‘Ll Lill'll'i Date La '1' 91 tag : This is to certify that the thesis entitled HIGH VOLTAGE SPARK EMISSION DETECTOR FOR CHROMATOGRAPHY presented by Clayton L. Calkin has been accepted towards fulfillment of the requirements for Ph.D. degree in Chemistry S. R. Crouch Major professor 7/29/81 04639 OVERDUE FINES: 25¢ per day per item RETURNING LIBRARY MATERIALS: Place In book return to remove charge from circulat‘lon records HIGH VOLTAGE SPARK EMISSION DETECTOR FOR CHROMATOGRAPHY BY Clayton L. Calkin A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Chemistry 1981 \me .m\\ \.......U ABSTRACT HIGH VOLTAGE SPARK EMISSION DETECTOR FOR CHROMATOGRAPHY BY Clayton L. Calkin A DC-powered, miniature nanosecond spark source (MNSS) for emission spectrochemical analysis of gas Chromatographic effluents is described. The spark is formed between two thoriated tungsten electrodes by the discharge of a coaxial capacitor. The spark source is mounted outside the GC oven and is coupled to the gas chromatograph by a heated transfer line. The GC effluent is introduced into the heated spark chamber where atomization and excitation of the effluent occurs upon breakdown of the analytical gap. A microcomputer-controlled data acquisition system allows the implementation of time resolution techniques to dis- tinguish between the analyte emission and the background continuum produced by the spark discharge. Multiple sparks are computer-averaged to improve the signal-to-noise ratio. Chromatograms demonstrating the element-selective detection Clayton L. Calkin capabilities of the MNSS for various metallic and non- metallic elements are presented. Feasibility studies on the use of an electronically-triggered spark source as an element-selective HPLC detector are also presented. TO My Family and April ii ACKNOWLEDGMENTS First, I would like to thank my research preceptor, Stanley R. Crouch, for both his guidance and for allowing me the freedom to pursue my research goals in my own way. I would also like to thank Dr. Chris Enke for serving as my second reader. I wish to thank the members of the Chemistry Department support staff, whose assistance has been in- valuable. Special thanks go to Martin Rabb and Len Eisele for their help and consulation. I wish to thank the members of the Crouch group and the other fellow graduate students for their friendship, advice, and support over the years I have spent at MSU. Special acknowledgments go to Jim, Frank, April, and Susan for all the great times we've had. I will miss the canoe trips, parties, and football games, but I will always cherish the memories of those crazy times that have made graduate school one of the most enjoyable periods of my life. iii Chapter LIST OF TABLES. . . . . . . . . . . . . . LIST OF FIGURES . . . . . . . . . . . CHAPTER I - INTRODUCTION. . . . . . . . CHAPTER II - ELEMENT-SELECTIVE CHROMATOGRAPHY DETECTORS. . . . . . . . . A. Introduction . . . . . . . . B. Atomic Absorption Detectors. . . . C. Atomic Emission Detectors. . . . 1. Flame Photometric Detector . . 2. Microwave Induced Plasma Detector . . . . . . . . . . 3. Direct Current Plasma Detector . . . . . . . . . . 4. Inductively Coupled Plasma Detector . . . . . . . . . 5. Spark Discharge Detector . . . D. Summary. . . . . . . . . . . . TABLE OF CONTENTS CHAPTER III - INSTRUMENTATION . . . . A. B. C. Introduction . . . . . . . . Spark Source Construction. . Gas Chromatograph-Spark Source Interface. . . . . . . . . . . 1. Analytical Chamber Heating iv Page viii ix 10 11 12 12 15 18 18 CHAPTER Page 2. GC-Spark Source Coupling . . . . . . . 20 3. Sample Flow Cell . . . . . . . . . . . 22 D. Liquid Chromatograph-Spark Source Interface. . . . . . . . . . . . . . . . . 24 E. Thyratron-Controlled Spark Source. . . . . 26 F. Introduction Systems . . . . . . . . . . . 29 l. Crossed Flow Nebulizer . . . . . . . . 29 2. Ultrasonic Nebulizer . . . . . . . . . 31 G. Chromatographic Components . . . . . . . . 32 CHAPTER IV - ELECTRONICS AND OPERATION. . . . . . . 34 A. System Overview. . . . . . . . . . . . . . 34 B. Microcomputer. . . . . . . . . . . . . . . 37 C. Real Time Clock. . . . . . . . . . . . . . 39 1. System Timing Controller Chip . . . . . . . . . . . . . . . . . 4O 2. Real Time Clock Circuit. . . . . . . . 43 D. Data Acquisition Electronics . . . . . . . 50 l. Microcomputer Interface. . . . . . . . 51 2. Timer Board. . . . . . . . . . . . . . 58 3. Gated Integrator . . . . . . . . . . . 61 4. ADC/DAC/Programmable Gain Amplifier. . . . . . . . . . . . . . . 61 E. Photodiode Trigger . . . . . . . . . . . . 65 F. Photomultiplier Gate Circuit . . . . . . . 66 CHAPTER V - SOFTWARE. . . . . . . . . . . . . . . . 69 A. Microcomputer. . . . . . . . . . . . . . . 70 CHAPTER B. CHAPTER Choice of Programming Language . . . . Operating Programs . . . . . . . . . . 1. Solution Analysis Program. . . . . 2. Chromatographic Data Acquisition Program. . . . . . . . . . . . . . 3. Plotting Program . . . . . . . . . VI MINIATURE NANOSECOND SPARK SOURCE AS AN ELEMENT-SELECTIVE GAS DETECTOR O O O O O O O O O O O O 0 Introduction . . . . . . . . . . . . . Selection of Model Organometallic Compounds. . . . . . . . . . . . . . . 1. Introduction . . . . . . . . . . . 2. Synthesis of Chelates. . . . . . . 3. Chromatographic Considerations . . Time Resolution Studies. . . . . . . . Evaluation of GC-MNSS Performance. . . . . . . . . . . . . . 1. Detector Performance . . . . . . . 2. Evaluation of GC-MNSS Interface. . . . . . . . . . . . . 3. Analytical Results . . . . . . . . 4. Conclusions. . . . . . . . . . . . Thyratron-Controlled Spark Source . . . . . . . . . . . . . . . . 1. Spark Current Measurements . . . . 2. Chromatographic Evaluation.. . . . 3. Conclusions. . . . . . . . . . . . vi Page 71 73 73 78 84 87 87 88 88 90 91 91 95 95 100 101 106 108 109 117 128 CHAPTER Page F. Perspectives . . . . . . . . . . . . . . . 132 CHAPTER VII - FEASIBILITY STUDIES ON USING A THYRATRON-CONTROLLED SPARK SOURCE AS A HPLC DETECTOR . . . . . . 134 A. Introduction . . . . . . . . . . . . . . . 134 B. Droplet Size Measurements. . . . . . . . . 135 1. Introduction . . . . . . . . . . . . . 135 2. Veillon Margoshes Nebulizer. . . . . . 137 3. Crossed-Flow Nebulizer . . . . . . . . 138 4. Ultrasonic Nebulizer . . . . . . . . . 138 C. Solvent Effects. . . . . . . . . . . . . . 140 D. Time Resolution Studies. . . . . . . . . . 143 E. Conclusions. . . . . . . . . . . . . . . . 147 CHAPTER VIII - COMMENTARY . . . . . . . . . . . . . 150 REFERENCES. . . . . . . . . . . . . . . . . . . . . 154 vii Table LIST OF TABLES Page Interface IOT Instructions . . . . . . . . 55 Detection Limits with the Free- Running MNSS . . . . . . . . . . . . . . . 107 Detection Limits with the Thyratron-Controlled MNSS. . . . . . . . . 127 viii Figure 10 11 12 l3 14 15 LIST OF FIGURES Diagram of MNSS-GC system. . . . . . . . . Miniature nanosecond spark source ......... GC-MNSS coupler. . . . . . . . . . . Sample flow cell . . . . . . . . . . . Simplified circuit diagram of thyratron—controlled MNSS. . . . . . . . Crossed-flow nebulizer . Block diagram of MNSS data acquisition system . Block diagram of internal architecture of AMD 9513 . . . . . Real-time clock circuit. . . . . . . . . . IM-6100 timing diagram . . . . . . . . . . Microcomputer interface circuit. . . Timing circuit . . . . . . . . . . . . . . ADC/DCA/programmable gain amplifier circuit. . . . . . . . Photomultiplier gate circuit . . . . . . . Signal intensity gs delay time for copper. . . . . . . . . . . . . . ix Page l4 16 21 23 27 30 35 41 44 45 53 59 63 67 94 Table Page 16 Signal intensity XE integration time for COpper. . . . . . . . . . . . . . 96 17 Separation of Al, Cu, Cr trifluoro- acetylacetonates . . . . . . . . . . . . . 98 18 Aluminum calibration curve . . . . . . . . 102 19 Chromium calibration curve . . . . . . . . 103 20 COpper calibration curve . . . . . . . . . 104 21 Carbon calibration curve . . . . . . . . . 105 22 Effect of support gas on spark current. . . . . . . . . . . . . . . . . . 112 23 Spark current waveform at 1 kHz. . . . . . 115 24 Comparison of MNSS detector to FID. . . . . . . . . . . . . . . . . . . . 118 25 Element-selective response Chromatograms. . . . . . . . . . . . . . . 120 26 Signal intensity gs delay time for oxygen and nitrogen . . . . . . . 123 27 Oxygen calibration curve . . . . . . . . . 129 28 Carbon calibration curve . . . . . . . . . 130 29 Silicon calibration curve. . . . . . . . . 131 30 Signal intensity YE delay time for copper, with free-running MNSS . . . . . . . . . . . . . . . . . . . 145 31 Signal intensity gs delay time for copper, with thyratron-con- trolled MNSS . . . . . . . . . . . . . . . 146 Figure Page 32 Signal intensity XE delay time for copper . . . . . . . . . . . . . . . . 148 xi I . INTRODUCTION Today, the analytical chemist is being called upon to analyze increasing complex samples. The analysis of such sample often involves the employment of some form of chromatography, if not for sample analysis itself, for isolation of the component of interest from the potentially interfering matrix. Although chromatography is a powerful separation technique, the analytical power of the method is generally limited by the detector. The most commonly used chromatographic detectors are rather unselective and provide little chemical information concerning the detected species. As a result, identification of the various sample components is achieved by either matching the retention time of the component to that of a standard, or by collecting the fraction containing the component of interest and submit- ting it to additional testing. Both means of identifica- tion are inconvenient and, in the case of the former, a standard is necessary for matching retention times. The analytical capabilities of the chromatographic method can be greatly enhanced by employing detection systems that provide chemical information about the de- tected species. This is evidenced by the GC/MS combination, which many presently feel is the ultimate analytical instrument. Such combinations can ease the task of sample identification, since the sample separation and identifica- tion steps are combined. Element-selective detection is another means by which the analytical power of the chroma- tographic method can be enhanced. Element-selective de- tectors, based on atomic absorption or emission spectros- c0py, offer high sensitivity and linear response for a large number of elements. The plasma atomic emission sources are perhaps the most versatile chromatographic detectors, since these detectors can detect not only metallic, but also non-metallic elements. When the element-selective detection capabilities of these plasma sources are coupled with the great separation power of modern chromatographic systems, a powerful analytical instrument results. These sources, when used as chromatographic detectors, provide not only elemental information about the detected component, but also, in the case of CO, the empirical formulae of components as they elute from the chromatographic column. Additionally, element-selective detectors can function as both universal and a selective detector. For universal response the detector can be, when feasible, tuned to respond to carbon, which is an element common to virtually all sample components amenable to chromatographic analysis. Additionally, by tuning the detector to respond only to those sample components containing a specific element, the element-selective detection capabilities of these detectors can be used to produce a simpler chromatogram. The major obstacle to the development of these systems is the high cost of the plasma sources. The cost of these plasma sources is substantially greater than the cost of the chromatographic systems for which they would serve as detectors. This work describes the use of a high voltage spark source as an element-selective GC detector. This source possesses many of the desirable characteristics of the more expensive plasma sources. The spark discharge has sufficient excitation energy to excite a large number of metallic and non-metallic elements. The spark source is but a part of an entire chromatographic detection system. A microcomputer—controlled data acquisition system allows the user to employ time resolution and signal averaging to gain the highest possible signal—to-noise ratio for each element. Also presented, are feasibility studies on using an electronically-triggered spark source as an element- selective HPLC detector. II. ELEMENT-SELECTIVE CHROMATOGRAPHY DETECTORS A. INTRODUCTION Both liquid and gas chromatography have experienced a rapid growth in recent years. This growth can be attribu- ted to the power of the chromatographic process to separate complex mixtures. Although cHromatography is a powerful separation technique, a serious deficiency of the chroma- tographic technique is the lack of chemical information provided by the detector. Additionally, the commonly used non-selective chromatographic detectors, when coupled to to- day's high performance columns, often result in the produc- tion of complex Chromatograms. Element-selective chromato- graphic detectors offer several advantages over their non- selective counterparts. These detectors respond only to those compounds containing the element of interest and, as a result, produce simpler, easier to interpret Chromatograms. In addition, element-selective detectors can reduce the chromatographic requirements necessary for separation. This occurs when two unresolved components have a different elemental composition. Under these circumstances, the detector can be used to resolve the two components if the detector can be alternately tuned to respond to a unique element in each component. A promising approach to the development of an element-selective detector is to use atomic absorption or emission spectroscopy to encode op- tically information concerning the elemental constituents of the sample. This chapter deals exclusively with a dis- cussion of atomic absorption and emission based chromato- graphic detectors. For a more extensive review of all types of chromatographic detectors, the reader is referred to the works of Lantz (1) and Koeplin (2). B. ATOMIC ABSORPTION DETECTORS Both flame and electrothermal atomic absorption (AA) spectrometers have been used as element-specific chromato- graphic detectors (3-8). Atomic absorption spectrometers are easily interfaced to both gas and liquid chromatographic systems. When flame units are involved, interfacing is straight forward. In the case of GC a heated transfer line carries the GC effluent to the burner head or to the burner auxiliary gas flow line. Liquid chromatograph effluent streams can be connected directly to the nebulizer input (9). However, frequently the optimum flow rates for the nebulizer and the chromatograph differ substantially and an auxiliary solvent flow is required for optimum operation (10). Electrothermal atomizers offer detection limits superior to those possible with flames. These lower detection limits often eliminate the need to preconcentrate the sample be- fore separation. Interfacing gas chromatographs to these atomizers is essentially the same as with flame units; a heated transfer line carries the GC effluent to the sample introduction port of the furnace. This approach has been used for the selective detection of lead compounds (11), and selenium compounds (12). The step-wise temperature programs used with electrothermal atomizers in the analysis of liquids makes interfacing of liquid chromatographs to these units more difficult. The usually interfacing ap- proach is to collect the LC effluent with a fraction col- lector and then to sample each fraction (13). A similar approach is to use a switching valve to inject a sample periodically from the LC effluent (14). Although both approaches are successful and have been used to determine a number of different metals in LC streams (15-17), both result in Chromatograms consisting of a series of discrete samples; this prohibits resolution of closely eluting peaks. The major problem associated with the use of atomic absorp- tion detectors is that they are only useful for the detec- tion of metals. Additionally, the single element at a time capability, and the expense of purchasing a hollow cathode lamp for each element to be determined, limit the employment of this method of detection. C. ATOMIC EMISSION DETECTORS A more promising approach in developing an element- selective chromatographic detector is to construct a detector based on atomic emission. The inherent sensi- tivity and selectivity of AB sources, and the ability to determine non-metallic elements would make an atomic emis- sion based detector a more versatile element-selective chromatographic detector. 