KINETICSOF DECOMPOSITION OF CHLOROFORMIC ACID
1803
conditions the transmitted radiation, X > 2800 A, is not absorbed by L-cysteine and is absorbed by hydrogen peroxide with the formation of hydroxyl radicals. In the absence of cysteine, the spectrum obtained is that typical of irradiated hydrogen peroxide, decaying with increasing temperature and disappearing completely in a few minutes at 150°K. In the presence of L-cysteine, a broad line with little evidence of structure was obtained at 77°K. On raising the temperature to 165'K the spectrum shown in Figure 7 was obtained. Although not identical with that obtained from pure solid cysteine, the similarity, particularly when compared to Figure 6b, is good evidence that thiyl radical
is formed. Thus, the secondary reaction is the abstraction of the sulfhydryl hydrogen atom by hydroxyl radical. These results confirm conclusions reached by HenriksenI6 from studies of X-irradiated aqueous solutions of L-cysteine. It may be emphasized that a major difference in these studies is that the X-irradiation of thiols in water yields both hydroxyl and thiyl radicals as primary irradiation products, whereas ultraviolet irradiation at X > 2800 A in solutions with hydrogen peroxide yields hydroxyl but not thiyl radicals as primary irradiation products. (16) T. Henriksen, J . Chem. Phvs., 38, 1926 (1963).
Kinetics of Decomposition of Chloroformic Acid
by Rapid-Scan Infrared Spectroscopy
by Reed J. Jensen and George C. Pimentel Chemistry Department, University of California, Berkeley, California (Received December 9, 1966)
The 768-cm-' absorption produced by flash photolysis of chlorine-formic acid mixtures has been reexamined under higher resolution and its kinetic behavior has been studied as a function of temperature. The assignment of this band as the G C 1 stretching motion of chloroformic acid is corroborated by the band width, 17.5 cm-', and the absence of frequency shift on deuteration. The decomposition is unimolecular over the temperature range 288-343°K and the rate constant equals 5 X 10la exp( - 14,00O/RT). The implied activation enthalpy and entropy indicate that the rate-limiting step is probably the cistrans isomerization of chloroformic acid. Bond length and energy considerations indicate that bromoformic acid will be similarly labile.
Introduction
COOH
West and Rollefson' studied the gas-phase photolysis of chlorine and formic acid and postulated the following mechanism to explain their observed quantum yield of several thousand. hu
Cln
--f
2c1
(1)
C1+ HCOOH
--f
HC1+ COOH
(2)
+ Cl2
--f
ClCOOH
+ C1
+ COS COOH +COZ + H H + Clz HC1 + C1 2C1 + M +Cl2
ClCOOH +HC1
4
(3)
(4) (5)
(6)
(7)
(1) W. West and G. I(. Rollefson, J . Am. Chem. SOC.,58, 2140 (1936).
Volume 71, Number 6 May 1967
REEDJ. JENSEN AND GEORGE C. PIMENTEL
1804
Herr and Pimente12z3 studied the reaction intermediates of the above reaction by rapid-scan infrared spectroscopy. They detected, at 768 cm-', the absorption of the C-C1 stretch of the ClCOOH molecule. The half-life of the unstable molecule at temperatures typical of their reaction conditions was estimated to be 50-70 psec. The present work further explores, by rapid-scan infrared spectrophotometry, the kinetics of the decomposition of this intermediate species in the chlorine-formic acid photochemical reaction.
