARD J. A m e r . Chem. Soc. 72, 5275 (1950). (12) One very rough estimation of a Henry's Law constant was made during the experiment which corresponds to the point near a mole ratio of 1.0 as explained in the experimental part. The data were used to correct the mole ratios plotted in Fig. 3.
Aluminium chloride as a solvo acid and Friedel-Crafts catalyst
37
and PEARSALL there is no evidence for any such formation here. This, of course, can be attributed to the weakness of phosgene as a Br6nsted base. An attempt was made to calculate equilibrium quotients for the processes ( 1)
HC1 @ COC12 :- AIC1 a ~ [HCOCI~!, A1CI4-]
(2)
HCI ~,--COCI,, @ 2AICI:~ ~ [HCOCI2 +, A12CI7 ]
(3)
HC1 _a COC1 z + A1CI 3 ~_ HCOCI2~ @ AICI 4 -
by means of the assumption that the ratio (aluminium chloride converted to ions) .'(residual aluminium chloride) is the same as the half-time ratio. None of these equilibrium quotients showed satisfactory constancy. The one calculated for process
1.0 i
0"8
0"6 0 f_
0.4
0.2
Moles
1"0 2'0 HCI-- moles
3'0 A[ CI 3
4"0
FiG. 3.
(3) showed the greatest deviation from constancy and certainly the degree of ionization in a solvent of so weak a dielectric constant can never be very substantial. More information on this system can be obtained by determining accurate Henry's Law constants for hydrogen chloride over solutions of aluminium chloride in phosgene, and such a study will be undertaken. Experiments were also performed to ascertain whether the rate of exchange of chloride between hydrogen chloride and aluminium chloride, both dissolved in phosgene, could be measured. It turned out that the rate of exchange was immeasurably fast from 0 ° down to --60 °, which was the lowest temperature at which a gas sample of suitable size could be obtained. At first it appeared that at 0 ° there was a measurable rate of exchange, of half-time approximately one minute, but the expected temperature dependence of this "half-time" failed to materialize, and it was, no doubt, a matter of equilibration between liquid phase and gas phase. It is clear then that the reversible equilibrium by which hydrogen chloride suppresses
38
JOHN L. HUSTONand CHARLESE. LANG
the phosgene-aluminium chloride change is both very rapid and reversible. (The two gases can easily be pumped off, leaving pure aluminium chloride behind.) We consider that our results with phosgene to be one in accord with the results of BROWN and PEARSALLwith toluene and constitute confirmation of their ideas. Before concluding it might be well to discuss briefly a minor point left unsettled at the time of publication of the previous paper. It was observed t4~ that when conventional plots were made of log (1-fraction exchange) versus time, the straight line did not intersect the zero-time ordinate at zero exchange. This was attributed to rapid exchange between aluminium chloride and hydrogen chloride impurity. It now appears that this effect is due rather to a small amount of radioactive hydrogen chloride produced by reaction of aluminium chloride with glass or with water adsorbed on glass, for it has not been possible to eliminate this effect by the most rigorous purification of the phosgene used. The effect has required the use of various modifications in the obtaining of data. At concentrations greater than 0.01, half-time values were obtained from the logarithmic plots and from the final specific activity of aluminium chloride, but at lower concentrations it was not possible to obtain satisfactory plots, because of this effect. Here, at the conclusion of a run, phosgene was fractionally distilled and only the least-volatile sample (presumed free of hydrogen chloride) was used for counting. Then a one-point plot was made to give a half-time value which was averaged with the value obtained from the specific activity of aluminium chloride to give the results reported. At the very lowest concentration (less than 0.0001) there was not sufficient aluminium chloride to give a sample of mercurous chloride large enough for counting, or indeed to permit an aluminium analysis. The half-time had to be obtained by comparison of the specific activity of phosgene (fractionally distilled) at some finite time with its infinite-time value. This latter value also served to furnish the aluminium chloride concentration. In this way we managed to investigate the exchange over a concentration range of a factor of 6000. EXPERIMENTAL
Preparation of materials These preparations differed materially from those previously described 14~ only in one significant detail. Since it was desired to obtain phosgene as free of hydrogen chloride as possible, the following step was added to its purification. It was observed by EPHRAIM(13) that hydrogen chloride forms stable compounds with silver sulphate and with various other sulphates of transition metals. After it had been ascertained that silver sulphate underwent no appreciable reaction with phosgene (silver sulphate in a drying tube showed no appreciable gain in weight when phosgene was passed through it for several minutes) a column was constructed containing silver sulphate and a small amount of metallic antimony, the latter being present to remove chlorine impurity. After evacuation and heating the solids to de-gas, phosgene was passed through at the rate of about 2 litres per hour. Mass spectrometric analysis of the phosgefie showed it to contain no more than 0"01% hydrogen chloride. Frequently, in the course of this work, preparations of radioactive aluminium chloride were obtained with a distinct yellow c010ur. This yellow colour was due to some substance somewhat less volatile than aluminium chloride itself, and for some ila) F. EPHRAIM Ber.58, 2262 (1925).
