Document npMx27V9er7L8a0aRLgNEOa58

The violet-red color of the dibenzantlironc may be stabilized by interaction with the H*SO. It may be hypothe sized that HPO,- and IIjSO+ con tribute t electrons to the dibenzanthronc. Crotonaldchyde reacts by a similar mechanism (2, 9). By condensation of crotonaldchyde with anthrone, li methyl-bcnzan throne is obtained (9), and dimerization may be assumed to follow. To test the proposed mechanism, benzanthrone (Caibiochcm, Los Angeles 63, Calif.) was dissolved in concentrated H, 2CH3CH2CHO + 2 0 sulfuric acid and further diluted to give suitable concentrations in a sulfuric acid-water mixture identical with that used in the anthrone test. The molar absorptivity of this solution at the X__ of 510 ma was 7.8 X 103, very close to the values found for the anthrone reaction products with malonaldehyde, acrolein and crotonaldchyde (Tabic I)- Evi dently the condensation of these com pounds with anthrone to give benzanthrone (or its methyl derivative) is stoichiometric and goes to completion under the test conditions. The deviation in the spectral data for the propionnldchydc-anthrone complex from that of other saturated aldehydes may be explained on the assumption that propionaldohydc forms a cyclic compound similar to that of malonaldchyde, except that the 12 and 13 positions in benzanthrone are saturated. colored dimer (^mu 466 mi) Other saturated aldehydes probably condense with anthrone without cyclization. The structures of the resulting pigments are not known, nor is there evidence on which to base an explana tion of the differences noted in the ac tivity of formaldehyde as compared to other saturated aldehydes. c LITERATURE CITED (1) Allen, C. F. H., Ovnrbaugh, S. C., J. Am.'Chem. Soc. 57, 1322 (1935). (2) Dally, 0., Scholl, II., Her. 44, 1G5C (1911). (3) Carrol, N. V., Longloy, It. \V., Hoe, J. H., J. Biol. Chem. 220, 583 (1956). (41 Huttel, R,, fler. 74. 1825 (1941). (6) Ivamlet, M, J. ed., "Organic Elec tronic Spectra I," Interscience, New York, 1900. (6) Mashio, F., Kimura, Y., Nippon Kagaku Zasshi 81, 434 (1961). (7) Meerwein, H., J. Frakt. Chem. 97, 235(1918). (8) Morris, D. L., Science 107, 254 (1948). (9) Rodd, E. H., ed., "Chemistry of Carbon Compound III6," Elsevier, Amsterdam, 1956. (10) Roe, J. H., J. Biol. Chem. 208, 889 (1954). (11) S&wicki, E., Stanley, T. W., John son, H., Robert A. Taft Sanitary Engineering Center, USPH, Cincinnati 26, Ohio, private communication, Au gust, 1962. (12) Schmidt, H., Felle, Seifen, Anslriehmittel 61,881 (1959). (13) Sinnhuber, R. O., Yu, T. C., Yu, Te Chang, Food Res. 23, 626 (1958). Received for review September 17, 1962. Accepted March 11, 1963. Work supported in part by a grant, EF-179 (C2), from the National Institutes of Health, U. S. Public Health Service. Determination of Toxic Organic Compounds in Admixture in the Atmosphere by Gas Chromatography F. R. CROPPER and SIMON KAMINSKY Dyestuffs Division, Imperial Chemical Industries Ltd., Hexagon House, Blackley, Manchester, England A variety of chemical and physical methods exists for detecting toxic organic compounds when present alone in the atmosphere, but these often prove inadequate when two or more contaminants are present in admixture* Gas chromatographic methods de scribed so far--e.g. for air pollutants in "smog"--have, in general, required the use of special cooled traps and have involved lengthy sampling periods. In this paper, o method is described which traps the components on a solid absorbent at room temperature and which requires only a relatively small sample of air; the total time for a test, including the sampling stage, is thus cr.ly 10 to 15 minutes. It has been applied, for example, to atmospheres ecr a chlorination plant where toluene, enzyl chloride, benzal chloride, benzotrichloride, benzoyl chloride, and benzaldehyde might be encountered. Because the procedure is so simple and rapid, it has also been used for detecting single components such as chlorinated hydrocarbons. These meth nitrobenzene. It is deduced that the ods usually involve bubbling the at method would be applicable to the mosphere through a suitable absorption majority of organic compounds in the solution and development of an ap list issued by the American Conference propriate color. They are often specific of Governmental Industrial Hygienists. for a certain compound (or at least for a class of compounds) but each test in volves quite a lot of work and requires a To ensure the safety of industrial fairly large sample of air, with a cor personnel who might be exposed to respondingly lengthy sampling period. toxic compounds in the atmosphere,More recently dry absorber tubes have threshold limit values have been recom been developed--e.g., by Dragerwerk, mended by the American Conference of Moislinger Allee 53/55, Lubcck, W. Governmental Industrial Hygienists Germany--which indicate how much of {17) and these values have been adopted an expected contaminant is present ac as maximum permissible concentrations cording to the length of stain seen in the in Great Britain by the Ministry of tube after drawing air through for the Labour (18). A considerable number of required time, but these tubes have so chemical methods has been developed far been applied mainly to inorganic for various likely contaminants--e.g., gases such as phosgene and carbon (10)--and these have been applied for monoxide and to a few common volatile many years; examples are the deter organic compounds. mination of nitrobenzene, of aniline and The detection of a single component other arylamines, of tolylene diiso in the atmosphere will, however, be in- cyanate, of formaldehyde, and of --(equate whenever there is a possibility SL 043&4A re o i l cm Figure 1. Flame ionization detector tli.it several toxic compounds may be present; in .such circumstances it is desirable that a single test should in dicate which of the likely contaminants is pn'M'ut. and give their approximate concentrations. Gas chromatography offers the best means of achieving this, but so far no publication has appeared describing a .simple and rapid procedure suitable for industrial application. This paper describes such a procedure. (las chromatography lias been ap plied to studios on the nature of con taminants in the atmosphere arising as a result of smog and the running of automobile engines. This work (6, 7, lf>) has generally involved the sampling of large volumes of air, the removal of atmospheric water with drying agents (which may interfere by retaining some of the components), the isolation of the components in cooled traps containing a solid ad'orbent--e.g., activated char coal--or a conventional GLC packing, and the use of a thermal conductivity cell as detector. The use of the flame ionization detector, with its very high .sensitivity to organic compounds and its insensitivity to air and water, enables samples of the contaminated atmosphere to lie taken in a gas sampling bulb and injected directly by means of a hypo dermic syringe (5), but we have found that this procedure may result in losses,, particularly with- higher-boiling cortipounds. We prefer, therefore, to sample the atmosphere by drawing a small volume through a short, straight ab sorption tube at room temperature packed with