Experimental Section The formic acid in this study was Baker and Adamson 98 to loo%,.the chlorine was Matheson 99.5%, the deuterium oxide was Bio-Rad 99.94%, and the calcium sulfate was Hammond indicating CaS04 (Drierite). The spectral region near 800 cm-l was studied at two spectral slit widths, 8 or 9 cm-1 and 40 cm-I, using the rapid-scan infrared spectrometer built by Herr and Pimente12g3 and modified by C a r l ~ o n . ~The instrunient n - used ~ with an NaCl prism, Nernst glower source, and a zinc-doped Ge detector at 20°K. With the 40-cm-1 spectral slit width, the scan rate was 900 cm-'/100 psec. For the smaller spectral slit width, the rotating Littrow mirror was replaced by a 40-line/ mm grating. With this grating and a scan rate of 200 em-'/] 00 psec, the 9.4-cm-1 interference fringes of an ,4gC1 film were resolved. For study of the spectral region near 1800 cm-', a IOO-line/inm rotating grating and a mercury-doped Ge detector at 20°K were substituted. Resolution of water lines indicated that the spectral slit width was about 5 cm" at scan rates of 1000 cm-'/100 psec. For photolysis, a bank of condensers (0.5, 9, 17.5, or 32.5 pf) ,charged to voltages in the range 7 to 18 kv was discharged through a 58-cm quartz flash tube filled with 20 mm of xenon. The flash duration (at half-height) mas about 30 psec for low capacitance and 60 psec for the higher capacitances. Throughout this work, a Pyrex reaction cell was used, so that very little light of wavelength less than 3100 A was admitted. The nunzbcr of photons in the effective region, 3100 to 3900 A, was about 40 pmoles per 1000-joule flashe5 The cell and flash tube were optically coupled by wrapping them together in aluminum foil. Tempera1 ure measurement and control were important in this experiment. Fortunately, the formic acid monomer-dimer equilibrium furnishes temperature information through the spectral intensities. About 40 psec after each measurement of the 768cm-' band of ClCOOH, the spectral scan encompasses the !tl7-cm-' band of formic acid dimer and 30 The Journal
01'Physical
Chemistry
psec later, the llO5-cm-' band of formic acid monomer. With the assumption of Beer's law, the following expression can be derived
TiAH
T2 =
r2
AH - RT, In -
T1
where ri = (Di/flflz),Di is the optical density of the dimer band at 917 cm-l at temperature T I ,Mi is the optical density of the monomer band at 1105 cm-I at temperature T I , and AH is minus the heat of dissociation per mole of dimer ( = - 14.5 kcal/mole6). This temperature calculation also tacitly assumes that equilibrium is rapidly established within the 20- or 30-psec period following the flash and preceding the spectral scan. This is undoubtedly so, since the halftime for dissociation of the dimer at 300°K is of the order of 10-l2 sec if AF for bond rupture is close to AF of dissociation of the dimer (1.64 kcal/mole).6 We see, then, that Tz can be estimated from a reference spectrum at room temperature, TI, before the flash and a measurement of the optical densities at 917 and 1105 cm-I at the unknown temperature. The applicability of this temperature T z in the present work depends upon one more factor. The temperature rise after cessation of the flash must be small enough so that the temperature is essentially constant (and equal to 2'2) for the period between flash termination and the recording of the band at 768 cm-l. For kinetic studies below 40", the heating that accompanies the flash sufficed to warm the gas to the desired temperature. Variation of flash energy and Clz pressure provided control. For reaction temperatures above 45") the cell was heated externally to a temperature 10 or 15" below the desired temperature. Thus the temperature rise due to flash heating and subsequent reactions never exceeded 13". Since a large fraction of this rise must have occurred during the flash, the error involved in the use of T z is at most a few degrees. In a typical kinetic experiment, a mixture of 20 mm of HCOOH, 20 mm of Cls, and 720 mm of S, was photolyzed with a 1300-joule flash to give a 10 or 15" temperature rise. Doubling the chlorine pressure increased the temperature rise to 40-50". If the chlo-
*
(2) G. C. Pimentel and K. C. Herr, J . Chim. Phys., 61, 1509 (1964). (3) K. C. Herr and G. C. Pimentel, App2. Opt., 4, 25 (1965).
(4) G. A. Carlson, Ph.D. Dissertation, University of California, Berkeley, Calif., 1966. ( 5 ) Based on actinometry performed by G. A. Carlson. (6) G. C. Pimentel and A . L. hZcClellan, "The Hydrogen Bond," W. H. Freeman and Co., San Francisco, Calif., 1960.
KINETICSOF DECOMPOSITION OF CHLOROFORMIC ACID
1805
r h e pressure were raised to 250 mm and Nz pressure dropped to 500 mm, the cell would explode. Time measurement was entirely based upon the time scale of the Tektronix 5358 oscilloscope horizontal sweep. The 5@psec/cm and 20-psec/cm sweep scales were used and each was calibrated to *l% with a Tektronix RM187 time mark generator. Time was measured from the beginning of the flash since this time is well defined, and for first-order kinetics, the error involved affects the intercept but not the slope. The slope, of course, fixes the rate constant.