Aluminium chloride as a solvo acid and Friedel-Crafts catalyst
39
time was believed due to traces of iron in the aluminium metal used for the preparation. It now develops that this colour appears only when the glassware in which the preparation is made has been cleaned with chromic acid solution. If alcoholic potassium hydroxide is used as the cleaning solution this ditficulty is avoided. Whatever the nature the coloured material may be, and its nature is quite obscure at the present time, it could be removed by distillation of the aluminium chloride over hot aluminium metal.
Procedure fi~r maldn£T runs Exchange experiments involving phosgene and solute aluminium chloride were usually performed in all-glass apparatus, fitted with breakoffs and seal points, though some of the earlier experiments were done in apparatus fitted with silicone-lubricated stopcocks. The general procedure has been described. I~/ Those runs where calcium chloride or sodium chloride was added to the aluminium chloride solution were pcrl'oNned in all-glass apparatus. At the conclusion of each run the solution was separated from residual solid by decantation, phosgene was distilled away, the residue from tile solution was taken up in water, and analyses were performed for aluminium and for chloride The solid residue from the phosgene solution was determined simply by weighing. The analytical procedures otherwise used have been described in the previous paper. If the chloro-aluminate ion formed under these conditions is indeed AICIa , than we should expect the number of milli-equivalents of chloride present in phosgene solution to be 4/3 the number of milli-equivalents of aluminium. The 3 sets of data are as follows: chloride expected 0.03;34: chloride found 0-0331. Chloride expected 0.460; chloride found 0.490. Chloride expected 0.801; chloride found 0-851. Better data could not be expected because the decantation afforded only a rough separation. In the first of the series of runs where hydrogen chloride was added to the alunainium chloride solution, the experiment was performed in an apparatus where liquids and vapours were contained behind a stock valve. The apparatus permitted both investigation of the rate of exchange between hydrogerl chloride and aluminium chloride and also of the suppression of the phosgene-ahn-ninium chloride exchange by hydrogen chloride. A rough value of the distribution ratio of hydrogen chloride between liquid phosgene and the gas phase at 25-' was obtained in the following manner: The total vapour pressure over the phosgene solution was balanced in one arm of the stock valve against a source of tank gas in the other arm, the pressure of this external gas being measured by an open-end mercury manometer. Using Gt!RMANN'S data ";/of the vapour pressure of phosgene at 25 ~', it was estimated that the ratio (millimoles phosgene per cc. liquid i millimoles phosgene per cc. gas) - : 150. This datum was used to correct the points plotted in Fig. 3 for the amount (5 ,~--20'!,,) of hydrogen chloride present in the gas phase. Runs to measure the rate of low-temperature exchange between hydrogen chloride and aluminium chloride were performed in apparatus fitted with stopcocks. Runs lo measure the suppression of phosgene-aluminium chloride exchange by HCI were performed in all-glass apparatus at 25 ° , the rate of exchange being calculated solely from the specific activity of aluminium chloride at the end of the run. It was felt that it would be useless to measure for this purpose the radioactivity in the gas phase, consisting as it did of two radioactive substances.