a suitable GLC filling or -ilica gel; this tube is then attached to a GLC column and the organic compound(s) are desorbed by electrical heating (.?) and detected in the usual way. Special apparatus such as gas bulbs, drying tubes, and cooling traps are thus replaced by absorption tubes which are easily portable and are im mediately ready, after desorption, for using again without repacking. This procedure, which could readily be adapted in an automatic apparatus, is specifically designed as a rapid and con venient means of checking that at mospheric conditions are safe. APPARATUS AND MATERIALS Chromatographic Unit. Any gas chromatograph with a flame ioniza tion detector could be adapted for use in the way described below. Our apparatus was adapted from that developed for general analytical work by A. Heywood (to whom grateful acknowledgment is made), based on the original detector described in British Patent 838189 (S-5), The essential parts of this apparatus are: Chromatographic Column. The folded 3-foot X 4-mm. bore glass chro matographic column has four 9-inch limbs, one of them (the inlet) protruding 0.5 inch longer than the other end (the exit); there is no special assembly on the inlet end of the column. It is inserted into a 12-inch X 2-inch i. d. electrical heater. Flame Ionization Detector (Figure 1). This consists of a glass bowl with a 1.5 inch length of 20 gauge stainless steel tubing (ground square at each end) sealed into the 2-mm. bore Figure 3. Absorption tube and desorption heater in posi tion on top of column SL 043645 capillary tubing fused at the bottom of ;!ii bowl. Thin copper wire is soldered I-' this stainless steel jet, which forms one electrode; the second electrode is a piece of 20-gauge platinum wire sealed at o to 7 mm, above the jet. The. glass capillary tulie is connected to the exit from the chromatograph column by silicone rubber tubing. Cathode Follower Circuit and Recorder. This is the single valve circuit of Figure 2, used with a recording galvanometer (Record Electrical Co. Ltd., Altrincham,,Cheshire, England) with internal resistance of 1500 ohms, 0.3-ma. full scale deflection, chart speeds of 3 inches and 12 inches per hour. The sensitivity switch gives full sensitivity when set at 1000 megohms and, respectively, Vio and 'Aoo sensitivities when set at 100 megohms and 10 megohms. If a potentiometer rerorder is used instead, a suitable attenuating circuit is added. The accessories required for use with the chromatographic unit are as follows: Absorption Tubes. These are glass tubes 2.5 inches long, 4.5-nun. bore, with a bulged constriction in the exit end (Figure 3). Each contains a 1-inch length of suitable packing, held in position by glass or quartz wool plugs, and packed so that the impedance is 4 to 5 cm. Ilg at 100- ml.-per-minute air flow. With the 100- to 200-mesh silica gel packing very little or no compression will be necessary, but with the 100- to 120-mesh Celitc stationary phase pack ings the final 1-inch length at the above impedance is obtained by filling a 1.5inch length of the tube by tapping, and then compressing by ramming with a glass rod, withdrawing or adding more packing as necessary. They are stored in glass-stoppered test tubes and can be used repeatedly without repacking. Desorption Heater. This is a cop per foil tube 1.25 inches long and "-mm. bore, insulated with asbestos tape and wound with resistance wire to give 12 watts; this fits snugly o'er the absorption tube so that it rests on the bulged portion (Figure 3). When in use it is connected in series with a variable resistance--e.g., 0 to 10 ohms. Sampling Apparatus. Most of the conventional units can be used for drawing the atmosphere through an absorption tube--e.g., a vacuum line and flowmeter, a D.S.I.R. hand pump (/<?), a nibber-bulb hand pump (9) or a hypodermic syringe of approxi mately 5- to 100-ml, capacity. A gas sampling bulb, with one or two taps and w ith volume equal to the volume of sample desired, can also be used by , evacuating it, connecting it to an absorption tube and opening the tap between the absorption tube and the bulb. The sampling time under these conditions is 1 to l'/i minutes. If an integrated sample over a longer time is required, a capillary tube of suitable dimensions is inserted between the absorption tube and the bulb; thus, a 7.5-inch length of 0.015-ineh bore capillary tubing increases the sampling time to 3 to 4 minutes. The rate and duration of sampling depend on the particular test in hand; ( olume of 100 to 500 ml., sampled auring 1 to 2 minutes, is usually suitable. Column Packings. Celitc 545 (Johns-Mnnville Corp.), graded to 30to 60-mcsh B.S.S., heated at 100 C. with concentrated hydrochloric acid for several hours, washed, dried at 300 C., and again graded to 30- to 60-mesh B.S.S., is used with these sta tionary phases, in the stated propor tions: 7:3 Silicone Elastomer E.301 (ex I.C.I. Ltd., Nobel Division, Stevenston, Ayr shire, Scotland). 9:1 Poly(ethylene glycol adipate), m.p. about 50 C., mol. wt. about 1200. 9:1 Octadecylaminc-ethylene oxide con densate (mol. wt. 1000). 9:1 Polyethylene glycol 400, 4:1 Adiponitrile. After packing, the columns are pre treated to remove volatile components by heating for several hours in a stream of nitrogen (50 to 100 ml. per minute) at the appropriate temperature--i.e., 320, 180, 180, and 110 C., respec tively, for the first 4 phases listed above; adiponitrile is used only at room tem perature and requires no pretreatment. Absorption Tube Packings. Celite 545 (Johns-Manville Corp.), graded to 100- to 120-mcsh B.S.S., heated at 100 C. with concentrated hydro chloric acid for several hours, washed, dried at 300 O, and again graded to 100- to 120-mesh B.S.S., is used with the following stationary phases in the proportions 7:3 Silicone Elastomer E.301. Polyethylene glycol 400. These packings are pretreated in bulk--i.e., before filling the absorption tube--by heating for several hours in a stream of nitrogen (50 to 100 ml. per minute) at, respectively, 320 and 110 C. to remove volatile components. Silica gel, Davison (U. S. A.) grade 923 (100- to 200-mesh). This is pre treated--i.e., freed from adsorbed im purities--after filling the absorption tubes by carrying out repeated blank tests (see below--usually 2 to 3 suffice) until the impurity peaks seen after each 2-minute desorption heating period finally disappear; a convenient GLC column for following the pretreatment is that containing Silicone Elastomer E. 301 at 120 C. PROCEDURE Assembly and Testing of the Appa ratus. The outlet end of an absorp tion tube, containing the required packing, is connected to the inlet of the chromatographic column (at the required temperature) by silicone rubber tubing. The desorption heater is placed in position so that it rests on the bulging constricted portion of the absorption tube, and the inlet end of the absorption tube is connected to the carrier gas supply (1:1 Hi/Ni at 50 ml. per minute) with silicone rubber tubing (Figure 3). A gap of about 1 cm. is maintained between the silicone rubber tubing at each end of the absorp tion tube and the desorption heater. Wif he apparatus at full sensitivity, the cV. ,-ption heater is switched on for 2 minutes at 12 watts--i.e., with