Results Identification of CICOOH. The 768-em-'
C-C1 stretching mode of 'ClCOOH was studied under higher resolution than was available in the earlier work2,g (about 20 cm-') to seek band contour verification of the identification. Figure 1 shows this band under the optimum conditions that could be obtained in this work with a spectral slit width of about 8 cm-I (trace a). Trace b in Figure 1 contrasts the spectrum of formic acid before the flash with a superimposed interference spectrum generated by a piece of AgCl sheet placed in the optical path. The 9.4-cm-1 spacings of the fringes furnish calibration marks and a qualitative measure of the resolution. An optical density plot of trace a shows that the band width at half-height, under this resolution, is about 17.5 f 1 cm-I. The uncertainty is connected with the difficulty in fixing the Io line and the zero transmission base line. Because of the chlorine atom mass, chloroformic acid is approximately a prolate symmetrical top and the axis of least moment of inertia is close to the carbonchlorine bond. Thus the C-C1 stretch is expected to exhibit the PQR structure of a parallel band. With reasonable bond lengths and angles, the value of '12(B C) is about 0.12 cm-l. At 300"K, the PR peak spacing is expected to be near 14 cm-l. With an 8-cm-' spectral slit width, it is unlikely that such a fine structure would be resolved. Furthermore, the measured band width of 17.5 cm-l is reasonably close to the expected value, certainly consistent with the interpretation. The 1600--2000-~m-~region was carefully examined for evidence of the carbonyl stretching motion of chloroformic acid. The HCOOH pressure was varied from 1.5 to 20 mm with a Cln pressure of 25 mm and an N2 pressure of 700 to 730 mm. The spectral slit width in this region was 5 em-'. No trace could be found of a carbonyl absorption outside of the region of the formic acid bands, 1690-1800 cm-I. There was, however, evidence that chloroformic acid may absorb within this region, perhaps near 1800 and 1760 cm-', though
+
0
'/a
tu
/20
/co
Cfyec) Figure 1. Band contour of 768-cm-1 band; spectral slit width, 8 cm-l: a, 768-cm-l band immediately after flash; time measured from flash trigger; b, before flash, formic acid and interference fringes from AgCl sheet; time measured from sweep trigger.
the interpretation is difficult if any temperature changes occur because of attendant displacement of the formic acid monomer-dimer equilibrium. As one further verification of identification, 18 mm of 9701, HCOOD, 25 mm of Clz, and 715 mm of Nz was photolyzed in the usual way. The transient spectrum was indistinguishable from that recorded in similar experiments with HCOOH. The absence of a noticeable frequency shift is consistent with the assignment of the 768-cm-I transient to a G C 1 stretching motion of chloroformic acid and it eliminates the possibility that the motion is connected with a hydrogen bending motion in such a species as carboxyl radical. Choice of ConditionsJor Kinetic Studies. Reproducibility was, a t first, quite a problem. It was discovered early that unless air was excluded and the sample was carefully dried with anhydrous CaS04 as the cell was filled, the transient absorption intensity was variable. Furthermore, after a replacement of a section of vacuum line or cleaning of the photolysis cell, the transient was diminished or lost until several experiments had been performed, after which the transient was again reproducible. The effect of flash energy cin intensity of the 768cm-I transient was investigated for two types of mixtures, chlorine in small excess (14 mm of HCOOH, 23.8 mm of C12, and 722 mm of N2) and chlorine in large excess (13 mm of HCOOH, 195 mm of Clz, and 550 mm of Nz). I n the latter case, the intensity of the 768-cm-1 band was found to increase to a maximum a t about 500 joules and then decrease a t higher Volume 71, Number 6 May 1967
REEDJ. JENSEN AND GEORGE C. PIMENTEL
1806
