MSU RETURNING MATERIALS: PIace in book drop to remove this checkout from LIBRARIES “ your record. FINES wiH be charged if book is returned after the date stamped below. Rat-“5.8441959 {I 30-3 r2 i; .0 ,J I ‘- t O .1. pt -J V.. . I ~ ' u! a a u “I... at n’ . \I, c v. cFI « DEVELOPMENT OF A COMPUTER-CONTROLLED HULTIDIMENSIONAL LIQUID CEROMATOGRAPE AND ISOLATED DHOPLET INTERFACE 3:! Patrick Mark Hiegand A DISSERTATION Sub-itted to Michigan State University in partial fulfillnent of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Chemistry 1985 ABSTRACT DEVELOPMENT OF A COMPUTER-CONTROLLED MULTIDIMENSIONAL LIQUID CHROMATOGRAPH AND ISOLATED DROPLET INTERFACE By Patrick Mark Wiegand This dissertation presents two instruments useful for the analysis of complex mixtures. The first is a computer—controlled liquid chromatograph incorporating two six-port, two-position valves for column switching. Column switching techniques have been used for increasing resolution , characterizing samples and simplifying chromatograms. The popularity of this technique has been limited, however, by the difficulty of reproducibly transferring eluent based solely on the time from injection. The multidimensional chromatograph presented here differs from previous designs in that valve control can be based on detector output as well as elapsed time. This method of eluent transfer is shown to be much more reproducible than that based solely on time. In addition, the instrument allows centralized control of all instrument functions using a unique operating system based on the FORTH computer language. This greatly simplifies the development of multidimensional chromatographic methods. Applications are Shown for backflushing, heartcutting and mode’switching. Chromatograms of both model compounds and petroleum samples are shown. The second instrument presented can assist in the interfacing of liquid streams to gas-based detection schemes such as atomic emission, atomic absorption or mass spectrometry. This instrument, called an isolated droplet generator. is based on the vibrating capillary principle of droplet production. It is capable of converting a liquid stream, such as that produced by an HPLC, to a sub-nanoliter sized monodispersed droplet stream generated at rates of up to 50 kHz. Selective charging and deflection can be used to select individual droplets or droplet packets from the main stream. This instrument is an improvement over older designs in that computer control imparts sufficient flexibility to make the device useful as a general-purpose high-resolution liquid handling system. In addition, droplet production parameters can be automatically altered to compensate for changing liquid streams, as in gradient liquid chromatography. Results are shown which indicate that droplet production is relatively unaffected by changes in viscosity and surface tension that typically accompany an BPLC gradient such as methanol to water. The effect of droplet introduction into a spark source is also discussed. To Marguerite and Christopher, my dearest friends ii LII 03! II! It: :3! Q m I H r ACKNOWLEDGEMENTS I have often thought that the sociology of a research group would make an interesting subject for a dissertation. I wouldn’t want to write it, of course, but it would be interesting nonetheless. During the five years or more that one spends doing research with fellow graduate students, many long-lasting relationships are developed. These relationships take many forms, but certainly teacher, student, mentor, friend and family are all adjectives which can be used to describe them. Graduate school has included some of the worst and some of the best times of my life; the people responsible certainly deserve some credit. Many thanks go to the following group members: Nelson Eerron, for showing me the proper way to prepare for cross-country skiing and other relaxing hobbies; Eim Ratanathanawongs, for not showing me the proper way to ride a bicycle (especially near my car); Keith Trischan, Frank Curran, and John Stanley, for helping me watch for sidewalk obstacles in Philadelphia; Max flineman and Tom Doherty, for being the dynamic duo that they are; iii Pete Wentzell, for the many products of Reptilian Lifeforms, Inc.; Helen Archontaki and Jimmy Wu, for their very entertaining conversations; and the rest of the group, Paul Kraus, Cheryl Stultz, Bob Harfmann, Ralph Thiim, Mark Phillips and others for helping celebrate good times. A very special thanks goes to Dr. Stan Crouch, for his advice, friendship and guidance throughout the past five years. I am also indebted to Dr. Chris Enke, for serving as second reader, and to the members of his research group, for always having Just the right part at the time when it was most needed. The staff of the MSU Chemistry Department also deserves credit, especially Marty Rabb, electronics designer extraordinaire, for his help and friendship, and all the people in the machine and electronics shops, for turning ideas into reality. My family deserves many thanks; for putting up with my infrequent visits, for providing financial support, and most of all, for always believing that I could do it. Last, but definitely not least, a very special thanks goes to Marguerite Danna Wiegand; astute collegue, loving spouse, and best friend; for making the bad times seem insignificant and the good times worth celebrating. iv PREFACE This dissertation is primarily concerned with the design and implementation of two analytical instruments used in the support of work done under NASA grant number NAG 3-93. Among the goals of this grant were the development of methods for the determination and speciation of various metals in petroleum matrices. Several possible approaches were investigated using chromatographic separation followed by selective detection. The first instrument is a computer-controlled multidimensional liquid chromatograph. The purpose of constructing this instrument was to investigate automated column-switching techniques for increasing the resolving power of liquid chromatography. In addition, such an instrument could provide more information for sample characterization. The second instrument is an isolated droplet generator. This device was constructed to investigate the isolated droplet technique as a possible interface between various gas-phase detectors and the liquid chromatograph. These two instrumental projects could have been separated and reported independently. Because the projects fl have a common origin, however, it was felt that a better :approach would be to include both in the same body of text. To aid future researchers who may be interested in only one of the projects, each chapter of this dissertation has two main sections; the first pertaining to the multidimensional chromatography and the second to the isolated droplet generator. Every attempt has been made to distinguish these two sections clearly in each chapter. In addition, researchers interested in further applications of the multidimensional chromatography instrumentation should consult the 1985 Michigan State University doctoral dissertation of Marguerite R. Danna. vi Table of Contents List of Tables....... ..... ...................... . List. of Figures......... .......... . ..................... I. Introduction........ ....................... .. ........ II. Historical Background.. ............................. .A. Multidimensional RPLC..... ........ ....... ....... B. Element Specific Detection. ..................... C. Isolated Droplet Generators.......... ........... III. Instrumentation .................................... A. Multidimensional RPLC ........................... 1. System Overview ............................ 2. The Microcomputer System......... ...... .... System Timing Controller ................. 3. Solvent Delivery Requirements ......... ..... Description of the SP8700 Solvent Delivery System....... ........... The Keypad Emulator Interface.... ........ 4. Valving Requirements for MRPLC ..... . ....... 5. Column Requirements.. ............ . ......... 8. Isolated Droplet Generation...... ........... .... 1. Theory and System Overview ................. vii ll 17 22 23 23 28 33 39 44 54 59 60 60 2. Capillary Production and Mounting .......... 66 3. Constant Pressure Liquid Delivery Vessel.... ........................ 69 4. Eimorph and Electrode Mounting Assembly....... ............................ 71 5. IDG Internal Electronic Circuitry ...... .... 73 Dimorph Driver Circuit ................... 77 Droplet Pulsing Circuit...... ............ 81 Strobe Synchronization Circuit ........... 85 Bimorph Frequency Selection Circuit .................................. 87 Pulsing Frequency Selection Circuit.......... ........................ 94 Adjustable 300 V Power Supply ............ 96 5 V Fixed Power Supply ................... 99 Frequency Meter/10 MHz Frequency Source...... ..... ...... .................. 99 6. Computer-controlled Droplet Generator .................................. 101 IV. Software...................... ......... . ............ 109 A. FORTE Operating System and Programming Language............ ............................ 110 B. MRPLC Operating Routines ........................ 113 1. Control of the SP8700 Solvent Delivery System ............................ 115 2. Valve Control............ .................. 118 3. Data Acquisition and Storage ............... 118 4. Peak Finding Routine ....................... 120 5. Event Control. ............................ . 122 Time-based Event Control......... ....... . 122 Detector-based Event Control ............. 123 viii 6. Recorder Control. .......................... 125 Digital Recorder Driver .................. 126 Regraphing Utility ....................... 127 7. Other Useful MHPLC Words ................... 128 C. Computer-controlled IDG Operating Routines... ..................................... 128 D. Support Software and Data Conversion ............ 132 l. FORTE Support Programs ..................... 132 2. Data Conversion Software ................... 134 V. Evaluation and Applications. ......................... 135 A. Multidimensional HPLC ........................... 135 l. Extra-column Band Broadening in MRPLC............. ...................... 136 2. Baseline Disturbances in MRPLC ............. 141 3. Effect of Imprecision on Valve Switching .................................. 143 4. Backflushing Applications .................. 148 5. One-valve Reartcut ......................... 152 6. Two-valve Heartcut ......................... 157 7. Multidimensional Backflush ................. 159 8. Selectivity Programming with Backflush................ .................. 159 9. Autoinjector/Fraction Collector ............ 165 10. Styragel Startup Routine .................. 165 B. Isolated Droplet Generator ...................... 167 1. Operation in the Stand-alone Mode.. ........ 168 2. Operation in the Computer- controlled Mode ............................ 173 ix 3. Use of the Constant Pressure Reservoir for Solution Delivery ........ . ............. 4. Use of a RPLC for Solution Delivery ........ 5. Use of the Droplet Generator with Nonaqueous Solvents. ....................... 6. Use of the IDG to Introduce Liquids Into a Spark ............. . ................. 7. Other HPLC Interfacing Techniques .......... 8. Use of the IDG with Flames ................ . Conclusion and Future Perspective ................... A. Multidimensional RPLC.. ......................... 8. Isolated Droplet Generation... .................. References................. .................... . ........ Appendix A. FORTE Software Listings ..................... Appendix B. MRPLC Operator’s Manual ...... . .............. Appendix C. IDG Operator’s Manual. ...................... 175 177 180 186 190 198 204 205 209 221 245 266 LIST OF TABLES 122.13.11.32 22x2 1 Selected applications of multidimensional RPLC.... 12 2 Microcomputer components.......................... 28 3 AM9513 master mode configuration options.. ........ 36 4 AM9513 counter mode register options .............. 39 5 Pulsing circuit current consumption.......... ..... 85 6 Key code definitions ...... . ............... . ....... 116 7 Event Controller Commands.................... ..... 124 8 Peak data for extra-column band broadening experiment................................. ..... .. 140 9 Reartcut precision comparison .............. ....... 146 10 Initial settings for stand-alone droplet generation ..... . ................... . .............. 168 11 Low frequency droplet production characteristics.. 173 12 Initial settings for computer-controlled droplet generation...................... ......... . 174 13 Solvent charging ability .......................... 185 14 Gas flows used for IDG burner ................ ..... 203 El Rey Code Definitions............... ........... .... 247 82 Event Controller Commands.. ....................... 250 xi LIST OF FIGURES Figure Title 1 2 10 11 12 13 14 15 A hierarchical computing environment...... ..... Block diagram of the multidimensional HPLC .yate-OOOOOOOOOOOOOOOOOOOOOOO0.0 00000000000 Microcomputer system wiring diagram... .......... Block diagram of the AM9513 system timing controller............. ......... ......... ....... Block diagram of an AM9513 counter logic group.. Block diagram of the SP8700 solvent delivery 'yste-OOOO0.00...OOOOOOOOOOOOOOOOOOOOOOOO ....... Keypad array for the SP8700 solvent delivery system......... ...... . ............. . ...... . ..... Keypad driver timing diagrams. Pulse length = 0.085 msec. Cycle period = 2.42 msec............ Keypad emulator circuit............. ............ Flowchart of keypad emulator operation. ......... Key codes for the SP8700 keypad emulator interfaceOOOOOOOOO0.0000000000000000000000 ...... Valve control and sensing circuits.............. Configurations of a 6-port, 2-position valve.... Basic components of the isolated droplet generator.OOOOOOOOOOOOOOOOOOOOOO0.0.0.0....00... An improperly fire-polished capillary tip (43 um diameter)........................ ........ xii 32 48 49 52 53 56 58 61 Figure Title 16 17 18 19 20 21 22 23 24 25 26 27 28 29 A properly fire-polished capillary tip (83 um diameter).. ...... . ....................... Solution delivery reservoir. 1: Aluminum cylinder, 5" x 3” o.d. x 2.6” i.d. 2: Brass cap 0.75” x 3.25” o.d. 3: Viton rubber seal. 4: Copper tubing 1/4". 5: 1/4" Swagelok fitting. 6: 1/16” inverted fitting. 7: 2 pm in-line filter. 8: 1/16” SS to 1/16” cheminert adapter. 9: 1/16” SS tubing. 10: Microline tubing. 11: Glass capillary. 12: 1/4” to 1/16" Swagelok union. 13: Inlet filter. 14: Polyethylene liner (optional)........... ..... Bimorph and electrode mounting assembly (shown approximately 1.2 times enlarged). 1: 1/4” Plexiglas. 2: High voltage deflection plates. 3: Screw clamp. 4: Insulation. 5: Gimbal mount for positioning capillary. 6: Bimorph. 7: Electrical connection for bimorph. 8: Bimorph ground connection. 9: Cylindrical charging electrode 1/4” x 1/4” o.d. x 1/8" i.d. 10: Charging electrode positioning bracket ...... ‘Block diagram of the stand-alone version of the isolated droplet generator........... ........... Bimorph driver circuit. ..... .. .................. Circuits used to measure bimorph capacitance.... Bimorph driver output waveform.... ...... ........ Bimorph driver amplitude for various dial settings. ( ) Peak-to-peak voltage. ( ) Flat-to- fl‘t valta‘QOOOOOOOOOOOO000...... ......... COO... Droplet charging circuit ........................ Droplet charging circuit output. A: 43 us. B: 13 us. C: 8 us. D: 20 us. E: 2 us. F:12.5".00000000000000000 ....... O. 00000000000 0 Strobe lamp synchronization circuit ............. Bimorph frequency selection circuit ............. Bimorph frequency divider circuit ............... Bimorph freqency selection timing diagram ....... xiii Page 68 70 72 74 78 so 32 83 84 86 88 90 91 93 Figure Title Page 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 Charging frequency divider circuit .............. 95 Adjustable high voltage power supply ............ 97 Schematic of 5-volt power supply.......... ...... 100 Block diagram of the computer-controlled version of the isolated droplet generator ......... . ..... 103 Configuration of the AM9513 system timing controller for isolated droplet generation ...... 104 Comparison of timing diagrams for the stand- alone mode and the computer-controlled mode of the isolated droplet generator .................. 107 Pictures of the isolated droplet generator (top) and bimorph mount (bottom) ...... .. ......... ..... 108 Organization of multidimensional RPLC operating routines ...................... . ....... .......... 114 Isolated droplet generator controller flowchart ...... . ................................ 129 Configuration used for band broadening evaluation ...................................... 138 Representative chromatograms from band broadening evaluation ........................... 139 Example of baseline disturbance due to sudden backpressure change ....... ........ ..... ......... 142 Examples of baseline disturbances due to refractive index change caused by mixing of heartcut solvent and developing solvent.. ....... 144 Top- Program used to automatically generate the previous series of baseline disturbance chromatograms. Bottom- Valving configuration used for baseline disturbance experiment........ 145 Valving configuration used to test detector- based valve switching .................... . ...... 147 Single-valve backflush configuration.. .......... 149 Examples of backflush chromatograms............. 150 xiv Figure Title 47 48 49 50 51 52 53 54' 55 56 57 58 59 60 61 Single-valve heartcut configuration. ............ Single-valve heartcut of residual fuel sample, amino column to silica column ................... Sequential heartcut configuration using a single valve .................................... Two-valve heartcut configuration..... ........... Configuration used for multidimensional bQCRflu.hin‘OOOOOOOOOOO ......... 0.0.0.... ....... Multidimensional backflush of model compounds, amino column to silica column ................... Configuration used for selectivity programming with backflush ........... . .......... . ........... Selectivity programming with backflush on heteroaromatics fraction of residual fuel ....... Two valves configured for autoinjection and fraction collection ............................. Top- Capillary jet without imposed oscillations Bottom- Capillary jet with longitudinal oscillations imposed by bimorph ................. Top- Droplet stream showing poor phasing adjustment. Center- Droplet stream showing 1 charged droplet deflected. Bottom- Droplet stream showing 1 neutral droplet separated ...... Top- Stream showing 10 charged and 10 neutral droplets. Center- Stream showing 5 charged and 15 neutral droplets. Bottom- Stream showing triangular droplet packet... .................... Flow rate/pressure relationship for a 30pm diameter capillary and in-line filter ........... Effect of a 48/minute water to methanol gradient on droplet formation ................... Effect of a 48/minute water to methanol gradient on droplet formation showing viscosity Chan‘OOOOOOOOOOOOOOOOOOOOOO. OOOOOOOOOOOOOOOOOO O. XV Page 153 154 156 158 161 162 164 166 170 172 176 178 181 183 Figure Title 62 63 64 65 66 67 68 69 70 71 72 Effect of a 48/minute water to methanol gradient on droplet formation showing surface tension change .................................. Droplet generator used with miniature nanosecond spark source.. ......... . ............. Effect of droplet introduction/spark trigger phase shift on emission signal .................. HPLC using AAS detection. Direct connect nebulizer interface .............. .. ............. Dripping cup interface ............. . ............ HPLC with AAS detection using the dripping cup interface. ...................................... Comparison of signal magnitude for dripping cup and direct interface methods ........ .. .......... Comparison of band broadening for dripping cup and direct interface methods .................... Top: Droplets being introduced into a flame. Bottom: Signal obtained from droplet introduction. Vertical axis: relative signal. Horizontal axis: 0.015 ms/div.... ..... .......... Design of a sheathed air-acetylene Meeker burner for isolated droplet emission spectroscopy ........ . ...... ........ ............. Configuration used for heartcut recycle chromatography..... ..... .. ...................... xvi Page 184 187 189 191 193 194 196 197 200 I. INTRODUCTION The term ”multidimensional chromatography” was first used in conjuction with the practice of re-developing thin-layer chromatography (TLC) plates (1). After the end of the first separation, the plate is rotated 90° and developed a second time. If the solvent always travels in the vertical direction, the rotation of the plate changes the direction of solute travel on the plate and gives rise to a "multidimensional" thin-layer chromatogram. This technique can be employed to obtain additional resolving power based on simple physical separation as a result of turning the TLC plate. An additional enhancement in resolution can be obtained by varying the second mobile phase so that a different selectivity is produced for the second dimension. When used in modern high performance liquid chromatography (HPLC), the term ”multidimensional” generally refers to a collection of techniques involving switching a portion or portions of the effluent from one column onto a second column for further separation. Other common terms currently in use to describe this technique include column-switching, multiphase, multicolumn, or coupled-column chromatography (2). The use of the same or different stationary phases packed into the same column or connected in series is not considered multidimensional chromatography, since no switching mechanism is involved. Several other related techniques which use similar instrumental setups are also commonly encountered in multidimensional chromatography. The most common of these are backflushing and trace enrichment. Backflushing refers to the practice of using a valve to reverse the mobile phase flow direction after some of the components have eluted from the column. Trace enrichment involves pumping a large quantity of sample through a column containing a stationary phase' which strongly adsorbs the components of interest. The concentrated components are then eluted with a different solvent, and a portion is transferred to a second column for further separation. Multidimensional chromatography can be carried out in an off-line or an on-line manner. Earlier approaches were usually of the former variety, where a portion of the first chromatogram was collected, evaporated, reconstituted, and finally, injected onto a second column. For convenience, automation, and better precision, on-line techniques are generally preferred. Instrumentation for implementing on-line multidimensional HPLC (hereafter referred to as MHPLC) has generally involved multiple pumping systems and time-based event controllers. There are several drawbacks associated with such systems. First, the instrument cost is primarily dependent upon the number of pumps, since the pump is the most expensive component of an HPLC system. Thus, multiple-pump MHPLC may be prohibitively expensive for some laboratories or applications. Second, time-based valve switching is often analytically unreliable. Retention times in HPLC can vary considerably from sample to sample, making it necessary to reprogram the event controller. This is especially a problem with silica and alumina columns, where slight changes in water content drastically affect chromatographic behavior. Third, without centralized control of the experiment, such systems suffer from a lack of versatility in coordinating the actions of the various components. This dissertation presents an MHPLC instrument which overcomes many of these difficulties. A single, intelligent solvent delivery system is described; this system can perform many of the functions which previously required multiple pumps. A central nicrocomputer controls and coordinates the entire experiment, including valve switching, data acquisition, and solvent delivery. Due to the added versatility of centralized control, intelligent valve switching can be performed based on the chromatogram itself. Valve switching can be based on the number of peaks detected, the percentage change in signal, the absolute signal level or the time after injection. A unique interactive operating system based on the FORTH language allows creative control procedures to be written. Such procedures can include repetition based on loop structures and conditional execution of events by using IF-ELSE-THEN constructions. Multidimensional Detection in MHPLC In the above discussion, the term ”multidimensional” was introduced to describe a method of column-switching in chromatography. This term also has a much broader use, which refers to the nature of the data obtained from such an experiment. One example may be the coupling of a size-based separation with a polarity-based separation. The first dimension of information concerns the size of the species eluting from the first column. The second dimension, obtained when fractions are transferred to the second column, contains polarity information. When used in this sense, the term "multidimensional” can be extended to the detection process as well. If a diode-array UV detector were to be employed, the third dimension of data would contain information on the UV spectrum of the eluting components. Additional non-destructive detectors could be added in series to give additional dimensions of information. The potential for maximizing the information content from each chromatogram in this manner is a very active field of research. One type of sample information useful to the chromatographer is a breakdown of the eluting components by element. Several researchers have used element-selective detectors to obtain empirical formulae of components eluting from a 60 column (3), but interfacing liquid eluent to such detectors has been difficult. Nebulizers are commonly used to deliver liquids to flames, plasmas, and sparks for emmission-based detection. Most nebulizers, however, are inefficient and contain dead volumes too large for use in HPLC. In addition, there is often a high correlation between nebulization efficiency and liquid viscosity, which would severely limit the use of nebulizers in gradient HPLC where mobile phase viscosity changes dramatically. In addition to the above MHPLC instrumentation, this dissertation presents an alternative to the nebulizer for introducing HPLC effluent into atomic excitation sources. The device is called an isolated droplet generator (IDG) and is based on principles similar to those used in some ink-jet printers. Liquid is forced through a capillary attached to a piece of piezoelectric material called a bimorph. The bimorph is vibrated, imparting longitudinal oscillations on the emerging jet. This causes the jet to break into reproducibly sized and shaped droplets. Selective charging and deflection is used to deliver the droplets to the excitation source. Isolated droplet generators have been used in some form for nearly 35 years but this is the first instrument constructed that has been placed under complete microprocessor control. There are several advantages to this design. The droplet streams produced are very stable and reproducible. Droplet production parameters can be altered under computer control to respond to experimental needs or changing liquid composition. While the main focus will be on the use of the IDG as a possible HPLC interface, the instrument is versatile enough to function as a general purpose liquid handling device capable of manipulating liquids at the sub-nanoliter level. II. HISTORICAL BACKGROUND In this chapter the historical progression of published research which pertains to this dissertation is presented. Specifically, three reviews of the literature were performed. First, an extensive review of the major chromatographic journals has been assembled with respect to those references dealing with multidimensional liquid chromatography. Second, a somewhat less extensive but fairly representative review of literature dealing with element-specific detection is presented. The last section contains a comprehensive presentation of published research dealing with the