the variable scries resistance at 0 ohms; the baseline is observed for 15 minutes and if straight (or containing only very minor peaks), the apparatus is con sidered ready for use. If the baseline is not straight and/or peaks are present, and if this condition persists after several desorption cycles, it probably indicates that the desorption heating is causing some volatilization from the absorption tube packing; this is elimi nated by keeping the heating period constant at 2 minutes and reducing the wattage by adjusting the variable series resistance until, the .resulting baseline is satisfactory (this step will be more necessary with the polyethylene glycol packing than with the silicone, because of the greater thermal in stability of the former). The reduced wattage is noted and used subsequently, after ensuring that it is still sufficient to desorb the compounds being deter mined. If reduction in the wattage does not give a satisfactory baseline, the cause is probably due to (a) the trapping of volatiles from the carrier gas line by the absorption tube, with subsequent release on desorption. This may be significant if there are appreciable lengths of rubber and/or plastic tubing in the line, and may be corrected by minimizing the length of such tubing in the line, making it as far as possible in glass or metal tubing, with butt joints. (b) thermal instability of the silicone rubber joints. This should be unlikely if approximately 1-cm. gaps are main tained between the heater and the rubber at each end of the absorption tube. If necessary, a fresh piece of silicone rubber should be used which has been preheated for several hours at a temperature which does not cause it to go brittle (approximately 250 C.), and/ or the gaps between the heater and the rubber should be increased. Calibration Test. The carrier gas supply line is detached and the stipulated volume (usually 0.02 ml.) of a standard solution (usually in CSi) of the expected toxic com pound^) is introduced on to the glass or quartz wool plug at the inlet of the absorption tube, using a micropipet or a hypodermic syringe; the carrier gas supply line is reconnected, the flow restarted, the detector flame lit, and the desorption heater switched on for 2 minutes. The slight signal due to the elution of the CSj solvent is ignored. During the next 15 minutes, the ex pected compound (s) are eluted in peaks approximately Vio to 3/< of full scale deflection; no other peaks should be apparent. The peak areas and retention times are noted. Collection of Sample. The req uisite volume of the atmosphere to be tested is drawn through an absorp tion tube in the di/ection of flow from inlet end to outlet end (see Sampling Apparatus). When an SL 043646 VOL 35, NO. 6, MAY 1963 737; it vinyi- j,f i './ t" t LiTiC'rr u*Vi it *(,/ r vVh'-n ti.t Tj.H.I.Tl. pump is used, at least 0.5 minute must be allowed for each stroke, with 10 seconds between each stroke, and the piston must be held during this 10-second interval so that the partial vacuum in the barrel does not suck it forward and result in a low sample volume. After sampling, each absorption tube is placed in its stoppered test tube, and is usually tested as soon as possible afterwards. Testing the Sample and Assessing the Results. The absorption tube is connected to the GLC apparatus at the required temperature as described above, the desorption heater is ^witched on for 2 minutes, and the carrier pas flow is started after the first ininiiU-. Tint record i f.fic/i ol>eery'd to see whether any peaks arc pn^-ut, at, the expected retention times. Pi nee the volume of calibration solu tion and the volume of air sample are fixed (and specified) in each particular case so that- each peak corresponds to a threshold limit value, it is usually sufficient to rely solely on visual in spection of the records for the test and the calibration to decide whether the atmosphere contained less or more than the limit for each component. When numerical results are required, the areas under the peaks in the calibration records are measured and the mean for each peak (dc) is calculated; the areas kinder the peaks in the test record "(A t) are aho measured. If for a given component /Jp ,, ra, (v./v.) is the thresh old limit value represented by the ralibration peak, then CoaeeniM'ion in sample under test = ^4- p.p.m If rim r- mi is required in milligrams per .me motor at 25 C., the result in I .-.i ts per million is multiplied by M/24, v. hero .\f is the molecular weight of the compound. THEORY The maximum permissible sample volume for quantitative retention of a compound by an absorption tube con taining a GLC [tacking could be related' to the breakthrough volume curve, de termined by carrying out a frontal analysis experiment in which a fixed flow of an atmosphere containing the compound is passed through the tube connected to a GLC detector. How ever, it was considered simpler to relate the maximum permissible sample vol ume to the readily determinable reten tion volume of the compound on the ubc in the following manner. Consider a sample volume Vg equal to the retention volume Vr, and let both equal 100 arbitrary units. It is clear that the compound will not be retained quantitatively on the tube, < ------------------ --- ------------------- Table ! Sertenfiar Vzirjmt .wnp I-lricii Griih*/SHicc"* ^ssarprfcr Tab* of Room Temperature Benzene Cyclohexane Toluene Pyridine n-Butanol p-Xylene Cyclohexanone Cyclohexanol Benzaldehyde Benzyl chloride Aniline Benzoyl chloride Benzal chloride Nitrobenzene^ Bcnzotrichloride B.P., C. 80 81 no 115 118 138 155 162 179 179 184 197 207 210 214 Vr (retention volume, ml.) 20 26 77 no 36 210 230 230 330 590 580 1100 2100 1200 5800 since the peak maximum corresponding to the first unit will have reached the end of the tube, and thus the compound in this first unit will be only 50% re tained on the tube; this percentage will, however, increase for the successive units until a unit is reached, correspond ing to the breakthrough volume, which is just retained 100% on the tube; all succeeding units until the 100th will be also 100% retained. The distribu tion of each unit will approximate to a Gaussian type towards the outlet end of the tube (which is the portion of in terest) of standard deviation a = Vr/-\/n, where n is the number of theo retical plates in the tube. Consider the distribution of the t`h unit of sample volume; the extent to which this is not retained on the tube is given by that fraction of the area under the curve of the probability integral outside the bounds of the tube--i.e., 0.5 -- l/V^x j* exp(--(V2)df, and the percentage of the total sample not retained on the tube is therefore t - 100 _ [0.5 - l/y/2r J exp( --0/2)dl], where t -- i/a Since a typical value of n is 32, a is 17.5, and the above expression is cal culated with the aid of a table of the probability integral to be 7%--i.e., if the sample volume is equal to the retention volume, 93% will be retained by the tube. Most of the loss is from the initial portion of the sample volume; thus if the sample volumes are reduced to 90% and 80% of the retention vol ume, then the losses are reduced to 3% and 1%, respectively. Similar figures were obtained by graphical integration of