good accuracy. However, as the compound purity is lowered to the 95% range, deviations from theory increase t>o an extent that accurate hydrogen counting would be difficult. The spectra of two of the compounds, ethanol and tetralin, were integrated three times a t each of several values of the r.f. field. This was done to see if selective saturation could be observed a t high r.f. values and if so, to determine the optimum operating range of ref. power. In Figures 1 and 2 the magnitude of the integral for each proton type and total hydrogen in ethanol is plotted against the different ref. field settings. Similar plots for tetralin are shown in Figures 3 and 4. Inspection of these figures shows that maximum integral amplitude is obtained a t r.f. settings of 0.2 and 0.4 milligauss and that the amplitude drops a t higher values of r.f. power and even more rapidly a t lower ref. power. The prob-
able explanation of drop-off a t low r.f. power is nonlinearity of the automatic gain control incorporated in the A-60 circuitry. (The automatic gain control is intended to attenuate the signal proportionally as the r.f. power is raised.) The gradual drop-off at high r.f. power is probably due to the -1/2yHI)/R term of Equation 8. Table I1 shows the per cent hydrogen found and the deviation from theory for each hydrogen type in ethanol, obtained from the integrals a t each r.f. setting. Figure 5 is a plot of these data showing evidence of selective saturation as the r.f. field is increased. Figure 6 is a plot of the deviation from theory of each proton type against the r.f. setting. There is a definite trend in the curves, demonstrating selective saturation a t r.f. settings greater than 0.2. The data for tetralin are indicated in Table I11 and shown in Figure 7. The results here are similar
to those for ethanol except selective saturation does not occur until the r.f. field is increased above 0.8 milligauss. CONCLUSIONS
It is concluded that apparent deviations from theory noted earlier were simply due to impurities in the samples under examination. Under optimum conditions of sample concentration, r.f. power, gain, and sweep rate, as employed in this paper, integrals should be accurate to within a few tenths of 1% of the total hydrogen content. LITERATURE CITED
(1) Reilly, C. A., ANAL. CHEM.30, 839
(1958).
(2) Varian Associates, Instrument Division, Technical Information Bulletin 3, No. 1 (1960). RECEIVED for review February 14, 1963. Accepted April 8, 1963. Presented at 4th OCEANS, Pittsburgh, Pa., February 28-
March 2, 1963.
The Physical Basis of Analytical Atomic Absorption Spectrometry The Pertinence of the Beer-Lambert Law KEllCHlRO FUWA and BERT 1. VALLEE Biophysics Research laboratory o f the Division of Medical Biology,Department o f Medicine, Harvard Medical School, and the Peter Bent Brigham Hospital, Boston, Mass.
b In this study the absorption of molecules in solution as expressed b y the Beer-lambert law has been used as a model for the investigation of atomic absorption. As predicted by this model, the sensitivity of measurement is a function of the length of the absorbing flame path and of the reflectivity of the surrounding cell surfaces. Employing absorption cells here described, the limit of detection of many elements has been lowered greatly. The limits of detection for zinc, cadmium, magnesium, copper, nickel, and cobalt have been found to vary from 0.4 to 16 p.p.b. The utilization of absorption cells for atomic absorption spectrometry should greatly enhance the usefulness of this analytic procedure.