A L U M I N I U M C H L O R I D E AS A SOLVO A C I D A N D FRIEDEL-CRAFTS CATALYST* JOHN L. HUSTONand CHARLESE. LAN61Chemistry Department, Loyala University, Chicago
(Received21 September 1956) Abstraet--A detailed kinetic study has been made of the rate of chloride exchange between liquid phosgene and solute aluminium chloride. At low concentrations the rate is first order in aluminium chloride, but it changes over to a rate intermediate between 1 and 2 at higher concentrations. The cross-over point occurs at high concentration when the temperature is low and low concentration when the temperature is high. The energy of activation is 14.1 kilocalories in the first order region and 18"3 kilocalories in the region of higher order. It is suggested that phosgene and aluminium chloride undergo a weak acid-base interaction leading to chloride exchange. It is also suggested that the cross-over to higher order is due to the appearance of ions in the solution, rather than to a specific chemical effect. In conformity to this picture, it is found that chloro-aluminate ions, for example, A1C14-, undergo much slower exchange of chloride with solvent phosgene than does aluminium chloride. It is also found that the presence of ions in this solution can have a drastic effect toward raising the rate of exchange between phosgene and solute aluminium chloride. Hydrogen chloride suppresses the rate of exchange of chloride between phosgene and solute aluminium chloride, possibly by means of such a reaction as A1Cla + HC1 + COCla ~ HCOCla + + AIC14It appears that while there is no compound formation between hydrogen chloride and aluminium chloride by themselves, yet aluminium chloride can force hydrogen chloride to donate a proton to a proton acceptor, even one as weak as phosgene. INTRODUCTION
A RATHER extensive study of the properties of phosgene solutions of aluminium chloride by GERMANN and co-workers ~1) led to the first formulation of what has come to be called the solvent systems theory of acids and bases. ~2) GERMANNobserved that a solution of aluminium chloride in phosgene conducts much better than the pure solvent, evolves carbon monoxide on electrolysis or on reaction with alkali metals, and is capable of reacting with ionic chlorides to give conducting solutions. On the basis of these observations he constructed the following formulation: A parent solvent is a substance from which a system of solvo acids, solvo bases, and solvo salts is derived. A solvo acid is any electrolyte which, in a given parent solvent, yields cations identical with the cations of the self-ionization of the parent solvent, and in addition possesses atoms characteristic of the anion of the parent solvent. Thus sulphuric acid is a solvo acid in water, while hydrochloric acid is not. He considered aluminium chloride to be a solvo acid in phosgene by means of the process AlzC1G ~
COC12~
Co ++ ~
A12C1 s -
* Presented at the Dallas Meeting of the American Chemical Society, April, 1956. t Taken in major part from the Master of Science thesis of Mr. Lang. cl) A. F. O. GERMANN and D. M. BmOSEL J. Phys. Chem. 29, 1469, et ante (1925). 12) T. MOELLER Inorganic Chemistry, Wiley, N e w York, p. 322ff. (1952). 30
Aluminium chloride as a sotvo acid and Friedel-Crafts catalyst
31
A solvo base was similarly detined. A solvo salt can then be made by the reaction o f a solvo acid with a metal or with a solvo base. CO ~' - - C a - - C a
~ ! ~ CO
CO ~- !.CaClz-:~Ca~i
+ COCIz
As part of a recent investigation of the validity of the solvent system postulates
AND
DISCUSSION
AICI:~COCI,2 exchange Data from these experiments are summarized in Table 1 and in Fig. 1,<~ a log-log plot of rate of exchange of chloride between aluminium chloride and solvent phosgene. The small numbers to the right of the various straight lines are the slopes of these lines and, therefore, the order of exchange rate with respect to aluminium chloride. The most striking feature of the plots is, of course, the abrupt change in the order fi-om about one to higher values. It will be observed that this change occurs at lowest concentration when the temperature is high and highest concentration when the temperature is low. lndeed, if the experinaents had been performed at sufficiently low or sufficiently high temperatures, no such change in order could have been observed. The exchange process would have been either entirely first order or entirely of higlner order. This observed change in order seems to be unique in the literature of exchange reactions. Probably the most thoroughly investigated exchange between a solute and a solvent of low dielectric constant is the chloride-catalysed exchange of thionyl chloride and solvent sulphur dioxide./a~) In this case the exchange order is one with respect to thionyl chloride. And it appears, on the basis of more meagre data, that the order is also one with respect to acetyl chloride in acetic anhydride, acetic acid in acetic anhydride, and acetic anhydride in acetic acid. lay) It does appear that the exchange between thionyl bromide and solvent sulphur dioxide is zero order with respect to the solute. (a<') But a change in order was observed in none of these cases. a,,~ R. E. Jolt',soy, T. H. NORRIS, and J. L. H t STON J. Amer. Clwm. &,c. 73, 3052 (1951). :~'~ E. A. FVA',S, J. L. HUSTON, and T. H. NORRIS d. A m e r . Chem. Soc. 74, 4985 (1952). :~'~ B. J. M,XSIERS and T. H. NORRis J. A m e r . Chem. Soc. 77, [346 (1955). ~) J. 1,. l t u s / o N J. k~ork,. Nucl. Chem. 2, 128 11956), ~:" Equivalent fraciion of a l u m i n i u m chloride : (cquivaIents a l u m i n i u m chloride) -; (equivalents aluminium chloride : equivalents phosgene). Equivalent fraction is used rather than mole fraction in order that concentration should be expressed in terms of the exchanging species, i.e. chloride. See G. FRIEDLANDER alld J. W. KENNI'DY (1955), Nuclear altd Radiockemi.strr, Wiley, New York, p, 315 If. 1i)r quantitative theory of exchange reactions. Rate of exchal~ge ~ a s calculated from the cxpression R~
0-693'a)(b) T.
