, .V. ‘. . "3.. r. . rw :yodm‘ O'DI‘ .-_._-_00 p- 1‘ J'— g...‘ .. ... . u. 'r -. ‘ . ~ A ,5 - I ‘ ”2:37. ‘ . 9311-313? (q...- ". U'" AW) 1. .1" ”.7115. N - ‘ ‘ 1:“... .‘ - '. 35.“ ‘14-}; A‘:~ . :‘3’ x "11" - , H Lv- . 3.. ,‘- .. :1“? - -v;- 7 51.3 ‘3.“ -- 5a. fiqu‘kflf . .; , . m. . ‘H 1 A . - - 2‘ . ' .1 '11.?“ ' “‘ - in f- . ‘ ~ ' -' - ~ T «ram-1313‘??? 75.2 m: » - ’ v“‘-' ““"‘- * ... 35k? a: :5 “LL-:5 W‘:““V n-“ « N *3. www.cg N. am V'Qif'fi ”H ”193,3“; ”33,1" N . _ .- .. ‘II'. ”3' wk: 3"“’1“'=W§“ l "4?." ' .. vGum \i‘ I. 33% W“ 3‘54? 1;. i§-g *j'EESV ' N u- 11'. '1 ‘V v..' 1,: h! ,k ‘ I '1 - n “4);? 4"“ ’5 5'3. LI BRA R Y ‘Michigan State a University This is to certify that the thesis entitled INVESTIGATIONS OF A SPARK EMISSION DETECTOR FOR GAS CHROMATOGRAPHIC ANALYSIS presented by Sandra Marie Koeplin has been accepted towards fulfillment of the requirements for Ph . D. degree in Chemistfl Major professor Z 7/ Date flj 7'5 0-7639 INVESTIGATIONS OF A SPARK EMISSION DETECTOR FOR GAS CHROMATOGRAPHIC ANALYSIS By Sandra Marie Koeplin A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Chemistry 1978 ABSTRACT INVESTIGATIONS OF A SPARK EMISSION DETECTOR FOR GAS CHROMATOGRAPHIC ANALYSIS By Sandra Marie Koeplin The design, construction and operation of a spark emission gas chromatographic detector is described. The effluent from the chromatographic column is run directly into a spark chamber. The characteristic emission which results from spark excitation is isolated by a mono- chromator, and the intensity is measured by a photomulti- plier tube. The spark emission detector (SED) is shown to be use- ful for detection of boron, bromine, carbon, hydrogen, iodine, nitrogen, oxygen, phosphorus, silicon and sulfur with sensitivities in the microgram to nanogram range. The atomic emission lines, used to monitor emission, detection limits and dynamic range for each element in- vestigated are presented. The SED is shown to be reliable to within 2.8% over five chromatographic separations and relatively free from effects due to carbon chain length or bond type. Sandra Marie Koeplin The usefulness of the SED for multielement analysis is demonstrated with several examples. The empirical formulas of several alcohols are determined using the SED. The SED is used to simplify complicated chromatograms ob— tained with universal detectors by responding to only those compounds containing a particular element, and to separate compounds that cannot be separated by a universal detector. The ability of the SED to uncover peaks hidden by larger interfering peaks is also demonstrated. The software system design and FORTRAN IV programs for the SED chromatographic system are described. The software package allows for control of a scanning mono- chromator for multielement analysis and a box-car integrator for time resolved spectroscopy. In addition, raw chromato- graphic data are obtained, and retention times and areas calculated under software control. ACKNOWLEDGMENTS I wish to thank Dr. Stan Crouch for serving as my research advisor and Dr. Andy Timnick for serving as my second reader. I also wish to thank the Crouch Group for many enjoy- able times over the past five years. My special thanks to Roy Gall, Gary Seng and Marty DiStasio for their many help- ful discussions, and friendship, without them graduate school would have been less meaningful and much more dif- ficult. And finally, but most of all, I wish to thank my family, especially my parents, for their love, support and encouragement throughout these many years. My degree would not mean as much if it weren't for them. ii To Mom and Dad Carol, John, Rob, Jim and Tom iii TABLE OF CONTENTS Chapter Page LIST OF TABLES. . . . . . . . . . . . . . . . . .viii LIST OF FIGURES . . . . . . . . . . . . . . . . . ix CHAPTER I - Introduction. . . . . . . CHAPTER II - Background . . . . . . A. Ionization Detectors. 1. Thermionic Detectors. comma-H 2. Electron Capture Detector B. Electrochemical Detectors . . . . . . . . 10 C. Optical Emission and Absorption Detectors . . . . . . . . . . . . . . . . 12 1. Microwave Plasma Emission Detector. . . . . . . . . . . . . . . 12 Atomic Absorption Detectors . . . . . 1“ Discharge Emission Detectors... . . . 15 Chemiluminescence Detectors . . . . . l6 U'IJZ'UUN Molecular Fluorescence Detectors . . . . . . . . . . . . . . 18 6. Ultraviolet Detector. . . . . . . . . 19 D. Miscellaneous Detectors . . . . . . . . . 2O 1. IR and NMR Detectors. . . . . . . . . 2O 2. Mass Spectrometer and Plasma Detectors . . . . . . . . . . . . . . 21 3. Piezoelectric Sorption Detector. . . . . . . . . . . . . . . 21 A. Other Detectors . . . . . . . . . . . 22 E. Summary . . . . . . . . . . . . . . . . . 22 CHAPTER III — Spark Characteristics . . . . . . . 25 iv Chapter A. E. CHAPTER A. B. C. D. E. F. Design Considerations . . . . . 1. Double Gap Design 2. Power Supply. 3. Spatial Stability Spark Chamber Construction. 1. Capacitor Construction. 2. Auxiliary or Control Gap Construction. . . . . . . 3. Analytical Gap Construction A. Spark Gap Measurement . Electrical Characteristics. 1. General Circuit Description 2. Capacitance Measurements. Spark Current Measurements. 1. Current Waveform Production 2. Current Characteristics of the Spark Argon Background Emission IV — Instrumentation. Overview of the SED System. Chromatographic Components. Spectrometer. . . . . . . . . Supporting Electronics. Optical Trigger Interface . CHAPTER V - Software. A. B. C. Choice of Programming Language. Overview of FORTRAN IV. FORTRAN IV Overlay Structure. V Page 25 26 27 28 29 31 32 32 33 3A 3A 36 37 37 HO “5 A8 A8 50 53 5M 57 61 65 67 68 69 Chapter Page D. Software System Description . . . . . . . 70 E. Data Acquisition Overlay. . . . . . . . . 7O 12 General Description . . . . . . . . . 7O 2. Parameter Input Overlay . . . . . . . 76 3. Line Finding Overlay. . . . . . . . . 77 A. Analog to Digital Acquisition Overlay . . . . . . . . . . 81 5. Data Storage Overlay. . . . . . . . . 85 F. Data Analysis Overlay . . . . . . . . . . 86 General Description . . . . . . . . . 86 2. Data Smooth Overlay . . . . . . . . . 9O Plotting and Interactive Overlay . . . . . . . . . . . . 92 N. Area Determination Overlay. . . . . . 9M 5. Report Generation Overlay . . . . . . 95 G. Error Analysis Overlay. . . . . . . . . . 97 CHAPTER VI - The Spark as a Selective G.C. Detector. . . . . . . . . . . . 100 A. Overview of SED Performance . . . . . . . 100 B. Effects of Gas Flow Rates on the SED. . . 10A C. SED Reproducibility and Stability . . . . 106 D. Time Resolution Studies for Elements of Interest. . . . . . . . . . . 110 E. Carbon Response . . . . . . . . . . . . . 117 1. Detection Limit and Dynamic Range . . . . . . . . . . 117 2. Structural Effects. . . . . . . . . . 118 F. Selective Detection Capabilities .. . . . 120 1. Boron . . . . . . . . . . . . . . . . 120 2. Halogens. . . . . . . . . . . . . . . 121 vi Chapter mummzw Hydrogen. Nitrogen. Oxygen. . Phosphorus. Silicon Sulfur. Summary CHAPTER VII - Multielement Studies. A. B. Multielement Detectors. Examples of Multielement Capabilities of the SED 1. Separation of TMS, Carbon Disulfide and Benzene 2. Separation of Methanol, 1— Propanol and 1-Butanol. 3. Separation of Ethyl Bromide Iodomethane and Nitromethane. A. Summary CHAPTER VIII - Commentary . . . . . REFERENCES. APPENDICES. A. B. Selected Program Listings Data Acquisition and Interface Circuits. . . . . . . . vii Page 123 123 123 12A 12A 127 127 129 129 131 131 133 133 137 138 1U1 1A8 1A8 183 Table LIST OF TABLES Computer Command Set for the Spark System . Wavelengths of Elements Detected With SED Solutions Used for Determination of Detection Limit and Linear Range. Typical Operating Parameters of the SED. Optimum Delay and Integration Times. Response of Carbon Containing Compounds at Carbon 2A78.6A Emission Line. Detection Limits with the SED. Empirical Formula Determination with the SED . . . . . . . . . viii Page 63 102 103 105 113 119 128 135 Figure 10 ll 12 LIST OF FIGURES Diagram of spark chamber construc~ tion . . . . . Simplified circuit diagram of the nanosecond spark An example of digitized oscillo- scope voltage vs time data . . . . Current dependence on carrier gas. . . . . . Current dependence on DC supply voltage. . . . Current dependence on gap length Time resolved spectra of argon background emission. Diagram of the Spark Emission Detector system. . . . . . Photograph of the SED inside the chromatographic oven . . . . . . . Block diagram of the SED data acquisition system . . . . . . Optical trigger systems for the SED. . . . . . . . . . . . . . . . Data acquisition overlay ix Page 30 35 39 A1 A2 AA A6 A9 51 55 59 71 Figure Page 13 Flow diagram of data acquisition overlay. . . . . .1. . . . . . . . . . . 7A 1A Data analysis overlay. . . . . . . . . . 87 15 Flow diagram of data analysis overlay. . . . . . . . . . . . . . . . . 88 16 Sample output from SKLIST. . . . . . . . 96 17 Flow diagram of error analysis overlay. . . . . . . . . . . . . . . . . 98 18 Sample output from SKEROR. . . . . . . . 99 19 Van Deemter plot for a Porapak Q column . . . . . . . . . . . . . . . . 107 20 An example of SED reproducibility. . . . 108 21 Signal stability with the SED. . . . . . 109 22 Signal stability with large sample quantities . . . . . . . . . . . . . . . 111 23 A graph of signal intensity vs delay time for phosphorus. . . . . . . . 11“ 2A A graph of signal intensity vs delay time for carbon. . . . . . . . . . 115 25 A graph of signal intensity vs integration time for phosphorus. . . . . 116 26 Calibration curve for iodine . . . . . . 122 27 Calibration curve for phosphorus . . . . 125 28 Calibration curve for silicon. . . . . . 126 Figure 29 3O 31 B1 B2 B3 BA B5 B6 B7 Separation of TMS, carbon di- sulfide and benzene. Separation of methanol, 1- propanol, and 1—butanol. Separation of ethyl bromide, methyl iodide and nitromethane. Delay Scaler Circuit Integrate Scaler Circuit Analog integrate circuit Clock/initialize/sample and hold circuitry. Primary interface circuit. Monochromator interface circuit. Monochromator encoder monitor. xi Page 132 13A 136 CHAPTER I INTRODUCTION Gas chromatography has come to be one of the most widely used techniques in the modern analytical labora- tory. Although, with the onset of increasingly more complex biological and environmental samples, the job of the chromatographer has become more difficult. The separations obtained are not always clean and easy to interpret; thus, there has been a greater emphasis on improving separation techniques through improved column material, column packing and liquid phase technology. In addition, there has been a move towards improving separation quality through the use of automated control of operating parameters and electronic data handling. Another area of investigation has been involved with new detector technology. The overall effort in chroma- tography has been towards making complex separations easier to accomplish. This work deals with an element selective detector to simplify chromatographic separations. Element selec- tive detectors have become very popular due to their ability to ignore all compounds eluting from the column except the compound containing the element of interest. The element selective detector in addition to being selective, must be reliable, accurate, easy to use and low in cost Just like any other chromatographic detector. One further characteristic that would be useful for an element selective detector would be tunability of response for any specific element of interest. This characteristic would eliminate one of the primary inconveniences of many element selective detectors; the need for a different detector for each element of interest. If the detector were tunable, only one detector would be needed to handle all elements of interest. Since elements emit radiant energy at characteristic wavelengths, the development of a detector involving some sort of emission process would be one approach to a tunable element selective detector. The emission for each element could be monitored by using a monochromator adjusted to the proper wavelength with a photomultiplier tube as the radiation transducer. This work deals with such a chromatographic detector. The detector to be studied can be called a spark emission detector since the effluent from the gas chroma- tographic column is run directly into a spark discharge. The characteristic emission which results from spark excitation is isolated by a monochromator, and the in- tensity is measured by a photomultiplier tube. If the spark detector is used along with a radiation transducer that monitors more than one wavelength, such as a linear diode array, true simultaneous multielement gas chroma- tography would be feasible. The primary advantage of this type of system would be the ease with which empiri- cal formulas for the compounds could be determined. The design, construction and detector characteris- tics of the spark emission detector are discussed in this work. The ordinary operation and computer control routines are also described. In addition, the tunability, selectivity, reliability and general usefulness of the spark emission detector for gas chromatographic analysis are demonstrated by experimental examples. CHAPTER II BACKGROUND Since gas chromatography can provide both qualita— tive and quantitative information about the composition of mixtures, it has become one of the most widely used modern analytical techniques. However, with the advent of increasingly more complex biological and environmental samples, the pertinent information from a chromatogram is becoming more difficult to extract. Chromatograms may have not only a large number of peaks, but also over— lapping peaks, or large background signals which may obscure many peaks. These things combine to make peak identification very difficult for the chromatographer. Clearly, a means of simplifying the chromatogram would greatly simplify and enhance chromatographic analyses. Many types of improvements in chromatographic science are available, but this discussion focuses on improving chromatographic analysis through the use of selective detectors, since the more selective the de— tector, the poorer the chromatographic separation can be. The best detector would be one which conclusively identified the specific compound and the amount of com- pound eluting from the column. Unfortunately, nothing quite this specific is available. Many methods have been tried such as using a male gypsy moth to detect the female gypsy moth sex attractant or employing pro- fessional "sniffers" to identify the compounds eluting from the chromatographic column. While these methods have merit for qualitative analysis, they are not quanti- tative and, in addition,app1y only to a very specific analysis. The completely specific detector that identi- fies only a single compound requires one detector for each compound of interest. This is impractical due to both cost and storage considerations. This would also be inefficient use of materials since only one detector would be in use for any given analysis and also ineffi- cient use of time due to the necessity of frequent changing of detectors. A better approach would be to develop detectors that respond to a selected class or family of compounds. A better solution still would be a detector which was independently selective for many different classes of compounds. The introduction of the "element selective" gas chromatographic detector was an attempt to simplify com- plex chromatograms. The element selective detector, as the name implies is selective for certain elements. Generally, the term "element selective" is reserved for detectors that respond to specific elements other than carbon, such as sulfur, iodine, etc. These detectors reduce complex chromatograms to chromatograms which contain peaks due to compounds with only the element of interest. There has been an increase in the development of types and applications of various element selective detectors as is apparent from several recent review ar- ticles (1-8). This section briefly discusses recent de- velopments in selective detectors for gas chromatography. A. Ionization Detectors There are several varieties of ionization type de- tectors available. The most popular of these is the flame ionization detector (FID). The FID is a universal detector that uses a hydrogen flame to burn the organic compounds and produce ionized molecular fragments. The resulting ions are collected by a pair of electrodes and produce the response signal. Since the FID has proven to be a reliable, sensitive and linear detector, it has served as a model for the class of selective ionization detectors called thermionic detectors. Another popular ionization type detector is the electron capture detector. Recent improvements in thermionic detectors and the electron cap- ture detector are discussed in this section. 