development of isolated droplet generators. Multidimensional HPLC Multidimensional chromatographic techniques have been carried out for many years to assist in the separation of complex mixtures. Prior to the commercial development of modern HPLC instrumentation, thin-layer chromatography (TLC) was used extensively in this regard. To perform a multidimensional separation using TLC plates, the plate was rotated 90° and re-developed, usually with a different mobile phase. Since the advent of high-performance TLC plates, this technique has seen a minor resurgence in popularity. A good review of current multidimensional TLC applications has been written by Zakaria, Gonnord, and Guichon (4). Modern column-switching techniques began appearing in the literature soon after the introduction of the first high-pressure, low dead-volume valve by Huber, Van Der Linden and Ecker (5) in 1973. A previous paper by Scott, Chilcote and Lee (6) in 1972 used a modified 6-port, 2-position injection valve to assist in the development of ion-exchange columns, but the time scale of the separation (14 hours) does not put it in the category of ”modern" HPLC. The original concept of using valves to switch columns can be credited to the inventors of commercial water purification systems. Perhaps it was the success of these systems. which prompted Snyder (7) and Liljamaa and Hallen (8) to suggest similar schemes for chromatographic applications. Successful implementation of column-switching using high-performance columns required the minimization of extra-column band broadening resulting from dead volume in the valves and associated tubing. Dolphin and Willmott (9) addressed this problem in a paper in 1976, and concluded that the contribution of the valve to overall peak widths was not significant. Their theoretical evaluation compared well with the experimentally determined broadening contribution presented by Huber (5). Both authors found that long pieces of capillary tubing can cause significant broadening effects, however, and thus tubing lengths should be kept to a minimum. As a complement to the band-broadening theory mentioned above, several papers dealing with the theoretical limits of resolution in multidimensional HPLC have been published. A recent report by Giddings (10) compares the separation power obtained from a variety of 2D separation techniques. The maximum separating power is obtained when the two separation mechanisms are totally independent. In this case, the predicted peak capacity is the square of the average capacity expected from a 1D separation. As the two separation mechanisms become more correlated, the peak capacity decreases. Snyder (11) reaches similar conclusions and also demonstrates several throughput advantages of 2D separation techniques. Guichon (12-13) has published several papers on this topic as well. Huber (14) has published a related paper which may be of interest on the application of information theory to multidimensional gas chromatography and low-resolution mass spectrometry. Another related paper explains the hysteresis effects observed when 60 columns are backflushed (15). This paper is interesting in that it also explains why this: phenomemon is not observed when backflushing HPLC columns. Several research groups have been quite prolific in the field of multidimensional HPLC. Frei of the Free University of Amsterdam has published many papers, several in conjuction with Little and Stahel of Kontron Ltd. (16-18); Werkhoven-Goewie, also of the Free University (17.19-20); Erni, of Sandoz Ltd. (21-22); and Nielen and Brinkman, also of the Free University (23-25). Another researcher who has made several contributions to the literature is Majors of Varian Corporation(26-28). He has also collaborated with Apffel of Virginia Polytechnic (29), and Johnson and Gloor, also of Varian (30). Other prominent researchers include Dolphin and Willmott (9,31-32), Huber and Van Der Linden (5,33-34), Miller (35-36), and Harvey and Stearns (37-38). Of the instrumental setups reviewed for column-switching, only one was found that used an on-line detector to trigger a valve change (32). The remainder used time-based sequencing. Approximately 503 employed two pumps (9,11,18-21,30,35,39,40-43), 31* used a single pump (5-6,29,32,37,44-46), and 193 used three or more pumps (16-17,47-49). Most systems employed a single valve (5-6,9,l9,21,30,32-33,35,37,39,42-43,45-46,50), to perform either a simple heartcut (5,9,19.21.30.32-33,39.42-43.50), backflushing (37,35), or recycling (45). Of the remainder of the systems reviewed, 188 used two valves (ll,29,4l,44,47) and 25x used three or more valves for more complex switching arrangements (16-18,20,23,48-49). Eleven 10 of the systems used two or more detectors (6.9.11,16-17,21,29,32,42-45); the remainder used only one. Most applications of column-switching fall into the three categories of sample cleanup (l7,20,43,46,5l-52), trace enrichment (17.19.22.24-26,41,53), or class separation (54-56). The most common matrix was biological fluids, usually where the analyte was a pharmaceutical product (6,16-18,20,22,27-29,39,41-42,44,47,50,52,57). Other common matrices were foodstuffs (16-17,21,27,30-31,33,40,44,46), petroleum (35,37,54-56,58-60), and aqueous waste streams (17,19,23-24). Table 1 summarizes applications from some selected references. In addition, several reviews and general discussions of multidimensional HPLC have also been published (16,26-28,35,61-63). Element Specific Detection The problem of positive identification has always existed in chromatography. Since the separating power of a column is limited, one always runs the risk of mistaking the identity of a peak for another compound having a very similar retention time. This is true even for the best case, where the peak identity is known and a standard exists for comparison. For those cases where an unknown sample having many components is separated, identification is tedious at best and impossible at worst. 11 Table 1. Selected multidimensional applications. Sample type Primary Mode Secondary Mode Reference Pesticides in milk LSC LSC 31 Herbicides in cereals n-BPC n-BPC 33 Drug metabolites in plasma and urine RPC RPC 44 Acids in wine RPC RPC 17 Chlorophenols in water RPC RPC 17 Malathion in tomatoes GPC RPC 30 Limonin in grapefruit peels GPC RPC 30 Additives in rubber stocks GPC RPC 30 Vitamins in food protein supplement GFC RPC 27 Antibiotics in serum 180 RPC 35 Sugars in candy GFC RPC 29 Ettre (64) has put the problem in perspective with a simple calculation. Assuming a 50-m long glass open tubular capillary column with a HETP of 0.33 mm for néhexadecane, the separation number for the pair of normal paraffins Cl and Cie would be 97. This means that about 100 peaks can be separated in the time interval between the two paraffins, in this case about 20 minutes. Although this is a remarkable performance, the number of possible compounds that could elute within this interval is probably at least ten times this amount. Fortunately, for industrial samples, the number of species is limited by the chemical process itself. However, for natural or biological samples this is not true and some 12 other means must be used for peak identification. Substance-selective detectors can play an invaluable role in this regard. Selective detection is not new to chromatography. The term ”chromatography" itself implies detection by color -- a fact which Tswett (65) emphasized in his first paper. Before going further, a point of distinction should be made between ”selective" detection and ”specific” detection. Specific detection refers to a process that only gives a response to a single compound. Selective detection, on the other hand, gives a response to a class of compounds. Egan (66) presents a good discussion of this distinction. Depending upon the point of interest, sometimes a detector can be either selective or specific. For example, an atomic absorption detector is specific for a given element, but selective for the class of compounds containing that element. The first limited-response detector was, in fact, a specific detector. In the late 1950’s, Bayer and Anders (67) extracted the glands of nine female silk moths and injected the extract into a 00. At the end of the 00 column, they placed a male silk moth in a small box. A number of peaks were obtained and most of the time the male moth remained motionless in the corner of the box. But when one particular peak eluted, he started to wiggle. his wings and run around, becoming very excited. This particular 13 peak, of course, corresponded to the pheromone of the female silk moth. Selective detectors for liquid chromatography first appeared around 1933 and were based on UV spectroscopy (64). Today, most LC analyses are done with the help of selective detectors. Commercial detectors are available based on UV or visible absorbsnce, electrochemical redox reactions, post-column chemical reaction, and fluorescence. Several reviews have been published on the current state of detector technology for HPLC (64,68-70). In contrast to the development of selective GC detectors, which were designed specifically for this purpose, most selective LC detectors actually represent existing analytical instruments modified to permit the direct coupling of the column effluent. Thus, the ability of the interface to deliver the column effluent efficiently and with a minimum of band broadening often determines the success of the detector. This is especially true of instruments where the detection process occurs in the gas phase. Most of the research concerning element-specific detectors for liquid chromatography has been centered around mass spectrometry or atomic spectroscopy. Of these two techniques, the classification of a mass spectrometer as a ”detector" for LC may be somewhat of a misnomer. It is probably more appropriate to call the less expensive LC an 14 51 IO 1. introduction technique for the mass spectrometer. The interfacing requirements of the two systems are quite different, since most mass spectrometry takes place at very low pressures. Nevertheless, some interesting work has been published concerning LC/MS interfacing techniques (71-73). Atomic spectroscopy has been interfaced with HPLC by a variety of researchers. Van Loon (74) has published an excellent review on this topic. Of the instrumental innovations noted, most dealt with novel ways of interfacing the HPLC effluent with the instrumentation used for atomic spectroscopy. Van Loon notes that HPLC is incompatible with most AAS nebulizers and with the stepwise operation of commercial electrothermal atomizers (74). The liquid flow rate used in HPLC is not compatible with most concentric nebulizers, but is often compatible with crossed-flow nebulizers used in many inductively coupled and DC plasma atomizers. Many approaches have been used in designing an interface with the following typifying some of the more common. Brinckman et a1. (75) used a carousel-type automatic sampler to collect fractions for subsequent analysis by furnace AAS. The advantages of this approach are said to be -better detection limits and small sample size. A modification of this approach was used by Vickrey et a1. (76) where a UV detector is used to identify the peak of 15 interest and the entire peak is collected for subsequent analysis by a furnace AAS unit. Because the flow rate typically used in HPLC is too slow (less than 2 mL min'l) to be interfaced directly to the nebulizer of an AAS burner, Slavin and Schmidt (77) used a discrete injection technique. In this approach, the effluent from the HPLC column is allowed to drip into a teflon funnel which is connected to the inlet of an AAS nebulizer. Each drop is then atomized separately, and gives a chromatogram comprised of a series of spikes. The authors claim that 0.1 mL droplets are sufficient to give a steady state AAS signal; thus the technique retains the full sensitivity of continuous AAS. Further investigations of this technique can be found in the ”Evaluation and Applications” chapter of this dissertation. Several different nebulizers have been evaluated in terms of dead volume and efficiency by Hausler and Taylor (78). They found that a spray chamber utilizing a drain that exited on the same side as the nebulizer decreased peak broadening by allowing a rapid and clean drain. A spray chamber of this type is now commercially available from Applied Research Laboratories (ARL). Another interesting nebulizer design was published by Lawrence et a1. (79). In this design, a microconcentric nebulizer is mounted directly under the plasma, thus avoiding dead volume associated with a spray chamber altogether. While peak broadening is minimized, the nebulizer was not operated above flow rates 16 of 0.2 mL min’l. Problems with plasma instability were also noted. Solvents are often a problem in LC using atomic spectroscopy detectors, especially when plasmas are used as the excitation source. Many applications that use reversed-phase or ion-exchange solvents, where a major component of the mobile phase is water, have been published (80-83). Nonaqueous mobile phases, however, can cause instability with plasmas and carbon buildup with both plasmas and flames (83). Microbore HPLC is an alternative which may relieve some of the solvent effects, but there appears to be a loss in sensitivity due to the smaller sample sizes associated with these columns (84). Isolated Droplet Generators From the preceding discussion it is obvious that a major limitation in utilizing LC-AAS or LC-AES is the lack of an efficient, low dead-volume interface. One device which may prove useful in this regard is the isolated droplet generator. These devices can be made with very low dead volume and the amount of sample transferred can be regulated to suit the excitation source. The discovery of the isolated droplet production phenomenon is usually attributed to Lord Rayleigh (85-86) in 1878. While these papers were the first attempt to explain the theory of droplet formation, the first experimental 17 observation of droplet production was reported by Savart (87-88) in 1833. Both of these authors reported conditions under which a liquid jet breaks apart in a uniform manner. Since the first reports of droplet production, many different types of droplet generators have been reported.. In 1947 Lane (89) constructed an apparatus in which a droplet was blown off the tip of a capillary by a jet of air when the droplet was of the desired diameter. Mason and Brownscombe (90) constructed a similar apparatus in 1964 which could also produce charged droplets. Cheng and Cross (91) also constructed an air-actuated droplet generator in which a concentric stream of air controlled the velocity of the droplet. Another type of droplet generator which several researchers have constructed is based on a vibrating reed principle. The reed, or stylus, is dipped into a reservoir where a small amount of liquid clings to its surface. Upon withdrawal, a filament of liquid is drawn out of the reservoir. The filament then collapses upon itself, forming a droplet, which is allowed to fall under the action of gravity. The first droplet generator of this type was reported by Wolf (92) in 1961. While the fastest droplet production rate was only 120 Hz, the droplet size could be varied between 10 and 50 microns by adjusting the penetration depth, diameter, and construction material of 18 the stylus. Abbott and Cannon (93) and Shabushnig and Hieftje (94) have also constructed droplet generators of this type. A disadvantage of this type appears to be the large dead volume associated with the solvent reservoir and the inability of the stylus to draw filaments from solutions with low wettability characteristics (nonaqueous). The third, and most common, type of droplet generator is based upon the induced breakup of a liquid jet. These droplet generators are most closely related to those originally described by Rayleigh (85-86). The first was described in 1964 and in subsequent papers by Schneider et a1. (95-96). This type is based on a capillary vibrated by a piezoelectric crystal. The vibrations are transferred to the emerging jet and cause it to break up into droplets under the action of surface tension. The size of the droplets is determined by the wavelength of the disturbance launched onto the surface of the jet. Lindblad and Schneider (97) published an improved version which employed a different mounting system for the capillary. All three versions were capable of producing charged droplets which could then be deflected into a trap when not needed. Hendricks and Schneider (98) published a paper which deals with the theory of droplet stability. This work is basically a detailed re-derivation of Rayleigh’s equations using modern notation. Hieftje and coworkers (99-105) produced several versions of droplet generators for 19 application in analytical chemistry. Bastiaans and Hieftje (99) used the droplet generator in flame spectroscopy to perform high precision "null point" measurements. Hieftje and Malmstadt (100-101) also used droplet generation to study fundamental flame spectrometric processes. Hieftje and Mandarano (102) and Lemke and Hieftje (103) used droplet generators as a basis for an automated titrator and a pH-stat, respectively. Steele and Hieftje (105) constructed a micro-titrator based on this method of liquid introduction. An updated controller for this type of droplet generator was presented by Russo et al. (104); it uses a crystal oscillator-based frequency source for enhanced stability. Further advances in control circuitry have been presented by Seymour and Boss (106). The "Instrumentation" chapter of this thesis also contains a description of the author’s control circuitry, which allows complete computer control of droplet production parameters. Two other variations of the jet-type droplet generator have also been published. The first variation uses a vibrating orifice instead of a vibrating capillary. This type was used by Joshi and Sacks (107) in a circular slot burner design. Some ink-jet printers use a similar vibrating orifice design (108) and TSI, Inc. of St. Paul, Minnesota manufactures a commercial instrument of this type for controlled aerosol production. 20 The second variation was published by Willoughby and Browner (73) as the ”MAGIC” interface for LC-MS. In this design, which uses a capillary jet, the vibrations are not artificially imposed. already present on disruption and an air they have time to changing the diameter Natural background frequencies the jet are allowed to cause jet stream separates the droplets before coalesce. Droplet size is varied by of the jet. 21 III. INSTRUMENTATION In this chapter, the instrumentation pertinent to the research in this dissertation is presented. The first section deals with the multidimensional liquid chromatograph. This instrument contains some modules which are commercially available; these are described to the extent that is necessary for comprehension of their function within the instrument. The other modules which have been designed in-house are presented in more detail. The second section discusses the isolated droplet generator. This instrument has been completely designed in-house and is presented here in two phases. The first phase in its development was an implementation as a stand-alone system with independent control circuitry. In the second phase, the instrument was placed under computer control, making it much more versatile and easy to use. Both phases are discussed in detail. 22 A. Multidipensional HPLC l. S t m 0 v ew Any HPLC system consists of four basic components: a solvent delivery system, a column to perform the separation, a detector, and a readout/recording device. A multidimensional HPLC system has two additional components: valving to alter the chromatographic flow path and a controller to properly sequence the experiment. The simplest control device for sequencing multidimensional chromatography is a chemist with a stopwatch and a quick hand. Due to the tedium of this approach, however, an) automated controller is highly desirable. Most controllers used in the past have been time-based event sequencers equipped with mechanical relays which can then be used to trigger valve changes or start and stop a pump. While this approach is very helpful, developing a reproducible time-based method is difficult. The main reason for this is that HPLC retention times can vary significantly from one injection to another. A much better approach, which is implemented in the system described herein, is to base the events on the chromatogram itself, as well as time. This has been accomplished by digitizing the detector signal and utilizing a microcomputer to analyze the data in real time. 23 A second consideration in constructing the multidimensional HPLC system was cost minimization. Previous designs delivered each solvent mixture with a separate pump and switching valve. These components add a great deal of expense to the system. An alternative is to mix the solvents prior to the high-pressure delivery stage with a single proportioning valve and deliver the resulting mixture with a single pump. Intelligent, microprocessor-controlled solvent delivery systems of this design are capable of gradient operation, solvent selection, flow rate programming and other advanced features needed in MHPLC. Since the main cost of a solvent delivery system is the hardware, one intelligent pump can be purchased for much less than several ”dumb” pumps of comparable quality. A third consideration in constructing this multidimensional HPLC was to provide full integration of all components of the instrument. This was accomplished by using an 8085 microcomputer as a centralized controller. Figure 1 illustrates how the microcomputer operates in the laboratory computing environment. Although this is a hierarchical computing environment, the microcomputer is actually the controller for the entire experiment, issuing commands to both lower and higher levels of the hierarchy. With intelligent centralized control, the term "event" takes on a much broader meaning than a simple valve switch. An experimental event could be a change in the data taking 24 304d mohm mi @000 .anolcosfl>so unease-co usofisonssowa < hzoo mum<4 m1 nmom ~III mutaazogzi nm\_p_wn mxm_o xmhmuroz_3 wu>_zo >Q¢Onm xuhz.¢m wo4<> hum >3 a: moon mo_rae can: ”snowecolwcmunsl on» no lsausan Joey: 2052(qu mmDth %od QIO .2. T--- 96 mZIDJOO xzw 96 mopounz. JHII fl u>4<> Q23 02.20:.moaomm a >mmumum mommmooma ...zw>..om IOKO.) OmN OONQQW Wu .N shaman mohouemo «momoomm >3 pm4<> npmm1< 0:“ fikDLx mmmm muhaazooomo:‘ \ mmom nupZ. . e ; amenazoo;:z m0u3 2 Ial m 3 m 2 3mm 35 Ba p N p N mmomoomm ..T.m2mm ezoo goes L I L I MPMP ozo P>n>e>n> w u w u :o ¢< cenmam R x/ R // , mzmm #200 E .I E 1. Z. .IJ T 1 r 1 >_ N >..> m 3> 1!. 0+ mZMm moemx 2_ EB E ozm 0.2 ._T.)>>l, _ _ _ xeomuaxe .20 some. mmeo nmmsnmeo SNDSN: .Zo> oo< It; 2; o :33 o :33 < Eon. msm.o _ m possum #212. saou.:m.. each J fl m w L ._ ¢~oe ooummw on» Me lsnwsuv aooua .m ousmuh azaa mmxmmx m>n<> xi 0220.50.55 >mmx 42 whose output is connected to the pump motor drive circuitry. This loop is used to provide short-term pressure control. The second loop is a digital feedback loop which monitors the number of motor steps and the cam marker. The pump progress is checked ten times each cam cycle by measuring the time required to complete a fixed number of motor steps. The difference between the measured time and a reference time is used to generate a new target pressure which is then sent through a DAC to the other input of the differential amplifier in the first feedback loop. The difference between the new target pressure and the actual pressure sensed by the pressure transducer produces a signal which controls the pump motor speed. Thus, the first loop provides short-term pressure control, while the second loop provides long-term flow control. After leaving the pump through an outlet check valve, the solvents pass through a static mixer to a column bypass valve. This valve is used for priming the pump or clearing the mixing chamber of old solvent. The static mixer is simply a 10 cm column packed with glass beads. A silica pre-saturator column has been used in place of the static mixer with no apparent change in pump performance. Solvent delivery parameters can be entered manually through an 11 x 4 keypad array on the front of the instrument, or automatically through the keypad emulator described below. The memory of the SP8700 is organized into 43 ten parameter files containing seven entries each. These files provide some limited programming capability which is usually adequate for simple, single-column operation. For multidimensional chromatography, however, seven entries are not sufficient to completely describe a run, and there is no provision for automatically linking one file to the next. There is also no provision for permanent storage of parameter files. For these reasons, and others discussed below, a keypad emulator was designed to interface the SP8700 to the 8085 microcomputer. The Keypad Emulator Interface The addition of a means to allow the 8085 microcomputer to control the SP8700 increased the capabilities of the system in several ways. A larger number of parameter changes can now be entered, and parameter files can be automatically linked. The parameters can be stored permanently, on hard disk, floppy disk, or as hard copy; disk-stored parameters can be retrieved and re-entered at will. The SP8700 clock and the microcomputer clock can now be synchronized so the entire experiment takes place in the same time frame. Finally, since the microcomputer is sensing the detector signal as well as controlling the SP8700, a feedback loop can be established so that the parameters of the HPLC experiment can be altered on the basis of the chromatogram itself. 44 Keypad emulation is a general method of control applicable to microprocessor-based instruments that use ”polled” keypad arrays (114). The method uses an integrated circuit (IC) multiplexer and an IC demultiplexer to simulate a keystroke by connecting the desired column of the array to the desired row under computer control. Keypad emulation utilizes fully the existing instrument circuitry, and allows the internal microprocessor to perform its built-in logic and error-checking functions. Manual keypad entry remains undisturbed. Operation of the interface uses simple parallel input/output, so there are no restrictions on the choice of laboratory microcomputer or programming language. Timing restrictions are minimized, since the keypad array circuitry is designed to accept asynchronous, variable-length activation (i.e., manual depression of a key). Other advantages include low cost and the need for only a few chips. By far the best feature, however, is that the emulator operates independently of the rest of the instrument, so only a minimal knowledge of the instrument circuitry is necessary for implementation. For use in a microprocessor-based instrument, a keypad array must have driver and receiver circuits as shown in Figure 7. The driver circuit scans across the columns of the array, activating each column in turn by sending it a logic-level pulse. If a key is depressed, the pulse is transmitted along the proper row to the receiving circuit, 45 .loumhm >ho>wuov aso>uem cabana OAu sou ashes stSOM .h ossuwh Iotsm 50x1... mnoza mgom macaw” (I EDOEO mmZmQ Qmx 46 where it is sensed and acted upon. By knowing the column that was activated and the row from which the pulse was received, the microprocessor can deduce which key was depressed. The keypad driver and receiver are often contained in a single 10 as in the National Semiconductor MM5740 (115). The activating pulses are active-high in some implementations, but active-low in others. Whatever the specific logic of the driver circuitry, the general strategy for keypad emulation is the same. The column selection is made by an IC multiplexer, and the row selection by an IC demultiplexer. This provides an alternate path for the activating pulse to travel from the driver to the receiver circuit. To implement the keypad emulator on the SP8700, it was only necessary to determine the logic level of the activating pulse. The characteristics of the signals entering the array were obtained with an oscilloscope. The driver circuit was found to supply active-low pulses 85 us in duration. An entire scan cycle was completed in 2.420 ms. The logic analyzer traces are shown in Figure 8. Figure 9 shows the complete interface scheme. The multiplexer is a one-of-sixteen data selector (74150). The input line is chosen by the logic levels established at control inputs A-D. In this way, four PIO lines are used to select which keypad column is the input to the multiplexer. A fifth PIO line is used to gate the interface on and off by 47 500 ps/div 125 [is/div Figure 8. Keypad driver timing diagrams. Pulse length = 0.085 msec. Cycle period = 2.42 msec. 48 .awsosfio sousaslo cannon .m ossumh :Domzo mm>_mo omx in. .Cbumfiu xuauu ch zo_._.ou.