Anal. Chern. 1982, 54,885-890
885
Instrumental System for Multidimensional Spectroscopic Characterization of a Radio Frequency Boosted, Pulsed Hollow Cathode Lamp Paul B. Farnsworth and John P. Waiters" Department of Chemistry, University of Wisconsin, Madison, Wisconsin 53706
Deslgn crlterla, characterization data, and sample performance data for a computer-controlled, multldlmenslonal spectrometer are presented. A stepper motor driven I-m monochromator Is operated with a 1200 Ilne/mm grating for low resolution work or wlth a 300 Ilne/mm echelle gratlng for hlgh resolutlon work. Wlth the echelle gratlng, wavelength resolution Is sufflclenlt to monitor atomic llne proflles in a hollow cathode discharge. Temporal resolution of 1 ws Is prodded by a gated photon counter. A chromatlc aberratlon compensated lens system produces spatial resolution of 50 pm. These comblned capabllltles make the Instrument a powerful tool for the study of pulsed electrical dlscharges.
Direct current hollow cathode discharges are used extensively a t low currents aa line sources for atomic absorption and atomic fluorescence spectrometry. Pulsed hollow cathode lamps are used le,ss extensively, primarily because the higher pulsed currents cause, in addition to increased sampling and excitation of the cathode material, severe self-reversal of the neutral resonance lines (I). The rf-boosted, pulsed hollow cathode lamp (2) shows promise as a means of producing the high intensities of a pulsed lamp without the attendant reversal of the analytical lines. Development of this lamp has been accompanied by an exploration of some of the more fundamental processes responsible for excitation of the cathode material by the discharge, which has provided some interesting insights into its complexity. Excitation of both the buffer gas and cathode species varies markedly as a function of time, position in the discharge, and the energies of the states involved (2-4). The complex dependence of the lamp's spectral output on space and time called for an instrument with a high degree of temporal, spatial, and spectral resolution. Combined with these requirements were the need for high sensitivity and broad wavelength1 coverage. Finally, the multidimensional nature of the experiments dictated that the instrumental system be capable of acquiring and processing large volumes of data. In this paper we describe the design and performance of a computer-controlled spectrometer constructed to meet the above requirements. Several descriptions of computer-controlled spectrometers have already appeared in the literature (5-8). These have been evaluated primarily in terms of the computer-controUed monochromator's ability to rapidly locate and scan analytical emission lines. Our system is evaluated in the context of a more fundamental experiment with two emphases: first, on the experimental leverage that results from the combination OB spatial, temporal, and spectral resolution in a single instrument and, second, on the factors which control and limit the performance of such instruments and which should be considered in their use for either basic or applied studies.
EXPERIMENTAL SECTION Instrumentation. The instrumental system consists of the elements shown in Figure I and the optical diagram in Figure 2.