the breakthrough curve derived from Glueckauf's treatment (11). It seems appropriate' therefore to stipulate a rk-fS;? * ? vn:ume " -9.fr,; tii- vC fs sure lAtl ibf erfieiuney is RESULTS Preliminary Experiments. A 1-inch length of Celite/silicone packing in the absorption tube was examined to see if it behaved as predicted theo retically in the extent to which it retained compounds at room tem perature. Silicone Elastomer E.301 was chosen because it was considered least likely to give rise to volatile impurities which would affect the base line. Values of V* (and thus of FTM") can be calculated from the specific retention volume, Vby means of the simplified expression: Vs - V,TWl/27Z (1) where T is the temperature of the tube(say 298 K.) and Wl is the weight of sta tionary phase in the tube (0.060 gram); the pressure gradient correction factor, j, has been taken as unity and the gas hold-up, VM, is negligible (15). Only few values of V, are given in the litera ture, especially at room temperature, but substitution of extrapolated V, values for benzene, toluene, and pxylene (IS) in the above equation gave VB values of 37, 97, and 220 ml. These were compared with values determined experimentally by connecting the ab sorption tube at room temperature directly to a flame ionization detector, loading a dilute CSj solution of benzene, toluene, and p-xylene, and passing car rier gas (1:1 Ht/Ni) at 50 ml. per minute; the retention volumes were 29, 77, and 210 ml., in satisfactory agreement with the above values, especially in view of the extrapolation and the fact that the V, values given are for Silicone 702. Hence, while a 5-ml. sample of air drawn through the tube should give good re covery for all three compounds (since V* is > 6 ml. in all instances), a 50-ml. sample would be expected to give a good recovery for toluene and xylene (Vr > 62 ml.) but a low recovery for benzene; this was found to be so. The retention volumes of a number of compounds were determined (Table I). Substituting V"** = 0.8 Vr, these re sults correspond broadly to the equa tion: -FT- + (2) The actual volume of air required to determine a given compound depends on the detector sensitivity and the maxi mum permissible concentration; if this volume is less than VTM" the silicone absorption tube can be used. Our sim ple apparatus (with sensitivity approxi mately 3 X 10' mv. ml. per mg.) gave full scale deflection signal for about 2 Mg- of a hydrocarbon compound with TOO * AWAlVTtrAI rHFMISTRY SL 043647 c a retention time of approximately 6 been declared for the other compounds, minutes; if the maximum permissible but it is likely that not more than a few concentration was 4 mg. per cubic p.p.m. would be tolerated for the other meter--i.e., 1 p.p.m. for a compound of chloro-compounds, and 1 p.p.m. has mol, wt. 100--men 500 ml. of air was been taken as a convenient figure for sufficient and the contaminants in this each chloro-compound. volume were in general trapped quan For the 1-p.p.m. level a 500-ral. titatively if their boiling points were sample of air was required, and it will over 170 C. When the maximum be seen from Table I that the retention permissible concentration was greater volumes for a silicone absorption tube than 1 p.p.m. the sample volume was were such that this volume should give correspondingly reduced and compounds quantitative recovery of the chloro- boiling below 170 C. were still retained. compounds, with benzyl chloride a Thus, the successful use of a silicone marginal case. This was found to be absorption tube packing depended on so; the compounds were easily sepa the particular determination envisaged. rated on the Silicone K.301 column and When a compound of relatively low the peaks appeared in the order: boiling point was being determined, toluene, bcnzaldehydo, benzyl chloride, especially at a low level, the volume of benzoyl chloride, benzal chloride, and sample required to give an adequate benzotrichloride. The toluene, if signal was sometimes greater than the present above about 1 p.p.m., gave an maximum permissible sample volume off-scale peak, hence, if its concentra for the silicone packing; in these in tion was reqtyred a separate test for stances, a number of alternatives were toluene (as above) was performed. possible, and these are treated fully in At the 1-p.p.m. level, the chloro-com the Discussion. pounds gave peaks about l/ to */i of Applications. The method de full scale deflection. Benzaldehyde was scribed above has been applied so separated from the other components far to 11 instances (19 substances), and was detected even at the 1-p.p.m. using the conditions given in Table level; but as expected the recoveries II. In general, silicone was satis were low, as the retention volume was factory both in the absorption tube only 330 ml. (see section on Recoveries). and in the column; in some instances Nitrobenzene. the packing in the absorption tube Aniline. and/or in the column was changed, for Cyclohexane. the reasons given below. These three substances are grouped Benzene, Toluene, and Xylene. together because each could be tested Since the threshold limit values are for using the silicone absorption tube relatively high, only 5 ml. of air was and the silicone column. Although the necessary, and from this volume these tests have been developed for detecting compounds were retained quantitatively only the one component in the atmos by the silicone absorption tube. The phere, the retention volume for aniline components were eluted in the order is about half that for nitrobenzene, so benzene, toluene, m- plus p-xylene, and that these two could be determined o-xylene. The result for o-xylene could easily in admixture. Nitrobenzene is, be obtained by calculation using the areas under the o-xylene peak in the test and under the p-xylene peak in the of course, a particularly toxic compound; 0.1 p.p.m. was easily discernible by this test. calibration. This result was reported Pyridine Bases. This is a crude separately when required, but normally mixture of lutidines, collidines, etc. No it would be added to the m- plus p- special attempt has been made to re xylenc result and reported as total xylene concentration. Tf the desorption stage cannot be car solve the individual components (in fact a single peak representing "pyridine bases" would have been preferred) but ried out within 2 hours of sampling the silicone column gave an irregular and/or the 5-ml. sample is too smal], a larger sample is ta::en on a silica gel absorption tube (see below). shaped peak showing that partial separation occurred. Cyclohexanone and Cyclo- Toluene Chlorination Products. ' hexanol. Since these two compounds The side-chain chlorination of toluene may well be present in admixture, a good produces benzyl chloride, benzal chlo separation on the column was desirable ride, and benzotrichloride, and from these are obtained benzoyl chloride and and for this reason polyfethylene glycol adipate) has been used as the stationary iwnz.'tldohyde. In an area where all of phase. these compounds arc produced, one or more arc liable to be present in the at mosphere, together with toluene, and it o-Dichlorobenzene and p-Dichlorobenzene. The separation of o- and p-dichlorobenzene is not unduly