whereaandharelheconcentrationsofaluminium chloride and phosgene. Values ol" T, w e r e o b t a i n e J f'rom plots of log (/-i'raction exchanged) versus time.
32
JOHN L. HUSTON and CHARLES E. LANG"
10 -~
10 -2
10 -3
10 - 4 .o ° 10- s Ju
iU
10- 2
Equivalent
fraction
10-~
AI CI3
Fro. 1.
l o - ~ L _~=
10
~0
~ ~_=
1() 4 =
1(~5
3'30
3'50
3'70
I/T x t03 FiG. 2.
3'90
33
Aluminium chloride as a solvo acid and Friedel-Crafts catalyst TABLE I.--PHOSGENE-ALUMINIUMCHLORIDEEXCHANGE
T~
R /, I 0
Slope
0-00423 0.00865 0.0396 0.0663 0.136 0-194
405 hr 460 hr 450 hr 420 hr 285 hr 185 hr
0.072 I 0.129 (I.586 1.20 ::.86 5.85
0'96 0"96 0"96 0"96 l '99
0"000503 0"00213 0-00419 0"00921 0-0192 0.0402 00694 0.132 0"173 0.300
48,0 hr 59,0 hr 58.0 hr 60,5 hr 52.0 hr 36.0 hr 25.0 hr 14.5 hr 10.5 hr 7.5 hr
0.0726 0.250 0.499 1.045 2.5 l 7.42 17.9 54.8 94-4 194
0'93 0"93 0'93 0'93 1 "60
0.000098 0.OOO479 0-00109 0'00236 0.00532 0.00570 0.0265 0.0739 0.211 0.417
5.20 hr 5"40 hr 4.70 hr 5.08 hr 3.70 hr 3.47 hr 2.13 hr 1.33 hr 0.57 hr 0.26 hr
0-131 0-616 1.61 3.20 9.90 11.3 84.1 358 2036 6546
1'03 1.03 1.03 1.03 1.34 1.34 1.34 1.68 1.68 1.68
Equivalent fraction
....21"1'
ff
= 25.0'
1 "99
1 '60 l "60 1 "60 1 "60 1 "60
Fig. 2 shows three plots for determination of activation energy, made at concentrations indicated in Fig. 1. It can be seen that good straight lines are obtained at regions 1 and 3 but n o t at region 2 which was deliberately chosen to include both first order and higher order plots. The fact that region 3 does give a good straight line indicates that a single exchange process is taking place in this region, and that the existence of orders intermediate between one and two is n o t to be ascribed to c o n c u r r e n t u n i m o l e c u l a r a n d bimolecular processes involving a l u m i n i u m chloride. It seems that the first order region corresponds at least approximately with the m i n i m u m of GERMANN'S conductivity curves) 61 a n d thus the rising order at higher concentrations may be due to a greater p o p u l a t i o n of ions in these regions, rather than to any specifically chemical effect. Perhaps the nature of the solutions in these regions of first order a n d higher order can be represented as 2A1CI a ~ - [A1C12+, A1Cla-] and (l~} k . F. O. GERMANN