The rf boosted hollow cathode lamp has been described in detail elsewhere (2). It is a commercial hollow cathode lamp into which a unidirectional current pulse is injected, followed at a fixed delay by a burst of rf energy at 150 MHz. A Westinghouse WL-23401 single element lamp (Westinghouse Electric Corp., Electronic Tube Division, Elmira, NY) with a copper cathode and neon gas full was used for the experiments described in this paper. The cathode of the lamp is imaged on the entrance slit of the monochromator by a 3-in. diameter, 2Lcm focal length fused silic.2 lens (Oriel Corp. Stamford, CT). An iris, mounted near the lens face, keeps the aperture of the imaging system at or smaller than that of the monochromator. Both the lens and the lamp can be precisely translated along the optical axis to allow for compensation of chromatic aberration over the wavelength range we are studying. The monochromator is a McPherson Model 2051 1-m stigmatic monochromator (GCA Corp., Acton, MA) equipped with a 1200 line/mm grating, which was used in low orders for survey and fixed wavelength work, and a 300 line/mm echelle grating, which was used in high orders for high resolution work. The grating mount has a 20Dextended wavelength position, which allows for grating rotations of up to 70'. The echelle grating is used Without a cross dispersive device. The sine bar of the monochromator is driven by a USM HDM-155-2000-8-HSstepper motor (2000 steps/revolution) (USM Corp., Harmonic Drive Division, Wakefield, MA). The motor is coupled to the lead screw with a gear ratio of 13.75:l. This ratio was chosen such that it would take at least 20 steps of the motor to pass the central image of the entrance slit across the exit slit. The resultant step size for the 1200 line/mm grating in first order is 1.8 X nm. Step sizes for the echelle grating are typically about 2.5 X mm. The motor and detection electronics are controlled by a CBM 2001 series microcomputer (Commodore Business Machines, Palo Alto,CA) through a parallel 1/0interface composed of four 6522 versatile interface adapters (VIA'S) and associated control logic. All of the optical hardware is mounted on a vibrationally isolated rail based optical bed (9). The detection system is a photon counter composed of an RCA 1P28A photomultiplier tube, a PAR Model 1121 preamp and discriminator (Princeton,NJ), a Systron Dormer Model 6151 seven digit counter-timer, an Evans Associates (Berkeley, CA) Model 4145-1 digital delay generator, a Tektronix Model 2101 pulse generator (Beaverton, OR), and an interface to the microcomputer. The photon counter can be operated in three modes: (1) an integrated mode in which the counter gate is held open for a preset interval at each wavelength point, (2) a fixed gate mode in which the gate is held open for a short period during each of a predetermined number of lamp pulses at each wavelength point, and (3) a boxcar mode in which the gate is scanned through a preprogrammed range of delays at a fixed wavelength. In the gated modes the gate delay is controlled by the Evans digital delay generator and its width is controlled by the pulse width setting on the pulse generator. Computer Software. The software to control the monochromator and the detection electronics is composed of a seriev of BASIC and machiie language subroutines all called by a central BASIC program. The monochromator wavelength is controlled by counting the number of steps in a positive or negative direction delivered to the stepper motor. This count is referenced to the zero-order position of the grating. For operation of the instrument with either grating, the wavelength setting is initialized by scanning across a line of known wavelength and then entering the wavelength and the number of steps taken past the line center
0003-2700/82/0354-0885$01.25/0 0 1982 Amerlcan Chemical Society
886
ANALYTICAL CHEMISTRY, VOL. 54, NO. 6, MAY 1982
I I I I I
L_
RF BURST GENERATOR
I
I
___________________
I
J
REFERENCE PULSE
I
r I
1 I I
I I
I
I
I I
I I I I
I I I I
I i I
I
I
I
I
1 Flgure 2. Optical diagram: (1) monochromator, (2) photomultipller tube, (3) plane folding mirror, (4) lens, ( 5 ) stepper motor, (6) hollow-cathode lamp, (7) xyz mount, ( 8 ) 1 In. diameter cross rails, (9) 2.5 in diameter optical rails. Solld lines show component positions for operation at 350 nm; dashed lines for operation at 200 nm.
into the computer. From these values the number of steps required to reach that lead screw position is calculated. Once the system is initialized, the starting and ending wavelengths and the number of points in a scan are entered into the computer via a BASIC subroutine. The wavelengths are converted to values of sin 8, where 0 is the grating angle, using the grating equation for a Czerny-Turner monochromator sin 8 =
mX
2d cos $J
(1)
If the grating is in the extended position, values for 0 are calculated
and 20' are subtracted from them. From the calculated values of sin 0, a beginning and an ending count are then given by dc dx c=-dx d(sin 0) sin where dc/dx is the number of motor steps required to move the lead screw nut a given distance along the length of the lead screw and 8 is the grating angle. The ending count is then adjusted to give the desired number of points with an integral number of motor steps between each point. A machine language routine then slews the monochromator directly to the starting wavelength by
ANALYTICAL CHEMISTRY. VOL. 54, NO. 8, MAY 1982
887
Table 1. Short Range Wavelength Accuracy" peak no. 1
A
(obsd)
A (lit.)
