Apparatus which permits repeated gas-liquid chromatographic

U. S. Army Natick Laboratories, Natick, Mass. 01760. An apparatus has been devised, utilizing the principle of cyclic enrichment, by means of which ma...
0 downloads 0 Views 734KB Size
An Apparatus Which Permits Repeated Gas-Liquid Chromatographic Separations of a Sample R. A. Prosser U.S. A r m y Natick Laboratories, Natick, Mass. 01760 An apparatus has been devised, utilizing the principle of cyclic enrichment, by means of which many gases and vapors diluted with an inert gas can be subjected to various reactions, concentrated, and transferred to the gas-liquid chromatograph with negligible loss or contamination. With this device a reasonable approximation to plug flow into the chromatograph is obtained. The entire sample or any portion represented by a set of chromatographic peaks can be recollected at the chromatograph exit, treated chemically, reconcentrated, and then reanalyzed by GLC numerous times using different liquid phase columns, carrier gas flow rates, detectors, column oven temperatures, etc. Utilizing this apparatus, i t is shown that powdered metals and other additives do affect the pyrolysis of powdered polyvinylidene fluoride at 450' C.

THECOMBINATION of a hot-filament or reaction vessel pyrolysis unit in line with a gas-liquid chromatograph (GLC) has been used very much in recent years to study the degradation of many polymers (1-14). This system offers three important advantages: the pyrolysis products are quickly removed from the pyrolysis chamber by the carrier gas of the G L C , thereby reducing the risk of further decomposition; there are no transfer losses between the pyrolysis unit and the G L C ; and plug flow results-i.e., the pyrolysis products enter the G L C column in the concentrated state required for good peak resolution. [The disadvantages and additional advantages of pyrolysis-gas chromatography are fully discussed elsewhere

(19.1 However, these advantages are obtained at a cost; the pyrolysis parameters are severely restricted by the operating conditions of the G L C . It would be desirable to separate the pyrolysis unit from the GLC and still retain the abovementioned benefits. Actually, in most cases where peak identification by infrared and mass spectra is uncertain, the pyrolysis must be carried out separately. This is especially true when relatively large quantities of material have to be decomposed t o provide enough of the degradation products

(1) D. A. Vassallo, ANAL.CHEM., 33,1823 (1961). (2) E. A. Radell, and H. C. Strutz, Ibid., 31, 1890(1959). (3) J. Strassburger, G. M. Brauer, M. Tryon, and A. F. Forziati, Ibid., 32, 454 (1960). (4) F. A. Lehmann and G. M. Brauer, Ibid., 33, 673 (1961). ( 5 ) D. Ettre and P. F. Varadi, Ibid., 34,752 (1962). (6) Ibid., 35,69 (1963). (7) R. S . Porter, A. S . Hoffman, and J. F. Johnson, Ibid., 34, 1179 (1962). (8) E. M. Barrall 11, R. S . Porter, and J. F. Johnson, Ibid., 35, 73 (1963). (9) C. E. Legate and H. D. Burnham, Ibid., 32, 1043 (1960). (10) S. B. Martin and R. W. Ramstead, Ibid., 33, 983 (1961). (11) W. H. Parriss and P. D. Holland, British Plastics, 33, 372 (1 960). (12) C . E. R. Jones and A. F. Moyles, Nature, 189, 222 (1961). (13) Ibid., 191, 663 (1961). (14) H. Szymanski, S. Salinas, and P. Kwitowski, Ibid., 188, 403 (1 960). (15) S . G. Perry, J . Gas Cliromatog., 2, 54 (1964).