6A1CI a ~
[AlaCI~]+ -~ [AlaCllo]--
J. Ph)'s. Chem. 29, 1148 (1925).
34
JOHN L. HUSTONand CHARLESE. LANG
It m a y be noted that GERMANN'S v a p o u r pressure data indicate aluminium chloride to be partially dimerized in phosgene solution and these equations could as well have been formulated in terms of A12C16.Iv) Phosgeno-aluminates
It has already been suggested 14) that solute aluminiurn chloride behaves toward solvent phosgene as a strong Lewis acid toward a slightly weaker one, the initial "neutralization" to form an aluminium chloride-phosgene complex being rapid, but the actual transfer of a chloride ion involving an appreciable activation energy. The kinetic data here reported are clearly compatible with this picture. In order to obtain collateral supporting evidence we have made a brief investigation of the exchange of chloride between solvent phosgene and chloro-aluminate ions, for example A1Cla-, AI.~Cls=, etc. These ions m a y be regarded as aluminium chloride partially neutralized, and would be expected to exchange at a slower rate with the solvent than does aluminium chloride itself. Table 2 summarizes data on the rate o f exchange of solvent phosgene with chloroaluminate and with mixtures of aluminium chloride and chloro-aluminate, the latter TABLE 2 Phosgeno-aluminates Reactants
1
2 3 4 5 6 7 8
9 10
A1C13= ÷ CaC12= A1Cla ÷ CaC12A1Cla= + CaCI2 A1Cla + CaC12A1CI3= + CaC12 A1Cla ÷ CaC12= A1Cla= + NaCI= AIC13= + NaCI= A1CI3= + NaC1 A1CI3 + NaCI =
Concentration A1C13 A1CI~0 0 0.0113 0.0104 0.0160 0.0193 0.0 0.0 0.0167 0.0169
T~. hr Observed
0.019 0.0398 0.0063 0.0059 0.0199 0.0209 0.0094 0-0352 0.0201 0.0201
180 190 59 62 73 72 1201~ 1520 39 41
A1CI3 alone ~al 55 37 54 54 39 37 59 40 37 37
~) That is, T½would have been this had no ionic chloride been added. ~b~LOWbecause of incomplete reaction. data taken f r o m the previous paper/41 The outstanding feature of these data is the great difference between the half-times reported in each case in the last two columns for runs 1, 2, 7 and 8. Clearly the chloro-aluminate ion, whatever its precise formula m a y be, is not nearly as active towards exchange with solvent phosgene as is aluminium chloride. A n d it is remarkable that there is so little difference between the last two columns for runs 2 and 4 and 9 and 10. In each o f these cases approximately one-half o f the aluminium chloride has been converted to the m u c h less kinetically active chloro-aluminate ion, yet the half-time has been only slightly effected. A n d even in the cases o f runs 5 and 6 the increase in half-time has been much less than one 17) A. F. O. GERMANNand G. H. MCINTYRE J. Phys. Chem. 29, 102 (1925).