1. Thermionic Detectors The thermionic detector (TID) exists in many physical designs. One basic type of thermionic detector is a modification of the popular universal FID. It is called an alkali flame ionization detector (AFID) since the design uses a flame and a coated tip, or alkali salt probe to produce an enhanced response for S, P, N, organometallics and halides. Recent studies have investi- gated new coating materials (9) and tip fabrication (10) techniques. The other basic type of thermionic detector is the flameless version (A,5,ll,l2) in which the salt container is electrically heated to yield an atomic vapor,and a plasma is used for ion production. The mechan- ism of the detector response is not well understood. Brazhnikov and Shmidel (13) attempt to explain the mechan- ism of the response of the TID. They attribute the back- ground current in a Cs coated-tip TID to the thermal ionization of Cs and Cs20 formed in an oxygen rich por- tion of the flame. When P or N containing compounds enter the flame, the combustion products form heavy ions with the Cs ions present in the background, which reduces the concentration of the alkali metal salts in the flame. Since alkali metal salts are active inhibitors of combustion, the decrease in concentration causes an increase in the flame temperature. A higher flame tem- perature causes more efficient ionization of Cs which in turn causes a change in the monitored current. Other workers have optimized the TID to improve the responses to phosphorous, nitrogen (1A,15) and organometallic compounds (16). The selectivity of the TID for P to C is about 10“, and for N to C, it is about 103. The response is linear over approximately 3 decades in concentration. This de- tector has absolute detection limits for P, its most 0'12 grams (17). sensitive element, on the order of 1 While the TID is not the most selective or the most sensi- tive detector available, it is being used regularly in the modern analytical laboratory. 2. Electron Capture Detector The electron capture detector (ECD) is probably one of the most widely used selective detectors (18). It has been improved in recent years from a temperamental device to one that is stable and extremely sensitive for halogenated compounds. This detector consists of a high energy electron source(usually 63Ni, Sc3H3, 3H, orlu7Pm)and an electron collector. The electron current decreases as halogenated compounds enter the detector and "capture" electrons. Recent improvements have extended the linearity and temperature ranges along with the sensitivity. The ef— fects of temperature (19), pressure (20), and voltage (21) on a dc electron capture detector have been in- vestigated. The use of a pulsed mode electron capture detector has been reported (22) and tested for effects of carrier gas on response (23). New electron sources have also been investigated by several workers (2A,25). Dwight and coworkers (2A) tested a 55Fe electron source. Wentworth and coworkers (25) investigated a photoioniza- tion source with the Lyman-a resonance line of hydrogen used with triethylamine to produce electrons, which eliminates the conventional radioactive source. The ECD has been coupled to a capillary column to take ad- vantage of the high sensitivity and selectivity of the ECD and the high separation efficiency of capillary columns (26). The mechanism of the ECD response has been investi- gated extensively in the last few years (5,27—29). Theo- retical studies have attempted to explain the empirical observations that the operational conditions for the coulometric mode, where the number of electrons not col- lected per unit of time is equal to the number of mole- cules entering the detector per unit of time, and constant current mode, where the ionization current is kept constant by varying the pulse frequency, are different. In fact, Bros and Page (30) calculate contradictory operating conditions for the two modes. They also explain why the constant current mode gives the best linearity and the coulmetric mode gives the best sensitivity. With a constant current ECD, a linear range of about 6 orders of magnitude is obtained (5). A reported 10 sensitivity for CClu is 50 pg (25). This type of sensi- tivity for halogenated compounds is one of the main ad- vantages for the ECD. A major disadvantage has been the large variation in electron affinity of each compound which results in a separate calibration being required for every compound. However, since the selectivity for halogenated and oxygenated compounds is good, and few other compounds have high electron affinities, the ECD continues to be used routinely, particularly for pesti- cide analysis (6). B. Electrochemical Detectors Three basic electrochemical methods, coulometry, conductometry and potentiometry, have been applied to produce electrochemical detectors. Generally, conductance detectors and coulometric detectors are discussed together since both require either an oxidation or reduction of the organic compounds eluting from the gas chromatographic column. The microcoulometer has been used for S, N and halogen analyses. The selectivity is very good (31,32); however, the sensitivity has generally been only in the nanogram range (2). Lately, research in the area of electrochemi- cal detectors has emphasized the electrolytic conductivity detector. This may be due to the greater use of the Hall (33) style electrolytic conductivity detector. This is 11 a smaller, more compact modification of the Coulson coulometric detector that has a linear range of 5 orders of magnitude, sensitivity of 0.1 ng for S, Cl and N and a selectivity for C to S of 105:1 (33). Modifications of the electrolytic conductivity detectors have included temperature control (3A), bridge circuitry (35), and sol- vent venting (36). The mechanism of operation along with numerous applications of this detector are discussed by Pape and coworkers (37). Electrochemical detectors using various types of electrodes have been described. Methods of detection of carbon monoxide by electrochemical means have been reported by several authors (38,39). In both these detectors, the CO is electrochemically oxidized at a platinum (38) or metallized-membrane electrode (39). The current produced is then proportional to the initial CO concentration. Hydrogen sulfide has also been detected electrochemically by Stetter and coworkers (A0) using methods similar to those employed for the determination of CO. Another type of electrochemical detector available is the ion selective electrode. An ion selective electrode has been used to analyze for sulfur compounds. The process consists of conversion of the sulfur compound to hydrogen sulfide which in turn precipitates silver ions from a solution. The monitoring signal is derived from a silver ion selec- tive electrode measuring the decrease in the concentration 12 of silver ions (Al). C. Optical Emission and Absorption Detectors Detectors based on an optical response have proven very useful in gas chromatography. These detectors monitor a characteristic line or band emission or absorp- tion by a particular species. This method provides a very selective analysis technique for a variety of ele- ments. These detectors can be divided into those which involve a response to an atomic species such as the micro- wave plasma emission detector (MPD), atomic absorption and discharge type detectors or those which respond to molecu- lar species such as the flame photometric detector (FPD), molecular fluorescence, chemiluminescence, and ultra- violet detectors. These detectors are discussed briefly in this section. 1. Microwave Plasma Emission Detector The use of a microwave plasma to excite emission from a gas chromatographic effluent produces a highly selective detector based on atomic emission of the ele- ments of interest. The MPD was primarily a research instrument, until the recent appearance of a MPD from Antek (A2) and Applied Chromatography Systems, Ltd. (A3). However, the MPD will probably not become as common 13 as the PPD or AFID since the detector cost is still ex- tremely high. There are many advantages to this detector. The largest purported advantage is the ability to monitor twelve elements simultaneously with this single detector (A2,A3) while also providing sensitive detection which can be used to determine empirical formulae. The mul- tiple element advantage is not unique to the MPD, but can be obtained with any atomic emission type detector outfitted with a multielement photodetector, probably at a lower cost. The Applied Chromatography Systems, Ltd. MPD is equipped with a direct reader and 12 photomulti— plier tubes to allow simultaneous determination of 12 elements. The microwave plasma gas chromatographic detector has been described by several authors (AA,A5). This detector has been optimized for ten elements by Van Dalen and coworkers (A6). The parameters which were optimized included the pressure, microwave power, observation height and concentration of scavenger gas. With their optimized system, a linear range of 3—A decades was reported for all ten elements except hydrogen,and detection limits were about 0.1 nanograms. Various improvements for this detector have been developed. Houpt (A7) found that selectivity could be greatly improved by the use of wavelength modulation. Kawaguchi and coworkers (A8) found that adding potassium 1A chloride as an ionization suppressantenhancedtflmespectral intensities and gave unity slopes for the calibration curves. Serravallo and Risby (A9) reported that doping the MPD with oxygen will increase both selectivity and sensitivity. A final improvement involves the construc- tion of a cavity that is cylindrical in shape and allows a stable low pressure helium plasma to be used. This makes use of the better atomization characteristics of the low pressure helium plasma over the atmospheric argon plasma (50). A mechanism for the response of the MPD is dis- cussed by several authors (A8,A9), although it is still not clearly understood. 2. Atomic Absorption Detectors The increase in the use of element selective detec- tors has led to a search for a single detector selective for many elements. Atomic absorption spectroscopy fits this description well. If atomic lines are observed, the selectivity should only be limited by the elements that the source can excite, thus creating a selective detector that is useful for many elements. The main disadvantage to this technique is that it is limited to a single element at any point in time due to the line source limitations. Both flame and non-flame atomic ab- sorption units have been adopted as gas chromatographic detectors (51-55). A flameless gas chromatographic 15 atomic absorption detector (NFGCAA), constructed by coupling the gas chromatograph to the AA unit by a simple tube connector, has been described by several authors (51,52). Segar (52) with an unoptimized system reported a sensitivity of 10-8 ng for tetra-alkyl lead compounds, While Parris and coworkers (51) described a system able to determine nanogram levels of astantine, selenium and tin. The flame GC-AA units (FGCAA) were constructed by Wolf (5A) by tapping a hole into the center side of a burner head and directly attaching the column to it. Chau and coworkers (55) connected the column and AA by attaching the burner nebulizer tip to the column and also inserting an all glass liner into the burner chamber to eliminate adsorption of the compounds on the burner walls. Absolute detection limits with these units have been reported as 1 ng for Cr (53,5A) and 0.1 ng for Pb. 3. Discharge Emission Detectors Another useful type of multielement selective detector is the arc, spark or glow discharge. Recent papers have described the use of both spark and glow discharges as gas chromatographic detectors. Lantz (8) reported a Spark emission detector that obtained sensitivities in the nanogram range for most elements studied. He re- ported detection of C, P, Si, B, O, S, N and H. The work 16 presented in this thesis is a new design of the detector described by Lantz. Two papers have recently been pub- lished (56,57) which discuss the use of the helium glow discharge emission detector. Feldman and Batistoni (56) describe a simple to construct, inexpensive glow dis- charge detector. It can detect halogens, S, P, C and metals in column effluents in the nanogram range. A back- ground correction system was used to enhance selectivity and sensitivity. The detector described by Rezchikov and coworkers (57) was used to detect hydrogen and water in gases down to l x 10'"5 % of the total solution. A. Molecular Emission Detectors Chemiluminescence has proven to be a useful tool in the detection of gas chromatographic effluents. The popular FPD operates by the detection of the molecular emissicwrproduced from the mixture of column effluent and the mixture of oxygen, air and hydrogen burned inside a shielded Jet. A photomultiplier tube and appropriate fil- ters provide the detection system. The FPD is most commonly used to analyze for phosphorus or sulfur con- taining