for identification by NMR, elemental analysis, or boiling point, because they are difficult to pyrolyze quickly enough for plug flow to result. Also, the G L C column can be obstructed by pyrolyzate which is not volatile at the temperature of the column oven. I n addition, if materials such as hydrogen fluoride, hydrogen chloride, or water are produced in relatively large quantities, not only may many peaks of the chromatogram be masked but the equipment may also be adversely affected. When the pyrolysis is carried out separately, the first advantage is retained: the pyrolysis unit can still be swept by helium. Separately operated, the pyrolysis can be carried out over a wider range of pressures, including vacuum conditions, over a wider range of inert gas flow rates, over a period of time (if the decomposition is slow at the given temperature), and hence over a broader range of temperatures, using much larger sample sizes, and in the presence of additives-e.g., hydrogen, oxygen, and sulfur dioxide. Furthermore, materials such as hydrogen fluoride and water can be removed from the sample prior to chromatographing. It is difficult to retain the second and third advantages when dealing simultaneously with gases and volatile liquids. The usual technique has been to use a hypodermic syringe, but this entails loss of material and generally makes plug flow impossible. As an example, suppose a polymer is pyrolyzed in a n evacuated 25-cc container with a rubber septum and the resulting pressure stabilized at 127 mm Hg. Helium is allowed to enter until a pressure of 760 m m is reached. This permits a 5-cc sample of gas to be withdrawn by means of a syringe and transferred to the GLC. Irreproducible amounts of material (some of which might be toxic) are generally lost during transfer. Sometimes, inexplicable peaks appear on the chromatogram and much time is wasted in ascertaining whether these peaks are attributable to the sample or t o air, moisture, or acetone vapor, entering the syringe during transfer. Because the sample is already 8 3 % helium, plug flow will be impossible and resolution will suffer accordingly. There are two other disadvantages inherent in the syringe technique; one involves time, and the other the sampling temperature. Approximately 17% of the total gases and vapors are contained in the syringe. If the major peaks require a n hour to issue from the G L C and if quantities of each peak are to be collected, multiple chromatograms must be made (using a 0.25-inch 0.d. column). If the entire gaseous pyrolyzate could be introduced into the G L C a t one time in a concentrated state, much time would be saved, the peak resolution would improve, and peaks might appear that would not be detected when the syringe was used. As for the temperature disadvantage, the best way to obtain a chromatogram showing the correct relative peak areas of the pyrolysis products is to introduce the entire undiluted solid, liquid, and gaseous pyrolyzate, or an aliquot thereof, into a G L C that can be programmed t o a temperature high enough so that all the major pyrolysis products are eluted. But sampling when the syringe and pyrolyzate are a t an elevated temperature is relatively inconvenient. However, if the VOL. 39, NO. 10, AUGUST 1967

1125

HELIUM

INERT GAS

MANOMETER

1 1 i r n

~

A

N

O

TO VACUUM PUMP M E T E R

t f

SAMPLE COLUMN

-

+OD 1

BLACK PEN

REFERENCE COLUMN

1

x

H

TERS

& OD

W G

Figure 2. Schematic of the concentratoi FM = Bellows pump motor Figure 1. Schematic of the gas-liquid chromatograph pyrolyzate is divided by removing the gases and other materials that are readily volatile at 120" C, the rest can be handled by routine methods at room temperature with negligible losses by evaporation. This would lead to an equivalent result by means of two overlapping chromatograms. In cases where a major pyrolysis product condenses a t room temperature, a chromatogram of a sample of the gaseous phase will not give a true picture of the relative amount of the material. Even when the sample is taken a t an elevated temperature, the same criticism holds, although a truer picture is obtained of the relative amounts of the lower molecular weight materials (on which the mechanism of decomposition is usually based). Further, when the pyrolysis is carried out in a closed vessel, secondary degradation can occur, the relative amounts of the degradation products may be severely affected, and the mechanism of decomposition can be obscured. Secondary degradation can be reduced by sweeping the pyrolysis products as they are formed into a trap with helium. It would be desirable t o keep this trap small enough so that, on attachment to the GLC, it would provide plug flow and thus eliminate the syringe. On the other hand, the trap should be long enough to collect all the pyrolysis products a t a reasonable sweep rate of the helium. In order to keep the surface-tovolume ratio of the trap sufficiently large to condense the pyrolyzate quickly, narrow-bore tubing should be used, but, unfortunately, such tubing is prone to clogging. Therefore, it would be better to construct the trap to serve its main function-i.e., the collection of the pyrolyzate without becoming blocked. It is apparent from the foregoing discussion that a separate device is needed by means of which the pyrolyzate, in the gaseous state at a n elevated temperature, can be transferred to the GLC with negligible loss or contamination and plug flow obtained. A description of such a device, which we shall call a concentrator, is given below, together with observa1 126

