JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY JUNE 1994 VOL. 9 697 Pulse Optimization Criteria for the Microcavity Hollow Cathode Discharge Emission Source Paul D. Mixon and Steven T. Griffin Department of Electrical Engineering Memphis State University Memphis TN 38 152 USA J. C. Williams Jr. Department of Anatomy Indiana University Medical Center Indianapolis IN 46202-5 120 USA Xiangjun J. Cai and J. C. Williams* Department of Chemistry Memphis State University Memphis TN 381 52 USA Criteria are presented that are important for optimizing the emission intensity in a microcavity hollow cathode (MCHC) discharge source operated in the pulsed mode. The d.c. component must be sufficiently large to ensure a continuous discharge inside the HC. It is shown that the pulse parameters that result in maximum light output for the spectral lines of the cathode material do not result in maximum light output from sodium as an analytical sample which has been mounted inside the HC.This lack of correlation may be due to either the physical and chemical properties of the species or to the manner in which the species enters the HC discharge plasma. Voltage-current characteristics are presented for the MCHC to illustrate the role of the d.c. component during pulsed mode operation. The d.c. current of the discharge must be >5 mA in order to maintain a discharge in the HC. At currents <5 mA for this system the discharge is restricted to the planar surface of the cathode and will not participate in the excitation of analytes thrown into the hollow space during the pulse.Keywords Glow discharge; hollow cathode discharge; atomic emission spectrometry The hollow cathode discharge (HCD) has been used extensively as a line source for atomic absorption spectrometry (AAS) atomic fluorescence spectroscopy (AFS) and as a radiation source for atomic emission spectrometry (AES). Djulgerova' has reviewed the application of the pulsed HCD to AAS AFS and AES. The HCD emission intensity is reported to be increased over that achieved in the d.c. mode by up to two orders of magnitude and instability and reproducibility prob- lems encountered in the analysis of dry residues from liquid samples are greatly reduced by operating the source in the pulse The idea of the pulsed HCD is to use short pulses (a few p) of high current (a few hundred mA) to spatially separate the vaporization and excitation.A few ps of diffusion time are needed for the analyte atoms in the HC to reach the negative The discharge current is reduced to several mA after the sputtered atoms have reached the negative glow to avoid possible discharge instability and self-reversal.6 Most of the work in this area has been directed toward improving the HCD as a source for AAS and AFS where narrow lines are an absolute requirement. Although line widths are less critical for AES sufficient self-absorption will affect the linearity of calibration curves. The Russian group,4 as reviewed by Djulgerova,' has investigated the analysis of dry residues in the HCD using pulses 0.02-2 A 5-1000 ps wide at frequencies of 0.02-10 kHz that were superimposed on a d.c.current of a few mA. The combined mode of d.c. and pulse discharge gave 10-100 times better detection limits than the d.c. discharge a10ne.~ Normally the high-current pulse is superimposed on a d.c. or pilot current that maintains a continuous discharge inside the HC. There are several parameters associated with pulsed mode operation that directly affect emission intensity. These include pulse amplitude Ip pulse width t pulse frequency f and the pilot (d.c.) component Id.c.. A widely used optimization technique for AAS and AFS applications maximizes the inten- sity gain ratio I / l o as a function of these four parameter^.^ In this technique I is the intensity of the spectral lines of the cathode material under pulsed excitation and lo is the intensity for the case of d.c.excitation at the same average value of current. The four pulse parameters (Ip Id,c. t andf) are selected and the emission intensity for these pulses is recorded over a specific period of time. The d.c. current is set to the average value of the pulsed excitation. Then the intensity is recorded over the same period of time for the case of d.c. excitation. The intensity gain ratio l / I o is calculated from the results. The pulse parameters are adjusted and the procedure repeated until an optimum is reached. Using the above procedure intensity gains of 50-800 have been reported for Ca Co Cu Pb Mg and Mn by applying pulsed mode operation to commercial lamps in absorption ~tudies.