Anal. Chem. 1987, 59,681-684
where pep is the electrophoretic mobility, po is the electrophoretic mobility of the buffer, L, is the length of the column, D is the diffusion coefficient, and V is the applied voltage. Since two band broadening factors are introduced with the microinjector, the spatial variance due to the injector ( u t ) can be represented as
where ti is the injection time, Li is the distance from the tip of the injector to the column, Vi is the injection voltage, and &b is a term describing the contribution due to the presence of turbulence within the injector. The injector turbulence term is experimentallyrelated to the difference between the column and injector tip inside diameters (dc- d,), and is also expected to be related to the distance from the tip of the injector to the column (L,),the electroosmoticmobility of the buffer (po), the injection time (ti), the length of the column (Lc), the applied voltage during injection (Vi), and the electrophoretic mobility of the species (pep). Work is currently being done to correlate the relationship of these parameters to the band broadening contribution observed. The total spatial variance ( uT2) expected when a microinjector is used is the sum of the individual spatial variances
(4)
The separation efficiency in capillary zone electrophoresis has been borrowed from chromatography (6) as suggested by Giddings (19). The number of theoretical plates, N , is defined as
N = L,/aT2
(5)
68 1
inside diameters of the injector tip and the column. This suggestion is in agreement with the data shown in Table I where an injector tip more closely matching the diameter of the separation capillary shows a loss in efficiency of only about 5 %. Thus, by minimization of the orifice diameter difference, the term appears to approach a negligible value. If this behavior is consistent with a wide range of capillary diameters, then efficiencies approaching those obtained with direct injection should be attainable with ultrasmall injectors as long as similarly small capillary bores are applied. ACKNOWLEDGMENT The authors are grateful to James Jorgenson and John Green of the University of North Carolina at Chapel Hill for their assistance in constructing the capillary electrophoresis apparatus. LITERATURE CITED (1) Microcolumn Separations; Novotny, M., Ishii, D., Eds.; Elsevier: New York, 1985, Preface. (2) Yang, F. J. HRC CC, J. High Resoiuf. Chromatogr. Chromafogr. Commun. 1980, 3 , 589-590. (3) Yang, F. J. J. Chromatogr. 1982, 236, 265-277. (4) Jorgenson, J. W.; Guthrie. E. J. J. Chromafogr. 1983, 255, 335-348. (5) Kennedy, R.; St. Clalre, R. L., 111; White, J. G.; Jorgenson, J. W.Plttsburgh Conference; Atlantic City, NJ, March 12, 1986; No. 538. (6) Jorgenson, J. W.; Lukacs, K. D. J. Chromatogr. 1981, 278, 209-216. (7) Jorgenson, J. W.; Lukacs, K. D. Science 1983, 222, 266-272. (8) Gebauer, P.; Deml, M.; Bocek, P.; Junak, J. J. Chromafogr. 1983, 267, 455-457. (9) Jorgenson, J. W.; Lukacs, K. D. HRC CC, J. High Resoluf. Chromatogf. Chromatogr. Commun. 1985, 8 , 407-411. (10) Lauer, H. H.; McManigill, D. Anal. Chem. 1986, 5 8 , 165-170. (11) Tsuda, A.; Norura, K.; Nakagawa, G. J. Chromatogr. 1983, 264, 385-392. (12) Deml, M.; Foret, F.; Bocek, P. J. Chromafogr. 1985, 320, 159-165. (13) Stone, T. W. Microiontophoresis and Pressure Injection; Wiley: New York, 1985; Chapters 1-3. (14) Guthrle, E. J.; Jorgenson, J. W. Anal. Chem. 1984, 5 6 , 483-486. (15) Tsuda, A.; Nomura, K.; Nakagawa, G. J. Chromatogr. 1983, 264, 385-392. (16) Fujiwara, S.;Honda. S.Anal. Chem. 1986,5 6 , 1811-1814. (17) Pretorius, V.; Hopklns, B. J.; Shieke, J. D. J. Chromatogr. 1974, 732, 23-30. (18) Massey, B. S. Mechanics of Fluids, 4th ed.; Van Nostrand Reinhold: New York, 1979; pp 197-200. (19) Glddings, J. C. Sep. Sci. 1989, 4, 181-189.