—uo oux Hf #30 (...< what; 0: .uu ousufib _.o mupzu . O mmummczama hm”: 0(10 m...“— 0.. m N P aoemmomiizoo 30: 23. :2. >55 4, ex... 3.... mi... us: I m m .v «~s> .NH ousuwm ”30> 0... AT buzo._<> 5.22:3 o<> oNFoiiiiLn/rw A:ozmnom emoaim zoEmoaiu “m2(4 56 is connected to the input of the relay; two positions are used for manual control of the valve position and the position labeled "TTL" is connected to the PIO port of the microcomputer. A low level at the PIO line turns the solenoid valve on. The position of the valve is determined by the state of a magnetically operated reed switch mounted on the valve housing. A magnet is mounted on the valve shaft and moves into proximity of the reed switch when the valve is in the clockwise position, thus closing the switch. A change of state thus occurs at the input bit of the PIO, enabling the microcomputer to determine if a valve rotation was successful. Several different styles of valves have been used for multidimensional HPLC, but we have found a combination of two 6-port, 2-position valves to be the most versatile. Figure 13 shows several ways a single valve of this type can be connected for a variety of applications. Two valves of this type allow the user to configure the system over 30 ways, which should accomodate a considerable number of switching applications. The autoinject and fraction collection configurations, although not strictly column-switching in nature, are nonetheless valuable applications of the system. The autoinject mode allows the user to repeatedly inject a sample contained in a_ sealed sample vessel. The sample is forced into the injection loop by the same air pressure that drives the actuator piston. 57 .o>~s> sonuaeoaiu .usonim e no esoaasaswuusou .m~ osswuh \ / \ / 223400 O._.ZO IwDiixo umnmmmmaiiw mna2m 223400 .m Homfiz_0._.3< .u Iwznux0f 13 i Figure 17. Solution delivery reservoir. 1: Aluminum cylinder, 5" x 3” o.d. x 2.6" i.d. 2: Brass cap 0.75” x 3.25” o.d. 3: Viton rubber seal. 4: Copper tubing 1/4”. 5: 1/4” Swagelok fitting. 6: 1/16” inverted fitting. 7: 2 um in-line filter. 8: 1/16” SS to 1/16” cheminert adapter. 9: 1/16” SS tubing. 10: Microline tubing. 11: Glass capillary. 12: 1/4" to 1/16" Swagelok union. l3: Inlet filter. 14: Polyethylene liner (optional). 70 a fitting in the cap, forcing liquid through a sintered metal filter to a 1/16” fitting also in the cap. Attached to the outlet fitting is a 0.45 um in-line filter of the type used in HPLC (Rheodyne 7335). After the filter is a union which adapts from 1/16” stainless steel tubing to 1/16" plastic tubing compatible with Cheminert fittings. The plastic tubing was 0.04" i.d. and 0.07” o.d., which is a size easily force-fit onto the end of the capillary. A number of tubing materials were evaluated for solvent compatability, and Microline tubing (Thermoplastic Scientifics, Inc., Warren, NJ) was found to be compatible with all HPLC solvents, yet still flexible enough to be force-fit. 4. B mor h and Electrode Mountin Assembl Figure 18 shows the method used to hold the electrodes and bimorph. The mounting plate was constructed of 1/4” Plexiglas. The high voltage electrodes were constructed of brass dipped in a plastic insulation (Plasti-Dip, PDI, Inc., St. Paul, MN) for safety. Brass bolts were soldered to the electrodes and used both for mounting to the Plexiglas and for electrical connection. The circular charging electrode was also constructed of brass and was mounted with a Teflon arm to a brass bracket attached to the rear of the Plexiglas plate. Electrical connection to the charging electrode was made by directly soldering a wire to the electrode. The 71 .aoxusaa uswsowuweom ovosaoo~o uswussao "cu .v.w :m\~ x .v.o :v\~ x :¢\~ ocosaoogo uswussao usuasvswmho "m .sowaoossoo nssouu canolwn "m .cauoluc sou scuuoossoo mesmhuoo~u up .nasolwn "m .husunwaso wsmsowuwaoa sch ussol uncluo "m .sowasasasn "v .alsuo zouom "a .aousua coauOOuuov owsa~o> dual ”N .esuuuxoam :¢\~ ”a .Avoussaso eelwu N.H >~ouslaxoaass szocsv hacienda unwassol onosaoodo can sane-«a .mu shaman m _ / u. &\ m n _ tri o\ _._ v.0 i MJ ..\\ \ e. ,1 ._ . iiiii vxiln Ln— //\ 72 bracket holding the charging electrode can be adjusted to give vertical, horizontal and side-to-side movement. The electrode itself can also be rotated by turning the Teflon connecting arm. The bimorph mount was constructed to provide rotation in the horizontal and vertical planes while maintaining the tip of the capillary stationary. This rotation ability can correct for errors in mounting the capillary or errors caused by uneven heating during fire-polishing of the capillary tip. Thus, even if the jet is not in perfect alignment with the capillary axis, it may still be usable. Electrical connection to the bimorph is made by clamping it between the rotating mount and the fastening bracket. The mount is held at ground potential, while the fastening bracket is connected to the bimorph driver. The present design was adapted from a mount proposed by Seymour and Boss (117), but is considerably smaller and has more movement of the charging electrode. 5. IDG Internal Electronic Circgitry Figure 19 shows a block diagram of the system as it is configured for stand-alone production and charging of droplets. A more versatile computer-controlled configuration has also been implemented and is discussed in detail below. 73 .uoussosow uoumouc coasnomu ens mo seasho> osousivssam one we lsnwsfiu :oofin .mm shaman sue you”; 6233.... 2. x030 z. cues...» 02.6 3...... Ca. $25.53 3...... .53 o my 000 pwoo pd cud + ... + rooms tooEo good i oz.m...... Ti/t oz>m rooms moomhouaw hmqmomo 4. ”.23.. oh 33.5 2.: F “.2525 a Juuxilaaxh WI can a .m... .22 . J A 52.9. 50> oon ”3:3 JGSEViI/W comm... 1.2.02.8 3:83 i m .>_o < < 59>... /.. (soon 2 .22 38> o5. ililliw oliliili 3 .— tooxa Kuhn! mm... o... 20:.0m;mm Ail 52.5 o n .3326 ....onm < mom.>_o 0.— hmnfin: fi .05... Hoar—4LI< I :LKO" H ”w Dona... < hao oz>m MHIGZ m3haun canola: .9." one»: 9.9% >oP+ o.o_ oxo. i , one no.o2m ooonzm oo. w a: >m.i oxo. oo. ~o.m2u eoonzw , at. >n—i. 78 sufficient current is available to meet the requirements of the piezoelectric element. An approximate current drain can be calculated from the bimorph capacitance. The capacitance of the bimorph was measured in two ways. The passive RC circuit shown in Figure 21 was driven at 10 kHz by a Wavetek function generator with the bimorph in place of the capacitor. The resulting bimorph voltage waveform was noted, and various values of known capacitance were then substituted until an approximate waveform match was obtained. The estimated capacitance by this method was found to be 1500 to 2500 pF. The second method involved connecting the 74121 monostable multivibrator as shown in Figure 21 and observing the pulse width obtained when the bimorph is connected in the place of Cezt. The product literature for the 74121 estimates the pulse width from the equation: Toulse = Cext Rext 1n 2 The measured pulse width of 0.0148 ms was substituted into this expression to solve for the capacitance of the bimorph. The value obtained by this method was 2170 pF. Using this value one can estimate the current required by the relation: i = C dV/dt where i is the current and dV/dt is the fastest voltage change, which will occur at the edge of the square wave. 79 FUNCTION é:———i GEN. -—WNVfi--dr-——:L Vcc ‘ £ +5V +5v o—rAi 7 4. A2 1 R/C A .6 2 l. 1 Con/Bimorph 1 kHz TTL O—-iB T 9.84 kg) Scope o—‘Q Figure 21. Circuits used to measure bimorph capacitance. 80 Estimating a 30 V change to occur in approximately 0.001 ms, the maximum current required would be 65 mA. Since this is a rough estimate, a 300 mA power supply was installed; this should provide a good margin for error in the estimate. Although the power supply should be capable of supplying current for operation up to 100 or 150 kHz, in practice the heat sinks installed on the transistors are only capable of dissipating the heat generated from a maximum operating frequency of about 80 kHz. For most experimental applications, however, droplet production frequencies of less than 50 kHz are sufficient. Figure 22 shows the shape of the waveform produced with the bimorph installed. Figure 23 shows the relationship of the peak-to-peak and flat-to-flat voltages for various settings of the front panel bimorph driver amplitude adjustment. Droplet Pulsing Circuit Figure 24 contains a schematic diagram of the circuit which is used to control the voltage present at the electrode used for droplet charging. A TTL input is applied to the base of control transistor SK3044. When the input is in the high state (+5 V) the control transistor turns on the pass transistor SK3053, allowing current to pass through the 30 k0 resistor to ground. In this manner, the voltage present at the collector of the pass transistor is presented to the charging electrode. When the TTL input is low (0 V), 81 > o .maouossz assume uoeqav canola: .NN enemas "|IA'|'I qulouluoE l lllllllllll l+lllllllllllll llill xooaiouixooa 82 30-i- q 0 e 25. o d 5 3 - . O E‘ ‘ ° ° 0 O O . e e o T I .I T '1 F l' T V fil o 10 Diol Setting (A0.) Figure 23. Bimorph driver amplitude for various dial settings. (0) Peak-to-peak voltage. (e) Flat-to-flat voltage. .uasouao wsuwssao season: .GN shaman conn ._:Q|\<.I nmonxm 84 both transistors are off, and the charging electrode is held at ground potential through the 30 k9 resistor. This circuit provides a good compromise between low current consumption and fast rise and fall times of the high voltage pulse. The relationship between the trigger waveform and the charging waveform is shown in Figure 25. The current consumption for various frequencies is shown in Table 5. Table 5. Pulsing circuit current consumption. Frequency (Hz) Current consumption (mA) 10 3.83 100 3.90 1000 4.00 4000 4.35 10000 ' 5.06 14000 5.60 constantly on 7.62 constantly off 0.00 Strobe Synchronization Circuit The Strobotac strobe lamp contains an input for firing the lamp externally. The recommended firing method is to use either a 6 V peak-to-peak signal or to simply short the input to ground. The input has approximately 32 V DC present due to the internal voltage divider network of the firing circuit. If a 6 V peak-to-peak trigger is used, the RPM knob must be manually adjusted at each setting to provide adequate bias for firing. The shorting method is 85 ..a o.u~ “a ..a N “a ..a on no ..a o no ..a a. .a .e: we .4 .usauso uasouuo uswussco Hodges: .mN ousuwh u. 1.1.] I: "(3.. a) .hnnv .>.I >, ON.F+. <.iiiiiiw NIx mép 4:. mmooz: 86 therefore superior, since adequate voltage is always present at the trigger input to induce firing, regardless of the range setting. The circuit shown in Figure 26 was constructed to fire the strobe by shorting the input to ground. Transistor 2N3053A acts as a solid state switch which is closed by an active-high TTL pulse. Because the trigger frequencies are often considerably higher than the maximum firing frequency of the Strobotac (500 Hz), provision has been included to divide the trigger frequency by factors of 10, 100, or 1000. Three 7490 decade counters are cascaded as shown, and a front-panel switch can select between the various dividers. The input to the decade counters is connected to the droplet pulsing circuit input. Bimorph Frequency Selection Circuit One disadvantage of previous designs that used digitally synthesized frequency generation was that the frequency was not continuously adjustable. When initially setting droplet production parameters a method of continuously adjusting the frequency is desirable, since the operator must be constantly observing the droplet stream for stability. The circuits shown in Figures 27 and 28 were constructed to provide this feature while also allowing the user to lock in the correct frequency once it was determined. When locked, the output is as stable as the 87 TO STROBE TRIGGER - 2N3053A V 470k!) F/10 F/100 F/IOOO CHARGING o—Lc—77 oo '2 1 7 '2 1 7 ‘2 SYNC 47>?) 14 4 i4 4 14 9 9 g 0 11 0 11 11 03 O 2 MR1 2 2 o—Z-iMSZ 7 7 <7 <7 <7 Figure 26. Strobe lamp synchronization circuit. crystal oscillator input. Because of the complexity of this circuit, the two parts are presented separately. Figure 27 contains those components primarily concerned with divisor selection, while Figure 28 contains the counters used for frequency division. Selection of a 12-bit frequency divisor is performed with an AD574 ADC. A voltage divider composed of a fixed resistor R, a 5 kn trimpot and a 1 kn front-panel potentiometer is used to vary the input voltage to the ADC. The value of the fixed resistor is adjusted to set the maximum upper frequency allowable by biasing the lower end of the divider away from ground. The 5 kn trimpot is adjusted to set the lower frequency limit and thus determine the total range. The 1 MHz crystal oscillator provides both the fundamental frequency to be input to the 74191 counters and a trigger waveform used to initiate conversions on the ADC and subsequently load the 74174 latches. A complete conversion/frequency selection cycle involves conversion of the input voltage, latching of the digital output, and loading of the output by the counters. The 1 MHz oscillator frequency is divided by a factor of ten by the 7490 decade counter. This signal is used to trigger a 74121 monostable multivibrator which provides a 38 us active-low pulse to initiate a conversion on the AD574. The monostable also provides an active-high pulse to latch the digital output of 89 .uwsoauo soauoofies aososvoum canola: .bN ohsmuh as h.N at No. >o+ nehzo SI 2 o. 2. o IE“ I snub ~l 340333 8 . mace o 083 .2 .n o « FNF¢N .vhfvh .thVB o . o mmwm o m "mu. .8 3 to no «oz. 2. ..o 3.5 no 332. . non H_ w . €122o+n _: _ o —hu —oa—u«—e nu «a puoua—ioph— c— w}. to So 8 no 8:. 33 33 .o om. :12. nopmwoo...o omen tnmo< So ..5 2 a. 5.: 2:. E: «a: omo >3 an m}. mm S. 9. on 5 N1, P ”p 0 N a .62. 3+ 9. o 3.7 .uusonao sooa>on hososveum canola: a. or ...o op .mN summon >n+ >o+ e. o. o o o om 0 ..IV. .2. PM soon .m.¢~ oo.¢o 2.30 K m o in :8 .im 3 o o _n F o v -4 - >o+ _ Q _ T. I c ,1 Wm H mm ..o elm elm ulo .8 mm mm Ho. .8 Pmpvh me¢h Pmp¢h no No 3 oo no «o .o oo no «o .o oo o o. . 3 mm: mm. WMIOH<4 OH — 91 the previous conversion. The latches are connected to both the counter data inputs and to a series of LEDs on the front panel through the 4050 display drivers. The LEDs can be used to determine the current frequency divisor. The frequency is locked in by breaking the connection between the latch pulse monostable and the latches via a front panel toggle switch. The purpose of the 7474 flip-flop is to provide an initial connection between the latch pulse monostable and the latches should the unit be powered up with the toggle switch in the frequency lock position. This is necessary so that the unit does not initialize with the latches unloaded, which would produce an initial frequency of 500 kHz, too high for the bimorph driver circuit to handle. On power up, the 01 input of the 7474 is at ground potential. The state of D1 is loaded into 01 on each clock edge at CPl. Eventually, D1 changes state as the capacitor charges, with the result that a single rising edge is sent to the latches with the proper timing relationship governed by the latch pulse monostable. The timing relationships are shown in Figure 29. Figure 28 shows the part of the circuit concerned with frequency division and waveform output. The input is selectable via a front panel toggle switch between the onboard 1 MHz oscillator and an external input. Three 74191 4-bit binary counters are cascaded by connecting the 92 .lsswsav msulua soaaoouos hososuonm canola: .mN shaman i— r a d fl — hll :3 3.7.3... i~ _ F _ — _ .Aiii «no oozioza _ _ b _ L _ _ .3530 see.— i=i b — = — =I b use .3530 3.2. Boo f \ 23> 2:3. Eco \ escapees; :3: /.|I|I euee ‘33 j /_i|._i|i Boo ..33 . E \AiIi—II te>coo temm save {233 team 93 active-low rippple count output to the clock pulse input of the next counter. The oscillator is connected to the clock pulse input of the first counter. The output of the last counter is inverted end then connected to both a 74109 flip-flop, which is used to make the output a square wave, and a 74121 monostable multivibrator, which generates a load pulse of approximately 70 us. When the most significant counter issues a ripple count, the three counters are simultaneously reloaded from the latches to begin the next count cycle. Two outputs are available from the 74109. The non-inverted output is connected to the bimorph driver input. The inverted output is made available at the front panel for synchronization of external devices. Pulsing Frequency Selection Circuit The selection of the frequency with which droplets are pulsed out of the main stream is made via three thumbwheel switches mounted on the front panel of the instrument. Because this frequency does not need to be continuously adjustable and the output does not have to be converted to a 50X duty cycle, this circuit, shown in Figure 30, is considerably less complex than the bimorph frequency selection circuit. Three 74190 4-bit BOD counters are cascaded by connection of the ripple counter output of one counter to the clock pulse input of the next. The input to the first 94 .aasouwo uovu>uv hoaosvouu maflmusao .on seamen ax «a ma co. $7.5? - rr fi._ o. u« o PN—¢h ¢o¢h m m «R W So 2. So 2. p m e. n a, n, o. _. H _ d — Jo .Eo .MADa O... _ 4 0023 p , muo::a :2.ch Rwy kwv 1H». P. ‘4 4. mm mm mm no mm ...m N. no Wm H... We. no oat: capes cage 3 3 S on no «a S on no «a E on a o. . me mm: mm... 323:5 .3238?» com ES... 95 counter is selectable from a front panel toggle between an external input or the output of the bimorph frequency selection circuit. The counters are loaded from the inverted outputs of the BCD thumbwheel switch when the inverted ripple count of the most significant counter triggers the 74121 monostable multivibrator. The latter issues a load pulse of approximately 1 us. The output of the 'most significant counter is also connected to a 7404 inverter, which converts the active-low ripple count to a high pulse which is connected to the droplet charging circuit input. The non-inverted waveform is available at the front panel. The inverted waveform (internal connection) is used to separate charged droplets from the stream. Uncharged droplets can be pulsed out by externally connecting the non-inverted waveform to the external pulsing circuit trigger input. The width of the either pulse is determined by the period of the input clock frequency. Adjustable 300 Volt Power Supply To supply the voltage for charging droplets, the 300 V adjustable power supply shown in Figure 31 was constructed. This is a modified version of a design published by Russo et al. (104). A capacitively filtered full-wave rectifier supplies 300 V to pass transistor 800238 whose resistance controls the voltage at the output. The resistance of the pass transistor is varied by the TLOBl amplifier and control 96 .humass hoses amends» dawn smasassflv< coon a; . IIIA§W1II22(I>@.: .Hm ohsumh wit 9.2 exam \ \C «‘00— p001.._. \ \ D} .7 .www oa— cxo. 4 .. o..~.~ / fin V\, Nooez. H] «03.2. 3: f Algl axon / >on+ 3N. uao+ .— [14— uaF.o noo+zp Ar L SaSoH :83. com BNp axon can in coon 30 500+:— omNIOIonN 05/0 — ¢ 97 transistor 806198 through the front panel 500 a potentiometer. In operation, the voltage presented to the non-inverting input of the amplifier is varied by the front-panel potentiometer. The amplifier corrects for the voltage difference at its inputs through the feedback loop, and the resulting output current varies the resistance of the control transistor. The base current to the pass transistor (and thereby the output voltage) is regulated by the resistance of the control transistor. The particular output voltage selected is displayed on an analog panel meter. Several modifications were made to the original design. A 250 a resistor was added between the rectifying diodes and the filter capacitor to drop the rectifier output voltage slightly, reduce clipping of the AC voltage and give the diodes some resistance to work against. A 1N4007 diode was added at the pass transistor 806238 to protect it against excessive emitter to base voltage. This transistor is rated for 5 V but it is feasible that a larger voltage could build up if the adjustment potentiometer is turned down too rapidly, or the transformer is turned off. A 40 uF, 450 V capacitor was substituted for a 20 uF, 250 V capacitor on the output for better stability and the higher voltage rating. The 30 k9, 12 fl load resistor was added for discharging of the output capacitor. 98 5 Volt Fixed Power Supply A fairly simple 5 V power supply was constructed as shown in Figure 32. A 12 V peak-to-peak transformer secondary was connected to an encapsulated bridge rectifier. The output was capacitively filtered and regulated. The ripple at e 33 mA load was measured to be < 2 mV or 0.042. With the digital frequency meter described below connected, the noise increased to about 15 mV, which was considered acceptable performance. Frequency Meter/10 MHz Frequency Source A counter/timer/frequency meter (CTFM) was constructed from an ICM7226A evaluation kit (Intersil, Cupertino, CA). The ICM7226 is a fully integrated universal counter and LED display driver. It combines a high frequency oscillator, a decade timebase counter, an 8 decade data counter and logic and driver circuitry to directly control a large 8 digit LED display. The ICM7226 can function as a frequency meter, period meter, frequency ratio counter, time interval counter, or a totalizing counter. It is supplied in the evaluation kit with a 10 MHz crystal oscillator, which is accessible through a buffered oscillator output on the chip. The CTFM is utilized in the IDG to perform various measurements on the generated waveforms. A front-panel rotary switch allows selection between the outputs of the bimorph or pulsing frequency selection circuits, or the 99 .>Anasa notch u~o>lm mo cuasloaom .Nn shaman gm ,3? H Hiooov >m+W « annZJ H #30 z; 0<> GNP N. oou (a: a> a. 100 inputs of the bimorph driver or droplet charging circuit. Inputs to the CTFH are buffered with a 7404 inverter. The most common use of the CTFM is for measurement of the droplet production frequency. The 10 MHz oscillator is also utilized as an external frequency source for the bimorph frequency selection input. The oscillator is buffered in the CTFM and can be connected directly to the external input for the bimorph frequency selection circuit. Care must be taken to ensure that the frequency selector is adjusted with the internal 1 MHz oscillator to a value less than 5 kHz so when the 10 MHz clock is connected the resulting frequency sent to the bimorph driver will be within the proper limits (i.e., less than 50 kHz). This higher clock frequency can be helpful in obtaining more resolution when high droplet production frequencies are used, because higher divisors can be used with the 10 MHz input. 6. Computer-controlled Droplet Generation The stand-alone version of the IDG is suitable for many applications, but suffers from two distinct drawbacks. First, only a single droplet can be pulsed out of the stream at a time. Second, no provision is made for adjusting the phase between the waveforms used for droplet charging and droplet production. The phase can be adjusted indirectly by decreasing the amplitude of the bimorph driver, which causes 101 the droplets to break off at a different point in time, but this has the unfortunate side effect of decreasing the stability of the droplet stream. Several additional features were desired in the droplet generator as well, especially for use as an HPLC interface. Frequently in HPLC the nature of the mobile phase varies with time. Such variation requires continuous readjustment of the droplet parameters to correct for variations in viscosity and surface tension. A means of making these adjustments automatically would be very advantageous. These problems were solved by placing the IDG under computer control as described below. Figure 33 contains a block diagram of the IDG configured for computer control of droplet generation. The bimorph driver, pulsing circuit, strobe synchronization circuit and power supplies are identical to those described previously. However, the controlling waveforms for these circuits are now produced externally by the AMD 9513 System Timing Controller (STC) described previously. The reader is referred to part A, section 2 of this chapter for a description of the STC and microcomputer components. The STC is configured as shown in Figure 34. Three counters were used to generate the waveforms for controlling the IDG. Counter 1 produces the waveform for droplet production and is configured to count down repetitively. A value of one-half the desired bimorph period is loaded into 102 coasaosa one no soaauoe vofinosusoOIansnlou 0:» he .m... .2: umomhm ...m:..a< m0(h._o> h40> con Oh H a b.3050 oz>m “momhm fl :aoza _ 3.3:“. Ne 1 A _ 02.2: zwhm>m .uOussosom asunouv lsuusmv gooun .nm assuah AL awhamzooomgz omm.mo mOh<44-Umo a Iamvam NI: p . . . ._. _ 2% 333...: m a 2: :2.an u0 IQ...— - (F. Don ._.< mh40> mp“.- 55.....0025 3). ..."... © uo... Vega .....So 592.. Icy—02.0 mac} m204<4uo ouxzq. <55 roads}. Le _ . .1 ‘ xm