A l u m i n i u m c h l o r i d e as a s o l v o acid a n d Friec2el-Crafts c a t a l y s t
35
might expect, considering that the aluminium chloride concentration is here in the region where the half-time is concentration-dependent (the order is greater than 1). It would seem that unreacted aluminium chloride is undergoing exchange more rapidly with the solvent in these solutions than it could if no ions had been present. Of course it is not surprising that in a solvent of such low dielectric constant as phosgene "ionic strength effects" should be very large. In any case, this may be taken as supporting evidence for the hypothesis previously offered to explain the change of order. It may be noted that in the case of run 7 there was apparently incomplete reaction between sodium chloride and aluminium chloride. In this run (which had only a very small excess of sodium chloride over the stoichiometric quantity) the plot of log (1-fraction exchange) versus time underwent a sharp break at about 40 ~ exchange and became almost flat. Runs 1, 2 and 8 gave good straight lines, and it may be assumed that there was substantially complete reaction in these three cases from the very beginning. It is of interest that the unreacted calcium chloride and sodium chloride (solids) of runs 1 and 8 underwent at least appreciable heterogeneous exchange with the solution, the specific activity diminishing in each case by about 11 ~/o. No doubt this exchange was by chloride ions immediately adjacent to the surface of the solid. Since these solids were prepared in just the same way as the solids used in the previous work, ca) it is clear that the failure of solid ionic chlorides to undergo heterogeneous exchange with pure liquid phosgene (containing no solute) was not due to any lack of development of their surface area, but rather to the absence of ionic species in pure phosgene. The great difference in rate between runs 2 and 8 may be due to a difference in solubility of calcium chloride and sodium chloride in phosgene. It is not unreasonable that solid sodium chloride should permit the higher concentration of chloride ions in phosgene and thus permit the equilibrium C1- + A1C13~ A1C1C to proceed further to the right. It may be asked whether we have any positive evidence that the formula of the chloro-aluminate ion is indeed A1C14- and not, for example AlzCl7-. In three runs the solutions were separated from unreacted solid ionic chlorides by decantation, and after removal of solvent phosgene the previously dissolved material was taken up in water for analysis. In all three cases the amount of chloride found was much closer to that expected for A1CI~- than AI~C17- ; this does not of course preclude the existence of such ions as AlzC1 s" or other polymers of A1CI4-Is)
AlCl.~HCl It has been clearly shown that aluminium chloride and hydrogen chloride form no compound (e.g. HA1CI~) of any substantial stability, lu/ At temperatures as low as 120 ° no appreciable amount of hydrogen chloride is taken up by aluminium chloride, the conductivity of a saturated solution of aluminium chloride in liquid hydrogen cs) See Experimental part for details. (9,) H. C. BROWN and H. PEARSALL J. A m e r . Chem. Soc. 73, 4681 (1951). (~") R. L. RICHARDSON and S. W. BENSON J. A m e r . Chem. Soc. 73, 5096 (1951). ,101 M. BLAU, W. T. CARNALL, and J. E. WILLARD J. A m e r . Chem, Soc. 74, 5762 (1952).
36
JOHN L. HUSTON and CHARLES E. LANG
chloride is about the same as that of hydrogen chloride alone, and at higher temperatures the mixed gases show negligible deviation from ideality. However, such exchange experiments as have been done on this system, (10) have left room for feeble compound formation between the two substances. There was substantial heterogenous chloride exchange at 25° and even at --80 °. Probably this heterogeneous exchange should be compared to the heterogeneous exchange between aluminium chloride and carbon tetrachloride,(m there being of course not the slightest indication of any compound formation between these two substances. Yet it is well known that hydrogen halides have a promoting action on aluminium halide operating as a Friedel-Craft catalyst. It has been four~d that aluminium chloride, which has no appreciable solubility in toluene, goes into solution in the presence of hydrogen chloride, apparently by means of the reversible reactions: (9~) CHaC6H5 + HC1 @ ½A12C16~-~ [CHaC6Ha][A1C14-] CHaC6H 5 -~- HC1 -}- AI~CI6~+- [CHaC6Ha+][A12C1c] (Higher complexes presumably can also form). BROWN and PEARSALLhave suggested that it is such complexes which promote the solubility of aluminium halide catalyst and that these complexes play an important role in most Friedel-Craft reactions by furnishing a highly polar medium in which ionic intermediates may form and react. In other words, it might be stated that aluminium chloride can react with hydrogen chloride provided there is some place for the proton to go; some Br~nsted base to accept it. The phosgene-aluminium chloride system is formally similar to the system used by BROWNand PEARSALL, but phosgene certainly is a much weaker Br~Snstedbase than toluene, and by using it we have a more drastic test of the ability of aluminium chloride and hydrogen chloride to react in the presence of a proton acceptor. Accordingly, we have done a few experiments to ascertain the effect of hydrogen chloride on the rate of exchange of chloride between aluminium chloride and solvent phosgene. In Fig. 3 are plotted the ratio T½ (HC1 not present) -- ~ (HC1 present) as a function of the mole ratio of hydrogen chloride to aluminium chloride, concentrations of aluminium chloride being maintained as nearly constant as possible (equivalent fraction _~_ 0.038). The data have no precise numerical significance since we do not have the Henry's Law constants needed to calculate precisely the concentration of hydrogen chloride in the liquid phase, t12) but it is clear that hydrogen chloride suppresses the rate of exchange. Since aluminium chloride is, as we have seen, much more active kinetically toward exchange than chloroaluminate ion(s), it is probable that the mechanism of this suppression is a reversible equilibrium such as A1C1a -? HC1 + COCI~ ~-HCOC12+ q- A1C14Although definite compounds are formed in the toluene experiments of BROWN (111 C. H. WALLACEand J. E. WILLARD J. A m e r . Chem. Soc. 72, 5275 (1950). (12) One very rough estimation of a Henry's Law constant was made during the experiment which corresponds to the point near a mole ratio of 1.0 as explained in the experimental part. The data were used to correct the mole ratios plotted in Fig. 3.