is clearly important to distinguish them difficult, but to make the peaks as from toluene and from one another. ; distinct as possible octadecylamine- The threshold limit values for benzyl ethylene oxide condensate was preferred chloride and toluene are 1 p.p.m. and as stationary phase, since this gave a 200 p.p.m., respectively; no limits have better separation factor than the other commonly used stationary phases. Pyridine. ji-Butanol. In both instances the silicone absorp tion tube was not quite sufficiently re tentive so the more retentive poly ethylene glycol 400 packing was used instead; this was quite satisfactory pro vided the desorption heating wattage was reduced to approximately 6 watts by suitable adjustment of the variable series resistance. Hydrogen Cyanide. It was desired to measure HCN in air which could be contaminated with appreciable con centrations of methanol, toluene, and HC1. Since the sensitivity of the flame ionization detector to HCN was only about l/io that of most organic com pounds, it was necessary to take a larger sample than hitherto, and it was also particularly important to separate it completely from the organic compounds likely to be present. The silicone ab sorption tube had a negligible retention volume for HCN at room temperature and therefore silica gel was used. This quantitatively retained HCN from a 1000-ml. sample, and released it com pletely on heating. To separate HCN from methanol and toluene and still obtain a peak for HCN at a reasonable retention time, adiponitriie was used as the stationary phase in the column, from which the order of elution was: HCN, methanol, toluene. The detector was, of course, insensitive to the HC1. Reproducibility. The reproducibil ity of the calibration procedure was assessed by carrying out a series of tests, and measuring the areas (by triangulation) under the appropriate peaks. Most of the results were within 15% of each mean, which is sufficient for this type of work; there was generally little change from day to day, but it was considered advisable to run a calibration test on the same day as the tests on the atmosphere. Results Using Standard Toxic Atmospheres. The reliability of the method was checked by carrying out trials using air containing known concentrations of toxic compound(s). These standard atmospheres were made by the controlled fluid-feed atom ization procedure of Gage (10), in which a dilute solution of the compound(s) in a volatile solvent in a syringe was ex pelled at a fixed rate into a metered stream of air by coupling the piston to a Slow Injection Apparatus (C. F. Palmer Ltd., 63A Effra Road, London, S.W.2). Care was taken to ensure that there was no significant leakage of solu tion past the piston and that adsorption equilibrium had been reached in the ap paratus by running it for a short while before sampling. The air supply was checked for freedom from interfering contaminants by carrying out a blank test prior to injection of solution. 043648 VOL 35. MO. 4. UAY 19A3 739 I I -i. li As an example of this procedure, a standard atmosphere containing 100 p.pjn. of toluene and 1 p.p.m. of benz- - .aldehyde, benzyl chloride, benzoyl chloride, benzal chloride, and benzotriddoridc was prepared by injecting a stock solution of these compounds in toluene at the appropriate concentra tion (respectively 0.94, 1.11, 1.25, 1.44, and 1.74% w./v.) at 2.5 X 10 ml. per minute into an n j* stream at 5.5 liters per minute. Tests' on these standard at mospheres were carried out by drawing -- the stipulated volume of the air through the appropriate absorption tube, which was then treated in the usual way im mediately after the sampling stage. The results were between 80 and 100% of the expected values, with the following ex ceptions: Benzaldehyde. The recovery was only 60 to 65%. From the retention ; volume (Table I) it was expected that some bcuznldehyde would emerge from the silicone absorption tube before sampling was complete, and this was shown to be so by carrying out further tests with two absorption tubes in series, when the first tube gave 60% recovery and the second tube about 20%. It was also probable that some of the benzaldehyde was oxidized in the air stream be tween the fluid-feed atomizer jet and the absorption tube. However, in this particular instance (Table II) it was the ^determination of the chloro-compounds which was regarded as particularly im portant, and since the results for these Were not affected by the presence of benzaldehyde, the method as a whole was regarded as satisfactory. Benzoyl Chloride. The results for benzoyl chloride were also only 60% of the expected values. This could not be due to incomplete retention, since its retention volume was greater than that for benzyl chloride, which gave a satis factory recovery. The slow formation of a white deposit (presumably benzoic acid) near the jet in the atomizer ma chine suggested the strong possibility that benzoyl chloride was being lost in significant amount before reaching the absorption tube, due to hydrolysis by the moisture in the air stream. With the other three chloro-com pounds, and indeed in all the other in stances given in Table II, the recoveries were 80 to 100%, using various meth ods of sampling, and this is regarded as quite satisfactory for this type of work. Storage of Absorption Tubes be tween Sampling and Testing Stages. Although it is preferable to arrange that the testing of an absorption tube should be carried out as soon as reasonably possible--e.g., within 2 hours--after sampling the atmos phere, it may so happen that testing must be delayed for 24 hours--e.g., if the sampling is necessary at a site some distance from GLC facilities. It was therefore considered necessary to ascertain whether the results on absorp tion tubes stored after sampling in glassstoppered test tubes were satisfactory; tests shewed that, with a few exceptions, the tubes could be stored for 24 hours without serious loss of the absorbed compounds. The exceptions were: Benzaldehyde and Benzoyl Chloride. Storage reduced the re covery to 45% for benzaldehyde and 5% for benzoyl chloride, presumably due to oxidation and hydrolysis, respectively. For a reliable result for benzoyl chloride it is therefore essential to test the absorption tube within 2 hours of sampling. Benzene. Gradual loss of benzene was observed from the silicone , tube on storage. Although this was not serious, it was considered better to use a silica gel tube, which retained all the. benzene (and toluene and xylene) even if 1000 ml. of air was drawn through the tube at room temperature, and which released all of these components rapidly when the tube was heated. It was most con venient to take a 50-ml. sample over 1 minute, to use the apparatus at l/io sensitivity, and to calibrate with a solu tion 10 times as concentrated as that given in Table II. The reproducibility was still within 15% in the calibra tion tests, and recoveries on standard atmospheres were close to 100% even after storage of the absorption tube up , to 48 hours. Gas Sampling Bulbs. In the intro duction it is stated that the taking of samples of air in a gas sampling bulb followed by direct injection on a GLC column using a hypodermic Table II. Experimental Conditions Compounds under test Threshold limit values (p.p.m.) Vol. of air sample, ml. Time of sam pling, min. Absorption tube packing Col umn pack ing Temp. of col umn, C. Sensi tivity range used Standard solution, 0.02 ml. corresponds to threshold limit value in specified air sample volume (except for HCN) Uen/.eno Toluene Xylene 25 200 200 5 0.5 A (see remarks A on storage) 0.0023% v./v. benzene, 0.022% v./v. 65 Max. toluene and 0.025% v./v. p-xylene in CSj .'ioii.'aldeiiydn lim/.yl chloride licn/.uvl chloride Hcnr.al chloride Mcn/ot rn-ldoride -- 1 '| Vnone stated, ,1 taken as 1 ' i! tolicn7pne V Hi i 1 (It; 1 5 ri,,.