tained liquid scintillators and yielded normal background counts. The inrrease in counts in these two samples over the previous ones was due to cosmic radiation and environmental radioactivity. Samples 11 and 12 were fabricated from optical plastics to differentiate the relative importance of the effects of potassium-40 and Cherenkov radiation in glass. However, from the large number of counts observed in these samples as compared with sample 3, i t can be seen that the plastic constitutes a copious source of Cherenkov radiation, possibly due to a greater optical transmission of light in the short wave-length region. This observation agrees with that of Arnold ( I ) . By assuming the normal background counts as unity, the per cent contribution from the above sources may be computed from the results shown in Table 111. Thus, for counting a t optimum settings for carbon-14. the contributions from chance coincidcnce, potassium40 (light sensitization of photocathodes), and Cherenkov radiation were approhimately 6, 31 (20), and 25% in the 10-to-50 volt channd and 3.2, 14.6 (9.21, and 12% in the 10-toinfinite volt channel, respectively. At optimum settings for tritium these values became 5, 74 (32), and 9% in the lower channel and 5, 52 (29), and 8% in the upper channel, respectively.
It may be noted that increasing the high voltage supply increases the sensitivity of the multiplier phototube, thereby resulting in a more efficient detection of potassium-40 in glass. The background counts of a liquid scintillation solution may now be considered t o be composed of a hard component and a soft component fraction. The hard component consists of the counts usually observed in solution media without an organic scintillator present and, for this reason, these counts are not fluorescence quenchable. I n contrast, the soft component is affected by quenching; however, the extent to which this ocrurs depends largely upon the concentration and the quality of the quencher. Only under extremely adverse quenching conditions can the soft coniponmt be completely eliminated.
(6) Hayes, F. S.,Ott, D. G., Kerr, V. X., hucleonics 14, No. 1, 42 (1956). (7) Hayes, F. N., Ott, D. G., Kerr, V. K., Rogers, B. S.,Ibid., 13, No 12, 38 (1955). ( 8 ) Kallmann, H., Furst, M., in “Liquid Scintillation Counting,” C. G. Bell, Jr., F. N. Hayes, eds.. p. 3. l’ergamon, New York, 1958. (9) Kallmann, H., Furst, 1\1 , Phys. Rev. 79.857 11950). (10) jbid.,‘81, 853 (1951). (11) Iierr, V. S . , Hayes, F. S . , Ott, D. G.. Intern. J . A.nw1. Radia/ion and
LITERATURE CITED
6 ) Swank, R. K., “Liquid Scintillation Counting,” C. C;, Bell, Jr., F. N. HayeE, edp., p. 93, Pergamon, ?;em- York, 1058. 7 ) Sn-ank, R . K . .l’ircleonics 12. S o . 3. 14 11954).
2 ) Lingham, iV. €I:, Eversole, W. J., Hayes, F. S . , Trujillo, T. T., J . Lab. Clin. X e d . 47, 819 (1956). 3 ) Packard, L. E., in “Liquid Scintillation Counting,’’ (7. G. Bell, Jr., F. N. Hayes,, eds., p. 50, Prrgamon, S e w York, 1058. 4) Paaeman, J. It.,Rntiin. S . S., Coouer. J. -4. I).. . ~ X . A L . C H m i . 28.
484 -( 1956).
5 ) Peng, C. T., in ,‘Liquid Scintillation Counting,” C. G , Bell, Jr., F. S . Hayes, e&., p. 108. Pergamon, Xcw York,
1958.
(1) Arnold, ,J. R., in “Liquid Scintillation Counting,” C. G. Bell, Jr., F. ?J. Hayes, eds., p. 129, I’ergamon, Xew York, 1958. (2) Arnold, J. It., Science 119, 155 (1954).
(3) Birks, J. B., “Scintillation Counters,” p. 62, McGraw-Hill, New York, 1953. (4) Furst, M., Kallmann, H., Brown, F. H., i4’z~cleonics 1 3 , No. 4, 58 (1955). (5) Guinn, V. l’., “Liquid Scintillation Counting,” C. G. Bell, Jr., F. N. Hayes, eds., p. 170, I’tlrgamon, S e w York, 19.58.
5) T’aughan, M.,Steinberg, D., Logan, J., Srzence 126,446 1957).