compounds, but it has also been used for boron (6), nitrogen (6), chlorine (58) and tin (59) containing compounds. Various instrumental improvements to the FPD have been described to eliminate the problems associated 17 with its use. Hasinski (60) designed a two section burner head that eliminates extinguishing of the flame during passage of the solvent peak. Instrumental changes have included a detector which premixes hydrogen and column effluent (61), movement of the detector inside the gas chromatographic oven (62), a new geometric arrangement to improve selectivity (63) and a two flame system, the first flame to decompose samples and the second flame to excite emission (6A). Flow optimization was also dis- cussed (65) and it was found that flow rates for all gases drastically effect the sensitivity. These authors also recommended that the use of air with the FPD be discon- tinued. Several authors combined the FPD with capillary columns to analyze for both phosphorous containing (66) and sulfur containing (67) compounds. One of the most serious problems with the FPD, has been related to the detection of sulfur containing com- pounds. Since S2 is the emitting species in this case, the response is roughly proportional to the square of the compound concentration. The fact that the response is not always a square relationship has been the subject of several papers. It has been suggested that flow rates (65,68) cause the varying proportionality. The FPD is a sensitive, selective and easy to use detector. The sensitivity for P is 5 x 10"13 gP/sec, for S it is 5 x 10'11 gS/sec, and the selectivity with 18 respect to hydrocarbon emission is 5 x 105 gC/gP and 103-106 gC/SS (6A). The detector has functioned well with only minor modifications; although careful flow rate control will enhance sensitivity further. A furthernmdification»of the FPD has produced a class specific detector that measures the chemiluminescence generated when ozone reacts with compounds emerging from a gas chromatographic column (69). The selectivity of this detector is derived from the reactivity of ozone with varying compounds. It has been demonstrated that it is selective for olefins (69), aromatics (70), saturated hydrocarbons (70), thiopenes (70) and nitrosoamines (71). The thermal and flow properties have been improved (71) by reducing chamber volume and changing the chamber geometry. The sensitivity of this detector is from micro- grams to nanograms depending on the compound. 5. Molecular Fluorescence Detector Molecular fluorescence gas chromatographic detection has also been attempted as a means of selective detection. Several authors have described a spectrophotofluorimeter detector (SPFD) (72-7A). Winefordner and coworkers (72) have described a SPFD which uses a flow through gass cell and silicon-intensifier-target (SIT) camera tube mounted at the exit of the emission monochromator. This system, however, provided sensitivities about three to five times 19 lower than a normal photomultiplier tube with a linear range from 50—900 ng for anthracene. One advantage of this system is the immediate subtraction of the background signal to enhance selectivity. Mulik and coworkers (73) described a system which used a heated transfer line to flow sample past the monochromator and photomultiplier tube for detection. They reported a sensitivity of 50 nanograms for benzopyrene. Cooney and Winefordner (7A) investigated instrumental effects on detection limits by the SPFD. They found that improved detection could be obtained by using sources with an intense UV output and a low stray light level. 6. Ultraviolet Detector While the coupling of an ultraviolet spectrometer and a gas chromatograph has been done, it has been a rarely used detection system. The recent concern about carcinogenic nitrosoamines in many commercial products has prompted the development of methods for quantitative determination of the amount present in a given product. The most recent use of the GC-UV detector (GCUV) has been as a method for the determination of nitrosoamines (75). .Although at present this detector is not widely used due to its limited applicability and low sensitivity, it has the potential for further use as a selective gas chromatographic detector. 20 D. Miscellaneous Detectors In addition to all of the classes of gas chromato— graphic detectors that have been discussed previously, there still remain detectors that do not fall into any of these classes. These detectors involve more specialized detection methods such as IR, NMR, or optoaccoustic spec- troscopy, piezoelectric crystal sorption, radiochemical methods, and gas chromatography-mass spectrometry. These detection methods will be briefly discussed in this sec- tion. 1. IR and NMR Detectors Infrared and proton NMR spectroscopic instruments have become common in many chemical laboratories; however, their usefulness for analyzing mixtures has been some— what limited. The combination of a gas chromatograph to either of these instruments can provide selective detec- tion. The gas chromatography/infrared systems that have been used have ranged from simple modifications of com- mercial modules (76-79), to more complex systems such as Fourier Transform IR detection (80-81) and multiwave- length IR detection (82-83). NMR has also been used as a detector for gas chromatography (BA—85). GC-NMR can also provide good qualitative information about complex mixtures. 21 2. Mass Spectrometer and Plasma Detectors A mass spectrometer (MS) coupled to a GC has proven to be invaluable in the analysis of many types of complex mixtures. The GCMS has undergone many modifications, but they will not be discussed in this review since it is thoroughly reviewed by Cram and Risby (7). Recent improvements in GCMS have involved more complex data acquisition and analysis through the use of micro- computers and stored spectral data (86-87). The MS remains the most selective and overall sensitive detection system available today. Various other multimode detectors which emulate the GCMS in terms of sensitivity and selectivity have been developed. The plasma chromatograph described by Cohen and Karasek (88) is one such detector. The plasma chromatograph is an ion detector that can also provide qualitative information about column effluents (89). However, Bird and Keller (90) have investigated the vapor concentration dependence and concluded that plasma chromatography at present seems inferior to other con- ventional detectors even though the high sensitivity seems attractive. 3. Piezoelectric Sorption Detector The piezoelectric sorption detector has been used primarily as a universal detector for gas chromatography, 22 but recent applications have reported its use as a selective detector for ammonia (91) and sulfur dioxide (92). The effects of temperature (92) and selective coat- ings have also been discussed (93). The mechanism for the functioning of this detector has been discussed by King (9A). This detector has not achieved wide usage, but it seems to have particular applications in the analysis of air pollutants. A. Other Detectors There are other gas chromatographic detectors avail- able which have not achieved the popularity of the de- tectors already described. These detectors are not less useful, but simply less well known or more specific in their operation. Radiochemical detectors are used for the detection of radioactive labelled compounds (95-98). A recent article by Kreuzer (99) described an optoac- coustic detector for gas chromatography. While this is similar to GC-IR, since an IR laser is used, the detection system is quite different. Kreuzer reports that the optoaccoustic detector has a poor selectivity. E. Summary The widespread use of gas chromatography has produced a need for methods to simplify the increasingly complex 23 samples that the chromatographer is called upon to analyze. The selective detector provides a method for simplifying chromatograms of complex mixtures. However, as can be seen from the number of detectors presented in the pre- vious sections, the decision as to which detector is best for any particular application is a difficult one. The decision must include considerations such as: elements of interest, sensitivity, linear range, ease of operation and maintenance, cost and availability. There is presently no single detector which is the best in all of these categories, so the most important considerations must be weighed most heavily. One of the most important considerations in choosing a specific detector is the decision of which element or elements it is selective for, although, this is not the only factor. For example, the TID, MPD, FPD, conducto- metric detector andcoulometriCuaa3m mwzfix_ m0 10$. 36 analytical gap is less than the resistance of the bridg- ing resistor (typically 11 M0). When RA < RB, the analytical gap breaks down, causing the analytical gap to become conducting. When the gap conducts,analyte material in the gap is atomized and exited. Subsequently, electro— magnetic radiation characteristic of the element is emitted. Illustrations of the current waveforms are shown in a later section. 2. Capacitance Measurements The capacitance of a particular spark is directly related to the power of the spark. The two spark capaci— tors used in this work have capacitances of 150 pF and 300 pF. The transmission cable used has a capacitance of 29.5 pF/ft for a total line capacitance of 190 pF and 350 pF respectively. The capacitance was determined by com- parison with precision capacitors in an RC circuit pulsed by a Wavetek pulse generator. The capacitance can be increased by using a longer capacitor or a thinner di- electric as can be seen by Equation (1) for the capaci- tance of a coaxial line (110): 2weoK£L = 2n((a+tD)/a) (l) where so is the permittivity, K is the dielectric constant, 2L is the length of the aluminum bar, a is the radius of 37 the aluminum bar and tD is the thickness of the dielectric. However, when increasing the capacitance, care must be taken to insure that the proper LRC ratio is maintained since neither an overdamped nor oscillating Spark will provide as efficient a power transfer as a critically or near critically damped spark. D. Spark Current Measurements The spark is a transient phenomenon which results from the rapid discharge of a capacitor through a resistance. The current waveform during the discharge can be useful in determining overall spark characteristics. Therefore, current-related waveforms for various spark conditions were obtained by measuring the voltage drop across a small wire. The resulting waveform was displayed and stored with a fast sampling, storage oscilloscope. The data were later digitized, and a current-time curve was generated from the differential equation which describes the current response. The measurement procedure and results are discussed briefly in the following sections. 1. Current Waveform Production The spark current was measured by placing a small nichrome wire in series with the two spark gaps. The voltage drop across the resistance wire was attenuated 38 with a resistor-attenuator network that was impedance matched to the input of a sampling, storage oscilloscope (56A, 3T2, 381). A voltage versus time waveform is ob- tained from this procedure. The data are then digitized to produce a waveform like the one shown in Figure 3. The relationship between the voltage versus time data and current-time data is shown by the following differen- tial equation: E(t)=L—-+iR+% (2) A modified Runge Kutte routine (111,112) based on these data was used to solve the differential equation for cur- rent. However, this is not an easy problem since the accuracy of the results depends upon the accuracy of the resistance, inductance and capacitance values of the small wire. The wire probe was attached to both the bottom elec- trode of the second gap and ground potential by a set screw connection into the brass mount that holds the electrode in place and the brass outer housing, respec- tively. The resistance of the wire and connectors was determined by attaching a large current source and measur- ing the voltage drop. The resistance was determined to be 0.0792 0. The capacitance of the wire was estimated to be zl pF based on normal values of capacitance for 39 0.028" '"' u t Q Q t 4 __. c * A 9' a (D .— n: .J __ at g n- V t g: m (D < * t .— “'3 _| an O > n- 3 "" " .1”: .«r n- I...“ a a: ‘1’.“ 4: f c t“ .. . ,1: «- ‘t‘am’ -.1111111111L111111 0 E, 55 l l' 1 l I l l l I I l l l *1 I 0.000 7.500 TIME (NANOSECONDS) X10 7 Figure 3. An example of digitized oscilloscope voltage vs time data. A0 small wires (113). The inductance was estimated to be on the order of l to 10 nH from literature values for the inductance of Nichrome wire and brass (11A,115). An average of 5 nH was used for the current calculation. The current calculations were made through the use of the modified Runge Kutte routine (SKRUKU) listed in Appendix A. As it turns out, the single most important factor in the determination of actual current values is the inductance. Since the inductance is not known ac- curately, the absolute current values obtained may be significantly in error. However, the current values relative to one another are useful in assessing changes in spark characteristics as various parameters are changed. 2. Current Characteristics of the Spark The effect of varying Operating parameters were studied with the use of the current waveforms. Figure A illus- trates the dependence of the current on the analyte car- rier gas. It can be seen that the use of argon instead of helium produces a higher current spark of a longer duration. Emission in argon is also more intense than in helium. Figure 5 illustrates the effect of supply voltage on current. A lower supply voltage gives a higher maximum current. It was also found that operating at the lowest Al .mmw swappmo co mocmpcmdmp unoppso .z mpswfim I- 0_x a- 0_x Amazouwmoz hadazw on so mocopcommp pcmhpso .m maswfim «.. an nu Sax Amazoummozv_ .