~ Djulgerova has reported intensity gains of 55 for Al,*?' 400 for Cu and 100 for Fe.l0 A number of elements were studied by Cordos and Malmstadt," who reported intensity gains of 40-200 when using pulsed excitation for AFS.An empirical formula for the ratio I / I o has been developed by Katskov et a1.I2 Good agreement was reported between pre- dicted and measured values of gain up to 70." Intensity gains over the d.c. mode of an order of magnitude have been reported in studies using a hot HCD source in pulsed mode.13 The I / I optimization technique has been used primarily to maximize the emission signal from HC lamps that are to be used in absorption and fluorescence techniques. It has been suggested that optimizing pulse parameters for maximum intensity gain ( I / I o ) may not yield the best conditions for analytical performance in the HCD as an emission s o ~ r c e .~ It has also been noted that physical or chemical properties of the analyte will cause unique optima for some ana1ytes.l' This is particularly the case when using the HCD as an emission source for AES where the analytical sample is mounted directly inside the HC. For some analytes such as the alkali metals the pulse parameters that result in maximum light output for the spectral lines of the cathode material may not maximize light output from the analytical sample. Provided that precision and sensitivity are maintained the maximum absolute emission intensity is desired in AES. This may or may not correspond to maximizing I/I@ Some aspects and character- istics of a particular HCD system can be studied by observing the emission intensity of the hollow material of the HC itself.This is much easier than observing the transient signals from samples deposited in the hollow space. However optimum pulse parameters must at least be verified for each analyte. The main thrust of the research in this laboratory is to698 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY JUNE 1994 VOL. 9 develop a suitable method for the determination of several elements (Nay C1 K Cay Mg and P) of physiological interest in 1 nl of renal fluid. Suitable detection limits for most of these elements have been demonstrated here using the HCD operated in the d.c. rnode.l4 However the improved stability and precision promised by pulsed-mode operation is very attractive. Thus the behaviour of Na in a pulsed discharge is of primary interest and must be investigated in spite of difficulties caused by environmental contamination. Fortunately highly con- trolled sample handling techniques developed for handling small samples of ubiquitous elements allows competent deter- mination of such elements.15 Even though the concentration of the components of renal fluid is relatively high the nanolitre volumes usually available make the determinations very challenging.Experimental The instrumentation used in the pulsed HCD atomic emission experiments is shown schematically in Fig. 1. Zero grade argon was used in the experimental work described here. The gas pressure was monitored using a Veeco model TG-270 pressure gauge. The demountable HCD source used in these experi- ments has been described in detail pre~ious1y.l~~~"~~ Br iefly a water-cooled HC emission source was mounted on an optical rail.The light from the source was collimated by a lens and then directed along the optical rail to a neutral beam splitter that divided the light between two channels. Each channel had a monochromator one at the end of the main rail and the other mounted at a right angle to the main rail. A lens before each monochromator was used to focus an image of the HC on the slit. Each monochromator employed a Hamamatsu R928 photomultiplier tube (PMT) mounted just outside the exit slit. Appropriate wavelength settings were selected and data were collected in two channels simultaneously. The output of each of the PMTs was fed into a multichannel integrator box before being sent to an analogue-to-digital converter for processing by a