1. Flame Photometric Detector The flame photometric detector (FPD) was the first emission based gas chromatographic detector. The FPD was originally utilized as a phosphorous and sulfur selective detector (18). In the sulfur and phosphorous mode of Operation, the FPD is not a true atomic emission detector, since the emitting species are 82 and HPO. The selective detection capabilities of the FPD have been expanded to include a number of transition metals (19,20), boron (21), and nitrogen (21). Common problems associated with the FPD are interference from broad band emission generated by hydrocarbons (22), and compound dependent response (23). This broad band emission has been exploited and used to construct a dual channel, selective, non-selective GC detector (24). Improvements in flame geometry (25) and the develOpment of a dual flame detector, where one flame is used for effluent combustion and the other for excita- tion, have reduced the hydrocarbon interference (26,27). Additionally, dual flame units have been successfully ap- plied as phosphorous specific HPLC detectors (28). However, the low excitation energy of the flame limits this detector to the determination of a few easily excited elements. Furthermore, despite the improvements in flame geometry and the dual flame configuration, this detector still suf— fers from interference from hydrocarbons. 2. Microwave Induced Plasma Detector The microwave induced plasma source (MIP) has found wide use as an element-selective GC detector. The in- creasing use of the MIP as an element-selective GC detector results from recent improvements in cavity design. Earlier cavity designs limited the choice of plasma gas and could only support a reduced pressure helium plasma (29-30). The new cavities can support either argon or helium plasmas at atmospheric pressure (31,32). The MIP is not only capable of detecting metals, but also a number of non— metallic elements (33-35), including the halogens (36). Furthermore, these sources, when supporting a helium plasma, possess sufficient energy to generate line spectra (37) and, as a result, suffer from fewer interferences than FPDs. The small plasma volume of the MIP has the advantage of keeping the detector volume to a minimum; however, because of the small size of the MIP, sensitivity to carrier gas flow rate and extinguishing of the plasma by the GC solvent front are common problems with these sources. 3. Direct Current Plasma Detector The direct current plasma source (DCP) with its higher energy is less sensitive to carrier gas flow rates and can tolerate high concentrations of organic material. As a result this source has found application as both a GC and an HPLC detector (38-41). Improvements in electrode con- figuration have reduced the drifting problems in earlier designs (42). New three-electrode configurations have not only increased the stability of the source, but also have movedtflmipoint of maximum analyte emission away from the high plasma background radiation region, improving the detection limits possible with this source. This source has been found to be useful as an HPLC detector using not only reverse, but also normal phase solvent systems (43). The relatively high cost of this instrument, however, pre- cludes its use as a routine chromatographic detector. 4. Inductively Coupled Plasma Detector Inductively coupled plasma sources (ICP) have also found applicability as chromatographic detectors. This source is normally used to analyze solutions and, as a 10 result, can be easily used as an HPLC detector (44,45). Recent reports of the use of an ICP as an oxygen and a nitrogen specific GC detector (46,47) illustrate the versatility of this source. The detection of ICP induced non-resonant, near infra-red emission from C1 and Br (48) should increase the applicability of this source as both a GC and HPLC detector. In addition to giving elemental information, the ICP can be used to obtain the empirical formulas of compounds as they elute from a GC column (49). As with the DCP with high cost of these instruments precludes their use as routine chromatographic detectors. 5. Spark Discharge Detector Work by Lantz (1) and Koeplin (2) demonstrated that a high voltage spark discharge could be used to generate atomic emission from CC effluents. Their work reported that a large number of elements could be excited, and de- tection limits as low as 10—12 g/s were possible for some elements. This source possessed many of the desirable characteristics of the plasma sources and yet was con- siderably less expensive than these sources. However, problems persisted with the reliability of the source and the degree of chromatographic band broadening caused by the large detector volume. 11 D . SUMMARY The element selective detection capability of atomic emission based detectors has the potential of vastly in- creasing the analytical power of modern chromatographic systems. These detectors not only produce simpler, easier to interpret Chromatograms, but also provide chemical information concerning the detected species. However, the plasma sources, that offer the high power necessary to excite a wide range of elements without producing spectral interferences, are prohibitively expensive for routine use. This work describes the use of a high voltage spark source and gas chromatograph-spark source interface that allows element-selective detection of GC effluents to be easily performed. Additionally, feasibility studies on using an electronically triggered spark source as an element-selective HPLC detector are also presented. I I I . INSTRUMENTATION The miniature nanosecond spark source (MNSS) has been successfully applied as an element-selective GC detector by Lantz and Koeplin (1,2). The original work by Lantz employed a low repetition rate, single gap spark source. Koeplin designed a new double gap source, which increased the breakdown potential stability, and included a DC power supply for higher repetition rates. This design was a considerable improvement over the earlier spark source; however, problems persisted in the GC-Spark Source inter- face. One goal of this work was to develop a GC-Spark Source coupling scheme that would allow the spark source to be moved outside of the GC oven, without compromising chromato- graphic performance. Additionally, work has been performed on the development of a viable HPLC—Spark Source interface. Preliminary results indicated that a new electronically triggered spark source was necessary to allow implementa- tion of the proposed interface. Such a source was sub- sequently constructed. This electronically triggered spark source was then evaluated as a GC detector. Feasibility studies were also carried out on the application of this source as an HPLC detector. This chapter describes the instrumentation and chromatography interfacing schemes developed in the course of this work. 12 13 A. INTRODUCTION A diagram of the MNSS-GC system is shown in Figure 1. In operation, the effluent of the GC column is introduced into the heated Spark chamber perpendicular to the inter- electrode axis. Atomization and excitation of the ef- fluent occur when the analytical gap breaks down. The photodiode trigger monitors the control gap and triggers the data acquisition electronics to begin the data acquisi- tion sequence as soon as emission from the control gap is detected. The data acquisition sequence employs time resolution techniques to discriminate between the analyte emission and the background continuum generated by the spark discharge. A selected time window is chosen for integration of the photocurrent during the time of maximum signal-to-noise ratio. Digitized data are received from the data acquisition electronics and are stored in the microcomputer memory. Simultaneously the digital data are sent to a digital-to-analog converter (DAC) so that the user can follow the progress of the chromatogram on a strip chart recorder. The data acquisition electronics are under the direct control of an IM 6100 based microcomputer. This microcomputer communicates with a PDP 8/e minicomputer on which Operating programs are created and compiled. The resulting binary program is then downloaded into the memory of the microcomputer via a serial link. A CRT terminal allows the user to input data acquisition control l4 442 Emu». Pm U EMFDQEOBZE mkam—zouomui 1 1 mu_zomkum._m ZO_._._m_DOU< (F40 mwkmzomkuwam mmogmk woo_oo...OIn_ \2 men5 0 A; » ban—3m mmBOQ m0<50> 10.1 .Ewumhm owlmmzz mo EMHmMHo .H wusmflm .U .0 00¢. mmem 15 parameters and communicate with both computers. B . SPARK SOURCE A diagram of the MNSS is shown in Figure 2. The source consists of three separable components: a coaxial capacitor, a control gap housing, and an analytical gap housing. The coaxial capacitor consists of an outer copper tube, a high molecular weight polyethylene dielectric (0.02 inch thickness), and an inner aluminum cylinder. A high voltage connector (Amphenol 82-843) is threaded into the top on the aluminum cylinder and provides the electrical connection to the high voltage DC power supply (Spellman Corp. Model UHR 10P100). The upper electrode of the control gap is mounted in the aluminum cylinder and is held in place with a set screw. To prevent possible thermal damage to the capacitor dielectric as a result of heat transfer from the heated analytical gap, a cooling coil is silver soldered to the outer surface of the capacitor. This cooling coil consists of four feet of 1/8" o.d. COpper tubing. Cooling water connections are made with 1/8" MPT Swagelock fittings. The upper control gap is formed between two 0.04" 2% thoriated tungsten electrodes (Union Carbide, Linde Division) and is contained in a Teflon housing. Argon is fed into the Teflon housing through a Gilmont flow meter. Removable quartz windows are mounted on either side of l6 Cooling Cooling Water\ ”Water IN _our """"" ‘71— l3.5" Control Gap Argon --> Teflon Spacer Hot Ar Sheath Gas —’ Analytical Sample—r Gap in Flow Cell Exhaust—r Exhaust Figure 2. Miniature nanosecond Spark source. 17 the Teflon housing to allow observation of the discharge and to provide light transmission to the photodiode trigger. The analytical gap is housed in the lower segment of the spark source housing. The upper electrode of the analytical gap also serves as the lower electrode of the control gap. The electrode is mounted in a Teflon holder and is held in place with a set screw. One end of a 2.2 MO resistor is soldered to the set screw while the other end makes contact with grounded spark housing. This resistor serves to maintain the cathode of the control gap at a true ground. Any leakage current which may develOp before breakdown is shunted to ground through this resistor. Upon breakdown of the control gap, a rapid voltage in- crease across the resistor occurs. Breakdown of the analytical gap results when the resistance of the gap becomes less than that of the bridging resistor. Argon is supplied to the analytical chamber through a port in the Teflon holder. The argon enters the analytical chamber through a series of holes surrounding the electrode. The cathode of the analytical gap is mounted in a threaded stage that can be adjusted to obtain different gap lengths. The analytical gap is housed in a small Teflon flow cell whose purpose is to reduce the effective volume of the analytical chamber. A one inch focal length lens is mounted on a movable stage and serves to collimate the discharge radiation. Two exhaust ports are drilled on either side of the analytical gap housing. 18 C. GAS CHROMATOGRAPH-SPARK SOURCE INTERFACE In their work both Lantz (l) and Koeplin (2) placed the spark source in the GC oven. This arrangement has the advantage of preventing the condensation of the GC efflu- ent in the analytical chamber. However, putting the spark source in the GC oven limits the choice of possible capacitor dielectric material, since the dielectric must be able to withstand the high temperatures encountered in gas chromatography. Of the materials tested by Koeplin, Teflon was found to be the best, because its high thermal stability allow it to withstand the GC oven temperature. However, the high impurity content of the Teflon caused frequent breakdown of the dielectric. This necessitated the frequent replacement of the capacitor dielectric, an arduous task. The high temperature inside the GC oven also prevents the user from easily making the fine adjust- ments to the spark source that are often necessary for optimum performance. For the above reasons, we decided to move the spark source outside the GC oven and to employ a heated chamber to prevent condensation of the GC effluent. 1. Analytical Chamber Heating Heating the analytical chamber was achieved by flush- ing it with hot argon gas. The first approach taken to generate the necessary hot argon utilized 25 feet of 0.25" l9 o.d. copper tubing packed with 20-30 mesh copper granules. This tubing was coiled and placed in a specially constructed oven held at a temperature of 400°C. Argon was passed through the coil at a flow rate of three liters per minute. The argon was found to emerge at a high initial tempera- ture (mZSOOC). However, this high temperature could not be sustained. The maximum sustainable temperature at the three liter per minute flow rate was found to be approxi- mately 900 C, clearly, this approach was not satisfactory. To generate the high temperatures necessary, an in- expensive commercial gas heater was purchased (Hotwatt Inc. Model AH3710F). This unit was designed to heat gases at high flow rates, i.e., two cubic feet per minute or greater. To prevent destruction of the heater at the low flow rates used in this application, the temperature of the unit was controlled with a Variac transformer. With this heater a three liter per minute flow of argon could be heated to 400°C. The hot argon from the gas heater is introduced into the analytical chamber through the sheath gas port. The sheath port serves to break up the argon stream into a uniform flow which surrounds the analytical gap, and provides uniform heating of the analytical chamber With this arrangement the analytical chamber was readily heated to approximately 190°C. Higher temperatures can be achieved by increasing the argon flow rate, however, at these higher flow rates loss of spark source stability occurs . 20 2. Gas Chromatograph-Spark Source Coupling When coupling any external device to a chromatographic system, the introduction of extra-column volume must be minimized in order to prevent the loss of chromatographic performance through chromatographic band broadening. For this reason a heated transfer line from the GC to the spark source was not used, since the dead volume introduced by the transfer line would compromise chromatographic per- formance. The approach taken in coupling the spark source to the gas chromatograph was to construct what is essentially an extension of the GC oven. This extension was then fitted to the spark source to allow the end of the GC column to be positioned very close to the Spark gap. The coupler is illustrated in Figure 3. The coupler was formed by inserting the last 18 inches of the GC column into a 0.25" o.d. aluminum tube, packing the volume between the GC column and the aluminum tube with 20—30 mesh copper granules. The tube was then wrapped with an electrical heating cord. A small hole was drilled into the side on the coupler and a chromel-constantan thermocouple was inserted to provide temperature monitoring capability. The temperature of the electrical heating cord was con- trolled with a Variac transformer. The coupler was fitted to the sample introduction port of the spark source with a Teflon Swagelock fitting. A Teflon fitting was used 21 .Hmamsoo mmzzluw .m wusmfim 3m; ozm 5:28 .3 0.036050 :0 .Obsm mm0_c_O.—m so: on nos. 8.3» .oo .mgo 298085;... 52288 0:33... .83an .0006 _< 6.0 ONO 3m; m9m 28 8:8: 22 since it was found that the brass spark source housing acted as a heat Sink causing condensation of the GC ef- fluent. The Teflon fitting eliminated this heat sinking characteristic of the Spark housing. 