u mill; onozmumz» H A 5oz... .2“... xEEVS converts a number on the stack to the key codes that represent that number. For example, 1.55 would be converted to the #1 #. #5 #5 series of key codes. OET# accepts up to a 6 digit number from the terminal and converts the number to key codes. The most useful word is ENTER# which functions like ENTER except that it first converts the top number on the stack to key codes. Another special case is the ”Edit Node/Run Node” key. Depression of this key toggles the instrument from one mode to the other. Because it is often desirable to switch to a particular mode regardless of what the previous state is, the words EDITMODE and RUNMODE were written for this purpose. This is accomplished by depressing the scroll button, which is used for viewing an SP8700 entry line and thus is not acknowledged in the run mode. If an acknowledge signal is received, the instrument must be in the edit mode and the appropriate action can be taken. Two SP8700 status words have also been defined. FLOW? delays until the "Flow Ready” light is on. CAN waits until the ”Cam Marker” light flashes. CAM is useful for synchronizing with the clock in the SP8700 and for coordinating flow rate changes. 117 2. Valve Control Two words control valve actuation and sensing. CCW turns the valve whose number is on the stack in the counterclockwise direction. If the valve is already in the counterclockwise position, this word has no effect. Otherwise, the valve position is sensed after a short interval to determine if the valve change was successful. If the valve does not switch within the appropriate interval, all foreground routines are aborted and control is returned to the terminal handler. C0 is the equivalent word for turning a valve in the clockwise direction. Valve numbers 1 or 2 are appropriate arguments. 3. Data Acquisition and Storage This routine operates as an independent background task. It is initiated by the word TAEEDATA. Acquisition can be temporarily halted by the word SUSPEND and resumed by the word MOREDATA. A data taking run is ended with the word ENDDATA. The data taking rate is controlled by the variable ’REST, which contains the number of hundreths of seconds between points. The delay is performed by the AH9513 STC, and is thus independent of software activity. Data are stored as time-data point pairs, where the time is the number of tenths of seconds since injection and the data point is a 12-bit number stored in a 16-bit cell. These 118 words can all be accessed from the event controller so changes can be performed while a run is in progress. The data are stored on the 11/23 in the FORTH pseudo-disk file DFLTDAT.FTH. The size of this file determines the number of data points which can be taken. The current default is 50 blocks which corresponds to 12,800 data points. The default sampling delay is 0.12 seconds. The easiest way for the user to set the appropriate delay is to use the word LENGTH, which prompts for the expected length of the chromatogram in minutes. The delay is then set to the rate which will fill approximately 802 of the available file space. A single data-taking run proceeds as follows. The I/O buffers are flushed, and the pseudo-disk file name is changed from the program file to the data file. Date taking is initiated at the rate specified, and the data are stored in the I/O buffer. When 256 points are taken (the capacity of the buffer) data storage is switched to the second buffer,’ and the data acquisition task starts the flush task to send the full buffer to the data file. In the event that the second buffer becomes full before the first buffer is finished being flushed, data taking is temporarily halted. This has never happened, but is feasible if the 11/23 is exceptionally busy. If data taking is SUSPENDed, the acquisition task is halted, but the state of the task is preserved. Nhen 119 acquisition is re-started with MOREDATA, a flag is stored in the data buffer consisting of a -1 followed by the current value of ’REST. The flag is not used directly, but is simply a comment noting the change in sampling interval. When acquisition is terminated with ENDDATA, both buffers are flushed, the acquisition task is halted, and the user is asked for a file name. The number of blocks acquired is then transferred to the new file, and the disk file specification is changed back to the program file (WICRO.FTH). 4. Peak Finding Routine This routine is an independent background task whose function is to detect a chromatographic peak and make an attempt to measure the peak height in real time. Three parameters control peak sensing: the sampling interval ’TIME (100ths of second units), baseline threshold lTHRESH (in ADC units) and peak maximum threshold 2THRESH (also in ADC units). These values may be changed directly by altering the above variables, through the prompting words 2THRESH and SETTINE, or through the event controller. The start of a peak is sensed by comparing the difference between the first two data points to lTHRESH. If the difference is less, another data point is taken, and the second and third is compared and so on. If the difference is greater than 1THRESH, the start of a peak is assumed, and 120 the routine begins watching for the peak maxi-um. The maximum is determined by comparing the difference of two subsequent data points to 2THRESH. If the difference is greater than 2THRESH, and in the downward direction, then the first data point is assumed to be the peak maximum. A report flag is then set for the event controller to process. The report contains the tins of peak start, time of peak maximum, peak height, and baseline value. The peakfinder also maintains a counter, #PEAES, which is updated each time a peak is recognized. This count can be used as a basis for event control. For example, a valve could be switched after the 5th peak has been recognized. Since sometimes a sample contains more or fewer peaks than expected, the counter can also be incremented or decremented or zeroed through the event controller or by the respective words I#PEAKS, D#PEAES, and Z#PEAES. In this manner, even if a sample is not well-behaved, a run can be salvaged by manual correction. The peak recognition algorithm is not perfect in that as the sampling interval is decreased, the baseline value is determined to be further up the beginning of the peak. This can be reduced to some extent by lengthening the sampling interval, but this occurs at the expense of properly sensing the peak maximum. The default values of 20, 40, and 50 for lTHRESH, 2THRESH, and ’TINE, respectively, do a fairly good job for well behaved chromatograms with fairly sharp peaks. 121 The peak maximum and time of peak maximum, which are of the most value to the chromatographer, are reported quite accurately. The baseline value should be checked manually by the word ?SVS, which issues a system status report including the current ADC value. 5. Event Control This classification is probably the most versatile group of words for creative MHPLC control. Events such as valve switching, solvent change, or flow change can be based on time, HPLC status, detector output or any combination thereof. Time-based Event Control Time-based commands include EVENT, the event controller which waits until the specified time after injection, and WAIT, which is an absolute delay. WAIT takes an argument off the stack (in hundreths of seconds) and simply delays for that time interval. EVENT takes two arguments off the stack, minutes and seconds, and delays until the real-time clock exceeds that value. Normally this is the time from the injection, where the clock is automatically started by sensing the position of the injection valve. This is accomplished with the word INJECT. SYNC is a similar word, except it prompts for injection at the appropriate time for synchronizing the microcomputer clock with the clock in the SP8700. Both clocks are then started. The microcomputer 122 clock can also be started without injection by use of the word CLOCK. EVENT is a very powerful word which allows access to a considerable amount of the operating system during a run. It is also responsible for processing the peakfinder report and displaying the present time and time of the next event. EVENT uses single-letter commands to pass control to a word through a process called vectored execution. Specifically, each letter is used as an offset into a table of addresses. The address is then used by FORTH to determine where the word resides in the dictionary. In use, a key is struck once, to get the event controller’s attention, and then the key representing the word desired is struck. A list of the keys and their corresponding words is contained in Table 7. Detector-based Event Control Event sequencing based on detector output is perhaps the most powerful method of control. Timing is very crucial in column-switching applications, and often HPLC retention times are not very reproducible from run to run. When limited to simple time-based sequencing, frequent timing adjustments are often needed. If detector output is used as a basis for decision-making, programming can be simplified and erroneous runs minimized. Several commands are available for detector-based sequencing. The peak count, as mentioned above, can be used 123 Table 7. Event Controller Commands i delay until the signal is greater or less than the specified percent are very useful for heartcutting. t signal changes by the specified percentage 0 these commands is SFSCHG. regardless )A/D, (AID, of full scale recorder deflection. and A/DCHG are similar, Key Word Function A ABORTRUN Abort run, return control to term. handler B TAEEDATA Begin taking data. C OLE Clear screen. D SETTIME Set delay for peak finder. E ENDDATA End a data taking run. F D#PEAES False peak, decrement peak count. H EHELP Lists event controller commands. I I#PEAES Increment peak count. E Z#PEAES Eill (zero) peak count. L LMARK Sets length of minute marker for recorder. N NOREDATA Restarts data taking after a SUSPEND. N NEWREST New data taking interval. 0 lSTOP Turns off recorder driver task. P ?PEAE Turns on peak finder task. 0 4STOP Turns off peak finder task. R DIGREC Turns on recorder driver task. S ?SYS Displays system status. T 2THRESH Sets peak finder thresholds. U SUSPEND Suspend data taking temporarily. V VIEW List a block on the terminal. X TOGGLE Stop/Restart the microcomputer clock. 2 ZERO Sets recorder zero for recorder driver task. n conditional execution. The command >XFS or (XFS will The relative of full the direction of the change. except these take units as arguments. 124 These commands counterpart This command delays until the scale, The commands it To assist in programming with detector-based event control, ten registers have been established which can be used to save and recall the ADC values corresponding to detector output. The word SAVE takes a number from 0 to 9 "off the stack and saves the current detector output in the specified register. RECALL takes the register number off the stack and replaces it with the previously saved detector value. Most applications use combinations of event control commands, linked in either a logical AND mode or a logical OR mode. FORTH allows easy implementation of either combination. For example, one may want to wait until six minutes after injection, wait for two peaks, wait for baseline, then switch a valve after a 102 change in the signal, which might be the rising edge of the next peak. With this type of strategy, very powerful and inteligent event sequencing can be accomplished. 6. Recorder Control The strip chart recorder is used both as a method of obtaining real-time output and as a means of re-plotting a previously stored data file. This method of graphics output produces lower quality plots than can be obtained by using a graphics printer, but it is considerably easier and faster. The chart motor drive can be started or stopped with the 125 words ON and OFF. The chart servo drive is controlled as described below. Digital Recorder Driver The digital recorder driver is an independent background task whose purpose is to convert the ADC output to a form compatible with the DAC output connected to the chart recorder input. The task is activated with the word DIGREC, or can be started through the event controller. The output scale is controlled by the two variables OFSET and RANGE, which are conveniently set using the word ZERO. ZERO prompts for the percent of the recorder scale which should be designated zero volts from the detector. The ADC value is then transformed to DAC units according to the equations: OFSET 2048 - (2048 x 3 zero input) / 100 RANGE - 4096 - OFSET DACunits = ((ADCunits - OFSET) x 256) / RANGE If the result is less than zero or greater than 256, the DAC is set to the respective extremes. This method of conversion always retains the complete positive range of the ADC, but allows the user to select what portion of the negative range is displayed on the recorder. The recorder driver also places a tick-mark on the output every time the real-time clock turns over to the next minute. This allows easy estimation of retention times. This option can be disabled by setting the length of the 126 minute marker to zero using the word LMARH. The default length is 10 DAC units. The word EVENT makes a similar mark upon initiation. In normal use, this allows the user to see where the injection point is and where each event occurs if the event is immediately followed by another EVENT cycle. The length of this mark can also be controlled by the user by changing the variable ENARE. Regraphing Utility The regraphing utility allows one to output a stored data file to the recorder with different scale factors. The output occurs at ten times the speed with which the data were taken, so this is a quick and easy method of displaying a data file. The user sets the chart recorder speed to a factor which is ten times faster than the original to obtain a scale factor of one in the x direction. The scale factor in the y direction is determined by answering the prompts for the minimum and maximum ADC values to graph. The complete range of the ADC is displayed by using values of 0 and 4095, respectively. Higher maximum values will compress the x output. A higher minimum and lower maximum cause the x output to be expanded. The regraphing utility is initiated by the word REGRAPH. A file name is prompted for and the file is copied into the default data file DFLTDAT.FTH. The number of 127 blocks transferred is displayed along with any error messages. If the file is not present, or any other file error occurs, the utility exits, and control is returned to the terminal handler. 7. Other Useful MHPLC Words Several other words have been defined for checking the status of the system. ?SVS was designed to display the status of the background tasks and their controlling variables. It also reports valve positions, SP8700 flow ready and cam signals, and current ADC and DAC outputs. ODAT is the fundamental word used for initiating a single data conversion and placing it on the stack. RUN will continuously display the ADC output until a key is hit. TFI and (TF1) return the time from injection, or since the clock was last started, in units of seconds and tenths of seconds, respectively. RESET can be used to reprogram the AN9513, and SETUP can be used to reprogram the 8255 PIO for use in the MHPLC system. Both of these words are executed when the software is first loaded, however, and do not necessarily need to be used again. C. Computer-controlled IDG Opegating Routines The software for operation of the IDG in the computer-controlled mode is mostly concerned with programming the AN9513 system timing controller. Figure 38 128 .uaenosouu aenqoaasoo soasaeseu ue—aoav veusuoe— Lt h.xu O .OMKu hmaac( htzm um:.<0mz ....tzm was...“ u>_.:moa ..(>¢u._.z. 9223230 um¢whz_ 62—08(10 um maupzaoo 32 .37.... pausao sum #4 mmmhzaoo 0(04 4553:: u0¢¢u...2. 02515.5 ub(3..—mw .mm essmuh g c<...<0m um.o tacos; uhzazou .Oucm hmumOJO uh: 200 .Oumu taco}; .(«kauz Home ‘ 129 illustrates the use of the word IDG, which is used to operate the droplet generator in the repetitious droplet production mode. The initial arguments placed on the stack are the number of drops to be charged, the number of drops uncharged and the frequency in kHz to two decimal places. The program first computes the closest frequency available from the 1 MHz oscillator in the AM9513. An upper frequency limit of 50 kHz is set in software. If the desired frequency is less than this upper limit, the appropriate frequency divider is calculated to be placed in counter 1. This frequency divider is 1/2 the period of the desired square wave. The phase scalar is then estimated by simply doubling the value used for the bimorph frequency divider. The charging and neutral widths are calculated according to the equations: CW NW (2N - 1) x t/2 (2M + l) x t/2 where t is the period of the bimorph frequency, N is the number of droplets to be charged, and M is the number of droplets left uncharged. After estimation of the proper frequency dividers and pulse widths, the counters are loaded, and the output bits on the AM9513 are set to the correct state. The counters are then armed, and the program goes into an interactive 130 adjustment loop. In this loop, the unshifted + (ASCII 61.) and - (45.) keys are used to adjust the charging and neutral widths. This is done by adding (subtracting) one microsecond from the charging width and adding it to the neutral width, then restarting the counters. Thus, this action does not affect the phasing or frequency of the charging waveform, only the duty cycle. The unshifted < (ASCII 44.) and > (46.) keys are used to adjust the phasing between the charging and bimorph waveforms. This is done by adjusting the phase scalar by one microsecond and restarting the counters. The H (ASCII 72.) and L (76.) keys are used to adjust the frequency of droplet production. This is accomplished by adding or subtracting one microsecond from the bimorph divider, re-estimating the required pulse widths and phase scalar, and then restarting the counters. The S key (ASCII 83.) exits the IDG adjustment task and returns control to the terminal handler. The IDG will keep producing droplets as defined by the last adjustment command. A second word which is similar to the above is RANPIDG, which is useful for locating the proper droplet production frequency. This word uses IDG to set an initial upper frequency and then ramps down by adding one microsecond to the bimorph divider at a rate of 1 Hz. The operator can watch the droplet stream until a good production frequency is found, and then a key is pressed. The routine then goes 131 into the frequency adjustment loop as described above. The ramp is continued by pressing the S key. This routine does not automatically adjust for the pulse widths, however, so the pulse voltage should be turned off for the best results. Operation of the IDG in the single-shot mode is accomplished with the word SINGLE-SHOT. This word takes as arguments a delay in hundreths of seconds. It is meant to be used immediately following IDG. Droplet production and pulsing parameters are set in the normal manner with IDG, and then IDG is exited. When SINGLE-SHOT is invoked, the specified delay is performed in software, and the counters for droplet production are started with the values for the bimorph frequency, phase scalar, and pulse widths that were set with IDG. In this case, however, counter three is reprogrammed to count only once instead of repetitively; thus, only a single packet of droplets are produced. D. Support Software and Data Conversion Several support programs have been written to run on the ll/23 minicomputer and are briefly mentioned below. 1. FORTH Support Proggggg As mentioned previously, the version of FORTH implemented herein does not use floppy disks for data and program storage, but rather, uses the LSI 11/23 as a pseudo-disk. The program which emulates the FORTH disk is called FORTHPIP and was written by Phil Hoffman (120). It 132 serves to transfer data between FORTH devices, standard FILES-ll format RSX files, RSX FORTH emulator pseudo-disk files, and serial data transmission lines. In this case, FORTHPIP functions in three capacities. First, it sends and receives data between the microcomputer serial line and the emulator file. This file is structured in 1024-byte segments called blocks. This is also the size of an I/O buffer on the microcomputer. Second, it generates hard copy of the programs contained in the emulator file by transferring data to the printer. The third function is to act as a translation program to convert data stored in the FORTH emulator file format to a FILES-ll format that can be accessed by FORTRAN programs and word processors. Use of FORTHPIP to generate hard copy is rather cumbersome, since a rather long command line is necessary, and the position of the block on a page must be previously known. A short command file called FTHPRT.CND was written to make updating of FORTH listings easier. This file accepts a block number as input, determines what two other blocks need to be listed on that page, and generates the appropriate command line to print this triad of blocks. Another useful utility was written to help locate the block in which a word was defined. This utility is a collection of programs run from the command file FTHINDEX.CMD. When run, the utility extracts words defined with :, CODE, CREATE, VARIABLE, and CONSTANT, along with their block number. The words are then sorted 133 alphanumerically and printed to provide a convenient index to a FORTH program file. 2. Data Conversion Software Graphics output or data reduction is performed by processing the data with the FORTRAN program CRUNCH and then using a graphics package such as MULPLT to obtain high—quality graphics output. The purpose of CRUNCH is to convert a data file stored in the unformatted FILES-ll compatible form produced by FORTHPIP to a formatted ASCII file containing records of the form: RD time,data with one record per line. This form was chosen to be compatible with MULPLT and many of the other programs written for data processing on the LSI 11/23 system. CRUNCH does not perform scaling, smoothing or any other manipulations which would alter the original data. These functions are left to the user. Typically data files are CRUNCHed only when necessary, as the formatted file takes up considerably more room than the unformatted file. Data are stored in the unformatted FILES-ll form produced at the time a data collection run is terminated. 134 V. EVALUATION AND APPLICATIONS This chapter is concerned with evaluating the performance of the previously described instrumentation. Several studies dealing with band broadening, baseline disturbances, backflushing, and detector-based valve switching are presented. Subsequently, applications pertaining to fuels separations are shown. These separations were performed in conjunction with M.R. Danna (122), and the reader is referred to her dissertation for additional information. Several miscellaneous applications not directly related to MHPLC are also presented. The isolated droplet generator is also evaluated and sufficient fundamental studies are shown to provide a good starting point for future work with this instrument. A. Multidimensional HPLC Some of the fundamental questions which arise when designing multidimensional HPLC instrumentation include the following: What are the effects of the valves and tubing on extracolumn band broadening? What types of baseline disturbances are caused by column switching and how can 135 these be minimized? How much imprecision can be expected when using totally time-based event control? These types of questions must be answered before applications can be implemented and interpreted properly. The applications which follow are mostly taken from the fuel work which has dominated much of the instrument’s use. Fuels are very complex samples, and therefore multidimensional chromatography is ideal for characterization and classification. This type of sample cannot be completely resolved, however, even with column switching techniques. Therefore, the samples shown do not produce chromatograms with the same degree of resolution as samples containing fewer components, such as those mentioned in the historical chapter of this dissertation. Nevertheless, it was felt that the most productive use of the instrument was to apply it immediately to the fuel analyses for which it was constructed. The chromatograms shown illustrate the use of the instruaent and software. 1. Extra—column Band Broadening in MHPLC In a multidimensional HPLC system a greater length of tubing is required than in standard HPLC because the outlet of each column must be returned to the valve to which the inlet is attached. One way of avoiding this problem would be to pack a column in a U-shaped tube. This is not as ludicrous as it sounds, since the shape of the column has no 136 effect on the separation, and this design would allow the inlet and outlet of the column to be closer to the switching valve. In the absence of such a column, however, one can expect to use a minimum of about 2" plus the length of the column in connecting tubing for each switched column. Figure 39 shows the setup used to make an extra-column band broadening comparison. In the first case, a mixture of toluene, anthracene and naphthalene were injected onto a 15 cm analytical silica column connected directly to a UV detector. The mixture was then injected again with two valves and lengths of tubing appropriate for connecting 25 cm columns placed between the silica column and the detector. The two valves were connected to the detector with a 1/16" low dead-volume union. The tubing used was 0.009” i.d. l/l6” stainless steel. The chart speed was 2.5” min-1. The resulting chromatograms are shown in Figure 40. The average results for 6 runs are shown in Table 8. An often-quoted measure of separation efficiency is the number of theoretical plates. In the absence of extra-column band broadening, the number of plates is fairly constant for different chromatographic bands. Because the number of plates calculated in Table 8 represents the separation efficiency of the entire system, including extra-column effects, the number of plates is not constant, but varies with retention time. Therefore, efficiencies for three reasonable (<10 min) retention times are shown. As 13? .so«»s:~s>s mausovsoaa vasa sou use: soaasaswumsoo mOhomhuo. .mm shaman 223400 ......n.)..-....- T , «Chomsz_ 02¢. ulna 138 .souass~s>o nausevsoan vase loam slssuouslosao o>musussessaem me..W m2u~2wm mm>.._<> 2.30153 >\I mu>._<> It; 139 Table 8. Peak Data for Extra-column Band Broadening Experiment. Peak Plates Half-width mm Retention Distance mm l-No Valves 5145 8.7 266 2-No Valves 5785 9.9 318 3-No Valves 6361 11.7 396 1-2 Valves 3885 10.7 284 2-2 Valves 4543 11.8 337 3-2 Valves #5345 13.4 416 2-3 Change -21.5 +19.2 3-8 Change ~16.0 +14.5 Measured Resolution Peaks 1-2:7 Peaks 2-3 Without Valves , 3.17 4.15 With 2 Valves ¥§.67 3.64 2 Change -l5.8 -l2.3 expected, the added tubing affects the narrower, early-eluting peaks more than the later eluting peaks when stated as a percentage decrease in efficiency. As an alternative measure of band broadening effects, the resolution decrease for each pair of peaks is also shown. As can be seen from these results, the relative effect of the band broadening depends upon the retention time of the peak in question. These results can be taken as the worst case using this size of tubing. In many applications, only a single valve is used, which would eliminate a lot of the extra tubing used in this study. 140 2. Baseline Disturbances in MHPLC There are two sources of baseline disturbance in MHPLC. The first is a fairly symmetric departure from the baseline that occurrs immediately upon switching a column into or out of line. An example is shown in Figure 41. This disturbance is a result of the finite response time of the solvent delivery system to a change in backpressure when attempting to keep the flow rate constant. The flow cell used in the UV detector consists of a quartz plate pressed against a teflon spacer with flow channels cut into it. Because the reference side of the cell is not in-line, it does not respond to flow rate changes in the same manner as the sample side. The UV detector, which is slightly sensitive to the flow rate of the solvent through the sample cell, then detects the re-equilibration period. This disturbance is usually quite small and has a positive deviation for switching a column on-line (increase in backpressure, temporary decrease in flow). Switching the same column out of line gives a negative deviation of similar size. The second type of deviation is only seen when a cut is made and developed with a different mobile phase composition. This type of deviation is a non-symmetrical, positive (higher absorbance) erratic disturbance due to the refractive index changes which occur as a result of mixing at the solvent boundaries. This disturbance can be 141 .owssao oasesssaaosa access on osv.eossnssas«v ssuueesa no adolsxu .uv oaswmh 142 minimized by beginning development of the cut with a solvent mix similar to that contained in the cut portion and then performing a fast gradient to the desired solvent. If this is not possible, the size of the cut should be kept to a minimum. Figure 42 shows an experiment in which a plug of 1002 THF was placed onto an ODS column and developed with 508 THF/HzO. The size of the plug was varied from 0 seconds (the valve was immediately switched back, but a small quantity of solvent was cut), to 30 seconds. The program which automatically made these cute is shown in Figure 43. A de-activated silica column was used as a backpressure source. Because this disturbance occurs at the void volume, it is seldom a major problem, but a blank should be run to distinguish this disturbance from actual peaks. 3. Effect of Imprecise Retention on Valve Switching To be analytically useful, an MHPLC system must be able to perform a reproducible heartcut. This is quite difficult with time-based event control due to retention time imprecision. To measure this imprecision, an attempt was made to cut the center of a single chromatographic peak consistently to a second column. The critical valve switch would thus occur on the rising edge of the peak. The reproducibility of the height at which the switch occurs provides a measure of the reproducibility of the heartcut. 143 10 sec 15 sec ‘ 20 sec H w ML Figure 42. Examples of baseline disturbances due to rOfractive index change caused by mixing of heartcut solvent ‘nd developing solvent. 