218.181 218.172 218.966 .~~ 218.963 219.227 219.224 219.567 219.573 219.957 219.958 219.975 219.975 220.050 220.053 220.975 220.980 221.025 221.027 221.275 221.274 221.459 221.458 221.515 221.510 221.570 221.565 221.811 221.812 221.859 221.852
I
~~~
c I E N P
0
8 9 10 11 12 13 14 15
I 0 N
a
LENS POIlllON
Fie*. 3. Images oi 50911mesh scrwn producad by lines In @aHg
~~~
~
~
AA
cO.009 +0.002 -0.003 +0.006 -0.001 0.000 +0.003 -0.005 -0.002 fO.001 fO.OO1 +0.005 f0.005 fO.OO1 +0.002
spectrum
Cu I c u I1 cu I1 cu I1 cu I cu I cu I1 CuII Cu I1 c u I1 cu I cu I1 cu I c u I1 cu I1
All waveleneths in nm.
I'
51
spemm: (1) 334.1 nm. (2) 313.2 and 312.6 nm. (3) 302.1 nm. (4) 296.7 nm. (5) 289.4 nm, (8) 275.3 nm. (7) 285.2 nm,(8) 248.2 nm. (9) 237.8 nm. (Axes marked in 1 cm intervals.)
running the stepper motor at ita maximum rate of 4000 stepsls. Software generated frequency ramps accelerate and decelerate the motor to minimize the probability of step IOSS. Once the starting wavelength has been reached, the photon counter is activated and the stepper motor is held inactive until an interrupt signals the end of the counting period. This period is determined in the integrated mode by a hardware gate in the Systron Donner counter. In the gated mode it is determined by the gate count registered by a 16-bit counter in one of the 6522 VIA'S. At the end of the counting interval, the counter is dieabled and the monochromator is stepped to the next point. While it is being stepped, the count is read from the counter, converted from BCD format to floating point, and displayed on a Nicolet Explorer I11 digital oscilloscope. The next counting interval is then started, during which the count from the previous interval is stored on a floppy disk for later acws. This cycle is repeated until the desired wavelength range has been covered. Once an experiment is completed, the data can be retrieved from the disk to be displayed on the digital oscilloscope, further processed by the microcomputer, or transferred via an intelligent terminal package (McTerm, Madison Computer Store, Madison, WI) and acoustic modem to the university's computer center for plotting or statistical analysis. Wavelength Calibration. The groove spacing for the 1200 liie/mm grating was determined by slowly scanning the monochromator from 200 to 271 nm and recordii the number of steps required to reach the peaks of 109 lines in the Cu I and Cu I1 spectra. T h a data were then fit to a line using a hear regreasion. From the slope of the line, a lump constant dc dx dx d(sin 6')
1 ~ ~ ( C O@) S
was calculated and used in the computer software an described above. The high angles required for operation of the echelle grating are obtained by rotating the grating 20° with respect to the sine bar lever arm. A consequence of this is that the relationship between lead screw revolutions and wavelength is no longer linear. Because of this nonlinear relationship, no attempt was made to calibrate the echelle grating directly. Instead, calculated values of de/dx and d and measured values of dx/d(sin 6') and @were used in eq 1 and 2 to position the grating. This resulted in poorer long range acnuacy with the echelle than with the 12M)line/mm grating, as is reflected in the systemic error in Table 111. Lens Positioning. Chromatic aberration is a serious problem when a lens is used as a high fidelity imaging optic, particularly a t short wavelengths. This problem was eliminated by com-
U R V E L E N G T H I NM I
Fi&we4. First& specbun d a C u hdlow.catha!.s lamp takm wim the 1200 linelmm grating. The cwent was 10 mA dc.