RECEIVED for review December 21, 1959. Accepted June 20, 1960. Work aided by grants from U.S. Public Health Service (C-3630) a n d .4merican Canwr Society (T-431,
Extraction and Flame Spectrophotometric Determination of Vanadium CORNELIUS M. STANDER African Metals Corp. Lid., P . 0. Box 66, Meyerton, Transvaal, Union of South Africa
b A flame spectrophotometric method that is sufficiently sensitive to determine vanadium in a variety of materials has been developed. The method consists of separation of vanadium from interfering elements by extraction with N-nitrosophenylhydroxylamine plus ethyl acetate, followed by aspiration of the organic phase directly into the flame and measurement of the emissivity of vanadium by using the band at 550 mp. With a slit width of 0.065 mm., the sensitivity is 1 p g . per ml. per scale division (% T ) . The working curve is linear between 0 and 100 pg. per ml. of vanadium. The interference of metals that are extracted with the vanadium has been examined and methods for their separation have been devised. 1296
ANALYTICAL CHEMISTRY
A
s PART of a study on the flame photometric determination of metals belonging to groups IV and V of the periodic table, a method for the determination of vanadium has been developed. Burriel-RIarti, Ramirez-MuBoz, and Asuncion-Omarrementeria (2) in their study of the emission characteristics of various metals in the flame reported that the response of vanadium in the flame was small and that i t was affected by large amounts of iron. The dissociation energy of vanadium oxide in the flame is given as 5.5 e.v. by Gaydon (5) and as about 6.4 e.v. by Mahanati ( 8 ) and Herzberg ( 7 ) . As a result of this high dissociation energy only a very small fraction of the vanadium oxide molecules present will be dissociated into vanadium atoms a t the
temperatilie of the oxyacetylene flame. Line spectra emitted by neutral or ionized vanadium atoms are therefore absent from the flame spectrum. The flame spectrum consists of a series of bands associated with a continuous emission starting at approximately 400 mp and extending into the infrared region. Bands that may be used for analysis occur in the region from 500 to 580 mp. I n the present investigation the bands occurring a t 529 and 550 mp were used (6). Many metals interfere in the flame photometric determination of vanadium, However, the interference of a large number of these metals can be obviated by extracting the vanadium with ethyl acetate plus S-nitrosophenylhydroxylamine (cupferron) from dl-
separatory fuiinc.1, shaking for 30 seconds to establish equilibrium, and then extracting two times with 20-ml. portions of ethyl acetate. T h e combined extracts were diluted to exactly 100 ml. with ethyl acetate. The background correction was obtained b y measuring the emissivity of an ethyl acetate solution prepared in exactly the same manner as the above solution but containing no vanadium. The emissivity of the solution containing no vanadium was subtracted from t h a t of the solution containing vanadium to obtain the net emissivity.
t
.
lut,e liydrochloric or sulfuric acid solution. Tanadium(1V) is nearly conipletely extracted from 1-to-9 hydrochloric acid by c,upferron plus ethyl acetatc (3, 4). The ethyl acctatc solution of vanadium is aspirated directly into the flame ailti interferences not removed during extraction can be circumvented by making prior separations. EXPERIMENTAL
Apparatus. T h e instrumcnt uscd was a Becknian Model D t T s p d r o photometer n i t h photomultiplier. Model 9200 flame attacshnient, and a standard Beckinan osyacrtylcnc~ burner. Special Reagent. T'anatlirim stantlart1 solution, 500 pg. PPI' nil. 1)issoIvc) 1.148 grams of pure ammonium metavanadate (analytical reagent grade) in 43.0 nil. of concentrated hydrochloric acid and dilutth to I 1itc.r with distilled watcr. Instrumental Settings. l'hc following instrumental srttings ivcre usetl : densitivitj- control delector switch dlit Spectral slit width at
Full!. riockn-iw 10 x 0.03 rnm.