- up. -|p—- _ _ _ .mwwufio> Saddam on so mocmczmdmn pamppso .m mpswfim H- 00x AmazouumozJan5m mwgoa w0 10.1 VII m<0 IFCNIM AI UWQIQ ”1:00: UJQECW VII 13 >¢O 0.0 50 the signal from the PMT is integrated, in addition to allowing for observations to be made an adjustable time after the spark has fired. The integrate analog signal is converted to a digital signal with the use of an analog to digital convertor (ADC). This signal is then analyzed and stored by a PDP 8/e computer equipped with 16K of core memory, an extended arithmetic element and real time clock. The following section briefly discusses each major component of the SED electronic supporting system. B. Chromatpgraphic Components The gas chromatographic oven used in this work was built at Michigan State University. The oven consists of a large insulated box, a squirrel cage fan and a series of heating coils. The power to the heating coils can be controlled by a Variac to maintain the oven at the desired constant temperature. The oven contains coiled tubing along the walls to allow all gases to reach thermal equilibrium. "On-column" injection is employed. The column is coiled inside the oven and attached directly to the spark de- tector which is also mounted inside the oven. The detector is mounted along the inner side wall of the oven and secured in place by a circular mount (Figure 9). Electrical connection from the high voltage power supply to the SED is made through a hole cut in the top of the gas chromatographic oven. Emission from the 51 Figure 9. Photograph of the SED inside the chromato— graphic oven. 52 detector is viewed through a hole cut in the side of the oven. The optical trigger is also attached through a hole drilled in the side of the oven. The trigger and optical connections are discussed in more detail in later sections of this chapter. The heart of any gas chromatographic system involves the column and mobile phases. The columns that were used varied depending on the separation desired. A discussion of the specific column used is included later with specific parameters of the separation. All injec- tions were made with Hamilton series 7000 syringes to insure that all of the analyte had been removed from the needle. In addition, all needles were cleaned by insertion into a heated syringe cleaner. This cleaner is constructed from an injection septum and a brass tube surrounded by heating tape, which is electrically heated. A vacuum is drawn on this assembly during the cleaning process to aid in the removal of any contamination left in the syringe. The mobile phase used most often in this work was argon gas. While helium provides a very stable spark, the emission is very low and difficult to detect. The tank argon was passed through a series of purifying tubes separated from each other by glass wool or glass frits. The purifying tubes were filled with a 5A molecular sieve to remove small hydrocarbons and water, ascarite to 53 remove carbon dioxide, activated charcoal for removal of hydrocarbons and BASF catalyst for removal of oxygen. The flow rate of the mobile phase was measured by using a bubble flow meter at the detector exit port. The flow rate through the control gap was found to be critical in the control of the spark. For this reason, it was necessary to regulate this flow rate closely. A Gilmont F9160—ll needle valve and rotameter combination was used for this purpose. The sheath gas flow rate for the SED was controlled by a Gilmont F-l3 flow meter and Delmar valve combination. C. Spectrometer The light emitted from the atoms excited in the spark is collimated by a lens situated so that the spark gap is at its focal point. The light is then directed through a brass light pipe and into the entrance slits of a GCA McPherson EU-701/E monochromator with a EU-700-56 program- mable filter attachment. A brass tube aligns the spark gap with the other components along the optical rail. An RCA IP28 photomultiplier tube (PMT) housed in a GCA McPherson EU-701-30 PMT module serves as the radiation transducer. The PMT is kept off for most of the time by keeping the voltage between dynodes 2 and 3 at 5 volts. The PMT is then turned on by applying a voltage of 150 volts to dynode 3, which causes the full gain to be 5A available again. This sequence is triggered by a signal from the supporting electronics and takes approximately 300 ns to accomplish. This circuit is discussed further in Zynger's thesis (101). The analog signal from the PMT is then the input of the gated integrator discussed in the next section. D. Supportinnglectronics The basic design of the supporting electronic system is shown in Figure 10. The design and construction of this system is discussed extensively by my coworker, Gary Seng(118). Thus, only a brief discussion is included here. The schematic diagrams for the entire system are included in Appendix B. The supporting electronics consists of four main com- ponents: l) a delay scaler, 2) an integrate scaler, 3) an integrator and A) the computer interface. The inter- face is discussed in Section F of this chapter. This sec- tion describes the operation of the remaining circuitry. Since the large continuum background signal from the spark decays faster than the analyte emission signal, maximum signal-to-noise ratios for a particular element are obtained some time after the spark has fired. Time resolution allows a signal to be integrated for a specific duration after spark initiation. Time resolution is accomplished in this work through the use of a gated 55 .Emuwmm coflpfimfizcom 0000 omm 0:» mo .0pmwfip xooam .oH myswfim 00502022.: I >300 0.5528 0 $00800 .5240 \0 non qu Azoma 00.03024 00.200 «.330 00E 0 z. I], Illll m A 004 .0 020005. 2.00 1 0940005. 9 4 6H t 002.368 00220010on2 0.5092 #550 0000.5. .433 56 integrator and hold system. This system allows a delay time and an integrate time to be set for each individual experiment. Once the spark has fired, the circuit delays the preset time and then the signal is integrated for the proper length of time. The integrated signal is then gated to the computer for further processing. A more detailed description of the gated integrator and hold system follows. The formation of a spark is detected by the optical trigger discussed in Section E of this chapter. The negative going pulse from the trigger starts a series of Schottky counters which count down from a preset delay time. The delay time is the time that must pass after the spark has fired before data are taken. When the delay counters have reached the preset count, a similar series of Schottky counters begins to count clock pulses for the time desired for integration of the analog signal. Both of these scalers can be set by either a PDP 8/e minicomputer or by rotary switches located on the front panel of the MSU built spark electronics box. A 20 MHz crystal oscillator is divided down to a 5 and 10 MHz clock. These are used to pulse, respectively, the front panel delay and integrate circuitry and the computer delay and integrate circuitry. With this circuitry, time resolution of approximately 100 ns is obtained. The actual analog signal from the PMT is integrated 57 by a gated integrate and hold circuit that is sequenced by the aforementioned counting sequencer. The signal can then be amplified by factors of 1 to 100 before being sampled by the analog-to-digital converter (ADC) or output to a strip chart recorder. E. thical Trigger Emission data are taken from the spark discharge after the initial background emission has decayed. In order to know the actual time that data are acquired, it is necessary to know the actual time of the spark formation. This problem is a difficult one since the spark is "on" for less than 100 ns, but is repeated at about a 5 kHz rate. The most reasonable solution to this problem was to employ an optical detector of some sort which was triggered from the intense continuum radiation present very early during the discharge. There were several types of triggers available. A phototransistor, photodiode, photodiode-op amp combination and phototube were all possibilities. A phototransistor provides a good means of detecting a short on time pulse due to its high sensitivity; however, the rather slow response is a definite disadvantage for measurements to be made early in time. The photodiode, on the other hand, provides a very fast response, but a low sensi- tivity. The photodiode op-amp combination seemed to be 58 a good solution, except for the high cost and slow response of the commercially available models. A photodiode op- amp was designed that incorporated an Motorola MRD SOO combined with a video amplifier as shown in Figure 11a. This combination was not employed since the RF noise from the spark caused problems with the wide bandpass video amplifier. The phototube was also considered since it could provide both a rapid and sensitive response, but due to the bulkiness of the supporting equipment was not utilized. The two types of trigger systems that are used regularly are shown in Figures llb and llc. Figure llb illustrates a photodiode-darlington circuit which incorporates an MRD 500 photodiode with a typical darlington set-up to provide a clean negative going pulse when an intense spark is provided. This trigger is normally used with a large capacitor or large secondary gap; since with a small capacitor, or lower energy spark, the photodiode tends to produce erratic pulses. A further problem arose due to the RF noise when the trigger circuit was placed physically close to the spark; therefore, the light from the spark is transferred to the trigger circuitry by a lucite light pipe. The advantage of this trigger is that the circuitry can be triggered in less than 100 ns; however, this system was only used for solution work outside the oven since the lucite rod is not thermally stable. 59 NW hv ; .‘Et MRO 500 .0”, F 3K9 1 F (a) T0 DELAY SCALER I009 -OMF Eu“? :1 it. (b) 4.7K!) TO DELAY SCALER +5V IOKQ (C) l TO DELAY SCALER hv QI§ TH. 64 Figure 11. Optical trigger systems for the SED. 60 The trigger system used for the majority of the work presented in this dissertation is shown in Figure 110- This circuit uses a Texas Instruments TIL-6N phototransistor and a pull up resistor to provide a negative going pulse when the transistor turns on. This circuit is very sensi— tive to light emitted from the spark and functions quite reproducibly when placed in close physical proximity to the spark itself. Since the spark resides within the GC oven, and the phototransistor does not function properly above 80 °C, two lenses were used to focus the spark image onto the phototransistor which was mounted on a water cooled plate placed between the inner and outer walls of the oven. This arrangement provides a stable, nega- tive going pulse from +5 V to within +.2 V. The primary problem with this detector is the fall time of typically 1.5 us. This then necessitates missing the first 1.5 us after the spark has fired. It was found, however, that this was not a significant problem in this work since the best signal-to-noise ratios occur after the high initial continuum emission has died down. An attempt was made to find the phototransistor with the smallest fall time relative to the photodiode. This produced a tran- sistor with a fall time of less than 1 us, and this one was then used. 61 F. Interface The computer used in this work is a PDP 8/e mini- computer with 16K of core memory, an extended arithmetic element and real time clock. The PDP 8/e is equipped with a KA8—E positive I/O Bus interface card (119) which converts the PDP 8/e internal OMNIBUS signals to PDP 8/I and PDP 8/L type bus signals for compatability with PDP 8/I or PDP 8/L peripherals in addition to transferring the internal PDP 8/e OMNIBUS signals to the external bus. The external bus is daisy-chained to the Sykes dual floppy disk, Heath interface buffer box, and a driver-transmitter to pass signals to the equipment in a laboratory on a different floor from the computer. The interface circuitry built specifically for this work provides the hardware link between the instrument and the computer. Programmed I/O provides the software link. When an octal 6 is indicated in the first digit of a four digit octal code inserted into the assembly language portion of the program, an I/O instruction is indicated. Digits 2 and 3 of the four digit octal code indicate which particular device is to be activated. This is called a device selector. The last digit indicates which IOP will be generated. The combination of the correct device selector and IOP will cause the proper electronic signals to be generated by the interface circuitry to elicit the required response. The interface circuitry for this 62 instrument was designed to control the various functions of data acquisition, timing, and movement of a GCA Mc- Pherson EU-70lE monochromator. The circuit diagrams for the interface are shown in Appendix B. A brief discussion of the interface commands follow. The analog data are converted to digital data by the ADC which is controlled by an interface instruction set of 3 instructions. These instructions are listed in Table l. The 6uu1 command causes the computer to skip the next instruction in the program if the ADC has not finished the conversion of the analog signal to a digital data point and another spark has already fired. This is used to indicate how many sparks have not been used. It also gives an indication of the stability and regularity of the spark. The 6A52 command causes the data from the ADC to be gated to the computer. The 6A51 command causes the ADC status flag to be cleared. These instructions inserted into the program SKADC (Appendix A) allow for the transfer of data from the experiment to the computer for storage and analysis. The timing circuitry is set by the rotary switches on the front panel of the supporting electronics cabinet or can be set under computer control through the use of the program SKSET (Appendix A). There are two instruc- tions which control the operation of setting the counters. These instructions are listed in Table l. TDEL sets the 63 Table 1. Computer Command Set for the Spark System. ADC Commands 6AM1:SKPAD-skip on enabled ADC flag 6A51zCLRFT-clear ADC flag 6H52zDRIVE-input digital output from ADC to computer Timing Commands 6Au2zTDEL-set delay counters 6AMA:TINT-set integrate counters Monochromator Commands 6A3l:STEPS-step the monochromator 6A32zUP-enable the direction flag 6A3A:DOWN—clear the direction flag 6H6leNCOD-skip on positive encoder flag 6A6A:CLREN-clear the encoder flag 6A proper delay time and TINT sets the proper integrate times. The monochromator is controlled by an instruction set of 5 commands when only the scan function of the mono- chromator is used. The five commands are listed in Table 1. The 6U3l command causes the monochromator to move one step in the proper direction. The 6A32 command is defined as UP and causes the flip-flop controlling the direction to be enabled indicating a movement to a higher wavelength. The 6A3“ instruction causes the direction flip-flop to be cleared indicating movement to a lower wavelength. The 6A6l instruction is called ENCOD and causes the next instruction in the program to be skipped if the encoder flag has been enabled. The 6A6A command clears the encoder flag to allow it to be enabled on the next monochromator step. CHAPTER V SOFTWARE Recent developments in integrated circuit technology have made computers affordable for many everyday applica- tions. The analytical laboratory is no exception to this phenomenon. Many modern instruments are microprocessor or computer controlled to afford the analyst ease in con- trol of mechanical operations, in actual data acquisition and in data analysis. The modern trend in instrumentation is towards total automation accomplished through a complex arrangement of both hardware and software. In many cases, the software greatly enhances the hardware system. This is particularly