\ ,t,p 1i : hue ; ciM'S 100 none staled, taken as 10 500 2 A 500 2 200 2 ' id 0.5 250 1 t A A A A 0.46% v./v. benzaldehyde, 0.52% v./v. benzyl chloride, 0.52% v./v. A 120 Max. benzoyl chloride, 0.58% v./v. benzal chloride and 0.64% v./v. benzotrichloride in dry toluene; diluted 44-fold with CSt A 120 Max. 0,010% v./v. nitrobenzene in CS A 105 Max. 0.019% v./v. aniline in CSj, freshly prepared A 45 */. 0.09% v./v. cyclohexane in CSj A 90 Max. 0.067% v./v. pyridine bases in CSj "t|C\ 'IK'IK1 : n- i ' ' `i, "henzenc >1 iuoi oheuzene r.o 50 50 75 Pyridine 5 ')-Hutanol 100 Hydrogen cyanide 10 125 1 125 1 125 125 1000 1 1 4 A A D D Silica gel B 80 Vs. 0.13% v./v. each of cyclohexanone and cyclohexanol in CSt C 95 `/10 0:19% w./v. o-dichlorobenzene and 0.28% w./v. p-dichlorobenzene in CSj D 75 Max. 0.01% v./v. pyridine in CSt D 60 V. 0.23% v./v. -butanol in CSt E 20 Max. 0.004 ml. of 0.28% w./v. HCN in methanol Code for packings: (details as under Apparatus and Materials) A = Silicone Elastomer E.301. B TM Poly(ethylene glycol adipate). C -- Octadecylamine-ethylene oxide condensate. D Polyethylene glycol 400. E -- Adiponitrile. 740 ANALYTICAL CHEMISTRY SL 043649 syringe results in losses, particularly C ( with higher-boiling compounds. To illustrate this, a standard atmosphere containing benzene, toluene, and p- .\ylcne (25, 100, and 100 p.p.m.) was stored in two oOG-ml. gas sampling lull's, with taps at each end; bulb A Contained a short central side-arm over which was fitted a rubber scrum cap and bulb B was fitted with a rubber serum cap over t'.p glass tubing at one end of the bulb so that the serum cap and the sample in the bulb were separated by __a tap. At intervals, 5-ml. samples were removed via the serum cap using a hypo dermic syringe (inserted also through the bore of the tap in bulb B) and in jected into the GLC apparatus, the difference between the two sets of tests was that with bulb A the sample was in contact with the serum cap during the entire storage period, whereas with bulb B the contact was only momentary during withdrawal of the samples for in jection. The recoveries were calculated by comparison with the results using a standard solution; the recoveries ob tained with bulb B were fairly satis factory up to 24-hour storage, but those obtained with bulb A became sub stantially loner--e.g., 20% for toluene and for xylene--after storage for 24 hours, presumably because of the pro- ionged contact with the serum cap. it is clearly inadvisable to use bulbs of jype A for collecting atmosphere sam- Jts prior to testing. FVlthoueh bulbs of type B may be .ti-iaetorv, adsorption losses may occur :tn iijulicr-boiling compounds when ansi'erring .samples by means of a vpodermic syringe. Thus, a 50-ml. - unple of air containing 20 p.p.m. nitrobenzene was injected into a GLC ilunui, the syringe was then filled with 'd ml. of uneontaminated air, and the 'intent' were injected after 5 minutes; :V syringe was again filled with 50 ml. 8. Pt "C. Figure 4. Relationship between b.p. and threshold limit values, for 1 -inch silicone absorption tube air and the contents were injected ter 5 minute-. The peak heights from ie second and third fillings were 27% id 10% of the original peak height, dio-itinn considerable retention of .iinbenzenc m the syringe after the -t mjeetion. Similar results were ob,ued ii-ing other compounds. To id ri-k ot ~uch hi--es, ue reject the -.imp'"m: bulb hypodermic syringe to predict whether a silicone absorption tube is likely to prove suitable in further applications. If the sensitivity of the detector and the operating conditions are such that full scale deflection will be given by ,, W fig, of compound and if the threshold * limit value is C mg. per cubic meter, it follows that full scale deflection will be given for a sample volume of W/C liters With our apparatus W may be taken as 2 (for the convenient column retention time of 5 to 10 minutes) and hence a silicone absorption tube can probably be used provided the boiling point is not less than 169 -- 70 log C. Figure 4 shows this relationship in the form of a demarcation line, the extrapolated por tion of which is shown as a broken line; if the point on this figure corresponding of air. Since l/% of full scale deflection to the boiling point and the threshold oCUSSlON may be taken as a convenient minimum limit value for a given compound lies ii .n\en m Table ! for a 1.. 'me i.o'orptiou tube show that, oh a fi-w exceptions, the maximum nni'-ible sample volume of a given igaine compound is related to the point of the compound according peak size for detecting C mg. per cubic meter, the corresponding sample volume should be ml., and the silicone tube is likely to prove suitable if this volume does not exceed V$"--i.e., to the right of the line, the silicone absorption tube can probably be used; it is necessary, of course, to establish experimentally in each case that the retention is complete using the ap propriate sample volume. If the point ^^Bapmtion 2 above; this is in acconiance with the known property of the 'iicone stationary phase of eluting com- provided Log 200TT C \\ // b.p. + 70 13 lies to the left of the line the silicone absorption tube is likely to prove un satisfactory, the degree to which this is !'"U!ii. generally in the order of their "iling points. It is therefore possible or b.p. 148 -f 70 log W - 70 log C so increasing with increasing distance of the point from the line. SL 043650 VOL 35, NO. 6, MAY 1963 741 ^i,nat the corresponding minimum boiling In the list of threshold limit values for 1901 (17)--including the tenta tive values--carbon monoxide, carbon dioxide, carbon disulfide, phosgene, and nickel carbonyl can be regarded ns in organic; there arc thus 104 organic gases and vapors in the list and of these only the following 21 compounds (Table III) lie to the left of the line (arranged in in creasing distance from the line). In addition to the 21 compounds, there are 7 others which (although they lie to the right of the line and thus are determinable in principle with the silicone absorption tube) have such low boiling points that the cor responding maximum permissible sample volume is inconveniently small to sample; the minimum convenient sample volume can be taken as 5 ml., and substitution in Equation 2 shows Table 111. Organic Compounds Lying to the Left of the Demarcation Line of Figure 4 Boiling Threshold point, limit, 8 C. mg./m. Butylaminc Dichiorodifluoro- methane" Perehlorotncthyl mercaptan Methyl mercaptan Ethylene oxide Ally1 alcohol Trilluoromono- bromomethane" Ethylamme Methyl chloride Ally1 chloride Methyl bromide Ethylene imine Tctranitromethane* Chloroacetaldehyde Isopropylamine Clilori ipicna Dimeth vlhydrazine Acrolein Fnrmaldohvde Hydrogen cyanide" Ketene 77 -30 159 8 11 96 -58 17 -24 45 4 56 126 85 34 110 62 52 -21 26 -50 15 4,950 0.8 100 90 5 6,100 45 210 15 80 9 8 3 12 0.7 1 1.2 6 11 0.9 * Appropriate allowance has been made for tetranitromethane, hydrogen cyanide, and the fluorinated hydrocarbons, which give a low response to the flame ionization detector (approx. */w of that of most organic compounds). Table IV. Compounds with Maximum Permissible Sample Volume less than ' 5 Ml. Boiling point. C. Fluoro t ri ch 1 oromethane 24 Ethyl chloride 13 Dicliloromonofluoro- methane Dich lorotet rafluoro- 9 ethane -2 Butadiene -3 Vinyl chloride -14 Methyl acetylene -23 Threshold limit mg./m.