550 mp Photomultiplier
1 45mp 1 64 111p 60 volts per
Grid resiitor
22 megohm3
520 m p
dynode
Fuel and Oxygen Pressures. Optimum acetylene and oxygen prwsure. weie determined by measuring th(n net emissivity of a n ethyl awtatc. 3olution containing 100 pg. pel nil. ok vanadium as vanadium cupfcriate. This solution was prepared by adding 0.30 gram of cupferron (E. AIeirk AG, Darmstadt) to 20 i d . of the standard vanadium solution in a conirentional
529rny I
5
1
6 7 8 9 1 OXYGEN PRESSURE, p s I
i I
0
Figure 2. Intensity of vanadium emission as function of oxygen pressure
The relation between net emissivity of the tn-o bands and acetylene pressure
for various fixed oxygen pi'essures is shown in Figure 1. The net emissivitjincreases with increase in acetylene pressure. Figure 2 shows a plot of net c,missivity us. oxygen pressure at various fixed acetylene pressurcs. As bht. net (missivity of the 529-1np band is practically independent of the oxygen pressure at fixed acetylene pressures, the> upper curve in Figure 1 is valid for oxygen pressui'es from 6 p.s.i. to 10 I1.s.i. For the 550-mp band the net cmissivity is independent of the oxygen pressure up to pressures of about 8 p.s.i. The middle curve in Figure 1 is therefore valid for oxygen pressures from 5 p.8.i. to about 8 p.s.i. At oxygen pressures above 8 p.s.i. thc emissivity of t,he 550-mp band falls with increase in oxygen pressure. Although the emissivity of this band is higher a t a n oxygen pressure of 9 p.s.i. than i t is a t a n oxygen pressure of 10 p s i , the relation between emissivity and acetylene pressure has been shon-n as a single curve in Figure 1. The lower curve in Figure 1 is thus not strictly valid but serves to show the ctc,pendence of emissivity on the acetylcnc pressure. Aiplot of o\.~geti-t,o-ac.et'yleiie pressure
ratio 2's. ciniasivitj. is shown in Figui,cx 3. At ratios highel' t,han 5 the net, emissivitj- is low and changes slowly with change in prtwure ratio, but a t ratios below 5 the emissivity incwases rapidly with decrease in ratio. At osygcn-toacetylene pressure ratios below 0.8, t'he background becomes incrcwingly unstable. An oxygen-to-acet,ylene pressure ratio of 0.8 could still be used without causing excessive backgi.uurid instability. However. the optimum ratio varies from burner to burnel, and .should he determined for each burnel'. Tlie emissivit,y of the 5.50-mp band is equal to that of t>he 529-mp band a t acetylene pressurc~sof about i . 5 p.s 2nd oxygen pressurrs below 8 p.s At acet,ylene pressures above and oxygen pressures below these values, the emissivity of the 550-mp band is larger than that of the 529-mp band. A t oxygen pressures above 8 p.s.i., the emissivity of the 550-mp band is appreciably l o m r t,Eian that of the 329-mp band for all pressures. Slit Width. T h r slit width was maintained bt.lo\v 0.1 m m . t o obviate t h e risk of spc~c~tralint~erferrnce. T h e minimum slit ividth used ]vas 0.03 inni., as x-r'ry little addcd resolution is gained a t smaller slit w-idths. The valucs givtn previously for t h e spectral slit widths should not be confused with the spCct,ral hand width. Calibration Curve. Transfer 2-, 4-, 6-. 8-, antl 10-nil. aliquots of t h e st,andard vanadium solut'ion t o 50ml. separatory funncT1s. Add 0.1 grain of cupfrrron antl shake for 30 seconds t o achieve equilibrium. .4dd 15 ml. of ethyl acetatr. shake for 60 seconds, and allow t8he phsscs t o separate. Draiv off t h e aqueous lager into another separator\- funnel and transfer t h e organic. lnyrr t o a 50-nd. volumetric flask. Atid 0.05 gram of cupferron to the aqueous solution and again exti,act, with a furt,her 15-ml. portion of et,hyl acetate. Allow the phases t o separate, discard the aqueous VOL. 32, NO. 10, SEPTEMBER 1960
1297
