3006
RICHARD J. MILLERAND JEROME DAEN
Shock Tube Studies on the Condensation of Various Vapors
by Richard J. Miller and Jerome Daen Department of Chemistry, Lehkh University, Bethlehem, Pennsylvania
(Received March 16, 1966)
Shock wave induced condensation of 2-butanol, ethanol, methanol, chloroform, carbon tetrachloride, hexane, pyridine, and water was followed on a microsecond time scale using light transmission measurements. In all instances, deposition began as a film; “rupture” of the film into droplets or a lenticular structure was inferred from a decrease in transmitted light at later times. Values of the condensation coefficient were computed for a steady-state model of the process and a discussion of several aspects of the rupture was made.
a condensation experiment, the tube pressure was reduced to 1 p and a leak rate of 0.3 p/min. was mainKinetic processes in phase changes have been of contained. Shock velocity measurements were made with siderable interest. Though much attention has been two photomultiplier light screens whose signals were paid to the vapor-liquid condensation transformarecorded photographically from a Tektronix Type tion,1-3 it appears that only one investigation has 585 oscilloscope. been made of early events, this being the shock tube study of water vapor condensation by G~ldstein.~ The optical system used to follow the condensation on the tube walls was composed of an Osram Model The present study was initiated to investigate the sigHBO 100 W/2 high-pressure mercury arc lamp, the nificance of various parameters in the condensation light from which after collimation was passed through and film growth in a variety of materials. a Bausch and Lomb interference filter and then through Condensation is induced in the shock tube by utithe shock tube. The incident slit width was 0.125 in. lizing the pressure increase across the shock wave. The The transmitted light impinged on a Dumont No. bulk of the gas behind the shock wave is compressed 6292 photomultiplier tube and was read out on a adiabatically and heated very rapidly. However, Tektronix Type 515 oscilloscope. The mercury arc the gas in the thermal boundary layer along the wall, lamp was powered by a d.c. supply having a ripple of essentially, will suffer an isothermal compression. less than 1%. Observations were made 8.83 ft. below Thus, if a gas is shocked and its resultant pressure is the diaphragm. greater than the equilibrium vapor pressure at the wall Modifications were made to measure the condensatemperature, the material in the thermal boundary tion rate simultaneously pt two different wave lengths, layer will become supersaturated and condensation on ie., at 5460 and 4000 A. Upon emerging from the the wall will ensue. shock tube, white light from the lamp fell onto a 4000In this study, weak shock waves were passed through 8. filter which was inclined at 45”. This acted as a various vapors at or near their saturation vapor beam splitter by transmitting the 4000-A. radiation pressures. The resulting condensation was monitored into the photomultiplier and reflecting all other-radiaby recording the intensity of a transmitted beam of tion upward onto a 5460-A. filter and then into a second monochromatic light passing through the shocked gas photomultiplier tube. The absorption spectrum of the since a thin coating of condensed material influences inclined filter showed negligible changes in its transthe light reflective properties of the system.
Introduction
Experimental The shock tube was constructed of two pieces of 1.5in. i.d. Pyrex pipe, the driver section being 3 ft. long and the channel 10 ft. long. Seals between the various sections of the tube were of Teflon gaskets. Prior to The Journal of Physical Chemisfry
(1) W. M. Nagle and T. B. Drew, Trans. Am. Inst. Chem. Engrs., 31, 605 (1934). (2) N. Fatioa and D. L. Katz, Chem. Eng. Progr., 45, 661 (1949). (3) J. F.Welch, Dissertation Abstr., 21, 3393 (1961). (4) R. Goldstein, J. Chem. Phys., 40, 2793 (1964).