pensating the lens system for chromatic aberration over the wavelength range to be studied. The focal length of the lens was calculated for nine linea of the mercury spectnun with wavelengths ranging from 237.8 to 334.1 nm. At each of these wavelengths the lens was placed a distance of Zf from the entrance slit of the monochromator. With a 1-mm wide entrance slit the monochromator was set in first order to the Wavelength of the line to he focused. A series of screen images were then recorded on a plate in the monochromatorfocal plane, with the screen translated along the optical axis in 0.05-in. increments. The plates were examined under a micnxupe,and the position of the best foeused image was noted. Because the monochromator itaelf had been eharaeterizedand found to be aberration free (IO),any degradation of the screen image could be attributed to the lens imaging system. Figure 3 is a plot of the focusing results. The straight line indicates that a constant relationship between image and object distances is maintained and that the magnification does not change with wavelength. The sharp screen images registered in exposures of up to 10 min illustrate the vibrational stability of the optical bed. The open squares on the plot mark the lens and screen positions for Cu I1 lines that were used to check the calibration at wavelengths below 237.8 nm. The lens and screen were positioned by using the calibration data, and a single photograph was taken. The images were faint because of poor plate aensitivity below 250 nm hut were intense enough to confirm that the lens and screen settings were correct.
RESULTS A N D DISCUSSION Survey Performance. A representative scan of 4 nm of the spectrum of a copper hollow-cathode lamp is shown in Figure 4. Table I gives the identification of the lines along with the difference between the observed wavelengths and literature values (11). The observed values were obtained by locating the line center on the digital oscilloscope and then
ANALYTICAL CHEMISTRY, VOL. 54, NO. 6, MAY 1982
888
TIME (HOURS)
XI 0 3
LEROSCREI..
REVOLUTIDUS
Figure 5. Residuals of the least-squares fit of the grating calibration
data plotted as a function of lead screw rotation.
07
Table 11. Long Range Wavelength Accuracya order 1 1 1 1 1 1 1
2 2 1 1 1
2 2
3 a
h
(obsd)
203.594 203.718 204.363 327.989 329.297 569.950 578.140 327.955 329.261 203.589 203.713 204.380 327.949 329.259 328.102
h
(lit.)
203.585 203.713 204.380 327.982 329.283 570.024 578.213 327.982 329.283 203.585 203.713 204.380 327.982 329.283 327.982
z
Ah
spectrum
-0.009 -0.005 t0.017 -0.007
CU I1 CU IT Cu I1 CUI CU I Cu I Cu I Cu I CU I CU I1 c u I1 c u I1 Cu I Cu I CU I
-0.014
t0.074 +0.073 +0.027 -0.021 -0.004 0.000 0.000
t0.033 t0.024 -0.120
All wavelengths in nm.
using the scan parameters stored in the computer to calculate the wavelength. The 219.975-nm line was used as a calibration reference. The precision shown in Table I is comparable to that attainable with a photographic plate. This type of scan can in many cases serve as direct substitute for photographic surveys made with a plate and densitometer. Precision in this case is not limited by uncertainty in locating the line center but by nonidealities in the lead screw mechanism. That this is the case is illustrated in Figure 5, which shows the residuals of the least-squares fit of the grating calibration data plotted as a function of lead screw revolutions. Instead of varying randomly, the residuals change periodically with the rotation of the lead screw. The long-range accuracy of the system is indicated by the data in Table 11. The wavelength calibration was initiated on the 249.215-nm line, and the lines were scanned in the order that they appear on the table. As is the case with precision, the accuracy is limited by nonidealities in the sine bar mechanism. The large deviations at high wavelengths are probably due to misalignment of the lead screw or to a slight missetting of the zero-order grating angle. Step loss in the drive mechanism is ruled out by the fact that the erros are noncumulative and are reproducible as the scanning process is repeated. The absence of step loss was confirmed by monitoring the position of a mark on the stepper motor drive gear as multiples of 2000 steps were delivered to the motor driver. Under both normal stepping and slewing conditions the mark always returned to its original position. Time Resolution. The gated photon counting system provides time resolution of about 1ps, which is sufficient to monitor most temporal changes in the rf boosted hollow cathode lamp. This limit is imposed by the step size of the
-
4
0.