3. Flow Cell The efforts to minimize the level of chromatographic band broadening introduced in coupling the spark source to the GC would be futile if the effective detector volume of the Spark source was not reduced. The volume of the analytical chamber of the spark source is approximately 50 ml. Clearly any gains in minimizing band broadening with the present coupling scheme are negated by this large volume. To reduce the effective volume of the analytical chamber, a flow cell was constructed. An illustration of the flow cell is presented in Figure 4. The flow cell is constructed from Teflon and has an entrance and exit port, and two holes for insertion of the electrodes. A 6 mm diameter Opening permits Observation of the discharge. No window material covers this Opening Since deposition of debris from the spark discharge would soon obscure the Spark channel. Leakage of the GC effluent into the analytical chamber from this Opening is minimized by the positive pressure exerted on the Opening by the relatively high flow of the hot argon sheath gas. 23 I _l r— F1 3MM I Q J FRONT VIEW Q 0 Figure 4. TOP VIEW Sample flow cell. 24 The flow cell is mounted in the analytical chamber, with one end fitted to the sample port and the other end fitted to an exhaust port. The end of the GC column fits into the end of the flow cell, the volume between the Spark gap and the GC column being approximately 300 ul. Thus the flow cell reduces the effective detector volume by more than 2 orders of magnitude. D. LIQUID CHROMATOGRAPH-SPARK SOURCE INTERFACE Early work by Lantz (1) demonstrated the feasibility of employing a Spark source as an element-Specific HPLC detector. In this work, a single gap pulsed spark source was used to detect amino acids as they eluted from an ion exchange column. The interfacing scheme developed by Lantz employed pneumatic nebulization of the LC effluent with subsequent desolvation of the sample aerosol. The resulting dry aerosol was then introduced into the spark source. Sample desolvation was achieved using the system described by Veillon and Margoshes (50). In this system the sample aerosol was introduced into a heated chamber where solvent vaporization took place. The vaporized sample then passed through modified Friedrich condenser where the solvent was removed. This interfacing approach, while successful, introduced significant dead volume into the chromatographic system. The volume of the desolvation system was approximately 0.5 liters. Although this is 25 not a true dead volume (since the sample was quickly con- verted into a gaseous form and carried rapidly by a stream of argon), it nevertheless is clearly unacceptable if optimum chromatographic performance is to be maintained. An alternate HPLC-MNSS interfacing scheme based on the spark and spray solution analysis technique (51,52) was developed. The current interface utilizes an ultrasonic nebulizer to generate a sample aerosol from the LC ef- fluent stream. The aerosol is then introduced directly into the Spark chamber. To facilitate aerosol desolvation and prevent accumulation of solvent in the Spark source, the spark chamber is heated and hot argon is used as the aerosol carrier gas. This approach eliminates the de- solvation system thereby significantly reducing the dead volume of the MNSS-HPLC interface. Pursuing this approach required substantial modifications to the MNSS system. The introduction of large amounts of sample aerosol was found to cause instability in the firing frequency of the free running Spark source. This instability demanded that spark breakdown be electronically controlled. Electronic control of the spark source was achieved by incorporating a hydrogen thyratron into the MNSS circuit. 26 E. THYRATRON-CONTROLLED SPARK SOURCE Thyratrons have been used by a number of workers for control of Spark sources (53-55). A schematic diagram of the Single gap, thyratron controlled source is presented in Figure 5. When the thyratron is in the non-conductive state the lower electrode is held at the same potential as the spark anode. Upon triggering, the thyratron rapidly switches the lower electrode to ground. A rapid voltage increase across the bridging resistor occurs and gap break- down results. Implementation of this scheme required modification of the original Spark source. Since electronic control over spark breakdown is gained with the thyratron, the control gap Of the original design was no longer re- quired. The control gap was removed and a 6 cm electrode was extended from the coaxial capacitor to the analytical chamber to form a Single gap with the cathode of the original spark source. High current multistranded wire was used to connect the 22 M0 bridging resistor across the Spark gap. One end of the wire was connected to the aluminum cylinder of the energy storage capacitor and is held in place with a set screw. The other end is attached to the cathode mounting stage with a set screw. The Spark source housing is essentially unchanged; the only modification involved replacement of the brass cathode mounting stage with one made of Delrin. Replacement was necessary in order to isolate the cathode electrically from the grounded spark 27 .anZ OOHHOHOQOOISOADHHSLH mo Emumwflo ufloouwo OOHMAHQEHm .m anamflm -3382 H $.er - ANS: -. , >o - A zomezs oi. Ll l. o W. 1. L 5&8 aqo xmqam .CSNN Milli: meowfiww . 7 :9: .C.s: .<lL 28 source housing. A brass rod within the threaded Delrin stem provides electrical connection between the spark cathode and the thyratron anode. An EG & G Model HY-6 hydrogen thyratron was employed. The hydrogen thyratron has several advantages over other gaseous thyratrons (e.g., xenon, mercury) (56). The low molecular weight of hydrogen results in little cathode oxide damage, and rapid deionization of the tube permits high repetition rates to be used. Heating of a hydride reservoir allows the in-Situ generation of hydrogen to replace that lost in normal Operation. For safety and RF noise reduction the thyratron was mounted in a small aluminum box. The thyratron requires no biasing; however, current must be supplied to the cathode and hydride reservoir heaters. This current is supplied by an external 6V DC power supply. A three minute con— ditioning period is required to heat the hydride and generate hydrogen before triggering of the tube can be initiated. To suppress RF noise production, high current 8.8 uH in- ductors are in series with the thyratron anode and heater connections. High voltage connections between the thyra- tron and the Spark source are made with RG/8 coaxial cable and high voltage BNC connectors (Amphenol 82-856, 82-320). Triggering of the thyratron is achieved by a fast positive pulse applied to the thyratron grid. This pulse ionizes the hydrogen in the tube and causes the thyratron 29 to become conductive. An EG & G thyratron driver (Model TM-27) is currently used as the pulse source. This driver provides an 800 V pulse with a rise time of approximately 100 ns. A rapid grid pulse risetime is essential since ignition jitter is strongly influenced by the rise time characteristics of the trigger pulse (56). F. INTRODUCTION SYSTEMS Introduction of liquid samples into the MNSS was achieved by dispersing the sample solution into an aerosol. Aerosol generation was performed using one of two avail- able nebulizers; a crossed flow and an ultrasonic nebulizer. A brief description of each is given below. 1. Crossed Flow Nebulizer A crossed flow nebulizer based on the design described by Donohue and Carter (57) was constructed. A diagram of the nebulizer is presented in Figure 6. This nebulizer utilizes two metal-jacketed glass capillaries mounted at right angles. The horizontal capillary directs argon across the tip of the vertical solution uptake capillary to gen- erate an aerosol. This nebulizer has been found to Operate efficiently at argon flow rates as low as 0.5 l/min. Proper alignment of the 2 capillaries is essential for Optimum efficiency. Earlier nebulizer designs restricted NYLON SCREW ARGON GAS , INLET . 30 SAMPLE UPTAKE I, TEFLON BODY—>- l/4 IN. NYLON SWAGELOK FITTING TEFLON INSERT RIGID PLASTIC FLANGE \ O-RING Figure 6. Crossed-flow nebulizer. 31 capillary alignment and breakage of the fragile glass capillaries was frequent. This new design permits the capillary position to be easily adjusted without the ap- plication of force on the capillary. The metal-jacketed glass capillaries are placed into Teflon inserts which are in turn attached to a rigid mounting flange with 0.25" Nylon Swagelok fittings. An 0 ring is placed between the mounting flange and the Teflon nebulizer body. Three Nylon screws fasten the mounting flange to the nebulizer. Adjust- ment of the capillary position is achieved by tightening one Of the screws. This compresses the O ring slightly and results in capillary movement. A Spray chamber patterned after the design developed by Plasma-Therm fits into the aperature of the nebulizer. This design minimizes the number Of large droplets reach- ing the Spark source. 2. Ultrasonic Nebulizer A Plasma-Therm ultrasonic nebulizer (Model UNS-l) was also used in this work. Ultrasonic nebulizers Offer several advantages over other types of nebulizers. Ultrasonic nebulizers produce finer aerosols with a smaller Size distribution than other nebulizers (58). Additionally, these nebulizers require no gas flow for aerosol generation. Therefore both aerosol production and carrier gas may be 32 independently varied. The ultrasonic transducer is a piezoelectric crystal contained in a Pyrex sleeve and is mounted in a Teflon body. Cooling water flows through a chamber behind the crystal and allows the crystal to be Operated at relatively high power levels. Power is sup- plied tO the ultrasonic transducer by a 50 W, 1.4 MHz radio frequency generator. Front panel control allows the frequency to be adjusted by :10 kHz. Aerosol generation is achieved by pumping the solu- tion through a delivery tube positioned very close to the surface of the Pyrex faceplate of the transducer. The solution makes contact with the surface and flows across it. Radio frequency power supplied to the transducer in- duces oscillations which cause the liquid to be dispersed into fine drOplets. Argon carrier gas is introduced into the spary chamber through a side arm in the drain tube and drives the aerosol out of the Spray chamber. G. CHROMATOGRAPHIC COMPONENTS The gas chromatograph used in this work was a Varian Aerograph Series 1400. This unit was equipped with a flame ionization detector (FID) and temperature programming. The FID was found to be quite useful when separations were being Optimized, and for determination of the effect of the MNSS on the chromatographic system. 33 A variety of chromatographic columns were used. Stain- less steel and Teflon were used as column materials, and all columns were packed in our laboratory. Purified argon was used in all separations. Purification of the tank argon was achieved by passing the argon through a 2 meter tube consisting of a series of segments with each segment containing a different purification material. The puri- fication materials consisted of: 5 A molecular sieves, activated charcoal, Ascarite, and BASF catalyst. Carrier gas flow rate measurements were made at the detector output using a bubble flow meter. The HPLC used in this work was a Spectra Physics 8700 Solvent Delivery System. This unit utilized a low pressure gradient formation system allowing ternary gradients to be produced with a single pump. Sample injection is performed with a Rheodyne Model 7120 syringe loading sample injector equipped with a 20 ul sample loop. A Chromatronix Model 220 dual wavelength (254, 280 nm) UV detector was also used. Prepacked reverse and normal phase columns were used in all experiments. IV. ELECTRONICS AND OPERATION A. SYSTEM OVERVIEW The microcomputer-controlled data acquisition system (DAS) of the MNSS was designed to provide time resolution and Signal averaging to gain the highest possible signal- to-noise ratio. The heart of the MNSS DAS is a microcom— puter which utilizes the Intersil IM6100 microprocessor. The remainder of the DAS consists of a microcomputer inter- face, a timer board, a fast photodiode trigger, a gated integrator, a programmable gain amplifier/analog-to-digital and digital-to-analog converter board, and a fast photo- multiplier gating circuit. A block diagram of the MNSS DAS is presented in Figure 7. The microcomputer communicates with the PDP 8/e mini- computer via a serial link. Programs used in the operation of the MNSS system are created and compiled on the mini- computer and then downloaded into the memory of the micro- computer. Upon the completion Of the downloading process, the microcomputer becomes independent of the minicomputer. If storage of the acquired data is desired at some later time, the minicomputer-microcomputer link can be reestablished. The data may then be transmitted to the 8/e, which can then 34 35 .Eoumwm cofluflmflsoom mumo mmzz mo EOAOMHO Roofim .n onsmflm mwbbmiogzi 342.55 m o o .3 non w mo mm «3828052 um” 86 s: u mucous? -- ._ _L :T I mufimmsz. 5252.2. u Soc: «9428:8202 85o moan 53296:“. , _ - , «0:338 - I - .5956 «33m $.28 $.er ~12 on 33.8.»! >53 398,610. ‘2 r - 36 write the data file onto a mass storage device. Operation of the MNSS system involves first loading the following operating parameters: number of data points desired, delay time, integrate time, and gain. The delay time refers to the time, relative to breakdown, the user wishes to delay before integration of the photocurrent is to begin. The integration time defines the integration period. The gain refers to the degree of amplification of the output Of the gated integrator desired. The user can chose from gains of 10, 50, 150. After entering all the necessary operating parameters the user types a carriage return to start the data collection process. The MNSS data acquisition sequence begins when a trigger pulse is produced by the photodiode trigger upon breakdown Of the control gap. This trigger pulse Opens a gate and allows 20 MHz clock pulses to reach the delay scaler. Pulses derived from the clock are counted by the delay scaler until the delay period has expired. The PMT, which is kept in a low gain state during the initial stages of the discharge by the PMT gate circuit, is returned to full gain upon expiration of the delay period, and integra- tion of the photocurrent begins. The gated integrator integrates the photocurrent for the duration of the inte- gration period loaded into the integrate scaler. Upon completion of the integration period, the PMT gate circuit lowers the gain of the PMT, and the amplified output of 37 the gated integrator is digitized by the analog-to-digital converter. The resulting data point is then stored in a temporary memory location of the microcomputer. The data acquisition process is repeated, and the resulting data point is added to the previous one to form a running total until 1024 points have been taken. At this time the running total is divided by 1024 to form a single data point. This data point is then stored in the common storage area of the microcomputer memory and sent to a digital-to-analog con- verter. The output of the DAC drives a strip chart recorder, which allows the user to follow the progress of the chromatogram. This process continues until the number of data points requested have been acquired. The user can then choose to send the data to the minicomputer for storage on a floppy disk or to start the process again. This brief overview was intended to give the reader an understanding of the MNSS-DAS. However, to understand the capabilities Of the system fully a detailed descrip- tion of the system is necessary. The remainder of this chapter is dedicated to such an explanation. B. MICROCOMPUTER The IM6100 microprocessor is the heart of the micro- computer. The IM6100 is a 12-bit CMOS microprocessor which executes the instruction set of the Digital Equipment Corporation PDP 8/e minicomputer. The highest link in 38 the computing chain within the research group is a PDP 8/e. The compatability of the IM6100 with the PDP 8/e was the primary consideration in the choice of the IM6100. The computing facilities of the research group can best be utilized by employing the minicomputer for upper level tasks such as program creation and compilation while using the microcomputer for data acquisition and experi- mental control. The common instruction sets of the two computers allow the user to employ the 08/8 operating system and its associated software to develop and debug programs on the 8/e and then to download the binary program into the microcomputer. This allows the microcomputer to control the experiment independent of the 8/e, and allows the 8/e to perform other tasks. The microcomputer and the data acquisition electronics are housed in Faraday cages to eliminate interference from the MNSS. Each cage consists of an aluminum box, one side of which is made of 14 mesh wire screen to provide ventila- tion. A plastic insulator covers the floor of the cage. The card racks holding the different circuit boards are placed on the plastic insulator so that they are elec- trically isolated from the cage. The cages are connected to a water pipe with heavy gauge wire to provide a good earth ground. The CPU board, 8K RAM, memory extender, and two serial ports were purchased in kit form from Pacific Cyber/Metrix, 39 San Ramon, CA. The microcomputer is equipped with a 2K PROM monitor developed by Christmann (59) and a real-time clock board. In operation, the microcomputer enters the transparent mode of the PROM monitor upon power-up. In this mode the microcomputer appears to be the 8/e as a terminal and the user can communicate to the 8/e. Pro- grams are created, compiled, and mapped on the 8/e. The binary program is then downloaded into the memory of the microcomputer. Upon completion of the downloading process the PROM monitor enters the local mode. From the local mode the user can begin execution of the program by enter- ing the starting address obtained from the mapping program and typing the chain command "G". C. REAL TIME CLOCK In order to obtain the retention time of the various components in a chromatogram, a real time clock was con- structed. The real-time clock (RTC) is on a printed cir- cuit board which fits into the card cage of the micro- computer. The RTC was designed with versatility in mind. The present design allows the measurement of elapsed time and can also be programmed to generate interrupts at almost any desired frequency. / 40 1. System Timing Controller Chip The RTC utilizes an Advanced Micro Devices (AMD) 9513 system timing controller as the timing element. This LSI circuit features five independent 16-bit counters, two of which are used in the RTC. One counter divides the 1.8 MHz external crystal frequency down to 2 Hz. The second counter counts the 2 Hz output of counter one and stores the accumulated count. To understand the Operation of the RTC fully one must first understand the AMD 9513 timing element. A simplified diagram of the internal structure of the AMD 9513 is presented in Figure 8. Each of the five counters in the 9513 is controlled by a counter mode register. Through this register the user can soft- ware-select the count source, binary or BCD counting, gating modes, and input-output polarity. Associated with each counter are separate load and hold registers. The load register is used to load the counter automatically with any predefined value; thus it can be used to control the period of the counter. The hold register is used to store, upon command, the count value of the counter without disturbing the counting process. A 16-bit master mode register controls those internal activities not controlled by the individual counter mode registers. The 9513 has an internal 16-bit bus connecting all control and data registers. The user may select through software instructions loaded into the master mode 41 g5 3‘“ mm” l6 - Bit Master }_ Load l CID Control Mode Register Register _ H , 1 I , Counter Control — Counter 4 1 Register I s I Bus Butter I: E 4 Hold and * Register Multiplexer} I «5-8:: I Data Pointer I 8 - Bit L a Command 90 Register Register FF 4 I Counter Control Counter 3 ‘ Register _ r J lasts} ~ ET .223... I Figure 8. Block diagram of internal architecture of AMD 9513. 