Ross A. Wallingford Andrew G. Ewing*
Substituting eq 4 into this expression results in
N =
L:
r
Our results suggest that the major parameter involved in determining Btwb is apparently the difference between the
Department of Chemistry The Pennsylvania State University University Park, Pennsylvania 16802 RECENEDfor review July 29,1986. Accepted October 20,1986. This material is based upon work supported by the National Science Foundation under Grant No. BNS-8504292.
Application of a Modulated Magnetic Field to a Flame Photometric Detection Burner in the Detection of Phosphorus Sir: Phosphorus is one of the important elements that needs to be detected in the semiconductor industry and in environmental, biological, and geochemical scientific work (1-3). Analytical techniques commonly used for phosphorus determinations include the colorimetry of phosphoromolybdenum peteropoly blue, N / P thermionic detection, and flame photometric detection (FPD). FPD has seen widespread use in gas chromatography as a specific detector for sulfurand phosphorus-bearing compounds. This detector operates on the principle that phosphorus and sulfur compounds emit characteristic green and blue colorations, respectively, in 0003-2700/87/0359-068 1$01.50/0
hydrogen-air flames. In the case of phosphorus, the green emission is due to the HPO molecules and the intensity of this emission is known to vary lineary with P atom flow into the flame (4). Recently, we found that the emission intensity of some excited species in hydrogen-oxygen flames changed under the effect of a static magnetic field; for example, the emission intensity of HPO was partially magnetically quenched, taking minimum value a t 0.45,0.81, and 1.58 T (5). The emission intensity from HPO decreased about 20% at 0.45 T. This phenomenon cannot be explained by the Zeeman effect, be0 1987 American Chemical Society
682
ANALYTICAL CHEMISTRY, VOL. 59, NO. 4, FEBRUARY 15, 1987
(b)
HPO
4
C
Y
l
-2
( cI )
G P
s2
a
0
FPD Burner
Figure 1. Outline of the experimental arrangement of the new background correction system (a) and the conventional FPD (b). I
I
0
L - 1 - u 0
10
20
30
Time (msec)
a modulated magnetic field (a) and the PM outputs in the aspiration of the (NH4)pHP04solution (P, 10 yg/mL) (b) and the (NH4),S04 solution (S, 50 yg/mL) (c). The emission from S2 was observed through a glass filter (Toshiba UVDlA). Figure 3. Transient behaviors of
I
H7
N.
1
1
Aerosol
Figure 2. Experimental arrangement of the burner, rotating magnetic system, and optical guide.
cause relative spectral distributions did not change. In this report, we propose a new background correction system for FPD detection of phosphorus (6) that uses the magnetic field effect found in HPO. In the present system, a modulated magnetic field, the strength of which varied from 0 to 0.45 T at 98 Hz,was applied to the FPD burner in order to modulate the emission from HPO selectively. Associated noises, i.e., flame noises and PM dark current noises, were not modulated. The output of PM was fed to a tuned band-pass amplifier (98 Hz)and a phase sensitive detector to reject associated noises and extract the signal of phosphorus from high noise levels. This new technique will be compared with a conventional FPD. EXPERIMENTAL SECTION O u t l i n e of E x p e r i m e n t a l Arrangement. The outline of the
experimental apparatus is shown in Figure la. Figure 2 illustrates the arrangement of the FPD burner, rotating magnetic system, and optical guide. A modulated magnetic field was applied to the FPD burner by using a rotating magnetic system (A). The magnetic field, flame (B),and optical guide (C) were perpendicular to each other. The flame emission was guided by a Pyrex glass rod (10 mm in diameter, 300 mm in length, covered by aluminum foil) through broad band-pass glass fiters (D) to a photomultiplier (PM) (HTV, R453). The PM supply voltage was 800 V. The outputs of PM and reference signals from the rotating magnetic
system were fed to a lock-in amplifier (PAR 124A). B u r n e r . A schematic illustration of the FPD burner is shown in Figure 2. It was made of Pyrex glass. Premixed hydrogen (540 mL/min), the aerosol of an aqueous solution of (NHJ2HPO4(0.16 mL/min), and nitrogen (820 mL/min) flowed through the central nozzle (E),the diameter of which was 3 mm. Air (300 mL/min) passed through the outer ring (F),the inner and outer diameters of which were 5 and 8 mm, respectively. The aerosol of the phosphorus solution was produced by an ICP nebulizer (Shiamadzu), which was operated by flowing nitrogen. The ICP nebulizer introduced a trace amount of phosphorus into the flame in these experiment. Of course, it is possible to use the present system as a detector for phosphorus in gas chromatographic