flash energies, diminishing by about 30% at a flash energy of 1150 joules. At still higher flash energies, cell destruction is likely to occur. The transient can be detected only for one or two flashes of a given mixture, indicating that the chains furnished by reactions 2 and 3 art? long enough t o consume the formic acid. Thus the high chlorine mixtures are quite unsuited to kinetic studies: the amount of transient depends sensitively upon flash intensity, there is a large temperature rise, and the mixture must be replaced after each flash. For chlorine in small excess (ClZ/HCOOH = 1.7) the transient intensity was constant and reproducible within ;yo over the flash energy range 400 to 1200 joules. Furthermore, the amount of heating was small arid reasonzbly constant over this flash range, about 10". JIost important, it was possible to flash a given sample five or six times and obtain the same optical density of the SG&cm-' transient repetitiously if delay time was held constant. The small temperature rise and the repeated appearance of the transient at constant intensity show that the chains cannot be long enough to vonsume a large fraction of the formic acid. Furthermore, the aniount of transient produced by the flash must be relatively independent of the more important experimental variables : the flash energy, the forniic acid pressure, and the chlorine pressure. I n view of these favorable features, all kinetic studies were conducted with chlorine in small excess and with flash energies ne:w 1000 joules. Kinetic Study. Figure 2 shows the optical densitydelay time data recorded in a typical kinetic study
of a low-excess Clz mixture. The five data points were obtained from a single filling of the sample cell and with delay times varied nonsystematically : 49, 170, 75, 100, and 135 psec. The linearity of the plot indicates that the 768-cm-' band is lost in a first-order process, presumably reaction 4. The slope gives the rate constant 1.1 f 0.5 X lo4 sec-', and the formic acid bands indicate that the reaction temperature was 308 A 2°K. Similar kinetic studies were conducted at nine temperatures in the range 288 to 343". The data, collected in Table I, provide a basis for a plot of log k - log T us. T-l, as shown in Figure 3. The leastsquares straight line shown has a slope of -3.0 & 0.4 X lo3. This line gives an activation enthalpy, A H * = 13.8 & 1.6 kcal/mole ( 2 u ) and activation entropy
Table I : Kinetic Data: Decomposition of Chloroformic Acid Log OD transient
48 f 2 71 f 2 83 i 2 110 f 2
-0.39 i 0 . 1 -0.94 f 0.08 -1.40 f 0.08 -3.00 0.8, - 3 . 0
5.04 f 0 . 2
338 f 2'
68 f 2 95 i 2 130 i 2
-0.78 i 0 . 1 -1.01 i 0.08 -2.73 f 0 . 4
4.86 f 0 . 2
323 i 2'
60 f 2 109 i 2 140 f 2
-1.06 f 0 . 1 -1.17f0.08 -2.11 i 0 . 2
4.61 f 0.15
318 f 2"
70 i 2 103f 2 131 f 2
-0.79 i 0 . 1 - 1 . O O i 0.08 -1.24 f 0 . 1
4.32 f 0.15
311 f 2"
70 f 2 109 f 2 135 f 2
-0.78 i 0 . 1 -0.80f0.08 -0.95 f 0.08
4.15 f 0.15
308 f 2"
49 i 2 75 zk 2 1OOf2 135 f 2 170 + 2
-0.72 i 0 . 1 -0.82 i 0.08 -0.91fO.08 -1.14iO.08 - 1 . 3 1 1 0.08
4.04 f 0.15
295 f 2"
85 i 2 149 i 2 200 f 2
-0.75 f 0 . 1 -0.84 f 0.08 -0.95 f 0.08
3.72 f 0.15
80 i 2
-0.69 f 0 . 1
3.38 i 0.15
343 f 2'
0 0 -1
-1.2-
-1.4
-
293 i 2'
200
100
TIME
(psrc)
Figure 2. Plot of log OD (768 cm-l) us. time ( T , 308°K); time measured from flash trigger to time of scan through 768 cm-'.
The Journal of Physical Chemistry
Delay time, psec
Temp, OK
288 i 2'
164 + 2 240 f 2
105 i 2 170 i 2 248 f 2 372 i 2
Log k
+
-0.88 2 ~ 0 . 0 8
-0.95 4 ~ 0 . 0 8 -0.63 f 0 . 1 -0.67 f 0 . 1 - 0 . 7 2 f 0.08 - 0 . 8 4 2 ~0.08
3.25 i 0.15
KINETICSOF DECOXPOSITION OF CHLOROFORMIC ACID
1807
Discussion
*
The low value of AH rules out bond rupture mechanisms. A bimolecular process is ruled out, of course, by the linear plot shown in Figure 2. An intramolecular mechanism is suggested. An obvious process is the two-step mechanism, (4a) and (4b). The barrier to rotation has been measured for formic acid by
c1 I
Cl
I
C\o/ H
oAo
-.--c
/
trans
H cis
complex
trans
Lerner and Dailey,' who found 17 kcal/mole. Our value for AH is suffciently close to this value to suggest that reaction 4a is the rate-determining step and that reaction 4b is rapid in comparison. Furthermore, the positive value of AS* can be associated with the bending mode of the OH group which, in the activated complex, is no longer restrained. The reasonableness of this deduction can be tested by examining the probable interatomic distances in the trans-ClCOOH as well as in the stable counterparts HCOOH and FCOOH and the as yet unknown counterpart, BrCOOH. These distances can be compared to those in the product molecule, HX, which must be formed. Table I1 shows these calculated distances using con-
*
I
,
I
I
I
3.0
3.2
3.4
3.6
I o3T-' Figure 3. Temperature dependence of ClCOOH decay constant.