0
VER THE COURSE of
a very few years since its discovery by Walsh (9) atomic absorption spectrometry has achieved major prominence as an exquisite analytical method. Exceeding, however, the proceeds of the pioneering in this field up to the present, is the tremendous potential for further de942
ANALYTICAL CHEMISTRY
velopment. Knowledge of the physics underlying the phenomenon of atomic absorption has permitted the prediction that marked improvements in the sensitivity of detection of many elements will be forthcoming (7, 9). Although the principles which are common to molecular and atomic absorption spectrometry are readily apparent, the pertinence of this similarity to the further analytical exploitation of atomic absorption spectrometry has not been fully appreciated. The investigation of various types of absorption cells, described in the present communication, demonstrates the pertinence of the Beer-Lambert law to atomic absorption spectrometry. The proportionality between the light entering a solution and to that absorbed as a result of the chemical and physical characteristics of its contents, to the concentration of the absorbing medium and the length of the absorbing path are so well known that they need not be dwelt upon further. The same principles were thought to apply both to atoms and molecules in gas or aerosol phases and hence to atomic absorption. This circumstance has led us to under-
take a systematic investigation to define the optimal characteristics of absorption cells for use in this system. As a result of these studies the sensitivity of detection of many elements has been increased from 10- to 100-fold over those reported in the literature (1-4, 7). DESCRIPTION OF THE INSTRUMENT
Light Source. Hollow cathode discharge tubes for zinc and copper, WL-22607, magnesium WL-22604, cadmium WX4936, (Westinghouse Electric Corp.); nickel F L 155 and cobalt F L 157, (Hilger and Watts Ltd.) are used as light sources. Individual tubes are connected to a constant-current power supply with 800-d.c. volts compliance and operated a t an appropriate current between 10 to 60 ma. The light from the hollow cathode lamp is collimated by a quartz lens of 12-cm. focal length placed 7 cm. from the tube window; it then passes through the absorption cell and is focused on the entrance slit of the spectrograph by another quartz lens of 15.5-cm. focal length placed 19.5 cm. from the slit. The distance between the two lenses is 113.4 cm. (Figure 1). Absorption Cell. Absorption cells have been constructed of various
Vycor,Cell; Length: 91 c m
1
Absorption Cell With ~ Magnesium Oxide Sheath
of Optical System
~
0
)
I
Burner/
Spectrograph Figure 1. Absorptiori cell assembly for atomic absorption spectrometer and absorption cell Cell covered with magnesilm oxide sheath and 3-mm. exit orifice i s shown in the insert
kinds of combustion tubing: Vycor (inside diameters; 1, 1.65, and 2.6 cm.; length: from 23 t o 91 cm.), asbestos (inside diameter: 1 cm.; length: 91 cm.), Alundum (inside diameter: 1.7 em.; length 34.3 cm.); graphite (inside diameter: 1.7 cm.; length: 38.1 cm.) or Zirconia (inside diameter: 2.0 cm.; len5th: 76.2 cm.). The absorption cell is held on a strip of asbestos shielding iiside a length of borosilicate glass tubing, with an inside diameter of 4.8 cm. and an over-all length of 99 cm., plmed horizontally between the two lenses). The open end of the cell is located near the focusing lens. The light from :he source passes through the center 0 ' the absorption cell. A side arm of 1,he outside glass tubing, 27 cm. from t i e end and close to the collimating lens, is connected to an exhaust system to protect the lenses from hot gases (Figure 1). Burner. A Beckman atomizer burner (KO.4020) is used t o spray the sample into the absorption cell. The fuel is a mixture of commercial hydrogen, a t a pressure of 10 t o 20 inches of water, and air, a t a pressure of 15 pounds per square inch. The hydrogen pressure i j adjusted for each type of absorpticln cell to obtain optimal flame length and maximal sensitivity The flow rate of solution through the burner i,3 approximately 1.5 ml. per minute. The burner is placed adjacent to the absorption cell near the focusing lens and is tilted a t an angle of 15' to the optical axis to allow the flame to enter the tube without interfering with the light beam; the upper surface of the burner is filed off to prevent such interfelence (Figure 1). Light Dispersing .Element. Two spectrographs, one a Hilger E 2 medium quartz instrument and the other a Bausch and Lomb, large Littrow spectrograph with exchangeable quartz and glass optics were adapted to serve as the light dispersing elements. For both spectrographs the entrance and exit slits were adjusted to a width of 0.4 mm. Detectors. The follcwing lines were employed t o detect the respective elements: zinc 213S.h6A., cadmium 2288.0188., magnesium 2852.129A., copper 3247.5408., cobalt 2424.9308., and nickel 2310.9668. Multiplier phototubes (RCA 1P28) were used as
0' 0
I
0. I
I I 0.2 0.3 Zinc, ,ug per ml
I
0.4
Figure 2. Repeatability of atomic absorption measurements with a Vycor cell; length 91 cm., diameter 1 cm. Twelve measurements were performed a t each concentration of zinc from 0.01 to 0.4 p.p.m.