SHOCK TUBESTUDIESON THE CONDENSATION OF V m o u s VAPORS
mission characteristics. Outputs from the two photomultiplier tubes were read out photographically on a Type 585 oscilloscope using a Type 82 dual trace plugin unit for simultaneous display of both signals. Materials studied were %butanol from Eastman Organic Chemicals (Eastman grade), methanol, CClc, CHCl,, and pyridine from the J. T. Baker Co., hexane from Mathesou Coleman and Bell (reagent grade), and absolute ethanol from United States Industrial Chemicals. All were used without further purification. Singly distilled water was studied in all water experiments.
3007
Results %Butanol. Experiments were camed out with 2butanol as the condensate, using helium gas as a driver and a diaphragm of 1.0-mil foil. A typical oscilloscope trace is shown in Figure 1. When this alcohol, at ita saturation vapor pressure, was shocked it was observed that the intensity of light transmitted through the tube increased by 2-5% within 5 pet. after the pasaage of the shock. This initial increase was followed by oscillations in the intensity which lasted for approximately 200 psec. At this point a large decrease in intensity occurred which marked the end of the useful portion of the record. At no time during the oscillations did the intensity return to its original level. That the behavior apparent in the trace in Figure 1 is indicative of the formation of a thin film of condensate on the tube wall may be seen from the following argument. Suppose a film of refractive index n1 exists between two media of refractive index no and ns, respectively; here no refers to the refractive index of the vapor and ns to that of the wall material. If unit light intensity is incident upon this film, then the transmitted light intensity can be shown to be6
I =
Figure 1. Typical timetransmission records for various materials: A, >butanol: Tz, 440"K., P, = 260 mm., P,/P. = 13; B,ethanol: T*,400"K.,P, = 250 mm., P2/P, = 5.37; C,ethanol; Tg, 350°K.. Pz = 87 mm., P,fP. = 1.50; D,wnter: TI,MO'K., PI = 165mm.. PdP. = 7.40; E, chloroform: T,,3OO0K., Pz = 530 mm., P,/P. = 2.40.
The tube wall was cleaned between series of runs but not between individual rum in a series on a compound. The test area was swabbed with methanolie KOH and rinsed three times with pure methanol. Vapors were evaporated into the evacuated shock tube from a glass bulb; initial preasum were measured on a Dubrovin gauge. Helium, oil-pumped nitrogen, and Freon-12 were the driver wes. Aluminum foils, 1.0- and 0.8-mil in thickness, were used as diaphragms; these broke at pressure differences of about 1.5 and 1 atm., respectively.
where z = 4n10/X. Here e is the thickness of the film and X is the wave length of the incident light. If n1 < n2, the film formed is an antiieflection film; the amount of light reflected at the surface is reduced and hence the transmitted intensity increases. As the film thickness increases, eq. 1 describes an oscillatory behavior whose minima are equal to the original intensity. Such behavior is as observed in Figure 1, with the difference that the observed intensity never returns to its original level. It is inferred that filling of surface irregularities with liquid and the subsequent removal of scattering centers from the glasa surface is responsible for this increase. It is seen from eq. 1 that maxima and minima in intensity occur at integral multiples of z = a. Presented in Table I are time-thickness data calculated for several runs for %butanol and other materials. In these calculations, the refractive index of the Dline of sodium was used for that at 5460 1. The un(6) A. Vmiwk. "Optios of Thin Films." Interscience publishers, Ine..
New Yark. N. Y..1980. P. 112.
Voluma 69,NunzAer 9 Ssplanbcr 1966
3008
RICHARD J. MILLERAND JEROME DAEN
Table I: Typical Time-Thickness Data c
Time (paec.) for thickness of (cm.
x
,
106)
Material
1
2
3
4
5
6
7
%Butanol %Butanol %Butanol %Butanol Ethanol Methanol Chloroform Carbon tetrachloride Hexane
5 5 5 5 15 45 15 10 10
20 15 12 15 45 250
55 30 30 50 120
90 55 50 80 200
135 75 75 110
200 110 110
300
30
60 50 80
100 80 140
140 110
225 150
320
25 30
PI,“
Pz?
mm.
mm.
