measuring the absorbance of the solution. Accordingly, an absorption cell attachment was designed and fastened to the Lucite case that contained the reflectance heat of the Densichron. A double-pole, double-throw switch was used to switch the light source from the internal lamp of the reflectance head to the lamp in the absorption cell attachment. The cell holder attachment is machined from a block of Bakelite. A Parr Laboratory No. 3003 thermometer reading lens collimates a light beam from a General Electric No. 46 lamp. The collimated beam passes through a 3/lsinch diameter hole in the Bakelite and then through a 2 X 2 inch interference filter that is held in a slot in the Bakelite by two stainless steel spring clips made from 0.035-inch stainless steel tubing. The light beam then passes into the cell compartment through another 3/16-in~h diameter hole. The cell compartment is formed on three sides by a 0.510-inch wide slot machined in the Bakelite block. The fourth side and the bottom of the cell compartment are formed by an L-shaped piece of Bakelite that is fitted into the slot. The cell compartment is square in cross section and will receive a Beckman square absorption cell of 1-em. light path length. After passing
through the cell, the light beam passes through another 3/16-in~h diameter hole in the Bakelite and then through a a/,-inch diameter hole in the Lucite case in which the reflection unit is mounted. The light beam then passes into the photoelectric receiver unit, which is not modified in any way. The power supply of the reflectance unit lamp can be switched alternately to provide power for the General Electric No. 46 lamp of the absorption cell attachment. A suitable rheostat is inserted in series with the lamp filament to decrease the intensity of the light beam. Details of the attachment are shown in Figure 1. Drawings of the component parts of the absorption cell attachment are available from the authors. Tests are made for nitrate-nitrite content by the brucine method, for manganese by oxidation to permanganate with persulfate and silver, and for reducing power by reaction with ceric sulfate solution and measurement of the diminution in color of the ceric solution. The precision obtained on these tests was better than =t12% standard deviation a t a concentration level of less than 0.08 mg. per ml. The calibration curves for the various ions have the following upper limits: Nos, 9 pg.; reducing power, 0.010 meq.; Mn, 4 pg. A light
filter of the interference type (Photovolt Co.), that has a maximum transmittance a t 410 mp is used for the nitrite-nitrate and reducing power tests. A similar filter that has a maximum transmittance a t 520 mp is used for the manganese test. Since these tests are primarily of the “go-no go” type, a high degree of precision is not required. The information desired from the analyst is only that the solution is within certain set limits. The combination of the techniques of measuring reflectivity and absorbance with the same instrument can facilitate the performance of the abovementioned analyses remotely within hot cell facilities. ACKNOWLEDGMENT
The authors are indebted to D. J. Fisher for suggesting the use of the Welch Densichron and for suggestions concerning the design of the absorption system. P. F. Thomason was also of assistance on the chemical phase of the problem. OAKRidge National Laboratory is operated
by Union Carbide Corp. for the U. S.
Atomic Energy Commission.
Rapid Analysis and Sample Introduction in Gas Chromatography D. L. Peterson and G. W. Lundberg, Shell Development Co., Emeryville, Calif.
developments in gas chroR matography have made possible such significant reductions in analysis ECENT
times as to make practicable nearly instantaneous analyses of volatile gas mixtures. High speed chromatography has been used in these laboratories for some time with very satisfactory results in kinetic studies where several analyses per minute were required over a period of hours. In one application the quaternary system ethane, propane, propylene, and n-butane was analyzed in only 10 seconds (Figure 1). This work has been greatly facilitated by the development of a simple criterion by which a column packing permitting a minimum analysis time may be selected and of a sample introduction valve capable of almost instantaneous sample injection. The minimum time criterion is derived purely through a consideration of equilibria; differences in band spreading through mass transfer limitations in the cases being compared are neglected. (A more sophisticated approach was recently discussed by Purnell and Quinn, Preprints, Third Symposium on Gas Chromatography, June 8 to 10, 1960, Edinburgh, Scotland, pp. R15kR162, 652
*
ANALYTICAL CHEMISTRY
Butterworths, London, 1960.) A comparison on this basis is generally
2
E
2E’ ‘ c
6 C
0
5 Time, seconds
10
Figure 1. High speed chromatogram of ethane, propane, propylene, and n-butane Column. 4-inch silica f &inch alumina Column, O.D. ‘/a inch Carrier gas. Nitrogen Upstream pressure. 225 p.r.i.g. Temperature. 75’ C. Fa2
= 0.82
adequate. That column packing for which the peaks appear the soonest in relation to their relative separation and for which the most even distribution of peaks is obtained will then give the shortest analysis time. In the (unattainable) limiting case, some single peak is being developed a t every instant during the analysis and the carrier gas volume is simply the sum of the peak widths. The approach to this ideal limit by a real column may be expressed by a “peak distribution factor,” F, which is defined as the ratio of the sum of the peak widths to the actual total volume. If d. and wi are the retention volume and peak width, respectively, of component i, then for z components
The assumptions are made that the elution curves are Gaussian and that the number of theoretical plates, n = 16d;/w;, is the same for all components. The number of required plates is fixed by the condition that the two con-
I
Carr:er Out
di lb-i
Packing Sut
Injection Sh-f:
I I
Figure 3.