ANALYTICAL CHEMISTRY

7

tions made showing that powdered metals and other additives d o affect the decomposition of powdered polyvinylidene fluoride and/or the pyrolysis products at 450" C . It was not found necessary to maintain the metal tubing connecting the pyrolysis oven to the trap at 120" C for these experiments. Heating the tubing with a heat gun occasionally during the course of the pyrolysis and a t the end enabled the helium to sweep the volatile products into a refrigerated collection trap. It will be shown that, using the concentrator, the entire sample or any portion represented by a set of chromatograph peaks can be recollected a t the chromatograph exit, treated chemically, reconcentrated, and then reanalyzed by GLC numerous times using different liquid phase columns, carrier gas flow rates, column oven temperatures, detectors, etc. This feature is especially important in those cases where a small amount of the sample is so difficult to obtain, prepare, or reproduce identically that its preservation is paramount. Before describing the concentrator and its use, we shall briefly describe the GLC. EXPERIMENTAL

The Gas-Liquid Chromatograph. As is apparent from Figure 1, the GLC is a dual-column instrument. The columns and katharometer ovens and their controls (isothermal) are from Research Specialties Co., Richmond, Calif. The recorder is a dual pen (red and black) Speedomax G from Leeds & Northrup Corp., Philadelphia, Pa., with a 5-mV span and a chart speed of 0.5 inch per minute. The red- and black-pen katharometers and their respective power supplies are from the Gow-Mac Instrument Co., Madison, N. J. Each power supply has a n attenuator (maximum attenuation is X51.2); however, a second attenuator has been installed on the black-pen power supply in parallel with the first. The sample travels from the concentration trap, D, which is in oven H a t 120" C, through the red-pen katharometer, sample column, and black-pen katharometer to the collection trap, C, which is immersed in liquid nitrogen. The red-pen katharometer provides information on the concentration of the sample as it enters the column. (The significance of

sample concentration and shape of the intake peak is discussed under Example 4, below.) After the sample has passed through this katharometer, the red pen of the recorder can be quickly shifted to the second attenuator of the blackpen power supply. The red pen can thereby operate either separately or in conjunction with the black pen of the blackpen katharometer. After registering the intake peak (usually on R X 512), the red pen is transferred to and left a t RB X 32-i.e., it is connected to the second attenuator of the blackpen power supply a t an attenuation of X32. By leaving the red pen a t one attenuation, the relative heights of the peaks can be determined by direct comparison. The attenuation of the black pen can be varied either to show detail or to draw the top of all peaks which are off-scale on the red pen. Also, the attenuations can be chosen so that the red pen will show all the large and the black pen all the small peaks, thus allowing the equipment t o operate unattended and to be shut down by a timer. The Concentrator. The concentrator operates o n the principle of cyclic enrichment utilized by Mackay (16). The sequence in which its parts are connected is apparent from Figure 2. A (see Figure l), B, C’, D’,and J’ are four-port switching valves made from stainless-steel and Teflon, from the PerkinElmer Corp., Norwalk, Conn. The maximum permissible temperature for these valves is 150” C. N o oil or grease is used anywhere in the system. Valve C’ is part of collection trap, C, which is made from 0.250-inch o.d., 0.030-inch wall copper tubing. Lengths u p t o 50 feet have been used. This trap is used to collect the pyrolyzate after it leaves the pyrolysis oven or the G L C . Valve D’is part of concentration trap, D,which is made from 0.125-inch o.d., 0.030-inch wall copper or stainless-steel tubing u p to 15 feet long or from 3-mm 0.d. glass tubing. (The glass concentration trap is used only t o permit observation of the pyrolyzate as it condenses. It is not used routinely, because of the fragility of the glass and because the pyrolyzate may react with the glass). After connecting the traps as shown in Figure 2 the material in the larger trap, C, is transferred to the smaller trap, D. Valve J’ is part of the reaction vessel, J (Figure 2), which can be made of glass or metal tubing and may contain a reagent such as calcium chloride, concentrated sulfuric acid, molecular sieve pellets, o r a platinum, palladium, or hydrogenation catalyst (using hydrogen in the lines). All connections are made with Swagelok fittings (not shown) (Crawford Fitting Co., Cleveland 10, Ohio) with Viton-A, Fluorel, or silicone O-rings in place of Swagelok ferrules. The flow indicator, E, is made from 3-mm 0.d. glass tubing which contains a stainless-steel ball 0.0625-inch in diameter. The straight section is so tilted that the ball will run slowly in the opposite direction to that of the carrier gas. The tubing is constricted near the ends to prevent the escape of the ball. If trap D or the copper connection tubing of the concentrator should become blocked, the ball ceases to move. The volume of the sample at the temperature and pressure of the G L C column inlet and the amount and boiling points of the materials in the sample will determine the length, diameter, and design of the concentration trap, the choice of coolant, and the pumping rate and duration (which can be overnight, especially when a 50-foot collection trap is used). From the intake peak of Chromatogram 1, it is estimated that the volume of the pyrolyzate was approximately 18 cc at the temperature (120” C) and pressure (110-mm H g above atmospheric) of the G L C column inlet. Because calibration of the 0.125-inch 0.d. copper tubing showed that 5 feet has a volume of about 3 cc, u p to 30 feet of tubing could be used instead of 6 feet. This would allow a faster pumping speed and would decrease the time required for the transfer of material from trap C to trap D. (16) D. A. M. Mackay, ACS Summer Analytical Symposium,