the detectors for the absorption measurements. The dynode voltages were identical for all stages, and were varied from 70 to 120 volts per stage. The camera of the original spectrographs were replaced by facilities for mounting multiplier phototubes. An aluminum plate was coated with a photographic emulsion and was exposed to the hollow cathode light source. The plate was removed and developed, and the area of a particular absorbing line was excised precisely, and the rest of the plate was then blackened. After returning it to the original focal plane, the excised lines of the plate served as exit slits. The mounting of multiplier phototubes was similar to that used previously for a multichannel flame photometer which has been described (8). The windows of the tubes used for the detection of zinc, cadmium, cobalt, and nickel mere covered by a thin layer of sodium salicylate to increase the tube sensitivity at the low wavelengths for which they were employed. 05,
1
I
V y c o r C e l l , Length: 91 cm
1
Zinc, pg per ml
Figure 3. Effects of diameter of absorption cell on absorbance
Amplifiers. The instrument provides a high voltage power supply and amplifier for the multiplier phototube. They were arranged on a common chassis together with the hollow cathode lamp power supply. The high voltage power supply provided a regulated voltage variable from 700 to 1200 volts d.c. for the multiplier phototube. The circuit consisted of a Y50HF rectifier followed by a voltage regulator. Output voltage was controlled by a total 1-megohm variable resistor, coarse and fine, on the control grid of a 12AX7 tube. The signal input to the amplifier was passed through an attenuator provided with precision resistors, so that the sensitivity of the amplifier might be varied by factors of 3 and 10 for each step of the attenuator switch. After passing through the attenuator the signal was sent to one side of a 12AU7 tube in a bridge circuit, which included a variable resistor to permit setting of the amplifier balance. Circuits for the high voltage power supply and amplifier were almost identical with those used in the multichannel emission flame spectrometer (8). Meter. The output signal of the amplifier was indicated on a direct current microammeter with a range of 0 to 100 pa. This could be read easily to =k1pa. Standard Solutions. These were solutions of zinc, cadmium, magnesium, copper, cobalt, and nickel chlorides. Rods of zinc, cadmium, magnesium, and copper, cobalt sponge, and nickel oxide, spectrographically pure, (Johnson, Matthey and Co., London) were weighed out, and dissolved in 6 N hydrochloric acid. The solutions contained 10,000 p.p.m. of metals. These concentrated stock solutions were diluted further with water purified by a mixed bed ion exchange resin column. The concentrations used covered a range from 0.0001 to 1.0 p.p.m. depending on the sensitivities of detection for individual metals. VOL 35, NO. 8, JULY 1963
943
1
I 0
/I
Vycor C e l l , Diameter l c m
1
001'
I
I,
30 50 70 Vycor C e l l L e n g t h (cm)
J 90
Figure 4. Effects of length of absorption cell on absorbance
EXPERIMENTAL
The air pressure to the burner is fixed a t 15 lbs. per sq. inch and the hydrogen pressure adjusted until the microammeter gives a reading half that obtained before the flame is lit. After allowing the system t o come to temperature equilibrium, the meter is zeroed by adjusting the dark current with the slit closed. Then, with the slit open, metalfree water is atomized through 8 cm. of polyethylene tubing attached to the burner capillary. During atomization, the meter is adjusted to an arbitrary light intensity, lo,of 80 pa., by adjusting the voltage applied to the photomultiplier dynodes. A series of standard metal solutions is then atomized and the absorbed line intensity, I , is obtained for each. RESULTS
The stability of the apparatus was tested by measuring the repeatability of a zinc solution (Figure 2). The relative standard deviation was less than 4y0 for all concentrations above 0.05 pg. per ml. Using the zinc line a t 2139A. the following characteristics of the absorption cell were examined: A. Effect of Cell Diameter. The results with 91 cm. long Vycor cells having diameters of 2.6, 1.65, and 1.0 cm. are shown in Figure 3, which demonstrates that the smaller the diameter of the cell the higher the sensitivity. In all cases, sensitivity is lost a t high concentrations, above 0.1 p.p.m. for a 1-cm. tube (Figure l), above 1 to 2 p.p.m. for those of larger diameter (not shown). Sensitivity is not increased when the inside diameter is reduced below 1 cm. E. Effect of Cell Length. This was eyamined with a Vycor cell having an inside diameter of 1 cm. which was found optimal (Figure 3). The results 944
*
ANALYTICAL CHEMISTRY
are shown in Figure 4. As the length of the cell is increased to about 70 cm., there is a corresponding linear increase in absorbance. On further increases of length, however, the rate of increase in absorbance diminishes until there is very little further change at a length of about 90 cm. C. Reflection from the Inner Surface of the Tube. A rough-surfaced, asbestos cell (91 by 1.0 cm.) was prepared, and the absorbance in this cell is compared (Figure 5) to that of a Vycor cell of similar dimensions. At a given concentration of zinc the absorbance observed in a Vycor cell is about 10 times that observed with the asbestos cell. D. Effect of a Magnesium Oxide Sheath. The Vycor cell employed in Figure 1 was surrounded by a layer of magnesium oxide powder, 5 mm. in thickness, to serve as an additional reflector from the outside of the cell. As seen in Figure 6, a 16% increase in response is obtained, when compared to that of the control Vycor cell. E. Effect of Pressure. To test the effect or sensitivity of increasing the pressure within the cell, the exit of the Vycor cell used in D (vide supra) was modified by inserting an asbestos stopper containing another cell, 0.3 cm. in diameter, a t the end of the cell employed in Figure 1. The results are also shown in Figure 6. At a zinc concentration of 0.1 p.p.m., a 30% increase in absorption was obtained. Quantitative Determination of Cadmium, Magnesium, Copper, Cobalt, and Nickel. To construct working