true of systems in a research laboratory which may undergo complete design changes during the course of the study. One would usually obtain savings in both time and money if the software can be used to implement these modifications. The spark emission gas chromatographic system generates data at a rapid rate. This requires the use of a fast data analysis and acquisition system. The data acquisi— tion hardware system has been described previously (Chapter IV) and this chapter discusses the software associated with the system, although, it should be noted that when a computer is unavailable, data can be taken with a strip 65 66 chart recorder and analyzed in the usual manner. The software system was designed for maximum computer control and minimum human intervention. A. Choice of a Programming Language The choice of a computer language is a difficult one since there are many different languages available to the programmer. The most logical way to approach the choice is to decide in the beginning what types of tasks need to be accomplished by the computer and then find a language which can be operated on your specific computer system and which accomplishes these tasks with a minimum of expended energy and resources. This application required a language which could accomplish both high level tasks, such as data analysis, and low level tasks, such as stepper motor movement. This required a language which accomplished high level tasks with a minimum of assembly language programming and an easy to use assembly language for the other tasks needed. The computer available for this work is a PDP-8E minicom- puter with 16K of core memory, an RKOS cartridge disk and dual floppy disks for mass storage. This computer used with an OS/8 operating system allows programming in several languages, but it was felt that a high level language such as FORTRAN II, FORTRAN IV, or BASIC would provide the most flexibility. BASIC, although it is a very popular hobbiest 67 language, was not considered due to the unfamiliarity of the author with the operation of the BASIC compiler and the difficulty in doing I/O in BASIC. That left FOR- TRAN II or FORTRAN IV, FORTRAN II with the SABR assembly language seemed to be an obvious choice since it affords the programmer the opportunity to use standard functions such as natural log and square root without any further programming and yet, still include assembly language code when necessary. The initial data acquisition programs for the spark emission detector system were written in FORTRAN II/ SABR. The ease of insertion of assembly level code was a distinct advantage. However, with FORTRAN II the user is allowed to open only one device independent file at a time and, for applications such as multielement chromato- graphic data, more than one file would be desirable. The other commonly touted disadvantages, such as limited integer values and slow multiplication and division due to not using the extended arithmetic element (EAE), can generally be overcome by proper programming. FORTRAN IV on the other hand allows the user to open multiple (up to A) device independent files simultaneously, uses the RAE, allows integers up to + or - A096, which is twice as big as FORTRAN II allows, accepts logical variables and uses the overlay structure to expand core space. The main disadvantage of 08/8 FORTRAN IV is the in- convenience associated with inserting assembly language 68 code into a program. This inconvenience was outweighed by the other advantages, and FORTRAN IV was chosen. Although FORTRAN IV was used in this application, it is not necessarily the best language possible for all ap— plications. The programmer must evaluate his own program- ming needs and decide for that particular application which language is the best, since one language may accomplish some tasks easier and make other tasks far more difficult. B. Overview of FORTRAN IV The OS/8 FORTRAN IV system functions through the use of the FORTRAN IV compiler, RALF assembler, FORTRAN loader and FORTRAN run time system (120). A user's main program and all subprograms are compiled independently by the FORTRAN IV compiler. The compiler translates each FORTRAN statement into Relocatable Assembly Language, Floating-Point or RALF code through the use of a three pass compiler. The RALF output produced is then trans- lated by the RALF assembler into a series of relocatable binary codes for each individual program. The FORTRAN loader then links the individual RALF modules together into a loader image file containing the relocatable main program linked with the relocatable versions of each subprogram. The loader image file is then an executable program with no need for additional assembly or compila- tion. This loader image file can then be executed at 69 anytime by using the FORTRAN Run Time System or FRTS. The run-time system allows device independent I/O at run time. The FORTRAN IV system allows easy program revision and rapid execution of programs desired. C. FORTRAN IV Overlay Structure The overlay structure operates by maintaining a small main program, all assembly language programs, library sub- routines and all common areas in core at all times with all other programs to be used stored on mass storage devices. The main core resident program functions as a controller to call up the other programs as needed. When another program is called by the main program, it is then swapped into core, where it operates and stores any necessary values. After the program has completed its function, the next program called by the main program can then be swapped into the same core position that was occupied by the last program, thus saving the core space that would normally be occupied by all but one of the sub- programs called by the main program. This system works most efficiently when subprograms are written to occupy approximately the same amount of space in core, since the most core space needed at run time will be the amount needed by the longest subprogram. It also functions most efficiently when very little is kept in common since all common areas always remain core resident. The overlaying 70 structure employs maximum core usage to permit operation of very large computer programs on minicomputers. D. Software System Description The software for control of the Spark Emission Gas Chromatographic System is written into two main overlay systems. This was necessary since large quantities of data must be stored rapidly. The speed requirements eliminated the possibility of transferring every data point directly to mass storage, and this created a large block of data stored in the common area of core. Since this large block of core was occupied, there was less space left into which other programs could be swapped. The FORTRAN IV plotting package, which is patterned after the DEC plotting package, occupies more core than available when all the common areas and assembly level programs are also core resident. For these reasons, the data acquisi- tion programs and data analysis programs were separated into two main overlay systems: the data acquisition overlay and the data analysis overlay. B. Data Acquisition Overlay I. General Description The design of the data acquisition overlay is shown in Figure 12 Level 0 contains any programs which must always 71 .zmahw>o coapamfisvom mama mahmxm m4 H 4w>u4 o 4m>m4 72 remain core resident. These include the main data acquisi- tion program, (SKMNDT), any FORTRAN IV library programs that are needed for program operation (FORLIB), and as- sembly language programs to control setting delay and integrate times (SKSET), analog to digital conversion (SKADC), real time clock functions (CLOCK), movement of the monochromator (SKUP,SKDOWN) and ringing of the bell (SKBELL). Level 0 also contains all areas designated as common. In this series of routines, all emission inten— sity data taken from the Spark System are temporarily stored in common and later transferred to a mass storage device. This was done to accommodate the time restric- tions on taking multiple element data in real time. Level 0 can make calls to and receive calls from any other level, but levels other than 0 can only call programs below them and in the same vertical overlay. Since level 0 is always maintained core resident, the programs it holds were purposely held as small as possible. For this reason when it was found that the encoder on the mono- chromator could not be used to control movement under the slew option, the slew programs were removed from the overlay. The first vertical overlay shown in FigureJIBis called SKPARM which allows one to set all experimental parameters for any programs needed in a gas chromatographic run. The second overlay SKLNFD is used to find the maximum 73 emission line intensity for each element of interest and also controls the actual operation of the monochromator. The third overlay scheme is called SKDATA since it ac- tually controls the transfer of analog information to the computer. The final overlay called SKSTOR stores the data obtained for each individual element in the proper file for later analysis. A flow diagram for the data taking portion of the Chromatographic programs is shown in Figurelb; The flow diagram illustrates the basic order of functions through- out this overlay. This series of programs is normally put into a LOAD module which can then be executed by the FORTRAN run-time system (FRTS). This load module is stored on a mass storage device and can be called and executed anytime by the run time system. The load module for this set of programs can be generated by compiling each program to get a relocatable file and then typing the following series of commands at the computer terminal: .R LOAD *SKMNDT.LD,SKMNDT.MP SKMNDT, CLOCK, SKSET, SKADC, SKUP/C *SKDOWN,SKBELL/O *o mammamcm mama anhxm gahxm m4 H 4u>m4 o 4w>m4 E38 START READ DATA FROM FILE yes SMOOTH DATA SMOOTH? INTERACT? no DEFINE PEAKS FIND AREA l OUTPUT RESULTS Ifiigure 15. Flow diagram of data analysis overlay. 89 generated by compiling each individual program and then typing the following series of commands on the computer terminal. .R LOAD *3 4.) H U) c all-- m 4.) c H o > .__ or-I 4.) 0 r-I o m: fib- O l i 1 l l I l I j I I I T I I r I l I 1 I I I I 0 Time (min) 160 Figure 21. Signal stability with the SED. 110 0.1 pi injections for 2 different compounds was made. The average of these runs for 5 different times is dis- played in Figure 22. In between each of these averages, approximately 5 pl of different compounds were also in- jected. As the electrodes became coated with carbon deposits, the signals for both nitromethane and acetone decreased; however, the ratio of the two compounds to each other remained relatively constant. In a typical analysis in which an internal standard is employed, the spark should perform well regardless of the condition of the electrodes. D. Time Resolution Studies for Elements of Interest The data acquisition system for the SED allows for time resolution of the analyte emission signal. This was incorporated into the data acquisition system since in a spark discharge, the background emission disappears before analyte lines become well defined (126). Thus, if the time window of observation can be chosen to in- clude the time window of maximum emission for the analyte and minimum emission for the background, maximum signal- to-noise ratios can be obtained. If the maximum emission for the analyte occurs during intense background emission, the maximum signal—toenoise ratio occurs later in time since the background emission falls off more rapidly than analyte emission. The electronic timing circuitry used 111 180 70.0- I-III- I ‘1‘ + fl .. ' I >. 4.) oI-I U) c a) ‘- 4.) c H (D > H ""- 4.) :3 A ‘“ e “ 4 m 4_ Li‘ii O l J 1 I 1 I n ' I L I l I _g J I4 l I I I I I I I ' I I I I I . ,1 IT I I 0 Time (min) Figure 22. Signal stability with large sample quantities. I Acetone, A Nitromethane, 0 acetone/nitro- methane. 112 in this work provides a delay after spark initiation and then controls integration for a preset time. This system is described extensively by Seng (118). Since the dis- charge is a transient phenomenon, there are continual changes in spark characteristics with time. Zynger (101,125) has shown that various elements have different time windows for maximum signal-to-noise ratios. For this reason, the best time window for each element was determined. These values are listed in Table 5. Each of these values were obtained by observing the emission signal obtained for a variety of delay times and .a constant integrate time for a particular element. Graphs of relative emission intensity for P and C vs delay time are shown in Figures 23 and 2A and are representative of the procedure followed for each element. The delay times reported are relative to spark initiation indicated by the phototransistor trigger. The best integration time for each element is found by holding the delay time constant at the optimum value and varying the integration time to obtain the maximum signal-to- noise ratio. A representative plot of P signal intensity vs integration time is shown in Figure 25. Observation of the elements early in time generally produced the maximum signal-to—noise ratio. Integrating for too long a time period was not generally advantageous because the bulk of the emission signal is of a short 113 Table 5. Optimum Delay and Integration Times. Element Delay Time (us) Integrate Time (us) Boron 0.1 1.0 Bromine 0.1 0.5 Carbon 0.2 1.0 Hydrogen 0.1 O.A Iodine 0.1 O.A Nitrogen 0.1 1.0 Oxygen 0.3 1.0 Phosphorus 0.1 0.6 Silicon 0.1 1.0 Sulfur 0.2 1.0 11A o.m -P 1' q- 4- up 4- .oEHp mmaoc .w> mpfimcoucfi Hmcwam mo comuw < Aoomav oEHB mmflom b — . _ n d P I D u q q mapocamocm H. .mm opswfim r u 0.0 Aqtsuequl aAquIau mm 1000 -II- Relative Intensity Figure 2A. 115 CARBON A l l J ' 1 L L l T I U i r l ‘— I I . I I I Delay Time (usec) A graph of signal intensity vs. delay time. 115 1000 1- —.I.. 4.) .,—I U) C. (I) 4-3 -L— C H G) .3 +3 'E (U H (D 0: _.I— -I- A A1 1 J 1 I 4 J l L l n 1 l 4 I O I T 1 I I I f l I I fi I ' I '1 0.0 Delay Time (usec) 3.0 Figure 2A. A graph of signal intensity vs. delay time. 116 .oEHu cofipmpmopcfi .m> mufimcouca chwflm mo compm < .mm opsmfim o.m Aomwnv oEHB coapmmmoch 0.0 n b n L L b P P b F b b I b P D b D b - P b b n C d) d i d u 1 1 q I - -‘ I I a d1 J 1 ‘1 c -I L. d I d .2. .0 1.. II "a 8 T... 11 Du 1. Tr .. A a I. I maponomonm w .. 0 S If H. II4II 11 K O .. .\ 41 117 duration leaving only the background emission. The delay and integration times listed in Table 5 were used for all further studies presented. E. Carbon Response 1. Detection Limit and Dynamic Range The primary advantage of the SED is its potential for selective detection. However, simultaneous universal detection is also useful. The carbon response of the SED provides a universal response similar to a FID or thermal conductivity detector (TCD) response. The SED carbon response is limited to a detection limit of 1.1 x 10"8 gC 5"1 due to the background carbon signal in the argon carrier gas. The detection limit can be reduced by using 99.9999% pure argon, but this would exclude the SED from routine laboratory use due to the expense involved. The SED does not compare to other universal detectors for carbon detection since most other universal detectors use helium carrier gas which has less carbon contamination. However, the SED does compare favorably with other emis- sion type selective detectors which are also used for carbon detection. 