computer.The integrator box included six integrator cards (Evans 41 30A) and three dual amplifier cards (Evans 4163A) under computer control through Metrabyte PIO-12 and Dash-16 boards. The electronic circuits used to drive the HCD in pulsed mode and d.c. mode have been described in detail previo~sly.l~*~~ For the work described here stainless-steel (type 304) HCs having dimensions 1.5 mm diameter by 5 mm depth were used. These dimensions were found to maximize the emission signal during previous studies using A1 and Cu e l e c t r o d e ~ . ~ ~ ' ~ ~ After sample deposition the cathode was dried under an infrared lamp or in a vacuum chamber heated to ~ 5 9 5 "C placed in the HC source chamber and discharged.In the case of microsamples the emission signal is transient. It builds up HCD To second channel enclosure PMT - I Monochromator Power supply I electronics Trigger I Multichannel gated integrator/amplifier Digitizing oscilloscope Fig. 1 intensity from the HCD Block diagram of the instrumentation used to observe emission rapidly and decreases quickly as the small sample dissipates. Most of the sample is sputtered from the hollow space in a few seconds producing the analytical signal. The remainder of the discharge period is used to clean the electrode for use with the next sample. Lapsed time between samples is 3-5 min. The argon passing to the HCD source was filtered to remove particulates and samples were prepared in a 'clean bench' by NUAIRE Model NU-201-S24 to reduce contamination.20 A 1000 point histogram of this transient emission signal was made by integrating the emission signal for 6-1 5 ms.The pulse width of 15-30 ps was much shorter than the integration periods used. Thus the emission signal collected during the integration period consisted of the signal from 30 to 75 pulses plus that from the d.c. pilot current during that time. Plastic [either polypropylene poly(tetrafluoroethy1ene) or polyethylene] containers were employed for the preparation and storage of all test solutions as recommended by Moody and Lindstrom.'" In order to minimize sample contamination the sample electrodes were thoroughly cleaned by HCD sput- tering handled only with stainless-steel tweezers and trans- ported in an enclosed container.Calibrated nanolitre-range pipettes are not available com- mercially and reproducible deposition of such small samples without contamination is difficult. Pipettes were fabricated according to the accepted practice developed by Bonventre et aZ.22 Williams and S ~ h a f e r ~ ~ give a more recent detailed description of the preparation and use of pipettes for handling nanolitre volumes. Because the volumes used in the electron probe method are less than 100~1 the technique used for handling samples is clearly adequate for the HCD with the present renal fluid samples. For delivering the sample to HC electrodes volumetric pipettes calibrated for 1- 10 nl were used. The pipettes weire fashioned on a microforge using capillary glass and consist of a chamber at the end of the capillary that lies between an opening in the drawn-out tip and a constriction a few hundred inicrometres from the tip. The opening at the tip and the narrowest part of the constriction are both about 20 pm in diameter.The pipette is silanized to reduce wetting of the glass surface by aqueous solutions and improve handling. The accepted procedure23 has been adapted to the HCD and successfully used in this laboratory for several years.14,16-18,24 Samples are delivered to the HC which is mounted on the stage of a low-power stereomicroscope using a micromanipulator upon which a calibrated glass micropip- ette is mounted. The pipette tip is bent at nearly 90" so that the sample can be deposited on the bottom of the HC.The sample is drawn up by syringe suction until its meniscus reaches the centre of the constriction and is immediately deposited on the HC bottom. In the recommended procedure for microsamples a small volume of oil is pulled into the tip after the sample to prevent evaporation following exposure of the pipette to air. However oil has not been used here as the small sample taken from a bulk standard solution is easily transferred from the silanized pipette tip to the more hydro- philic metal surface without loss of analyte. The volume of