its presence did not intcrfere with quantitative results. Analyses of both mixtures (Psc13PC13 and P0Cl3-PCl3) were quite reproducible, and choice of operating conditions was not critiral. The method should be satisfactory for routine analysis. Dimethyl Phosphite-Diethyl Phosphite. Four mixtures of dimethyl
phosphite-diethyl phosphite were analyzed. Each analysis of the first three mixtures listed under Table I was made at a different flow rate (57 to 91 ml. min.-’) using compounds immediately after purification, and a t a later date several analyses of the fourth mixture were run a t equal f l o ~rates. By this time the concentration of the impurity in each component had increased. Components of the mixture were run separately through the chromatograph and the impurity present in each was found to have the same retention time and in each case n-as present in the same proportion to the major component. For this reason the errors introduced by these impurities cancelled out Fluoropak 80 as solid support and either di-n-butyl phthalate or Apiezon K oil as liquid phase proved suitable as
the column packing. After repeated analyses some tailing occurred. This was overcome by occasionally cleaning the detector cell and replacing the glass mool. It was not found necessary to change the column packing. The use of Chromosorb as the solid support produced tailing to such a n extent that quantitative analysis was impossible. Examination of Figure 3 shows complete separation of the compounds u a s not achieved under the conditions employed. The minimum between peaks n a s used as a dividing line for area measurements (4). Phosphorus Pentachloride. An attempt was made t o analyze phosphorus pentachloride which sublimes a t 160’ C. and would, therefore, seem suitable for analysis by gas chromatography. Samples of PClj were dissolved in carbon tetrachloride or benzene and run through the chromatograph. There n a s no indication that PCls was eluted a t temperatures ranging from 120’ to 180’ C. while normal peaks were obtained for both solvents. It is not considered likely that the PCls was eluted simultaneously with its solvent as these solvents had considerably different elution times. A heavy yellow deposit
was observed inside the cell after these attempted analyses, indicating that the compound decomposed or reacted within the system. ACKNOWLEDGMENT
The authors thank the VirginiaCarolina Chemical Corp. for supplying the organophosphorus compounds and the Fansteel Metallurgical Corp. for the tantalum wire. They also gratefully acknon ledge the assistance of R. W. Kiser and Emilio Gallegos m-ith the mass spectrographic analyses. LITERATURE CITED
(1) Ellis, J F.: Iveson, G., “Gas Chro-
matography, D. H. Desty, ed., p. 300, Academic Press, Sew York, 1958. ( 2 ) Keeler, R. A., Anderson, C. J., Satriania, D., ANAL. CHEW 26, 933 (1954). (3) Keulemans,,, A. I. &I., “Gas Chromatography, p. 58, Reinhold, New York, 1957. (4) Pecsok, R. L., “Principles and Practice of Gas Chromatography,” p. 146, Wiley, New York. 1959. RECEIVED for review October 17, 1960. Accepted January 3, 1961. Taken from the M.S. thesis of S. H. Shipotofsky, February 1961. Contribution 30.607, K.A.E.S., Manhattan, Kan. Work performed under Contract AT( 11-1)-584 with the U. S. Atomic Energy Commission.
Double-Column Programmed Temperature Gas Chromatography EDWARD M. EMERY and W.
E.
KOERNER
Research Department, Organic Chemicals Division, Monsanto Chemical Co., St. Louis 77, Mo. A versatile double-column programmed temperature gas chromatograph permits the use of selective column packings whose vapor pressure at elevated temperature precludes their use in a single-column programmed temperature gas chromatograph. A specially designed column block i s provided to ensure sample vaporization and avoid condensation of column packing in the line between the column and the detector. Two completely independent gas flow control systems facilitate minor flow adjustments to compensate for lack of identical performance of the two columns. Automatic hold, cool, and reset features require the attention o f the user only for sample injection and signal attenuation.
