Sept., 1934
THETHERMAL DECOMPOSITION OF DEUTERIUM IODIDE
1885
[CONTRIBUTION FROM THE DEPARTMBNTOF CHEMISTRY, COLUbiBI.4 UNIVERSITY]
The Thermal Decomposition of Deuterium Iodide BY D. RITTENBERG AND HAROLD c. UREY
Shortly after the discovery of the hydrogen isotope of atomic weight 2, we calculatedl the equilibrium constants of the hydrogen-chlorinehydrogen chloride and the hydrogen-iodhehydrogen iodide reactions as a function of the isotopic composition of the hydrogen. These calculations showed that a marked dependence of the values of these constants on isotopic comgosition should exist. In the case of the first seaction, the percentage dissociation of deuterium chloride should be less than that of protium chloride, while in the case of the second reaction, the equilibrium is shifted in the opposite direction with increasing deuterium concentration in the reaction substances. The theoretical predictions based on well-established theory indicated such marked differences in chemical properties that we were led to conclude that marked variations in the properties of the hydrogens and their compounds were certainly to be expected. It is the purpose of this paper to report the experimental confirmation of the theory in the case of the hydrogen-iodine-hydrogen iodide reactiom2 The thermal dissociation of hydrogen iodide was extensively studied by M. Bodensteina from 302-516'. He found that the dissociation, in this range was reversible and equal to about 20-25%. We have repeated two of his values, and are able to check his results within the limits of both his and our experimental error. Experimental Materials.-Deuterium rich hydrogen was prepared either by the reaction of water with a magnesium amalgam or by the electrolysis of water containing a high concentration of deuterium oxide. Hydrogen iodide was prepared from its elements: 35 g. of iodine was placed in a five-liter flask containing a platinum catalyst. The vessel was evacuated until all air had been displaced by iodine vapor. Hydrogen was pumped in with a Toepler pump to a pressure of 12 cm. of mercury, and the flask sealed off. It was heated in an air bath at 370' for six hours. More than 90% of the hydrogen reacted. This was purified by distillation before use as will be described below. Thermostat and Reaction Vessel.-A diagram of the thermostat and reaction vessel is shown in Fig. 1. The reaction vessel was kept at constant temperature by im(1) H. C. Urey and D. Rittenberg, J . Chcm. Phys., 1, 137 (1933). (2) A preliminary note reporting these results has been published in the J . Chcm. Phys., 8, 106 (1934). (3) M. Bodenstein, Z . Bhysik. Chcm., 18, 56 (1894); 22, 1 (1597); 29, 296 (1899).
mersion in the vapor of a liquid boiling a t constant pressure. Two liquids, mercury and sulfur, were used t o obtain a wide temperature range. A device to hold the pressure constant over the boiling liquid was developed in this Laboratory and independently described by C. C. Coffin.* It consists of a 100-watt heater placed in a five-liter bulb connected to the thermostat. The heater current is controlled by a relay actuated by a pair of contacts on a mercury manometer. When the pressure falls below the predetermined value, the heater is turned on and increases
!
IP
H-
D
J
All measurements In niillimcters
Fig. 1. the pressure. When the pressure rises, the heater is shut off, the gas in the five-liter flask cools and the pressure in the system falls. The fluctuations in pressures were about 0.2 mm. I n the range of temperatures used, the fluctuations correspond t o about 0.008' for mercury and to about 0.014' for sulfur. The pressure fluctuations have a period of about two to three seconds and since it is scarcely possible for the temperature of the thermostat to follow such rapid changes, the fluctuation in the temperature of the reaction vessel was probably considerably less than the above figures. (4) C. C. Coffin, THISJOURNAL, 66, 3646 (1933).
A is the outer j a e k t of &e rttermwtat in' w&i& the liquid is boiled by the 250-watt heater, C. B is the reaction vessel. The entire.thennastat' u*as-plaed'iii ttie iron can, D, and then packed with diatomaceous earth. The diatomaceous earth packing served' tapi d o u b i a f & a t b of; a thermal insulator and of a rigid support for the thermostat. The r&uxing liquid, was$ condensed by. the air condenser, Qj which leads to the presura control device. F is a thermometer well. Gfand H ara 1-mm. capillary tubes through which the reaction.vesse1isiilledand emptied, The temperatures were measurzd by a chromel-alumel therinomqde which had.been stanilardized6 against boil. ing. water,.mexury and sulfur. A- vertical seation of the thermostat outside the reaction v-1 was,explored with the thermooouple for temperature uniformity. The.thetmocouple was read bs means of a Lee& and Northrup type K potentiometer. The maximum difference between the,readings was 0.015 m n , corresponding tor0.35'. The fluctuations were random andldid not indicate any superheating of the vapor. Further, the temperawe in the thermometer well was observed' and found to, agree closely with the temperature autbide the reaction vessel. 1
Fig. 2.