144 ( BASELINE DISTURBANCE EXPERIMENT) : THF SC ; : H20 8A ; . SILICA 2 CW ; : ODS 2 CCW ; EDE DIGREC EDITMODE EDFI l ENTERO HZO 50 ENTERO TEF 50 ENTEE# FLOW 1 ENTERO EDFI 2 ENTER# TRF 100 ENTERS FLOW 1 ENTERO RUNMODE ODS 7 0 DO INIT 1 ENTER# FLOW? CLOCE 10 0 EVENT ODS CLOCK 10 0 EVENT SILICA INIT 2 ENTER# FLOW? CLOCE 10 0 EVENT ODS I 500 8 WAIT I . .” SEC CUT" CR SILICA LOOP STP PRESS ; 005 AND - DETECTOR INJECTOR SILICA Figure 43. Top- Program used to automatically generate the previous series of baseline disturbance chromatograms. Bottom- Valving configuration used for baseline disturbance experiment. 145 The alternative method of performing the same heartcut is to specify a time window (or count peaks) and switch the valve when the detector output rises above a particular value. This method was accosplished using the command sequence: INJECT 5 0 EVENT 20 XFSCHG 2 CCW The command sequence for time-based switching was: INJECT 5 48 EVENT 2 CCW As can be seen from a comparison of the two command sequences, one disadvantage of time-based seqencing is that the exact time of the valve switch must be known. In both cases, the goal was to switch the valve at 308 above baseline (approx. 75 mm height). The results are contained in Table 9. The valve configuration used is contained in Figure 44. Table 9. Heartcut precision comparison. Run number Time-based Cut Ht. (pg) Detector-based Ht. mm 1 135.1 95.6 2 110.7 94.8 3 128.9 94.9 4 113.4 93.1 5 118.1 95.4 Mean 121.2 94.8 Std. Dev. 10.41 .986 x Rel. Std. Dev. 8.63 1.08 146 SHJCA ]——— PUMP 005 AND —* 1‘— % INJECTOR DETECTOR Figure 44. Valving configuration used to test detector-based valve switching. 147 As can be seen from these results, detector-based valve switching offers substantial improvements in the precision of peak cutting. The accuracy of the cut is more difficult to ascertain, however, due to the amount of sample trapped in the section of tubing between the valve and the detector. The amount trapped in the valve itself is negligible. Fortunately, the amount trapped in the tubing can be quantitated from a single run and the cut adjusted by subtracting the difference between the actual cut height and the desired cut height. 4. Backflushing Appliggtiogg One of the first applications of the system was a single-valve backflushing experiment. In this experiment, a 15 cm analytical silica column was connected as shown in Figure 45. With the valve positioned as shown, solvent flows in the forward direction through the column and on through the valve to the detector. With the valve in the switched position, the solvent flows in the reverse direction through the column, through the bypass loop, and again on to the detector. A mixture of anthracene (1), nitrobenzene (2), and ethyl benzoate (3) was injected to give the first chromatogram shown in Figure 46. Without backflushing, peaks 1 and 2 elute in a reasonable time period, but peak 3 does not elute until nearly 44 minutes after injection. The 148 .souusaswumaoo Aesmnxosa s>~s>|o~msum .mv sasuun OE. mubamzoo ION—0.2 ka..<> II II 23052 22:32.. co a... zoo_ _;o 20:02.2. zuhmrm >mw>3mo ...2w>._om 149 . Backflush After WITHOUT Bockmsnnq Fm Connonen I 2 l 2.3 a; ‘ L ]\ {'c’ as '3 TIME—’ 80¢kadeth I 2 a; ( L§__L a g Figure 46. Examples of backflush chromatograms. 150 chromatogram in the lower left demonstrates how the situation is improved by incorporating a backflush after the second peak. The third peak now elutes at twice the backflush time, for a time savings of 17 minutes. This chromatogram also demonstrates one of the fundamental principles of the backflush. All components not irreversibly adsorbed to the column will elute at approximately twice the backflush time. In other words, it takes the peaks just as long to backflush off the column as it did for them to get into the column in the first place. This principle is often used in gasoline analysis to quantitate a group of components as a single peak, because all species will co-elute in a backflush. In the third chromatogram a backflush is performed after the first peak, and the second and third peaks are co-eluting. Not only is an additional time savings realized, but the chromatographer can be sure that everything that is going to elute from the column has, in fact, come off after the 9:44 peak. This third chromatogram also demonstrates another point concerning backflushing. Very little time savings is realized for the second peak. This is due to the fact that the second peak had already gone about halfway through the column when the backflush was performed. Once a peak has gone more than halfway, it is faster to continue the normal elution process for that peak. An additional time savings can still be realized, however, 151 if a synchronous flow increase is instituted along with the backflush as in the fourth chromatogram. With the help of the keypad emulator, a higher flow rate was employed to bring the total time down to only 7 minutes and 10 seconds. Although backflushing is not as widely applicable as gradient elution, it can often achieve similar results with much less expensive equipment. The other advantages include minimal baseline drift and no need for a column re-equilibration period as in gradient operation. One of the nicest advantages of backflushing is that all compounds not irreversibly adsorbed will elute at a definite, pre-determined time. Thus, there is no danger of a strongly-adsorbed component eluting in the middle of a subsequent chromatogram. 5. Q_g-va1ve Heartcut The simplest heartcutting system employed was one using a single valve, as shown in Figure 47. With the valve in the position shown, a cut is being made from column 1 onto column 2. With the valve in the switched position, column 2 is isolated from the system and column 1 can finish being developed in the normal manner. When all components have eluted from column 1, column 2 is switched back in-line and development of the previously stored cut is performed. Figure 48 shows a chromatogram obtained with this valving arrangement. The top chromatogram was developed on 152 .souasasuduaoo «senses: s>~s>ls~usum .Nv seawah 0... 3340528 momaom $5.5". IIIIU u>._<> II II 252:2 nomoi 82:32.. mo m... [Illalll 35.5 260— :6 20:73.. _ _ / K \mfl _\ u>..<> mohouhuo Tl TITIIII . 2:38 zozouaz. N 223300 Zuhm>m >¢w>3wo hzm>30m 153 SINGLE COLUMN tumow (HEXAflt/METHYLENI CHLORIDE) I GRADIENT HEARTCUT ELUTION I mm I 00/20 I I IOO/O Figure 48. Single valve heartcut of residual fuel sample, amino column to silica column. 154 a 25 c- amino bonded phase column with the gradient shown. While this separation is quite good for a sample of this nature, it was desired to know if the largest peak was a single component or several unresolved peaks. To obtain some additional resolution, the peak was cut off the amino column (column 1) and placed on a 15 cm silica column (column 2) for further separation. After performing a solvent changeover to 1002 hexane, the silica column was switched back on line for development. The resulting. chrmoatogram shows evidence of several components being present in the heartcut. The single-valve heartcut configuration has the disadvantage that the solvent used to develop the second column must flow through the first column. This configuration thus limits the choice of mobile phases to solvents which are compatible with both columns. Another disadvantage is that multiple heartcuts cannot be performed; the sample must be re-injected to perform a different heartcutting experiment. Fortunately, with a 2-va1ve system an autoinjector can be implemented along with the heartcut valve as shown in Figure 13. This can be used to perform an unattended multiple heartcut experiment. The only other way multiple heartcuts could be performed with a single valve is shown in Figure 49. In this configuration, the first column is placed on the valve and switched into and out of line. The second column is 155 .s>~s> saunas s wows: sowusasmumsou assessed usuusssusm mohouhuo Illllfi J L N 253300 .mv essmfim F 223300 mohowuz_ 02¢ m23m 156 always in line after the valve. A heartcut is performed by switching the first column in line, letting some effluent be swept onto the second column, switching the first column out of line, developing the second column, and repeating until all components have eluted from the first column. This configuration is limited to sequential heartcuts of the entire first column. 6. Two-valve Heartcut Two valves provide a much more flexible heartcutting arrangement. With the system configured as in Figure 50, either column can be isolated from the system. This allows the user to perform multiple heartcuts, but there isn’t the limitation of having to perform sequential cuts of the entire first column. Another very practical advantage is the ability to develop the second column with a solvent that is incompatible with the first column. An example of this which has arisen in fuel characterization is the heartcutting of a portion of the effluent from an Ultrastyragel size exclusion column onto a reversed-phase column (see Danna (122) chapter 5). The Ultrastyragel column cannot withstand exposure to water, which was used as part of the reversed-phase solvent. 157 .aowusasuwmsoo escapee: o>~s>lozs .on oasuwh o... 3335on season 55.28 U 33.; . II II 23052 tome... 22:32.. mo «2 J A A N _ 1 3 N L 3. we. no. Boo_ _;o 300_ _30 u>3<> mohouhu0 N 223300 20:03.2. — 223300 Iuhm>m >¢u>.3u0 h2u>30m 158 7. Multidimensional Backflush A two-valve configuration which is especially appropriate for the separation of mixtures of widely differing polarities is the multidimensional backflush configuration shown in Figure 51. In this configuration column 1 is hooked up in the backflush mode, while column 2 is plumbed in the column isolation mode. Those components which are easily eluted are separated in the forward direction on column 1. The more polar species are backflushed off of column 1 onto the second column for separation with a different stationary phase. Figure 52 shows a chromatogram which was developed with multidimensional backflushing. The top trace shows a four-component mixture eluted in the forward direction off of an amino bonded phase. The amino column can separate different ring numbers easily, as evidence by the resolution of the first three peaks, benzene, naphthalene, and anthracene. The fourth peak, however, contains a heteroatom which places its retention time close to that of anthracene. By backflushing the last two peaks onto a silica column and changing to 1002 hexane, a much better separation of these two components is achieved. 8. Sglectivity Programging with Backflush A variation of the multidimensional backflush which uses three columns is shown in Figure 53. This 159 .usuaesdnaosa usaoaesosuvuansl new use: soaussswausoo .un shaman 160 o... «$.32:on momaom «35..on HHV u>..<> I.. II zmoomtz :28... 0.2332“. «o «2 «fl 7 ._ 3. mo. ;Qo_ _3o zoo_ _;o zuem>m u>3<> «0.6qu , 5.32.3 20:05.2. h2m>30m N 223300 — 223300 U 8 U 1 I h S SINGLE COLUNN ELUYION (OT/3 NluNF/MITNVLINI CNLOIIOI) U 8 U 8 ‘ 3 «‘- § 3 3 i u ° u n 8 "‘ s u s 3 s 2 3 5 3 ‘ -.- '2' . < I r l vso/O ”‘71.“... 7° SECOND COLUMN IITN SOLVENT CNANOI Figure 52. Multidimensional backflush of model compounds, amino column to silica column. 161 .Assnuaosa Ava: wag-Issues: huusauosuse sou pee: sodassswuusoo .mo shaman 162 OE mmu3302h200 mom—30m mwhamaoo, u)3(> IIIIII 2u00¢=2 1020.2 0....(33m2n— 20 m; «H I ._ 3 . won 300- _30 300- _30 e _ R - ZUhm>m mohbuhuo u>3<> >¢u>.3u0 20.._.0u_.2_ hzw>30m p 223300 223300u¢l configuration uses a pro-column on the first valve in a backflush mode while the second valve is implemented in a column selection mode. The pre-column roughly separates strongly polar components from those easily eluted. The easily eluted compounds are allowed to pass to an analytical column for normal separation. The strongly retained components are backflushed off using a strong solvent and separated in a different mode on a different column. Figure 54 shows a chromatogram obtained with this valving configuration. The pre-column was packed with an amino bonded phase, as was column 1. Column 2 is packed with an ODS reverse-phase packing. The sample is a portion of a residual fuel which had been previously separated off-line by silica low-pressure column chromatography. The fraction containing heteroaromatics was collected and injected onto the amino precolumn. The first large envelope is eluted in the normal phase with a gradient from 992 hexane/THF to 608 hexane/THF. Then a step gradient is performed to 1008 THF with a couple of other components of medium polarity eluting. The remainder is backflushed onto the ODS column with a 308 acetonitrile/THF mobile phase and developed. Sharp peaks are obtained for the extremely polar moieties with this method, whereas they would have been extremely broad had they been developed solely in the normal phase environment. 163 2 a a J O 0 U. C O O .- 2 O AMINO COLUMN a 003 COLUMN 3 J lb 8 0 ‘ I 1 um # ”/1 (HEXAWF/THF) to 00/40 to was THF :o/7o ACITONITRILE/THF) Figure 54. Selectivity programming with backflush on heteroaromatics fraction of residual fuel. 164 9. AutoinjectorZFractiop,Collggtor Another interesting application of the system which has been implemented is the use of the system as an automated sample purification apparatus. By configuring the two valves as shown in Figure 55, a sample can be repeatedly injected, separated on a high resolution analytical column, and a particular fraction collected which contains the component of interest. The autoinjector was implemented successfully, but the software had to be modified slightly. The words which turn the valve, CW and CCW, contain a software delay loop to allow the valve time to turn before checking to see if the valve has reached its new position. The addition of the sample vessel to the air line actuating the valve has a capacitive effect which causes the valve to turn more slowly. Therefore, two new words were written for use with the autoinjector which have longer delay loops. These words are SCCW and SCW and are loaded at the user’s option from block 38. 10. Styragel Startup Routine An excellent application of the SP8700 communication software is a routine which has been implemented to bring the Ultrastyragel column up to a flow of 1.0 mL min"1 gradually, according to the manufacturer’s instructions. Because this column is packed with a gel, it is somewhat 165 .souuoenuoo souuosau use soauoenauoass sou vsasuumsoo sspnss 089 .00 ensign mom 0.0 mmu33och200 «Samson. U u>..<> I020; 0:.(23w2n_ a .... _30 mohouhua L: 300_ _so u0¢30m I ..II chomtz 20 . a; _ Til. 2n \ AI 3mmwm> 03a2m >mw>_3u0 h2u>30m 3 166 fragile and subject to bed compression when flow is increased too rapidly. The manufacturer recommends bringing the column up to flow in steps of 0.1 mL min'1 every 30 seconds. Unfortunately, this is very difficult to do on the SP8700 because the pump does not properly keep track of the flow rate unless at least one cam cycle is allowed per flow setting. At low flow rates, (e.g., 0.1 ml min‘l) one cam cycle takes several minutes. Therefore, a simple time-based flow increase is unsatisfactory. The styragel startup routine limits the flow rate increase at the low end by waiting for the cam marker, thus allowing the SP8700 to update the flow properly. The changes at the upper end are time-based, since several cam cycles occur in 30 seconds. In this manner the column is brought up to flow in the minimum amount of time with no operator intervention. This routine is loaded at the user’s option from block 39. B. Isolated Droplet Generator In this section, the effects of various instrumental parameters on droplet generation are presented. Operation of the instrument in the stand-alone mode is described and compared with operation in the computer-controlled mode. The use of the pressurized liquid delivery vessel is then contrasted with HPLC as a liquid source. Specifically, the effect of a gradient on the droplet stream is examined in detail. 167 The primary application for which the droplet generator was constructed was for use as a liquid introduction system to a miniature nanosecond spark source. The results of this experiment are presented. Several other applications are also described. 1. Operation in the Stand-alone Mode The instrument settings for droplet generation in the stand-alone mode are shown in Table 10. Table 10. Initial settings for stand-alone droplet generation. Source A: Internal Source B: Internal Bimorph Driver Source: Internal Pulsing Trigger Source: Internal Bimorph Driver Amplitude: 10.0 (Fully clockwise) Pulsing Divider:l-5 drops/drop pulsed Pulsing Voltage: 0 V Frequency Adjust Toggle: Unlatched Frequency Meter: Freqency, 0.1 sec Frequency Meter Select: Bimorph Driver In Deflection Voltage: 4 kV Strobe Lamp: External Input, Low Intensity The droplet generation procedure in the stand-alone mode is as follows. First, the liquid flow rate is adjusted to the minimum amount necessary to produce a jet of the desired velocity. The instrument is then turned on, and the bimorph driver frequency adjusted to the desired production rate. If the production frequency is not known, the frequency is slowly turned up from an initial setting of 168 about lkHz until a stable droplet stream is obtained. The frequency is then locked by placing the latch toggle in the ”A/D Latched” position. Droplets are then pulled out of the main stream by increasing the pulsing voltage until deflection can be seen. The phasing is then adjusted by decreasing the bimorph driver amplitude slightly until only a single droplet is deflected, and the deflection is maximized. The bimorph driver amplitude should remain as high as possible for maximum stream stability. The number of drops/drop pulsed is then adjusted on the BCD thumbwheel. Although this procedure is rather complicated, good results can be obtained with a little patience. Figure 56 shows the difference between the capillary stream with and without the applied bimorph oscillation. The top photo shows the stream with the instrument turned off and the irregularities and uneven size distribution is easily seen. The bottom photo shows the same stream with the instrument turned on and a bimorph oscillation frequency of 14.71 kHz applied. The droplets are uniformly spaced and monodisperse in size. In this case, the capillary size used was approximately 60 um, and the flow rate was about 1 mL min‘l; this gave droplets of about 1 nL in volume or 64 pm in radius. The frequency of 14.71 kHz is lower than predicted for the optimum droplet production frequency. Based on Rayleigh’s equations presented in the instrumentation 169 Figure 56. Top- Capillary jet without imposed oscillations. Bottom- Capillary jet with longitudinal oscillations imposed by bimorph. 170 chapter, the optimum droplet production frequency for a flow rate of 1 mL min‘1 and a 60 um capillary size should be about 25.1 kHz. In practice, it was found that imposing a square wave oscillation allowed much lower frequencies of droplet production than that predicted for a sine wave. It is believed that this is a result of the higher harmonics present in the square wave contributing to droplet formation. Upon close inspection, one can see that the stream first breaks up into pairs of small droplets which then quickly coalesce into large droplets produced at the fundamental frequency. This harmonic effect has the result of extending the lower range of droplet production frequency. For example, Table 11 lists some production frequencies obtained with a 30 um capillary and a flow rate of 0.43 mL min'l; these are well below the expected minimum frequency of 45 kHz. Although not all frequencies produce a stable stream, certain windows can be found which produce droplets at the fundamental frequency quite well. Pulsing of droplets in the stand-alone mode is shown in Figure 57. The top photo shows the effect of poor phasing. More than one droplet is pulsed at a time and neither has good deflection. When the bimorph driver amplitude is adjusted as described above, the stream looks like the second photograph, where a single droplet is pulsed out of the stream. The droplet that is pulsed out in this manner is a charged droplet. Uncharged droplets can be produced as 171 Figure 57. Top- Droplet stream showing poor phasing adjustment. Center- Droplet stream showing 1 charged droplet deflected. Bottom- Droplet stream showing 1 neutral drop separated. 172 Table 11. Low frequency droplet production characteristics. Bimor h Fre uenc kHz) Stream Stability 1.0 Unstable 1.2 Unstable 3.2 Marginal 5.0 Stable 8.0 Stable 10.0 Stable 12.0 Stable 14.0 Unstable 16.0 Stable 18.0 Stable shown in the third photograph. In this case, the rest of the stream is charged and deflected, and the charging electrode is turned off for a single droplet. This is accomplished by externally connecting the pulsing divider synchronization output to the external bimorph driver input and placing the bimorph driver source toggle to external. In either case, however, only a single droplet can be separated from the main stream in the stand-alone mode. 2. Operation in the Computer-controllgg Meg; The instrumental settings for production of droplets in the computer-controlled mode are shown in Table 12. Generation of droplets in the computer-controlled mode is much less painstaking than in the stand-alone mode. The frequency of droplet production is set using one of the two words IDG or RAMPIDG described previously in the software 173 Table 12. Initial settings for computer-controlled droplet generation. Source A: Internal Source B: Internal Bimorph Driver Source: External Pulsing Trigger Source: External Bimorph Driver Amplitude: 10.0 (Fully clockwise) Pulsing Divider:Not used Pulsing Voltage: 0 V Frequency Adjust Toggle: Not used Frequency Meter: Freqency, 0.1 sec Frequency Meter Select: Bimorph Driver In Deflection Voltage: 4 kV Strobe Lamp: External Input, Low Intensity External Connections:AM95l3 counter l to Bimorph Driver In AM9513 counter 2 to Pulse Trigger In chapter of this dissertation. If the frequency is not correct, it is adjusted in the highest resolution jumps possible by depressing a key. There is no need to lock or unlock the frequency adjustment. Phasing is automatically adjusted to approximate what is needed at each frequency. It is further adjusted from the terminal or under software control. This phasing adjustment is independent of the bimorph amplitude. Therefore, the bimorph amplitude can be kept at the maximum. Once the droplet production frequency is set, the pulsing voltage is turned up until a suitable deflection is obtained. The phasing adjust keys are then used to maximize the deflection and center the pulse on the drop. If more than one drop is being charged the pulse width keys are used in combination with the phasing keys to optimize the pulse 174 width and center it on the droplet packet. Utilizing these two sets of keys, a wide variety of droplet streams can be obtained. Figure 58 shows some of the droplet streams which can be created under computer control. The top photograph shows a square wave of droplets, with 10 drops charged and 10 uncharged. Uneven duty cycles of droplets can also be produced as in the second photograph, which shows 5 drops charged and 15 uncharged. In fact, any number of drops can be charged and any number left uncharged, within the limits of the l6-bit counters used to produce the waveform. The longest pulse width that can be produced under computer control is 65.5 msec. ' In the third photograph, the ability to control the charging pulse precisely is demonstrated. This droplet stream contains three drops which have received a full charge and two outer drops which have received a partial charge. This was accomplished by purposely shortening the charging pulse with the duty cycle keys and then interactively shifting the phase until the charge was properly balanced on the two outer droplets. 3. Use of tppf Constant Pressure Reservoir for Solution f _— Delivery The constant pressure reservoir described in the instrumentation chapter was constructed to deliver solution 175 Figure 58. Top- Stream showing 10 charged and 10 neutral droplets. Center- Stream showing 5 charged and 15 neutral droplets. Bottom- Stream showing triangular droplet packet. 176 in non-chromatographic applications. A regulated nitrogen or air source at 10 to 70 psi is introduced through the cap. The bulk flow rate through the capillary is adjusted by increasing or decreasing the pressure. Flow rates were measured by collecting the effluent in a 10 mL graduate over a known interval. The relationship between flow rate and pressure is quite linear as shown in Figure 59. In practice, a calibration curve can be established for each capillary to aid in setting a particular flow rate. One suggestion which may prove useful in maintaining the reservoir is to add a preservative to aqueous solutions to inhibit bacterial growth. The in-line filter was found to clog rapidly if old solutions were used. Of course, all solutions should be filtered prior to being put in the reservoir. A 0.45 pm nylon membrane filter of the type used in HPLC is recommended. Another solution to this problem may be to use disposable filters instead of the in-line variety. The filter should be able to withstand backpressures of 60 to 70 psi. 4. Use of an HPLC for Solution Delivery The use of a reciprocating piston HPLC pump for solution delivery differs from the pressure reservoir in two ways. First, because the flow is constant regardless of backpressure, the flow rate is much easier to reproduce on a day-to-day basis. Unfortunately, most reciprocating HPLC 177 .no333e cansua. ea- assuuuaso asuelsuv annual on s new Aunesoaasnes enseeena\eusa loan .00 emav Samoa... .900 .20, P p P n p p n p - D snoo .I 0005. shaman (uguJ/jtu) 910.: MOIJ 178 pumps have some degree of pulsation associated with the flow due to the need to refill the piston chamber. The degree of pulsation is dependent upon the quality of the pump, the amount of backpressure the pump is pumping against, the compressibility of the liquid, and the ability of the system to dampen the pulses. The effect of the pulsations on droplet production was, of course, a major concern. The SP8700 was set up to pump against a backpressure of approximately 1000 psi, and the aqueous effluent directed through the capillary. The droplet production parameters were set with the computer in the normal manner. For this system, the amount of pulsation was almost negligible, but a very slight visible movement of the drops along the direction of the jet trajectory was noticeable. If the droplets were charged in the normal manner, the result was a slight variation in the amount of charge delivered to each drop. Because this effect was so small, it was possible to compensate for it under computer control by lengthening the width of the charging pulse and using the phasing adjustment to center the pulse on the drop until no charging variation was observed. If the pulses were larger, as with a Milton-Roy single piston reciprocating pump, complete removal of the charging variation would be difficult, if not impossible. Fortunately, if neutral droplets are satisfactory for the experiment, all that is necessary is to impose enough charge on the droplets to deflect them into the trap. For this purpose, slight deviations in the 179 deflection do not matter. The neutral droplets will follow a linear trajectory regardless of the pulsation. 5. Upg;of the Droplet Generator witp Nonagueous Solvents Another important consideration for use of the droplet generator as an HPLC interface is the effect of nonaqueous solvents. All previous work on droplet making has been done with aqueous solutions. Three quantities important to droplet-making change when using different solvents. The first is the viscosity of the solvent. As the viscosity increases, one would expect droplet production to become more difficult due to increased damping of the applied oscillations. The second consideration is surface tension, which is one of the critical forces responsible for droplet formation. As surface tension decreases, droplet-making once again becomes more difficult. The third quantity is the ability of the solvent to conduct current. This is not an important consideration for droplet formation, but it could affect the charging of droplets. To test the effect of changes in surface tension and viscosity, the SP8700 was programmed to deliver a binary gradient which goes through a large change in the above quantities. Methanol and water is one such system for which the surface tension and viscosity are known over a wide variety of compositions. The distance from the tip of the capillary to where the stream first coalesces into stable 180 HOAA—Ohv BO HEON—vflh. HOHOANOI 0a hoafll OHSH«I\&¢ I *0 NOOMMI 6239 .6. 