T
HE extension of single-column programmed temperature gas chromatography (PTGC) to higher temperatures has greatly restricted the use of selective column packings (3-6).
Severe base-line drift is encountered when the increasing vapor pressure of the column packing is detected in the carrier gas stream entering the detector (see Figure 6). These base-line drift problems actually set the upper usable temperature limit for column packings in PTGC independent of the actual thermal stability of the column packing. Apiezon oils and greases, asphaltenes, various distillation residues, and silicone oils are used with fair success up to 300’ C., but above 300’ C. silicone gum rubber is used almost exclusivcly. The useful range of thermally stable but slightly volatile column packings is extended by the double-column programmed temperature gas chromatograph described in this paper. By programming the temperature of a pair of symmetrical columns and passing the effluent from the second column through the reference side of the thermal conductivity detector, compensation for the vapor pressure increase of the
column substrate is achieved and an acceptably level base line at much higher temperatures is achieved. I n common with single - column programmed temperature gas chromatography, column packing loss is minimized by exposing the packing to higher temperatures for only a short period of time during each analysis. Araki ( 1 ) has reported the successful use of the double-column principle, but the systems analyzed (gases and light oils) suggest that maximum upward extension of programmed temperature operating limits was not a prime goal of the work. APPARATUS
Gas Flow System. A schematic diagram of t h e gas flow system is shown in Figure 1. Moore constant differential-type flow controllers (Model 63BU-L, Moore Products Co., Philadelphia, Pa.) are used in the reference and sample VOL. 33, NO. 4, APRIL 1961
523
4 I
SAMPLE COLUMN
COLUMN THERMOSTAT
TEMPERATURE PROGRAMMER
1
1
f,RECORDER
2
SECTION A-A
1
DETECTOR POWER SUPPLY DETECTOR THERMOSTAT \
U
EX'TS
Figure 1. Schematic diagram of double-column programmed temperature apparatus
All electrical heater wiring omitted for clarity 1. Line from iniection block 2. Reference helium supply 3. Circu ating fan and motor 4. Transite tap plate, '/z inch thick 5. Column housing thermostat 6. Exploded view of column connection 7. Opening for cooling blower 8. Column housing heater mounted on stand-off insulators 9. Column housing, 6 inches in diameter X 2 6 l / ~ inches high, made of 20-gage stainless steel sheet 10. Connections from injection block cartridge heater to variable transformer 1 1 . Injection block 12. Columns 13. Sample helium supply 14. Column block, aluminum, 26/8 inches in diameter X 3I/r inches high 15. Openings for electrical wiring 16. Detector oven Transite end plates, '/zinch thick 17. Connections from column bock cartridge heater to variable transformer 18. Detector oven magnesia insulation, 1"s inches thick 19. Detector oven interior wa I, constructed from 6-inch galvanized stovepipe, 8 inches high 20. Detector block 21. Exit lines
streams. The two incoming helium lines are connected t o the columns in the instrument through the heated column block. A detailed drawing of the column block is included in Figure 2. This figure is drawn to scale but is only schematic. Section -4-A shows the exact geometric arrangement of the openings in the column block. A gas-tight column connection which will withstand elevated temperatures was made by drilling a 1/4-inch hole through the body of a Crawford Swagelok No. 400-C cap. The recess in the column block was machined to match the interior dimensions of a l/rinch Cra\\ ford Swagelok N o . 402-1 nut. Teflon ferrules are used to make a gas-tight connection n-hich can readily be disassembled to change columns. The use of recessed fittings in a heated block to make the column connections provides a convenient means of keeping these connections a t the de-
524
ANALYTICAL CHEMISTRY
d 0Fig. 2 . Detailed diagram of column housing and detector oven