19.0 18.8 19.7 15.0 19.6 19.8 219 138 21.8
145
140 260 284 252 247 158 527 304 270
Subscript 1refers to conditions ahead of the shock. Subscript 2 refers to conditions behind the shock. ation ratio relative to the vapor pressure of the liquid at wall temperature.
certainity in 0 introduced by this approximation was estimated to be less than 0.5% based on dispersion data for several liquids. Gas properties behind the shock were calculated using the method of Bethe and Teller6 with the required enthalpies being obtained from various s o ~ r c e s . ~To substantiate further that the condensation of a film on the tube wall was being observed, simultaneous measurements were made of the intensity behavior using radiation of two different wave lengths. The interference peaks in the intensity should occur earlier for the shorter wave length, but the calculated thickness at any time should agree for the two wave lengths. Curves for both wave lengths are seen to coincide on the thickness-time graph for 2-butanol given in Figure 2. All photographs showed the characteristic large decrease in intensity which occurred at 160-200 psec. after the shock arrival and which marked the end of the oscillatory portion of the trace. Calculations demonstrated that this decrease could not be assigned to the arrival of the cold front into the viewing area. Consequently, this decrease is ascribed to the formation of a lenticular structure such as would accompany the rupture of the film to lenses or droplets which scatter light and thus lead to a diminution in transmitted light. Ethanol and Methanol. In general, the results for ethanol qualitatively resembled those of 2-butanol, the difference being that the condensation rate was lower for ethanol. Typical traces are shown in Figure 1. Supersaturations (Pz/P,) ranged from 5 to approximately 1.2. Down to a supersaturation of 2, there was a gradual decrease in observed condensation rate. From supersaturations of 1.2 and 1.5 a rather different type of trace was obtained. In these instances, there was an increase in intensity as usual which was quickly The Journal of Physical chemistry
Tz, OK.
Pz/PaC
380 460 440 450 490 500 300 337 370
7.0 13.0 10.4 15.9 5.37 1.44 2.40 2.20 1.80
Pl/P. is the supersatur-
0
0 0
0
50
100 Time, mea.
150
200
Figure 2. Simultaneous thick-time mwurements for %butanol: 0, 5460 1.; 0, 4000 A.
followed by a large decrease, below the original intensity, which is ascribed to the rupture of the film, The condensation of methanol appears to be less extensive than that of either 2-butanol or ethanol, generally affording only one maximum and one minimum in intensity before film growth ceased. Hexane. In view of its nonpolar nature and lack of association in the vapor phase, the behavior of hexane was compared with that of the alcohols. At supersaturations of about 2, behavior similar to that of ~
~~~~
(6) H. A. Bethe and E. Teller, Ballistics Research Laboratories, Aberdeen Proving Ground, Md., Report X-117(1945). (7) N.S.Berman and J. J. McKetta, J.Phys. Chem., 66, 1444 (1962) (%butanol); K. 8. Pitzer, J. C. M. Li, and E. V. Ivash, J . Chem. Phys., 23,1814 (1955 (methanol); J. H.S. Green, J . A p p l . C h m . , 11, 397 (1961) (ethanol); F. D. Rossini, et al., “Selected Properties of Physical and Thermodynamic Properties of Hydrocarbons and Related Compounds,” Carnegie Presa, Pittsburgh, Pa., 1953 (hexane) ; K. Li,J . Phys. Chem., 61,782 (1957)(pyridine) ; E.Gelles and K. 8. Pitzer, J. Am. Chem. SOC.,75, 5259 (1959) (carbon tetrachloride and chloroform); A. S. Friedman and L. Haar, J. Chem. Phys., 22, 2051 (1954) (water).