2.
4.
5.
8.
IO.
TIME [HOURS)
Flgure 6. Long term stability of the instrumental system: (upper trace) entrance and exist slits both set at 10 pm; (lower trace) 10 pm entrance slit, 100 pm exit slit.
digital delay generator controlling the gate, although other factors discourage attempts to reduce it. Most important of these is the fact that even at a count rate of 1 MHz, above which pulse pileup becomes a problem, an average of only one count will be registered during a 1 p s gate. If the gate is narrowed further, the time required to accumulate an acceptable number of counts becomes unmanageably long. Even with gate times of 2 ws, an experiment to map the intensity of an emission line as a function of both space and time may take several hours. This requires that the entire system be stable over that time period. Such intensity maps were generated by scanning the monochromator across the desired line and then repeating the scan, stopping at the line center. A series of time scans were then made with the lamp displaced normal to the optical axis in 0.13 mm increments. To check the system stability, the monochromator was stopped on the Ne 1344.8-mm line, and the intensity was monitored over an 11-h period. The entrance and exit slits of the monochromator were both 10 pm wide. The results are shown in the top half of Figure 6. The large intensity fluctuations correspond to the air conditioner cycle in the laboratory. These changes in intensity correspond to changes in temperature of about 1 "C. The source of the fluctuations was isolated by repeating the experiment with the exit slit opened to 100 pm. The resultant stability indicates that the drift was almost entirely mechanical and caused by temperature-induced changes in the monochromator. The problem was solved for our fiied wavelength experiments by simply opening the exit slits after first checking to ensure that doing so would not cause interference from neighboring lines. The problem is much more serious for sequential scanning direct readers, which deal with more complex spectra and which require small, equal slit widths for adequate spectral resolution. The problem of mechanical drift also most be considered with scanned experiments. It dictates that the time required to scan across a spectral feature must be short compared to the period of temperature fluctuations in the laboratory. Failure to meet this criterion could result in distortion of the feature being studied. The combination of time resolution with the spatial resolution afforded by high fidelity imaging optics does reveal features that would be completely missed if the spectrum were
-
ANALYTICAL CHEMISTRY, VOL. 54, NO. 6. MAY 1982 889 Table 111. Line Identification with EcheIle Grating' 1st order A (obsd)
5902.80 5903.10 5903.42 5907.12 5910.16 5910.86 5911.41 5913.23 5913.66 5914.36 5915.77 5917.60 5918.48 5921.68 5923.73 5924.04 5926.13 5926.29
1st order
line
order
N e 1 347.26 cu I 327.98 cu II 203.56
17 18 29 29
Cu I1 203.71 Ne 11 369.42 Ne I1 295.57 Cu I1 218.96 Cu I1 211.21 Cu I1 227.47 Cu I1 257.18 Cu I1 268.93 Cu I1 227.63 Cu I1 219.23 Cu I 296.12 Cu I1 246.85 cu 11 236.99 Cu I1 237.08 Cu I1 204.38
16
20 27 28 26 23 22 26 27 20 24 25 25 29
A
(calcd)
5903.37 5903.66 5903.97 5907.68 5910.72 5911.46 5912.00 5913.88 5914.32 5915.05 5916.46 5918.28 5919.13 5922.34 5924.40 5924.75 5926.88 5927.02
ah
-0.57 -0.56 -0.55 -0.56 -0.56 -0.60 -0.59 -0.65 -0.66 -0.69 -0.69 -0.68
-0.65 -0.66
-0.67 -0.71 -0.75 -0.73
Ail wavelengths in nm.