each pipette was determined by the use of a calibrated solution of a radioactive solute. Three to five samples of this solution were taken with each pipette and ejected into liquid scintillation counting fluid and counted to determine pipette volume.The coefficient of variation of the measure- ments on a pipette had to be less than 1% for a pipette to be considered acceptable for use. The measurement of the volume delivered by a pipette is not crucial to the technique as samples and standards were handled with the same pipette for a given experiment. The precision of sample handling is far more important and the calibration method provides a way to select only those pipettes that handle samples in a reproducible manner. Results and Discussion The results of intensity gain ( I / I o ) measurements for two major constituents of a 304 stainless-steel HC (Fe at 371.99 nm andJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY JUNE 1994 VOL. 9 699 Ni at 341.48 nm) are shown in Fig. 2 The gain was measured for values of peak pulse current (I,) at 50 80 and 120 mA.The other pulse parameters were held constant at f= 5.1 kHz t = 15 ps and I,,.=6 mA. The pulse parameters used for these measurements do not represent an attempt at optimization; however the parameter values used are within the range of previously reported optimum val~es.~-ll The argon pressure was held constant at 5 Torr (1 Torr = 133.322 Pa). The results in Fig. 2 show that I / I increased monotonically for each constituent as the current was increased from 50 to 80 to 120 mA. Fig. 3 shows normalized emission intensities from Li (670.8 nm) Na (589.6 nm) K (766.5 nm) Fe (371.99 nm) and Ni (341.48 nm) as a function of current. A solution containing 100 pg each of Na Li and K in 2.5 nl was prepared from doubly de-ionized water (Millipore de-ionized H20 18 Mi2 cm resistivity) and Spex plasma emission standards. The data were normalized after subtracting the water blank.A 2.5 nl pipette was used to deposit an aliquot of solution on the bottom of the HC. The stainless-steel cathodes were conditioned in three steps using 6 kHz pulsed discharges of 100 mA pulses that were superimposed on a 7mA d.c. pilot current. First a discharge for 2.5 h with 60 ps pulses (47 mA average current) was used to produce a nearly spherical shape in the HC. Next the HC was conditioned for 1.5 h using a discharge of 30 ps pulses (25 mA average current). Finally the HC was discharged for 10 min using the exact parameters (15 ys pulses in this case) planned for the analysis.After sample deposition the cathodes were placed under vacuum for 2min in the HCD enclosure before discharging. The emission intensity from the alkali metals is relatively constant compared with that from 4.0 3.6 3.2 2.8 2.4 2.0' ' I I I I I I I I 50 60 70 80 90 100 110 120 Peak pulse current/mA Fig.2 Emission intensity ratio for A Fe (371.99 nm) and B Ni (341.48 nm) as a function of discharge current. The stainless-steel HC used was 1.5mm in diameter by 5mm deep. The RSD for three replicates was less than 1% for all points shown. Pulse parameters f 5.1 kHz; t 15 ps; and Io 6 mA 1 .o > 0.9 .I- .- 2 0.8 .'= 0.7 u .- 2 0.6 E 0.5 2 0.4 0.3 .I- - m I 40 60 80 100 120 140 160 180 200 Pulse current/mA Fig. 3 Normalized emission intensities from a 1.5 x 5 mm stainless- steel HC for A Fe (371.99 nm); €3 Ni (341.48 nm); C Na (589.6 nm); D Li (670.8 nm); and E K (769.8 nm).Pulse parameters:f=6.0 kHz; t 15 ps; and Id.c. 7 mA. The emission signals come from either solution residue deposited on the bottom of the HC (alkali metals) or the stainless-steel HC itself (Fe and Ni). The RSD for three replicates was less than 1 % for all points shown the transition metals of the cathode itself. The wider range of emission intensities from Fe and Ni tend to compress the alkali data on the plot thus minimizing the peak. However there is a distinct peak for each alkali metal emission intensity. This is consistent with previously reported results for Na Li K and Ca in which all showed an intensity maximum at a d.c. discharge current of 75 mA.I4 The emission intensity from the alkali metals shows little correlation with that from cathode components.This lack of correlation may be due to fundamental differences between alkali metals and transition metals. It