SHOCK TUBESTUDIESON THE CONDENSATION OF VARIOUS VAPORS
ethanol was observed. At a supersaturation of about 1.3, the behavior resembled that of methanol. Water. Initial experiments on water, with supersaturations ranging from 7 to 9, resulted in traces which were quite different from those of hexane and the alcohols and also different from those obtained by Goldstein. An example is shown in Figure 1. At these higher supersaturations a flm is observed to grow for a relatively long time, 100 to 200 Bsec., although the growth rate was quite small. What occurs after this period is not entirely clear, but it involves the rupture of the film as indicated by the large decrease in intensity at about 200 psec. The effect of lowering supersaturation and collision rate with the wall was seen to be an earlier rupture as with certain ethanol films. The difference in appearance between traces for water and the other materials previously studied is believed to be due to the relatively large surface tension of water which results in a less stable film. At supersaturation ratios of about 5, the water traces resembled Goldstein’s more closely. In experiments discussed in this work, the flms were stable for a larger period of time; reduced turbulence in the present shock tube is probably responsible for the difference. Carbrm Tetrachloride and Chloroform. Since CCL is symmetrical and nonpolar, and CHC1, has low polarity, a comparison of their behavior with that of the rest of the materials was of interest. Typical results for CHC& are shown in Figure 1. The traces qualitatively resemble those of the alcohols and hexane, with one difference. For the previously studied materials, the envelope of the oscillatory portion was a horizontal straight line. Here it is a straight line with a slope toward greater intensities. Pyridine. To investigate the effect of surface tension, experiments were performed on pyridine, which has a surface tension of 38 dynes/cm., intermediate to that of water and the other materials. Pyridine exhibited a decrease in intensity as condensation was initiated, as predicted by eq. 1 for a material of refractive index slightly higher than Pyrex. This observation, of course, is additional evidence for the production of a film on the wall, as a “reflection coating” is formed. Since the refractive indices of liquid and glass were so close the oscillations in intensity were of small amplitude. Curves of thickness us. time for typical runs with various materials are displayed in Figure 3.
Discussion A point which should be discussed is the thermal stability of the various compounds at the temperatures that exist behind the shock. In experiments
3009
a, 2-BUTANOL
L 1
.o 7
-
CARBON TETRACHLORIDE
6
5
3
1 0
0
50
100
150 200 Time, @eo.
250
300
350
Figure 3. Deposition rates for various materials.
with CCL, CHCl3: and hexane, temperature rises of only 50 to 100’ were obtained; decomposition would be negligible here. Pyrolysis in 2-butanol, methanol, and pyridine has been observed at temperatures above 550°,8-10 considerably higher than those obtained in these experiments. It is thus clear that pyrolysis of the gases behind the shock wave in these experiments cannot be responsible for the observations on light intensity. The compounds do not absorb light in the visible region, their absorption peaks being far into the ultraviolet. Any decomposition, however small, which might occur here could not give products with drastically different spectral characteristics froF the original material, so that absorption of the 5460-A. light would be negligible for any species present. In other words, there could be no enhanced transmission resulting from chemical breakdown. Close inspection of the transmission records, as in Figure 4A, reveals that for 2-butanol a very sharp decrease (of about 1 psec.) occurs immediately after the arrival of the shock front. To investigate this more closely, photographs were obtained at higher sweep speeds. Interestingly enough, under these conditions it is possible to resolve the initial spike into two parts: a decrease followed by a less obvious abrupt increase. The decrease can be interpreted in terms of the known (8) F. Someno, Bull. Phys. Res. (Tokyo), 21, 277 (1942).
(9) E. Kuss, Angew. Chem., 49, 483 (1936). (IO) C. D. Hurd and J. L. Simon, J. Am. Chem. SOC.,84, 4519 (1962).