filtiate in a mercuiy cathode cell until layer, and transfer the organic layer to the total iron content is less than 0.1 t'hr volumetric flask containing the m g . Bdjust the electrolyte volume to first ext,ract. Dilute t o exactly 50 ml. 25 ml.. neutralize to pH 7 with 1-to-1 n-ith ethyl acetate. I'r(~parc~t h r blank ammonium hydroxide solution, and solution by treating 10 nil. of 0.5;V make i t 0.5X in sulfuric acid. Transfei hydrochloric acid in exavtly tlic same way as t,lic vanadium solutioris. the solution to a 6o-nil. srparatorjfunnel and extract the solution three Aspirate thc coxittat's of t,hc standard times in succession with 0.025 gram of flasks into the flanic. Jteasurc. t h c cupferron and i ml. of ethyl acetate a+ intensity of the oxidr I)and at 550 mp. -. described for the calibration curvo. I o obtain tlir net emissivity. suht'ract Combine the extracts in a 25-m1. the blank ni(~asurrinrntfi.oni tlic t'otal volumetric flask and dilutr to the mark emissivity of t h e vanadium band. with ethyl acetate. Determincl the net 'The solutions u s r d caitainrtl 20. 40! emissivity of the solution a t 550 nip 60, 80, and 100 p g . pcr nil. of vanadium. using rxactly the samr method ant1 respectively. 'The slit was adjusted so conditions as dcsci ibed for the calihrathat' the solution cont'aining 100 pg. tion riiivr. per ml. gave a reading of 100% t,ransmittance. With tht. photomultiplier sensitivity set a t position 5 and the sensitivity control set at its clockwise limit. slit widths sniall(~i~ than 0.1 mm. could br maintained n.itliout difficulty. For rach sample, fiw wadings ~ v e i ~ takrn in rapid succcwioii, with tlir cup lowered froni thc capillary between rradings. 'l'he first tn-o i ~ ~ a d i n gofs a stit m r e usually low owing t o hubblc rntrapnirnt and w r e ignored. Thr nican of t,hc remaining t h i w readings the final measurrnirnt of \vas taken rmission, At 550 ink tlir calitwation curve is linear up to 100 pg. per ml. of vanadium. The detection limit. defined as thr caoncentration of vanadium that produc,es a net response equal to 170of t#h(' I)nc+ground reading. is 1 p g . prr ml.
PROCEDURES
'The theory of cupferron ext'ractions, as discussed by Furman, Mason, and Pekola ( 4 ) , indicates that both the hydrogen ion concentration in the aqueous phase and the cupferron concentrat.ion in bhe organic phase have a marked effect on the degrrc of estraction of a metal. The fffect of these two variables on thc dcgree of estraction of vanadium was investigated by the following procetlurc: 'r 7
1 n-cnty milliliters of n solut'ion, containing 0.20 mmolr: (10.19 nig.) of vanadium(1T') and a known amount of hydrochloric acid, and saturated with ethyl acetate. was treated with a known w i g h t of cupferron in a separatory funncl. After mixing, 20 ml. of ethyl acetatr was addcd and tJhemixture was csquilihratrd. After srparation of the phasrs, 10 ml. of the aqucous layer was taken and evaporated almost to dryncss. The residue was treated with sulfuric acid and nitric acid t o destroy organic mat,ter. The residue was dissolvrd in n-ater, and the vanadium was tletermincd specti,ophotometricall!- 1)y t h r hydrogen perosidc method ( 9 ) .
The results shown in Figure 4 indicate that vanadium(1V) is rxtracted quantit'ativrly froni solutions 0 . 5 s in hydi.ochloric acid if the cupferron concentration in the organic phase is larger than 9 x 10-2JI. With a free cupfrrron concrntration of 6 X 10-*M. ahout 95% of the vanadium is rxtract'ed with one extraction.
PH
IXssolvc, 1.0000 gram of strel in t h e minimum amount (about 1 5 nil.) of 1-to-9 sulfuric arid. Filter, if necessary, ant1 trniisfrr the filtrate to n mercury c~atliodr vlectrolysis ( 3 ~ 1 1 . Elwtrolyze until the iron content of t h e elwtrolj.t(~is lvss than 0.1 mg. Adjust t h r rlrrtrolyte volumr to 25 nil. and neutralizt> with 1-to-1 a~mmoniumhydi,oxidt. solution to pH i . :\dd 4 ml. of I-to-9 sulfiiric a&I. 'Transf r i . thv solution to a 60-nil. separatoqfrinnrl and rsti,act it t h c v t,inics in ion with 0.05 gi'ani of cupferron and 10 ml. of ethyl ac-rtatc as described for the calibration c u i ~ v t ~ .Conibine thr d r a c t s in ~t 50-inl. volumetric flask and tiilutc to the> niark with ethyl aretatr. Determine the cniissivity of the solution a t 530 nip using exactly the same met.hod and cmiditions as used for the calibration c~irvv. Ferrochromium. Ti,amfrr 0.5000 gi'ani of ferrochroiniuni t o R 40-ml. iron crucible containing 6 grams of sodium peroxide. M i s thoroughly and fuse. Transfrr t h r rooled mrlt to n beaker and add 30 nil. of water. Aftrr t h e rc.ac*tion has subsided. dissolve t h e precnipitate in the mininiuni amount of I-to-4 sulfuric acid. Reduw t>hechromiiim (VI) to hromium (111) with 30% Iiydrogrn peroxid(. diluted to give a I-to-1 aqueous solution. Rrniove the rxcrss hydrogcn primidr by boiling the solution for 5 minutw. Filter, if necessary, and rlwtrolyze thr, Steels.