523 585.. 00— a W p . p L . . p . n r 0 e . o o o o o . o o .t e e o r e e e . O .. O r .souuslnou (mm) BONVLSIO aouaosznvoo easwah 181 droplets was noted at several points along the gradient. The results are plotted in Figure 60. The first point to note concerning this graph is that the longest coalescence distance is only about 14 mm, which is well within the limits of the pulsing electrode. The other interesting feature is that there are two maxima instead of the one which was expected based solely on viscosity. The viscosity of the solution has been plotted on the same graph in Figure 61. The major trend is correlated with viscosity, but a deviation occurs at the 1008 methanol end of the gradient. If the surface tension is plotted on the same graph as shown in Figure 62, one can see that this is the end of the gradient where the surface tension is very low. Since surface tension is a driving force for droplet formation, this may explain the second maxima observed near 908 methanol. The effect of the varying coalescence distance on droplet charging is similar to that produced by improper phasing. The time interval for droplet coalescence is changing as well as the distance, with the result that a phase adjustment is necessary to keep the droplets properly charged. Using computer control, the necessary commands to adjust the phasing can be performed automatically. As in the case of pulsating flow rates, however, a better trajectory will be obtained by using the neutral droplet stream as opposed to the charged stream. 182 usuaohv so assavsaw Hosanna- .omssao haueooeus Indians sagas-sou on sous: sauna-\uv s «o 0233. $23 Smog“... aoshhu COALESCENCE DISTANCE (mm) >2m 0m.) ON 00 . 0N .am ALISOOSIA 3N1V'138 ens-«h 183 COALESCENCE DISTANCE (mm) .sussao sousssu eosuass ususoae scans-sou usuaoae so assavsuu nossnasl ou sous: sasswl\ae s mo aoshhu $2302 $5.3 530mm... - P D b b b b m m b Dec“ O“ + d d ‘1 1T lid 4 d 4 Dec“ .Ne unease (ma/saufip) NOISN31 aovaans 184 The ability to charge the solvent is dependent upon several factors. The primary factor is probably the conductance of the solution, since the only way for the current to be conducted to ground when a charge is imposed on the droplet is through the solution itself. The other factor which may be important is the dielectric constant of the solution, since this should influence the ability of the liquid to stabilize the charge separation necessary for a net charge to be imposed on the droplet when it separates from the jet. Table 13 shows some of the solvents which are suitable for charging droplets and some which were not able to be pulsed with charging voltages of less than 300 V. Table 13. Solvent charging ability. Solvent Dielectric Specific Chargable? Constant Conductance (mho cpfl) Water 78.3 6 x 10‘8 yes Methanol 32.7 1.5 x 10'9 yes 2-Propanol 19.9 6 x 10'8 yes Acetonitrile 37.5 1 x 10'7 yes Hexane 1.88 < 10‘16 no CH2C12 8.93 4 x 10"11 no Tetrahydro- furan 7.58 4 x 10‘10 no The addition of a small amount of a solvent with a high specific conductance can drastically alter the nature of the 185 charging characteristics. For example, a 58 water/THF solution was able to be charged successfully. 6. 059 of the 129 to Iptrodpce Liquid to a Spark One of the original applications of the IDG was to introduce liquid into the miniature nanosecond spark source (MNSS) shown in Figure 63. This spark source had been used successfully as a GC detector (123), but attempts to interface it to HPLC had been hampered by the lack of a suitable interface. The firing of the spark was controlled by a thyratron trigger module which could be operated externally by a TTL-based input waveform. The strategy for droplet introduction was to send the entire stream through the spark gap without pulsing out drops. The unused drops would simply exit through an opening in the opposite side. The firing of the spark could then be phased with the droplet production waveform so that a drop could be positioned exactly between the spark gap when the spark plasma forms. To test this introduction technique, a 100 ppm calcium solution was delivered to the capillary using the pressurized sample reservoir. A 30 um capillary was used at a flow rate of approximately .25 mL ' min‘1 and an introduction frequency of 2.19 kHz. The droplet radii were approximately 50 pm. The spark was fired at the same rate, 2.19 kHz, so in this case all droplets were used within the 186 Efl-Bmu \ - Teflon 8.75' -_ § ? § . § 3 I325“ (Senator : § / ‘ '5 JL - - -. Sheath Gas 5 WWW/WWI _. W I 7////// qV/ /%§ ”mull”! III Ell (III/W II a. W. l/l' To Thyrstron Trigger Figure 63. Droplet generator used with miniature nanosecond spark source. 187 spark chamber. Phasing was accomplished using the AM9513 timing controller as described previously. The results of the phasing are illustrated in Figure 64. This part of the experiment does appear to be successful as noted by the minimum and maximum of the relative emmission intensity for calcium. The absolute magnitude of the signal, however, was very weak. Several different droplet introduction frequencies were tried, but none produced a strong enough signal to warrant use of this method in HPLC. There are two possible explanations for this weak signal. The first is simply an intuitive feeling that the spark may not be of sufficient duration to completely desolvate and atomize the sample. While the spark itself is of very high energy, the duration of the spark is only a few nanoseconds. Therefore the total amount of energy dispersed may be too small for desolvation of droplets of this size. The second explanation is based on a phenomenon observed visually in the spark gap when introducing droplets. Although great care was taken to line up the droplet stream with the spark gap, when the spark actually fired, remnants of the droplets were ejected from the sides of the spark plasma. Whether this is caused by sonic disruption or electrostatic repulsion, it is suspected that a large portion of the liquid did not remain in the vicinity of the spark plasma for a time sufficient for desolvation. 188 BO unwAe .usswus soussals sssna asumuaa assae\so«aosvoausu segment mo ausmun .wm hfiim wm53 58...... 25am shaman 189 7. Other HPLC Intgrfacing Techniques A much more conventional excitation source for use with HPLC-AAS or HPLC-AES is a flame. The standard method of liquid introduction into a flame is the nebulizer. Several experiments were performed to assess the efficacy of this method of liquid introduction for HPLC interfacing. The easiest way of interfacing the HPLC to the nebulizer is by direct connection. This method has the disadvantage of requiring the HPLC flow rate to approximately match the natural uptake rate of the nebulizer. Since most concentric nebulizers of the type used in AAS instrumentation have natural uptake rates on the order 'of 6-10 mL min'l, a compromise must sometimes be made by raising the HPLC flow rate to a point above the optimum for the column. If the mismatch is too great, noise spikes may appear on the AAS signal. In certain cases, depending on the solvent used, a good match can be obtained. The chromatogram shown in Figure 65 demonstrates the results obtainable from direct connection using methanol as the solvent. In this case, normal HPLC flow rates were able to be used without the appearance of noise spikes. A nickel diethyldithiocarbamate complex, Ni(DTC)2, was synthesized in-house and checked for purity by HPLC-UV as shown in the top trace. It was not immediately apparent which peak was the complex and which was an impurity. By using nickel (232 nm) AAS detection immediately following 190 UV w k T INJECT 1 MIN 44 MIN AA Figure 65. HPLC using AAS detection. Direct connect nebulizer interface. 191 the UV detector, the first peak was conclusively identified as a compound containing nickel. Two other observations can be made concerning this chromatogram. First, it is apparent that the signal-to-noise ratio for AAS detection is considerably lower than for UV detection of this complex. While this trend is not valid for all compounds, for most metal complexes the signal obtainable from UV detection has a higher signal-to-noise ratio than that obtainable by AAS. The second observation is that the width of the first peak has been increased considerably, even though a minimum amount of tubing was used to connect the UV detector to the nebulizer inlet. This band broadening is characteristic of nebulizer introduction and is dependent upon the exact nebulizer design used. A cross-flow nebulizer was also evaluated (122) and displayed much more broadening effects than the concentric nebulizer. Another approach which has been employed with nebulizers is shown in Figure 66. The dripping cup interface allows the nebulizer to operate at its optimum flow rate independently of the HPLC optimum flow rate. Since the HPLC optimum for analytical columns is lower than the optimum for most nebulizers, each drop falling into the cup is immediately swept into the nebulizer and aspirated. This type of interface should increase the efficiency of nebulization by letting the nebulizer work in a "starved" 192 TO HPLC DROP FORMER———> °(——COLUMN EFFLUENT 0 PLASTIC CUP -—-) TO NEBULIZER Figure 66. Dripping cup interface. 193 o M d 00. EC“: 3:0 U: a“ he “AH UH .5 HO HOOHOV GG EH“: OH:- NH 22 0.NnN ._.< 20....0mhw0 3wxo_z 2.2}: u 2.2\ .... . 194 node, and also decrease the amount of band broadening by allowing the spray chamber to clear between droplets. The resulting chromatograms, shown in Figure 67, consist of a series of spikes, one for each droplet nebulized. The droplets should be large enough to give the same signal as in continuous introduction, but small enough to minimize mixing within each drop. Other researchers (77) have reported that for aqueous introduction, 0.1 mL droplets are sufficient in size to give the same signal as for continuous introduction. Unfortunately, for nonaqueous solvents, the surface tension is too low to form drops of this size. Figures 68 and 69 compare results obtained with the direct approach with that obtained using the dripping cup method. Figure 68 compares the signal obtained by the two techniques for various flow rates. If the droplet size is large enough, the dripping cup method should be relatively insensitive to flow rate changes. The direct method is quite dependent on flow rate at low flow rates, but eventually levels off at high flow rates. The natural uptake rate of the nebulizer was about 2 mL nin'1 for this solvent mix. In each case the best signal is obtained at higher flow rates. Figure 69 shows the effect of different flow rates on the number of plates measured for the different techniques. The cup technique does appear to give higher plate counts, probably due to better clearing of the spray chamber between 195 .avoAHOI oosnuoasa avenge ens aao wounnusv sou ovauuswsl daswua no soauusaloo .mm unawau Assigns E5. 26.: ~20 n50 . 'NNOIS alums 196 ooshuoasw acouwv was 3:0 usummaue you wamsovsoua vans mo assauOAloo 36}; BE :3... .am SEJV'Id (EHHSVBW .avOAaOI shaman 197 droplets. The chromatographic efficiency drops off considerably at higher flow rates due to operation of the column at flow rates above the optimum. This is evidenced by the plate counts measured from the UV detector signal. The above results indicate that if signal magnitude is a primary consideration, the flow rate should be increased to a region where high sensitivity can be obtained. If chromatographic efficiency is the primary consideration, however, flow rate should be kept at a value which will maximize plate counts for the column being used. The inability to maximize both simultaneously is one of the fundamental problems of interfacing HPLC to element-specific detectors with nebulizers. g. Ug_fof the IDG witthlgggg The isolated droplet generator has been used by several researchers as a method of introducing liquids into flames (99-101). The type of burner used has a large effect on the signal-to-noise ratio observed when using this method. Three types of burners were evaluated for use with the IDG. The first was a slot-type burner commonly used for AAS. Although this burner was able to desolvate, atomize and excite the sample, the emission occurred very high in the flame. In this region, the flame position is very susceptible to drafts. Thus, the portion of the flame 198 viewed by the monochromator varied drastically, resulting in a large change in signal intensity. The second type of burner used was a standard Meeker-type laboratory burner with an air-natural gas flame. This flame had the advantage of being taller than the slot burner and less susceptible to room drafts. However, it was also slightly cooler than the air-acetylene flame used in the slot burner. Nevertheless, this flame gave considerably better results. A picture of the droplet introduction into the flame is shown in Figure 70. These droplets are being produced at a rate of about 30 kHz in the stand-alone mode, with one out of every 900 droplets produced deflected into the flame. The capillary size was 47 pm with a flow rate of approximately 0.5 mL/min of 100 ppm aqueous calcium. As can be seen from the top photograph, the droplets enter the flame at an angle slightly above horizontal. The hot flame gases sweep the droplets in the vertical direction and desolvation begins. Near the top section of the flame can be seen the emission from calcium. If the diameter of the droplets are measured and compared to the introduction rate, the rate of desolvation can be calculated. The signal obtained from this method of introduction is shown in the bottom photograph. Each peak represents the emission from one drop of calcium solution. The signal between peaks represents the flame background. Thus a signal obtained with widely spaced droplets could be used to 199 Figure 70. Top: Droplets being introduced into a flame. Bottom: Signal obtained from droplet introduction. Vertical axis: relative signal. Horizontal axis: 0.015 ms/div. 200 obtain the sample enission corrected for flame background. If the droplet rate is increased, eventually the emission peaks merge and continuous emission is obtained. For this combination of production rate and droplet size continuous emission was obtained at about one out of every 100 droplets introduced. In an effort to improve the reproducibility of droplet emission even further, a new burner was designed. This burner incorporates the best features of the Meeker burner, but can use a hotter air-acetylene flame. In addition, provision for a nitrogen sheath gas has been provided to further stabilize the flane against room drafts. The flame gas settings and flame appearances are shown in Table 14. A diagram of the burner construction is contained in Figure 71. Preliminary work with this burner has been successful. Emission can be observed with larger droplet sizes than could be used in either of the two previous burners. The characteristics of the flame are easily altered to suit the particular type of droplet stream used. The nitrogen sheath gas has been shown to provide an effective shield against moderate room drafts. This burner should provide a stable, well-characterized environ-ant for flame/droplet work in the future. 201 .mnouaouaoona seas-“Io pennant voyages“ sou banana sexes: occuhuooeluus voausoaa no amuse: .Hp own-um 0.3.7”. X95193... ...DOldom e Ihhu0( ~.___( Azzozm hoz munozv EE OF do» ene— 202 Table 14. Gas flows used for IDG burner. FuelgL min'll A1;(L min‘12 Height(ig) Appearance 1.0 17.0 7 cone shaped, oxidizing 1.2 17.0 8 cylindrical, oxidizing 1.5 17.0 9 blue outer, cyl., ox. 2.0 17.0 12 blue outer, cyl., ox. 2.5 17.0 14 cylindrical, oxidizing 1.0 14.5 6 cone, oxidizing 1.5 14.5 9 cylindrical, oxidizing 2.0 14.5 12 cylindrical, reducing 2.5 14.5 14 cylindrical, luminous 1.0 12.5 6 cone, oxidizing 1.5 12.5 10 cylin., stoichionetric 2.0 12.5 12 cylindrical, luminous 203 VI. CONCLUSIONS AND FUTURE PERSPECTIVES In this research dissertation, two instruments have been presented which can aid the analytical chemist in the separation and characterization of complex mixtures. Several unique approaches have been utilized in the instrumental design which make the intruments both powerful and easy to use. These design initiatives are consistent with current research trends and represent a significant contribution to modern analytical instrumentation. As in previous chapters, the multidimensional HPLC is discussed first. Important features of the instrumental design are briefly sum-arized, and conclusions drawn from the performance data are presented. Suggestions for future research using the instrument are also given. The important features of the isolated droplet generator are then summarized, its probability for success as an HPLC interface is evaluated, and several recommendations for future applications are outlined. 204 Michel HPLQ The multidimensional HPLC described in this dissertation has been successfully used for a wide variety of column switching and backflushing applications. Several design factors can be identified which have significantly contributed to this success. Detector-based event control was shown to improve significantly the precision of valve switching and synchronous parameter changes. Lack of such precision was one of the biggest problems preventing widespread use of the technique. Time-based sequencing had been used successfully, but the event times had to be determined on a trial-and-error basis. Run-to-run precision was shown to be quite poor for this method. Since the day-to-day variation in retention time is even greater, development of a consistent analytical lethod with valve switching based solely on time was very difficult, if not impossible. A well written detector-based event control strategy can adapt to changes in retention tine and thereby maintain the correct chromatographic relationship. In addition, it is possible to detect the presence or absence of various peaks and alter the event sequencing accordingly. Many other conditional event control sequences are possible using this approach. Indeed, the implementation of detector-based event control opens a new dimension of creativity in multidimensional chronatography. 205 Another factor which has prevented the technique fron becoming more popular was the lack of an integrated instrument. In the past, separate sequencers or tiners were used for each valve and had to be synchronized by hand. Even when multiple channel timers became available, solvent delivery was asynchronous with valve switching. Centralized computer control, when coupled with an intelligent solvent delivery system, per-its very detailed separation schemes to be employed where all events are synchronized. Such events can include solvent changes, flow rate changes, starting a gradient or even custom gradient construction through the use of solvent composition stepping. Other parts of the instrument can be controlled as well. The scale factor with which the data are plotted or the data taking rate can be altered using the same commands which control valving events. Such integration is necessary for truly automated operation. Perhaps the most important factor which has contributed to the success of this instrument is the development of an easy method for generating and saving the separation schemes for each experiment. Although centralized computer control is possible with any computer language, some are better suited to instrument control than others. The choice of the FORTH operating system for implementation of the system com-ands allows the user a choice of either entering a line of commands directly, interactively compiling a new command 206 frol existing words, or loading a previously defined command sequence from disk. The language is such more readable than machine code, faster than BASIC, and more easily modified than FORTRAN. The basic operation of the instrument is easily learned by novice users, yet the language is powerful enough to handle very advanced programming techniques. Thus, the ”user interface” can adapt to a wide variety of resercher’s needs and preferences. The applications presented show only a small portion of possible valving configurations. These do demonstrate the capabilities of MHPLC, however, as well as sole of the chromatographic requirements for successful implementation. Several observations are immediately apparent. First, most MHPLC separations require more extensive solvent manipulation than traditional HPLC. This is especially true if mode sequencing or stationary phase programming is being employed. Some proponents of MHPLC have claimed that column switching techniques can completely replace gradient separations. This may be true in certain cases, but this research did not find such a state-ant to hold true in general. Multidimensional HPLC is best thought of as a way to increase the power of gradient separations, not as a way to replace them. The cases where isocratic operation can be successful are in backflushing applications or heartcutting to a column of the same stationary phase. 207 Many avenues for future research with this instrument are available. Because the instrument operates in a hierarchical computing environment, much more computing power is available than in a stand-alone system. An excellent use of this computing power is in the application of artificial intelligence methodology to separation optinization. Simplex optimization of isocratic separations is a feature available on many commercial HPLC systems. This approach to optimization requires that several experinents be performed to define the initial simplex. With .the added computing power of a minicomputer, an expert system could be developed which could predict the initial separation parameters based on sample information. Optimization of gradient separations or even multidimensional separations may also be possible. The hardware for a closed-loop optimization schene is already in place. The microcomputer would function as an instrument controller and data acquisition device. The minicomputer would analyze the data, evaluate the separation, write a new separation program in the FORTH control language and send a command to the microcomputer to run the computer-generated separation. This could be a very exciting research project for someone interested in artificial intelligence as applied to chromatography. A possibility for future work in the valve-switching area is the use of the instrument to perform infinite-length 208 coluln separations by using heartcut recycling. This could be accomplished by the valving diagram shown in Figure 72. In this diagram, heartcuts are taken from column 1 onto column 2. Column 2 is then developed and the resulting cut is recycled back onto column 1. This process can be repeated indefinitely. For example, the sample is injected with both valves in the position shown. Solvent is flowing through column 1 only. When the detector senses the desired point in the chromatogram, valve 2 is switched, and a cut is placed on column 2. Valve 2 is then returned to the original position, and column 1 is allowed to clear. Then both valves are switched, and the cut from column 2 is recycled to column 1. Both valves are then returned to the positions shown, and the process is repeated. A detector capable of withstanding the column backpressure would be very helpful in this application, since it could be put at position A and the entire chromatogram detected at each recycle. B. Isolated Droplet Generation The isolated droplet generator presented represents the most versatile form of this instrument constructed to date. Nhils the stand-alone version is certainly capable of producing droplets in a variety of frequencies and size ranges, computer control enables the user to manipulate the droplet stream to a much greater degree. In addition, computer control enables the droplet stream to be stabilized 209 .>Ansswoaslouao uncanny ascends: son was: sonuauawuusoo .Np shaman N 223400 KOFUMs2_ oz< Q2DQ A — 223400 «obouhmo 210 much lore easily by allowing more precise control of the droplet production and charging waveforms. The fundamental studies with the HPLC as a liquid delivery source for the droplet generator are encouraging enough to support continued research in this area. It should be possible to produce droplets with all common HPLC solvents. Initial results with selected solvents show that both surface tension and viscosity affect the speed with which the Jet coalesces into a monodisperse droplet strean. However, these effects are reproducible and can be compensated. Droplet charging of nonaqueous solvents was not possible for all solvents, but should work for lost reverse-phase HPLC mobile phase systems. For those cases where the entire droplet stream can be sent to the excitation source (or other experimental destination), all common HPLC solvents should be usable. The droplet generator instrumentation was constructed to be versatile enough to act as a general-purpose high precision sub-nanoliter liquid handling system. Because of this versatility, many uses of the droplet generator in future research can be envisioned. Some work has already been done using the droplet generator as a micro-titrator or micro-pB stat (103-105). The computer-controlled version of the droplet generator could easily be modified to do this type of work. In addition, a feedback loop could be established between the 211 device detecting the endpoint (or pH) and droplet control. Thus, the rate of introduction of droplets into the titration vessel could be automatically altered on the basis of the state of the titration. With the advent of high performance TLC plates, coupling of column HPLC and BPTLC becomes an interesting way to do lultidimensional chromatography. The droplet generator could be used in this regard, both to select a portion of the column effluent to be sent to the TLC plate and as a means of automatically spotting the plate at the sane time. An application which could be implemented in this manner is the coupling of reversed-phase HPLC to normal-phase BPTLC. This type of mode sequencing is very difficult to do with multidimensional column HPLC because the types of solvent used in reversed-phase chromatography usually deactivate nor-a1 phase columns. By using the droplet generator to spot the HPTLC plate with reversed-phase solvent and allowing the solvent to evaporate, normal-phase development could be accomplished without solvent incompatibility problems. Another use of the droplet generator in HPLC may be as a possible detector based on droplet charging. The nonaqueous solvent charging data presented earlier shows that the solution characteristics drastically affect the ability to charge a droplet. Since charging of the droplets represents a current drain from the charging electrode, a 212 pico-ammeter connected in series with the electrode may be able to detect a change in the solution characteristics as a peak elutes. This detector would probably be similar in selectivity to a conductivity detector, but may exhibit lower detection limits. One final application of the droplet generator which may prove beneficial is as a liquid introduction technique‘ for laser induced breakdown spectroscopy (LIBS). LIBS is a technique which uses a sharply focussed laser beam to generate a plasma which can be used as an excitation source for atomic emission. Normally, temperatures in the plasma are as high as 10,000-100,00008, although the region of the plasma is quite small. The droplet generator could be used to position a droplet in the plasma region. If a pulsed laser is used to form the plasma, and if the laser could be triggered by the droplet generator, the phasing of droplet introduction with plasma formation could be optimized for maximum signal. Some initial studies have been done using droplet introduction without phasing, and the results were encouraging (124). Although the precision of the signal was quite poor, when a droplet was in the right position for plasna formation, observed emission was very intense. With synchronized laser triggering and better data acquisition electronics, IDG-LIBS could be a useful analytical technique. 213 1. 10. 11. 12. 13. REFERENCES Kralove, B. J. Chromatogr. 