42 register an 8- or l6-bit external bus width. The present design of the RTC uses an 8-bit external bus width. In this 8-bit bus width configuration the l6-bit information is transferred one bytead:a time to and from the 8 low order data pins. An 8-bit, write-only command register provides soft- ware control over many of the general counter functions. These functions include the loading and arming of the counters and the transfer of the counter contents to the hold register. Another important function of the command register is the loading of the data pointer. The data pointer is a 6—bit register used to control internal ad— dressing for data bus transfers via the data port. The data pointer uses 3 of its 6 bits for addressing the five different counter groups, and 2 bits for the address of the different registers associated with each counter. When configured in the 8-bit bus width mode the last bit of the data pointer is used to indicate if the present 8 bits of data is the LSB or the MSB of the 16-bit data word. The bus interface control lines RD, WR, CS, and C/D, control the read, write, and transfer operations of the AMD 9513. The active low read (RD) and write (WR) lines, when conditioned with the chip select (CS) line, control the read and write operations of the 9513. The control-data line (C/D) is used to select the destination of I/O in- formation within the 9513. To write to the command 43 register the C/D line is exerted high and the write line is Strobed. Reading and writing of information to the other registers is performed by loading the address of the desired register into the data pointer via the command register, exerting the C/D line low, and, depending on the action desired, strobing the read or write line. 2. Real Time Clock Circuit A schematic diagram of the RTC circuit is presented in Figure 9. The RTC circuit can be broken down into three functional components; decoding circuitry for microprocessor I/O transfers, 9513 data input circuitry, 9513 timing element and output circuitry. To understand how the decoding circuitry for micro- proceSSor I/O transfers operates one must first understand the timing and the sequence of events which occur during the execution of an input output transfer (IOT) instruc- tion by the IM6100. A timing diagram illustrating the dif- ferent states of an IOT instruction is presented in Figure 10. The instruction sequence begins when the instruction address is placed on the data-address lines (DX lines) denoted as state one in the diagram. The memory responds by placing the instruction on the DX lines at state two. The CPU then reads the instruction at state three. At this time the CPU recognizes the instruction as an IOT instruction by its operations code (Opcode). In all IOT 44 OX0- “ II LXMAR m Figure 9. Real—time clock circuit. 45 .emusmae measeu ooeoqu .oe magmas at,“ _ 4mm>mo _ll_ «(SXJ IUFME. E.. 226 m5. 226 <5. 226 :20“. . c2833.:- 46 instructions, the first 3 bits (octal digit) of the 12- bit Opcode word is a binary Six. Upon recognition of the IOT instruction, the CPU sequences the IOT instruction through a two-cycle execute phase, referred to as IOTA and IOTB. The instruction can be latched into the device interface during IOTA using the falling edge of LXMAR, denoted as state five. A data word from the device is placed on the DX lines by the device interface during state Six. The data can then be written into the IM6100 when DEVSEL is low and XTC is high. The device controls the action to be taken by the CPU by exerting control over the four control lines designated as C0, C1, C2, SKP. The state of these control lines determines whether a CPU read, write, or skip the next instruction is requested. The CPU reads these control lines during state seven. Data to be written into the device are present on the IM6100 data lines (DX lines) during state eight in the timing diagram. The data can then be written into the device when DEVSEL and XTC are low. The RTC circuitry responsible for the decoding of the IOT instructions of the IM6100 is composed of: two hex latches 74174-1, 2, three BCD to decimal decoders 7442-l,2,3, and two quad two input OR gates. Three IOT instructions are recognized by the RTC decoding circuitry: 47 6553 latch input instructions into 9513 input latches 6554 read data out of the 9513 timing element 6555 write data into the 9513 timing element In operation, the IOT instruction is latched into the two hex latches 74l74-l,2 on the falling edge of LXMAR. The outputs of these latches are connected to the three BCD-to-decimal decoders. The decoders labeled 7442-2,3 decode bits 3-8 to produce the device number, while decoder 7442-1 decodes bits 9-12 to produce the assignment number. The outputs of the decoders are then OR'ed together to produce the decoded 6553, 6554, 6555, IOT instructions. The 9513 data input circuitry is composed of two hex latches 74l74-3,4, and two tristate buffers 74365-1,2. As was stated earlier the 9513 timing element is con- figured with an 8-bit external data bus. The lZ-bit word length of the IM6100 allows 4 bits to be used for control purposes. The most significant bit (MSB), DXO, is used to enable the tristate input buffers 74365-l,2. The tristate buffers utilize an active low enable signal; therefore DXO must be set low to write commands into the 9513. DXl is unused, while DX2 is used to control the chip select (CS) line of the 9513. Therefore, to write information into the 9513 DXO, and DX2 must be low. DX3 is used to control the control/data (C/D) line of the 48 9513. This bit, when set high, directs the data on the data bus to the command register. When this bit is set low, the information on the data bus is directed to the register currently addressed by the data pointer. In Operation, commands are loaded into the accumulator of the IM6100 and 6553 IOT instruction is executed. The 6553 instruction is decoded by the decoding circuitry and a negative pulse is produced. This pulse is OR'ed with the DEVSEL and XTC lines to produce a pulse which causes the data input latches 74l74-3,4 to latch the accumulator con- tents. The latched information is placed on the data bus of the 9513 through the tristate input buffers 74365-l,2. The information is then written into the 9513 by executing a 6555 IOT instruction to produce the necessary write pulse. In this manner counter 3 of the 9513 is programmed to divide the 1.8 MHz Signal produced by the external crystal down to a 2 Hz signal. Counter 4 is then programmed to countup (in binary) the output of counter 3. Counter 4 is used to store the elapsed time. When the 2 Hz Signal is used, the RTC has a resolution of 0.5 seconds. To read the elapsed time, the following operations must be performed: First the data pointer is loaded with the address of the hold register of counter 4 by issuing the necessary commands to the command register. The command to save the contents of counter 4 is then loaded into the accumulator of the IM6100 and then latched into the input 49 latches by executing the 6553 IOT instruction. The SAVE command is then written into the 9513 command register by exerting the C/D line high and executing the 6555 IOT instruction. This causes the contents of counter four to be transferred to the hold register. To read the elapsed time the data input buffers must be disabled. This is done by loading 40008 into the ac— cumulator of the IM6100 and writing this word into the data input latches with a 6553 IOT instruction. This causes DXO to be set high, and thus disables the data input buf- fers. The elapsed time can now be read into the accumulator by executing the 6554 IOT instruction and enabling the out- put buffers. The read command (6554) must be executed twice since the 16 bit time word is placed on the 8-bit data bus one byte at a time. Execution of the read com- mand causes the 9513 to place the LSB of contents of the counter 4 hold register onto the data bus. The output tristate buffers are then enabled by OR'ing the 6554 read command, DEVSEL, and XTC together. The OR'ing of these lines enables the output buffers and properly times the data transfer between the 9513 and IM6100. The first byte of the elapsed time is then stored in a temporary memory location and the second byte of the contents of the counter 4 hold register is read into the accumulator by repeating the 6554 read command. A 12-bit time value is then formed by saving the lower 4 bits of the second data byte and 50 rotating these 4 bits into the upper four bit position. The first data byte is then retrieved from memory and added to the four bits from the second byte. This Simple procedure allows time intervals Of approximately 34 minutes to be measured. This is the limit imposed by using a 12- bit word length. To measure longer time intervals the modulus of counter 3 can be changed to one hertz. This doubles the measureable time interval, but at the expense of resolution. Longer time periods can be measured, without sacrificing resolution, by implementing software that utilizes all 16 bits of counter information. Using a two hertz signal and all sixteen bits of information measure- ment of time periods of greater than nine hours are pos- Sible. D. DATA ACQUISITION ELECTRONICS The data acquisition electronics (DAE) consisting of the microcomputer interface, the timer board, the gated integrator, and the ADC/DCA-amplifier board, are housed in a Faraday cage separate from the microcomputer. To inter- face the data acquisition electronics to the microcomputer, the data address lines (DX lines), and the various control lines of the IM6100 must be brought to the backplane of the DAB. This essential communication between the DAB and the microcomputer is achieved by using two printed circuit boards containing the necessary logic elements. One board 51 plugs into an edge connector on the microcomputer back- plane, while the other plugs into the DAB backplane. The two boards communicate through three feet of forty-conductor shielded ribbon cable. The bus board located in the DAB is responsible for decoding the IOT instructions issued by the IM6100. The details of the timing and decoding of the IM6100 IOT instructions are given in the real time clock segment of this chapter; therefore only a brief description of the decoding process will be given here. The IOT instruction is latched into the two hex latches of the bus board using the falling edge of LXMAR. The outputs of the latches are then decoded by three BCD-to-decimal decoders. The outputs of the decoders are then placed on the backplane of the DAB. The bus board located in the microcomputer contains four hex tristate buffers. These buffers serve to buffer the ox lines of the IM6100.and allow transmission of data to and from the DAB. A complete description, including diagrams, of both bus boards is given by Gano (60). The remainder of this chapter is devoted to a detailed description of the various components of the data acquisition electronics. 1. Microcomputer Interface The primary duty of the microcomputer interface is to coordinate data transfers between the DAB and the micro- computer. The interface is mounted on a wire wrap board. 52 Wire wrap was used because the interfacing requirements often change. The ability to add new circuits or to change the circuit wiring with wire wrap is a distinct advantage over printed circuit boards. A functional diagram of the microcomputer interface is presented in Figure 11. The process of producing the different write pulses necessary for latching control in- formation into the various latches begins with the active low outputs of the decoders being brought onto the inter- face board where they are connected to the inputs of dual- input OR gates. The unique state of an OR gate is a logic level low. This state is produced when both inputs to the gate are low. When an IOT instruction is decoded by the BCD-to-decimal decoders, the active low outputs of the decoders cause a low logic level at the output of the OR gate. As can be seen in the circuit diagram, one gate is used to produce device code 44 and another is used to produce device code 45. The outputs of each of these gates serve as an input to three separate dual-input OR gates. The other input to each of these gates is the assignment digit of the IOT instruction, which results in the decoding of six IOT instructions. As was mentioned in the real- time clock section of this chapter, information from the IM6100 is available to a device when the IM6100 control lines DEVSEL and XTC are low. These two lines are OR'ed together to produce a low logic level when data is available ASI l__:£§i:::iE§E>a£vun:coaster 4- 53 LATCH iii; TO 9A; ' LATCH . TOINTE TE LATCH A A52 A54 EOC A53 €333 Figure 11. IOKO 5.250 of 745l40 TO PMT GATE A TO INTEGRATOR GATE 4 ADC CONVERT Microcomputer interface circuit. 54 from the IM6100. This output is OR'ed with the different IOT instructions to produce write pulses for the different control latches. The interface recognizes Six IOT in- structions issued by the IM6100. These instructions and their actions are presented in Table I. In addition to generating the write pulses for the different control latches in the DAB, the interface board also generates the control signals required by the gated integrator. The integrator requires two control signals; integrate gate Signal, and a reset signal. These two sig- nals are generated with two RS flip-flops constructed from high Speed NAND gates (74H00) located on the interface board. To insure that both flip-flops are in the reset state, an initialize pulse generated by a monostable located on the timer board resets both flip-flops upon power-up. The integrate flip-flOp is used to control not only the integrator but also the PMT gate and the analog- to-digital converter (ADC). The integrate flip-flop is set by the end of delay pulse generated by the delay scaler on the timer board. When set, the Q output of the flip- flop goes high. The first function of the Q output is to drive the reset line of the gated integrator high. This allows integration of the photocurrent, since the integrator can only integrate when the reset line is high. The Q output also activates the PMT gating circuit. The PMT gate circuit utilizes negative logic. Therefore the Q 55 Table I. Interface IOT Instructions. IOT INSTRUCTION ACTION 6451 6452 6453 6442 6443 6444 Load delay scaler Load DAC input latches Load integrate scaler ADC Skip test Transfer data to accumulator Select amplifier gain 56 output must be inverted to provide the correct logic level. A 745140 quad input NAND gate line driver was used as the inverter. The line driver was necessary Since the cur- rent required by the PMT gating circuit could not be pro- vided by a normal TTL inverter. The third function of the flip-flop is to provide the gate input signal for the gated integrator. The integrator utilizes negative input logic; therefore, a low logic level must be generated with a dura- tion that controls the integration period. This is achieved by setting the flip-flop with the end of delay pulse (to start integration) and resetting the flip—flop with the end of integration pulse. The 6 output is then low for the duration of the desired integration period. The Q output is also used to trigger the ADC to start the digitization of the analog signal. The ADC requires a convert pulse that is between 100 nS and 2 uS long to initiate the digitization process. This pulse is generated by using the rising edge of the 6 output to trigger a monostable which produces the necessary start conversion pulse for the ADC. Another function of the interface is to coordinate the data transfer from the ADC to the IM6100. The ADC used in the DAB requires 25 uS to digitize the analog input. During this time the IM6100 Sits in a wait loop, checking the status of the ADC by issuing the 6442 instruction. The ADC Signals the end of conversion (EOC) by exerting 57 its status line low. This signal is inverted and the rising edge is used to trigger a monostable. The monostable then produces a pulse of shorter duration. This pulse is OR'ed with the 6442 command, using the internal OR gate of the 74365 tristate buffer. When both of these lines go low, the tristate buffer is enabled. The outputs of the buffer are connected to the control lines of the IM6100 and, when the buffer is enabled the buffer causes the CPU to Skip out of the ADC wait loop. Data are then ready to be jam-transferred into the accumulator of the IM6100. The CPU is ready to accept data from a device when DEVSEL is low and XTC is high. The XTC line is inverted and OR'ed with the DEVSEL line to produce a low logic level when the CPU is ready to accept data. This output is OR'ed with the 6443 command to produce a pulse labelled ADC write in the circuit diagram. This pulse is used as the enable pulse for the data output tristate buffers located on the bus board, and thus causes the ADC data to be transferred to the IM6100. The monostable pulse derived from the EOC signal is also used to clear the gated integrator. This pulse is used to reset the RESET integrator flip-flop. This sets the reset line of the gated integrator low, causing the integrating capacitor to be discharged. 