Fraction of Column Length as Silica
Figure 2. columns
Peak distributions of composite silica and ahmina
Curves calculated from end points, circles experimental. Subscripts 2, 3, and 4 refer to propane, propylene, and n-butane, respectively. Silica. Davison Type 62 Alumina. Alcaa grade F-20 Column, O.D. inch Column length. 12 to 30 inches Carrier gas. Nitrogen Upstream pressure. 40 p.s.i.g. Temperature. 25' C.
secutive components, p and q, for which d,/dp has the smallest value are just resolved dp
+ wp/2
=
dq
- wq/2
(2)
An expression may now be derived for F i n terms of retention volumes only:
Equation 3 is a convenient guide in screening adsorbents and solvents for high speed chromatography, even though it is based on somewhat crude assumptions. The retention volumes (or the known distribution coefficients) are employed in a calculation of F for each packing material considered and for promising combinations. That case for which F is a maximum involves the smallest redundant carrier gas volume and would be expected to permit the shortest analysis time. If the volatility range of the compounds is not too great, solid sorbents should be considered because they
may possess higher values of F than liquid solvents, owing to their higher selectivity and hence lower required carrier volume for resolution. Figure 2 shows the performance expected from combinations of silica and alumina geh in the determination of ethane, propaqe, propylene, and n-butane. m l e alumina appears superior to silica, a column comprising 0.20 of its length as silica and the remainder as alumina is expected to yield a higher value of F (0.80). This improvement in peak distribution was observed in high speed analyses over composite columns, though at a slightly higher silica content, probably because of the failure of the assumptions underlying Equat,ion 3. A simplified version of MeWilliam's hydrogen flame ionization detector [McWilliam, I. G., Dewar, R. A., Nature 181, 760 (1958)l was chosen as most appropriate for high speed use by virtue of its high sensitivity, extremely low holdup, and rapid response. The recording system is a Model 150 Sanborn galvanometer recorder (0.1 second for full-scale deflection),
Sample introduction valve
The requirements of the sample introduction system were the charging of 1 pl. of gas a t atmospheric pressure into a carrier at nearly 20 atm. in a time short in comparison with the width of the narrowest elution curve (0.6 second). I n addition, an interruption or variation of the carrier f l o may ~ disturb the base line and confuse the early portion of the chromatogram. A simple version of a sample introduction valve which meets these demands is shown in Figure 3. The sample is introduced through a hole in a movable steel injection shaft which may be situated in either of two slightly oversized chambers (the chamber size is exaggerated in Figure 3) through which either carrier or sample is flowing. The valve body is turned from brass. Seals are provided by Teflon held under compression (by bolts not shown).
It is thus possible to transfer a sample of fixed size and a t low pressure into the carrier stream under high pressure without interrupting the flow of either. With a set of exchangeable shafts the sample volume is conveniently varied. The only critical features of the valve are the sample hole, which must be chamfered and polished a t its entrances, and the seal and shaft surfaces, which must be circular and smooth. The valve introduces samples of constant size to within =t0.5%. With proper seals it is possible to maintain a static vacuum of under 1 mm. in the sample chamber with a carrier gas pressure of as much as 300 p.s.i. The valvelis easily made automatic by attaching the injection shaft to that of a pneumatically operated valve whose supply pressure is remotely controlled by a cycle timer.
VOL 33, NO, 4, APRIL 1961
e
653