sired temperature. If conventional Swagelok fittings, which would protrude from the block, were used, it would be necessary to remove and then replace heater windings over these connections each time the columns n-ere changed. Both ends of each column are connected to the column block in a similar manner. Each of the exit gas streams passes from the respective column through the column block into a heat exchanger consisting of a 36-inch length of coiled stainless steel tubing l/s-inch 0.d. 0.035 inch-n-all thickness located inside the detector oven. After passing through the heat exchangers, the gas streams flow through the two identical passages in the hotwire thermal conductivity detector and thence through stainless steel exit tubes, 1 's-inch o.d., 0.035-inch wall thickness, that extend through the bottom of the detector oven. The lower extremity of the sample line has a machined taper
which permits the ready attachment of standard hypodermic needles for fraction collecting. Injection Block. The injection block in the sample line was taken from a Perkin-Elmer Model 30.154-B Vapor Fractometer having the older type of externally heated block. A half-inch-thick aluminum block having the same n i d t h and height as the injection block was attached to the back side of the injection block. A 75-vatt cartridge heater (Katlon No. 3d35J, 3/9 X 2t/4 inches) was inserted in a 3/8-inch-d~ameterhole in the aluminum block and the whole assembly was insulated. Power for the cartridge heater is provided from a separate variable transformer to provide independent control of the injection block temperature. A thermocouple is inserted in a small hole drilled into the injection block itself to provide an accurate indication of the injection system temperature.
The line from the outlet of the injection block to the column block consists of stainless steel tubing, l/B-inch o.d., 0.035inch wall thickness. The line is wrapped with a layer of asbestos tape, then with 28-gage asbestos-covered Chrome1--4 wire, and finally with another layer of asbestos tape. Power for the heater winding is supplied through a variable transformer. The temperature of the line is monitored with a thermocouple which was silversoldered to the bare tubing before the heater winding mas applied. Column Block. A 175-watt cartridge heater (Katlow Firerod No. x 21/4 inches) was inG2E21A1 serted in a 3/g-inch-diameter hole in the column block (Figure 2). Power for the cartridge heater is supplied through a variable transformer. The temperature of the column block is monitored n i t h a thrrmocouple inserted in a small hole in the column block.
I
Figure 3. Top view of Variac drive mechanism 1.
Column Housing. T h e column housing, which is cylindrical in shape, is shown in Figure 2.
,4 100-c.f.m. centrifugal blower is inserted in the hole near the top of the housing. The cooling air can escape through the openings cut in the bottom of the housing for the gas and electrical lines. The top of the housing is capped with a Transite cover on which is mounted a Fenwal No. 17300-0 Thernioswitch which controls the initial column temperature, four standoff insulators for the 30-ohm bare wire heater controlled by the thermostat, and a small fan motor (Delco Model No. A-8430) n-ith a 4-inch-diameter fourbladed don ndraft fan attached to the shaft. Detector Oven. The detector is housed in a separately heated cylindrical air bath, shown in Figure 2. Heaters, consisting of 20-gage asbestos-coT-eied Chromel-A wire, are wound on the outside of the stovepipe interior nall of the oven and on a doughnut-shaped Transite form a t the bottom. The wall winding has a resistance of 24 ohms and the bottom n-inding 4 ohms. These windings :ire connected in series to a variable transformer 11-hose primary n-inding is connected t o the output of a constant voltage transformer. This simple arrangement provides satisfactory temperature uniformity. The top and bottom of the detector oven are constructed from circular pieces of Transite. The column block is mounted in a hole in the center of the cover. Detector Circuit. The detector is of the four-filament hot-wire type ( G o n - N a c Model 9255 pretzel geometry). This detector block meets the requirement t h a t both gas paths should have identical geometry, so t h a t changes in concentration of column packing vapors will be detected by both sets of wire filaments at the same time. A Perkin-Elmer Part 154-0366 hot-wire detector power supply (including signal attenuation circuit) is used. The output of the
2.
3. 4. 5. 6.
7. 8.