curves, measurements were also made using the following elements and lines: Zn 2139A., Cd 2288.4., PIig 2852A., Cu 3248A., Co 2425A., and Ni 2311A. The results of these experiments, using a Vycor cell (91 by 1 cm.), are shown in Figure 7 . DISCUSSION
The analytical utilization of atomic absorption rests on the principle that the total absorption of light, expressed as log Io/Z,is directly proportional to the concentration and absorptivity of metal atoms in the flame. This has been well documented in flames of varying dimensions and characteristics (5, 7). The absorption of light by a homogeneous species of molecules in solution is similarly dependent upon their absorptivity and concentration. In addition, however, the absorption of molecules in solution is directly dependent ulion another variable, the path length. In solutions this relationship is expressed by the Beer-Lambert law which states that
'o----Lz7J C e l l L e n g t h 91cm
A
1
"0
/ 01
7 02
1 03
04
Zinc, pq per ml
Figure 5. Effects of reflections from inner surface of Vycor and asbestos absorption cells on absorbance
Where a = absorptivity, b = the path length, and c = the concentration. It might be predicted, therefore, that atomic absorption would also be proportional to the path length of the flame. As shown by the present studies, when both absorptivity and concentration are held constant, the path length can be examined as the single variable of the experimental system. Increasing the path length of the flame was accomplished by recognizing that a flame is markedly lengthened when it is inserted into a suitable tubing. If the long asis of such a directed flame is employed as the absorption cell, its path length is increased as a function of the length of the tubing. Using this system the predicted dependence of absorption on the path length of the flame is clearly borne out. As expected, however, the dynamic equilibrium of atoms in the flame cannot be maintained beyond certain limitations-e.g., the flame cannot be propagated infinitely. This is shown by the eventual failure to increase sensitivity as the path length is increased beyond 70 t o 90 cm. (Figure 4). To define the precise limits of the contributions of the path length of absorbing flames to a modified Lambert's law for atomic absorption, it will be necessary to delineate the contributing physical factors which differentiate absorbing systems in liquid and gaseous phases from thope in solution. Such studies are under way. In molecular absorption spectrometry scattering of light from the surfaces of the absorption cell introduces significant errors of measurements. The absorbance varies as a function of wavelength, and under the conditions which usually prevail in practical absorption spectrometry, the light entering the absorption cell is not strictly monochromatic; the band width of the light beam entering the solution is a function both of the re-
"1
Vycor Cell; L.bngth: 91 c m Diameter: I c m ~
0
002
004 006 008 Zinc, pg per ml
010
0.I
Figure 6. Effects of magnesium oxide sheath and increaslsd pressure of absorption cells on absorbance
solving power of the dispersing element and of the entrance and exit slits of the spectrometer. Herice, there is background due to hetsrochromatic radiation. Light scattering will further contribute to this background absorption, as a fourth power of the wavelengths of the entering beam, and also as a function of the absorbance 0 " the material under study. I n contrast, in atomic absorption spectrometry the in tense radiation from the hollow cathode tube which enters the flame cell is strictly monochromatic since a specific electronic transition of the light source is selected for absorption in the gas or liquid-aerosol phase. Hence, this scattered monochromatic radiation, when reflected into the cell, will actually enhance absorption, while background will not be increased correspondingly. The superiority of the Vycor over the asbestos cell demcnstrated by these experiments may bl: accounted for on this basis. The effective path length is increased tenfold in the Vycor cell over the actual cell length. This is apparently due to a tenfold increase in circular reflections from the inner snrface of the Vycor tubing over that obtained with the rough-surfaced, nonreflecting asbestos cell. The use of a magnesium oxide clover demonstrates that outside reflection may be employed similarly to enhance the effect. The loss in sensiti1;ity observed with tubes having a diameter greater than 1 cm. may be attributed to a decrease in circular reflectivity of the larger cells. It may be predicted from the BoyleCharles law that the concentration of arl absorbing species, b , will increase with increased vnl)or prersure in the absorption cc.11. The increased sensitivity resulting from conqtriction of the open end of the absorption cell bears out this anticipation euperimentally. The higher absolute sensitivity ob-
0.2 0.3 0.4 Concentration, pg per ml
0.5
Figure 7. Typical calibration curves for cadmium, zinc, magnesium, copper, cobalt, and nickel as measured b y atomic absorption
Table I.