* 5,600 2,600 4,200 7,000 2,200 1,300 1,650 point is 30 C. The compounds in this category are shown in Table IV. Subtracting the above 28 substances from the 1G4 listed leaves 136 sub stances--i.c., 83%--which it is con cluded can be determined in principle with the silicone absorption tube. In practice it is envisaged that certain ex perimental difficulties may arise in a few instances--e.g., The substance may be oxidized or hydrolyzed in the silicone absorption tube during the interval between sampling and testing; benzoyl chloride has been shown above to behave in such a manner, and possible compounds falling in this category are acetaldehyde, acetic anhydride, dimethyl sulfate, and tolylene di-isocyanate. The substance, being a mixture, may give several partially separated peaks, thus making calibration difficult; coal tar naphtha, petroleum naphtha, gaso line, turpentine, and Stoddard solvent fall into this category, but since the mixture constituting pyridine bases has been determined successfully above, so by analogy should these other mixtures. If it is desirable in certain investiga tions to determine actual concentrations below the threshold limit--i.e., not merely to report that the atmosphere contains less than the threshold limit--it would be necessary to arrange conditions so that the compound at the threshold limit gives say a full-scale instead of a */i deflection signal. With those compounds which lie well to the right of the demarcation line this could be done simply by increasing the sample volume accordingly, since this will generally be much lower than the maxi mum permissible sample volume; for high-boiling compounds, the sampling time could be kept short by increasing the sampling rate to for example 1 to 2 liters per minute, by replacing the compressed 100- to 120-mesh Celite in the absorption tube by loosely-packed 30- to 60-mesh Celite. For those com pounds which lie just to the right of the demarcation line, however, it would be necessary to use one of the modifications suggested below for detecting the re maining 28 compounds. It has been shown above that the majority of the organic compounds in the list of threshold limit values (17) can be retained by the silicone absorption tube. The method could be extended to cover some of the 21 compounds of Table III by using a more sensitive ap paratus (with a smaller sample volume), but such apparatus would be more ex pensive, less portable and more sub ject to noise arising from thermal in stability of the packing. It is considered that it would be preferable to modify the conditions so that the maximum permissible sample volume is increased, and this can be done in three ways; ,'educing the temperature of the absorption lube below ambient by special cooling. This is regarded as the least attractive since it would involve re designing the shape of the absorption tube and desorption heater, and would result in considerable inconvenience when sampling at a distance from laboratory facilities. By increasing the amount of stationary phase. The amount of stationary phase on the Celite 545 (30% w./w.) is already near the maximum loading, but it would be possible to replace the Celite by a solid support of greater packing density; thus Silocel C22 Firebrick (L. Light & Co. Ltd., Colnbrook, Bucks.* England) and Chromosorb P (Johns-Manville Corp.) are both approximately 1.5 times as dense as Celite 545. However, it has been claimed that firebrick and some varieties of kieselguhr are unsatisfactory as solid supports, particularly in work on nonhydrocarbons, since they exert a catalytic effect (probably dehydration) on the decomposition of certain oxy genated compounds (14). Although various methods of treating the support have been reported, these have not been tried in this work. An alternative way of increasing the amount of stationary phase would be to use a larger absorption tube. However, by increasing the length of the tube and correspondingly increasing the sample , volume to a value near the maximum permissible, the compound would be distributed over a longer length of the packing, and desorption would neces sarily result in the introduction of the compound on to the GLC column over a more prolonged charging period, with a consequent increase in the peak width; this would ultimately affect the separa tion between two adjacent peaks--e.g., of o- and p-dichlorobcnzene. A similar effect would be obtained by increasing the diameter of the tube, since the packing will take longer to heat towards the center. If the central portion did not become hot enough, recovery would be incomplete, and if the desorption wattage were increased to overcome this, the packing next to the tube wall might become hot enough to cause thermal breakdown (with consequent reduction in sensitivity and increase in noise). By using a more retentive stationary phase or a solid adsorbent. Silicone gives significantly lower specific reten tion volumes than the other common stationary phases, as exemplified by the values for p-xylene at 100 C. on 6 stationary phases (H). The difference between silicone and the other stationary phases is, of course, more marked if both the solute and the other stationary phase are polar--e.g., the retention volume of pyridine on polyethylene glycol 400 has been shown to be five times that on silicone. However, these other stationary phases are less thermally stable---i.e,, they and their associated impurities are more easily volatilized and/or pyrolyzed during the desorption stage--and Silicone Elas tomer E.301 is therefore the first choice in most instances. Nevertheless, other stationary phases, particularly polar 742 ANALYTICAL CHEMISTRY SL 043651 ones, cau prove very useful m certain circumstances--for example, for nbutanol and pyridine--for which poly ethylene glycol 400 has been used (Table II). The polar stationary phase of greatest thermal stability so far ^examined is Ucoplcx 400 (ex Geigy Ltd.) ind this would now be regarded as first choice when seeking an alternative to silicone. Of the 2S compounds for which sili cone is probably unsuitable, most of those in Table IV and some of those in the upper part of Table III could probably be detected using a suitable polar stationary phase. The remainder would probably prove insufficiently retentive on any gas-liquid partition' system and for these a gas-solid adsorp tion system may be effective. This work contains one example in this category--i.e., IICX--which was ad sorbed on