Experimental'Procedure Figure 2 is a diagram of the apparatus used in purifying the hydrogen io& aad in Wing the reaction vessel. The five-liter flaskt; A, i9 which the hydrogen iodide had been prepared wasisealed to stopcock D. The entire system was tEen evacuated. Stopcock D was dased and the break seal, B, crushed with the magnetic hammer C. E was immersed in liquid.&, after which A was slowly pumped out. When the hydrogen iodide had all been condensed in E, .the system was sealed off 'at 1, and evacuated'through R. After H was clbsed, F was immersed in liquid air and E in solid carbon dioxide-alcofiol'mixture. The vapor pressure of liydkogen iodfde is about 7'cm. at -79"' and it therefore rapidly distiIs over into F. Tlie hydrogen iadide after tliis dktillition was pure white. The system was seakd off at 2 and'pumped out through' HI To determine the concentration of deuterium iodidi in the hydrogen iodide, a sample of*hydrogen iodfd6 was secured'at this point b y sealing off' the one-liter Bask J ( 6 ) P. D.Foote, C. 0. Fairchild and T. R. HarrisoB, *'firem&& Practice," U. S. BIm~&Fit&P.dr, TcchmsidgbPspcr No. lm.
ae alont 4U om.p w w e .
'Plm hydrogen iodide was again condensed in F. In sealing off J, the hydrogen iodide in t&? &@a@' bet%Wen3: a&t€km&in line was decomposed, but due to the small volume of the capillary, the amount deoempbsed'was &la& The hydrogen was pumped out and the iodine sublimed back into F. Following the same procedure as above, the hydrogen iodide was sublimed into the trap G. The system was sealed off at 3, pumped for thirty miflutes a d ' sealed off at 4. The hydrogen iodidk i n 4 a t tliis stktge waspure wiiite, showing no signs OPiodine. Ta~admibf%eh y w g e n iodide to the reaetibn wskl, the liquid air wes rernaed from trap G and the pressure allowed to,rise t o about 60 cm. The capillary tube leading to the reaction vessel was then sealed off at 6. n e seal'was niadp a3 close-t o the thermosthf as possible. The t2iepm&sseaEtl was set ihtoi operation and mm until equilibrium had k * a t t i a i n e & I n general, the time of heating was from 3 to 40 times more than that required by ordinary hydrogen iodide t o come within 1 part in zD0b of e q d i b r i m . The h e t i c data of fi.€%ddenst'ein* w usrd iri calcttlatiohs. Wan eqt?iHbriw htrd been e s t e M i h d , the reaction vessel was pumped out through the system shown irl Fig. 3; 40 cc: of a 574 sebtitm of p o t a s s a io&& deaerated1 Ebnd frozen in H by a*d i d carbon dioxide-alcohol bath. Tkie function of the trap, L, @as to pt-event fl@Fltrjr mpbf f?m*the Toe@r ptmfp ff6Sf1 dshsing babkhihto the trags, F, G, HJ and J. The entire system was evacuated through C. #, G ahd 5 were nier*d%ff lipaid a h S t t p cwkX waWtclW, matic Toepler pumps M,st&ed,.and the'break&al B was crushed by the magnetic hammer. The rate of flow of the gases %?iss!med"down by tlie tapmary leak, e,wkddr uhder a pressme drop-of'oirestniosplferepassedabant 2 cc. of gas pwsecoird. T'lie l'$&ugen iadtde and iodine were frozenaut iff the Uiaps E: G and' J: and t&e &&age& -ped into the bulb N. When t&e rate d%owof the gases had slowed d6m (as'stkwnl& the mmt orndrugen pufnped diuing a' s m k E of tW TdepTer pmnp), the break seal D w%w'ctuslfed: D ms a l&?ljreal s&l with an openfng of- 4'mm. w h W wfmr crushed allowed tlie system to be pumped out errsUjr. W e n most of' tlie hydrogen' had been pumped'into N,the system W&J sealed ofl at 1 and 2. m e remainder of the €iT&gen was remwed thkough'P by the HigW vacutim pnmps. ?he system was a l l m d * t o stand overnight to permit dl'ibdine wklich had condensed in t& tulves ICadtrrg fhtn B to F to sublime over tb F. The traps, G and J, were used for protection against loss of i d n e or hydmgeniodidein the first surge of ges when B was crushed. No condensate, h m v t r , was ever sewlin (6) D. MacGillavry, "Dissertation," Columbia Unberdty; 1938.'