Atomic Absorption Enhancement of Sensitivity for Zinc Analysis Zn detectable, Log I J I = 0.1 Detection limit, pg./ml grams pg. /ml. 3 x 10-7 1. Elongated burnera 7.5 0.3 2. Beckman burner 3 x 10-8 8.2 0.3 a. Vertical flame 4 x 10-9 b. Horizontal flame 1 .o 0 04 1 x 10-9 c. Asbestos cell 0.3 0.01 d. Vgcor cell 6 X lo-'" 0.16 0.006 Diameter 2.6 cm e . Vycor cell 6 X lo-" 0.016 0.0006 Diameter 1 cm. f . Vycor cell 4 x 10-11 0.01 0.0004 Diameter 1 cm. MgO sheath 3 x 10-11 g. Vycor cell 0.005 0,0002 Diameter 1cm. MgO sheath a
Higher Pressure Hilger-Watts Ltd., attachmeiits No. H909 and H1090.
tained by the use of the optimal conditions here defined permits the measurement of extremely small samples. This is of great importance when the size of the original sample and the concentration of the element undei examination is limited, as is often the case with biological material. I n this regard, the total consumption burner used in the present apparatus is greatly superior to those of the atomizerchamber type where the volume of samples required is about 10 times greater. The definition of the detection limit employed in Tables I and I1 is the concentration in micrograms per milliliter of metal which results in 1% absorption, or approximately one division of the needle deflection on the microammeter used for read-out. The absolute amount of detectable metal, in gramq, therefore, is the smallest quantity required to give a reading of 1% absorption. The values would vary according to the sample flow rate and the rate of response of the detection system. In the present system, 0.1 ml.
Table 11.
Atomic Absorption Detection Limits Vycor cell, length 91 em. Diameter 1 cm. Elements pg. per ml.
Cadmium Zinc Magnesium Copper Cobalt Xickel
0.0004 0.0006 0.005
0,007 0.013 0.016
of solution, which has a metal con-
centration equal to that defined above as the detection limit, constitutes the absolute amount of detectable metal. Apparatus currently available commercially is equipped with a chamberatomizer that feeds into a n elongated multiple Bunsen burner having a flame width of only about 10 cm. ( 5 ) . Table 1 compares the sensitivity of the present arrangement (lines 2a to g) with that of a commercial instrument line 1 using the conditions defined to yield the highest sensitivity reported for both. For our studies a 91- X 1VOL. 35, NO. 8, JULY 1963
945
cm. Vycor tube covered with magnesium oxide powder, the exit of which is constricted, was employed. The lowest detection limit for zinc under these conditions is lo-” gram, or 0.1 p.p.b. In the early stages of our investigation of absorption cells a T-tube flame adaptor was placed over the burner, the flame being deflected in both directions. This system in our hands failed to yield any increase in sensitivity for the metals studied. Presumably this is due to the fact that only the upper part of the flame is deflected into the part of the tube which serves as the absorption cell. Thus there is a major disturbance of equilibrium of the flame due to the sudden change in the combustion brought on by the abrupt change in direction. Our negative results with this system are in accord with the conclusions of Robinson that such an “adapter could not be used for the measurement of the alkali metals or some of the transition metals” (6). Table I1 shows the detection limits for six elements using a plain Vycor cell of 91 X 1 cm. Obviously, at these low concentrations corrections for simultaneous emission of the absorbing species are unnecessary. This is true
particularly for elements such as zinc or cadmium which are excited with difficulty at these flame temperatures. Although Vycor has been used in this work, the use of other materials for absorption cells is not precluded, of course. The only requirements for the cell material are that they have high temperature resistance, be chemically inert, and have a highly-reflecting surface. Other materials may well allow the use of higher flame temperatures. If the sensitivity requirements are not stringent, a nonreflective cell may be manufactured of materials like alundum or other ceramics, graphite, or asbestos. This might be an advantage since deposits on the surface are of no importance when reflectivity is not being exploited to enhance sensitivity. The absorption cells may also be made to serve as a combustion tube, such that the organic matter with which the metals of interest may be combined may be burned while the sample is being atomized. In the process the metal atoms are liberated for analysis without prior ashing. Further studies of the physical factors affecting atomic absorption are in progress. From these data it may safely be
