ANALYTICAL CHEMISTRY, VOL. 50, NO. 12, OCTOBER 1978
1703
Corrections to the Third-Order Formula for X-ray Fluorescence Intensities Sir: In a recent Monte Carlo investigation ( I ) of X-ray fluorescence matrix interferences in three-component samples, it was found t h a t neglect of the third-order or tertiary fluorescence can introduce a significant error into the X-ray analysis of such samples. Therefore, the authors have initiated a study (2) of the accuracy of the theoretical formulas as derived by Sherman (3, 4 ) when applied to typical X-ray spectrometers as compared t o the more general, but calculationally less efficient Monte Carlo models ( I , 2, 5 , 6). In so doing, two additional errors in the original Sherman derivation ( 4 ) have been discovered and are corrected here. In Equation 6 of Ref. 4, the term given as should be corrected to log2 ( F f 3 / B 2 ) This error was apparently introduced by incorrectly simplifying the equivalent term given in Equation 5 . This error was particularly difficult to locate because it was negligible a t the exciting energies lower than or equal to those used in the examples calculated in the Sherman papers. Likewise the term of Equation 6 given as
T h e pertinent nomenclature is given in Ref. 4. T h e only numerical result reported by Sherman (3) that was incorrect was the third-order effect for Sample 2 . The true value should be 0.0024 instead of 0.0041. A computer program has been written by the authors for the calculation of the corrected first, second, and third-order formulas for X-ray fluorescence. This program is written in standard FORTRAN and is called SHERMA. It is available as a documented listing of FORTRAN statements from the authors.
LITERATURE C.[TED (1) A. R. Hawthorne and R. P. Gardner, Ana/, Chem.. 48, 2130 (1976). (2) J. M. Doster and R. P. Gardner, presented at the Twenty-Seventh Annual Denver X-Ray Conference, to be published in A&. X-Ray Anal. ( 3 ) Jacob Sherman, Spectrochim. Acta, 7 283 (1955). (4) J. Sherman, Spectochim. Acta, 15, 466 (1959). (5) R. P. Gardner and A. R. Hawthorne, X-Hay Spectrom., 4, 138 (1975). (6) A. R. Hawthorne and R. P. Gardner. Anal. Chem., 47, 2220 (1975).
R. P. G a r d n e r * J. M. Doster Nuclear Engineering Department North Carolina State University Raleigh, North Carolina 27650
-IC(WA2) should be corrected to
-IC ( - 0 3 / A2 )
RECEIVED for review May 1, 1978. Accepted July 10, 1978. Environmental This work was supported in part by the U.S. Protection Agency under Grant No. R-802759.
Dense Gas Chromatograph/Mass Spectrometer Interface Sir: It has been known for many years that dense gases-gases a t temperatures greater than their critical temperatures and a t pressures (20 to 2000 atm) sufficient to give densities near liquid densities-often are excellent solvents (1-3). Thus, a dense gas moving phase in a chromatograph can act as both solvent and carrier as in liquid chromatography whereas in conventional gas chromatography the gas is a carrier only. Three research groups have thoroughly studied some aspects of the increased solubility (or enhanced volatility) of a solute in dense gas chromatography (DGC) (4-10), but the general development of DGC has been slow partly because of the lack of a sensitive and selective detector such as a mass spectrometer. Because of the requirements of sufficient compound volatility and temperature stability in both gas and high pressure liquid chromatograph/mass spectrometer interfaces, a DGC/MS interface would be particularly valuable for studies of involatile, thermally labile compounds. In an earlier paper Giddings et al. (11) proposed using a supersonic molecular beam system (12) for such an interface and discussed possible problems arising from solute-solvent and solute-solute interactions during the expansion. Subsequent work has led to the design of a differentially pumped molecular beam interface and some preliminary results. The molecular beam is formed by expanding the dense gas (20 to 250 atm in our work to date) in a nozzle-skimmer-
collimator system (Figure 1). The nozzle orifice is a laserdrilled pinhole in nickel shim stock and may be easily changed to permit a wide selection of nozzle diameters. Presently we are using a 12.5-fim diameter nozzle, a conical skimmer with a 1-mm entrance diameter, and a 2.5.mm hole in a flat plate as a beam collimator. Both the skimmer and collimator can be easily changed. The nozzle-skimmer distance may be varied from near zero (about 0.3 mm) to greater than 70 mm. The skimmer-collimator distance is fixed at 102 mm and the molecular beam travels another 68 mm before intersecting the perpendicular ionizing electron beam of a modified PerkinElmer Model 270 mass spectrometer. Ionized molecules are accelerated orthogonal to both the molecular and electron beams into the electrostatic and then magnetic sectors of the double-focusing, Nier-Johnson type mass spectrometer. Two repeller electrodes a t adjustable potentials have been added to the ion source to increase the ion flux from the source: the first, the now isolated back of the ion chamber, increases the normal draw-out field arising from accelerating voltage field penetration; the second, located near the molecular beam exit hole, was to compensate for ion velocity components acquired in the molecular beam formation. Un-ionized beam molecules travel through the ion source and are detected by an ionization gauge. Another detection option is to allow ionized beam molecules to travel straight through the ion source (zero accelerating voltage) and impinge upon a current sensing