58 2. Timer Board The transitory nature of the Spark discharge requires the use of time resolution techniques in order to achieve the highest possible Signal-to-noise ratio. To employ these techniques, accurate measurement of Short time periods is essential. The timer board is responsible for these time measurements and is therefore perhaps the most important component in the DAB. The timer board consists of two parts: the delay scaler, and the integrate scaler. A functional diagram of the timer board is presented in Figure 12. a. Delay Scaler The purpose of the delay scaler is to delay integration of the photocurrent for a time period selected by the user. This delay period is measured relative to the breakdown of the control gap of the Spark source. In principle, the delay period allows the continuum radiation produced in the initial stages Of the Spark discharge to decay be- fore integration of the photocurrent is initiated. The heart of the delay scaler is made up of a 20 MHz oscillator, three 748161 binary counters, three 74585 magnitude com- parators, and three 7475 quad latches. Schottky TTL integrated circuits are used in all high frequency count- ing portions of the delay scaler. In Operation, the 59 0- 55's .Ow updao 0-05vh NXO nxo Gun .5 Nxo nxo n-050L ><4wo n . nwmvs Dow n i .9ka n-.@.m¢s v n-00mvk n- nkvk 0x0 hxo Vic g 08 8x0 N-nkvk u N- 553. .uflsouflo >n+ 9.5505 «-9396 o u vunhvk .3 one 9‘0 :3 0x0 03 93 :x0 70505 450.! 4:5»! v _- 8ka u 7.0.ka _-N_.mQ~. OCHEHB .NH magmas mwgih 60 desired delay period is written into the 7475 latches by the microcomputer. The counting process is then initiated when the inverted output of the photodiode trigger sets the RS flip-flop. This gates the 748112 JK flip-flop on, which divides the 20 MHz clock Signal down to 10 MHz. This 10 MHz TTL signal is then used as the count source for the three cascaded 748161 binary counters. The outputs of these counters are continuously compared to the value con- tained in the 7475 latches by the 74885 magnitude compara- ters. When the count value equals that in the latches the magnitude comparators produce a high logic level at the "equals" output pin. This output is then used to initiate the integrate scaler and to clear the counters of the delay scaler. When the 10 MHz clock signal produced by the JK flip-flop is used, time increments of 100 ns can be measured. b. Integrate Scaler The integrate scaler defines the period of integration for the gated integrator. The integrate scaler operates in the same manner as the delay scaler except that the counting process is initiated by the end of delay pulse produced by the delay scaler and terminated by the end of integration pulse produced by the magnitude comparator of the integrate scaler. With this circuit integration periods from 0.1 us to 409.6 us can be implemented. 61 3. Gated Integrator The gated integrator used in the DAB is a Model 4130 Gated Integrator Module, manufactured by Evans Associates, Berkeley, CA. This low-cost unit is a fast, low-leakage integrator that is TTL compatible. This unit utilizes negative input logic and has a minimum gating time of 30 ns. The input impedance of the integrator is 10 KO, and a 3800 pF integrating capacitor was used. This results in a 38 us time constant. The small physical Size of the gated integrator card required the mounting of the inte- grator card on another card whose size was compatible with the card cage of the DAB. A BNC connector was attached to the front edge of the mounting card. This connector feeds the photocurrent to the analog input of the integrator. This approach was used SO that the analog Signal could be kept off the relatively noisy DAE backplane. 4. Programmable Gain Amplifier-ADC-DAC Board The amplifier-ADC-DAC board combines as many of the analog circuits as possible on one board. Ideally, all analog circuitry as well as the ADC Should be on one board. This eliminates the need to place the analog signal on the system backplane where it is susceptible to noise pick-up. The use of a commercial integrator prohibited the placing of all analog circuitry on one board. Therefore, a 62 printed circuit board containing the amplifier, ADC, and DAC was constructed. To reduce the introduction of cir— cuitry noise into the analog section of the board a large ground plane passes beneath the ADC and surrounds the analog components. Further attempts to limit noise in- cluded severing the analog Signal line from the integrator so that it only communicates with the adjacent card slot. Thus the analog Signal is only carried approximately 1.5 inches on the DAB backplane. A functional diagram of the amplifier, ADC, and DAC board is presented in Figure 13. The programmable gain amplifier section consists of: a 7475 quad latch used for gain selection, a Teledyne Philbrick 1026 FET input Opera- tional amplifier, a precision resistor ladder, and Magne- craft W107 SPST reed relays for selection of the different feedback resistors. Reed relays were used as the switch elements in the resistor ladder because of their low "on" resistances. FET switches could have been used; however, FET'S have a finite on resistance. This on resistance contributes to the total resistance in the feedback loop and can thus lead to uncertainty in the gain factor, Since the gain of the amplifier is proportional to the total resistance in the feedback loop. Reed relays are much slower than FET switches; however, their on resistance is virtually zero. Since switching Speed is not important in this application, 63 h:&h:0.ll Udoi. .uflsouflo MOMMAHQEO cflmo manmEEmuooum\¢Oo\OQ¢ .me magmas uOm h¢m>ZOu _.x~ 3.. I: . g . uI I - 8 CL too—~ 00 I l I I . I . I . Ij on to 20 so +0 so INTEGRATION TIME (useo) Figure 16. Signal intensity XE integration time for copper. 97 was then separated on the OV-lOl column described pre- viously. The injection port, oven, coupler, and detector were maintained at 175°, 125°, 140°, and 1800 C, respectively. An argon carrier gas flow rate of 30 ml/min was used for all separations. The resulting Chromatograms are shown in Figure 17. In all metal Chromatograms, peaks represent 1 ug of metal. When the 2478.6 A carbon emission line is monitored the MNSS acts as a "universal" detector, since carbon is a com- I' ‘ '._ '21—"! I" i mon element in all the metal chelates. The carbon chroma- togram shows only two peaks, indicating that possibly two of the metal chelates coelute. Identification of these peaks can be easily performed by repeating the sample in- jection and monitoring the different metal emission lines. The aluminum chromatogram, indicates that the first peak in the carbon chromatogram is the Al(TFA)3 chelate. The copper and chromium Chromatograms indicate that the second peak in the carbon chromatogram is due to unresolved Cu- (TFA)2 and Cr(TFA)3. This demonstrates the usefulness of the element-selective detection. In this case the two co- eluting chelates have been resolved, without changing the chromatographic conditions, by making use of the element Specific response of the MNSS. Attempts to chromatograph the Ga(TFA)3 and In(TFA)3 were unsuccessful. Failure was a result of the poor chromatographic behavior and the poor spark sensitivity for the metals. INTENSITY 98 0 TIME(MINUTES) Figure 17. acetonates. 2000.0 CARBON 3000 J 10000 m r E m 2 1000.0 Lu I: 10000 .. 000.0 000.0 . , . I e I , fl 000.0 0.0 2.0 4.0 0.0 so 10.0 TIME(MINUTES) 2500.0 e 2500.0 Cr(TFAl3 20000 7 2000.0 I 9% 10000 ~ E 1000.0 1000.0 — 1 1000.0 0000 . 1 . ‘ I fl 000.0 0.0 2.0 0.0 0. e0 10.0 aunmu 1 I . I v I _] 0.0 2.0 £0 .0 ‘0 10.0 TIME(MINUTES) (mums . l y w k I r i . . , I *1 0.0 2.0 0.0 0.0 0.0 '00 TIME(MINUTES) Separation of Al, Cu, Cr trifluoroacetyl- 99 Acetonitrile was chosen as the solvent used in the preparation of the chelate mixture. Of the solvents tested, CH3OH, CHC13, and CH3CN, the latter had the least influence on the MNSS. Chloroform and chlorinated hydrocarbons in general had the greatest effect, causing erratic firing of the MNSS. Methanol and the other Short chain alcohols tested caused the firing frequency of the MNSS to increase Significantly. This frequency change results in a Shift in the observed baseline, since the background continuum produced by the MNSS is frequency dependent. At higher firing frequencies the background continuum is lower. This lower background, at higher repetition rates, may result from the persistence of ions in the spark gap causing a lower breakdown potential. Acetonitrile was well tol- erated by the MNSS, and volumes of up to 10 ul could be injected without causing Significant firing frequency changes. The Spurious peak, corresponding to the solvent front in the A1, Cr, and Cu Chromatograms, is a result of per- turbation of the MNSS by the relatively large amount of solvent. The larger size of this peak in the Al and Cu Chromatograms is due to a contribution by CN band emission (from the acetonitrile solvent), which occurs in the wave- length region where Al and Cu emission is detected. i , ._.—. .F. .14.... 100 2. Evaluation of GC-MNSS Interface The volume of the MNSS analytical chamber is approxi- mately 50 ml. Clearly, this large volume would result in a loss of chromatographic performance by introducing Sig- nificant extra-column band broadening. This large detector volume is effectively reduced to approximately 300 ul by housing the analytical gap in the Teflon flow cell. To measure the effect of the MNSS detector volume on the ef— ficiency of the chromatographic system the chromatographic bandwidth obtained with the MNSS was compared to that ob- tained with a flame ionization detector (FID). AS in HPLC the effect of extra-column band broadening is more severe for early eluting components (74). For this reason the Al(TFA)3 band was chosen for comparison. The full width at half maximum for the peak obtained with the MNSS was found to be 6% greater than that obtained with the FID. Thus, the MNSS does not severely comprise the performance of the chromatographic system. Noticeable peak tailing was observed with the copper chelate. This tailing could possibly be the result of partial condensation of the chelate in the coupler. To determine if indeed condensation was taking place within the coupler, the GC column was removed from the coupler, placed entirely within the GC oven, and connected to the FID. A Cu(TFA)2 standard was injected and the peak tailing 101 was still observed, suggesting that the tailing was the result of non-ideal chromatographic behavior of the chelate. This is in agreement with other workers who have found that in less than microgram quantities, the chromatographic behavior of the copper chelate is highly non-ideal (75). The non-ideal chromatographic behavior of Cu(TFA)2 is not surprising. Of the metal chelates chromatographed, Cu(TFA)2 was the only divalent metal complex. Other workers have found that complexes that can form a stable coordina- tion state greater than twice their oxidation state may polymerize or react with unsilanized surface hydroxyl groups on the chromatographic support (71). COpper(II) can indeed form octahedral complexes (76), albeit distorted ones, which may account for its poor chromatographic behavior. 3. Analytical Results A useful GC detector must provide a linear response with sample concentration. Calibration curves for the three metals and carbon were generated and are presented in Figures 18-21. All possess good linearity and, with the exception of COpper, extend from the nanogram to micro— gram range. The poor chromatographic behavior of Cu(TFA)2 prevented quantitative measurement at lower levels. The rather limited concentration range covered in the calibra- tion curves is a result of two factors. The light loading 102 100.0 ~ 80.0 — ._ _ 5 {3 00.0 — I E I 0. 40.0 _ o 20.0 — O °’° ‘ I . I I l fl 0.0 1,0 3.0 *0 2.0 MICROGRAMS INJECTED Figure 18. Aluminum calibration curve. PEAK HEIGHT 100.0 — 0.0 103 0.0 Figure I I I l I I 1.0 20 3.0 MICROGRAMS INJECTED l9. Chromium calibration curve. "% 104 PEAK HEIGHT 0.0 . , 0.0 4.0 3.0 12.0 MICROGRAMS INJECTED I ' I I Figure 20. Copper calibration curve. 105 100.0 — I'— 75.0 — I 9 m - I X 11‘) 50.0 —— 0. 25.0 — 0.0 0.0 Figure 21. . I I I I I I i 5.0 10.0 15.0 20.0 MICROGRAMS INJECTED Carbon calibration curve. I 25.0 fl 106 of the stationary phase, 1.5%, prevented large sample con- centrations from being injected, since column overload would result. Additionally, the weight percentage of metal in the chelate is relatively small. For example, Al(TFA)3 is only 5.5% A1 by weight; therefore, large amounts of the chelate have to be injected to realize a high metal concentration. Detection limits, defined as the quantity of sample per second that gave a signal-to-noise ratio of 2, were determined for the three metals and carbon. These detec- tion limits were obtained by calculating the average peak area, after background subtraction, from four separate sample injections. The signal-to-noise ratio was cal- culated from the relative standard deviation of the resulting data (S/N = l/(RSD)). The signal-to-noise ratio obtained for the least concentrated standard was used to calculate the concentration that would yield a signal-to-noise ratio of 2. This sample quantity was then divided by the peak width (in seconds) of the least concentrated standard. Results are presented in Table II. These values are com- parable to those obtained with a direct current plasma source (39). 4. Conclusions The MNSS has been demonstrated to be a useful metal- specific GC detector. The MNSS system offers good sensi- tivity and linear response without introducing significant 107 Table II. MNSS Detection Limits ELEMENT WAVELENGTH, K DETECTION LIMIT, (g/s) Carbon 2478.5 6 x 10-9 Aluminum 3961.5 6 x 10-10 Copper 3247.5 4 x 10'-9 -10 Chromium 4254.3 4 x 10 108 band broadening into the chromatographic system. The most serious deficiency of the free running MNSS system is in- stability in the firing frequency. This instability not only makes the analysis of certain compounds difficult (e.g., alcohols, chlorinated hydrocarbons), but also requires the operator be present to make the necessary adjustments to correct for any frequency changes. It was felt that elec- tronic control of the spark firing frequency should sub- stantially increase the applicability and ease of Opera- tion of the MNSS. Additionally, proposals to use the MNSS as an HPLC detector demanded that electronic control of the firing frequency be incorporated into the MNSS system, since the large amounts of solvent entering the spark source would certainly result in frequency instability. The re- mainder of this chapter deals with the evaluation of the electronically triggered spark source as a GC detector. E . THYRATRON-CONTROLLED MNSS Electronic control of the MNSS firing frequency was achieved by incorporating a hydrogen thyratron into the MNSS system. Since electronic control over spark break- down was achieved with the thyratron, the control gap in the original Spark source was no longer required. Removal of the control gap is desirable, since with a single gap spark source the total energy is now dissipated in the analytical gap. The control gap was removed and a 6 cm 109 electrode now extends from the coaxial capacitor to the analytical chamber to form a single gap with the cathode of the original spark source. For further information on the design and the theory of operation of the thyratron- controlled spark source, the reader is referred to the instrumentation chapter of this work. 1. Current Measurements The energy dissipated in a spark gap is proportional to the spark current (77). Therefore the spark current is of great importance in determining the spark plasma energy. Because of the importance of this parameter, the spark current waveform generated by the thyratron controlled spark source was measured. Current measurements on the free running MNSS made by Seng (61) involved measuring the voltage drop across a segment of nichrome resistance wire that was placed in the MNSS circuit. The data from the resulting voltage-time profile were used to solve the first order differential equation which related the voltage-time profile to the current-time profile. This method required the capacitance, resistance, and inductance of the resistance wire to be accurately known. Determination of the resistance was straightforward. However, measurement of the capaci- tance and inductance were more difficult. Therefore esti- mated values were used. Thus current values obtained using this method, while useful for comparative purposes, 110 may be in error. An alternate approach that allowed a more direct measure- ment of the spark current was employed for current measure- ments on the thyratron-controlled MNSS. A current monitor (Ion Physics Corp. Model CM-lO-M) was purchased and used to make all current measurements. The current monitor is a toroidal shaped transformer that allows the current through a conductor to be easily measured. Current measure- ments are made by passing the current-carrying conductor through the central opening of the monitor. This con— ductor acts as the primary of the transformer and the cur- rent in the conductor induces a current in the monitor. This current is then terminated into 509 to yield a voltage drop with a sensitivity of 0.1 V/A and a risetime of 10 ns. Current measurements on the MNSS were performed by passing the conductor connecting the thyratron anode and the spark cathode through the current monitor. The output of the monitor was connected to a 509 attenuator to reduce the signal within the range of the Tektronix 564 storage scope used for voltage measurements. The scope was equipped with a 381 dual trace sampling plug-in, and a 3T2 sampling time base plug-in. The voltage waveforms produced on the oscilloscope were photographed and digitized. The result- ing data were plotted. 