could also be due in part to the manner in which the species of interest enters the HCD plasma. In both cases the sample is sputtered into the plasma. However sputtering rates may differ at different locations in the hollow space and sputtering rates for samples deposited on the HC may differ from the sputtering rates of the cathode material that is present uniformly in the hollow space. Thus excitation conditions sputtering rates vaporization temperatures and other parameters will differ for the two cases. Voltage-Current Characteristic for the Microcavity Hollow Cathode The term microcavity hollow cathode (MCHC) is used to describe electrodes having hollow diameters of 2 mm or less.25 The discharge parameters used for analysis in this laboratory place the MCHC discharge well to the right of the minimum point on the Paschen curve.The result is that the discharge initially breaks down over the shortest possible path. When performing analysis with the MCHC an insulating quartz disc is placed over the mouth of the HC. This quartz disc has an opening which is larger than the diameter of the hollow space. As a result there is a front planar surface present which results in a 'lip' effect. The presence of this front planar surface has been found to affect the evolution of the discharge.26 It is important to determine the effect of this front planar surface in terms of the evolution from planar discharge mode to HCD mode in the MCHC.An experiment was conducted to measure the I/-i characteristic for the MCHC over the range of current from 1 to 20mA. Fig. 4 shows a two-piece MCHC which was constructed of two different materials. The top section was made of 304 stainless steel which is identical to the cathodes used for the data shown in Figs. 2 and 3. The bottom section was composed of spectrally pure aluminium (99.9995%) and was designed to snap into the top section to form a one-piece electrode. The dimensions of the hollow were 1.5mm in diameter by 5mm deep. Fig. 5 shows the results of measuring the run voltage and emission intensity versus discharge (d.c.) current for the two- piece cathode.The solid line represents the voltage across the discharge and the broken line represents the emission intensity 6.35 mm 1.5 mm T 1 T x m Fig. 4 Drawing of a two-part HC. The bottom is A1 while the upper part is 304 stainless steel700 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY JUNE 1994 VOL. 9 330 t A 320 310 300 290 - - 2 - 4- - 0 - 3 (r 4 9000 2 6 - 7000 u) 0 - 5000 .z 3000 - B b) c 280 0 2 4 6 8 10 12 14 16 18 20 Current/mA Fig. 5 Breakdown voltage as a function of discharge current for the two-part cathode shown in Fig.4. The solid line represents voltage across the discharge and the broken line represents the emission intensity of the Al 396.2 nm line of the A1 line at 396.2 nm. As is evident from the figure at low values of current (0-4 mA) both the voltage and the emission signal showed a steady increase with increasing current.The abrupt drop in run voltage that occurs at about 5 mA marks the point at which the discharge transfers from the planar face of the cathode down into the hollow. This is evidenced by the sharp rise in the emission intensity of the A1 line. Similar behaviour has been noted by White27 in a study of spherical cavity hollow cathodes. The sharp decrease in voltage is due to a combination of two separate effects. The increase in cathode surface area that occurs when the discharge transfers to the inside of the hollow results in a decrease in the required discharge voltage. In addition the on-set of the HC effect contributes to a decrease in discharge voltage.28 It is interesting to note that the discharge in this MCHC exhibits a slight but well-defined negative resistance characteristic over most of the range of current from 6 to 20 mA. The V-i characteristic for the MCHC yields valuable infor- mation with regard to the optimization of pulse parameters particularly the pilot current (Id.c.). The primary purpose of pulsed excitation is to separate temporally the processes of atomization and excitation in the discharge.When the pulse parameters are set correctly the relatively large pulse current does the job of atomization by sputtering the proper amount of atoms into the hollow space. The appearance of a large