silica gel, and it is therefore relevant to discuss certain aspects of the behavior of such adsorbents. The two most useful adsorbents are silica gel anti activated carbon, which differ in several respects. Silica gel is polar and will consequently show pro nounced selectivity in preferentially adsorbing polar molecules, in particular water. lienee, when sampling large volumes, the atmospheric moisture may compete for the adsorption sites and dis place the toxic compound being sought, especially if this is nonpolar. Thus, it can be shown that the silica gel absorp^tan tube described above may become [^Bturated with water when the sample volume is only 3 liters, and the retentive properties of the silica gel may be im paired as sample volumes of this size are approached. Desorption from silica gel is relatively easier than for carbon, but will be more difficult for the more strongly held polar compounds; thus it has been shown that o-dichlorobenzene can be readily desorbed from silica gel with 2 minutes heating at 12 watts whereas `ompounds of similar boiling point but with chlorine in the sidechain instead of the nucleus--e.g., benzyl and benzal chloride--cannot be desorbed even with 5 minutes heating at 18 watts. Activated carbon is nonpolar and will consequently generally adsorb organic vapors in preference to water, so that * atmospheric moisture will not interfere with the sampling of large volumes; however, desorption is generally more difficult than from silica gel. Although the adsorptive capacities of both silica gel and activated carbon will be much greater in general than the amount of toxic compound to be adsorbed, in practice it will be Uieir retcntivitics--i.e., the maximum ^^unt of the compound retained when ^BJrosed to uncontaminated air (IS)-- tint will decide their suitability, since ; he concentration of toxic compound in the sample volume will always be V Retentivity values for a number of compounds on 6- to 14-mcsh carbon have been tabulated (/). Since the system is a dynamic one, the rate of adsorption will limit the sampling flow rate. From the information in (I), it can be shown that, with our absorption tube, the sampling rate should not exceed 70 ml. per minute and in many instances, par ticularly with compounds boiling below 08 C., it would have to be much lower than this, necessitating excessively long sampling times. In Borne instances the tube would have to be lengthened and/or the carbon cooled or impreg nated--e.g., with bromine, for retaining olefines. Similar considerations will no doubt apply to silica gel. Losses due to irreversible adsorption at the desorption temperature, or to catalytic reaction on the adsorbent surface--e.g., olefin polymerization on silica gel--may render an adsorbent un suitable; the catalytic effects may be reduced by using specially purified gel (Davison grade 923) although even with this material, evidence was obtained that n-butanol was catalytically de hydrated to butene. Activated carbon which has been specially processed to minimize catalytic activity is available (from Carbide and Carbon Chemicals Co,, New York). It is clear from the above considera tions that the use of solid adsorbents as tube packings cannot be treated in as general a manner as the gas-liquid partition packings, and that each toxic compound must be tested for retention volume, sampling time, storage stability, and completeness of desorption. Toxic Dusts, Fumes, and Mists. The method is in principle applicable to some of the organic dusts, fumes, and mists in the list of threshold limit values (17), since such particulate matter is expected to be trapped by the absorption tube, probably in the inlet plug. Desorption should not prove dif ficult, since even chrysene (b.p. 450 C.) can be 80 to 90% desorbed from a sili cone absorption tube under the stand ard conditions. However, the desorp tion heater should be made longer so that the inlet plug is properly heated. Automation of the Method. The method described above could be adapted into a unit which would monitor a plant atmosphere every 15 minutes, for example. It is en visaged that a sampling valve would be made with appropriate slots or grooves cut into a polytetrafluoroethylene sliding plate sandwiched be tween two stationary metal plates, one of which carries a thin metal loop, say 5-mm. bore and l1/* inches long. The other plate would carry ports to allow the carrier gas stream to pass into the column, and to allow the air stream to be drawn through the loop by a vacuum pump. The loop would be charged with v .tc/stationary phase or silica gel, and would be wound so that it could be electrically heated when de sired. A glass or metal column, with the appropriate packing, would be at tached to a flame ionization detector fitted with a flame-proof housing with nitrogen flowing in the annular space. Nitrogen would be used as carrier gas and the required flow of hydrogen would be fed in between the column exit and the detector jet, thereby ensuring that the flame would not be extinguished when the sample valve operates. A motor-driven timing device would actuate the sampling valve and also switch on and off the heater around the loop. The signal from the detector would preferably be fed via a relay into an audible alarm system, adjusted so as to operate only when the signal exceeds a specified value during a specified period. LITERATURE CITED (1) Am. Soc. Testing Materials 1961 Book of Standards, Part 10, pp. 1672, 1675. A.S.T.M. D.1605-60, Philadel phia, 1961. (2) Deaty, D. H., Warham, T. J., Whyman, B. H. F., "Vapour Phase Chroma tography," D. H. Desty, ed., p. 346, Butterworths, London 1957. (3) Dewar, B- A., McWilliam, I. G., British Patent 836,189, July 4, 1957. (4) Dewar, R. A., McWilliam, I. G., "Gas Chromatography 1958," D. H. Desty, ed., p. 142, ButterworthB, London 1958. (5) Dewar, R. A., McWilliam, I. G., Nature 181, 760 (1958). (6) Eggerteen, F. T., Nelsen, F. M., Anal. Chem. 30,1040 (1958). (7) Farrington, P. 8., Pecsok, R. L., Meeker, R. L., Olson, T. .T., Ibid., 31, 1512 (1959). (8) Feinland, R., Andreatch, A. J., Cotrupe, D. P., Ibid., 33,991 (1961). (9) Gage, J. C., Analyst 84, 509 (1959). (10) Gage, J. C., J. Sci. Instr. 30, 25 (1953). (11) Glueckauf, E,, Trans. Faraday Soc. 51,34(1955). (12) Littlewood, A. B., Phillips, C. S. G., Price, D. T., J. Chem. Soc. 1955,1480. (13) Mantell, C, L., "Adsorption," 2nd ed., pp. 157, 164. McGraw-Hill, New York, 1951. (14) "Materials for Gas Chromatog raphy," 2nd ed., May and Bnker Ltd., Dagenham, England, 1961. (15) Scott, R. P. W. (ed.), "Gas Chroma tography 1960," p. 423, Butterworths, London 1960. (16) Strafford, N-, Strouts, C. It. N., Stubbings, W. V., "The Determination of Toxic Substances in Air--A Manual of I. C. I. Practice." Heffer, Cam bridge, England, 1956. (17) "Threshold Limit Values for 1961," Am. Conf. Governmental Industrial Hygienists, Detroit, April 1961. Arch. Environ. Health 3, 489 (1961). (18) "Toxic Substances in Factory At mospheres," Ministry of Labour pam phlet, New Series No. 8. H. M. Sta tionery Office. London, September 1961. (19) West, P. W., Sen, B., Gibson, N. A., Anal. Chem. 30,1390 (1958). Received for review October 1, 1962. Accepted February 1, 1963. SL 043652 VOL 33, NO. 6, MAY 1963 743