111 a. Effect of Support Gas on Spark Current The effect of spark support gas on spark current was investigated. Current measurements were made with both helium and argon as the support gas. Figure 22 shows current-time profiles taken under identical conditions with the exception of support gas. As the figure illustrates, the duration of the Spark current is approximately 70 ns, virtually the same duration as observed by Seng with the free-running MNSS. Additionally, as was reported with the free-running MNSS, higher peak currents were observed with argon than with helium. Helium was investigated as a possible support gas be- cause it offers a number of advantages over argon. The background continuum generated by helium is substantially lower than that produced by argon. Additionally, the back- ground produced by helium is primarily in the "red" region of the Spectrum, where PMT sensitivity is low. Helium is also much easier to heat than argon; therefore the ana- lytical chamber of the MNSS could be heated to temperatures higher than those possible with argon. This is important when high boiling compounds, where condensation is likely, are being chromatographed. The lower peak current ob- tained with helium was also reflected in the analyte signal produced in helium. Generally, the signal obtained with helium was approximately one quarter that obtained with argon. The magnitude of the analyte signal produced in a 112 .ucmuuso xnmmm co mom UMOQQSm mo vowmwm .NN mnsmflm Ambcooomocos m2: AmUCooomOCOcv NEE. 0.0x. 0.00 0.00 0.3 0.0» 0.0a 0.0. 0.0 0.05 0.00 0.00 06¢ 0.00 0.0N 0.0. 0.0 Ti _, p — p — b — p b p h » h » coo plb p — p — p b p — p _ p _ » 00° - 0.0. ,. 0.0— 3 n , 0.0m Mu Nu m I. I 0.0m ) .w. 0.0m d I 6 I 0.00 - 0.9. .0231 >0. Maw >v. m6 .. I 0.0+ 0.8 (sdwo) lNBHaflO 113 particular gas is a function of many processes and cer- tainly the current is not the only factor affecting signal production. However, the energy dissipated in a spark is related to the spark current and, therefore, it is not surprising that a lower peak current produced a lower Signal. b. Effect of Bridging Resistance on Spark Current The effect of the bridging resistance on the spark cur- rent was also investigated. Four different resistors, ranging from 2.2“M2to 22.0DM2were inserted into the MNSS circuit, and the current waveform measured. These experi- ments revealed that the bridging resistor had virtually no effect on the peak spark current. However, observations did indicate that the smaller 2.2DM2reSistor yielded a smoother current pulse. However, when the 2.20M2resistor was used at lower repetition rates (1 kHz), the spark dis- charge would occasionally extinguish, and the capacitor discharge current would pass through the bridging resistor. Evidently at the lower repetition rate the residual ions from the previous discharge were removed from the gap, and the resistance of the gap became greater than the bridging resistor. This problem could be reduced by increasing the power supply voltage and thus the capacitor voltage. How— ever, arcing problems, which will be discussed later,pre— vented raising the potential above 5.5 kV. 114 c. Effect of Repetition Rate on Spark Current The influence of repetition rate on the discharge cur- rent was also investigated. Current measurements in argon and in helium were made at a repetition rate of 1 kHz, half the 2 kHz repetition rate normally used. No effect on peak current was observed with helium. However, an in- crease in peak current was observed in argon. Additionally, current oscillations were observed in argon at the lower repetition rate. This is illustrated in Figure 23. The Slightly higher peak current in argon is probably the result of the longer time between sparks, allowing the energy storage capacitor to be charged to the power supply potential. The observed current oscillations are probably the result of the capacitances and inductances associated with the thyratron and the conductor joining the thyratron to the MNSS (78), Since thyratrons are unidirectional switches and should conduct an oscillatory current (56). In the course of making the current measurements, peculiar capacitor charging behavior was observed when argon was used as the support gas. A high voltage probe was used to monitor the capacitor charging waveform, at the thyratron anode. When helium was used, the expected ex- ponential capacitor charging waveform was observed. How- ever, when argon was used as the support gas, this smooth charging waveform was not observed. The waveform observed 115 20.0 - CURRENT (amps) 10.0 _. -10-O I 1 fl 1 I I 7 I I I I I r I 00 250 50.0 75.0 100.0 1250 150.0 1750 TIME (nanoseconds) Figure 23. Spark current waveform at 1 kHz. 116 with argon appeared to have glitches, where the capacitor discharged slightly. It was noted that arcing within the MNSS analytical chamber was observed much more frequently with argon than with helium, suggesting the longer persist- ence of ions with the argon support gas. It was hypothe- sized that the persistence of ions in the analytical chamber provided a current leakage path during the capacitor recharging cycle, and thus causing the observed waveforms. To test this hypothesis, a high intensity xenon arc lamp was used to illuminate the spark gap. Such sources have been used by other workers to initiate gap breakdown, by inducing ionization (79). It was theorized that the ions produced by the xenon lamp would cause a reduction in capaci- tor potential and thus peak current by providing a leak— age path for the capacitor charging current. Results of the experiment revealed no change in capacitor potential or peak current. However, the spark discharge was shifted approximately 25 ns earlier in time, relative tc>thyratron triggering, as a result of the ionization produced by the lamp. The cause of the peculiar charging behavior ob- served with argon is still unknown and further study is necessary. 117 2. Chromatographic Evaluation a. Element-Selective Detection The element-selective detection capabilities of the thyratron-controlled MNSS were investigated by separating a mixture of organic compounds containing various hetero- atoms. The mixture prepared for testing the MNSS contained dibromomethane, cyclohexanone, tetraethyl orthosilicate, and iodotoluene. Methanol was used as the solvent. Methanol was chosen as the solvent Since this alcohol caused sig- nificant frequency changes in the free running MNSS, and the possible effects of this compound on the electronically- triggered MNSS were of interest. Separation of this mixture was achieved using a six-foot stainless steel column packed with 10% Carbowax 1540 on Chromosorb W. The volatilities of these compounds differed substantially. Therefore temperature programming was necessary to obtain satis- factory resolution within a reasonable time period. All Chromatograms presented were obtained using an initial oven temperature of 75°C, programmed to a final tempera- ture of 1600 C at a heating rate of 100 C/min. To il- lustrate the performance of the MNSS as a GC detector, Figure 24 shows separations of the same mixture obtained with a FID, and the MNSS when the 2478.6 A carbon emission line was monitored. As can be seen the two Chromatograms are virtually identical. However, some resolution has .on ou Hepomuwp mmzz mo COwAHwQEou .wm musmflm 02:58. E: 118 .§§E5§» 3. o... 3 2. 3 3 ca 3 2. em on 0.0 an o.~ o._ 0.0 _ _ L . _ p _ . b p 0.0 p p p h b p n p p j -095 fi/ I 0.80p M II o m m N a :98? B a u m r E I 0.000N I 07.; I. 0.08N 119 been lost in the MNSS chromatogram. This is a result of the need to maintain the coupler at the upper limit of the temperature program (1600 C), thus reducing the efficiency of that segment of the column housed in the coupler. Iden- tification of the various peaks in the carbon chromatogram is easily achieved by repeating the sample injection and monitoring the emission line of the various heteroatoms. Silicon emission was monitored at the 2516.1 A Si(I) line using an R166 solar blind photomultiplier tube. As can be seen from the silicon chromatogram, in Figure 25, the first peak after the solvent front in the carbon chromato- gram is the tetraethyl orthosilicate component. The peak represents 12.5 pg of silicon. Iodine emission was monitored using an R 166 PMT. The sensitivity of this PMT is substantially greater than that of a 1P28A at the 2062.4 g I (I) emission line used for iodine determinations. The iodine chromatogram identifies the last peak in the carbon chromatogram to be the iodo- toluene component. The peak represents 199 ug of iodine. The bromine chromatogram identifies the later eluting portion of the unresolved components in the carbon chromato- gram as the dibromomethane component. Bromine emission was monitored at the non-resonant 4477.7 A Br(I) emission line, with a 1P28A PMT. Poor bromine sensitivity was observed, and therefore high gain levels had to be employed. This high gain level resulted in the production of the noisy INTENSITY INTENSITY 790-0 2500 290-0 120 'OD'NE I- SILICON 4 15000 II I: y II I I I W i‘. t 10000 ~ g I II ('5 ‘ . .I ' Z I J g : z I I ’ .... . a i I 4 I I I (VI I K. I A _ ,5 j \A NYK J I; 00 h-r'V . wvw‘wwhw k...— 0.0 2D #0 6-0 8.0 1 00 0.0 2-0 4.0 6-0 M 1 00 TIME (minutes) TIME (minutes) . BROMINE 25000 . i ‘ OXYGEN _ I I '0] 20000 .4 ‘I 1 I E . I ~ I _ In I I I I I Z I I I _ II! E 15000 “‘ I I A. .3 . ' I .I -yH II H II” 'I III WNW . I I . I I. f I ‘ I k! Mr H. , II I. '7 " \ 1 “ I I III “IA“ II moo I III, .‘I ’4” I" . I . I ' VIII I .hmg4fi7 I I. ; . V f\ V'“ 'Vflw u L¢fi\\ I . I I . a 500-0 I I Y r r . 0-0 20 4-0 60 8-0 10-0 00 2.0 0.0 6-0 80 100 TIME (minutes) Figure 25. Element-selective response TIME (minutes) Chromatograms. 121 chromatogram. The peak represents 900 ug of bromine. Obviously, with this emission line, the MNSS is a poor bromine detector. Attempts to monitor bromine emission at one of the more intense non-resonance Br(I) lines reported by Fry et a1. (48) met with limited success. Fry reported 2 ICP excited Br(I) lines at 8247.4 3, and 6350.7 5. Both of these lines were more intense than the 4477.7 A line. Of these two lines the 8247.4 A line was the most intense, and was 2.5 times more intense than the 4477.7 A line. The energy of the upper state of each of these transitions is virtually the same as that of the 4477.7 A line, and all three transitions are allowed. Consequently, the spark source should be capable of generating bromine emission at all three of these emission lines. However, significant bromine emission could not be Observed at either the 8247.4 A or 6350.7 A lines. The inability to obtain significant emission at the two more intense emission lines was at- tributed to poor PMT sensitivity. The PMT used to monitor the "red" emission lines was an R666 PMT. This tube has a wide spectral response (1850 m 9100 A) with peak response at 3500 A. However, the sensitivity of this tube is ap- proximately a factor of 40 less than a 1P28. Additionally, the sensitivity of this tube drops off rapidly at wave- lengths longer than 8000 A (80). The ability to achieve reasonable bromine emission at 4477.7 A again is probably due to PMT response. The peak wavelength response of the 122 1P28A used is 4500.0 A, virtually the bromine emission line. Measurement of bromine emission at 4477.4 A is far from ideal. Background emission from the spark discharge is high in this Spectral region, and the background interferes with analytical determinations. This interference can be minimized, but not entirely eliminated, by selection of the proper integration time window. Improved bromine detec- tion with the MNSS may be possible using the more intense "red" emission lines, since the background continuum is very low in this region. However, further research, and a better red-sensitive PMT are necessary. The oxygen chromatogram shows three peaks; the methanol solvent, the oxygen containing tetraethyl orthosilicate, and the cyclohexanone components. The silicon chromatogram identified the first component after the solvent to be the orthosilicate component, therefore the third peak is the cyclohexanone component. Oxygen emission was monitored at the 7771.9 A O (I) line with an R666 red-sensitive PMT. The cyclohexanone peak represents 31 ug of oxygen. Of the elements analyzed, oxygen was the most peculiar in that the emission signal persisted for a relatively long time. This is illustrated in the plot of emission in- tensity versus delay time presented in Figure 26. For comparative purposes the lower plot in the figure is the time emission profile for nitrogen measured at 8216 A. As can be seen the oxygen Signal persists for Significantly 123 .smmouuH: Ucm cmm>xo How mafia mmHmc MN wuflmcqufl Hmcmfim 00003 mi; limo 0.0 9n meomt Z 0.0 06' odN 0.00 , 0.0.? .mm musmee 00003 mi; Elmo AllSNElNI BMW-Ba AllSNHlNI 3N1V'IEIEI 124 longer than the nitrogen signal. This is advantageous in that it permits long delay times, to allow the background continuum to decay, to be used without sacrificing signal. b. Background Correction As can be seen in all Chromatograms other than carbon, baseline disturbances occur whenever a component passes through the spark gap. This same effect was observed with the free-running MNSS, although it may not be quite as apparent in the Chromatograms presented earlier. This is a result of the amount of sample used in these earlier Chromatograms. In all the metal chelate Chromatograms, each peak represented 1 pg of metal. In the Chromatograms Obtained with the thyratron-controlled MNSS, the amount of each component was substantially greater, up to 3 orders of magnitude. Therefore this baseline disturbance is, as would be expected, concentration dependent. Furthermore, the magnitude of this disturbance is dependent on the spectral region being monitored. This baseline disturbance is certainly the result of a combination of a number of dif- ferent processes. However, two factors which most certainly play a role are plasma cooling, and molecular band emission. Plasma cooling results when components pass through the gap, and spark energy is consumed in the atomization and excitation processes. This results in a decrease or wave— length shift in the background continuum. This was 125 frequently observed when blanks were being run. When a blank was injected a negative peak would often result. Depending on the spectral region being monitored this negative peak would be smooth dip in the baseline (observed in the "red") or a negative dip followed by a positive deviation, usually observed in the UV-visible. Molecular band emission would account for the Observed positive baseline deviations. The occurrence of molecular band emission with atomic emission-based chromatographic detectors is not uncommon. This has been reported with both microwave-induced and inductive-coupled plasma sources when used as chromatographic detectors (81,47). Cyanogen band emission was Observed with the free running MNSS in the aluminum and copper determinations, when acetonitrile was used as the solvent. Moreover, attempts to use the MNSS as a chlorine-selective detector lend credence to the band emission hypothesis. When a mixture of