population of ions results in an increased sputtering rate. When the atomic cloud in the hollow space reaches the optimum density the high current is switched off and excitation continues as the pilot current excites the cloud.Thus the function of the pilot current is for both sample excitation and ionization maintenance. From Fig. 5 it can be seen that the sharp decrease in voltage across the discharge is accompanied by the on-set of emission from the aluminium bottom of the hollow. Apparently the pilot current for this particular MCHC geometry must be greater than 5 mA for the discharge to enter the hollow space. For values of pilot current less than approxi- mately 5 mA the discharge is planar and is restricted to the uninsulated flat cathode surface around the mouth of the hollow. If the d.c. component does indeed contribute signifi- cantly to analyte excitation Id.c must be sufficiently large to ensure a continuous discharge within the hollow space.Conclusion Analytically optimal pulse parameters for the alkali metals are clearly different from those that maximize light output when using the cathode metal as the analytical sample. The HCD plasma is known to interact non-homogeneously with the HC s ~ r f a c e . ~ ~ * ~ * Thus an analyte deposited entirely on the bottom of the HC may be subjected to very different sputtering conditions from the remainder of the HC surface. The alkali metals have a much lower ionization potential than Fe and Ni and also differ markedly from these constituents in terms of melting-point thermal conductivity and other factors which influence sputtering rates. A broader study is needed to deter- mine the optimum pulse parameters for analysis of samples deposited from solutions in the MCHC and to determine if the effect seen for alkali metals is generally observed.The d.c. current must be > 5 mA for this system in order for a discharge to be maintained in the hollow space between pulses. At currents < 5 mA the discharge is restricted to the planar surface of the cathode and will not participate in excitation of analyte that was deposited on the bottom of the hollow. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 References Djulgerova R. B. in Improved Hollow Cathode Lamps for Atomic Spectroscopy ed. Caroli S. Ellis Horwood Chichester 1985 pp. 52-73. Borkowska-Burnecka J. and Zyrnicki W. Spectrosc. Lett. 1987 20 795. Maksimov D.E. and Rudnevskii N. K. Zh. Prikl. Spektrosk. 1983 39 5. Rudnevskii N. K. Pichugin N. G. and Maksimov D. E. Zh. Prikl. Spektrosk 1975 25 921. Piepmeier E. H. and deGalan L. Spectrochim. Acta Part B 1975 30 263. Pichugin N. G. Rudnevskii N. K. and Maksimov D. E. Zh. Anal. Khim. 1977 32 12. Dawson J. B. and Ellis D. J. Spectrochim. Acta Part A 1967 23 565. Djulgerova R. B. Bulg. J. Phys. 1977 5 569. Djulgerova R. B. Bulg. J. Phys. 1980 7 91. Djulgerova R. B. Bulg. J. Phys. 1977 4 459. Cordos E. and Malmstadt H. V. Anal. Chem. 1973 45 27. Katskov D. A. Lebedev G. G. and L'Vov B. V. Zh. Prikl. Spektrosk. 1969 10 215. Rudnevsky N. K. Maksimov D. E. and Pichugin N. G. Zh. Prikl. Spektrosk. 1973 19 5. Ryu J. Y. Davis R. L. Williams J. C. and Williams J. C. Jr. Appl. Specrosc. 1988 42 1379. Tolg G. Talanta 1972 19 1489. Williams J. C. McDonald J. T. and Davis R. L. Anal. Instrum. 1987 16 241. Chen F. Y. and Williams J. C. Anal. Chem. 1990 62 489. Tseng J. L. Williams J. C. Bartlow R. B. Griffin S. T. and Williams J. C. Jr. Anal. Chem. 1991 63 1933. Mixon P. D. Bray C. W. Griffin S. T. and Williams J. C. Proc. IEEE Southeastcon 1992 2 586. McDonald J. T. Williams J. C. and Williams J. C. Jr. Appl. Spectrosc. 1989 43 697. Moody J. R. Lindstrom R. M. Anal. Chem. 1977 49 2264. Bonventre J. V. Blouch K. and Lechene C. in X-Ray Microscopy in Biology ed. Mazat M. A. University Park Press Baltimore 1981. Williams J. C. and Schafer J. A. Meth. Enzymol. 1990 191 232. Tseng J. L. Kung J. Y. Williams J. C. and Griffin S. T. Anal. Chem. 1992 64 1831. Czakow J. in Improved Hollow Cathode Lumps for Atomic Spectroscopy ed. Caroli S. Ellis Horwood Chichester 1985 p. 36. Mixon Paul Ph.D. Thesis Memphis State University Memphis 1993. White A. D. J. Appl. Phys. 1959 30 711. Slevin P. J. and Harrison W. W. Appl. Spec. Ret.. 1975 10 201. Puper 3104899C Received August 12 1993 Accepted March 9 1994