0003-2700/78/0350-1703$01.00/0C 1978 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 50, NO. 12, OCTOBER 1978
1704
I
I
A/
1
F
i
2400 USEC
4
Figure 1. Dense gas chromatographlmass spectrometer interface. F, filters, sintered stainless steel; D, preheat column; B, sampling valve; C, sample chamber; A, chromatograph column: E, shut-off valve, zero-dead-volume; V, sliding vacuum seal; N, nozzle; S, skimmer: L, collimator; I,ionization chamber; J, ion exit slit; K, filament; R, variable limiting orifice for ionization gauge; M, ionization gauge beam detector
3
2
B U
L
I
9
I
I
10
11
I
C
5
4 "
+ 12
h
I
13
P6
7 L 1
A
I
14
15
8
a
I
16
* I
17
1
I -L-,-L I
M *-tui
152 pix
289 A06
Figure 2. Mass spectra. A, B. Cluster ions from ethane, 35 atm. 35 OC. Pressure in nozzle-skimmer region is 1.1 X Torr. Number of molecules in cluster is given. Peaks in region labeled T are attenuated by a factor of 4.6. C. Anthracene and 1,3,54riphenyIbenzenein ethane, 171 atm, 33 "C. Pressure in nozzle-skimmer region is -5 X Torr. The m l e ratios are given. Peaks labeled P are from the potyphenyl ether pump oil background (CGH50C6H,+). Peak heights are on arbitrary
scale
electrode placed in the beam path. Each region is separately pumped with the following pressures typical with beam operation: nozzle-skimmer, 5 X to lo-' Torr; skimmer-collimator, Torr; ion source, 5 X to 10 s Torr; and molecular beam detector ionization gauge, lO-'to 5 x Torr. We are unable to find any prior references t o a molecular beam sampling system operating between such extremes of pressure. A uniform temperature from sample introduction to nozzle was found t o be essential. The solvent dense gas preheater, sample chamber and associated valves, column, and the nozzle assembly are closely thermostated by a water bath to a temperature just above the dense gas critical temperature for maximum solvent power. A heater a t the tip of the nozzle assembly can be heated to 60 "C above the thermostated temperature. Typical beam intensity vs. nozzle-skimmer separation curves (whether intensity is measured by the ionization gauge pressure or the mass spectrometer signal) have established the generation of a molecular beam. For dense gases a t 60 to 200 atm, we have found the mass spectrometer to be more sensitive than the ion gauge to the nozzleskimmer separation. As expected, condensation clusters were observed for several gases: nitrogen (room temperature) and ethane, ethylene, and carbon dioxide (just above critical temperatures). Mass spectra have been obtained with evenly spaced cluster mass +, Figure 2A, 2B, (CZH4)z01, peaks out to ( N Z ) ~(CzH,),,+-see and (CO,),,+. As ionization by the electron beam occurs in
a low pressure region and the ions are immediately drawn out of the molecular beam by a drawout field, it seems certain that these were true condensation clusters and not the result of ion-molecule reactions. Molecular beams of argon, helium, and hydrogen at room temperature produced no clusters. For reasons unknown, the ion source potentials in the present ion source had to be adjusted differently for each gas for optimum detection of the clusters: the ion focusing voltage varied by 60 V, the voltage difference between the two halves of the focusing lens varied from 2 to 18 V (0- 26 V possible range), the new repeller electrode potentials varied from 0 to 24 V above the ionization chamber voltage, and the electron energy varied from 30 to 70 eV. Observable cluster formation was a sensitive function of gas temperature, nozzle tip temperature, stagnation pressure, and, most particularly, background pressure of the nozzle-skimmer region-in which an increase in pressure by a factor of only two destroyed all clusters. With the present marginal nozzle-skimmer pumping speed, variation in gas flow through different pinholes of the same nominal diameter gave different background pressures at the same stagnation pressure resulting in very different observed cluster formation. We have easily dissolved and detected the parent and fragment peaks of naphthalene, anthracene, triphenyl benzene, phenanthrene, and azobenzene using as the solvent either ethylene a t about 17 " C or ethane or carbon dioxide at about 35 "C. While these compounds can all be easily separat,ed by a conventional GC a t a moderately elevated column temperature, at our low temperature, 35 "Cmaximum, detection was not possible until the solvent gas pressure was increased to 70 to 90 atm. Some small mass peaks have been observed that indicate that anthraquinone and inosine may have been dissolved and eluted using dense ethane. Inosine is not, to the best of our knowledge, separable in any standard GC. However, the reproducibility of the peaks has not been good and these results need verification. Figure 2C shows a mass spectrum with parent and fragment ions of anthracene and 1,3,5-triphenyl benzene. It should be noted that unusually high parent ion intensit,ies are commonly