Houston, Texas, June 14, 1960.

The pump, F, should have a low chamber volume, operate at 150” C, use no oil or grease, withstand 50 psig without leaking, provide a n abrupt, pulsating, spurting flow, have an adjustable stroke, and in this case withstand the action of hydrofluorocarbons. The Micro-Bellows Pump (Research Appliance Corp., Route 8, Allison Park, Pa.) with stainlesssteel bellows and Monel check valves most nearly met the above specifications. To understand why spurting flow is needed, it is necessary to consider what occurred when a steady flow of helium from the tank was allowed to sweep the contents of collection trap C directly into a glass concentration trap D that was immersed in liquid nitrogen, with valve J’ (see Figure 2) on “bypass”. The pyrolyzate condensed quickly in D where the glass entered the liquid nitrogen, G, thus blocking the tube. However, when the pulsating Micro-Bellows pump was used, the glass concentration trap, D, did not become blocked, because most of the material condensed quite evenly as a snowy layer over the inside portion of the trap beneath the liquid nitrogen; some did condense on the walls above the liquid nitrogen on the inlet side. When valve D ’of the glass concentration trap was turned from “through” to “bypass,” the liquid nitrogen bath was removed and the trap and contents were allowed to equilibrate at room temperature, and a significant fraction of the pyrolyzate remained a liquid. In this case, sampling the gaseous phase with a syringe would have produced grossly misleading results. If the blocking of trap D becomes a problem because of large amounts of material condensing a t room temperature, it can be avoided by one or all of the following changes. Two concentration traps can be placed in series, using coolants a t different temperatures; a series of coolants at increasingly lower temperatures can be used with one trap. A loop of the concentration D trap can be placed above the level of the liquid nitrogen, as shown in Figure 2, preferably in the vertical plane. Readily condensible material will collect in the loop and will not run into that portion of the trap that is cooled by contact with liquid nitrogen. The period of time for raising the temperature of the oven (Figure 2, I ) to 120” C can be increased. A pump meeting the above requirements but providing an even more abrupt pumping stroke can be used. If the charge is a mixture, its size can be reduced by removing components (peaks). Using the procedure described under Example 1 (below), make a chromatogram of the material that has blocked the concentration D trap, and catch the eluent from the G L C in a collection C trap. Decide which peak(s) or section(s) of the chromatogram are t o be abstracted to alleviate the blocking. Again transfer the material from the collection trap to the concentration trap using the concentrator as described under Example 1, and start a second chromatogram. Place two collection traps in series at the exit port of the G L C ; leave the second open all the time, and open and close the first t o remove the desired peak(s) or section(s). To date, raising or lowering trap D in the liquid nitrogen coolant has removed all blocks without the system being opened. The use of 0.125-inch 0.d. stainless-steel tubing instead of 0.125-inch 0.d. copper tubing for the concentration trap D appears to provide more trouble-free operation. If the sample consists of relatively large amounts of material that condense at room temperature and the concentrator is used a t room temperature, there may be serious losses in the copper connecting tubing due to condensation. This can be avoided by installing the concentrator, with or without the reaction vessel, J, in oven H (Figure 1) and using it at a n elevated temperature. Holes would have to be made in the oven to permit attachment to the bellows pump motor and the D trap (in a coolant). Before use, the concentrator must be free from leaks and contamination. The bellows and check valves, all tubing, and the four-port switching valves must be cleaned with solvents and dried before assembly. Also, the traps should be VOL. 39, NO. 10, AUGUST 1967