trichloroethane in acetonitrile was injected into the MNSS, a well defined gaussian peak was observed when the 7246.7 A C1(I) emis- sion line was monitored. Injection of an acetonitrile blank produced only a small negative dip as it moved through the Spark. To verify that the suspected chlorine peak was indeed due to atomic chlorine emission, the monochromator was moved 10 A off the emission line (well outside the 3 A bandpass Of the monochromator), and the sample injection repeated. Surprisingly, a well defined Gaussian peak, 126 although of smaller area, was produced. This suggests that in addition to atomic chlorine emission, molecular band emission was contributing to the peak area. Injection of bromine and iodine compounds with retention times close to that of trichloroethane produced small baseline dis- turbances; however, no well defined peaks were observed, suggesting that some molecular chlorine containing species was being formed from the trichloroethane. This points out the need for background correction with the MNSS for true element-selective response to be realized. c. Analytical Results Detection limits, defined as the quantity of sample per second that gave a signal-to-noise ratio of 2, were determined for Br, C, I, O, and Si. Calculation of these detection limits were performed in the same manner as those Obtained with the free-running MNSS. Results are presented in Table III. These values are generally superior, by a factor of 5 to 10, to those obtained by Koeplin (2) with the free-running MNSS. These superior detection limits are probably more the result of the re- duced detector volume of this spark source than an increase in Spark power. The poorer carbon detection limit Obtained with the thyratron-controlled MNSS compared to those Ob- tained in the metal chelate analysis was the result of 127 Table III. MNSS Detection Limits. ELEMENT WAVELENGTH, A DETECTION LIMIT, (g/s) Bromine 4477.7 7 x 10'5 Carbon 2478.6 2 x lo‘8 Iodine 2062.4 1 x 10'7 Oxygen 7771.9 2 x 10-7 Silicon 2516.1 2 x lo"9 128 higher carbon background emission with the thyratron-trig- gered source. Arcing problems encountered with the thyra- tron-triggered MNSS required the inner brass chamber walls be covered with Teflon insulation. This substantially re- duced the arcing problem. However, addition of this insula- tion produced a high carbon background signal when the analytical chamber was heated. Calibration curves for oxygen, carbon, and silicon are presented in Figures 27-29. The use of a higher loading of stationary phase in the chromatographic column (10% Carbowax 1540) allowed a wider concentration range to be covered. As can be seen the silicon calibration curve is linear over the 0.25 to 25.0 mg range covered. However, both the carbon and oxygen calibration curves can be seen to begin to curve at higher concentrations. This is at- tributed to self absorbtion, since the curvature occurred when large amounts of analyte (mg range) were injected. 3. Conclusions The thyratron-controlled MNSS is a substantial improve- ment over the earlier free-running MNSS. No adverse ef- fects, on the MNSS, were observed with methanol or chlorinated hydrocarbons. Thus, this source can be ap- plied to the analysis of a wider range of compounds. Furthermore, this new source is truly easy to operate. 129 6.0 — o ‘5 ‘ o: (D O _J 4.5 — 4'0 T I r I I I I T I I l I 00 0.5 1.0 1.5 2.0 2.5 30 LOG (GRAMS INJECTED x105) Figure 27. Oxygen calibration curve. 6.7—- 6.2— 57-- L06 PEAK AREA 4.7-— 4.2 l 130 Figure 28. I I I I I ' I l 0.5 1.0 1.5 2,0 LOG (GRAMS INJECTED x105) Carbon calibration curve. 20 — 15 — 100 '— LOG PEAK HEIGHT 05‘— 00 131 Figure 29. I ’ I ' l ‘ l 05 1O 15 2a LOG (GRAMS INJECTED x 107) Silicon calibration curve. "J 'F . 132 The source no longer requires the constant attention of the Operator and can be left unattended for long periods of time. The new source is less sensitive to sheath gas flow rates and consequently higher hot sheath gas flow rates can be used. This allows higher analytical chamber temperatures to be achieved thereby allowing higher boiling compounds to be chromatographed without fear of condensa- tion in the MNSS. ‘;-¢.. .-i_.i . VI .A' .WWA- ‘L._... ~ F . PERSPECTIVES These studies have demonstrated the feasibility of employing a thyratron-triggered MNSS as an element-selec- tive GC detector. However, redesign of the MNSS is neces- sary. The present source was designed to use two spark gaps and, as a result of the modifications made to incor- porate the thyratron, it is difficult to change gap lengths and the bridging resistor. Also, the long electrode, now used to form a single gap with the original spark cathode, frequently breaks, and requires replacement. A new design should incorporate features that would permit replacement of electrodes, and changes in gap length to be easily performed. Additionally, the new source should allow higher wattage bridging resistors to be used. The present design limits the bridging resistor to 2 watts, since larger resistors Cannot fit in the MNSS housing. The new 133 analytical chamber should be smaller, with no exposed, internal metal surfaces. This smaller chamber would reduce argon consumption, and perhaps even further reduce the chromatographic band broadening caused by the MNSS. The removal of exposed metal surfaces within the chamber should eliminate the arcing problems that persist with the present design. Improved insulation of the analytical chamber would also be beneficial, in that higher chamber tempera- tures could be attained allowing higher boiling compounds to be chromatographed. Incorporation of an improved light collection system should substantially improve the detec- tion limits obtainable with the MNSS. Finally, some form of background correction would greatly enhance the element- specific detection capabilities of the MNSS. However, incorporation of background correction would substantially increase the complexity of the MNSS system and implementa- tion may prove difficult. VII. FEASIBILITY STUDIES ON USING A THYRATRON- CONTROLLED SPARK SOURCE AS A HIGH PERFORMANCE LIQUID CHROMATOGRAPHY DETECTOR A. INTRODUCTION High performance liquid chromatography (HPLC) has ex- perienced a rapid growth in recent years. This growth can be attributed to the power of HPLC to separate complex mix- tures that are either thermally unstable or exhibit in- sufficient vapor pressure for gas chromatographic separa- tion (82). One of the most serious deficiencies that has kept HPLC from becoming even more widely used is the lack of sensitive and selective detectors. Presently the two most popular detectors for HPLC systems are UV absorption and refractive index. Both detectors are rather unselec- tive, in that they respond to a wide variety of compounds, and both give little information about the chemical com- position of the LC effluent. Clearly the element-selective detection capability of the MNSS would make this source a valuable HPLC detector. The work of Lantz (1) demonstrated the feasibility of using the MNSS as an HPLC detector. In this work, coupling of the HPLC and the spark was achieved with a nebulization/desolvation sample introduction system. 134 ‘-".C. ‘. I 135 This approach, while successful, introduced significant dead volume and produced a loss in chromatographic ef- ficiency. This chapter reports the results of feasibility studies on an alternate HPLC-MNSS coupling scheme based on the spark-in-spary technique (51,52). B . DROPLET S I ZE MEASUREMENTS 1. Introduction Sample introduction, in the coupling scheme studied here was achieved by nebulizing the LC effluent to generate an aerosol. This aerosol was then swept into the heated spark chamber with a stream of hot argon. Heating of the spark chamber and carrier gas serves to facilitate aerosol desolvation and to prevent solvent accumulation within the spark chamber. This approach maximizes the amount of sample introduced and allows the spark source to be coupled to the HPLC with minimal dead volume. Efficient sample introduction is of the utmost importance in the HPLC- MNSS interface because of the small quantities of analyte present in the LC stream. In the course of the separa- tion process, the various mixture components are diluted. This dilution is a function of the retention time and can be quite significant. For example, if a mixture were 10-4 M in Ni, using a 25 pl sample, and assuming a peak volume of 0.5 ml at a 1.0 ml flow rate, then the average amount 136 of Ni flowing through the detector per second would be 4.81 x 10'9 g Ni/sec. This low analyte level demands that as much of the sample be introduced into the detector as possible. It has long been realized that where sample introduction is achieved by nebulization, not only nebulizer efficiency, but also the droplet size distribution can have a dramatic effect on the signal-to-noise ratio obtained by a par- ticular method. This is well documented for atomic absorp— tion and emission spectrometry (83,84). Two different nebulizers were used in the course of this work; a crossed—flow and an ultrasonic nebulizer. In order to characterize these nebulizers, measurements of the droplet size distributions of aqueous and methanol aerosols generated by the nebulizers were performed. For comparative purposes the droplet size distribution produced by the Veillon and Margoshes nebulizer (50) used by Lantz was also measured. Aerosol drOplet sizes were measured using the magnesium oxide method described by Kay (85). This method allows droplet sizes as low as 10 um to be determined by measuring the impressions made by the droplets when they strike an MgO layer deposited on a microscope slide. The microscope slides were prepared by burning Mg ribbon under the slide and moving the slide back and forth. This de- posits a smooth layer of MgO on the slide surface. The droplet size measurement was made by allowing the aerosol 137 to strike the slide surface. The slides were then examined with a 400x compound microscope. At this magnification, the droplet impressions appear as bright spots on the slide surface. Impression sizes were measured using a calibrated reticule eyepiece, and converted to droplet sizes by mul- tiplication by the calibration factor, given in Reference (85). The sizes measured were close to the failure limit of this method (10 um). Therefore, the absolute size measurements may be in error; however, they are still valuable for comparative purposes. 2. Veillon-Margoshes Nebulizer The Veillon—Margoshes nebulizer is of the concentric flow variety and as such requires a high gas flow rate for efficient operation (86). This high gas flow prevented the collection of the aerosol from being performed at the same distance as that used for measurements with the crossed- flow and ultrasonic nebulizers. The collection distance used was approximately 18 inches, three times that used for the other nebulizers. Attempts to collect the aerosol closer to the nebulizer aperature resulted in destruction of the MgO layer by the turbulent gas flow. As a result, the drOplet size measurements are perhaps biased in favor of the Veillon-Margoshes nebulizer, since it was found that the further away the aerosol was collected the smaller the mean drOplet diameter. Nevertheless, the aerosol 138 generated by the Veillon-Margoshes nebulizer had the largest mean drOplet diameter and standard deviation. The mean droplet diameter and standard deviation were 17.9 um and 6.3 um for the aqueous aerosol. For the methanol aerosol, the mean droplet diameter and standard deviation was 12.5 um and 3.3 pm respectively. The smaller mean droplet diam— eter for the methanol aerosol was observed with all the nebulizers tested and can be attributed to the lower sur— face tension of methanol (87). 3. Crossed-Flow Nebulizer The crossed-flow nebulizer performed quite well, even at gas flow rates as low as 0.5 l/m. The main problem en- countered with this nebulizer was that the small diameter solution capillary frequently clogged. However, the mean droplet diameter and the standard deviation of the droplets produced by this nebulizer were nearly equal to that pro- duced by the ultrasonic nebulizer. The mean droplet diameters and standard deviations for the aerosol, col- lected at a distance of 6 inches, were 14.6 um and 3.4 pm for the aqueous aerosol and 11.6 um and 2.3 pm for the methanol aerosol. 4. Ultrasonic Nebulizer As would be expected this nebulizer performed better than the others tested. The ultrasonic nebulizer produced 19-‘+';-_ 139 a very dense aerosol with a mean droplet diameter of 11.4 um and a standard deviation of 2.0 pm for the aqueous aerosol. The results of the methanol aerosol measure- ments revealed a mean droplet diameter of 8.9 um and a standard deviation of 2.6 pm. These results are slightly better than those obtained with the crossed-flow nebulizer. One major advantage of the ultrasonic nebulizer over the crossed-flow nebulizer is its efficiency. Examination of r1 the slides exposed to the ultrasonically produced aerosol g revealed many more droplet impressions than those ob- h tained with the crossed-flow nebulizer. As alluded to previously, efficient sample introduction is critical in the development of a viable HPLC-MNSS interface; thus, the higher efficiency of the ultrasonic nebulizer is perhaps as important as the smaller droplet size distribution generated by the nebulizer. An additional advantage of the ultrasonic nebulizer is the independence of aerosol genera- tion and gas flow. This is a distinct advantage over the pneumatic nebulizers. The only problem noted with the ultrasonic nebulizer was that the nebulization efficiency was, in some cases, highly dependent on the solution composition. For example, it was noted that a 65:35 mixture of water and tetrahydro- furan could not be efficiently nebulized. Nebulization efficiency increased as either of the components dominated (e.g., 20:80 water, THF). This is attributed to anomolous 140 viscosity effects. This type of anomolous solvent viscosity behavior, where the viscosity of a binary solution is greater than the viscosity of either pure component, is frequently observed in HPLC (74). This behavior is well documented for water:methanol solutions. Methanol:water combine to form a solution whose viscosity is substantially greater than the more viscous water component. The peak viscosity is observed at a 40:60 methanol water mixture and decreases as either component increases. This type of viscosity behavior may explain the loss of nebulization ef- ficiency with the 65:35 tetrahydrofuran-water solution. The change in solvent viscosity, and its effect on nebuliza- tion efficiency with certain solvent compositions, may make the use of the ultrasonic nebulizer difficult with gradient elution. C . SOLVENT EFFECTS It was noted in the course of this work that certain solvents were tolerated by the spark much better than others. For example, water and acetonitrile and combina- tions of the two were well tolerated by the MNSS. However, tetrahydrofuran caused the spark discharge to become un- stable when Operated at the same potential where aceto- nitrile was well tolerated. Spark stability could be restored by increasing the power supply voltage, and thus overvolting the spark gap. There appeared to be a certain 141 potential range where the solvent effects on the spark discharge were minimized. That is, for different solvents maximum spark stability was attained at different power supply voltages. Further research is necessary into the effects of overvolting the spark source, since the per- sistent arcing problem associated with the MNSS limited the degree of overvolting which could be achieved and, therefore, prevented a detailed study. In addition to destabilization of the spark discharge by certain solvents, the degree of arcing observed within the spark source was solvent dependent. Again water and acetonitrile were the best solvents, in that they did not aggrevate the arcing problem. Arcing problems became more pronounced when methanol or tetrahydrofuran were introduced into the spark source. The solvent dependence of the arcing is perhaps due to the persistence of ions in the chamber when certain solvents are used. A potential problem with the use of the MNSS as an HPLC detector is the fouling of the spark electrodes with car- bonaceous deposits from large amounts of organic solvents. Proposals to use the MNSS as element-selective HPLC detector for metal speciation studies on petroleum products demand that the MNSS be capable of tolerating not only reverse- phase, but also normal-phase solvent systems. Therefore, the possible electrode fouling effects of both types of solvent systems were studied. Fouling measurements were 142 made by connecting the LC effluent stream (1.0 ml/min) to the ultrasonic nebulizer and introducing the resulting aerosol into the heated spark chamber. The MNSS was allowed to run for 1.5 hours and the electrodes were then inspected. When pure acetonitrile, a common reverse phase solvent, was introduced into the spark source, large carbon deposits were formed in less than an hour. These deposits actually became so large that they virtually bridged the spark gap. The electrodes were cleaned, and the