THETHEBNAL DX~XBMROSISIQIR @aDEUTEBWM IODIDE
sept 1934
them. Neither the-m w u r y irr the. Z'&qlas.g ~ m pnor t h e dronlets of. rnwhish were put. into N s h w d any signs of reaction with hydrogen ibdide. After standing overniglit, the system was sealed off at 3 a i d 4 and the i d n e and l&dmgen iodlde diSt&d'm& J. It tlreo sealadd.at5' 'Dhe water in'H wss allowed melt and tlie h y k q p n ~ iodide and i d i e diesolved br it. The solution W a S W a s h e d back a d forth H s and J until it was of uniform comgosition. The equilibrium was not shifted dking the proct:ss of pumping out the reaction vessd, since it is not &cted b y pressure. The rate of reaction at?3Wa is va-ysmdl~'and since the time mquired for &e g a w to eool.tD WP is-also very small, it is highly improbable that the eqllilibrium proportions changed while the ms was being pumped out.
Apalyticd Metfiod.-The qwtity which. we seek to evaluate from our analytical data is c, the fraction of hydrogen iodide decomposed a t equilibrium, namely
t
= 2(1e)/2(12)
+ (HL)
where (12) is moles of I2 present a t equilibrium, and (HI) is moles of HI present a t equilibrium. Two different analytical methods were used. Method I-The iodine was titrated with an approximately 0.1 N s d u m %sulfate &atfm. T3e hydmgeniodirfir-A determined in the same solution bx titmtion with 0.3 L?A sodium hydroxide. Both titrations can be done in the same solution since tlie titration ofiodine. by sodium t t l i d f a t e a k s no$ afFect ttle hndmscn+ ion conaentratim All titra-: tions were done with.a weight buret, The 0.1 N sodium thiosulfate solution.was standardized against pure iodine.' Tlte starch indfCator was made up fresh every day from a sample of1 fine& ground. pcttata star&$ 025 g, of star& was added!to,26~ccof water, &I 75". This was agitated until sdbtion ww.oornpkte,and~ rapidly cooled to room temperature. Thesodium&ytta-m ide solution was standardized against potassium acid phthalate. The blank was 0.3 cc. of 0.015 N?iydrogen iodide. Both the sodium hydroxide and; the sodhnn tltiosulf&te d t i o n s were restandanlizsd! t k . h y B&am th-y used. The analysis of known s o ~ u ~ o nshowed s th;aR in mixtures of iodine and hydrogen chloride the iodine could be determined with a precision&€ 3,parts in 10,000 and the H + to within 10 parts in 10,000. Metkod.2.-TJ~ free iodine wfts titratedrdth approximately 0 1 N sodium thiosulfate. Two grams of potassium iodide was added and after fifteen minutes the liber(7) TreadwdI and Hall. "AadFticai Chemistry," VoL 11, p 552. 6th ed.
1887
atsd iodine wgsr titmted.*> It i3 clear that the sodium thiosul€ate dws not b v e to be standardized sinoa t is given br Weight of &&Os d u h n used to titrate free iodine = Weight OF ma,S;O, solution used in both titrations The analysis of known solutions of iodine and hydrosaid by;i ttfis,metfmd ratia z(~)./(zfr4 w&h,atpregi&n,of 8 parts + could be
(sa))
in 10,000.
,&
Indl determinations of iodine, the end-point was overu,oo5 N iodne solMim titrated mebpanidmmeftion,was of o.m5
~= sorutton,
*
iodine. The bast. end-points, as W a s h b e werep&
3heoretid C:alculationcof"r.-The theoretical investigation1, gives1 &e reidonship between the fcation of hydrogen iodide deemposed a$ e q ~ l i b r i u m fthe wncad'ox Of deuterium iodide in the original. hydmgen. iodide, the concentration of deuterium iodide in the hydrogen iodide remaining a t equilibrium, and the concentration of deuterium in the hydrogen pro-
Fig.. 3.
d k e d by the decomposition of tfle fiydfogen iodide. We may experimentally determine any pair of these qpantities and compare the relationship Between them with that tlieoretically predicted. At equilibrium there are four equilibria involved a
g + d12"+%Hr d h Dz + Iz -2DI b d~ g
Hz
(3)
c
HD f Is,+-HI a
Hi
c
+ DI+
(4)
k f DI
6
2HD
(6) (6)
The concentration in moles of the varims mlecular species at equilibrium will be denoted by the letters immediately above them. Let &, (8) Kolthoff and Furman, "Volumetric Analysis," Vd. 11, P. 38%: I. Mc k5ldhho5, Phslrm, W&hW 67# 68.1(1m). (9) E. W. Washburn, THISJOURNAL, SO, 31 (1808).