118 2. Structural Effects The proposed mechanism of response of the SED is via complete decomposition of molecules to atoms (atomization), excitation of the atoms and emission of radiation from atoms produced by compounds entering the spark channel. For this to be true, a compound with only one carbon should give a signal one sixth as big as a six carbon compound regardless of the structure of the compounds. A variety of compounds with different types of structures were investigated for structural effects with respect to carbon. The results are shown in Table 6. In general, there seem to be few carbon structure effects. The chain length or bond type seem to have little effect on the response except in the cases of car- bon tetrachloride, chloroform, chlorobenzene, m-fluoro- toluene and carbon disulfide. The low response for chloro- benzene and fluorotoluene may be due to the added stability of the benzene ring. The enhanced response due to compounds with high carbon/chlorine ratios may be due to the electron capturing ability of the chlorine atoms changing the plasma characteristics. Effects similar to these were obtained by Winsor and Denton (127) using an inductively coupled plasma (ICP) for detection of organic compounds. 119 Table 6. Response of C rbon Containing Compounds at Carbon 2A78.6 Emission Line. Compound Response nitromethane 100* methanol 97 carbon tetrachloride 16A carbon disulfide 118 acetonitrile 99 chloroform 152 formic acid 93 bromethane 100 acetone 107 2—propanol 9O 2-propyn—l—ol 92 isobutyl alcohol 92 pentane 101 hexane 10A cyclohexanone 100 benzene 91 chlorobenzene 81 m—fluorotoluene 78 * 1 pg of carbon from nitromethane produces a response of 100. 120 F. Selective Detection Capabilities Since the SED is an emission type detector, selec- tive detection is obtained by monitoring a specific atomic emission line for the element of interest. This section reports the detection limit and dynamic range for several elements of chromatographic importance. In most cases reported, the dynamic range is limited by the amount of compound that can enter the spark before it is ex- tinguished. All data presented in the following sections were obtained by averaging areas obtained from three repli- cate chromatograms of the mixtures listed in Table 2. l. Boron Chromatographic analysis for boron in mixtures has not been a common method of analysis due to the lack of boron specific detectors and the small need for this type of analysis. However, with recent investigations of boron analogs of various amino acids, boron chromatographic analysis may become a useful technique. The investigations of Spielvogel (128) and coworkers on laboratory mice have shown that various boron analogs of some amino acids can be used as anticancer, antiarthritic and cholesterol lower- ing agents. Boron derivatives of various compounds (129-131) are also becoming more widely used. For these reasons, the boron analysis capability of the SED should 121 prove useful. Boron was detected with the SED at 2A97.7A B(I) line with a 0.1 us delay time and a 1.0 us integration time. This gave a detection limit of 3.0 x 10'9 g B/s and a linear dynamic range of 3 orders of magnitude. 2. Halogens Detection of halogens with the SED is limited to bromine at the Au77.8A Br(I) line and iodine at the 2062.AA I(I) emission line. Chlorine and fluorine emission are not observed since there are no emission lines in the 2000A-9000A range that have excitation energies less than approximately 1A eV (132). It was empirically found that this was the maximum energy that can be excited with the SED. Both bromine and iodine emission lines are observed early in time. However, the sensitivity of the SED is much greater for iodine than for bromine. The detection limit for iodine was A.2 x 10'7 g I/s and the detection limit for bromine was 1.7 x 10"5 g Br/s. The bromine calibration curve was linear over a 2 decade concentration range and iodine over a 3 decade concentration range. A calibration curve for iodomethane in nitromethane observed at the iodine line is shown in Figure 26. 122 .ocfipofi Lou o>LSo cofiumpnfifimo AmEmLmV coapmpucoocoo NIOH MIOH . :IOH mIOH mIOH IOH . 1 n w n e I. 0H "d 9 TL 9 1 T... A a lv I e u S T... q .A 1.. MOH I OH . mm whfififlm 123 3. Hydrogen The ability to analyze for hydrogen is of primary importance since most organic compounds contain hydrogen. Thus, it is important in empirical formula determina- tion. The SED has the most sensitive response to the H(I) emission line at A861.3A. Hydrogen was observed with a 0.1 us delay time and a O.A us integration time. The 6 detection limit for hydrogen was 2.6 x 10- g H/s and the dynamic range was 100. A. Nitrogen Selective analysis for nitrogen is useful in applica- tions such as environmental analysis for nitrogen contain- ing compounds like ammonia or nitrous oxides, pharmaceuti— cal analyses for various barbituates and amino acid analysis. The SED is used for N analysis by monitoring the A110.0A N(I) emission line with a 0.1 s delay time and a 1.0 ps integration time. The detection limit for N was 5.0 x 10"7 g N/s with a dynamic range of 100. 5. Oxygen The detection of oxygen with the SED requires the use of a red sensitive PMT (R666) in order to detect emission from the O(I) emission line at 7771.9A. The R666 PMT is less sensitive than the 1P28 by a factor of 12A about A0 which requires further amplification of the al- ready weak oxygen signal. Oxygen was monitored with a delay time of 0.3 us and an integration time of 1.0 us. 6 The SED detection limit for oxygen is 2.6 x 10' g O/s with linear range of 100. 6. Phosphorus Selective analysis for phosphorus is most useful for simplifying the complicated chromatograms obtained for organophosphate insecticides. Phosphorus emission from the SED is monitored at the P(I) 2535.7A emission line and is the most sensitive element that the SED is useful for. A working curve for triethylphosphite in DMSO ob- served at the phosphorus line is shown in Figure 27. A A linear range of 1 x 10 and a detection limit of 1 x 10'10 g P/s were obtained. 7. Silicon Selective analysis for silicon is useful due to the ease of formation of silyl derivatives. Silicon detec- tion can provide a less complicated chromatogram than that obtained with a universal detector. Silicon emis- sion was monitored at the 2516.1A Si(I) line. As can be seen from the calibration curve for silicon shown in Figure 28, the SED is linear over A orders of magnitude 125 .ngocawoso pom o>uso coflumpnfiawo .00 005000 Amempwv cowpmhpcoocoo mIOH mloa Flea wloa mloa OHIOH .1 w u w n 0 I..- OH "H e T.. e q Tr A a II NOH m. 1. e u s T. 1. AA .1 LI OH 125 .mspozawoso Lou o>LSo cofiumppfiamo .nm opswfim Amemhwv cowpwppcmocoo mIOH mIoH wloa wloa mloa OHIOH O. I: OH O H a I B Q. T. A a 11 NOH m. 1. e u S T: a K 0 .fi moa LI OH 126 OH IOH m. .cooflafim Lou AmEmLmv cofipwmpcmocoo wIOH - % NIOH _ m>pso coapmpnfiamo OH I]. MOH .wm mpsmfim qusuequl eAqutea 127 for silicon. The detection limie was 6.6 x 10"7 g 81/5. 8. Sulfur Sulfur analysis is useful for amino acid analysis since it can provide an easy separation between sulfur containing and non-sulfur containing amino acids. It can also be useful for simplifying complicated chroma- tograms obtained from various pesticide mixtures. The S(I) 2168.93 line was used to monitor sulfur emission. 8 The detection limit was 2.5 x 10. g S/s and the dynamic range was 1 x 102. 9. Summary The SED is chromatographically useful for at least 10 different elements. The detection limits for each of these elements are summarized in Table 7. The SED cannot compete with the FID for carbon detection or the ECD for chlorine analysis. However, it does compare favor- ably with detection limits obtained with other emission type detectors such as the MPD or inductively coupled plasma. The multielement advantage of this detector, however, may serve to offset lower detection limits. 128 Table 7. Detection Limits With the SED. Element Detection Limit (g s'l) Boron 3.0 x 10‘9 Bromine 1.7 x 10'5 Carbon 1.1 x 10-8 Hydrogen .6 x 10—6 Iodine .2 x 10"7 Nitrogen 5 o x 10’7 Oxygen 2 6 x 10"6 Phosphorus l x 10-10 Silicon 6.6 x 10‘9 Sulfur . 2.5 x 10‘ CHAPTER VII MULTIELEMENT STUDIES The SED is a versatile and sensitive detector for gas chromatography. However, it's greatest asset is the possibility for simultaneous multielement analysis. The SED can provide the capability of empirical formula de- terminations without the problems and expense associated with a gas chromatograph-mass spectrometer. This chapter discusses types of multielement detectors available in addition to examples of the SED as it is used for multi- element detection. A. Multielement Detectors Several methods have been employed or suggested to accomplish multielement atomic emission measurements. Several review articles are available which contain ex- tensive discussions of these methods (133-135). Multi- element detectors can be classified into two basic varie- ties: l) mechanical scanning and 2) array. The mechanical scanning methods involve mechanical movement of a grating or mirror past a stationary detector. Rapid scanning mono- chromators, programmable monochromators and filter wheels are of this type. This arrangement provides sensitive detection since the radiation transducer most often 129 130 employed is a PMT. The primary limitation involves the time necessary for scanning the required wavelength range. The second variety of multielement detectors includes various array type detectors. These systems are basic- ally a series of discrete detectors each corresponding to a particular wavelength range. PMT arrays provide very sensitive detection, but are limited in resolution to a single wavelength per tube. Other types of arrays such as linear diode arrays or vidicon tubes are limited in resolution to the number of elements in the array and for this reason are generally applicable over a small wave-length range. In addition, they are limited in sensitivity relative to a PMT. However, these types of detectors do have the advantage of providing true simul- taneous multielement or multiwavelength detection. Each individual application that requires multiele- ment analysis must be individually evaluated as to the type of detector needed. Sensitivity, wavelength range, resolution, the time required and cost must be significant factors in any decision. In this application a program- mable monochromator was used for multielement analysis since it provides a wide wavelength range and the greatest sensitivity; however, the slow scan rate is a significant disadvantage. 131 B. 1Examples of Multielement Capabilities of the SED It has been shown that the SED is a promising method for GO detection. This chapter demonstrates the most lucrative characteristic of the SED - the multielement capability. Several representative experimental examples are shown to demonstrate this ability. 1. Separation of TMS,,Carbon Disulfide and Benzene The separation of TMS, carbon disulfide and benzene with an SED detector is illustrated in Figure 29. The carbon response shown in Figure 29a is the same type of chromatogram obtained with a TCD. As can be seen, the carbon chromatogram contains only two peaks; however, the first peak has a shoulder indicating incomplete separa- tion. Different chromatographic conditions can be em- ployed to improve the separation or the SED can be used to separate the peaks. This can be seen in Figures 29b, 29c, and 29d with silicon, sulfur or hydrogen detection respectively. It can be seen that the first peak is a combination of TMS which elutes at 1 minute and carbon disulfide which elutes at 1.“ minutes. The final peak is due to the benzene. The sulfur and hydrogen modes are less sensitive than the carbon mode, producing a noiser, but nonetheless useful signal. ];32 3 2 III b d 8 U ’ - Z < .J U c _ 0-020 .L i % e i + i 0.999 3.590 9.799" + > : -- S IL.1 CIJPJ 3 U --r- h :— 3 3 : x-)- < J 2 i 0.140 Av 1 L 1 1 ‘1 i J 9.009 ' ' ' ' ' 1 3.599' 0.866" ). : -f SULFUR 03 2 U h— z 3 3 Z >< < .1 U u- C o.q59 1 L 1 1 L J J 0.909 ' 1 ' ' Ti 3.500' 9.2221 I ‘ iiY’DFRO(3EIV 2 4 U .- z - W _ > 2 i J U a .4 9.999 3.500 Tint (MINUTES) Figure 29. Separation of TMS, carbon disulfide and benzene. 133 2. Separation of Methanol, l-Prqpanol and l-Butanol An illustration of the separation of methanol, 1- propanol and l-butanol is illustrated in Figure 30 Figure 30a is the carbon signal, Figure 30b is the hydrogen signal and Figure 30c the oxygen signal obtained with the SED. Due to the low sensitivity of the SED for oxygen, a high gain is needed, producing a noisier signal when compared to the carbon signal. The mixture in Figure 30 consists of an equimolar concentration of each of the compounds. The areas for each of the three peaks are determined by averaging three replicate chromatograms and then computing the empirical formulas using methanol as an internal standard. The empirical formulas determined are listed in Table 8. From this example it can be seen that the SED is especially useful for the determination of empirical formulas of unknown compounds. 3. Separation of Ethyl Bromide, Iodomethane and Nitromethane The chromatographic separation of ethyl bromide, iodomethane and nitromethane is listed in Figures 3la-f, in the carbon, iodine, bromine, hydrogen, oxygen and nitro- gen modes respectively. The carbon chromatogram indicates only two peaks. However, upon examining the bromine and iodine chromatograms, it can be seen that the peaks 13“ 0.1327" > «- p. ; di- 2 .‘1’ . i u d- > Z < di- .J w a: q.- dr- 8'922 I A I % 1F IL TL % fl "_ ;“m 0.000 . 0.2221” qp HYDROGEN y — ch- 9! z u . .- z - 4. \M U > 2 .J --r- u t: 1 1 1 A 0.168 0 90911 + fl] 1' § ‘ ' T 1.099 0.132T' .. OXYGEN 3 z u 0- «L- 1 U > 4,. '..'.' < .a u. a: .. l owl -— i r 1 a : i~ 4 '1 “as 0.000 Figure 30. TIHE (HlNUTES) x10 Separation of methanol, l-propanol, and l— butanol. 135 Table 8. Empirical Formula Determination with the SED. Peak Relative Areas Empirical Number C H 0 Formula 1* 1.0 14.0 1.0 CHL‘O 2 3.1 8.0 0.95 C3H80 3 “.0 1001 1.2 Cquoo * Methanol (Peak 1) served as the internal standard. 0495“" 8.165" ” CARBON ” HYDROGEN RELA11VE INTENSITY RELAYIVE INTENSIYV F «J 0.011 1 1L 1 I fl 0. 0 000 7 000 9.1391" 9.133T I OXYGEN ~r IODIFVE I >- )- 5 ‘* 'J. 5 .3, .J_ ; dr- 2 ‘;‘ 4- ‘;' : 1 3 W I” C C “ WW H“ a 90635 J; J»— 1 J L 34 00123 1 ‘ I % % % L. 9.990 T T ' 1f 7'.000 9.09'9 ' 7.999 3.0291,. 