Synchronous motor with two-way friction clutch for manual adjustment of Variac setting Brass p ate, 4l/2 inches square X '/8 inch thick Limit microswitch for drive motor Brass bushing End of Variac shaft Adjustable arm for setting voltage limit Brass hexagonal bar, '/p inch wide (one at each corner of brass plate) Front panel of temperature programmer cabinet
detector bridge is transmitted to a 5-mv. strip chart recorder. Temperature Measurement. In addition to the above-mentioned thermocouples, one also is located on the detector cell block. All of the thermocouples are 7-foot-long 26-gage iron-constantan and are connected through a 12-point Leeds & Northrup Model 31-3 rotary selector switch to a n Assembly Products Model 461 thermocouple pyrometer. Temperature Programmer. The columns are wrapped with resistance heaters. Originally asbestos-insulated 20-gage Chromel-A a ire was wound over a single layer of glass sleeving into which the 1/4-inch 0.d. stainless steel column had been inserted. Prior to filling the column and inserting i t into the glass sleeving, an iron-constantan thermocouple (26-gage) insulated with Teflon ryas silver-soldered to the outside of the stainless steel tubing about 6 inches from the exit end. The silver solder connection was made in a small depression in the tubing which & a s made with a center punch. This minimized the size of the joint, so that the column could easily be inserted in the glass sleeving. S o problem of toxic vapor eyolution from Teflon was encountered n-ith programming up to 380" C. I t is understood that Teflon is safe up to 400" C., but if one were still concerned, the thermocouple could be passed out between two turns of the column heater ninding or asbestosinsulated wire could be used.
The asbestos insulation frayed badly in handling after the columns had been repeatedly programmed t o high temperatures. A more durable column heater was developed which consists of x 0.0031 inch Chromel-A ribbon inserted into 1/4-inch glass sleeving of the same type used to jacket the stainless steel column. Close but not overlapped Tyinding m-ith the ribbon and sleel-ing yields a column heater for a 1-meter column n i t h a resistance of about 25 ohms. K h e n parallel 1-meter columns are used in the instrument, the windings on the two columns are connected in series. The column windings n-ill be connected in parallel if 2meter columns are used t o maintain the same power input per unit length of column per volt. The pon-er for the column heaters is supplied by a motor-driven ball bearing Variac. The details of the construction of the motor drive for the programming Variac are shown in Figure 3. Drive motors of different speeds 1, and 2 r.p.h.) and different initial voltage settings of the Variac are used to produce a wide variety of temperature programs. The small drive motors are Type 117 synchronous motors from the Cramer Controls Corp., Centerbrook, Conn. A 0" to 500' C. strip chart temperature recorder (Leeds & Northrup Speedomas blodel H) is used to record the exploratory temperature programs. The recorder is not used once a suitable temperature program for a given analysis has been established and is being repeated routinely. Linear programming of the column heater voltage does not give a linear temperature rise, since the power dissipated is proportional to the square of the voltage (assuming the resistance of the heater nindings remains constant). Horvever, the heat loss is greater a t higher tcmperatures, so the resulting temperature program is almost linear. Sonlinear temperature programs ( 7 ) can also be achiel-ed by suitable choice of operating conditions. The circuit of the temperature programmer detailed in Figure 4 is arranged to disconnect the Variac drive motor when a preset heater voltage has been applied, t o continue to apply maximum voltage to the column heaters for a predetermined time period, to disconnect the pover to the column heaterq and simultaneously to actuate a cooling fan, and to turn off the cooling fan after a preset cooling pwiod has elapsed, so that the thermostat and hrater located in the top of the column housing can come into control to ready the instrument for the introduction of the next sample. &% 0- to 30-minute and a 0- to 15-minute Type 412 KD-1123 time delay relay timer from Cramer Controls Corp. are used to set the heating period at maximum voltage, to set the cooling period, and to actuate the necessary relays to accomplish the actions described above. For compactVOL. 33, NO. 4, APRIL 1961
* 525
YAR A i
troduced t o a 1-meter column of 25 weight % Apiezon-L on 60- to 80-mesh Chromosorb-W a t 155" C. and the column temperature was programmed to 380" C. The Variac was driven a t 1 r.p.h., starting a t 40 volts when the sample was introduced and with the Variac drive cutoff switch set for 113 volts. The injection block, connecting line, column block. and detector were all maintained a t 300" C. The detector current was 200 ma. The sample column helium flow rate was 82 cc. per minute and the reference column 65 cc. per minute.