(3
1298
ANALYTICAL CHEMISTRY
Figure 4. Extraction of vanadium as a function of p H Cupferron concentrotion in organic phase A. 9 X 10-*M B. 6 X 10-2M
Table 1. (100
Tolerance Limits for Interfering Metals
per ml. of vanadium present i n ethyl :irct:itt, ph:tse) Amount Interfering Permissible, Metal pg. per hll. Cerium 400 Copper 50 10 Iron Siobium 20 Tantitluni 400 Thorium 400 Tin 200 10 Tit:tniuni Tungstrti 400 Urttnir!m 400 Zirronirini 40
pg.
Magnetite-Ilmenite. Transfw t h r saniplr containing I t o 4 nig. of vanadiuni t o a nickel cmrible. Mix i t ivit'h about 10 tinies its ~-crlumrof sodium p r r o d t , ttnd fiisc. ('001 arid dissolvr the niclt in 40 nil. of cold water. ;itld 0.1 gram of sodium peroxidr t o the susprnsion and boil for 10 miiiutc,s. Allow to stand on the steam bat'h for 45 mini1tc.r. Filtci, and wash with a roltl solution containing 27, of sodium hydmxitlc and 1 yc of sodium su1fat.c. Seut,ralizc. the filtrate and washings to pH i with 140-9 hytlrochloiic acid The cupfei I on concrntration should anrl adjust t,he volume to ahout 40 ml. >lake, t h r solut,ion 0.5.Y in hydioc~hloric~ not br, too high. as riratic flame conditions may result. The frc,e cupferron arid anrl p r n c w d as for stctls. is coniplctrly extracted into the ethyl acetate phasr nith the vanadium cupfei r a t r ; the tiistiibution roefficient of DISCUSSION hydrogen wpferi ate beta ern acidified aqueous solutions and ethyl acetate way Extraction of Vanadium. T'anafound to hc 285 accoiding to local ohdiuni can lw qcparated from many v i vation. The cupferron cwncentrametals by selective extraction Ivith cupferion plus rthyl acetate ( 3 , 4). tion in the final solution was kept a t 0.1 5
Table
II.
Material Cr-I’-steel Fcrrochromium
KO.
Certified Vanadium Found, Value, % % Av., yo
XBS 64a
S I:~gnetite-ilmenite 0
values and confirm the validity of the method.
Analysis of Samples for Vanadium
Std. Uev.
0 154
0 162, 0 154, 0 142, 0 154
0 153
0 005
1 100
1 09, 1 07, 1 18, 0 99
1 08
0 045
‘The author wishes to thank the hIaiiagement of -4frican Metals Corp. Ltd. for permission to publish this work. LITERATURE CITED
Ol>taintd by photometric method.