effluents. Rotating Magnetic System. A modulated magnetic field was obtained by rotating a pair of soft iron disks to which four pairs of magnets (Sumitomo Special Metals Co., NEOMAX-30,25 mm in diameter, 9.5 mm in thickness) were attached. The disk in front was omitted in this figure. The pole gap was 13 mm. The direction of the magnetic field was from back to front. This system delivered a modulated magnetic field, the strength of which varied from 0 to 0.45 T at 98 Hz, as shown in Figure 3a. The reference signals were synchronized to coincide with peak and zero field. Of course, it is possible to use an electromagnet instead of the rotating magnetic system. Measuring System. A modulated magnetic field was applied to the spot 20 mm above the nozzle head of the burner. The flame emission from the part where a modulated magnetic field was applied was guided by the glass rod through broad band-pass glass filters (D) (Toshiba VY-50, Corning 4-96) to a PM. The output of the PM was fed to a lock-in amplifier, which consisted of a tuned band-pass amplifier and a phase-sensitive detector. First, the former amplified a narrow band of frequency (98 Hz, Q = 12.34), Le., the modulated signal from phosphorus compounds, rejecting associated noises. Then the latter provided the means to extract low-amplitude signals from high noise levels. In this report, the present background correction technique was compared with a conventional FPD (Figure lb). In the latter case, the flame emission was observed through a narrow band-pass interference fiter (G) (Aw, 528 nm; AX1,*, 8.5 nm; T-, 42.5%), and the output of the PM was directly traced by a recorder. Other experimental conditions were similar to those described previously. The transient behaviors of the emission from HPO and S2 were recorded by a signal averager (TN-1505).
ANALYTICAL CHEMISTRY, VOL. 59, NO. 4, FEBRUARY 15, 1987
(a)
0 '
1
I
(b)
I
I
0 A 0
10
20
30
Time (msec) (C)
P
H2 0
Flgure 4. Transient behaviors of the PM output (a), the signal from a tuned band pass amplifier (b), and the output of a lock-in amplifier (c) in the aspiration of the (",)PHPo, solution (P, 0.5 pg/mL).
Reagents. The standard solutions of phosphorus (P, 1000 pg/mL), sulfur (S, lo00 pg/mL), and carbon (C, 100 ,ug/mL) were prepared by dissolving ammonium phosphate, diabasic ((NH4)zHP04),ammonium sulfate ((NH4)2S04),and ethanol in distilled water, respectively. Other concentrations used were prepared fresh from the standard solutions for every experiment. All of the reagents were special grades (Wako Chem. Ind., Ltd.). RESULTS AND DISCUSSION Effects of a Modulated Magnetic Field. The output of the P M was composed of the signal from phosphorus compounds (HPO) and associated noises. The latter were due to flame noise and P M dark current noise. When the concentration of the phosphorus solution was high, the signal from HPO was much larger than the background noises. Figure 3b shows the transient behavior of the emission intensity from HPO at a P concentration of 10 pg/mL. Magnetically induced quenching (about 30%) was observed around the peak of a magnetic field. On the other hand, associated noises, i.e., flame emissions (OH, C2,CH, S2,etc.) and P M dark currents, were observed not to be affected by a modulated magnetic field. Figure 3c shows the transient behavior of the emission intensity from S2molecules at a S concentration of 50 pg/mL. These experimental results indicate that the signal from phosphorus can be selectively modulated by applying a modulated magnetic field. As the sample concentration decreased, the signal from phosphorus decreased relative to the associated noise. At the low sample concentration (P, 0.5 pg/mL), a modulated HPO signal was not observed, as shown in Figure 4a. When the output of the, P M was filtered by a tuned band-pass filter, a modulated signal from phosphorus was observed, with most of the associated noises being rejected (Figure 4b). The output of a lock-in amplifier is shown in Figure 4c. The signal from P was clearly observed, as compared with the blank signal obtained with distilled water. Detectability. Most prior FPDs have used a narrow band-pass interference filter in order to cut off the associated flame emissions (OH, Sz, Cz, CH, etc.). In the present system, a modulated magnetic field, a tuned band-pass amplifier, and a phase sensitive detector were used to select the emissions from HPO. Therefore, only broad band-pass glass filters were used since the emission intensity of HPO observed through
683
0
c
I
A
0
C
Flgure 5. Comparison of the new background correction technique (a) with the Conventional FPD (b): (A) aspiration of 0.1 @g/mL phosphorus solution, (6) aspiration of distilled water, and (C) flame off.