rAR8AC
STAR-
WWER
DRIVE
C C 3 L UG
SWITCH
I
Figure 4.
Circuit of temperature programmer
Pilot lamps which indicate switch position omitted
uc
315
~~
4
5
4
8
Figure 5. Double-column programmed temperature chromatogram of 50% solution of polyphenyls in benzene 5-mv. recorder with detector signal a t maximum sensitivity, except for biphenyl peak, which is attenuated 16 times. Column temperature shown along top of tlgure
153
200
252
I
TEUOETATURE 3m
i't]
SQ
375
50
I
l i I
i
c' O!
I:
S'SGLE CCLLMN B A S E LINE
Ll
I -
DOJBLE COLUMN BASE
/
cL---6 IS 2 p 3 p-p30 -UE
.
35
40
45
LINE
48
Figure 6. Comparison of single- and double-column base-line performance of a well conditioned Apiezon-L column. 5-mv. recorder with detector signal at maximum sensitivity. tlgure
ness, the carrier gas flow control system is contained in the same cabinet as the temperature programmer. OPERATING PRECAUTIONS
In PTGC, it is necessary t o avoid excessive condensation of both column packing and higher boiling samples beyond the injection system by keeping the lines, connections, and detector a t a high temperature. With a column packing such as Apiezon-L, one can successfully program the column to as
526
ANALYTICAL CHEMISTRY
Column temperature shown along top of
much as 100" C. above the detector temperature. Apparently the small amounts of column packing that are deposited in the detector during the period of very high column temperature are re-evaporated during the cooling cycle. RESULTS AND DISCUSSION
A typical chromatogram of a wideboiling sample is shown in Figure 5 . A 5-pl. sample of a 5091, solution of a polyphenyl mixture in benzene was in-
In double-column programmed temperature gas chromatography it is necessary to provide independent control of the sample and reference column carrier gas flow rates, so that base-line performance can be optimized by adjusting the tn-o flows to compensate for minor variability in the column packings and column heater nindings. The use of PTGC facilitates the ready analysis of this wide-boiling mixture and also achieves acceptable resolution of mand p-terphenyl iyith readily available packing materials (bpiezon Grease L). The boiling points of the m- and pisomers are 363" and 376" C., respectively. Some specialized column packing materials have been reported (2) which give superior resolution of the two isomers, but these materials are not easily obtainable. The use of a 2meter column of the Apiezon-L also would have given increased resolution. The high boiling components were not collected and identified on this system. The tentative identifications shown in Figure 5 are based on retention time observations on triphenylene and available quaterphenyl isomers. The nomenclature used for the quaterphenyls assumes a linear structure comprised only of phenyl and phenylene groups. In Figure 6, a comparison is shown of the base-line performance obtained when the Apiezon-L column is operated under the normal double-column conditions and under a single-column arrangement where the second reference column is replaced by an empty tube. The superior performance of the doublecolumn arrangement is readily apparent. In both cases, the column, voltage program, and gas flon- rates were the same as in Figure 5 . -4 comparison of the column temperature scales on the two figures indicates the repeatability of the progrnniined temperature operation. Similar results have been obtained using columns packed nith Don-Corning silicone 710 oil. Tests with other slightly volatile but highly selective liquid phases are in progress. Utilization of the double-column principle should greatly broaden the utility of PTGC. The apparatus described in this paper
was successfully demonstrated early in 1960. Discussions with makers of gas chromat,ographs later led t o the recent introduction of two additional commercial double-column programmed temperature gas chromatographs.
of Lewis Fowler, A. J. Bindbeutel, and J. W. Sharp in constructing certain components for the instrument is also gratefully acknowledged. We thank G. Forrest Woods of the Universitv of Maryland for his generous donati& of the quaterphenyl isomers.