D. RITTENBERG AND HAROLD C. UREY
1888
Vol. 56
K2, Ka and Kd be the respective equilibrium con- These values have been confirmed experimentstants of (3), (4), (5) and ( 6 ) . Only three of ally.ll By interpolation and extrapolation we the four equilibria are independent since Ka can have found that the values K I / K and ~ K4 for the be expressed in terms of K1,Kn and Ka temperatures which we used are for 398O, 1.225 and 3.781 and for 46S0,1.215 and 3.823, respectively. Ka = KiKa/K4 From these equilibrium constants, it is possible The values of K1, KB and Kq have been calculated previously by the authors, for several temperatures. If we experimentally determine 5 and the concentration of deuterium iodide in the original hydrogen iodide, we can set up six independent equations which will permit us to solve for the six unknowns, a, b, c, d, g, h. Let x , y and u be the number of atoms of H and D and I, respectively, in the system. From the conditions of our experiment, x y = u. Then
+
g’
- Ktad
gh b2 2a 2c 2d
+ + + + +
0 0 0 b -j- g = x b h = y g h = u
(7) (8) (9) (10) (11) ( 12)
- Kabd = - K ~ U C=
where equations (l), (11) and (12) express the condition that the number of atoms of each kind is constant. Three unknowns can easily be eliminated by use of equations (lo), (11) and (12) to give
to calculate the fraction of the hydrogen iodide decomposed as a function of the isotopic composition and temperature. If we determine { and the concentration of deuterium in the hydrogen produced in the reaction, we may set up a set of analogous equations which will give us { in terms of the atomic concentration of H and D in the hydrogen instead of being given in terms of x, y and u as in the previous case. The equations are
-
gE &ad = 0 gh Kabd = 0 ba K4ac = 0 a + b + c - d = 0 2a b 2(a b c) = a 2c 4- b 2(a b c) =
-
+ + + + +
( 16) ( 17) ( 18)
’
where a! and are the atomic fraction of H and D, respectively, in the hydrogen. These equations g’ -- ( K z d / 2 ) ( ~ - g b) = 0 (13) may be solved in the same manner as the previous g ( U - 2d -- g) - Kabd = 0 (14) b’ -. (K1/4)(2d g y - b - U) = 0 set. (15) I n practice the uncertainty of A t calcd. due to These three equations can be solved by straightthe method was reduced to less than 0.00004 by a forward methods since they contain but three suflicient number of approximations. unknowns, but only with very great difficulty. The concentration of deuterium in ExperiWhen two of the unknowns are eliminated, one ments 2 and 4 was determined by Dr. Walker gets an equation of such complexity that i t is Bleakney on his mass spectrograph.12 The practically impossible to find the roots. The concentration of deuterium in Expt. 5 was deterequations (131, (14) and (15) can be solved, however, to any degree of precision by Newton’s mined by burning 400 cc. of the hydrogen iodide method of approximations.l0 Whittaker and used in the run and determining the concentraRobinson illustrate the method for two un- tion of deuterium in the water by the interknowns, but i t may easily be extended to three. ferometric method of Crist, Murphy and Urey.13 The ratio of the equilibrium constants cal- The interferometer was calibrated with samples of water whose isotopic composition was deterculated are given in Table I. mined by measuring the density and assuming TABLE I that the density of deuterium oxide is that given T,OK. 0 298.1 400 575 700 by H. S. Taylor and P. W. S e l ~ o o d . ’ ~ KdK2 0.0 1.164 1.212 1.234 1.222 A summary of the results of our experiments is Also, the values of the equilibrium constant given in Table 11. K4were calculated to be (11) D. Rittenberg, W. Bleakaey and H. C. Urey, J. Chem. Phys.,
+ +
T,OK. K4
298 1 3.269
400 3.494
575 3.710
700 3.800
(10) Whittaker and Robinson, “The Calculus of Observation,” p. 90, 2d ed.