giam per 50 nil. (about 2 x 10-*.U). Interferences. Table I shows a list of metals t h a t are also extracted from a 0 . 5 5 hydro(-hloric or sulfuric acid solution by &hyl acetate plus cupferron. T h e interference of each of these nicbtals \\-as tested by taking kiioan quantities of vanadiuni aiid each interfering mctal through the whole procedure and ascertaining the amount of each metal t h a t caused a ielativt, trior of 2% in th(x determination of vanadium. The permissible amount shoivn in the table is tho amount piesent in the aqueous bolution prior to mtraction. The amount of intwfcring metal extiacted with the vanadium was not determined. Cerium, tantalum, tin, thorium, tungh t m , and uranium can be tolcratcd in fairly large amounts. Hand and rantinuum spectral inteiference i6 euprv ic w c d whmi copper, iron, niobium. tita-
ACKNOWLEDGMENT
nium, and zirconium are present in amounts exceeding the limits indicated in Table I. Copper, iron, tin, a i d molyljdcnum are conveniently removed by electi,olysis with a mercury vathode cell. Fusion of the sample with sodium peroxide and subsequent leaching of tlie melt with water serves to separat,e copper, iron, niobium, tit~aiiium,and zirconium from vanadium. Iron(II1) and molybdenuni(V1) may IJC sc,parated from vanadium(1V) 12)- extraction from aqueous solution 6.0.11 in hydrochloric acid wit’h diethyl ct’hcr. The vanadium remains in the aqueous phasc ( 1 ) . RESULTS
Samples of three diffrreiit materials were analyzed using the method described. The results shown in Tablr I1 agree satisfactorily with the critified
(1) Bock, It., Kusche, H., Bock, K., Z. anal. Chem. 138, 167 (1953). (2) Burriel-Llarti, F., Ramirez-AIufioz, J., iisunciori-0marrenienteria, 31. C.. Inst hzerro y acero 9 , I l i (1956). (3) Foster, 11. l)., Grimnldi, F. S., Stevens. R. E.. I-. S. Geol, Survrv, Rept. 2 11944). ’ (4) Furman, S . H., Mason, R. B., Pekoltt, J. 6 . , h A L . CHEhI. 21, 1325 (1949). ( 5 ) Gaydon, A . G., “Dissociation Energies arid Spectra of 1)iatomic Molecules,” Chapman & Hall, London, 1947. (6) Gilbert, P. T., “Flame Photonwtry-
Sew Precision in Elemental Analysis,!’ Beckman Reprint, R-56, reprinted from Industrial I,aboratories, Beckman Instruments Inc., Fullerton, Calif., August 1952. (7) Herzbrrg, G., “l\Iolecular Spectra and llolecular St,ructure. I. Spectra of Iliatomic Molecules,” 2nd ed., Van Sostrand, S e a York, 19.51. (8) Mahanati, P. C., PTDC.Phy.s. SOC. (London) 47, 433 (1935). (9) Sandell, E. B., “Colorimetric Determination of Traces of Metals,” 2nd ed., Intersciencr, London, 1950
RECEIVEDfor review ,\larch 8, 1960. .4ccepted June 20, 1960. Extracted from part of a thesis t o he pul)mitted by C. ,Ll. Stander in partial fulfillment of requirements for the degree of dortor of vience at thr University of Pretoria.
Gas-Solid Chromatographic Separation of Some Light Hydrocarbons THOMAS A. McKENNA, Jr., and JOHN A. IDLEMAN Firestone Synthetic Rubber and latex
Co., Orange, l e x .
>
A combination column of silica gel and alumina may b e used to resolve some permanent gases and light hydrocarbons through the C4’s. Analyses of hydrocarbons heavier than Ch are feasible at elevated column temperatures, depending upon the sample composition. These same solids may be liquid-modified to shorten analysis time a t room temperature without loss of resolution.
T
allalysis Of light hydrocarbon streams from butane drhydrogrnation requires columns capalde of sepai,ating gases from hytlrogc~n through thc common Cq olefins and 1 ,3-hitadicw~. The authors H E CHROMATOGRAl’HIC
have previously reported ( 2 , 3 ) a twocolumn arrangrment for this analysis which involves a gas-solid and a gasliquid column. A single column operating at r,levatcd temperatures 11-ould simplify the procedure. and make it adaptable to a pi’occss c.hi,oinatograph. Patton, Lewis, and Kaye (4) reported the use of several active solids to separate gas mixtures containing hytlroc.arl)ons through prntanes. Greene and Pust ( 1 ) compared the separating properties of silica gel (Davison c h i c rant grade) and alumiiia (Alcoa F-I) using a mixture of low boiling gasrs and hydrocarhons through 1.3-hutadiwe. ‘I’hc use of silica gel (Davison G l d c 12j at reduced trmperatures is r ~ -
ported by Szulczcivski and Higuchi (6) for the separation of some oxides of nitrogen and low boiling gases. This paper is concerncd primarily n ith the development of a column composed of a combination of alumina and silica gel. This columii is capable of separating the components of a butane dehydrogenation stream, excluding air and caibon dioxide. APPARATUS
-1 Perkin-l