these glass filters was about 12 times as large as through a narrow band-pass interference filter. The increase of the signal obtained by using broad band-pass glass filters was expected to be larger than the decrease obtained by using only magnetically modulated signals in the present system. The contributions from associated flame emissions were observed to be negligible in the present system. For example, the signal from sulfur compounds, S2,was observed to be negligible in comparison with that from phosphorus (1pg/mL), when the aqueous solution of (NH4)2S04(S, 50 pg/mL) was nebulized into the flame. Similarly, the sensitivity of the present method to hydrocarbons was less than 0.01% of its sensitivity to phosphorus. One of the major features of the present system is the potential for increase in sensitivity. This is demonstrated in Figure 5a, which shows the analysis of the phosphorus solution (0.1 lg/mL). The signal produced by P was clearly recorded, compared with the blank signal which was obtained in the aspiration of distilled water. As a comparison, a measurement was performed by the conventional FPD. Direct tracing of P M output is shown in Figure 5b. The signal from P was about twice as large as the peak-to-peak noise level of the blank signal. Comparison of these figures shows that the ratio of the P signal to the peak-to-peak noise level (distilled water) of the present system is larger than that of the conventional FPD. S t a t i c Background Correction. Accurate background correction of the present system is demonstrated in Figure 5a. The background intensities were measured in the aspiration of distilled water and in the absence of the flame. Both of these cases caused no base-line shifts. On the other hand, large base-line shifts were observed both in the absence of the flame and in the aspiration of distilled water in the case of the conventional FPD (Figure 5b). The former base-line shift is considered to be caused by the PM dark currents, while the difference between these two kinds of base-line shifts is due to the flame emissivity. The signal intensity from phosphorus was observed to be much smaller than the blank signal (distilled water), the ratio of the former to the latter being about 0.05. Comparison of parts a and b of Figure 5 shows the ability of the present system to correct for static backgrounds completely. Calibration Curve and Detection Limit. The calibration curves were linear in the range from 0.02 to 200 pg/mL in the present system, as shown by the equation log I(mV) = 1.086 log C(pg/mL)
+ 0.139
The standard deviations for the slope and intercept are 0.009 and 0.008, respectively. The detection limit of the present system was about 0.008 pg/mL defined as the concentration corresponding to twice the standard deviation in the blank signal (distilled water). The present system was about 5 times
684
Anal. Chem. 1987, 59,684-687
as sensitive as the conventional FPD, the detection limit of which was about 0.04 pg/mL. In conclusion, the major advantages of the present technique are the improvement of the sensitivity by about a factor of 5 as well as the accurate correction for static backgrounds. Nevertheless, the improvement of the sensitivity by the present method is not as good as expected. This may be due to the incompleteness of the modulated magnetic field used, because the accuracy of the period and peak strength is estimated to be about *1%. The purpose of this work is to compare the performance of this new technique relative to that of conventional FPD. In order to get performance higher than that presented here, it will be necessary to use an improved modulated magnetic field (higher modulation frequency, a more completely modulated magnetic field), a burner, optical system, sample introduction system, combustion conditions, etc.
and K. Fukuda of the National Chemical Laboratory for Industry for useful discussions.
LITERATURE CITED (1) Nozoye, H.; Someno, K. Bunseki Kagaku 1985, 8 , 508-509. (2) Hashlmoto, S.: Fujiwara, K.; Fuwa, K. Anal. Chem. 1985, 5 7 , 1305- 1309. (3) Murphy, J.; Riley, J. P. Anal. Chim. Acta 1962,27, 31-36. (4) Brody, S.S.;Chaney, J. E. J . Gas Chromatogr. 1988,4 , 42-46, (5) Wakayama, N. I.; Ogasawara, I.; Nishikawa, T.; Ohyagi, T.; Hayashi, H. Chem. Phys. Lett. 1984. 107, 207-211. (6) Wakayama, N, I.; Nozoye, H.; Ogasawara, I.; Fukuda, K. Japanese Patent 60-43516.