Volunte 65,Number 9 September 1566
3010
~ Ficurc 4. 8imull;ineour intensity-lrml g:mgc r c ~ . o r dlor 2-butanol: PI = 11 mm.; 1; = Z!),S"K.; .VI= 3.13; P$IP, = 10.0; distance between light source and heat gauge is 1.98 em.; distance from diaphragm to heat gauge is 14.3 ft. The initial rising spike in A has been inked over to improve reproduction quality.
small though measurable reflectivity of the shock and a slight tilt with respect to the wall; the shock passage time over the slits is quite comparable to the width detected'. However, the second (increasing intensity) spike is associated with the condition of the wall surface. If the wall was cleaned with methanolie potassium hydroxide and rinsed with methanol, the second spike disappeared and the first intensity maximum broadened very slightly. After six consecutive runs without recleaning in the manner described, the second spike reappeared. What is the responsible factor is not clear. However, it might be noted that this behavior is observed with less polar materials such as %butanol, chloroform, and carbon tetrachloride rather than with polar molecules such as water, ethanol, and methanol. Some plots of thickness us. time have been displayed in Figure 3. In each case there seem to be two general regions to the curve, a region of decreasing growth rate from the onset of condensation continuing for 50 to 100 psec. followed by one of relatively constant growth rate. One factor in the change of growth rate with time in the very early stages of the deposition and especially at times considerably less than those corresponding to the first datum point would lie in the changing nature of the surface as the condensation proceeds. As the film thickn e s increases, the condensationprocess changes from one of vapor condensing on a glass surface to one of vapor condensing on monolayer through multilayer and terminating on bulk liquid. It would be unexpected to find that the probability of a molecule condensing on the liquid is the same as for all of the earlier stages involved. As the growth continues, the rate approaches a constant value as condensation takes place on bulk liquid. Such effects would be expected to be most significant at very short times, although work by Shereshefsky" on evaporation of liquids from capillaries suggests that
RICHARD J. MILLERAND JEROMEDAEN
the wall might have an effect on evaporation-condensation processes at distances up to lo00 A. from the wall. Beyond this "substrate influenced region" phenomenon, one would expect that mme time would be required to reach a steady state in flow. During this time, transient effects in heat transfer, mass flow, etc. would be smoothed. Following this transient period, in the steady state, condensation would proceed on an underlaying layer of molecules of liquid, with heat transfer taking place through the bulk. A t present, the importance of these transient effects is not completely understood. Heat flux measurements which are in progressLzwill be useful in clarifying the situation. To summarize, oscilloscope traces for the longer lasting films showed two regions in the condensation process, first the region of rapidly decreasing growth rate followed by a region of relatively constant growth rate. Calculations based on the latter were made since any errors in choosing the slope will be small. A condensation coefficient is defined as the fraction of gas molecules striking a cold surface which actually stick and, according to Mortensen and Eyring,13 is given by the ratio of partition functions for the molecules on the surface to those in the vapor phase. Because the vibrational degrees of freedom are the same for these two states, these do not have to be considered. However, the molecules in the surface experience restricted rotation and have a smaller rotational partition function. Therefore, the condensation coefficient is given by the ratio of these rotational partition functions. Experimentally determined values of the condensation coefficient generally agreewith the calculated ones.18 In the past, rather than the condensation coefficient itself, the quantity measured has been the evaporation coefficient, the analog of the condensation coefficient for the evaporation process. The two coefficients are equal at a state of equilibrium between liquid and vapor. It will h t prove helpful to have at hand an estimate of the temperature rise in the film that would be expected under steady-state conditions. An approximate calculation of the temperature rise can be obtained from a heat balance. The heat p given off in the condensation of vapor on unit area is (11) M. Folman and 3. L. Shereshefsky, J . Pkya. Chem.. 59, Bo7
(1955). (12) W. R. Smith and P. Kicska, private communication. (13) E. M. Mortamn and H. Eyring, J . Pkya. Chem.. 64, 846 (19Bo).