Figure 7. Intensity of the Cu I 239.3-nm lina plotted as a functlon of
II
~
position and time. integrated in either of these dimensions. Figure 7 shows the intensity of the Cu 1239.3-nm line from the rf boosted hollow cathode lamp plotted as a function of space and of radial position in the cathode bore. This profile was produced by a 4oo-mA current pulse, 10 pa in duration, followed 75 pa later by a 10 ps rf burst of approximately 360 V. Results from ouf spectroscopic investigations of the lamp will be described in a separate paper (4). System Sensitivity. A potential problem associated with the use of a high degree of spatial and temporal resolution arises from the fact that only a small fraction of the light generated by the light source is actually detected. This combined with the inherently digital nature of photon counting dictated our choice of photon counting over analog detection. Even with the inexpensive photomultiplier tube and a small entrance slit into the monochromator (10 X 500 pm) the sensitivity was more than adequate. In the boxcar mode it was necessary to stop down the lens to avoid piling up the counter on all hut the weakest lines. Even with the lens stopped to its minimum aperture (about 5 mm), the Cu I 324.8-nm line generated count rates far beyond the linear range of the photon counter. In the scanned modes the high sensitivity of the system worked to our advantage in producing clean spectra with high signal to noise ratios. High Wavelength Resolution. Measurement of the line profiles in the rf-boosted hollow cathode lamp required instrumentation capable of resolving spectral features 0.001 n m wide with simultaneous spatial and temporal resolution. DeJong and Piepmeu ( 1 ) have used an interferometric system to make such measurements. An alternative approach is to make use of the high dispersion of an echelle monochromator. A computer-controlled scanning echelle monochromator with a CK)(YJ dispersive prism has been used succeasfdly to separate c l w l y spaced transitions in complex atomic spectra (12). This type of system was unsuitable for our purposes because the insertion of the cross dispersive device into the optical train degrades both the line quality and the spatial fidelity of the system. Consequently, we used the McPherson as an echelle monochromator and sacrificed spectral simplicity for the sake of optical quality. The 300 line/mm grating used for these studies is blazed at 63'. At this angle 20 orders of the hollow cathode spectrum
10
3
6
t
7
m
6-
c
z Y
c
z
54-
3-
2I250
WRVELENGTH 1 N n I
Figure 8. Profile of the Cu I 324.8-nm line recorded with the 300 linelmm grating In 18th order.
are sensed by the photomultipler tube. Table I11 lists the lines registered in a narrow spectral region near the Cu I1 204.3-nm line in 29th order. Despite the complexity introduced hy the overlapping orders, positive identification of almost all lines is possible because of the high degree of precision with which peaks can be located. The use of the echelle grating without a cross dispersive device is practical only on light sources that produce narrow line spectra with low backgrounds. Broad features would overlap in the focal plane and the contributions from different orders would not be separable. The performance capabilities of the system with the echelle grating are illustrated in Figure 8, which is a scan of the Cu 1324.8-nm line from the hollow cathode lamp operating a t 5 mA dc. The scan was taken in 18th order with 10 pm slits. The line is separated into two components by the nuclear spin splitting of the ground state (13). The half-widths (0.0012 nm) and the separation of the peaks (0.0039 nm) agree well with the values measured interferometrically by DeJong and Piepmeir ( I ) on a copper hollow cathode lamp with similar operating conditions. The base line resolution of the two peaks
ANALYTICAL CHEMISTRY, VOL. 54, NO. 6, MAY 1982
890
EDGE OF C A T H O D E
limitation of this type of echelle system. The grating is always operated near its blaze angle, so the value of sin I9 in eq 1varies by only a few percent. Substituting n = W / d into eq 1,where W is the ruled width of the grating, and rearranging yields
mXn=
v)
2
757
Z
'
-
50-
4
25
~
0 ,
.oo
, , , ,
,
.01
, , , ,
, ,', .02
I
; \ ; ; ', , , , , ,I
.03
I
.01
,
INN
.06
1507
j
125
x
(5)
Since variations in I9 are small and d, is covalent, the resolution of the grating is proportional to its width and inversely proportional to its wavelength. I t is independent of groove density. At wavelengths above 350 nm, the grating limits the resolution of this system, a situation that could only be improved by increasing the gratings ruled width. The effectiveness of the echelle system in performing the task for which it was assembled is illustrated by Figure 9. The solid lines are scans which were taken during the peak intensity produced by a 10 ps,400 mA current pulse (10 ps gate width). The dashed lines are of the emission produced during the 10-ps rf burst, delayed from the current pulse by 75 ps. These data are illustrative of the type of profile information that can only be gathered by a combination of spatial and temporal resolution.