observed-e.g., chloroform, diethyl ether, anthracene, and azobenzene. Since the ionizing electrons are nominally in the energy range of 50-70 eV, it is believed that the predominance of parent ions is due to the abundance of lower-energy secondary electrons in the ion source. The mass spectrometer has an electron beam collimating magnet that has a nonuniform field in the region of intersection of the electron and molecular beams; this, we think, is why unusual potentials are required and strange behavior is noted. A new source design, under construction, should alleviate this problem. One would expect the molecular beam system to give a very significant increase in solute-solvent peak ratios as is usual in seeded beams but measurements of this effect have been postponed until the new ion source is installed. To date, no separations have been observed with an uncoated 6-m stainless steel capillary column (0.51-mm id.); the studies of Sie and Rijnders (.5. 6 ) , which show a maximum about 1.04 to 1.1 times the dense gas critical temperature in
ANALYTICAL CHEMISTRY, VOL. 50,
retention time vs. temperature plots, indicate that while solubilities would be smaller, separations would be increased by operation at a slightly higher temperature. The linear flow rate through this column a t 90 atm ethane is observed to be 7 cm/s-slightly slower than the optimum 20 cm/s for a standard capillary column and within the range of acceptable values. The major problem encountered thus far has been plugging of the nozzle orifice caused both by foreign particles and by sample deposition. Even with carefully filtered (multiplestage, 5 pm) solvent gas and a dual 512 bm filter placed directly between the final zero-dead-volume shut-off valve and the nozzle orifice, we have observed slow closure (after 1 to 10 h of operation) of the nozzle orifice from particle accumulation correlated with actuation of the preceding valve, the packing of which is graphite-filled fluorocarbon polymer. Inspection of the orifice once showed a symmetrical black deposit about the entire circumference with a long, thin particle ( - 2 pm by 15 pm) centered along a diameter. Installation of a triple 2./0.5/0.5 bm filter appears to have eliminated particulate buildup. A particulate plug could be dislodged only by removing the nozzle assembly from the system, whereas a solute plug of the compounds studied could be removed by alternately evacuating and pressurizing from the upstream side of the orifice and/or rapid heating of the nozzle tip. The nozzle tip was usually operated at bath temperature because heating a dense gas which is essentially a t the critical point decreases its solvent power with resulting deposition of solute. We have also seen solute deposited on the vacuum side of the orifice in such a way that it deflects gas flow from the beam center-line instead of stopping it. A t this point our test quantities (0.03 mg to 0.3 mg) of different solutes appear to have been too large to maintain an open orifice. There was typically plugging or partial plugging. T h e partial plugging was detected by lowered solvent gas intensity, increase in skimmer mid-point temperature. lowered expansion chamber pressure, and quite often an increase in solvent gas polymers. With a pure gas beam, a thermocouple mounted a t the mid-point of the beryllium copper skimmer gave a temperature reading 20-30 "C lower than one on the skimmer base (-6-mm separation). Upon sample elution the mid-point temperature at times dropped erratically to as low as -35 "C with the base temperature close to +30 "C. With visual inspection through a viewing port, we have seen neither deposits on the skimmer nor aggregates in the beam even a t the very low skimmer temperatures. While we have seen some clustering of solute with solvent gas-(C6Hd (CO,), (C6H6)(CO,),, (CI&H,)(GH,), (C,&H~) (CpHJ3 with the cluster intensity less than 20% of the solute intensity; it has not been seen with the larger, less volatile molecules
NO. 12, OCTOBER 1978
1705
and the higher solvent gas densities required to elute them. There appears to be a definite effect of solute upon solvent cluster formation, in that cluster formation often is inhibited by the presence of solute. This is reasonable in view of the greater number of collisions in the beam when species of greatly different molecular weight are present. Neither these collisions nor those with background gas which also break up solvent clusters seem to have major effect on the heavier solute molecules in the molecular beam over the limited pressure range of our studies to date. In fact, it appears that by proper adjustment of background gas pressure the weakly bonded solvent- and solutesolvent complexes may be destroyed with resultant simplification of the higher molecular weight spectrum to that of solute molecules only. While it is clear that further refinements and studies are needed, e.g., small volume sample introduction system, solvent gas and column selection, solvent gas density programming, use of mixed solvent gases, beam enrichment measurements, etc., the preliminary results indicate that a molecular beam interface between a dense gas chromatograph and a mass spectrometer is quite practicable and should have many applications.