SHOCK TUBESTUDIESON THE CONDENSATION OF VARIOUS VAPORS
q = hdpdt
where h is the heat of vaporization per gram of liquid formed, B is the growth rate of the film, and p i s the density of the liquid. This, of course, is equal to the sum of the heat which flows into the wall (here assuming constant wall temperature) and that which causes the temperature rise in the film; i.e. hdpdt = K(dT/dB)dt
+ 6CppdTdt
where K is the coefficient of thermal conductivity, dT/d6 is the temperature gradient across the film, 6 is the film thickness, and C, is the heat capacity per gram. Values of B were taken directly from curves of thickness us. time, and heat transfer between the film and gas and boundary layer effects were neglected. Temperature rises calculated by this method for 2butanol, ethanol, methanol, hexane, CCL, and CHCla are of the order of 6, 2, 0.5,1.5,3, and 3O, respectively. In the calculation of the condensation coefficient, these temperature rises were included. If the following computation of the condensation coefficient is to be applicable, the rate of flow from the vapor to the film must be small in comparison to the sound speed in the vapor. Under typical conditions with 2-butanol, the deposition rate in the later stages mole/sec./cm. of of the process is about 3.3 X wall. If all of the molecules deposited come from near the wall (maximum depletion) and the density of the gas behind the shock is 0.01 mole/l. (at conditions of the typical run quoted above), then the deposition rate is 3.3 1. of gas/sec. Consequently, there must be a velocity directed toward the wall of magnitude equal to 280 cm./sec., which is considerably less than the sound velocity. In a typical run with 2-butanol, the film reached a thickness of about 7 X cm. before the useful portion of the trace ended, this corresponding to a maximum removal of only about 7% of the gas in the tube. One may then, for simplicity, take the conditions at the center of the tube to be essentially the same as if there were no wall-directed transport; ie., a steady state is assumed. The net deposition onto the wall is given by A = '/dP(pwC,
- PLCW)
(2)
where the first term in the parentheses accounts for the number of wall collisions from the vapor and the second gives the number evaporating from the liquid film a t the temperature of the wall. The subscript w refers to the wall and p~ is the density of the saturated vapor at the temperature of the wall. As is usual, the evaporation and condensation coefficients have been taken to be equal and are represented by P.
3011
Using the experimentally determined values of A, the estimated wall temperature rise, and the condition of the shocked gas, the values of ,8 summarized in Table I1 were obtained. Table I1 : Measured Condensation Coefficients Material
B
%Butanol Ethanol Hexane Carbon tetrachloride Chloroform Water" Wateru Water16 Water" Carbon tetrachloride18 Chloroform18 Methanol18 Hexane'* Water"
0.032 f 0.005 0.026 f 0.005 0.04 f 0.01 0.05 i 0.01 0.017 Z!Z 0.004 0.036 0.35 to 1 . 0 0.243
Wall temp., OK.
312 297 298 307 305 283
0.042
1 0.16 0.045 0.70 (calcd.) 0.036
273 275 273 273
Comparison of the condensation coefficients in Table I1 with those given by Mortensen and Eyring shows reasonable agreement for ethanol. For 2-butanol, @ is of the same magnitude as their value for propanol, as might be expected. However, a value of of 0.05 was obtained here for CCL, much less than the previously reported value of unity.13 The other values of p are similarly lower than the previously obtained values. Discrepancies similar to these exist between values of ,8 of other materials obtained by various methods. Alty and McKay14 measured p for HzO by an evaporation method, obtaining a value of 0.036. In a recent study, Nabavian and Bromleyl5 obtained a value of @ ranging from 0.35 to 1.0 from a measurement of the heat transfer coefficient of a HzOfilm on a copper tube. In still another study, Hickman,lsobserving the evaporation from a jet flowing under vacuum, obtained a value of 0.243 for P, which may reach unity if surface cooling is considered. This paper was discussed by Delaney, Housten, and Eagleton,I7 who themselves found /3 to be only 0.027. They suggested that the rate of vaporization from a rapidly renewed surface, (14) T. Alty and C. A. McKay, Proc. Roy. Soc. (London), A149, 104 (1935). (15) K. Nabavian and L. A. Bromley, Chem. Eng. Sci., 18, 651 (1963). (16) K. C.D. Hick", I d . Eng. Chem., 46, 1442 (1954). (17) L. R. Delaney, R. W. Housten, and L. C. Eagleton, Chem. Eng. Sci., 19, 105 (1964).