1980, 187, 466. . Majors, R.B.; Barth, B.G.; Lochmuller, C.H. Anal. Chem. 1984, 56, 317R. . Calkin, C., Ph.D. Thesis, Michigan State University: E. Lansing, MI, 1982. Zakaria, M.; Gonnord, M.F.; Guichon, G. J. Chromatogr. 1983, 271, 127-192. Huber, J.F.; Van Der Linden, R.; Ecker, E. J; Chromatogr. 1973, 83, 267. Scott, C.D.; Chilcote, D.D.; Lee, N.B. Anal. Chem. 1972, 44, 85. Snyder, L.R. J. Chromatogr. Sci. 1970, 8, 692. Liljamaa, J.J.; Ballen, A.A. J. Chromatogr. 1971, 57, 153. Dolphin, R.J.; Willmott, F.W. J. Chronatogr. Sci. 1976, 14, 584. Giddings, J.C. Anal. Chem. 1984, 56, 1258A. Snyder, L.R.; Dolan, J.W.; Van Der Wal, S.J. J; Chromatogr. 1981, 203, 3-17. Guiochon, G.; Gonnord, M.F.; Zakaria, M.; Beaver, L.A.; Siouffi, A.M. Chromatographia 1983, 17, 121. Guichon, 6.; Beaver, L.A.; Gonnord, M.F.; Siouffi, A.M. ; Zakaria, M. J. Chromatogr. 1983, 255, 415. 214 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. Huber, J.F.K.; Kenndler, B.; Reich, G. J. Chromatogr. 1979, 172, 15-30. Stewart, 6.3. J. Chronatogr. Sci. 1981, 19, 216. Little, C.J.; Stahel, 0.; Lindner, W.; Frei, R.W. Amer. Lab. 1984, OCT, 120-129. Eatz, S.; Pitt, W.W.Jr. Anal.Lett. 1972, 5(3), 177. Werkhoven-Goewie, C.B.; Deruiter, C.; Brinknan, U.A.T.; Frei, R.W.; Dejeng, C.J.; Little, C.J.; Stahel, O. J; Chromatogr. 1983, 255, 79. Saner, W.A.; Jadamec, J.R.; Sager, R.W.; Killeen, T.J. Anal. Chem. 1979, 51, 2180. Werkhoven-Goewie, 0.3.; Brinkmann, U.A.T.; Frei, R.W.; Deruiter, C.; DeVries, J. J. Chromatogr. 1983, 276, 349. Erni, F.; Frei, R.W. J. Chromatogr. 1978, 149, 561. Schauwecker, P.; Frei, R.W.; Erni, F. J. Chromatogr. 1977, 136, 63. Nielen, M.W.F.; Brinkman, U.A.T.; Frei, R.W. Anal. Chem. 1985, 57(4), 806. Nielen, M.W.F.; Brinkman, U.A.T.; Frei, R.W. Anal. Chem. 1985, 57, 806. Van Vliet, B.P.M.; Boatsman, T.B.C.; Frei, R.W.; Brinkman, U.A.T. J. Chromatogr. 1979, 185, 483. Majors, R. Lg 1984, 2 5, 358. Majors, R.E. J. Chronatogr. Sci. 1980, 18, 571. "BJOPI, Rene LC— 1984’ 2’ 1-4. Apffel, J.A.; Alfredson, T.V.; Majors, R.B. J; Chromatogr. 1981, 206, 43-57. Johnson, B.L.; Gloor, R.; Majors, 8.3. J. Chromatogr. 1978, 149, 571-585. Dolphin, R.J.; Willmott, F.W.; Mills, A.D.; Boogeveen, L.P.J. J. Chroaatogr. 1976, 122, 259-268. Willmott, F.W.; Mackenzie, 1.; Dolphin, R.J. J; Chromatogr. 1978, 167, 31-39. 215 33. 34. 35. 36. 37. 38. 39. 4o. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. Huber, J.F.H.; Fogy, I.; Fioresi, C. Chromatographia 1980, 13, 408. Huber, J.F.K.; Vodenik, R. J. Chromatogr. 1976, 122, 331. Miller, R.L.; Ogan, K.; Poile, A.F. Amer. Lab. 1981, July, 52. Miller, R.L. Perkin-Elmer Lig. Chrom. App. 1981, LC213, 1-18. Harvey, M.C.; Stearns, S.D. Amer. Lab. 1981, Feb., 151. Harvey, M.C.; Stearns, S.D. Lg 1983, 1(3), 172-174. Edholm, L.E.; Kennedy, B.M.; Berquist, S. Chronatographia 1982, 16, 158. Gfeller, J.C.; Stockmeyer, M. J. Chromatogr. 1980, 198, 162. Hageman, H.J.J.; Greving, J.B.; Jonkman, T.H.G.; Dezeeuw, R.A. J. Chromatogr. 1983, 274, 239. Pullen, R.H.; Cox, J.W. J. Chromatogr. 1984, 307, 155. Ranstad, T.; Mahle, N.H.; Metalon, R. Anal. Chem. 1977, 49, 386. Erni, F.; Keller, H.P.; Morin, C.; Schmitt, M. J; Chromatogr. 1981, 204, 65-67. Henry, R.A.; Byrne, S.H.; Hudson, D.R. J. Chromatogr. Sci. 1974, 12, 197. Tindle, R.C.; Stalling, D.L. Anal. Chen. 1972, 44, 1768. Lecaillon, J.B.; Souppart, C.; Aabadie, F. Chromatographia 1982, 16, 158. Niederwieser, A.; Staudenman, W.; Wetzel, E. J; Chromatogr. 1984, 290, 237. Ransteiner, R.A.; Bohn, K.H. J. Chromatogr. 1983, 260, 33. Smith, H.T.; Robinson, W.T. J. Chronatogr. 1984, 305, 353. 216 51. 52. 53. 54. 55. 56. 57. 58. 59. 60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. Hulpke, H.; Werthnann, U. Chronatographia 1980, 13, 395. Conley, D.L.; Benjamin, E.J. J. Chronatogr. 1983, 257, 337. Soner, W.A.; Jademer, J.R.; Sager, R.W.; Hilleen, T.J. Anal. Chen. 1979, 51, 2180. Radke, M.; Willsch, H.; Welte, D.H. Anal. Chen. 1984, 56, 2538-2546. Dark, W.A. J.;Liq. Chromatogr. 1982, 5(9), 1645. Hajibrahin, S.K. J. Liq. Chronatogr. 1981, 4(5), 749- 764. Ong, H.; Marleau, S. J.;Liq. Chromatogr. 1984, 7(4), 779. Hatz, H.; Ogen, H. "Chemical Analysis and Biological Fate: Polynuclear Aromatic Hydrocarbons, 5th International Symposium”, Battelle Press: Colunbus, OH, 1980; pp 169-178 ‘ Sonnefeld, W.J.; Zoller, W.H.; Wise, S.A.; May, W.E. Anal. Chen. 1982, 54, 723. Suatoni, J.C.; Garber, H.R.; Davis, 8.8. J. Chronatogr. Sci. 1975, 13, 367. Freeman, D.H. Anal. Chem. 1981, 53(1), 2. Majors, R.; Barth, H.; Lochnuller, C. Anal. Chen. 1982, 54, 323R. Majors, R.E.; Barth, H.G.; Lochnuller, C.H. Anal. Chem. 1984, 56, 300R-349R. Ettre, L.S. J. Chromatogr. Sci. 1978, 16, 396-417. Tswett, M.S. Ber. Deut.;§ot. G25; 1906, 24, 313. Egan, H. Proc. Soc. Apgl. Cheg; 1972, 9(12), 283. Bayer, H.; Anders, F. Naturwiss. 1959, 46, 380. Borman, S. Anal. Chen. 1982, 54, 327A. Dicesare, J.L.; Ettre, L.S. Chron. Rev. 1981, 251, l— 15. Mcnair, H.M. J. Chromatogr. Sci. 1984, 22, 521. 217 71. 72. 73. 74. 75. 76. 77. 78. 79. 80. 81. 82. 83. 84. 85. 86. 87. 88. 89. 90. McFadden, W.H. J. Chromatogr. Sci. 1979, 17, 518. larger, D.L.; Kirby, D.P.; Vouros, P.; Foltz, R.L.; Hidy, B. Anal. Chem. 1979, 51, 2324. Willoughby, R.C.; Browner, R.F. Anal. Chem. 1984, 56(14), 2626. VanLoon, J.C. an. J. Spect. 1981, 26(4), 22A—32A. Brinckman, F.B.; Blair, W.R.; Jewett, R.L.; Iverson, W. P. J. Chromatogr. Sci. 1977, 15, 493. Vickrey, T.M.; Howell, 8.8.; Paradise, M.T. Anal. Chen. 1979, 51, 1880. Slavin, W.; Schmidt, G.J. J. Chromatogr. Sci. 1979, 17, 610. Hausler, D.W.; Taylor, L.T. Anal. Chem. 1981, 53, 1223. Lawrence, K.H.; Rice, G.W.; Fassel, V.A. Anal. Chem. 1984, 56, 289-292. Billiet, H.A.H.; Van Dalen, J.P.J.; Schoennakers, P.J.; Fraley, D.M.; Yates, D.; Manahan, 8.8. Anal. Chem. 1979, 51(13), 2225. Fraley, D.M.; Yates, D.A.; Manahan, 8.8.; Stalling, D.; Petty, J. Appl. Spec. 1981, 35(6), 525. Gast, C.H.; Kraak, J.C.; Poppe, H.; Maessen, F.J.M.J. J. Chromatogr. 1979, 185, 549-561. Jinno, H.; Tsuchida, H. Anal. Let. 1982, 15(A5), 427. Rayleigh, L. Proc. Lond. Math. Soc. 1878, 10, 4. Rayleigh, L. Proc. Roy. §oc. (London) 1879, 29, 71. Savart Ann. Chim. 1833, t.1iii. Berglund, R.N.; Liu, B.Y.H. gpvir. Sci. Tech. 1973, 7(2), 147. Lane, W.R. J. Sci. Ipptrun. 1947, 24, 98. Mason, B.J.; Brownsconbe, J.L. J. Sci. Instrum. 1964, 41, 258. 218 91. 92. 93. 94. 95. 96. 97. 98. 99. 100. 101. 102. 103. 104. 105. 106. 107. 108. Cheng, L.; Cross, W.G. Rev. §pi. Ipptrum. 1975, 46(3), 263. Wolf, W.R. Rev. Sci. Inst. 1961, 32(10), 1124. —__ Abbott, C.H.; Cannon, T.W. Rev. §ci. Instrum. 1972, 43(9), 1313. Shabushnig, J.G.; Hieftje, G.M. Anal. Chim. Acta. 1983, 148, 181-182. Schneider, J.M.; Hendricks, C.D. Rev. Sci. In_trum. 1964, 35(10), 1349-1350. Schneider, J.M.; Lindblad, N.R.; Hendricks, C.D. l; Colloid Sci. 1965, 20, 610-616. Lindblad, N.R.; Schneider, J.M. J. Sci. Instrum. 1965, 42, 635. Hendricks, C.D.; Schneider, J.M. J. Phys. 1963, 31(6), 450-453. Bastiaans, G.J.; Hieftje, G.M. Anal. Chem. 1973, 45(12), 1994. Hieftje, G.M.; Malmstadt, H.V. Anal. Chem. 1968, 40(12), 1860. Heiftje, G.M.; Malmstadt, H.V. Anal. Chem. 1969, 41(13), 1735. Hieftje, G.M.; Mandarano, H.M. Anal. Chem. 1972, 44(9), 1616. Lemke, R.F.; Hieftje, G.M. Anal. Chim. Acta 1982, 141, 173-186. Russo, R.R.; Withnell, R.; Hieftje, G.M. Appl. Spec. 1981, 35, 531. Steele, A.W.; Hieftje, G.M. Anal. Chem. 1984, 56(14), 2884. Seymour, R.J.; Boss, C.H. Anal. Chem. 1982, 54, 1037. Joshi, B.M.; Sacks, R.D. Anal. Chem. 1979, 51(11), 1781. Kuhn, L.; Myers, R.A. Sci. Amer. 1979, 162. 219 109. 110. 111. 112. 113. 114. 115. 116. 117. 118. 119. 120. 121. 122. 123. 124. H nstrum. 1984, Newcome, B.; Enke, C.G. Rev. §ci. 55(12), 2017. Ratzlaff, E.H., Ph.D. Thesis, Michigan State University: E. Lansing, MI, 1982; pp 74-78 Wiegand, P.M.; Crouch, S.R. Anal. Inst. (In Press). MacMillan, D. Rlectronic Design 1981, Nov. 26, 195-201. ”The AM9513 Systen Timing Controller”, Advanced Micro Devices: Sunnyvale, CA, 1983. Wiegand, P.M.; Crouch, S.R. Talanta 1985, 32(1), 37-40. ”MOS/LSI Datebook”, National Seniconductor Corp.: Santa Clara, CA, 1977; pp 10/2-10/9 Goldstein, S.A., B.S. Thesis, University of Illinois: Champagne-Urbana, IL, 1967. Seymour, R.J.; Boss, C.H., Unpublished Manuscript, 1982. Solit, M.L.; Parsons, M.L. Anal. Chem. 1985, 57(6), 715A. Brodie, L. ”Starting FORTH”, Prentice—Hall, Inc.: Englewood Cliffs, NJ, 1981; pp 1-348 Rather, B.; Brodie, L. "Using FORTH", Forth, Inc.: Hermosa Beach, CA, 1981. Hoff-an, P., Ph.D. Research in Progress, Michigan State University: E. Lansing, MI, 1982. Danna, M.R., Ph.D. Thesis, Michigan State University: E. Lansing, MI, 1985. Trischan, H.H., Ph.D. Thesis, Michigan State University: E. Lansing, MI, 1985. Archontaki, H., Ph.D. Research in Progress, Michigan State University: E. Lanisng, MI, 1985. 220 APPENDIX A I?()IITPII EIIBI?I§III§T¢EET ... ........................ ..Create ....... ... ...... 12 O>IEY8 ............................ Colon........ ..... .... 12 OPEAIS ..... ....... . ........... ....Create... ............. 25 SA .... ............................ Constant .............. 11 SB ....... ..................... ....Constant... ...... ..... 11 SC .... ......................... ...Constant... ........... ll SFSCHG ... ...................... ...Colon................. 30 ’1STOP ........................ ....Variable ............ .. 20 ’ZSTOP ............................ Create.... ..... ....... 22 ’3STOP . ........................... Create.. .............. 10 ’4STOP ............................ Variable.... .......... 24 ’80! ....... ..................... ..Variable ............. . 22 ’BUF .............................. Variable. ............. 68 'RIST ............................. Create ................ 66 ’REST ............................. Create ............... . 21 ’TIME .......................... ...Variable .............. 24 +xrs .............................. Colon ................. 30 +A/D .. ..... . .................. ....Colon........... ...... 30 -xrs .................... . ......... Colon ................. 30 -A/D ..... ........ .................Colon...... ........... 30 .0018! . ......... . ................. Colon ................ . 68 .DDISH ............................Colon... .............. 22 .MARE ..... ..... ....... ......... ...Colon ................. 20 .UP ...............................Colon.... ........ ..... 4 .VAR .......... ....................Co1on ................. 62 0) ...... ...... ......... ........ ...Colon...... ........... 2 0ADC ... ....... ......... ........ ...Constant ............. . 18 1ADC ........ ..... . ....... . ........ Constant .............. 18 lDAC ..... ... ................ ‘......Constant... ........... 19 lDATA ...... .......... .............Variab1e .............. 24 IR ......... ..... ........ ...... ....Colon ................. 16 lSTOP ...... ...... ........ ...... ...Colon...... ........... 20 lTHRESH .. ........ ........ ..... ....Variable ............. . 24 1TIME ....... ........ ....... ....... Variable ..... . ........ 24 ZBYTES .................. ...... ....Colon........... ...... l5 ZDATA ..... ............ .. ....... ...Variable.... .......... 24 2R ......... ....... ................Colon................. 21 2THRESH ... ...... ..................Variable.... ......... . 24 2TIMI ..... ...... ... ..... ..........Variab1e... ...... ..... 24 3DATA ......... ...... ... ........... Variable.. ............ 24 222 3PROG . ............................ Colon ................. 3TIMI . ............................ Variable .............. 4STOP ............................. Colon ................. 95138U ............................ Colon ................. (BRO .............................. Colon ................. (TFI) . ............................ Colon ................. (TF1) ............................. Colon ................. >DAC .............................. Colon ................. >DRG .... .......................... Colon... ............. . )R ..... ........................... Create ................ ?OPBAES ........................... Colon. ............... . ?COM ... ........................... Colon.. .......... ..... ?CRBATE ........................... Colon ................. ?DT ... ............................ Colon. ................ ?MIN . ............................. Colon ................. ?PEAI .. ........................... Colon ...... ........... ?SY8 ... ........................ ...Colon........... ...... ODAT ............................ ..Colon ................. A/D .. ............................. Colon ...... . ......... . A/DCHG ............................ Colon ......... .. ...... A/DVALUBS ......................... Create.. .............. ABORTRUN .. ........................ Colon...... ........... ACCEPT .... ....................... .Colon ................. ADD . .............................. Variable ...... . ....... ADJ ............................... Colon ................. ADOFF .................... . ...... ..Varisble.............. ADRNG ............................. Variable.... .......... APORT ........................... ..Constant... .......... . ARM? ... ........................... Create ................ BACKGROUND ........................ Colon..... ........ .... BIMDIV . ........................... Variable .............. BOX . .............................. Colon. ................ . BRO ... ............................ Colon ................. BUILD ......................... ....Colon....... ........ .. BYE . .............................. Colon ................. CALDI ........................... .Colon ................. CAM .... ........................... Colon ................. CCW .... ............... . ........... Colon.... ............. CHIFRO ............. .......... .....Colon ................. CHTI .... .......... ...... .......... Constant.............. GLEN .. ............. ... ............ Constant .............. CLOCK .... ..................... ....Colon ................. CLR ... ..... ............ ....... ....Colon ........... ...... GLB ... .......... ........ ..... .....Colon ................. CMD ....... ...................... ..Colon .......... . ...... COMAND ...... .............. ........Constant... ........... CONDAC ... ....... ...... ............ Colon. ....... . ........ CONT ... ....... ....................Constant....... ...... . CPNTR ... ..... ...... ........... ....Variable...... ........ CPNTR ... ...................... ....Variable... .......... . CPORT ...... ............. ..........Constant.... ......... . CW ........ ........ . ...... .........Colon...... ........... CWORB ...... ...... . ..... .... ....... Constant... ........... DOPEAIS ........ ........ ..... ...... Colon....... ....... ... DAC ........ .......... ........ ..... Colon................. BATARIG .......... ..... . ..... ......Constant... .......... . BBISI ....... ........ .... ..... .....Colon ........ . ...... .. WORD TYPE INITIL ............................ Colon ................. INJECT ............................ Colon ................. INJECT ............................ Colon..... ........... . INJST ............................. Colon ................. LC ................................ Colon ................. LEN .............................. .Create ................ LENGTH ............................ Colon ................. LMARE ............................ .Colon ................. LOADEMUP .......................... Colon ............... .. LOOE-SEE .......................... Colon... ......... ..... LOPUL ............................. Variable .......... .... MAEEOUT . ...................... ....Colon. ................ MAEEOUT ........................... Colon ....... . ..... .... MAPR ............................. .Constant ............. . MARE ............................ ..Variable... ...... ..... MM ......... ...... . ....... .........Colon................. MMARE .......................... ...Colon................. MOREDATA ......................... .Colon................. MULTITASEING ...................... Colon............ ..... NEWNUM ............................ Colon................. NEWREST ......................... ..Colon................. MEET ............................. .Colon................. NF . ................... . ........ ...Colon..... ...... . ..... NOACTION .............. ............Colon................. NOBRO ..... . ....................... Colon............ ..... NOW ........... . .................. .Colon..... ............ NOWBLE ............................ Create ............ .... NOWBLE ............................ Variable.. ........... . NET ......... .................. ....Colon................. NXTPNT ............................ Colon... .......... .... NXTPNT . ....................... ....Colon....... .......... OFF . .............................. Colon ............. .... OFSET ............................. Variable ............. . ON ................................ Colon ................. ONE-SHOT .......................... Colon. ......... ....... PARAMS? ....................... ....Colon ................. PER .............. ..... ............ Code .................. PF ................................ Colon ................. POS . .............................. Colon ............. .... PRES .............. . ............... Constant. ............. PRESS ................ .. ........... Colon ....... .. ........ PRNMIN ............................Colon................. PSTORE ............. . ..... .........Colon ......... . ...... . PURG ..... ......................... Constant... ........... R12 . ............................ ..Code. ................ . RAMPIDG ............... .... ...... ..Colon....... ......... . RANGE ..... ...................... ..Variable ....... . ...... RATE? ... .............. .... ..... ...Colon....... ...... .... RD ..... ......................... ..Colon............ ..... READ ...... ...................... ..Code. ................ . RECALL .... .............. ..........Colon.... ..... ........ REGRAPH ... ............. ...........Colon... ............. . RENAME .... ...... . ...... ...........Colon.... ....... ...... REPORT ..... ............. ..........Colon................. RESET ....... ...... ................Colon................. RESET ... ........... .. ......... ....Colon................. RESET ........... ....... .. ....... ..Constant.............. 224 DEL ............................... Create ................ DELAY ............................. Colon ................. DELAY ............................. Variable. ............. DELI .............................. Constant .............. DELSC ............................. Variable .............. DIGREC ............................ Colon ...... . .......... DISP .............................. Constant.. ............ DLY ............................... Create ........ . ....... DOZBACE ........................... Colon... .............. D03BACE ........................... Colon. ................ DPORT .... ..................... ....Constant... ........... DRG . ............... .. ............. Colon ................. DSPEC . .......................... ..Create..... ........... DT ........................... .....Colon................. DTFC .............. ....... ...... ...Colon... ...... ....... DTI ............................. Colon ........ ......... BUMPER . ........................... Colon................. EDFI .............................. Constant........ ..... . EDITMODE ..... ...... . ............ ..Colon............ ..... EDRU .......................... ....Constant... ....... .... EHELP ..... ...... . ..... ............Colon................. EMARE ................ ........... ..Variable.............. ENDDATA ... ............ ............Colon................. ENTE ..... ............. . ....... ....Constant.............. ENTER .......................... ...Colon................. .ENTERO ... ......... . ........ .......Colon...... ..... ...... EVENT .... .......... ...............Colon................. FASTDT .. ..................... .....Colon........... ..... . FFLAG ... .......................... Create..... .......... . FLOW ............................. .Constant .............. FLOW? ............................. Colon. ............... . FLOWST ........... . ................ Colon..... ............ FREE .................. . ......... ..Colon.. .............. . FTDC ............. . ................ Colon... .............. GET. .... ....... ...... ............ Colon........... ...... GETCHR ............. ............... Code..... ............. GETFIRST ................. ..... ....Colon... .............. GETIT ....... .......... ....... .....Colon.... ........... .. GLDT ...... ....... .................Colon.... ........... .. GO ................................Colon. ................ HEAD ................... ..... ......Colon....... ........ .. HELP .................... ....... ...Colon..... ............ HIPUL ..... ...... ..................Variable.... .......... HOLD ........... ...... .... ...... ...Constant.... ......... . HT ......... ...... .................Colon................. ICPEAES ...........................Colon................. IDG .... .......... .................Colon... ...... ........ IDG ......... ....... ...... ...... ...Colon.... ........ ..... IN ........... ..... ................Colon................. INGLDT ...... ...... ................Colon.... ..... ........ INIT ..............................Constant.............. INITBUFF ..........................Colon... ....... ....... INITBUFF ..........................Colon................. INITDAT ..... ........... . ...... ....Colon................. INITI .... ........ ........ ...... ...Colon..... ........ .... 225 WORD TYPE REST .............................. Colon ................. REST .............................. Colon ................. RESUME ............................ Code .................. REVERSE ........................... Colon ................. RFLAG ............................. Variable .............. RNAME ............................. Colon ................. RPT . .............................. Colon ................. RPUNC ............................. Variable .............. RST75 ............................. Code..... ............. RSE .. ............................. Colon..... ............ RT. ............................... Colon ................. RUGR .............................. Constant .......... .... RUN ............................... Colon................. RUNIDG ............................ Colon. ................ RUNMODE . .......................... Colon. ................ S .. ............................... Colon.... ............. 812 ...... ....................... ..Code.. .......... . ..... SAVE .............................. Colon... ....... .. ..... SCCW ....... ................... ....Colon........ ......... SCW ..... ........... . .............. Colon... ..... . ........ SEC ..... .......................... Colon.... ...... . ...... SELECT ........... . ................ Colon... ....... ....... SELECT . ................ . ..... .....Colon................. SETDELAY .......................... Colon .......... .. ..... SETEMUP ........................... Colon. ...... . ..... .... SETTIME ........................... Colon.... ............. SETUP .. ........................... Colon...... ........... SHOW ..... ......................... Colon ........... ...... SLOWDT .. .......................... Colon....... ........ .. SREST ... .......................... Colon. ............. ... SREST ............................. Colon. .............. .. START .. ........................... Colon...... ...... ..... START .. ........................... Variable... ......... .. STARTEMUP ......................... Colon ........ . ........ STARTIDG .. ........................ Colon ............ ..... STARTTFI .. ........................ Colon. ................ STFI ....... ............ ...........Constant. ............. STORE . ........... ........ ......... Colon..... ........ ..... STP ... .......... ..... ............. Constant.............. STS ....... ............. ....... ...Constant .............. STURN ................... .......... Colon ................. STY .................... ........ ...Colon ............ ..... SUSPEND .................. ....... ..Colon.. ..... .... ...... SW ................................Colon.... ............. SYNC ........ ...... .... ............ Colon... .............. SYNC ...... .......... ... ........ ...Colon.... ..... ........ STS ....... ........... . ........ ....Colon.... ............ . SYSTIM ........ ..... ... ........ ....Colon.... ......... .... TAEEDATA ..... . ...... . .......... ...Colon...... ........... TALE . ......... . ....... . ...... .....Colon.... ......... .... TEST ........... ............. ......Constant.... .......... TFI ............... ............. ...Colon...... .......... . TIME ....... ....................... Constant.............. TIMEOUT ...........................Colon................. TOGGLE ........ ...... ...... ...... ..Colon........... ...... TURN ...... ........ ..... ....... ....Colon.... ............ . UPBL ... ................ ... ..... ...Colon... ....... . ..... . 226 24 28 17 UPDDISE ........................... Colon ................. 28 UPLIM .......................... ...Create ................ 67 UPNUM ... ...... ....... .......... ...Colon .......... . ..... . 28 UPTIM .. .......... ... .............. Colon.... ............ . 4 V09 ............................... Colon ................. 35 VALVE .. ........................... Colon . ............... 26 VIEW . . ........................... Colon. ................ 27 VTAB .............................. Variable .............. 31 W ...... ........................... Colon...... ........... l4 W . ........ ........ ......... ..Colon. ................ 37 WAIT ............................ ..Colon ................. 21 WFPIP .. ........................... Colon ................. 35 ENE ............................... Colon ................. 9 XXX ............................... Colon. ................ 69 XXX ..... ..... . .................... Colon........... ...... 69 ZOPEAES ................ ...........Colon.... ...... ....... 25 ESE ...... ....... ......... ......... Colon.... ..... . ..... .. 20 ZERO ..... ........ ........ ......... Colon.... ............. 20 [?PEAE] ........................... Colon.... ............. 24 \/ ... ............. . ............... Constant .............. 12 227 Block Nunber: l 0 ( MULTITASEING WORDS) 1 2 BACKGROUND CREATE )R OVER + DUP HERE + 2+ 2+ , SWAP 3 HERE + 2+ . R) + ALLOT ; 4 5 BUILD OPERATOR O OVER 0 96 MOVE 20 DUP OPERATOR 6 0 1+ ! 5 + ! ; 7 8 HEX CODE RESUME H POP M E MOV H INE M D MOV XCHG 9 F7 0 M MOV NEXT JMP DECIMAL 10 ll 12 13 14 15 Block Nunber: 2 ( COMMON UTILITIES) 0 l 2 ( SHOW: BEG BLE S,END BLH O -- [CNTRL/Z8STOP, =NEET BLE] ) 3 : SHOW 1+ SWAP DO I DUP CR ." BLE O" . CR 4 LIST I 1+ BLOCE DROP HEY 26 8 IF LEAVE THEN LOOP ; 5 : 0) MINUS 0< ; 6 REVERSE ( n--> , 0=all) DUP 0) NOT IF DROP DEPTH ELSE DUP 7 DEPTH > IF ABORT” BAD ARC" THEN THEN 1+ DUP 0 DO PAD I 24 + ! 8 LOOP PAD O 1 DO PAD I 23 + O LOOP ; ( REVERSES STACE) 9 ° NOBRO UPCR WPRMPT UP" SET /NOBRO=TI:” UPCR ; 10 : BRO UPCR WPRMPT UP” SET /BRO=TI:” UPCR ; 11 NOBRO 13 : TALE BRO WPRMPT TALE : l4 : SEC CR DUP . ." second delay ...” 0 DO 14230 0 DO LOOP LOOP ; l5 : HT L 0 VV ; Block Number: 3 ( IDG LOAD BLOCE) 10 LOAD ( MULTITASEING WORDS) 15 LOAD ( 9513 WORDS) 61 65 THRU ( IDG WORDS) : SLOWDT 66 LOAD ( DATA TAEING COUNTER WORDS) 18 LOAD ( ADC BASIC ROUTINES) 22 LOAD 23 LOAD ( DATA TAEING ROUTINES) 28 LOAD 29 LOAD ( DATA TAEING CONTROLLING WORDS) ; ‘ FASTDT 67 69 THRU ; 10 : ?DT ." Enter F for fast data taker, S for slow?” KEY 11 CR DUP 70 = IF FASTDT THEN 83 2 IF SLOWDT THEN 3 12 ?DT madmuauNo—O 228 Block Nunber: 4 QQQQMQUNMO ( GET SYSTEM TIME AND DATE AND PUT ON STACK) CODE GETCHR BEGIN UICMD LDA 2 S ANA 0= NOT END UlDATA LDA A L MOV 0 S H MOV HPUSH JMP .UP ." TIM ” ; UPTIM [’] .UP 3 + 4 UPCHAR ; NOW UPTIM 13 UlDATA C! 24 0 DO GETCHR LOOP 24 REVERSE 2DROP 2DROP 2DROP ; : SYSTIM NOW 18 0 DO EMIT LOOP CR 1 FREE 30238 HERE - 3 BYE BYE TALE 3 ?CREATE )IN O -’ NOT IF HERE COUNT TYPE ." isn’t unique ” CR THEN DROP >IN ! (CREATE) ’ ?CREATE ’CREATE 2 Block Nunber: 5 ODQGGbUNl-‘O ( MEMORY DUMP USE: START [add --> ]) ( Cntrl/ZIEEIT,=FORWARD ONE SCREEN, (SP>=BACE SCREEN) ( F=FORWARD CONTINUOUS, B=BACE CONTINUOUS, ANY EEY STOPS) VARIABLE ADD NEXT CR 23 0 DO ADD O I + DUP CO SWAP ." Address: . DUP 9 EMIT 9 EMIT ." Value: " . DUP 32 > SWAP DUP 127 < HOT AND IF 9 EMIT 9 EMIT .” Ascii: ” EMIT ELSE DROP THEN CR LOOP ADD O 23 + ADD ! ; START ADD ! BEGIN NEXT HEY DUP 26 = IF ELSE DUP 32 = IF DROP ADD O 46 - ADD ! 