1127

cleaned with a flowing stream of hot water to remove any residual inorganic solder flux. The presence of a leak can be detected by means of a manometer which is attached at valve B (Figure 2). If a leak is present, it can be quickly located by placing the system, except for the bellows pump motor, under 50 psig of helium and then temporarily submerging it in water. After all leaks are repaired, the system (except for the bellows pump motor) is then placed in oven H (Figure 1) at 140” C with valve B connected to valve A . Helium is allowed to flow from the G L C through the concentrator and traps while additional heat is applied with a heat gun (Master Appliance Corp., Racine, Wis.) to all parts of the system except those containing Teflon. The heat gun is then removed and the base line of the recorder is allowed to stabilize. By comparing the base line of the recorder when the valve is on “through” with its position when valve A is on “bypass,” one can determine when the system is clean enough for use or, by judicious use of the valves, which part of the system needs further cleaning with solvents. Because the concentrator (with minor adaptations) can be operated while in oven H , contamination checks can be easily made after each use.

EXAMPLES OF THE USE OF THE CONCENTRATOR

In Examples 1, 3, 4, and 5 the sample column was 2 3 z didecylphthalate (DDP) deposited on 60/80 mesh, acid-washed, fire brick in a 30-foot length of 0.250-inch 0.d. stainless-steel tubing. In Example 2, the sample column was 16.7z by weight polyalkylene glycol (LB-550-X) on 60/80 mesh, acidwashed, fire brick in a 30-foot length of 0.250-inch 0.d. stainless steel tubing. The collection trap, C, was 25-foot x 0.250-inch 0.d. copper tubing; the concentration trap, D, was’6-foot X 0.125-inch 0. d . copper tubing. For all runs, the columns and valve A were held isothermally at 120” C , the katharometers at 130” C, and the helium flow rate was held a t 30 cc/minute. Example 1. The concentrator was used t o determine whether a given chromatogram is reproducible. The following steps were carried out. The C trap, containing the pyrolyzate of plasticizer-free polyvinylidene fluoride, was disconnected from the pyrolysis oven (or GLC), and removed from the liquid nitrogen bath; it and a clean D trap were placed o n the concentrator as shown in Figure 2 without J. Valve B was turned to “through” and the connecting tubing was heated with the heat gun and flushed out with pure helium with the pump turned on. Valve B was turned t o “bypass,” and valves D ’ and C’ in that order to “through.” The temperature of the C trap in oven I was slowly raised to 120” C ; the pyrolyzate was pumped to the D trap over a period of 2 to 3 hours. Valve C’ was turned to “bypass,” the Dewar was raised, and liquid nitrogen was added until all the loops of the D trap were covered. The helium was removed from the D trap using valves D‘ and B and a mechanical vacuum pump. (If this is the first time a chromatogram of a given charge is being made, it is recommended that this step be omitted because evacuation of the helium may simultaneously remove low-boiling constituents. The identical material should then be prepared for a second chromatogram and the helium removed from the D trap. Generally the peaks will be sharper and the resolution better. If comparison of the two chromatograms shows that no peaks are missing from the second, the helium removal step may safely be included in the preparation procedure.) Valve D‘ was turned to “bypass” and trap D connected a t valve A (Figure 1) with oven H a t 120°C. Then valve A was turned t o “bypass,” and the connecting loop was flushed out with helium from the GLC 1 128