0.110T' L a. NITROGEN BRO-M I NE > > .0- » "P O- : ‘é’ a. U W '- -4r- 0- l— 2 a- _ di— ;‘ S dy- : . I x E j .1 2’ ~r— 2 ' _ 1 1 _ 0. 11 1 1 _j1 0.990 1 ' 4 7.990 9.999 7'.909 YIHE «MINUTES\ YIN" UHNUTES) Figure 31. Separation of ethyl bromide, methyl iodide and nitromethane. 137 representing iodine and bromine containing compounds overlap. This example illustrates the usefulness of the SED for especially complicated chromatographic problems. It is highly probable that there will be many interfering peaks in a given chromatogram and the use of selective detection for many elements can uncover hidden peaks and give information as to the identity of the compound itself through the determination of the empirical formula. H. Summary The three chromatograms shown earlier in this section serve as examples of the advantages of the SED as a gas chromatographic detector. The SED can serve to simplify chromatographic techniques by separating compounds with the detector as opposed to the column or operating condi- tions, simplify chromatograms to only those compounds containing a particular element, eliminate large inter- fering peaks and provide empirical formulas to aid in compound identification. It can be seen that the SED can be a valuable gas chromatographic detector. CHAPTER VIII COMMENTARY The overall usefulness of the SED as a gas chroma- tographic detector has been demonstrated in this work. The SED can be used to monitor compounds of ten non- metallic elements with sensitivities in the microgram to nanogram range. It serves as one of the few chromato— graphic detectors suitable for selective determination of B and Si containing compounds. The SED is in most cases not affected by carbon chain length or bond type. How- ever, with the present design of the SED there are some structural effects and an improvement in the sensitivity for most elements would definitely make the SED even more attractive. The double gap design of the SED is a reasonable way to obtain controlled spark of nanosecond duration since electronically triggered spark gaps generally are of a longer duration. An increase in the power attainable by the spark would probably allow higher excitation energies and improve atomization capabilities. This would serve to improve sensitivity and reduce some structural effects. Several methods can be employed to increase the power attained by the spark. The first method would require increased size of the capacitor to allow greater charge 138 139 build-up before breakdown. However, since the dielectric is already as thin as possible, the length of the capaci— tor would have to be increased, which would destroy some of the compactness of the present SED design. Another method involves allowing a greater separation of the con- trol gap; this may cause the capacitor dielectric to break— down sooner causing an increase in the need for maintenance and repair. Another alternative is to run a different gas such as nitrogen or air in the control gap. Either of these gases would have a greater breakdown potential than argon. This alternative may also cause early capacitor breakdown in addition to requiring a larger DC power supply than is presently used. A final alternative would be to use a hefty pulsed power supply to create an over- volted spark. Incorporation of source correction into the SED system would enhance SED response. Any source correction employed would need to be fast enough to allow correction of every spark. The source correction system described by Gall (136) which employs a pelicle beam splitter with dual channel integration would be a possible system. Since the SED system electronics have several sophisticated and expensive timing circuits to allow for variable delay and integrate times, dual channel electronics would make the SED rather expensive for routine use. However, the timing circuitry can be reduced by using the same delay and lLIO integration times for all elements with only a small loss of signal and a large cost savings. The source correct- ed system would improve the precision of the SED. The use of a programmable monochromator for simul- taneous multielement chromatographic analysis is not the optimal system due to the time needed to slew between wavelengths of interest. The most practical system for routine use would employ an array of PMT's at the wave- lengths of all elements of interest. If there were only a few close lying elements of interest, a silicon intensi- fied target Vidicon tube may be of use. Maximum advantage of the multielement system can be obtained by incorporating a microcomputer or a system of microcomputers to inject samples automatically adjust chromatographic parameters, analyze data and finally, compute empirical formulas or through use of libraries of chromatograms, report the actual compounds separated. REFERENCES 11. 12. 13. 1M. 15. 16. 17. 18. REFERENCES Taylor, J. F., Proc. Anal. Div. Chem. 800., ii, 168 (1976). Adlard, E. R., Crit. Rev. Anal. Chem., 2, 13 (1975). Cochrane, W. P. and Greenhalgh, R., Chromatographia, g. 255, (1976). Pigliucci, A., Averill, w., Purcell, J. and Ettre, L., Chromatographia, g, 165 (1975). Karasek, F. w. and Field, L. R., Res./Dev., ii, March (1977). Aue, W., J. Chromatogr. Sci., J;, 329 (1975). Cram S, and Risby, T., Anal. Chem., 2g, 236R (1978). Lantz, R., Ph.D. Thesis, Michigan State University, East Lansing, MI (1977). Greenhalgh, R. J., Chromatogr. Sci., $6, 8 (1978). Johnson, B. M., Kaiman, D. and Lambrecht, R. w., Anal. Chem., ii, 1271 (1976). Kolb, B., Auer, M. and Pospisil, P., J. Chrom., 135, 65 (1977). Burgett, C., Smith, D. and Bente, B., J, Chrom., 15“, 57 (1977)' Brazhnikov, U. V., and Shmidel, E. B., J. Chrom., 122, 527 (1976). Hoodless, R. A., Sargent, M. and Treble, R., J. Chrom., 136, 199 (1977)- Lubkowitz, J., Glajch, J., Semonian, B. and Rogers, L., J, Chrom., 133, 37 (1977). Hill, H., and Aue, W., J, Chrom., 122, 515 (1976). Verga, G. R. and Poy, F., J. Chrom., 116, 17 (1976). Farwell, S. O. and Rasmussen, R. A., J, Chromatogr. Sci., ii, 22“ (1976). 1&1 19. 20. 21. 22. 23. 2M. 25. 26. 27. 28. 29. 30. 31. 32. 33. 3M. 35. 36. 37. 38. 39. 192 Poole, C. F., J, Chrom., 118, 280 (1976). Kapila, S. and Aue, W., J, Chrom., 118, 233 (1976). Kapila, S. and Aue, W., J. Chrom., 108, 13 (1975). Patterson, P., J. Chrom., 13H, 25 (1977). Burgett, C. A., HP Application Note, ANGC—l—76. Dwight, D. J., Lorch, E. A. and Lovelock, J. E., J, Chrom., 116, 257 (1976). Wentworth, W. E., Tishbee, A., Batten, C. F. and Zlatkis, A., J, Chrom., 112, 229 (1976). Rejthar, L. and Tesarik, K., J. Chrom., 131, MOM (1977). — Sullivan, J. J. and Burgett, C. A., Chromatographia, g. A (1975). Siegel, M. W. and McKeown, M., Res./Dev., $8, 101 (1977). Rosiek, J., Sliwka, I. and Lasa, J., J, Chrom., Bros, E. and Page, F. M., J, Chrom., 126, 271 (1976). (1976). N Lawrence, J. F., J. Chromatogr. Sci., U, 55 Vitenberg, A., Kuznetsova, L., Butaeva, I. and Inshakov, M., Anal. Chem., :2, 128 (1977). Hall, R., J. Chromatogr. Sci., 2, 152 (1974). Winnett, G., J, Chromatogr. Sci., Ji, 255 (1976). Lawrence, J., J. Chrom., J28, 15” (1976). MacDonald, J. and King, J., J, Chrom., l2u, 36A (1976). Pape, B., Rodgers, D. and Flynn, T., J. Chrom., 13M, 1 (1977). Stetter, J. R., Rutt, D. and Blurton, K., Anal. Chem., 1;. 92a (1976). Bergmann, I., Coleman, J. E. and Evans, D., Chromato— graphia, g, 581 (1975). U0. “1. U2. “3. an. “5. U6. “7. A8. “9. 50. 51. 52. 53- 5D. 55. 56. 57. 58. 59. 60. 1H3 Stetter, J. R., Sediak, J. M., Blurton, K. F., J. Chrom. Sci., J2, 125 (1977)- Kojima, T., Seo, Y. and Sato, J., Bunseki Kagaku, .311, 772 (1975). Antek Instruments Application Notes, Jan. 1978 Applied Chromatography Systems, Ltd., MPD850. 0, 13 (1975). Simpson, C. F., Process Biochemistry, Beenakker, C. I., Spectrochimica Acta., 32B, 173 (1977). Van Dalen, J., Lezenne Coulander, A. and Galan, L., Anal. Chimica Acta., 2:, l (1977). Houpt, P. M., Anal. Chimica Acta., 129 (1976). Kawaguchi, H., Atsuya, I. and Vallee, B., Anal. Chem., 5;, 266 (1977). — Serravallo, F. and Risby, T., Anal. Chem., ii, 21A1 (1975). Benakker, C., Spectrochimica Acta B, 31B, “83 (1976). Parris G., Blair, W. and Brinckman, F., Anal. Chem., 12, 375 (1977). — Segar, D., Analytical Letters, 1, 89 (197“). Wolf, W., Anal. Chem., ii, 1717 (1976). Wolf, W., J. Chrom., $33, 159 (1977). Chau, Y. K., Wong, P. T. S. and Saitoh, H., J. Chrom. Sci., ii, (1976). Feldman, C. and Batistoni, D., Anal. Chem., :2, 2215 (1977). — ‘— Rezchikov, V. G., Kuznetsova, T. S. and Zorin, A. D., Zhur, Anal. Khimii, 3;, 60 (1977). Versino, B. and Vissers, H., Chromatographia, Q, 5 (1975). Au, W. and Flinn, C. G., J, Chrom., 1U2, 1A5 (1977). Hasinski, S., J, Chrom., 119, 207 (1976). 61. 62. 63. 6A. 65. 66. 67. 68. 69. 70. 71. 72. 73. 7“. 75. 76. 77. 78. 79. INA Joonson, V. A. and Loog, E. P., J. Chrom., 120, 285 (1976). Pearson, C., J, Chromatogr. Sci., JJ, 15“ (1976). Sevick, Jr., Thao, N., Chromatographia, g, 559 (1975). Patterson, P., Howe, R. and Abu—Shumays, A., Anal. Chem., 2g, 339 (1978)- Eckhardt, J. G., Denton, M. B. and Moyers, J. L., J, Chrom, Sci., J3, 133 (1975). Krijgsman, W. and Van DeKamp, C., J. Chrom., 117, 201 (1976). “"—' _“ Blomberg, L., J, Chrom., JJJ, 389 (1976). Burnett, C. H., Adams D. F. and Farwell, S. O., J. Chrom. Sci., Jg, 68 (1978). Bruening, W. and Concha, F., J, Chrom., 112, 253 (1975). “‘—“‘ Bruening, W. and Concha, F., J. Chrom., 1A2, 191 (1977). Gough, T. A., Webb, K. and Eaton, R., J. Chrom., 131: 293 (1977)- Cooney, R. P., Vo-Dinh, T. and Winefordner, J. D., Anal. Chim Acta., £2, 9 (1977). Mulik, M., Cooke, M., Guyer, G., Semeniuk, G. and Sawicki, E., Analytical Letters, 9, 511 (1975). Cooney, R. P. and Winefordner, J. D., Anal. Chem., 12, 1057 (1977). Betteridge, D., Hasanuddin, S. and Rees, D., Anal. Chem., ii, 1078 (1976). Louw, C. W. and Richards, J. F., Applied Spec., 3;, 15 (1975). Shaps, R. and Varano, A., Am. Lab., J, 77 (1975). Esposito, G., Lillian, D., Podolak, G. and Tuggle, R., Anal. Chem., ii, 177” (1977). Katlafsky, B. and Dietrich, M., Appl. Spec., J2, 2“ (1975). 80. 81. 82. 83. 8“. 85. 86. 87. 88. 89. 90. 91. 92. 93- 9“. 95. 96. 97. 98. 99. 100. 1“5 Liebman, S., Ahlstrom, D. and Griffiths, P. R., Applied SpeC-. 33, 355 (1976). Coffey, P., Mattson, D. and Wright, J., 2m. Lab, J2, 126 (1978). "" Hausdorff, H., J, Chrom., 13“, 131 (1977). Hausdorff, H. Amer. Lab, J2, 77 (1978). Shoolery, J. and Majors, R., 5m. Lab., 2, 51 (1977). Shaps, R., Simons, W. and Varano, A., 2m. Lab., 2, 95 (1977). Budde, W. and Eichelberger, J., J, Chrom., 13“, 1A7 (1977). ““" “— Strauss, P. and Hertel, R., J, Chrom., 13“, 39 (1977). Cohen, M. and Karasek, F., J, Chrom. Sci., 2, 330 (1970). Karasek, Hill, H., Kim, S. and Rokushika, S., J. Chrom., 135, 329 (1977). Bird, G. and Keller, R., J. Chrom. Sci., (1976). Webber, L. and Guilbault, G., Anal. Chem., 28, 22“5 (1976). _—' u, 57“ Karasek, F., Guy, P. Hill, H. and Tiernay, J., J, Chrom., 12“, 179 (1976). Karasek, F. and Tiernay, J., J. Chrom., 22, 31 (197“). King, W., Anal. Chem., 22, 1735 (196“). Kiricsi, I., Varga, K. and Fojes, P., J, Chrom., 123. 279 (1976). Karmen, A. and Longo, N., J, Chrom., 112, 637 (1975). Tykva, R. and Seda, J, Chrom., 108, 37 (1975). Volpe, P. and Castiglioni, M., J. Chrom., ll“, 23 (1975). —"" Kreuzer, L., Anal. Chem., 22, S97A (1978). Lantz, R. K. and Crouch, S. R., Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy, Paper #286 (1976). 101. 102. 103. 10“. 105. 106. 107. 108. 109. 110. 111. 112. 113. 11“. 115. 116. 117. 118. 119. l“6 Zynger, J., Ph.D. Thesis, Michigan State University, East Lansing, MI (1973). Levy, 8., J. 22 Opt. Soc., 35, 221 (19“5). Bardocz, A., Spectrochim. Acta., J, 307 (1955). Mika, J. and Torok, T., "Analytical Emission Spec- troscopy", Crane and Russak, New York (197“). Coleman, D. M. and Walters, J. P., Spectrochim. Acta., 31B, 5“7 (1976). Hess, H., J. Phys., D., Appl. Phys., 2, 685 (1975). Walters, J. P. and Goldstein, S. A., S. T. P. 5“O, ASTM (1973). "Cadillac Plastic Buyer's Guide", Detroit, MI (1976). Hess, H., J, Phys. 2: Appl. Phys., 2, 685 (1975). Halliday, D. and Resnick, R., "Physics", John Wiley and Sons, Inc., New York (1966). DiStasio, M., Michigan State University, personal communication (1978). Ralson, A. and Wilf, H., eds., Mathematical Methods of Analysis, John Wiley and Sons, New York (1967). Rabb, M., Michigan State University, personal com- munication (1978). Lyman, T. (ed.) "Metals Handbook", The American Society for Metals, Cleveland, OH (19“8). Skilling, H. H., "Transient Electric Currents", McGraw Hill Book Co., Inc., New York (1937). Pearse, R. W. B. and Gaydon, A. G., "The Identifica- tion of Molecular Spectra", John Wiley and Sons, Inc., New York (1963). Gaydon, A. G., "Dissociation Energies and Spectra of Diatomic Molecules", Chapman and Hall, London (1968). Seng, G. S., Ph.D. Thesis, Michigan State University East Lansing, MI (1978). "Small Computer Handbook", Digital Equipment Corp., Maynard, MA (1973)- 120. 121. 122. 123. 12“. 125. 126. 127. 128. 129. 130. 131. 132. 133. 13“. 135. 136. l“7 "OS/8 Handbook", Digital Equipment Corp., Maynard, MA (197“). Johnson, E., Michigan State University, personal communication (1976). Savitzky, A. and Golay, M., Anal. Chem.,__§, 1627 (196“). —— Joseph, M., Michigan State University, personal communication (1978). Zynger, J. and Crouch, S. R., Appl. Spec., 22, 631 (1972). Zynger, J. and Crouch, S. R., Appl. Spec., 22, 2““ (1975). Shuster, A. and Hemsaleck, G., Trans. Roy. Soc., 193, 189 (1900). Windsor, D. and Denton, M. B., Appl. Spec., 22, 366 (1978). ‘——" “‘_‘ '“‘ Rawls, R., C. and E. News, 22, 21 (1978). Eisenberg, F., Carbohydrate Res., J2, 135 (1971). Brooks, C. J. W. and Maclean, I., J. Chrom. Sci., 2, 18 (1971). Belvedere, G., Pachecka, J., Cantoni, L., Mussini, E. and Salmona, M., J. Chrom., 118, 387 (1976). Zaidel, A. N., Prokofev, v. K., Raiskii, s. M., Slavnyl, V. A., and Shreider, E. Ya., Tables of Spectral Lines, IFI/Plenum Press, London (19707. Cooney, R. P., Boutilier, G. D. and Winefordner, J. D., Anal. Chem.,g22, lO“8 (1977). Santini, R., Milano, M. and Pardue, H., Anal. Chem., 35. 915A (1973). Talmi, Y., Anal. Chem.,gil, 658A (1975). Gall, R. S., Ph.D. Thesis, Michigan State University, East Lansing, MI (1978). APPENDICES OCSOCSOCSO 70 50 110 I20 130 141 142 143 144 310 840 APPENDIX A Selected Program Listings TRIS PROGRAM SCANS THE SPECTRUM WHILE TAKING DATA TO PROVIDE FULL DIGITIZED SPECTRA. PROGRAM NAME:SESPEC.F4 DIMENSION HEADR(II) DIMENSION XDATA4096 STEPS IGO’TO EXTI IF (4096 STEPS /SET DIRECTION DOWN /SET’EXTRA LOOP IN CASE ENCODER TAKES LONG /GET READY FOR 5 MSEC WAIT /STEP /CHECK FOR ENCODER FLAG /INCREMENT 4096 LOOP /INCREMENT # OF 4096’s /THIS BEGINS SECTION FOR (4096 STEPS /IT WORKS THE SAME WAY AS THE FIRST‘SEGTION \\\\\\\ 1614 DCA LPB TAD TWAIT DCA WAIT ISZ WAIT JMP .-1 TO]. CLA CLL TAD TLOOP DCA LOOP STEPS ABI. ENCOD JMP .+2 JMP .+4 ISZ LOOP JMP .-4 JMP TDII ISZ STP8D JMP EXTI JMP ENDI TDII. ISZ LPB JMP .+2 STEPS STEPS TAD TLOOP DCA LOOP JMP ABI TDI, 152 LPB JMP .