Aluminium chloride as a solvo acid and Friedel-Crafts catalyst
37
and PEARSALL there is no evidence for any such formation here. This, of course, can be attributed to the weakness of phosgene as a Br6nsted base. An attempt was made to calculate equilibrium quotients for the processes ( 1)
HC1 @ COC12 :- AIC1 a ~ [HCOCI~!, A1CI4-]
(2)
HCI ~,--COCI,, @ 2AICI:~ ~ [HCOCI2 +, A12CI7 ]
(3)
HC1 _a COC1 z + A1CI 3 ~_ HCOCI2~ @ AICI 4 -
by means of the assumption that the ratio (aluminium chloride converted to ions) .'(residual aluminium chloride) is the same as the half-time ratio. None of these equilibrium quotients showed satisfactory constancy. The one calculated for process
1.0 i
0"8
0"6 0 f_
0.4
0.2
Moles
1"0 2'0 HCI-- moles
3'0 A[ CI 3
4"0
FiG. 3.
(3) showed the greatest deviation from constancy and certainly the degree of ionization in a solvent of so weak a dielectric constant can never be very substantial. More information on this system can be obtained by determining accurate Henry's Law constants for hydrogen chloride over solutions of aluminium chloride in phosgene, and such a study will be undertaken. Experiments were also performed to ascertain whether the rate of exchange of chloride between hydrogen chloride and aluminium chloride, both dissolved in phosgene, could be measured. It turned out that the rate of exchange was immeasurably fast from 0 ° down to --60 °, which was the lowest temperature at which a gas sample of suitable size could be obtained. At first it appeared that at 0 ° there was a measurable rate of exchange, of half-time approximately one minute, but the expected temperature dependence of this "half-time" failed to materialize, and it was, no doubt, a matter of equilibration between liquid phase and gas phase. It is clear then that the reversible equilibrium by which hydrogen chloride suppresses
38
JOHN L. HUSTONand CHARLESE. LANG
the phosgene-aluminium chloride change is both very rapid and reversible. (The two gases can easily be pumped off, leaving pure aluminium chloride behind.) We consider that our results with phosgene to be one in accord with the results of BROWN and PEARSALLwith toluene and constitute confirmation of their ideas. Before concluding it might be well to discuss briefly a minor point left unsettled at the time of publication of the previous paper. It was observed t4~ that when conventional plots were made of log (1-fraction exchange) versus time, the straight line did not intersect the zero-time ordinate at zero exchange. This was attributed to rapid exchange between aluminium chloride and hydrogen chloride impurity. It now appears that this effect is due rather to a small amount of radioactive hydrogen chloride produced by reaction of aluminium chloride with glass or with water adsorbed on glass, for it has not been possible to eliminate this effect by the most rigorous purification of the phosgene used. The effect has required the use of various modifications in the obtaining of data. At concentrations greater than 0.01, half-time values were obtained from the logarithmic plots and from the final specific activity of aluminium chloride, but at lower concentrations it was not possible to obtain satisfactory plots, because of this effect. Here, at the conclusion of a run, phosgene was fractionally distilled and only the least-volatile sample (presumed free of hydrogen chloride) was used for counting. Then a one-point plot was made to give a half-time value which was averaged with the value obtained from the specific activity of aluminium chloride to give the results reported. At the very lowest concentration (less than 0.0001) there was not sufficient aluminium chloride to give a sample of mercurous chloride large enough for counting, or indeed to permit an aluminium analysis. The half-time had to be obtained by comparison of the specific activity of phosgene (fractionally distilled) at some finite time with its infinite-time value. This latter value also served to furnish the aluminium chloride concentration. In this way we managed to investigate the exchange over a concentration range of a factor of 6000. EXPERIMENTAL
Preparation of materials These preparations differed materially from those previously described 14~ only in one significant detail. Since it was desired to obtain phosgene as free of hydrogen chloride as possible, the following step was added to its purification. It was observed by EPHRAIM(13) that hydrogen chloride forms stable compounds with silver sulphate and with various other sulphates of transition metals. After it had been ascertained that silver sulphate underwent no appreciable reaction with phosgene (silver sulphate in a drying tube showed no appreciable gain in weight when phosgene was passed through it for several minutes) a column was constructed containing silver sulphate and a small amount of metallic antimony, the latter being present to remove chlorine impurity. After evacuation and heating the solids to de-gas, phosgene was passed through at the rate of about 2 litres per hour. Mass spectrometric analysis of the phosgefie showed it to contain no more than 0"01% hydrogen chloride. Frequently, in the course of this work, preparations of radioactive aluminium chloride were obtained with a distinct yellow c010ur. This yellow colour was due to some substance somewhat less volatile than aluminium chloride itself, and for some ila) F. EPHRAIM Ber.58, 2262 (1925).