Nobuko I. Wakayama* Hisakazu Nozoye Ichiro Ogasawara National Chemical Laboratory for Industry Tsukuba Research Center Yatabe, Ibaraki 305, Japan
ACKNOWLEDGMENT The authors thank T. Katayama of the Electro Technical Laboratory for sharing his knowledge of magnets and S. Nishi
RECEIVED for review July 22,1986. Accepted November 12, 1986.
AIDS FOR ANALYTICAL CHEMISTS Purge Gas Enhancement of Peak Resolution in Differential Scanning Calorimetry Guang-Way Jang, Ranjana Segal, and Krishnan Rajeshwar* Department of Chemistry, The University of Texas a t Arlington, Arlington, Texas 76019-0065 Overlapping peaks are a nuisance in the analyses by differential scanning calorimetry (DSC) of mixtures and compounds that undergo more than one thermal event. The strategy for enhancing peak resolution in DSC usually is based on a reduction in heating rate or sample mass. This method, however, suffers from the handicap associated with prolonged analysis times and lowered sensitivity, respectively. This paper describes an alternative method based on the use of a thermally conductive purge gas (e.g., He) to enhance peak resolution in DSC. The theory is verified through use of computer simulations and experiments on model compounds and mixtures.
THEORY The equivalent-circuit model previously developed for DSC cells of the heat-flux type (1) is shown in Figure 1. The slope, d(dq/dt)/dTsH, of the leading edge of a DSC peak may be written as d(dq/dt) - 1 dTsH Rs
1 RD'
1 2 1 +-KRD' ++(1) KRG KRG
.
In eq 1 and as before ( I ) , dq/dt is the heat-flow rate, TSH is the sample-holder temperature, Rs is the thermal resistance between sensor and sample, RD' is the thermal resistance of the disk between the sample and reference platforms, RG is the thermal resistance of the gas phase separating the heater block from the sample and reference, RG' is the gas thermal resistance between sample and reference, and K is the thermal lag term containing sample and instrumental contributions. It is important to emphasize that Rs also contains contributions from the thermal resistance of the interfacial region between the sample pan and the sample platform. Narrowing of the peak width and thus better separation of overlapped DSC signals can be achieved by minimizing the magnitudes 0003-2700/87/0359-0684$0 1.50/0
of Rs and the other component resistances (cf. eq 1) through the use of a purge gas with facile heat conduction characteristics.
EXPERIMENTAL PROCEDURES All experiments were performed on a Du Pont Model 1090 thermal analysis system fitted with the Model 910 DSC accessory module. Fusion endotherms were recorded usually after one or two initial "conditioning" heat-cooled cycles through the transition. Commercial samples of In and adipic acid (99.99% purity or better) were used as received. The purge gas (either Ar or He) was flushed through the DSC cell at the rate of ca. 80 mL/min. Sealed A1 sample pans were used in all the cases. Computer simulations of DSC thermograms were carried out on an IBM PC-XT fitted with a Hewlett-Packard Model 9575A graphics plotter. These simulations were facilitated by decomposing the overall thermograms into pretransition, transition, and posttransition regimes. Input parameters comprised the scanning temperature limits, resolution element size, transition enthalpy, transition temperature, and the heat capacities of sample plus container. Multiple transitions were individually simulated prior to summation. Parameters such as sample mass, heating rate, and the component thermal resistances in eq 1 were then systematically varied to assess their influence on DSC peak shapes and resolution. Further details of the simulation protocol will be published elsewhere (2).
RESULTS AND DISCUSSION For initial investigations on purge gas enhancement of DSC peak resolution, we chose the melting transitions of indium (mp 156.6 "C; heat of fusion, AHf = 28.4 J/g) and adipic acid (mp 152.0 "C, A",= 253 J / g ) as model systems. The simulated thermograms in Figure 2 parts a and b, for In and In/adipic acid mixtures, respectively, confirm our expectations (vide supra) on (a) increase of the slope of the leading edge of the DSC endotherm, (b) narrowing of the endotherm, and ( c ) enhancement of peak resolution when the purge gas is (C 1987 American Chemical Society