C E N T E R OF C A T H O D E
XI ' 0
2 W sin 8 cos 4
25-
ACKNOWLEDGMENT We are grateful to Ted Weigt for his assistance in the design of the computer interface electronics. - - - - - - - - - - . RF
LITERATURE CITED
BURST
CURRENT PULSE
Figure 9. Profiles of the Cu I 324.8-nm line from the rf-boosted lamp. The upper traces originate from a region 0.46 mm from the edge of the cathode and the lower traces from the center of the cathode bore.
is somewhat poorer than that recorded on the interferometer. The reason for this is that the above listed scan conditions produce a resolution figure for the monochromator that approaches the theoretical resolution limit of the grating. The reciprocal linear dispersion of the monochromator at 324.8 nm in 18th order is 0.07 nm/mm, which combined with 10-pm slits gives a spectral slit width of 0.0007 nm. The instrument resolution is then given by
A/
ax
(3)
(1) DeJong, G. J.; Piepmeir, E. H. Spectrochim. Acta, Part 8 1974, 298, 159. (2) Araki, T.; Walters, J. P.; Minami, S. Appl. Spectrosc. 1980, 3 4 , 33. (3) Farnsworth, P. B. PhD Thesis, University of Wisconsin, 1981. (4) Farnsworth, P. B.; Walters, J. P., submitted for publication in Spectfochim. Acta. (5) Floyd, M. A.; Fassel, V. A.; Winge, R. K.; Katzenberger, J. M.; D'Siiva, A. P. Anal. Chem. 1980, 52,431. (6) Spiiiman, R. W.; Malmstadt, H. V. Anal. Chem. 1976, 48, 303. (7) Johnson, D. J.; Plankey, F. W.; Winefordner, J. D. Anal. Chem. 1975, 4 7 , 1739. (8) Kawaguchi, H.; Okada, M.; Ito, T.; Muzuike, A. Anal. Chim. Acta 1977, 95, 145. (9) Coleman, D. M.; Waiters, J. P. Spectrochim. Acta, Part 8 1978, 338, 127. (IO) Kiueppel, R. J.; Coleman, D. M.; Eaton, W. S.; Goidstein, S. A., Sacks, R. D.; Waiters, J. P. Spectfochim. Acfa, Part 8 1978, 338, 1. (11) Striganov, A. R.; Jventitskii, N. S. "Tables of Spectral Lines of Neutral and Ionized Atoms"; Plenum: New York, 1968. (12) Anderson, D. L.; Forster, A. R.; Parsons, M. L. Anal. Chem. i981, 5 3 , 770. (13) Brlx, P.; Humbach, W. 2.Phys. 1950, 128, 506.
and is 463000. The resolution limit of the grating is given by mXn (4)
RECEIVED for review September 24, 1981. Accepted January
where m is the order and n is the number of rulings on the grating face. For the 102 mm ruled width of the echelle grating this value is 550 800. These values point to a fundamental
28,1982. Portions of this work were funded by the National Science Foundation under Grant CHE79-15195 and the Graduate School of the University of Wisconsin.