experiment was repeated with a 10:90 solution of water—acetonitrile. The results indicated that the addition of the small amount of water substantially reduced deposit formation; however, significant deposits were still formed over the 1.5 hour experiment period. Addition of 20% water, however, vir- tually eliminated deposit formation, and, after 1.5 hours of Operation, no significant carbon deposits were observed. This cleansing effect of water is attributed to the formation of free oxygen in the Spark discharge, which then oxidizes the carbon deposits on the electrodes. These results indi- cate that no significant problem with electrode fouling is foreseen with reverse-phase chromatographic solvent systems. The need to analyze petroleum products demands that the MNSS be capable of analyzing normal-phase LC effluents. A typical solvent system used in the separation of petroleum products consists of a saturated hydrocarbon with a small amount of polar modifier (5-10%), like chloroform. To 143 determine the effect of such a solvent system on the MNSS, a solution of 90% heptane and 10% chloroform was prepared and nebulized into the spark source at a rate of 1.0 ml/ min. As expected, carbon deposits were formed in less than one-half hour. To prevent the formation of carbon deposits on the walls of quartz plasma cavities used with microwave induced plasma GC detectors, oxygen has been used as a scavenger gas (88). Experiments were performed to evaluate the effectiveness of this technique with the MNSS. A small amount of oxygen (250 ml/min) was added to the 4.0 l/min argon sheath gas, and the experiment repeated. This did not eliminate deposit formation, but did reduce deposit formation to a tolerable level over the 1.5 hour time period of the experiment. The reduction of the deposit formation, however, was at the expense of spark discharge stability. The addition of oxygen did result in some dis- charge instability, but the degree of instability was tol- erable. Thus, by adding small amounts of oxygen, the MNSS should be capable of handling normal—phase solvent systems; however, frequent cleaning of the electrodes will be re- quired. D . TIME RESOLUTION The effects of the liquid sample form and the influence of the thyratron on the time of maximum analyte emission 144 were investigated. Initial studies on the MNSS-HPLC interface were performed with the free-running, double-gap MNSS. To determine the time of maximum COpper emission an aqueous solution of copper nitrate was prepared. This solution was aspirated into the MNSS and the copper emis— sion monitored at different delay times. The results were plotted and are presented in Figure 30. As can be seen in the figure, the time of maximum COpper emission is the same as that observed in the gas phase determinations. However, the persistence of the signal appears to be significantly shorter than in the gas phase study. This shorter signal duration for samples introduced as aerosols was also ob- served for aluminum and carbon, and may be the result of less energy being available for excitation, since the spark discharge must now desolvate the sample aerosol. Alternatively, the solvent may cool the spark plasma rapidly and lead to a loss in excitation energy. The copper time-resolution experiment was repeated with the thyratron-controlled MNSS. Additionally, in order to determine the influence of support gas, this experiment was performed with both argon and helium. Results ob- tained using argon are presented in Figure 31. As can be seen in the figure the time of maximum copper emission is approximately the same with the thyratron-controlled MNSS as with the free-running source. The most significant difference observed was that the signal intensity observed 145 1200.0 - 1000.0 -— t 000.0 ~— ('7) g .I Z 800.0 - m 2 is n LIJ 0: 400-0 -— 200-0 — 000 I I l I 1 I I _I 0.0 0.5 1.0 1.5 2.0 DELAY TIME (usec) Figure 30. Signal intensity XE delay time for copper, with free-running MNSS. 146 1600.0 -— o 1400.0 — 1200.0 - 1000.0 -4 800.0- RELATIVE INTENSITY 400.0 -— 200.0 — 00 . I . I , I I ¥_j 0.0 1.0 2.0 3.0 *0 DELAY TIME (usec.) Figure 31. Signal intensity gs delay time for copper, with thyratron-controlled MNSS. 147 with the thyratron-controlled source is large early in time and persists for a longer time period than observed with the free-running source. This is attributed to a higher energy spark discharge with the thyratron-controlled source, since the capacitor discharge energy is now dissi— pated in a single spark gap. The results obtained using helium are presented in Figure 32. As can be seen in the figure, the analyte signal observed in helium reaches its maximum value immediately after the Spark discharge and decays very quickly. This suggests a less energetic spark discharge, which correlates with the lower peak currents observed in helium. E. CONCLUSIONS This work has indicated that the thyratron-controlled spark source can tolerate large amounts of solvent without serious perturbation of the spark discharge. However, the source must be redesigned to eliminate the arcing problem that persists in the present source. The proposed MNSS- HPLC interface appears to be viable. The use of the high efficiency ultrasonic nebulizer should allow maximum sample introduction, which is crucial for achievement of the desired detection limits. The introduction of large amounts of solvent does not appear to result in significant problems with electrode fouling. Incorporation of the 148 3000—1 — HELIUM 250.0 - __'_ It! .' ‘I-_ if.” -" 150.0 -— RELATIVE INTENSITY 100.0 — 00 I I . I , I , I i °°° 0’5 w 1.5 2.0 DELAY TIME (useo) Figure 32. Signal intensity XE delay time for copper. 149 features described in the preceding chapter into a new spark source design should make the thyratron-controlled MNSS a viable HPLC detector. Further discussion of pos- sible improvements in both the spark source and the chromato- graphic system is presented in the following chapter. VI I I . COMMENTARY This work has demonstrated the value of the MNSS as an element-selective gas chromatographic detector. The present interface allows the MNSS to be placed outside of the GC oven and does not seriously compromise chromatographic performance. The addition of electronic control of the spark firing frequency has made the source truly easy to operate. A newly designed source, incorporating the features outlined in the perspectives of Chapter VI, should allow easier source maintenance and perhaps improve detec- tion limits. These improvements should make the MNSS a truly valuable gas chromatographic detector. The viability of the thyratron-triggered MNSS as an element-selective HPLC detector is still uncertain. Again redesign of the source is necessary. Elimination of the persistent arcing problem with the present design should allow a more extensive investigation of the potential of this source. Beyond the spark source itself, much work remains to be performed on the HPLC-MNSS interface and the HPLC system. The HPLC-MNSS interfacing approach based on nebulization of the LC effluent into the heated spark chamber appears viable. The major thrust in interface development should be directed toward increasing sample 150 m: 151 input. The low analyte levels involved demand that vir- tually all the LC effluent stream be nebulized and that all of the resulting aerosol reach the spark source. In the course of this work it was noted that a substantial portion of the aerosol would collect on the walls of the tubing coupling the nebulizer to the MNSS. This accumula- tion, and resulting sample loss, could be minimized by employing a heated tube to connect the nebulizer and the MNSS. An ideal material for this purpose would be a seg- ment of fused silica, which is currently finding a great deal of use for GC capillary columns. Fused silica can withstand high temperatures, is flexible, and has very low catalytic activity. Thus, this material could be heated to high temperatures without fear of reactions taking place on the hot tube walls. Additionally, the heat supplied by the coupling tube would promote aerosol desolvation, since it was noted that the hot argon carrier gas did not possess sufficient thermal energy to desolvate the large amount of sample aerosol generated at the nominally employed 1.0 ml/min flow rate. The recent developments in column switching and micro- bore column technology should be exploited to enhance the capabilities of the MNSS. Microbore HPLC columns are now commercially available and can be used with the exist- ing HPLC pumping system without substantial modifications. An advantage of the microcolumn technique, in this 152 application, are the low flow rates (50-100 ul/min) used with these columns. This low solvent flow rate would allow better desolvation of the sample aerosol by the hot argon carrier gas and would free the spark from the energy con- suming desolvation step. Additionally, by virtue of this low solvent flow, perturbation of the spark discharge will be reduced and electrode fouling should be eliminated. However, the use of these columns requires that dead volumes, and extra-column volumes in the chromatographic system, be minimized. Thus, modification of the present sample in- jector, UV detector, and connecting lines must be performed. The dead volume of the HPLC-MNSS interface will also have to be minimized, although the conversion of the LC stream to an aerosol and the rapid sweeping of the aerosol through the detector reduces the dead volume constraints on this portion of the chromatographic system. The use of micro- bore HPLC columns is not without sacrifices. The small bed volumes of microbore columns prevent large samples from being loaded onto the column, and this could result in analyte levels below the detection limits of the MNSS. The loading restriction can be overcome by implementation of column switching techniques. A possible scheme would be to use a standard analytical HPLC column, onto which a relatively large sample could be charged, for initial separa- tion. Then that segment of interest in the resulting chromatogram could be cut and switched to the microbore 153 column. This would isolate those components Of interest and allow a relatively large amount of these components to be charged on to the microbore column. Additionally, the compound group type of interest could be concentrated on the analytical column and then be eluted onto the micro- bore column, permitting higher analyte levels to be intro- duced into the MNSS. With the incorporation of these tech- niques and the redesign of the spark source, the HPLC-MNSS has the potential of being a valuable tool in metal specia- tion investigations. REFERENCES l. 10. 11. 12. 13. 14. 15. 16. REFERENCES R. L. Lantz, Ph.D. Thesis, Michigan State University, East Lansing, MI 1977. S. M. Koeplin, Ph.D. Thesis, Michigan State University, East Lansing, MI 1978. Y. K. Chau, P. T. S. Wong, and H. Saitoh, J. Chrom. Sci. 14, 162 (1976). D. Segar, Analyt. Letters 7, 89 (1974). W. Wolf, J. of Chrom. 134, 159 (1977). G. Parris, W. Blair, and F. Brinckmann, Anal. Chem. 49, 378 (1977). M. O. Andree, Anal. Chem. 49, 820 (1977). D. R. Jones, and S. E. Manahan, Anal. Chem. 48, 503 (1976). J. C. Van Loon, Anal. Chem. 51, 1139A (1979). N. Yoza, and S. Ohashi, Analyt. Letters 6, 595 (1973). Y. K. Chau, P. T. S. Wong, and P. D. Goulden, Anal. Chim. Acta 85, 421 (1976). Y. K. Chau, P. T. S. Wong, and P. D. Goulden, Anal. Chem. 47, 2279 (1975). E. J. Parks, F. E. Brinkmann, and W. R. Blair, J. of Chrom. 185, 563 (1979). A. Y. Cantillo, and D. A. Segar, Proceeding of the International Conference on Heavy Metals in the En- vironment. Toronto, (1975). T. M. Vickery, H. E. Howell, G. V. Harrison, and G. J. Ramelow, Anal. Chem. 52, 1743 (1980). H. K. Koizumi, R. D. Laughin, and T. Hadeishi, Anal. Chem. 51, 387 (1979). 154 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 29. 29. 30. 31. 32. 33. 34. 35. 155 F. E. Brinkmann, W. R. Blair, K. L. Jewett, and W. P. Iverson, J. Chrom. Sci. 15, 493 (1977). S. S. Brody, and J. E. Chaney, J. of Gas Chrom. 4, 42 (1966). W. A. Aue, and C. G. Flinn, Anal. Chem. 52, 1537 (1980). R. S. Juvet, Jr., R. P. Durbin, Anal. Chem. 38, 565 (1966). W. Aue, J. Chrom. Sci. 13, 329 (1975). D. A. Clay, C. H. Rodgers, and R. H. Jungers, Anal. Chem. 49, 126 (1977). S. O. Farwell, and R. A. Rasmussen, J. Chrom. Sci. 14, 224 (1976). F. M. Zado and R. S. Juvet, Jr., Anal. Chem. 38, 569 (1966). T. L. Chester, Anal. Chem. 52, 638 (1980). P. L. Patterson, R. L. Howe, and A. Abu—Shumays, Anal. Chem. 50, 339 (1979). P. L. Patterson, Anal. Chem. 50, 345 (1978). T. L. Chester, Anal. Chem. 52, 1621 (1980). F. C. Fehsenfeld, K. M. Evenson, and H. P. Broida, Rev. Of Sci. Instr. 36, 294 (1965). S. Greenfield, H. McGeachin, and P. B. Smith, Talanta 22, 553 (1975). C. I. M. Beenaker, Spectrochim. Acta Part B 31, 483 (1976). C. I. M. Beenaker, Spectrochim. Acta Part B 33, 53 (1978). D. C. reamer, W. H. Zoeller, and T. C. O'Haver, Anal. Chem. 50, 1449 (1978). B. D. Quimby, P. C. Uden, and R. M. Barnes, Anal. Chem. 50, 2112 (1978). Y. Talmi, and D. T. Bostick, Anal. Chem. 47, 2145 (1975). 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 156 B. D. Quimby, M. F. Delaney, P. C. Uden, and R. M. Barnes, Anal. Chem. 51, 875 (1979). C. I. M. Beenaker, Spectrochim. Acta Part B 32, 173 (1977). . P. C. Uden, R. M. Barnes, and F. P. DiSanzo, Anal. Chem. 50, 852 (1978). R. J. Loyld, R. M. Barnes, P. C. Uden, and W. G. Elliot, Anal. Chem. 50, 2025 (1978). C. H. Gast, J. C. Kraak, H. Poppe, and F. J. M. M. Maessen, J. of Chrom. 185, 549 (1979). P. C. Uden, and I. E. Bigley, Anal. Chim. Acta 94, 29 (1977). R. J. Decker, Spectrochim. Acta Part B 35, 19 (1980). P. C. Uden, B. D. Quimby, R. M. Barnes, and W. G. Elliot, Anal. Chim. Acta 101, 99 (1978). D. M. Fraley, D. Yates, and S. E. Manahan, Anal. Chem. 51, 2225 (1979). M. Morita, T. Uehiro, and K. Fuwa, Anal. Chem. 52, 349 (1980). R. M. Brown, Jr., and R. C. Fry, Anal. Chem. 53, 532, (1981). R. M. Brown, Jr., S. R. Norway, and R. C. Fry, Anal. Chem. 53, 934 (1981). S. K. Hughes, and R. C. Fry, Anal. Chem. 53, 1111 (1981). D. L. Windsor, and M. B. Denton, Anal. Chem. 51, 1116 (1979). C. Veillon, and M. Margoshes, Spectrochim. Acta Part B 23, 553 (1968). A Schlge, Ph.D. Thesis, University of Illinois, Urbana, IL (1959). R. G. Scholz, and H. V. Malmstadt, Anal. Chem. 27, 881 (1955). A. Bardocz, Spectrochim. Acta 7, 307 (1955). D. A. Sinclair, and R. N. Whitten, Spectrochim. Acta 16, 760 (1960). 55. 56. 57. 58. 59. 60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 72. 157 T. Nakajima, and H. Kawaguchi, Spectrochim. Acta 18, 201 (1962). D. M. Coleman, and J. P. Walters, Spectrochim. Acta Part B 31, 547 (1976). D. L. Donohue, and J. A. Carter, Anal. Chem. 50, 686 (1978). M. B. Denton, and D. B. Swartz, Rev. Sci. Instrum. 45, 81 (1974). D. R. Christmann, Ph.D. Thesis, Michigan State Uni- versity, East Lansing, MI (1980). J. T. Gano, Ph.D. Thesis, Michigan State University, East Lansing, MI (1981). G. T. Seng, Ph.D. Thesis, Michigan State University, East Lansing, MI (1978). J. Zynger, Ph.D. Thesis, Michigan State University, East Lansing, MI (1973). G. Gonzalez "FORTX", Hearing Research Lab., University of Minnesota, June (1978). A. Savitzky, and M. J. E. Golay, Anal. Chem. 36, 1627 (1964). J. F. Holler, Ph.D. Thesis, Michigan State University, East Lansing, MI (1977). F. E. DeBoer, Nature 185, 915 (1960). C. S. G. Phillips, and P. L. Timns, Anal. Chem. 35, 505 (1963). R. W. Moshier, and R. E. Sievers, "Gas Chromatography of Metal Chelates", Peramon Press, Oxford, (1965). R. D. Hill, and H. J. Gesser, J. of Gas Chrom. 1, 11 (1963). W. J. Bier, and H. J. Gesser, Anal. Chem. 32, 1525 (1960). P. C. Uden, and D. E. Henderson, Analyst 102, 889 (1977). B. L. Karger, L. R. Synder, and C. Horvath, "An Introduc- tion to Separation Science", Wiley-Interscience, New York, NY (1973). 73. 74. 75. 76. 77. 78. 79. 80. 81. 82. 83. 84. 85. 86. 87. 88. 158 S. Mandelstam, Spectrochim. Acta 15, 255 (1959). L. R. Synder, and J. J. Kirkland "Introduction to Modern Liquid Chromatography", 2nd Edition, Wiley-Interscience, New York, NY (1979). R. Belcher, R. J. Martin, W. I. Stephen. D. E. Hender- son, A. Kamalizad, and P. C. Uden, Anal. Chem. 45, 1197 (1973). K. M. Mackay, and R. A. Mackay, "Introduction to Modern Inorganic Chemistry", Intext, New York, NY (1969). E. L. Grove, "Analytical Emission Spectroscopy", Part II, Marcel Dekker, New York, NY (1972). M. Rabb, Michigan State University, East Lansing, MI, personal communication (1981). H. Kaiser, Spectrochim. Acta 1, 477 (1941). Hamamatsu TV Co. LTD. Photomultiplier Catalog (1979). A. J. McCormack. S. C. Tong, and W. D. Cooke, Anal. Chem. 37, 1470 (1965). R. J. Hamilton, P. A. Sewell, "Introduction to High Performance Liquid Chromatography", Wiley-Interscience, New York, NY (1978). J. B. Willis, Spectrochim Acta Part A 23, 811 (1979). K. W. Olsen, W. J. Haas, Jr., V. A. Fassel, Anal. Chem. 49, 632 (1977). K. R. Kay, J. of Sci. Instr. 27, 128 (1950). J. W. Novak, Jr., and R. F. Browner, Anal. Chem., 52, 792 (1980). R. Mavrodineau, and H. Boiteux "Flame Spectroscopy", J. Wiley and Sons, New York, NY (1965). W. R. McLean Slanton and G. E. Penketh, Analyst. (London) 98, 432 (1973).