+2 STEPS STEPS TAD TLOOP DCA LOOP JMP ABII ENDI. CLREN ION CDF CIF 0 JMPR *SCNDN DEFINE NECESSARY CONSTANTS WAIT. 0000 TWAIT. 3500 STPB. 0000 0000 STPBD. 0000 TLOOP. 5000 LOOP. 0000 LPBB. 7777 LPB. 0000 CAL8. 0000 0000 CALED. 0000 CALZB. 0000 0000 CALED. 0000 TA096. 0000 \‘\\‘\\‘\ CRET. #BASE. CAL. CALZR. 'GOBAK. 'ARGS. ‘TMP. 'LIT. CRT“. THIS IS THE RALPH MACHINE LEVEL PROGRAMCTO THE SCAN UP FUNCTION OF THE MONOCHROMATOR. 165 TO FUNCTION WITH THE ELECTRONICS BUILT BY G. SENG. 800? #ST .+10 +SCUP + #XR *BASE 0+3 .+6 .+3 .+3 .+3 #BASE+30 #RET .+3 .+0011 lI‘LBL *BASE *GOBAK #GOBAK50 #XR #BASE 0.1 #BASE *ARGS 'BASE.I+ 0002.0 0003.0 'LIT40000 STEP 0 STP8 CAL 0 CAL8 CALZR 0 CALZ8 *SCNUP 0004.0 #NE #RTN 0005.0 l”NE 'RTN ICHANGE F4 ARGUMENTS TO RALPH /GO TO SCAN UP SECTION CONTROL IT IS DESIGNED \\\\\‘ \\\\\\\\\\ STRT2. WAI2. A822. 166 DEFINE I/O COMMANDS ENCOD=6461 STEPS=6431 UP=6432 DOWN=6434 CLREN=6464 BEGIN EXECUTABLE COMMANDS SECTION TO SCAN UP FIELDI ’SCNUP O IOF CLA CLL ISZ GALZD TAD CALZD SZA JMP EXTZ CLREN UP CLA CLL TAD LPBB DCA LPB TAD TWAIT DCA WAIT ISZ WAIT JMP .‘I CLA CLL TAD TLOOP DCA LOOP STEPS ENCOD JMP .+2 JMP .+4 182 LOOP JMP 0-4 /INTERUPT OFF /CHECK 1F NEED TO MOVE >4096 STEPS /IF < 4096 GO TO EXT2 /CLEAR ENCODER FLAG /SET DIRECTION FLAG UP /EXTRA LOOP IN CASE ENCODER PULSE TAKES LONG /BEGIN WAITNG ABOUT 5 MSEC /STEP MONO /CHECK FOR ENCODER PULLSE IEEEP CHECKING IINCREMENT 4096 COUNTER /INCREMENT ' OF 4096 COUNTER /CLEAR.ENCODER AND DO AMOUNT < 4096 / IN THE SAME WAY AS ABOVE \\\\\ 167 JMP O-‘ JMP TD22 ISZ STPBD JMP EXT2 JMP ENDZ TD22. ISZ LPB JMP .+2 STEPS TAD TWAIT DCA WAIT ISZ WAIT JMP .-I STEPS TAD TWAIT DCA WAIT ISZ WAIT Jr“. 0-1 STEPS TAD TLOOP DCA LOOP JMP A32 TD2. ISZ LPB JMP .+2 STEPS TAD TWAIT DCA WAIT ISZ WAIT JMP .-I STEPS TAD TWAIT DCA WAIT ISZ WAIT JMP .-I STEPS TAD TLOOP DCA LOOP JMP A322 ENDQ. CLREN ION /INTERDPT OM CDF CIF O JMP“ 'SCNUP /GO BACK DEFINE NECESSARY CONSTANTS WAIT. 0000 TWAIT. 4100 STP8. 0000 0000 STPBD. 0000 TLOOP. 4000 LOOP. 0000 LPBB. 7777 LPB. 0000 CAL8. 0000 0000 CALED. 0000 CALZR. 0000 0000 CALZD. 0000 T4096. 0000 168 THIS IS THE ASSEHBLY LANGUAGE PROGRAH WHICH CONTROLS THE HONOCHROMATOR. PROGRAM NANE:SKSLWT.RA \\\\\\\ SECT SLWTO JA #ST lXR, ORG .+10 TEXT +SLWTO+ ’RET. SETX' J"MR SETB #BASE JA .+3 'BASE. ORG .+6 ARC. ORG .+3 STEP, ORG .+3 CAL. ORG .+3 CALZR. ORG .+3 ORG 'BASE+30 FNOP JA 'RET FNOP IGOBAK. 0:0 'ARGS. ORG .+3 ITMP. ORG .+0011 'LIT. 0001 2000 0000 0002 3000 0000 0003 3000 0000 0417 2716 4040 2520 4040 4040 #LBL=. ORG 'LBL 'RTR. BASE *BASE JA 'GOBAK 'ST. STARTD 0210 F STA dI‘GOBAK, 0 0200 SETX #XR SETB #BASE LDX 0.1 FSTA #BASE FSTA #ARGS FLDAZ #BASE.1+ FSTA ARG FLDAZ lBASE.l+ FSTA STEP FLDAX *BASE.1+ FSTA CAL FLDAX #BASE.I+ FSTA CALZR STARTF LDX 0002.0 LDX 0003.0 IGOOOI. #00002. \\\\\\ \\\\\\\\\\\ 169 FLDAX ARC FSUB #LIT+0011 JNE 1G0001 FLDA *LIT40000 itgAZ :TEP /BEGIN CHANGING E4 ARGUMENTS INTO RALPH FSTA STP8 FLDAZ CAL ALN 0 FSTA CAL8 FLDAK CALZR ALN 0 FSTA CALZB TRAP4 #SLDWN /GO TO SLEW DOWN LDX 0004.0 FLDAK ARG FSUB #LIT+0014 JNE J'60002 FLDA *LIT+0000 FLDAZ STEP /CRANGE F4 ARGUMENTS TO RALPH ALN 0 FSTA STP8 FLDAfi CAL ALN 0 FSTA CAL8 FLDAX CALZR ALN 0 FSTA CALZR TRAP4 #SLUP IGO TO SLEW UP LDX 0005.0 EXTERN :NE JA #RTN LDX 0006.0 EXTERN d"NE JA #RTN DEFINE OPDEFS ENCOD'6461 SLUP=6421 SLDN=6422 CLREN=6464 CLRSW=6424 8 ROBE CODE SECTION TO SLEW DOWN FIELD! ISLDWN CLA CLL ISZ CALZD ICREGK IF OVER.4096 TAD CALZD 82A JMP EXTRZ /JHP IF LESS THAN 4096 CLREN SLDN ENCOD §\\\\\\\ STPED, CAL8. CALED. CALZB. CALZD. TLOOP. \\\\\\\ EXTRI. JMP o-l ISZ TLOOP JMP 0-5 ISZ CALEB JMP .‘7 CLREN SLDN ENCOD JMP 0-1 ISZ STPED JMP .-5 CLRSW CDF CIF 0 JHPZ #SLDWN 170 /CRECK FOR ENCODER.PULSE ICOUNT TO 4096 /INCREHENT NUMBER.OF 4096 LOOPS /DO THE NUMBER OF STEPS LESS THAN 4096 /CHECK FOR ENCODER.PULSE DEFINE NECESSARY CONSTANTS 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 SECTION TO SLEW UP 'SLUP.0 CLA CLL ISZ CALZD TAD CALZD SZA JMP EXTRI CLREN SLUP ENCOD J”? 0"! I82 TLOOP JMP .“5 I52 CALEB JMP 0-7 CLREN SLUP ENCOD JMP .‘I 182 STPHW JMP .‘5 CLRSW CDF CIF 0 JMP“ 'SLUP /TRIS SECTION OPERATES BASICALLY THE SANE /AS THE SLEW DOWN SECTION. ONLYHTRE DIRECTION /FLAG ON THE INTERFACE IS DIFFERENT 00000000000 15 50 10 40 171 TRIS PROGRAM CONTROLS TIE ACTUAL'DATA ACQUISION FROM THE ADC. IT'CAN BE USED FOR.UP TO FOUR ELEMENTS. IT WILL SCAN 'N) THE LINE OF MAXIMUM EMISSION FOUND BY SKLNFD AND THEN PROCEED TO ACQUIRE DATA. PROGRAM NAME: SKDATA. F4 SUBROUTINE SKDATA( NEL.DELY, AINT. 'ITIINUT. NGO. ITHRU) DIMENSION DELY( 4) .AINT(4) .N00( 4) IDIR=1 T181. ITHRU-=1 'I'SEC=TMINU'I*60. IF(NEL .EQ.1) GO TO 50 CALL SKSEDAI TI . DELAY. AIN’I‘G. ITHRU) DO 20 I=1.NEL-1 ITHRU= ITHRU+1 NGON=NGO( I) DELAY= DELY( I) AINTG=AINT( I) CALL SCNR( NGON. IDIR) CALL SISEDIH TI . DELAY. AINTG, ITHRU) TI=TI+1.5 CONTINUE IDIR=0 NGON=NGO(NEL) CALL SCNR( NGON. IDIR.) IF(NEL .EQ. 2)TI=TI+3. IF(NEL .E0. 3) TI=TI+6. IF(NEL .E0. 4) TI=TI+9. IF(TI .EQ. TSEC) GO TO 40 GO TO 15 DELAY= DELY( l) AINTG=AINT( 1) CALL SKSEDA(TI . DELAY. AINTG, ITHRU) IF(TI .GE. TSEC) GO 'IT.) 40 ITHRU= I'I'HRU+I TI=T1+1. GO TO 10 CONTINUE RETURN END 00000000000000 30 20 108 201 98 172 THIS PROGRAM STORE ANY CHMMATWRAPHIC DATA TAKEN BY SKMNDT.F4. IT WILL STORE DATA FOR UP TO FOUR DIFFERENT ELEMENTS IN SEPARATE FILES UNDER NAMES THAT ARE SPECIFIED AT RUN TIME. THE AMOUNT OF TIME BETWEEN EACH DATA POINT DEPENDS UPON THE NUMBER OF ELEMENTS. IT IS CALCULATED TO ALLOW MOVEMENT BY THE GCA MCPERSON MONOCHROMATOR.TO CLOSE LYING ELEMENTS. PROGRAM NAME:SKSTOR.F4 SUBROUTINE SKSTOR(NEL,ITHRU.HEADR.EL.DELY,AINT) DIMENSION ARRAY(I500).HEADR(11) COMMON ARRAY TMIN1=1. TMIN2=2.5 TMIN3=4.0 TMIN4=5.5 IF(NEL .E0. 1) GO T010 IF(NEL .E0. 2) GO TO 20 IF(NEL .EQ.3) GO TO 30 WRITE(5.200)HEADR WRITE(6,200)HEADR WRITE(7.200)HEADR WRITE(8.200)HEADR WRITE(5.201)ITHRU.NEL.TMIN1 WRITE(6.201)ITHRU.NEL.TMIN2 WRITE(7.201)ITHRU.NEL.TMIN3 WRITE18.201)ITHRU.NEL.TMIN4 DO 6 I81.ITHRU.4 WRITE(5.202)ARRAY(I) WRITE(6.202)ARRAY(I+1) WRITE(7.202)ARRAY(I+2) WRITE(8.202)ARRAV(I+3) CONTINUE GO TO 98 WRITE(5.200)HEADR WRITE(6,200)HEADR WRITE17.200)HEADR WRITE(5.201)ITHRU.NEL.TMIN1 WRITE(6.201)ITHRU.NEL. TMIN2 WRITE(7.201)ITHRU.NEL.TMIN3 DO 7 I=I.ITHRU.3 WRITE(5.202)ARRAY(I) WRITE(6.202)ARRAY(I+I) WRITE(7.202)ARRAY(I+2) CONTINUE GO TO 98 WRITE(5.200)HEADR WRITE(6.200)HEADR WRITE(5.201)ITHRU.NEL.TMIN1 WRITE(6.201)ITHRU.NEL.TMIN2 DO 8 I=1.ITHRU.2 WRITE(5.202)ARRAY(I) WRITE(6.202)ARRAY(I+1) CONTINUE GO TO 98 WRITE(5.200)HEADR WRITE(5.201)ITHRU.NEL.TMIN1 WRITEI3.108)EL.DELY.AINT FORMAT(A6.2GIO.1) WRITE(5,202)(ARRAY(I).I=1.ITHRU) FORMAT£11A6J FORMAT1216.F8.1) FORMAT(4G15.8) CONTINUE RETURN END 0000000000000 586 90 95 100 173 THIS PROGRAM.CONTROLS THE DATA ANALXSIS PORTION OF THE S. KOEPLIN CHROMATOGRPAHY PROGRAMS. IT ALLOWS FOR READING OF A DATA FILE. SMOOTRING OF DATA . PLOTTING OF THE DATA ON A TEKTRONIX GRAPHICS TERMINAL.INTERACTIVELY DECIDING ON THE PEAK LIMITS. AREA COMPUTATION AND FINAL REPORT GENERATION. PROGRAM NAME:SKMNAN.F4 COMMON ARRAY.INDX DIMENSION ARRAYC1500).INDX(21).READR(11).TMSEC(1) DIMENSION RETIME(7).HITE(100).TME(100).RESULT(?) READ(5.75)HEADR FORMAT(11A6) FORMAT(216.F8.1) READ(5,80)NDPT.NEL.TMIN ITHRU=NDPT/NEL READ(5,100)EL.DELY.AINT FORMAT£I6) DO 586 I=1,ITHRU READ(5.85)ARRAY(I) CONTINUE FORMAT(4GI5.8) FORMAT(‘ TYPE WHICH YOU WANT PLOTTED-SM OR RAW? ’.0) WRITE(0.90) FORMAT(A6) READ(4.95)TYPE IF(TYPE .EQ. ’SM’) GO TO 10 IF(TYPE .EQ. 'RAW') GO TO 20 CALL SKSMTH(ITHRU) CALL SKTPLT(NPEAK.ITHRU,NEL.TMIN) IF(TYPE .EQ. 'RAW’) GO TO 50 REWIND 5 READ(5.75)HEADR READ(5.80)ITHRU.NEL.TMIN READ(5.100)EL.DELY.AINT READ(5.83)(ARRAY(I),I=1.ITHRU) CALL SKAREA1NPEAK.BESULT.RETIME.NEL.TMIN,XCOUN) CALL SKLIST(NPEAK.RESULT.RETIME.HEADR.EL.DELY.AINT.XCOUN. TMIN.ITHRU) FORMAT(A6.2010.1) END 0650(50630W50f50 500 499 501 502 ~C§0C§0W50CDO 11 12 15 13 0(50 “H THIS PROGRAM PERFORIB A SAVI’IZKY AND COLAY SIDOTH ON THE DATA SUPPLIED TO IT. IT IS A RUNNIGN SMOOTH THAT ONLY REQUIRES ONE EXTRA ARRAY OP FIFTEEN ELEMENTS IN ADDITION TO THE DATA ARRAY. PROGRAM NAME: SKSMTH. F4 SUBROUTINE SKSMTH(ITHRU) COMMON ARRAY DIMENSION ARRAY(650).X(I5) WRITE(0.500) FORMAT(’ 5.9. OR 15 POINT SMOOTH? '.0) CONTINUE READ(4.501)ITYP FORMATtI3) IF(ITYP .E0. 5) GO TO 10 IF(ITYP .E0. 9) GO TO 20 IF(ITYP .E0. 15) GO TO 30 WRITE(0.502) FORMAT(‘ ONLY 5.9.15 POINT SMOOTHS ARE ALLOWED.TYPE NUMBER AND RETURN’) GO TO 499 ********************************8*********8**************8*8*********t**8* 5 POINT SMOOTH ISMPT= ITHRU-4 ICOUN=1 X( 1)=ARRAY( 1) X( 2) =ARRAY(2) JKL= I X( 3) = 17.*( ARRAYI ICOUN-+2) ) +12. *(ARRAY( ICOUN+ 1)+ARRAY( ICOUN+3) ) +-3. *( ARRAY( ICOUN)+ARR.AY( ICOUN-1'4) ) IF( ISMPT .EO. JKL)GO TO 12 JKL8JKL+1 X(4)=17. *( ARRAY( ICOUN-+3) ) + 12. *( ARRAY( ICOUN+2) +ARRAY( ICOUN+4)) +-3 . *( ARRAY( ICOUN+1) +ARRAYI ICOUN+5)) JIG..=JIG..+1 ARRAY( ICOUN)=X(1) ARRAY( ICOUN+ I ) =X( 2) X(1)=X(3)/35. X( 2) =1“ 4) I35 . ICOUN= ICOUN+2 IF( ICOUN .LT. ISMPT) GO TO 11 K= 1 DO 13 I=ICOUN. ISMPT-O-Z ARRAY( I)=X( K) K=K+1 CONTINUE RETURN C*****#****ti*****88¥t$£¥8¥83*38*I*8*$****$*8#8**$88838$883838338ttiittt8tfl 175 9 POINT SMOOTH I°0C50650 0 ISMPTleHRD-B ICOUN=1 X(1)=ARRAY(1) X(2)=ARRAY(2) X(3)=ARRAY(3) X(4)=ARRAY(4) JKL=1 21 10:0 Il=1 12:2 l3=3 14:4 15:5 16:6 17:7 18:8 D0 25 185.8 X(I)=59.*ABBAY(ICOUN+14)+54.*(ARRAY(ICOUN+13)+ARRAY(ICOUN+15)) 1 +39.*(ARRAY(ICOUN+12)+ABRAY(ICOUN+16))+14.*(ARRAY(IGOUN+II)+ 1 ARRAYtICOUN+I7))+-21.*(ARRAY(ICOUN+IO)+ARRAY(ICOUN+18)) 10=10+1 Il=ll+1 12=12+1 13=Is+1 14=I4+1 I5=15+1 16=l6+1 17=17+1 18=IB+1 IF(ISHPT .EQ. JKL) GO To 26 JKL=JKL+I 25 CONTINUE 26 AREA}? ICOUN)=X( I) ARRAY(ICOUN+1)=X(2) ARRAY( ICOUN+2)=X(3) ARRAY(ICOUN+3)=X(4) X( l)=X(5)/231. X(2)=X(6)/231. X(3)=X(7)/231. X(4)=X(8)/231. ICOUN=ICOUN+4 IF(ICOUN .LT. ISHPT) so To 21 £31 DO 27 I=ICOUN.ISMPT44 ARRAY“ I)=X( K) K=K+l 27 CONTINUE RETURN C C*********************************838*******3*****3************$***$ 15 POINT SMOOTH @CH50CH5 0 ISMPT= ITHRU-14 ICOUN=1 X(1)=ARRAY( I) X(2)‘-'-’ARRAY(2) X(3)=ARRAY(3) X( 4)=AR.RAY(4) X(5)=A.RR.AY(5) X(6)=ARRAY(6) X(7)8ARRAY(7) 31 35 36 HM- 176 JKL-l JO=O JI=I J2=2 J3=3 J4=4 J5=5 J6=6 J7=7 J8=8 J9=9 JIO=IO J11=Il J12=12 J13=13 JI4=I4 DO 35 I=8.14 X(I)=167.*ARRAYIICOUN+J7)+162.*(ARRAY(ICOUN+J8)+ARRAY(ICOUN+J6))+ 122.x(ARRAY(ICOUN+JIO)+ARRAY(ICOUN+J4))+B?.*(ARRAY(ICOUN+JII) +ARRAY=X(5) ARRAY(ICOUN+5)=X(6) ARRAY(ICOUN+6)=X(7) X(1)=X(8)/1105. X(2)=X(9)/1105. X(3)=X(IO)/1105. X(4)=X(II)/1105. X(5)=X(12)/1105. X(6)=X(13)/1105. X(7)=X(14)/1105. ICOUN=ICOUN+7 IF (ICOUN .LT. ISNPT) GO TO 31 K=I DO 40 I-ICOUN,ISMP'I‘+7 ARRAW I)=X(I() K?K&I CONTINUE RETURN END OOOOOOOOOOOOOOO 101 301 510 511 512 59 60 61 63 62 177 THIS PROGRAH PLOTS THE DATA PASSED '1‘) IT ON A mu: GRAPHICS TERMINAL. THE FIRST PLOT CONSIS'IS OF A PLUI' OF EVERY FIFTH POINT AND AFTER THAT THE 08“ IS ALLOWED 1'0 INTERACTIVELY DEFINE THE POINTS TO BE PLOTED. INDICE CAN BE SET '11) DEFINE PEAKS. THIS SET OF PROGRAIB USES THE FORTRAN IV PLO'I'I‘ING PACKAGE DEVELOPED BY H. JOSEPH. PROGRAM NAHE: SKTPLT. F4 SUBROUTINE SKTPLT( NPEAK. NDPT. NEL. THIN) COMMON ARRAY. INDX DIMENSION ARRAY(1300).INDX(21) IDIR=0 ITHRU=NDPT/NEL IF(NEL .EO. 1)XCOUN=1. IF(NEL .EQ. 2)XCOUN=3. IF(NEL .EQ. 3)XCOUN=6. IF(NEL .EQ. 4-)XCOUN39. MAX: FLOAT( ITHRU+XCOUN) XMIN=TMIN YI‘IIN=AR.RAY( 1) YMAX=ARRAY( 1) DO 101 J=1.ITHRU IF(ARRAY(J) .GT. YHAX) YI'IAX‘ARRAYLI) IF(ARRAY(J) .LT. YMIN) YHIN=ARRAY(J) CONTINUE CALL 'I'PLINT(XI‘IIN.XI‘IAX.YI‘IIN.YHAX, 1,1) CALL TAXIS( 1.0.0) XVAL=XMIN DO 202 K=1.ITHRU.5 XVAL= XVAL+( SXXCOU'N) CALL 'I‘PLT(-1.XVAL.ARRAY(K)) CONTINUE CALL TPOSIT(34.60) WRI'I'E(0.301) ITHRU FORMAT( 16. ’POINTS BY 3‘ ) CALL TPOSIT(0.0) WRITE(0.520)YI‘IAX CALL TPOSIT( 32.0) WRITE(0.520)YMIN WRITE(0.510) FORMAT( ’ INITIAL POINT 4' N PLUI".S) READ(4.511) IPT FORMAT( I6) WRITE(O.512) FORMAT( ‘ FINAL POINT # TO PWT'HO) READ(4.511) IFPT IF( IFPT-IPT)59.59.61 WRITE(O.60) FORMAT( ‘ TRY AGAIN') CO '11) 1 NU= IFPT-IPT IF(NU .LT. 975) GO TO 62 WRITE(0.63) FORMATC ’ CANT DO MORE THAN 975 POINTS’) GO TO I IDIR=0 XMAX= ( FLOAT( IFPT- 1) *XCOUN)+TIIIN MIN-‘- ( FLOAT( IPT- 1 ) *XCOUN) +TMIN YHIN=ARRAY( IPT) YHAX'ARRAY( IPT) DO 10 J3 IPT.IFPT 178 IF(ARRAY(J) .GT. YEAH) YMAX‘ARRAYQI) IF(ARRAY(J) .LT. YHIN) YMIN=ARRAY(J) 10 CONTINUE CALL TPLINT(XHIN.XHAX.YHIN,YHAX.1.1) CALL TAXIS(I.0.0) XVAL=(FLOAT(IPT‘1)*XCOUN)+THIN DO 20 J=IPT.IFPT CALL TPLT(-I.XVAL,ARRAY(J)) XVAL=XVAL+XCOUN 20 CONTINUE CALL TPOSIT(0.0) WRITE(O.520)YMAX 520 FORMAT(G]5.8) CALL TPOSIT(32.0) WRITE(O,520)YHIN CALL TPOSIT(34.2) WRITE(O.511) IPT CALL TPOSIT(34.70) WRITE(O.511) IFPT J=IPT 25 READ(4.501) KEYED 501 FORMAT(A2) IF(KEYBD .EQ. ‘ ’ .AND. IDIR..EOm 0) J=J+1 IF(KEYBD .EQ. ’ ' .AND. IDIR .E0. 1) J3J-1 IF(ICEYBD .EQ. ’ ') GO TO 30 IF(KEYBD .EQ. ‘R') IDIR?! IF(KEYBD .EQ. ’F') IDIR=0 IF(KEYBD .E0. ‘8') GO TO 100 IF(KEYBD .EQ. 'A‘) GO TO 21 IF(KEYBD .EQ. 'N') GO TO I IF(KEYBD .EQ. 'P’)GO TO 99 IF(KEYBD .E0. ’1’) INDX(1)=J IF(KEYBD .E0. ‘2') INDX(2)=J IF(KEYBD .E0. '3’) INDX(3)=J IF(KEYBD .E0. ‘4') INDX(4)=J IF(KEYBD .E0. ’5') INDX(5)=J IF(KEYBD .EQ. '6') INDX(6)=J IF(KEYBD .EQ. '7‘) INDX(7)=J IF(KEYBD .E0. '8’) INDX(8)=J IF(KEYBD .E0. ‘9') INDX(9)=J IF(KEYBD .E0. ’10‘) INDX(10)=J IF(KEYBD .E0. '11') INDX(11)=J IF(KEYBD .E0. '12') INDX(12)=J IF(KEYBD .E0. '13’) INDX(13)=J IF(KEYBD .E0. '14’) INDX(14)=J IF(KEYBD .EQ. '15') INDX(15)IJ IF(KEYBD .E0. '16’) INDX(16)=J IF(KEYBD .E0. ’17‘) INDX(17)=J IF(KEYBD .E0. ’18') INDX(18)=J IF(KEYBD .EQ. ’19’) INDX(19)=J IF(KEYBD .E0. ’20‘) INDX(20)=J IF(KEYBD .E0. '21') INDX(21)=J GO TO 25 30 IF(J .GT. IFPT) J=IFPT IF(J .LT. IPT) J=IPT XVAL=FLOAT(J) CALL TPLT(-1.XVAL.ARRAY(J)) GO TO 25 21 READ(4.511) XADD JSIPT*KADD GO TO 30 99 WRITE(O.515) 515 FORMAT( ‘ TOTAL " OF FEARS TO INTEGRATE(I)’) READ(4.518)NPEAK 518 FORI‘IAT( I3) 100 CONTINUE RETURN END 00000000000 32 20 98 108 118 113 179 THIS PROGRAM FINDS THE AREA OF A PEAK. AND‘CORRECTS FOR BASELINE DRIFT BY SUBRACTING OUT'A.TRAPEZOID. THE AREAS ARE FOUND ON THE PEAKS DEFINED BY THE INDICES PASSED FROM SKTPLT. PROGRAM NAME:SKAREA.F4 SUBROUTINE SKAREA(NPEAK;RESULT.RETIME.NEL,TM1N.XCOUN) COMMON ARRAY.INDX DIMENSION ARRAY(1500).INDX(21) DIMENSION RESULT(7).RETIME(Y).EITE(100).TME(100) J=NPEAK*3 NUMB=1 IF(NEL .EQ. I)XCOUN=I. IF(NEL .EQ. 2)XBOUN=3. IF(NEL .EQ. 3)XCOUN=6. IF(NEL .EQ. 4)XCOUN=9. D0 25 I=1.J,3 KSKP=I KKzI K=INDX(I) KlsINDX(I+1) K2=INDX _< m» r< N—umrh wu<4 o9: CG 3.: psz on: pzu 59.5... x :mm “.331. “2“ 1 933.1. mzu 1 USS: mzu 725.1. a 3 a; 3.3 3 «5 V39 .3 ad 3.: a 3.3 d a a _ w _ _<\ m _ M L1. 5“ w N: 3... a... 2186 .pfisopfio mumpumpcfi moamcm cab .wm mpsmfih mxupzaou n. up<¢mm—z. u a ~< m cm ~- _~.rh uu» can u: .< L¢Hu O\< .>2— :C. 0 Iv x— .no »» o on.rs 4mz<¢ .pCu wh .3. l omen:— no.3 was... 4%. a .< u: .. .~.rn ¢_, no» 2— ts .2. _ _ c m up<¢ou»z. aoarn coach a. .l\/\/\(no . % oomrn . »<4un 0 new; I: 2 187 .zppH309Ho vaonlvcmlmHQEmm\mNHHmfiuficfi\x00Ho was .zm mhswfim H lHHWllll,oco <~m0nx <4¢hx uu>lllllln9 MUOJ ~=z o~ a .4 no mine. 1 wNH4<~HHzH uo< 0»! ~30 lHHWlloco ra—DIL naoflI :— muxtm Jup<4ua somcn oJOz . .: u4¢xzu mar<4uo a» «9'5 x; 849 mhrh 1 each r u C «a; x— xnuu< no nuuu Mpnm 2~ ¢JZ< _ A _ n n“- a“. xauu< mo n-& wh—Q u" .no m¢u>uxa owh