LITERATURE CITED (1) J. E. Hannay and J. Hogarth, Roc. R . SOC. London, 30, 178 (1880). (2) H. S.Booth and R. M. Bidwell, Chem. Rev., 44, 477 (1949). (3) P. F. M. Paul and W. S.Wise, "The Principles of Gas Extraction", M & B Monograph CE/5, Mills & Boon Limited, London, 1971. (4) S. T. Sie and G. W. A. Rijnders, Anal Chim. Acta. 38, 31 (1967). (5) S.T. Sie and G. W. A. Rijnders, Sep. Sci., 2 , 729 (1967). (6) S. T. Sie and G. W. A. Rijnders, Sep. Sci., 2 , 755 (1967). (7) N. M.Karayannis,A. H. Cwwin, E. W. Baker, E. Klesper, and J. A. Walter. Anal. Chem., 40, 1736 (1968). (8) J. C. Giddings, M. N. Myers. and J. W. King, J . Chromatogr. Sci.. 7, 276 ( 1969). (9) J. J. Czubryt, M. N. Myers and J. C. Giddings. J . f h y s . Chem., 74, 4260 (1970). (IO) L. M. Bowman, "Dense Gas Chromatographic Studies", Ph.D. Thesis, University of Utah, Salt Lake City, Utah, 1976. ( 1 1) J. C. W i n g s . M. N. Myers and A. L. Wahhfty, Int. J. Mass Spechorn. Ion Phys., 4, 9 (1970). (12) Many varied applicationsin literature. One survey is "Molecular Beams and Low Density Gasdynamics", P. P. Wegener, Ed., Marcel Dekker, New York, N.Y., 1974,primarily Chapter 1.
L. G . Randall A. L. Wahrhaftig* Department of Chemistry University of Utah Salt Lake City, Utah 84112 RECEIVED for review May 15, 1978. Accepted July 20, 1978. This project has been financed in pari, under PHS Grant No. RR07092 (7/17/74-5/31/75) and in part under Grant No. R804335010 from the U S . Environmental Protection Agency (7/21/76-7/20/78).
Calibration and Performance of a Thermal Converter in Continuous Atmospheric Monitoring of Ammonia Sir: The important nitrogen-containing compounds in the atmosphere are N 2 0 , NO, NOz, and NH3, and salts of NO2-, NO;, and NH4+. The oxides of nitrogen (NO,.), namely, nitric oxide (NO) and nitrogen dioxide (NO,), are significant atmospheric pollutants. Although primarily emitted by natural sources, ammonia (NH,) from anthropogenic emissions contributes significantly to local concentrations. Atmospheric NH, results naturally from biological decay at the Earth's surface. The following processes account for
the fate of NH, in the atmosphere ( I ) : (1)Absorption on wet surfaces t o form NH4+,(2) Reaction iwith acidic material in either gaseous or condensed phases i,o form NH4+,and (3) Oxidation to NO,-. Routes 1 and 2 account for the fate of approximately 75% of the NH,,and Route 3 for the remaining 25% (1). Wet chemical methods for determination of ammonia (e.g., Nesslerization) are tedious and not easily adapted to on-line monitoring of emissions. The efficiency of ammonia uptake
This article not subject to U.S. Copyright. Published 1978 by the American Chemical Society