Harden, J. C., Ibid., r
flrnnnninnr4
ACKNOWLEDGMENT LITERATURE CITED I,,
As-,.:
a t .
T^
*__I_.^,L)
*T"
RECEIVED for review November 21, 1960. Accepted February 13, 1961. Division ?fn&nn;iytioal,Chemistry, . r ~ 139th Meeting, T~
lr~~~~lLin^.
tiectrocnromatogrciphic Separation of Silver and Thallium Ions from Each Other and from Mixtures of Vu rious PoI yva lerl t Cations HAROLD H. STRAIN, JOHN F. BINDER,' G. HARLOWE EVANS,%HARLAN D. FRAME, Jr., and JOHN
J. HlNES
Argonne National laboratory, Argonne, 111.
b Differential electricol migrotion in on ommoniacal solution of oxalate plus cyanide provides a complete separation of thollium ions from silver ions and from various polyvalent cotions. Migration in an ommoniacal solution of ammonia-triacetic acid provides a complete separation of silver ions from thallium ions and from various poly-
The addition of buffered citrate t o the background solution, for example, converted the divalent cations to anionic complexes leaving the alkali metal ions as uncomplexed cations. This observation indicated that all polyvalent ions might he removed from mixtures as
moniocol oxalote solution provides o separation of silver plus thallium ions from various polyvalent cotions. These migrations in the presence of complexforming solutes a r e rapid and complete even ot minute concentrations of the ions.
T,
HE n I F F E R m T m L electrical migration of ions from a narrow zone in a stabilized background electrolytic solution provides a sensitive and effective method for the separation of mixtures of various substances (5-6, 9-1s). These electrochromatographic separations depend upon the solvent ( I S ) , the background electrolyte (3,7), the stabilization medium (81, and the electroosmotic flow of the solution (1). They are effective over wide ranges of concentration, from about 0.10M to the lowest concentrations that can be detected by the most sensitive tests (6). Evans and Strain have shown that the separation of the alkali metal cations from the alkaline earth cationsis improved by the addition of complexing
Present address, Laminated Products Demrtment. General Electric Go.. Coshocot,. Ohio. ' Present address, Department of Chemistry, Illinois State Normal University, Normal, Ill.
anions ieavlng ail monovalein elemems as cations. T o exploit thc possibility of separating monovalent mercurous, silver, and thallium ions from polyvalent cations, we examined the effect of many complexing agents, mixtures of these snhstances, and variation of the pH on the differential electrical migration of various cations. Procedures have been found that permit the separation either of silver or thallium from multicomponent mixtures of various ions, and also the separation of silver and thallium from one another and from any one of several of the polyvalent ions. As indicated by the most sensitive tests, these separations, based upon electrical migration in opposite directions, are complete wit.hout any cross contamination. EXPERIMENTAL
Stabilization Medium. The background solutions were stabilized with commercial filter paper made of wood pulp (Eaton-Dikeman Co., Grade 301; thickness 0.030 inch). Several sheets were washed by downward percolation with 6 N nitric acid for 1 day and then v i t b distilled water for 4 days using the arrangement shown in Figure 1. The mashed , . > sheets . . were separated an$ . 1~
.~~/-
~~
Fisure 1. Filter DO PE pGIlyethylene sheeGngI Supported for washing wiih Wok,
I
U"
4
n,,.
a p L b
'"'Y'L&
UUlU
,1115.
VOL. 33, NO. 4, APRIL 1961
lllr
527