AS* = 5 f 5 cal/mole deg (2u). These results correspond to a frequency factor, A = 5 X 10". The uncertainty in the activation energy is rather large, as is appropriate to the scatter of the data in some of the determinations of k. However, a more sinister experimental error lies, perhaps, in the temperature measurement. The decomposition of C1COOH is undoubtedly exothermic, probably by 10 or 12 kcal/mole. It is implied that there might be a 2 or 3" temperature rise during the time of study and, indeed, there was evidence from the formic acid dimerization that this was probably so. Such an effect would, of course, cause a deviation from linearity in the plots such as Figure 2, and it would tend to give a k that is too large. On the other hand, there is a tendency for the implied error to be about the same at various temperatures since the optical density range is about the same. Hence, the temperature dependence (leading to AH*) is more reliable than the absolute values of k (which lead to A S or to A ) . I n any event, the implied curvature is not in evidence in Figure 2, so it must be smaller than the uncertainty introduced in the photometry.
*
Table 11: Calculated X.* . H Distances in trans-XCOOH rXealod
rHX,
X
A
A
Discrepancy, A
H F
1.86 2.00 2.24 2.47
0.74 0.92 1.28 1.41
1.12 1.08 0.96 1.06
c1 Br
ventional G X bond lengths, the formic acid G O distance, and hypothetical bond angles ( L X - G O = 120°, L G O - H = 90"). The last column shows that the bond length discrepancy is a minimum for X = C1. Another quantity of importance is the trend in the energy needed to rupture a G X bond minus the energy released in formation of an H-X bond. Of (7) R. 5. Lerner and B. P. Dailey, J . Chem. Phys., 26, 880 (1967).
Volume 71,Number 6 M a y 1967
REEDJ. JENSEN AND GEORGE C. PIMENTEL
1808
course there are other energy contributions to the over-all process, but they may be relatively constant as X varies. Table I11 shows these bond energies and, again, the difference. For consistency, the C-X bond energy in CF3X molecules is listed in the second column.*
Table 111: Bond Energy Differences between C-X and H-X Bonds (kcal/mole) X
H F
c1
Br
Do(C-X) in FICX
DdH-X)
Do(C-X) Do(H-X)
103 121 83 65
103 134 102 87
- 13 - 19 -22
0
Both Tables I1 and I11 provide a basis for understanding the ease of decomposition of ClCOOH relative to FCOOH and HCOOH. For chlorine, the X * H distance discrepancy is smallest, facilitating the hydrogen atom transfer. At the same time, the bond energy discrepancy contributes most to exothermicity. Using these quantities as guides, we see that the bromine counterpart is sufficiently like ClCOOH that
The Journal of Physical Chemistry
its decomposition should be similar. In particular, if the cis-trans isomerization is the rate-determining step for ClCOOH, it will probably also be so for BrCOOH. It is implied that BrCOOH should also have only transient existence under normal conditions.
Conclusions This study confirms the earlier identification of chloroformic acid2esas a transient intermediate in the photolytic reaction between chlorine and formic acid. Hence, it verifies further the reaction mechanism of West and Rollefson.‘ The energy of activation provides a basis for concluding that the unimolecular decomposition is probably ratelimited by the cis-trans isomerization around the G O bond. In addition to the specific result, we note that this study provides an encouraging prototype in the use of rapid-scan infrared spectroscopy. The versatility of the infrared technique now is available for the direct detection and kinetic study of reaction intermediates.
Acknowledgments. We gratefully acknowledge postdoctoral fellowship support (for R. J. J.) provided by the National Institute of Health and research support by the Air Force Officeof Scientific Research. (8) T . L. Cottrell, “The Strengths of Chemical Bonds,” 2nd ed Butterworth and Co. Ltd., London, 1968.