'volums 69, Number 9 September 1966
3012
RICHARD J. MILLERAND JEROME DAEN
such as Hickman’s, is much higher than it is from a stagnant surface. Littlewood and Rideall8 have discussed the dependence of the condensation coefficient on the experimental method used. These authors feel that the heat and mass transfer fluxes of evaporation experiments could have some effect on interfacial equilibrium, In other words, lack of knowledge of the surface temperature is a large contributor to the uncertainty and the value of the experimentally measured condensation coeEcient is dependent on the supply of heat to the surface (when the evaporation coefficientis being measured). When heat transfer to the surface is sufKcient, 13 will always be unity. They attributed much of Hickman’s success to the fact that surface cooling is almost eliminated in the jet. Another aspect of the deposition process merits some attention here. All films observed in these experiments, as mentioned earlier, exhibited an abrupt turbidity increase that preceded the arrival of the contact surface as calculated from idealized shock tube theory. Though it has been shown that the actual arrival of the cold front can under certain conditions occur earlier than predicted,lg the diminished transmission intensity herein noted is assigned to a rupture of the film. In Figure 4 are displayed simultaneous light transmission and platinum thin film heat gauge measurements on a 2-butanol film; the time scale in Figure 4A is 50 psec./cm. while in Figure 4B it is 500 psec./cm. In Figure 4A, it is evident that the heat gauges and the intensity data correlate. That is, both indicate the shock arrival. Moreover, the optically observed growth is accompanied by a concurrent increase in temperature. When the turbidity abruptly increases, about 150 psec. after the shock arrival, the temperature begins to drop. Displayed in Figure 4B is the same heat gauge record simultaneously recorded at a slower sweep speed. The first maximum corresponding to the ascribed rupture is apparent. This is followed by an oscillatory profile which terminates in a sharp drop about 600 psec. after the entry of the shock into the observation station, this being ascribed to the contact surface. Idealized shock tube calculations for this experiment show that the latter should arrive 1.09 msec. after the shock front. Rupture of the film into a lenticular structure would result in enhanced light scattering as observed. Except for a few cases of early breakup, the final breakup of the film appears to be the result of a flow-enhanced instability based on observations of the time-torupture. For the large range of conditions achieved in experiments on 2-butanol, it is evident that the film stability decreases with increasing particle velocity
Thickneaa et rupture, om. X 105. 8 7 6 5 4 3
2
1
I I t l I
1
1
The Joumd of Physical Chemistry
I
1
I 0
I
I
I
3 4 Relative particle velocity ( Uz/al). 1
2
I 5
Figure 5. Time-to-rupture behavior of 2-butanol vs. particle velocity.
behind the shock front. This behavior is displayed in Figure 5 where the time between the onset of condensation and the rupture of the film is plotted against the particle velocity Mach number. Concomitant with the decrease of the time-to-breakup is a decrease in thickness at breakup as shown in Figure 5. Since breakup occurs a t no particular thickness, the downward trend in stability with increasing particle velocity leads to the conclusion that the stability is dynamic in nature in the experiments summariaed. Experiments are included here in which the supersaturation ratio was varied while the flow speed was held constant. All of the observations appear to fall on a common curve. To summarize very briefly, all materials first deposited as a continuous film which later broke up. Time-to-rupture data indicate that increasing particle velocity above the deposited film can be correlated with the rate of rupture. That is, films above which a highvelocity gas is flowing break up sooner than those in a low-velocity environment. Condensation coefficients computed from these shock tube studies, though probably low, are in fair agreement with those determined by other techniques. The major source of error in this method appears to be the determination of the film temperature.