Aluminium chloride as a solvo acid and Friedel-Crafts catalyst
39
time was believed due to traces of iron in the aluminium metal used for the preparation. It now develops that this colour appears only when the glassware in which the preparation is made has been cleaned with chromic acid solution. If alcoholic potassium hydroxide is used as the cleaning solution this ditficulty is avoided. Whatever the nature the coloured material may be, and its nature is quite obscure at the present time, it could be removed by distillation of the aluminium chloride over hot aluminium metal.
Procedure fi~r maldn£T runs Exchange experiments involving phosgene and solute aluminium chloride were usually performed in all-glass apparatus, fitted with breakoffs and seal points, though some of the earlier experiments were done in apparatus fitted with silicone-lubricated stopcocks. The general procedure has been described. I~/ Those runs where calcium chloride or sodium chloride was added to the aluminium chloride solution were pcrl'oNned in all-glass apparatus. At the conclusion of each run the solution was separated from residual solid by decantation, phosgene was distilled away, the residue from tile solution was taken up in water, and analyses were performed for aluminium and for chloride The solid residue from the phosgene solution was determined simply by weighing. The analytical procedures otherwise used have been described in the previous paper. If the chloro-aluminate ion formed under these conditions is indeed AICIa , than we should expect the number of milli-equivalents of chloride present in phosgene solution to be 4/3 the number of milli-equivalents of aluminium. The 3 sets of data are as follows: chloride expected 0.03;34: chloride found 0-0331. Chloride expected 0.460; chloride found 0.490. Chloride expected 0.801; chloride found 0-851. Better data could not be expected because the decantation afforded only a rough separation. In the first of the series of runs where hydrogen chloride was added to the alunainium chloride solution, the experiment was performed in an apparatus where liquids and vapours were contained behind a stock valve. The apparatus permitted both investigation of the rate of exchange between hydrogerl chloride and aluminium chloride and also of the suppression of the phosgene-ahn-ninium chloride exchange by hydrogen chloride. A rough value of the distribution ratio of hydrogen chloride between liquid phosgene and the gas phase at 25-' was obtained in the following manner: The total vapour pressure over the phosgene solution was balanced in one arm of the stock valve against a source of tank gas in the other arm, the pressure of this external gas being measured by an open-end mercury manometer. Using Gt!RMANN'S data ";/of the vapour pressure of phosgene at 25 ~', it was estimated that the ratio (millimoles phosgene per cc. liquid i millimoles phosgene per cc. gas) - : 150. This datum was used to correct the points plotted in Fig. 3 for the amount (5 ,~--20'!,,) of hydrogen chloride present in the gas phase. Runs to measure the rate of low-temperature exchange between hydrogen chloride and aluminium chloride were performed in apparatus fitted with stopcocks. Runs lo measure the suppression of phosgene-aluminium chloride exchange by HCI were performed in all-glass apparatus at 25 ° , the rate of exchange being calculated solely from the specific activity of aluminium chloride at the end of the run. It was felt that it would be useless to measure for this purpose the radioactivity in the gas phase, consisting as it did of two radioactive substances.