ANALYTICAL CHEMISTRY

while the D trap was allowed to stabilize at 120” C. Meanwhile the C trap was removed from the concentrator, placed in liquid nitrogen, and connected to the end of the sample column of the G L C as shown in Figure 1. Valve C’ was then turned to “through.” When the base line was again steady, valve D ’ also was turned to “through” and Chromatogram 1, Figure 3, was obtained. The C trap (Figure 1) containing the pyrolyzate and the D trap were placed back on the concentrator and the process was repeated for Chromatogram 2. (All subsequent chromatograms were made by the same procedure.) Comparison of Chromatograms 1 and 2 shows that reproducibility is excellent. That the first two peaks (at the right of point A ) of Chromatogram 2 are larger than those of Chromatogram 1 indicates that some of the components of the pyrolyzate had decomposed and that two of the products of decomposition were of the same material as those of the first two peaks. (Because the pyrolyzate was a mixture of hydrofluorocarbons and because the solid support in the column contained silica, some decomposition WRS expected.) The rest of the peaks of Chromatogram 2 were smaller than the corresponding ones of Chromatogram 1 , the amount of decrease varying from peak to peak but always less than 4 %. Because subsequent pyrolyses of different samples of the same polymer yielded practically the same chromatogram, a new pyrolysis need not be made for each peak to be examined. When 20-25 peaks for several different polymers are involved, this amounts to a considerable saving of time and effort. The intake peaks are quite sharp, indicating that practically all the material in the concentration D trap in oven H (Figure 1) was in the vapor state at 120” C when admitted to the GLC. Mixing with the carrier gas is thus minimized, which results in improved peak resolution. Example 2. Chromatogram 3 was obtained by putting the pyrolyzate through the LB-550-X column. Comparison of Chromatograms 2 and 3 shows that all peaks from points A up to points B are somewhat similar. However, t o the right of B , the Chromatograms are quite different; a peak in this region on Chromatogram 2 would probably be shifted to a different retention time on Chromatogram 3, thereby revealing any peaks that had been hidden beneath. Additional chromatograms, using many other columns, could be made until those needed to shuffle the peaks of interest are found. This separation of adjacent or overlapping peaks will facilitate the collection of individual peaks. Also, individual peaks can be tested for purity. Example 3. I n order t o determine if any of the pyrolysis products contained multiple bonds, the pyrolyzate was circulated for 2 hours through 1 cc of concentrated sulfuric acid in reaction vessel, J , which was a glass U-tube, (concentration D trap temporarily on “bypass”) and Chromatogram 4 was made. Comparison of Chromatograms 1 and 4 shows that a major peak, W , of Chromatogram 1 has disappeared (this peak was subsequently found to be acetone which had remained o n the sample holder of the pyrolysis oven after cleaning), revealing two smaller peaks, X , on Chromatogram 4. The considerable decrease in the size of the rest of the peaks, and also the blackened state of the sulfuric acid, indicated that a multiple bond might be present in most of the pyrolysis products. A new peak appeared in Chromatogram 4 (at Y) but this was not investigated. Example 4. I n order t o assess the effectiveness of the concentrator, Chromatogram 5 was made without using it. The material of Chromatogram 4 was caught in collection trap, C, which was immersed in liquid nitrogen at the end of

CHROMATOGRAM I

CHROMATOGRAM 2

A

m

B

TIME-

CHROMATOGRAM 3

A

TIME-

B

CHROMATOGRAM 5

C

TIME

-

CHROMATOGRAM 6

Figure 3. Chromatograms of pyrolyzate under various conditions the G L C sample column, and the helium was evacuated. The C trap was then connected at valve A (Figure l), the arms were flushed with helium from the GLC, and Chromatogram 5 was made as before. Next the material (from Chromato-

gram 5) was caught in the C trap at the end of the sample column and, using the concentrator, Chromatogram 6 was made. Chromatograms 4 and 6 (both made by means of the concentrator) agreed closely. Any differences between VOL. 39,

NO. 10, AUGUST 1967

1129

them and Chromatogram 5 were due solely t o the use of the concentrator. U p to points C (the first 70 minutes), when about 95% of the material had eluted, there is n o resemblance between Chromatogram 5 and Chromatograms 4 and 6. The chance of trapping any of the pyrolyzate components in a state pure enough for identification, using the equipment and technique which produced Chromatogram 5 , is so remote that it is not worth attempting. However, for the remaining time (about 2 hours, not all of which is shown) there is quite a strong resemblance. This can be explained by assuming that the factors which cause the peak to broaden are statistically independent and hence their variances are additive (17). O n the basis of this assumption, the equation:

can be derived (18), where N is the true plate-number of the column, N' is the empirical plate-number, sr2is the variance of the input distribution, and VR" is the retention volume. For Chromatograms 4 and 6, the variance of the input distribution was quite small because, when valve D' was opened, there was apparently very little mixing between the sample and the helium from the GLC in the 0.125-inch 0.d. copper tubing; thus a reasonable approximation t o plug flow resulted. The above equation then reduces to N' = N , and optimum resolution, starting with the initial peak, is obtained at fixed GLC operating conditions. There is some tailing at the end of the intake peak. This may be due either to turbulence or to high-boiling material which was not entirely in the vapor state at 120" C when the D trap was opened, or to both. The variance of the input peak of Chromatogram 5, 2, is comparatively large. This indicated that there was considerable mixing between the carrier gas and the sample when the 0.250-inch 0.d. tubing was used. This is to be expected inasmuch as the volume of the sample a t the column inlet temperature and pressure was estimated to be about 18 cc, whereas the volume of the C trap was 140 cc. N' is therefore much less than N , and resolution suffers accordingly. It is not until a retention time of 1 hour is reached that the retention volume becomes large enough for N' to effectively equal N , and Chromatograms 4, 5 , and 6 become similar. Example 5. One of the objects of this research (which necessitated the development of the concentrator) was t o determine whether powdered metals affect the decomposition of the powdered, plasticizer-free, polyvinylidene fluoride. Accordingly 4 samples (3.0 grams each) of the pure, powdered, polyvinylidene fluoride were pyrolyzed at 450" C, and chromatograms were made of the more volatile products. The mean and standard deviations of the peak heights of 22 of the large and medium peaks were computed. Three grams of the polymer were thoroughly mixed with 32 grams of Zn dust and the mixture was pyrolyzed a t 450' C. Comparison of the chromatogram obtained showed that 10 peaks were greater than the corresponding ones of the pure polymer by (17) A. B. Littlewood, "Gas Chromatography," Academic Press, New York, 1962, p. 141. (18) A. B. Littlewood, "Gas Chromatography," Academic Press, New York, 1962, p. 162.

1 130

ANALYTICAL CHEMISTRY

six times the standard deviation, and that there was one new peak. Using 40 grams of Cu dust (same volume and somewhat the same surface area as the Zn dust), six peaks were found to exceed the corresponding ones of the pure polymer by more than 6 times the standard deviation. Hence both Zn and C u d o affect the decomposition of polyvinylidene fluoride and/or pyrolysis products a t 450 " C. Other materials, such as AI, AIF3, Fe, and Cr200,were also used, but the above critical comparison was not made. Both AI and AIFB decreased many of the peak heights by 5 0 z . Iron caused 6 peaks to be over 100 larger and 3 new peaks to appear. When C r 2 0 3 was used, the water produced was removed by using the J trap (Figure 2) containing CaCI2. The resulting chromatogram showed that C r s 0 3 affected the decomposition very strongly; most of the peaks were 0.1 their normal size. The new peaks appeared in regions where the peaks on the chromatogram of the pure polymer were quite small. The new peaks as well as those of increased height may be due t o the formation of new compounds o r simply to the formation of more of those which appear on the chromatogram of the pure polymer. In all of the above cases there were many peaks of normal height. This rules out the presence of a leak or the accidental use of more than 3 grams of polymer as the cause of the deviations. A description of the pyrolysis oven is given in Reference (19). An attempt was made t o identify the major peaks in the chromatogram of the pure polymer. The largest peak (labelled B X 128 on Chromatogram 4) amounts to about 25 of the material chromatographed. Its infrared spectrum matched that of sym-trifluorobenzene (20), and the base and maximum peak of its mass spectrum was at mass 132, the proper molecular weight. Because the low molecular weight materials would be the easiest to identify, a major peak with a shoulder occurring near the beginning of the chromatogram obtained using the DDP column was selected for purification and identification. A DC silicone oil 200 column (21) broke the peak and its shoulder into 13 peaks. Other major peaks were also resolvable. Matching infrared or mass spectra for most of these materials could not be found in the literature. Fluoroform, CH2=CF2, CH2=CHF, and CF2=CHF were easily identified from their mass spectra. Some other compounds have been tentatively identified from their mass spectra as CF2=CH-CF2H, CF2=CH-CF3, CHF= CH-CFj, CFz=CF-CF:H, CF2H-CF2-CH3, CFa-CFz -CH3, and CF3-CH2-CF2--CF2H. Patent rights on this apparatus are reserved. ACKNOWLEDGEMENT

The very capable assistance and many helpful comments of V. T. Stapler are gratefully acknowledged. I also thank Maurice Bazinet for help in interpreting mass spectra. RECEIVEDfor review August 1, 1966. Accepted January 30, 1967. (19) R. A. Prosser, J. T. Stapler, and W. E. C. Yelland, ANAL. CHEM., 39, 694 (1967). (20) J. R. Nielsen, Liang Ching-Yu, and 0. C. Smith, Discussions Faraday SOC.,No. 9, 177 (1950). (21) H. T. Rein, M. E. Milville, and A. H. Feinberg, ANAL.CHEM., 35, 1536 (1963).