predicted that even greater sensitivity without interference may be anticipated with further development of absorption cell characteristics and the design of special burners for use in this system. ACKNOWLEDGMENT
It is a real pleasure to acknowledge the valuable comments and advice of W. E. C. Wacker, both in the course of this work and in the preparation of the manuscript. LITERATURE CITED
(1) Allan, J. E., Analyst 83,466 (1958). (2) Allan, J. E., Spectrochim. Acta 17, 459 (1961). (3) David, D. J., Analyst 83,655 (1958). (4) Ibid., 85,779 (1960). 32. 898 151 ~, Mensies. A. C.. ANAL.CHEM. (1960). (6) Robinson, J. W., Anal. Chim. Acta 27,465 (1962). (7) Russell, B. J., Shelton, J. P., Walsh, A., Spectrochim. Acta 8, 317 (1957). (8) Vallee, B. L., Margoshes, M., ANAL. CHEM.28.175 (1956). (9) Walsh, ’A., Apectrochim. Acta 7, 108 (1955).
Work supported by the Howard Hughes Research Institute and a grant-in-aid from the Petroleum Research Fund of the American Chemical Society.
Application of Time of Flight Mass Spectrometry and Gas Chromatography to Reaction Studies E. J. LEVY,l E.
D. MILLER,
and
W. S. BEGGS
The Atlantic Refining Co., Philadelphia, Pa.
b A technique i s described which permits the rapid qualitative and quantitative identification of the products from a complex chemical reaction. The system employed consists of a small scale reactor, a two-position indicating sample loop, a temperature programmed gas chromatographic unit, and a Bendix time of flight mass spectrometer. Identification of the gas chromatographic peaks is accomplished by mass spectrometry whereas the quantitative analysis is based on the chromatographic peak areas. The problems involved in operating the Bendix time of flight mass spectrometer under these conditions are discussed. Some results obtained in the catalytic cracking of nonane over silica alumina are presented as an example. The techniques described are applicable to any thermal, photochemical, or catalytic reaction, which can be carried out in a flow system. 946
ANALYTICAL CHEMISTRY
T
HE WORK to
be described represents an application of time of flight mass spectrometry and gas chromatography to the study of the changes in product distribution, as a function of cracking cycle time ( I ) , during the cracking of n-nonane over a silica alumina catalyst. Cracking cycle time may be defined as the period that a catalyst is exposed to “feed” before it is regenerated. This paper will describe the experimental techniques and some of the initial results. In the usual microcatalytic studies a hydrocarbon feed is allowed to pass over a catalyst for a given period of time-e.g., 15 minutes and then the products are collected and analyzed and the per cent conversion for the 15minute period determined. The per cent conversion determined in this way is an integral conversion and represents the total conversion for the 15-minute period.
In the present investigation, the cracking period was divided into short intervals of approximately 28 seconds (corresponding to one slug of feed) and then the conversion and product distribution for each 28-second period was determined separately. APPARATUS A N D EXPERIMENTAL PROCEDURES
The equipment used in this investigation is shown schematically in Figure 1 and consists of: a Fisher introduction valve; a temperature-controlled reactor containing 5 grams of silica alumina catalyst; an indicating sampling system including a two-position, six-ported Aerograph valve, a sampling loop, and a Cow-Mac thermal conductivity cell; an F & M Model 500 gas chromatograph; and a Bendix Model-14 time of flight mass spectrometer. 1 Present address, F & M Scientific Corp., Starr Rd. & Route 41, Avondale,
Pa.