2, 48 (1934). (12) W. Bleakney, Phys. Rcv., [Z] 41, 32 (1932). (13) R. H. Crist, G. M. Murphy and H. C. Urey, J. Chem. P h y s , 2, 112 (1934). (14) H. S. Taylor and P. W. Selwood, THII JOURNAL, 56, 998 (1934).
Sept., 1934
STANDARD POTENTIAL OF THE
FERRIC-FERROUS ELECTRODE
1889
TABLE I1 The results of our experiments are given in this table. A t is defined as the change in I caused by the presence of deuterium. The last column gives the time necessary for the HI to come within one part in 2000 of equilibrium calculated from M. Bodenstein's data. The values for experiment 5 differ slightly from those previously reported2 because of a more exact calibration of the interferometer readings used in the analysis for deuterium. Expt.
T,'C. 398 398 468 468 468
Concn. D
t Expt.
I Calccl.
....
A{ Expt.
T i m e of run
A t Calcd.
Time for HI to come to equilibrium
1 0.0% 0.20719° .... .... 70 hours 26 hours 2 14.3d .20838 0.20894 0.00119 0.00175 144 36 3 0.0 .22724b .... .... .... 30 0.9 4 41.7d .23287 .23326 .00563 .00602 14' 0.8 5 75.5" .23877 .23904 .01153 .01222 41 0.9 Bodenstein's value for at this temperature is 020703. Bodenstein's value for a t this temperature is 0.22772. Concentration of deuterium in the hydrogen iodide. Concentration of deuterium in the hydrogen produced by the decomposition. e The reaction vessel was kept a t 500" for two hours at the beginning of the run.
r
Precision of Results.-We estimate that the it would lower our calculated values of [ by about errors in the temperature a t which the equilibria 0.00050 for the 75.5% sample, by 0.00030 for the were measured to be 1' for the lower tempera- 41.7% sample and by 0.00007 for the 14.3% ture and 3' for the higher. No attempt was made sample. This would bring our experimental to determine the temperature more precisely values into slightly better agreement with the since the rate of change of AT with temperature calculated ones. is small. For run 5, d(Af)/dT = 0.000017. Summary We estimate the accuracy of { to be O.OOC60 The percentage of thermal dissociation of mixfor runs 1 and 2 and to be 0.00040 for the re- tures of protium and deuterium iodide has been mainder. Our experimental results agree with determined at 398 and 468O, and compared with the calculated ones to this precision. It is to be the theoretical predictions. They are found to noted that all our experimental results are lower be in agreement within the limits of experimental than the calculated ones. Our results do not error. A determination of the equilibrium conhave a sufficient precision to warrant any con- stant for the protium iodide decomposition a t clusion in regard to the reality of this deviation these two temperatures agreed within our limits between experiment and theory. If, however, of experimental error with the values given by instead of using our formula (11), we had used Bodenstein. formula (I) for the energy states of HD and Dz") NEWYORECITY RECEIVED JUNE 9, 1934
[CONTRIBUTION FROM THE CHEMICAL L.4BORATORY OF THE UNIVERSITY OF CALIFORNIA]
The Hydrolysis of Ferric Ion. The Standard Potential of the Ferric-Ferrous Electrode at 25'. The Equilibrium Fe+++ C1- = FeCl++
+
BY WILLIAMC. BRAYAND ALLENV. HERSHEY Dilute aqueous solutions of ferric salts are known to contain considerable amounts of colloidal ferric hydroxide and hydrogen ion, but the possibility of isolating the first step in the hydrolysis Fe+++
+
H20
=
FeOH++
+ H+
(1)
has received little attention. Thus, although in 1907 Bjerrum,l from a study of the conductivity of 0.025 to 0.0003 molal ferric chloride solutions, calculated the equilibrium constant of this reaction to be K , = 25 x 1 0 - 4 a t 250, chemists Seem to (1) Bjertum,
2. physzk. Chcm., 69, 350 (1907).
prefer the theory that FeOH++ is unstable with respect to ferric ion and the colloid. In other words, it is usually assumed that equilibrium lies far to the right in the reactions 2FeOH++ = P e + + + 4- Fe(OH)2+, and (lb) FeOH++ -t Fe(OH)z+ = Fe+++ 4- Fe(OH), (colloidal) (IC)
Our approach to this Problem is based upon the following ideas. when the hYaOlYSiS of a ferric salt is Padually decreased by the addition of acid, the quantities of colloid and Fe(OH)z+ must decrease more rapidly than that of FeOH +';