Acknowledgment. We are indebted to Professor W. R. Smith and Mr. Paul L(;icska for the records of Figure 4 as well as for many discussions. Thanks (18) R. Littlewood and E. Rideal, Trans.Faraday SOC.,52, 1598 (1956). (19) R. E.Duff, Phys. FZuids, 2, 207 (1959).
3013
PYROLYSIS OF CARBONATES IN THE GASPHASE
also are due to the Union Carbide Corporation for a Fellowship for the 1961-1963 period for R. J. M.,
and to the National Science Foundation for partial support of this research.
A Study of the Pyrolysis of Methyl Ethyl and Diethyl Carbonates in the Gas Phase
by Alvin S. Gordon and William P. Norris Chemistry Division, Research Department, U.S. Naval Ordnance Test Station, China Lake, Calijurnia 03667 (Received May 16, 1066)
A number of carbonates-dimethyl, methyl ethyl, and diethyl-have
been pyrolyzed in quartz reaction vessels with different surface-volume ratios. Unlike some recent reports, we find dimethyl carbonate to be extremely stable to pyrolysis and photolysis to at least 350". On the other hand, both methyl ethyl carbonate (MEC) and diethyl carbonate (DEC) pyrolyze with homogeneous, unimolecular kinetics in a temperature range where kDEC = DMC is stable. The specilic rate constants are: ~ M E C = 1018+710-'B~m/2.808RT; 101a.910-~~0@J/~.~~R~ (sec.-l). These kinetic factors suggest that the slow step in the reaction sequence is the same as that in esters with a P-H atom in the alkyl group, i.e., the transfer of the 8-H atom to the carbonyl oxygen with the concurrent breaking of the C-0 bond to form.the cr-olefin and a carboxylic acid. The mechanism is analogous to the mechanism for xanthate pyrolysis. Our results show no evidence that alkyl radicals attack either the carbonyl oxygen or the ether oxygen of a carbonate in a radical displacement reaction.
Introduction Wijnenl and Thynne and Gray2 have studied the pyrolysis of dimethyl carbonate and report that it is a heterogeneous process on a quartz surface. Wijnena has also reported that dimethyl carbonate photolyzes with the radiation from a medium pressure mercury arc. We have not been able to reproduce these results, indicating that the surface of our reaction vessel and our light source differed from those of the previous investigations. In a quartz vessel of 45-cc. capacity, with a surface to volume ratio (s/er) of 1.55 cm.-I, we find dimethyl carbonate to be stable at 350" for 1200 sec. at 10-mm. pressure. A significant percentage of both diethyl carbonate and methyl ethyl carbonate decomposes in 600 sec. at this temperature. Dimethyl carbonate has been photolyzed with full intensity of a Hanovia medium pressure mercury arc at various temperatures
up to 350" for 1200 sec. at 10-mm. pressure with only traces of CO, CHI, and C02 as photodecomposition products. Dimethyl carbonate vapor, at 25", shows no absorption from 2000 A. to the visible within the experimental accuracy of our a p p a r a t ~ s . ~ The mechanism for ester decomposition has been quite firmly established. For carbonates, the gas phase6 decompositions of phenyl and benzyl ethyl carbonates have been shown to be first order and homogeneous in a "conditioned" stainless steel reactor. (1) M. H. J. Wijnen, J. Chem. Phys., 34, 1465 (1961). (2) S. C. S. Thynne and P. Gray, Trans. Faraday Soc., 58, 2403 (1962). (3) M.H.J. Wijnen, J. Chena. Phys., 35,2105 (1961). (4) We wish to thank Dr. R. H. Knipe for obtaining the absorption spectrum of dimethyl carbonate. (6) (a) G. G. Smith, D. A. R. Jones, and R. Taylor, J.Org. chm., 28, 3547 (1963); (b) P.D.Ritchie, J. Chena. SOC.,1054 (1935).
Volume 60,Number 0 September 1086