0 ELSE DUP 66 = IF DROP BEGIN ADD O 46 - ADD ! NEXT ?TERMINAL END 0 ELSE DUP 70 = IF DROP BEGIN NEXT ?TERMINAL END 0 THEN THEN THEN THEN END 3 Block Number: 6 CDQGQfiGNHO ( NEW DEFINITION FOR NUMBER) VARIABLE RPUNC RT. 0 RPUNC ! 15 0 DO DUP I + CO DUP 32 = IF RPUNC O IF I RPUNC O - RPUNC ! THEN LEAVE ELSE DUP 58 8 IF I RPUNC ! THEN DUP 44 48 WITHIN IF I RPUNC THEN TEEN DROP LOOP RPUNC O IF -1 RPUNC +2 THEN NEWNUM RT. (NUMBER) ; ’ NEWNUM ’NUMBER ! 229 Block Nunber: 7 O ( SEARCH ACROSS BLOCE BOUNDARIES) l ( TO USE: endblk 8 xxx SEARCHES FROM CURRENT TO endblk) 2 3 EDITOR DEFINITIONS 4 5 . S OF STRING DROP BEGIN -FOUND 6 IF SCR O DUP . 1+ DUP BLOCE DROP SCR ! TOP 7 ELSE LINE SCR O . CR THEN 8 ?TERMINAL IF DUP 1+ SCR ! THEN 9 DUP SCR O < END DROP ; 10 ll FORTH DEFINITIONS 12 13 14 15 Block Number: 8 '0 1 2 3 4 5 6 7 8 9 10 ll 12 13 14 15 Block Nunber: 9 0 ( [65,67IMICRO.FTH DISE DIRECTORY) 1 2 ( COMMON UTILITIES) 2 LOAD 4 LOAD 3 DUMPER 5 LOAD : ( MEMORY DUMP) 4 MULTITASEING 1 LOAD ; 5 LC 6 LOAD 10 32 THRU 35 37 THRU ; ( HPLC CONTROLLER) 6 : IDG 3 LOAD ; 7 : XXX ; 8 9 10 ll 12 13 14 15 230 Block Nunber: 10 0 ( BEGIN MHPLC CONTROLLING ROUTINES) 1 2 . DELAY 0 DO 14230 0 DO LOOP LOOP : ( n --> ) 3 1 LOAD 4 CREATE ’SSTOP l C. 5 50 50 50 BACKGROUND lBACE lBACE BUILD 6 50 50 50 BACKGROUND 2BACE 2BACE BUILD 7 50 50 50 BACKGROUND 3BACE 3BACE BUILD 8 50 50 5O BACKGROUND 4BACE 4BACE BUILD 9 10 : ACCEPT DUP PAD SWAP EXPECT 1+ 0 DO I PAD + CO 0: IF 11 32 I PAD + C! THEN LOOP PAD 1- NUMBER ; ( n -->n2) 12 13 ’ (CREATE) 'CREATE : ( TURN OFF NOT UNIQUE MESSAGE) l4 15 Block Number: 11 O ( SP8700 KEYBOARD CONSTANTS) DECIMAL 1 96 CONSTANT CLEN 114 CONSTANT CHTI 2 112 CONSTANT O4 4 CONSTANT O9 3 120 CONSTANT O5 24 CONSTANT O8 4 100 CONSTANT O6 16 CONSTANT O7 5 5O CONSTANT INIT 0 CONSTANT MAPR 6 42 CONSTANT RUGR 8 CONSTANT FLOW 7 34 CONSTANT HOLD 20 CONSTANT SC 8 52 CONSTANT CONT 2 CONSTANT SB 9 4O CONSTANT PURG 10 CONSTANT SA 10 32 CONSTANT STP 18 CONSTANT TIME 11 48 CONSTANT Ol 12 CONSTANT EDRU 12 56 CONSTANT O2 36 CONSTANT O3 13 76 CONSTANT TEST 74 CONSTANT DISP 14 66 CONSTANT PRES 80 CONSTANT O0 15 88 CONSTANT O. 68 CONSTANT ENTE Block Number: 12 0 ( SP8700 KEYBOARD CONSTANTS, CONTINUED) 1 106 CONSTANT EDFI 116 CONSTANT DELI 2 98 CONSTANT STFI 104 CONSTANT \/ 3 4 CREATE O>EEY OO O, O1 C, O2 C, O3 C, O4 C, O5 C, 5 O6 C, O7 C, O8 C, O9 C, 6 7 ( CONVERT AN UNSIGNED 16-BIT NUMBER TO ONE EEY CODE PER DIGIT) 8 : O>EEYS PUNCT O IF DROP THEN DUP (O OS O) 9 DUP HOT HOT 0 DO DUP I + CO 48 - 10 PAD I + C! LOOP DROP 0 DO PAD I + CO O>EEY + CO LOOP ll RPUNC O IF RPUNC O 0 DO PAD I + C! LOOP O. 12 O RPUNC O 1- DO PAD I + CO -1 +LOOP THEN : 13 14 GETO 6 ACCEPT O>EEYS : 15 231 Block Number: 13 ( BUTTON PRESSER WITH RIM CHECK) DECIMAL 8 BASE ! ‘DQQGGOQNHO t-u—t-o NOD-0° 13 14 15 177000 CONSTANT APORT 177002 CONSTANT CPORT 177003 CONSTANT DPORT 212 CONSTANT CWORD ( A-) DECIMAL B(-, CLOW->, CHI<- ) 37 CONSTANT RESET ( BASE 2-- 00011111) CODE RST75 RESET O A MOV SIM NEXT JMP CODE READ 0 O H MOV RIM A L MOV HPUSH JMP ' SETUP CWORD DPORT C! 1 APORT C! ; 2 PRESS 33175 DUP APORT c: IF . ." Acknowledged ” ELSE l APORT C: Block Number: 14 ‘DOQQOIIDOONHO ( VALVE TURNING ROUTINE CPORT CO DUP 16 / SWAP 15 AND - SETUP 2000 0 DO LOOP READ 64 AND ABORT” Not Acknowledged -- Run aborted ” THEN CR 1 APORT C! 1500 0 DO LOOP ' VALVE O -- ) TURN CPORT C! 5000 0 DO LOOP ABORT" VALVE FAILURE -- RUN ABORTED" 3 SELECT HOT 0 DO 2 O SWAP DUP ROT + LOOP 2 / 3 MAEEOUT SELECT CPORT CO 16 / CR ; CCW DUP 0 1 MAEEOUT OR TURN DROP .” Valve O" . .” switched CCW” CR CW DUP 1 254 MAEEOUT AND TURN DROP .” Valve O” . .” switched CW" CR ; CAM RST75 BEGIN READ 32 AND END 3 ENTER ENTE 0 REVERSE DEPTH 0 DO PRESS LOOP 3 3 ENTERO O>EEYS ENTER ; FLOW? RST75 BEGIN READ 16 AND END 7 EMIT .” Flow ready" CR WCR ." Wait for beep... Block Number: 15 ( CLOCK ROUTINES - CONSTANTS & BASICS) DQQQGDQNHO OCTAL 177771 CONSTANT COMAND 177770 CONSTANT DATAREG DECIMAL CMD ( N -> ) COMAND C! DRG ( N -> ) DATAREG C! ( Delay till cam marker) ( 9513 command register address) ( 9513 data register address) ( send data to conmand register) ( send data to data register) ELSE DUP 256 / THEN SWAP 255 AND 3 : )DRG ( N--> ) 2BYTES DRG DRG 3 HEE : RESET FF CMD 5F CMD 17 CMD ; ( NOTE--RESET LEAVES D.P. AT MM REGISTER) RESET 232 (DRG ( --> N) DATAREG CO DATAREG CO 256 O + ; 2BYTES ( n--) cHI,cLO) DUP 0< IF DUP 32767 AND 256 / 128 DECIMAL + Block Nunber: l6 0 ( 9513 SETUP FOR SECONDS ELAPSED) BINARY 1 . 1R 11111111 CMD 01011111 CMD ( Initialize) 2 00010111 CMD ( MM reg) 11110000 DRG 11001010 DRG ( TOD,BCD div) 3 00000011 CMD ( O3 lode reg) 00100001 DRG 00001110 DRG ( node D) 4 00001011 CMD ( O3 load reg) 01100100 DRG 0 DRG ( = 100.) 5 00000100 CMD ( O4 mode reg) 00001000 DRG 0 DRG ( mode A) 6 00001100 CMD ( O4 load reg) 0 DRG 0 DRG ; DECIMAL 7 : INJST CPORT CO 128 AND NOT 3 ( 1=Inject, O=Load) 8 INJECT CR INJST IF ." Load sample” BEGIN l DELAY 9 INJST 08 END CR THEN ." Inject sanple" CR BEGIN l DELAY 10 INJST END 108 CMD ( load I arn counters 3 L 4) ; 11 (TFI) 168 CMD ( save O4) 20 CMD (DRG ( --> 10ths secs) ; 12 TFI (TFI) l 10 O/MOD SWAP DROP ( secs from inj) ; 13 SYNC CR INJST IF .” Load sample” BEGIN 1 DELAY INJST 0: END 14 THEN CR .” Ready to inject?” KEY W DROP CAM 108 CMD RUGR PRESS 15 CR ." Inject now" BEGIN 7 EMIT 1 DELAY INJST END CR ; Block Number: 17 0 ( EVENT CONTROLLER) l : CLR 27 EMIT 69 EMIT 27 EMIT 106 EMIT ( clear screen) ; 2 POS 31 + >R 31 + R) 89 27 4 0 DO EMIT LOOP ( col,row--> ) ; 3 PRNMIN DUP 60 MOD SWAP 60 / 4 DUP 10 < IF 32 EMIT THEN . 8 EMIT 58 EMIT 5 DUP 10 < IF 48 EMIT THEN . ( eec--> ) ; 6 : RPT 0 DO DUP EMIT LOOP DROP 3 7 : BOE 27 EMIT 70 EMIT 70 1 POS 102 EMIT 97 6 RPT 99 EMIT 8 70 2 POS 96 EMIT 77 2 POS 96 EMIT 70 3 POS 118 EMIT 97 6 RPT 9 116 EMIT 70 4 POS 96 EMIT 77 4 POS 96 EMIT 70 5 P08 101 EMIT 10 97 6 RPT 100 EMIT 27 EMIT 71 EMIT CR 3 ll ABORTRUN CLR ABORT” ABORTED" ; 12 13 EHELP CLR llSPEC CO 77 8 IF 34 LIST l4 ELSE .” Help unavailable while taking data.” THEN ; -15 Block Number: 18 0 ( ADC BASIC ROUTINES) 1 OCTAL 2 177700 CONSTANT STS 3 177710 CONSTANT 0ADC 4 177711 CONSTANT 1ADC 5 DECIMAL 6 CODE R12 1ADC LHLD HPUSH JMP 7 CODE 812 0ADC STA NEET JMP 8 ° RUN BEGIN $12 1000 0 DO LOOP R12 16 /MOD l6 U.R CR 9 DROP ?TERMINAL END 3 10 : ODAT 512 R12 16 IMOD SWAP DROP 3 ll 12 l3 14 15 233 Block Nunber: l9 0 ( DAC BASIC ROUTINES) 1 OCTAL 2 177720 CONSTANT 1DAC 3 DECIMAL 4 !lDAC 1DAC C! ; 5 6 ( SP8700 MODE SELECTION) 7 8 EDITMODE RST75 \/ APORT C! 2000 0 DO LOOP READ 64 AND NOT 9 1 APORT C! 1500 0 DO LOOP IF EDRU PRESS THEN 3 10 11 : RUNMODE RST75 \/ APORT C! 2000 0 DO LOOP READ 64 AND 12 1 APORT C! 1500 0 DO LOOP IF EDRU PRESS TEEN 3 Block Nunber: 20 ( DIGITAL RECORDER DRIVER) VARIABLE OFSET VARIABLE RANGE CVARIABLE ’lSTOP : lsTOP l ’lSTOP C! ; CVARIABLE MARK : :MARE MARK C: ; 10 :MARK ° ZER 2048 DUP ROT SWAP 100 O/ - DUP OFSET : 4096 SWAP - RANGE : ; 20 ESE ZERO CLR .” E OF FULL SCALE? " 4 ACCEPT ZER 3 CONDAC OFSET O - DUP 0( IF DROP 0 THEN 256 RANGE O O/ ; : .MARE ODAT CONDAC DUP MARK CO DUP ROT + 255 > 10 IF - ELSE + THEN :lDAC 3 ' ll : ?MIN BEGIN TFI DUP 0= IF DROP 1 THEN 60 MOD DUP 0= IF .MARE 12 THEN PAUSE END 3 13 : LMARE CLR .” LENGTH OF MARK (2.2MM/UNIT)? ” 4 ACCEPT :MARK ; l4 : DIGREC 0 ’lSTOP C! lBACE ACTIVATE BEGIN PAUSE ODAT CONDAC 15 :lDAC ?MIN ’1STOP CO PAUSE END STOP 3 @UQQUbuND-‘O Block Nuaber: 21 ( VARIABLE DELAY TIMERS n -—> ; n=100ths secs.) 2R 5 CMD ( O5 MODE REG) 0 DRG 15 DRG ( MODE A) 1 CMD 0 DRG l5 DRG 2 CMD 0 DRG 15 DRG 3 CREATE DEL 100 , CREATE ’REST 10 . CREATE ARM? 1 C, 0 1 2 3 4 5 : SETDELAY ( n--)) 13 CMD ( O5 LOAD REG) ZBYTES DRG DRG 6 112 CMD ( LOAD I ARM O5) 13 CMD 0 DRG 0 DRG ; 7 : TIMEOUT ( n-->) SETDELAY BEGIN PAUSE 8 176 CMD ( SAVE O5) 21 CMD (DRG ( GET O5) -1 = END 3 9 : SREST ( n-->) 9 CMD ZBYTES DRG DRG 97 CMD 9 CMD 0 DRG 0 DRG 3 10 : REST ( n-->) SREST BEGIN PAUSE 161 CMD 17 CMD (DRG -1 = END 3 11 : SW ( n-->) 10 CMD ZBYTES DRG DRG 98 CMD 10 CMD 0 DRG 0 DRG ; 12 : WAIT ( n-->) SW BEGIN PAUSE 162 CMD 18 CMD (DRG -1 = END 3 13 . RESET RESET 1H 2H 3 RESET ( REPROGRAMS 9513) 14 . TOGGLE ARM? CO IF 196 CMD 0 ARM? C! ELSE 36 CMD 1 ARM? C! 15 THEN 3 234 Block Number: 22 0 ( DATA TAKING BUFFER INITIALIZATION ) 1 CREATE LEN 20 , ( DEFAULT LENGTH IN MINUTES FOR DT INT. CALC.) 2 .DDISK .” DFLTDAT.FTH " 3 3 4 VARIABLE ’BUF VARIABLE CPNTR 5 CREATE FFLAG 1 C, CREATE ’ZSTOP 1 C, CREATE NOWBLK 1 , 6 ' DDISE IISPEC 25 BLANK [’l .DDISH DUP 7 3 + SWAP 2+ CO IISPEC SWAP MOVE 3 8 9 . INITBUFF FLUSH EB DDISK BUFFER 'BUF 2 10 1 IDENTIFY UPDATE 0 ’3STOP C2 11 2 NOWBLK 2 4 CPNTR 2 -1 ’BUF O 2 ’REST O ’BUF O 2+ 2 ; 12 : RATE? LEN O 60 O 100 / DUP 10 ( IF DROP 10 THEN 'REST 2 ; 13 : LENGTH ." Chromatogran length in minutes?” 3 ACCEPT CR DUP l4 LEN 2 RATE? ’REST O ." DT Int.= ” . .” hs" CR ; 15 RATE? Block Nusber: 23 ( BACKGROUND TASKS FOR DATA TRANSFER ) : DOZBACE 2BACK ACTIVATE BEGIN STOP 0 ‘ZSTOP C2 0 FFLAG C2 BUFFER ’BUF 2 NOWBLK O IDENTIFY 4 CPNTR 2 l FFLAG C2 FLUSH UPDATE 1 NOWBLK +2 1 ’ZSTOP C2 0 END ; . NXTPNT ’BUF O CPNTR O + 2 2 CPNTR +2 CPNTR O ‘1024 = IF BEGIN PAUSE ’ZSTOP CO END 2BACK RESUME BEGIN PAUSE FFLAG CO END -1 'BUF O 2 PAUSE ’REST O ’BUF O 2+ 2 PAUSE THEN ; , 10 : DOSBACK 3BACE ACTIVATE BEGIN (TFI) ODAT NXTPNT ll NXTPNT ’REST O REST ’SSTOP CO END STOP 3 OQQQGhNNO-‘O 13 3 2PARAM -1 NXTPNT 'REST O NXTPNT 3 l4 : TAKEDATA INITBUFF DO3BACH DOZBACE 3 Block Number: 24 ( OOOPEAE FINDEROOO RFLAG=1,REPORT READY; lDATA=BASELINE3 ) ( 2DATA=TOP OF PEAK: 1TIME=START OF PEAK; 2TIME=TOP PEAK) VARIABLE lDATA VARIABLE 2DATA VARIABLE 3DATA VARIABLE ’TIME VARIABLE 1TIME VARIABLE 2TIME VARIABLE 3TIME CVARIABLE RFLAG CVARIABLE 1THRESH CVARIABLE 2THRESH CVARIABLE ’4STOP MMARK MARK CO SWAP 2MARK 100 0 DO .MARE LOOP 2MARK ; [?PEAE] BEGIN ODAT TFI 1TIME 2 lDATA 2 BEGIN PAUSE ODAT TFI 2TIME 2 DUP 2DATA 2 lDATA O - 1THRESH CO > DUP NOT IF 2DATA O lDATA 2 2TIME O 1TIME 2 THEN ’TIME O TIMEOUT 10 END 5 MMARK BEGIN PAUSE ODAT TFI 3TIME 2 DUP 3DATA 2 2DATA O 11 SWAP - 2THRESH CO > DUP NOT IF 2DATA O SDATA O - 0( IF 12 3DATA O 2DATA 2 3TIME O 2TIME 2 THEN l3 THEN END 1 RFLAG C2 BEGIN PAUSE RFLAG CO NOT END ’4STOP CO 14 END ; 15 : ?PEAK 0 ’4STOP C2 4BACE ACTIVATE [?PEAK] STOP ; DQQQOIDNNHO 235 Block Number: 25 ( PEAK FINDER AUXILIARY ROUTINES, OPEAKS: CURRENT PEAK COUNT) GOQGGéQNr-O 2THRESH CLR .” BASELINE THRESHOLD (A/D UNITS)?" 4 ACCEPT 1THRESH C! CH .” PEAK THRESHOLD (A/D UNITS)?" 4 ACCEPT 2THRESH C2 ; CREATE OPEAKS 0 , 20 1THRESH C2 40 2THRESH C2 0 RFLAG C2 50 ’TIME 2 REPORT 27 EMIT 107 EMIT 1 OPEAKS +2 .." PEAK START3" 1TIME O PRNMIN ." MAX3" 2TIME O PRNMIN ." HEIGHT (A/D UNITS):" 2DATA O lDATA O - . lDATA O .” BASE:" .” COUNT:" OPEAKS O . 0 RFLAG C! CH 27 EMIT 106 EMIT 3 INJECT INJECT 0 OPEAKS 2 ; : SYNC SYNC 0 OPEAKS 2 ; 4STOP 1 ’4STOP C2 ; SETTIME CLR .” 100ths of seconds delay?” 5 ACCEPT ’TIME 2 ; IOPEAKS 0 OPEAKS 2 ; IOPEAKS 1 OPEAKS +2 ; : DOPEAKS -1 OPEAKS +2 ; Number: 26 SYSTEM STATUS REPORT FOR LC) RD .' RECORDER DRIVER: " ’ISTOP CO IF ." OFF" ELSE ." ON” THEN CR ." Offset (A/D units): " OFSET O . CR ." Range (A/D units ): " RANGE O . CR .” Zero (8 neg. volts allowed): ' 2048 DUP OFSET O - 100 ROT O/MOD SWAP IF 1+ THEN . ." 3” CR 3 PF .” PEAK FINDER3” ’4STOP CO IF ." OFF" ELSE ." ON” THEN CR ." Base thresh.(A/D units):” 1THRESH CO . CR .” Peak thresh.(A/D units):" 2THRESH CO . CR ." Sampling int erva1(100ths of secs.):" 'TIME O . CR ; DAC ODAT CONDAC ." DAC O1 :" . ." " ; A/D R12 16 /MOD SWAP DROP ." A/D :” . CR ; FLOWST .” Flow status: RST75 READ 16 AND IF .” Ready” ELSE .” Not ready" THEN CR ; VALVE DUP ." Valve ” . ." :” CPORT CO 8 HOT 0 00 2O LOOP AND IF .” CCW” ELSE .” CW" THEN ." ” 3 Block Number: 27 ( MORE SYSTEM STATUS WORDS) OCQQGhQND-IO HEAD .” OOO S Y S T E M S T A T U s it!” CR CR ; MM ." Marker length (DAC units):” MARK CO . CR ; IN ." Injector position:" INJST IF .” Inject" ELSE ." Load" THEN CE ; ?OPEAES .” Current peak count:” OPEAKS O . CR CR ; DT ." DATA TAKING:” ’ESTOP CO IF .” OFF” IISPEC CO 68 8 IF .” (SUSPENDED)" THEN ELSE ’ZSTOP CO IF .” ON" ELSE .' FLUSHING” THEN THEN CR .” Blockz" NOWBLK O l- . CR 3 DTI ." D.T. Intervalz" 'REST O . ." he” CR ; STS 27 EMIT 72 EMIT HEAD RD MM CR PF ?OPEAES DT DTI CR DAC A/D l VALVE 2 VALVE CR FLOWST IN 3 ?SYS CLR SYS ; VIEW ’SSTOP CO CLR IF .” Block?" 3 ACCEPT CR LIST ELSE .” Can’t view a block while taking date." CR THEN ; 236 Block Nunber: 28 ( DATA FILE RENAMING ROUTINE) OQQmUéle-‘O CREATE DSPEC 25 ALLOT RNAME CR .” Enter file nane (no .ext)” DSPEC 25 ZDUP BLANK EXPECT ; UPBL UP” ” ; UPNUM (- ) UPCHAR UPBL 3 RSX UP” .RSX/R ” ; UPDDISE [’l .DDISE 2 + COUNT UPCHAR 3 RENAME DSPEC CO IF UPCR WPRMPT COMMAND- STR UPDDISK SWITCHES 1 UPNUM NOWBLK O 1- UPNUM DSPEC 25 -TRAILING UPCHAR RSX UPCR THEN 11RESET 3 Number: 29 MORE DATA TAKING ROUTINES) SUSPEND 1 ’3STOP C2 ; : MOREDATA 2PARAM 0 ’3STOP C2 DO3BACK 3 ENDDATA 1 ’3STOP C2 -1 NXTPNT -1 NXTPNT 1022 CPNTR 2 -1 NXTPNT BEGIN PAUSE FFLAG CO END FLUSH RNAME RENAME 3 NEWREST ( n-—>) CLR DEL O ." Enter delay for data taking (” . ." units per second)" 5 ACCEPT ’SSTOP CO 0: IF SUSPEND ’REST 2 MOREDATA ELSE ’REST 2 THEN 3 Nunber: 30 EVENT CONTROL BASED ON A/D CHANGE) +A/D ( u -- ) .” Event will occur at ” DUP . ." A/D units" CR BEGIN PAUSE DUP ODAT - 0< END DROP 3 -A/D ( u -- ) .' Event will occur at " DUP . ." A/D units” CR BEGIN PAUSE DUP ODAT SWAP - 0< END DROP ; A/DCHG ( n -' ) DUP ODAT + SWAP 0< IF -A/D ELSE +A/D THEN 3 +SFS ( u -- ) RANGE O 100 O/ OFSET O + +A/D 3 -EFS ( u -- ) RANGE O 100 O/ OFSET O + -A/D ; SFSCHG ( n -- ) RANGE O 100 O/ A/DCHG ; 237 Block Number: 31 0 ( EVENT’ S VECTORED EXECUTION TABLE) 1 VARIABLE VTAB 50 ALLOT 2 : NOACTION CLR .” NO SUCH COMMAND" ; 3 ° INITI 26 0 DO I 2O VTAB + [' ] NOACTION SWAP 2 LOOP ; INITI 4 : 2VTAB ' SWAP 2O VTAB + 2 5 ( R) 17 2VTAB DIGREC ( S) 18 2VTAB ?SYS 6 ( O) 14 2VTAB lSTOP ( U) 20 2VTAB SUSPEND 7 ( C) 2 2VTAB CLR ( B) 1 2VTAB TAKEDATA 8 ( A) 0 2VTAB ABORTRUN ( M) 12 2VTAB MOREDATA 9 ( H) 7 2VTAB EHELP ( E) 4 2VTAB ENDDATA 10 ( Z) 25 2VTAB ZERO ( N) 13 2VTAB NEWREST 11 ( L) 11 :vran LMARE ( r) 5 2VTAB nsrsans 12 ( T) 19 2VTAB 2THRESH ( I) 8 2VTAB IOPEAKS 13 ( P) 15 2VTAB ?PEAK ( E) 10 2VTAB ZOPEAKS l4 ( O) 16 2VTAB 4STOP ( V) 21 2VTAB VIEW l5 ( D) 3 2VTAB SETTIME ( X) 23 2VTAB TOGGLE Block Number: 32 0 ( EVENT CONTROLLER FRONT-END ROUTINE) 1 : ?COM KEY DUP 65 91 WITHIN NOT IF DROP 72 THEN 2 65 - 2O VTAB + O EXECUTE 3 3 CVARIABLE EMARK 20 EMARK C2 4 : EVENT 27 EMIT 106 EMIT SWAP 60 a + EMARK CO MMARK 5 BEGIN ?TERMINAL IF ?COM 6 THEN RFLAG CO IF REPORT THEN 7 DUP 71 4 POS PRNMIN DUP TFI DUP BOX 71 2 POS PRNMIN 1 + 8 < END 27 EMIT 107 EMIT .” Event O ” PRNMIN ." cossencing” 9 CR ( sin.sec--> ) 3 10 11 CLOCK 108 CMD 3 12 HELP CLR llSPEC CO 77 = IF 33 LIST 13 ELSE .” Help unavailable while taking data" THEN ; HELP 14 ’ ?CREATE ’CREATE 2 ( TURN ON NOT UNIQUE MESSAGE) 15 Block Number: 33 0 WORDS AVAILABLE FOR HPLC AND VALVE CONTROL 1 PRESS < k -- > Presses one button wth constant k. 2 ENTER < kl.k2,...kE -- > Presses given keys, then ENTER key. 3 ENTERO ( n or d -- > Converts number to key codes and ENTERs. 4 INJECT ( -- > Waits for load/injection, then starts nicro clock. 5 SYNC ( -- > Syncs w/SP8700 by pressing RUGR A cueing for inject. 6 CCW or CW ( n -- > Turns valve n in spec. direction. 7 EVENT < m,s -- > Delays till sin,sec after injection. 8 WAIT ( hs -- > Delays for he hundreths of seconds. 9 +SFS or -SFS ( n -- > Delays till DAC is n3 of full scale 10 SFSCHG ( n -- > Delays till DAC changes by nS of full scale 11 CAM ( -- ) Waits till cam marker is sensed. 12 FLOW? ( -- > Waits till ”flow ready” is sensed. 13 HELP ( -- ) Lists this help frase on the tersinal. 14 CLOCK ( -- ) Resets micro clock to zero. 15 See docusentation and SCR 30 for sample run. 238 Bloc DQQQGQQNHO Bloc DmdmuéuNo-oo k Nunber: 34 EVENT CONTROLLER COMMANDS A -- Abort run -- Clear screen -- End date taking -- Lists this info -- Kill peak out (set to 0) -- Take sore data -- New d.t. interval -- Turns off recorder driver Turns on peak finder i -- Turns off peak finder -- Turns on recorder driver 3 -- Displays system status -- Sets p.f. threshold values -- Begin taking data -- Sets delay for peak finder False peak.decrement count -- Increnent peak count -- Sets length of sin. sarker r'H'IIDw I I AW “GHWNO'UOZZHINO I I -- Suspend d.t. temporarily V —- View (list) a block ‘ -- Stop/Restart clock 2 -- Sets recorder zero I Nusber: 35 . savs s RECALL A/D vanuzs) 3 CREATE A/DVALUES 20 ALLOT INITIL 10 0 DO 2048 A/DVALUES I 23 + 2 LOOP 3 INITIL VO? 0 9 WITHIN NOT ABORT” Save/recall error -- invalid erg” ; SAVE ( n -- ) DUP VO? ODAT SWAP 2O A/DVALUES + 2 ; RECALL ( n -- d) DUP VO? 2: A/DVALUES + O ; ( START OF REGRAPHING UTILITY) : WFPIP BEGIN GETCHR DUP 93 = END DEPTH DEPTH REVERSE DROP DROP DEPTH 1- 0 DO EMIT LOOP CR 70 ) ABORT” ABORTED -- FPIP ERROR” ; k Number: 36 ( REGRAPHING UTILITY) VARIABLE ADRNG VARIABLE ADOFF . INITDAT FLUSH RNAME UPCR WPRMPT COMMAND-STE DSPEC 25 -TRAILING UPCHAR RSX UPBL UPDDISK SWITCHES l UPNUM UPCR WFPIP EB DDISK 3 OFF CPORT CO 16 / 4 OR CPORT C2 ; ON CPORT CO 16 / 11 AND CPORT C2 ; GETIT OFF NOWBLK O BLOCK ’BUF 2 ON ; : PARAMS? ." MINIMUM A/D VALUE TO GRAPH?" 5 ACCEPT CR DUP ADOFF 2 .” MAXIMUM A/D VALUE TO GRAPH?" 5 ACCEPT CR SWAP - ADRNG 2 3 >DAC ADOFF O - DUP 0< IF DROP 0 THEN 256 ADRNG O O/ DUP 255 > IF DROP 255 THEN ; GETFIRST 1 NOWBLK 2 1 BLOCK ’BUF 2 0 CPNTR 2 ; . NET ’BUF O CPNTR O + O 2 CPNTR +2 CPNTR O 1024 = IF 0 CPNTR 2 l NOWBLK +2 TOGGLE GETIT TOGGLE THEN 3 239 Block Number: 37 COQQODOIIBOONHO ( REGRAPH UTILITY FRONT END STUFF) ( 9513 SPEED UP ) 9513SU 11 CMD 10 >DRG 2 CMD 0 DRG 14 DRG 3 W BEGIN DUP 1- (TFI) < END DROP 3 REGRAPH OFF 0 2lDAC INITDAT PARAMS? GETFIRST ON 9513SU CLOCE BEGIN NXT DUP 0< IF DROP NET DUP 0( IF EB 11RESET 7 EMIT 0 2lDAC RESET OFF ABORT" (EOF)" THEN DROP NXT THEN NET SWAP W )DAC 2lDAC 0 END 3 Nunber: 38 SLOW VALVE TURNING ROUTINE VALVE O -- ) STURN CPORT C2 25000 0 DO LOOP CPORT CO DUP 16 / SWAP 15 AND - ABORT" VALVE FAILURE -- RUN ABORTED” 3 SELECT HOT 0 DO 2 O SWAP DUP ROT + LOOP 2 / 3 MAEEOUT SELECT CPORT CO 16 / CR 3 SCCW DUP 0 l MAEEOUT 0R STURN DROP ” Valve O” . .” switched CCW" CR ; SCW DUP 1 254 MAEEOUT AND STURN DROP .” Valve O" . .” switched CW" CR ; 240 Block Number: 61 comqmuauNp—o ( PHASE SHIFTED FREQ GENERS. FOR IDG -- BLOCE 1/3) ( TO USE: BIM DIV..DELAY SC.. LOW PULSE. HI PULSE GO) ( UNITS ARE MICROSECONDS. DO SETEMUP ONCE AT START) VARIABLE BIMDIV VARIABLE DELSC VARIABLE LOPUL VARIABLE HIPUL ' DTFC 10 O 16960 15 HOT M/ 2/ 3 . NF BIMDIV O DTFC CR CR .” NEW FREQ: " 10 O U. CR 3 HEX : SETEMUP 17 CMD 5000 >DRG 1 CMD 322 >DRG 2 CMD B02 >DRG 3 CMD 8B62 >DRG 3 LOADEMUP ( C1,C2,C3L,CBH--> ) 13 CMD >DRG 0B CMD >DRG 0A CMD >DRG 9 CMD >DRG 3 : STARTEMUP C7 CMD 47 CMD E1 CMD E2 CMD 83 CMD 27 CMD 3 : GO LOADEMUP STARTEMUP 3 DECIMAL SETEMUP ° CHEFRQ DUP 5000 > IF CR .” OOO FREQ TOO HIGH OOO" 7 EMIT DROP 5000 CR THEN 3 Block Number: 62 ( PHASE SHIFTED FREQ.GENERATORS FOR IDG -- BLOCE 2/3) : 2VAR ( N,N,N,N--> ) HIPUL 2 LOPUL 2 DELSC 2 BIMDIV 2 3 : .VAR 13 EMIT HIPUL O LOPUL O DELSC O BIMDIV O ." BIMORPH DIVIDER=" U. .” DELAY SCALAR!” U. .” LOW PULSE =” U. .” HIGH PULSE =” U. 3 : CALDIV BIMDIV O 2O DELSC 2 2DUP 2O 1+ BIMDIV O O LOPUL 2 2O 1- BIMDIV O O HIPUL 2 3 STARTIDG BIMDIV O DELSC O LOPUL O HIPUL O GO 3 RUNIDG STARTIDG BEGIN HEY DUP 61 = IF LOPUL O 2 > IF 1 HIPUL +2 -1 LOPUL +2 THEN THEN DUP 45 = IF HIPUL O 2 > IF 1 LOPUL +2 -1 HIPUL +2 THEN THEN DUP 46 = IF 1 DELSC +2 THEN DUP 44 = IF DELSC O 2 > IF -1 DELSC +2 THEN THEN DUP 72 = IF BIMDIV O DUP 10 > IF 1- BIMDIV 2 HOT HOT CALDIV ROT NF ELSE DROP THEN THEN DUP 76 = IF 1 BIMDIV +2 HOT HOT CALDIV ROT NF THEN STARTIDG .VAR 83 = END 2DROP 3 Block Nunber: 63 DDQQGIDOONO-‘O PHASE SHIFTED FREQ GENERS. FOR IDG -- BLOCE 3/3) TO USE: OHI DROPS , OLO DROPS , FREQ TO 2 DECIMAL PLACES) THEN IDG -- ,( AND .) KEYS ARE PHASE SHIFTERS) -_ AND 3+ ADJUSTS DUTY CYCLE -- S FOR STOP -- OTHERS RESTART) H AND L ADJUST FREQ. AND RESET DEFAULTS) AAAAA : FTDC DROP CEEFRO DUP DUP 16960 15 HOT M/ 2/ 10 / DUP BIMDIV 2 DTFC BIMDIV O 1+ DTFC 2DUP CR ." LO FREQ:” 10 O U. ." HI FREQ3” 10 O U. SWAP ROT - HOT HOT - > IF 1 BIMDIV +2 .” USED LO FREQ" ELSE ." USED HI FREQ" THEN .” :DIV=" BIMDIV O . CR 3 IDG ( OHI,OLO,FF.FF) FTDC CALDIV RUNIDG 3 241 Block Nusber: 64 QQQmeNNI-‘O ( FREQ. RAMP FOR DROP GEN. ) ADJ BEGIN KEY DUP 46 = IF -1 BIMDIV +2 THEN DUP 44 = IF 1 BIMDIV +2 THEN STARTIDG .VAR 83 = END ; RAMPIDG IDG BEGIN BIMDIV O 1+ BIMDIV 2 STARTIDG .VAR 1 SEC ?TERMINAL IF ADJ THEN 0 END 3 Block Number: 65 @OQQGDQNHO ( ONE-SHOT DROPLET PRODUCTION) ( USE IDG FIRST TO SET PARAMETERS FOR BIMORPH FREQ., ) ( PHASING AND DUTY CYCLE. THEN SAY n ONE-SHOT, WHERE n IS) ( A DELAY IN HUNDRETHS OF SECONDS) HEX . 3PROG 3 CMD 8B42 >DRG 3 DECIMAL ONE-SHOT 3PROG 0 DO 142 0 DO LOOP LOOP STARTIDG .VAR CR ." Hit a key to exit" KEY DROP SETEMUP 3 Block Number: 66 madame-coupe ( CONFIGURE COUNTERS 4 O 5 FOR DATA TAEING/IDG) 16384 >DRG ( MM REG BY PREVIOUS RESET) 4 CMD 3328 >DRG ( 4 MR DOWN. ONCE, BINARY. F3) 5 CMD 3880 >DRG ( 5 MR UP, ONCE, BINARY, F5) 13 CMD 0 >DRG CREATE DEL 3906 , CREATE ’REST 100 , . SREST ( N-->) 12 CMD ( 4 LD) >DRG 72 CMD 12 CMD 0 >DRG 40 CMD 3 REST ( N--)) SREST BEGIN PAUSE 168 CMD 20 CMD (DRG -1 = END STARTTFI 112 CMD 3 (TFI) ( --> N) 176 CMD 21 CMD (DRG 3 CLR 27 EMIT 69 EMIT 27 EMIT 106 EMIT ( CLEAR SCREEN ) 3 242 Block Number: 67 DflfimubNNr-‘O ( GREASED LIGHTING DATA TAEER) 18 LOAD CREATE DLY 1 , VARIABLE START CREATE UPLIM 30236 , ( TOP OF AVAIL. RAM) INGLDT HERE DUP START 2 UPLIM O HERE - 2/ .” O PTS=” . CR UPLIM O H 2 ; GLDT UPLIM O SWAP DO 812 DLY O 0 DO LOOP R12 I 2 2 /LOOP 3 2 ( LOOK-SEE HERE START O DO I O 16 /MOD SWAP DROP I I 23 /M00 DROP 0= IF KEY DROP THEN) 2 /LOOP 3 LOOK-SEE LETS YOU VIEW DATA WITHOUT STORING ON 11/23) Block Nusber: 68 DQQGUMFNNHO ( DATA TRANSFER FOR GLDT) VARIABLE ’BUF VARIABLE NOWBLK VARIABLE CPNTR é .DDISE ." DFLTDAT.FTH " 3 DDISK llSPEC 25 BLANK [’l .DDISE DUP + SWAP 2+ CO llSPEC SWAP MOVE 3 INITBUFF FLUSH EB DDISK BUFFER ’BUF 2 1 IDENTIFY UPDATE NOWBLK 2 6 CPNTR 2 -1 ’BUF O 2 0 ’BUF O 2 + 2 DLY O ’BUF O 4 + 2 ; IF NXTPNT ’BUF O CPNTR O + 2 2 CPNTR +2 CPNTR O 1024 = .” NOW FLUSHING BLE O" NOWBLK O 1 - . CR FLUSH BUFFER ’BUF 2 NOWBLK O IDENTIFY UPDATE 1 NOWBLK +2 0 CPNTR 2 THEN 3 28 LOAD Block Nusber: 69 DQQQGDNNHO ( AAAAA DATA TRANSFER FOR GLDT) STORE INITBUFF HERE START O DO I O 16 /MOD SWAP DROP NXTPNT 2 /LOOP -1 NXTPNT -l NXTPNT NOW FLUSHING LAST BLOCK” CR FLUSH EB RNAME RENAME TART O H 2 ; PSTORE INITBUFF HERE START O DO I O 16 /MOD SWAP DROP DUP . I . CR NXTPNT 2 /LOOP -l NXTPNT -l NXTPNT NOW FLUSHING LAST BLOCK" CR FLUSH EB RNAME RENAME START O H 2 ; TO USE: l.INGLDT initializes memory. 2. Store delay in DLY) 3. GLDT takes data and stores it in aesory. 4. STORE ships) data up to the 11/23. PSTORE prints on terminal and stores) File structure: 12bit Os in 16bit cell, -1 -1 = EOF ) clear menory with FORGET XXX : XXX ; before next run) : XXX ' 243 U g—o O O I" OGQOODNNHO A Nunber: 70 EXAMPLE OF HOW TO CHANGE DROPLET PARAMETERS IN SOFTWARE) TEST-RUN ( OHI, OLO, FF.FF --> ) IDG ( GENERATE INITIAL PARAMETERS AND THEN CONTINUE) ( BY HITTING THE 5 KEY) 20 SEC ( 20 SECOND DELAY IN SOFTWARE) 5 DELSC +2 ( ADD 5 MICROSECONDS ONTO DELAY SCALAR) STARTIDG ( AND RESTART 9513) 150 SEC ( ANOTHER DELAY) 2 LOPUL +2 -2 HIPUL +2 ( ADJUST DUTY CYCLE BY 2 MICROSECONDS) ( MAKING LOW PULSE LARGER AND HIGH PULSE SMALLER) STARTIDG 400 SEC (RE-START AND DELAY) 4 50 12.25 IDG ( GENERATE A WHOLE NEW STREAM WITH 4 HIGH) ( AND 50 LOW DROPS AT 12.25 kHz, CONTINUE BY HITTING S KEY) 100 SEC ( ANOTHER 100 SECOND DELAY) ( MORE ADJUSTMENTS, ETC., ETC., ETC.) 3 244 APPENDIX B An Operator’s Guide for the Multidinensional HPLC APPENDIX B An Operator’s Guide for the Multidinensional HPLC This appendix is meant to serve three purposes. A concise listing of FORTH commands and their arguments is given as a means for easy reference. Detailed explanations of the commands are contained in the ”Software” chapter of this dissertation. Where appropriate, exanples are shown. A brief start-up and connection procedure is also given in the event that the equipment should be dismantled. Finally, since research is never as easy as one expects, a short troubleshooting guide will be presented for those nasty ”unexpected problems.” A. FORTH Commands The following notation is used: A word in all capitals is a FORTH command. The effect of the word on the stack is shown with the notation: . ( before ——> after ) where ”before” represents the arguments for the word, and "after” represents any numbers produced by the word. Single length signed integers are represented by the letter ”n”, double length by the letter ”d”, unsigned single length by the letter "u”, SP8700 key codes by the letter "k", a single 245 byte by the letter "b", and logical flags by the letter "f." Defaults and ranges are designated with the notation: [default,1ower limit:higher linit] A user input is designated by the symbol ”xxx” inmediately following the word. The word, stack effect, default/linits (if any) are given, followed by the action and an example where appropriate. SP8700 Communication PRESS ( k --> ) [ , : ] Sinulates one keystroke for the key with code k. Aborts if not acknowledged. Ex: Press edit/run button. EDRU PRESS ENTER (k1,k2,...kN -—) ) [ , : ] Presses all keys given, in first in first out order. Assunes all numbers on stack are key codes. ’ Aborts if any key is not acknowledged. Ex: Enter a flow of 1.5 mL/min. FLOW O1 O. O5 ENTER ENTERO (n or d with valid punctuation --> ) [ , : ] Converts the top number on the stack to key codes representing that nunber and ENTERs them. Stack should be empty prior to using. Numbers with decimal points are acceptable. Ex: Enter a flow of 1.5 mL/min. FLOW 1.5 ENTERO GETO xxx ( --> k1, k2, ... kN) [ , 0:65535] Waits for up to 246 6 digit input. other series of SP8700 key codes. O>KEYS ( u --) kl,k2, of GETO. loop to key codes. EDITMODE ( --) ) FORTH punctuation. If already in edit node, nuswons ( --> ) [ . already in run mode, FLOW? ( --> ) [ , lit. CAI ( --> ) [ . flashes. ] Places the SP8700 in run mode. has no effect. Valid punctuation RN) [ Useful for converting an is Converts index a decimal point 01‘ the input number to a No error checking is performed. 0:65535] Stack equivalent iteration ] Places the SP8700 in edit mode. has no effect. If ] Delays until the "flow ready" LED is ] Delays until the ”cam marker" LED Table Bl. Hey Code Definitions Key Mnemonic Change Time Clear Entry Continue Display Delete Line Edit File Edit Mode/Run Mode Enter Flow Hold Initialize Max. Pressure Pressure Purge Run Gradient Store File Stop Rey CHTI Test TEST GLEN Time TIME CONT 1 O1 DISP 2 O2 DELI 3 O3 EDFI 4 O4 EDRU 5 O5 ENTE 6 O6 FLOW 7 O7 HOLD 8 O8 INIT 9 O9 MAPR 0 O0 PRES . O. PURG SA SA RUGR SB 88 STFI SC SC STP (down arrow) \/ 247 Mnemonic XSLYG Control CW ( n --> ) [ , 1:2 ] Turns the valve specified in the clockwise direction. If already in the clockwise position, this word has no effect. If not turned in the clockwise direction within the tine-out period, the word aborts. Ex: Turn valve 1 in the clockwise direction. 1 CW CCW (n --> ) [ , 1:2 ] Same as CW, but turns valve n in the counterclockwise direction. VALVE ( --> ) [ , 1:2 ] Reports the valve position. INJST ( --> f ) [ , : ] Returns a 1 when the injector valve is in the inject position, a 0 for the load position. IN ( —-> ) [ , : ] Reports the injector valve position. SETUP ( --> ) [ , : ] Re-programs the 8255 PIO to the state after the software was first loaded. Sets valves to CCW position. Tiling Words CLOCK ( --> ) [ , : ] Starts or re-starts micro clock. TOGGLE ( --> ) [ , : ] Turns clock on/off without resetting. INJECT ( --> ) [ , : ] Starts micro clock when sample is injected. SYNC ( --> ) [ , : ] Starts micro clock and synchronizes with SP8700 clock by cueing for injection and PRESSing the run gradient button at the proper point in the cam cycle. 248 TFI ( --> n ) [ , : ] Returns the number of seconds since micro clock was started. (TFI) ( --> n ) [ , : ] Returns the nunber of tenths of seconds since micro clock was started. RESET ( --> ) [ , : ] Re-prograns AM9513 to the state after software is first loaded. WAIT ( n --> ) [ , 1:65535] Delays for n hundreths of seconds. Uses AM9513. EVENT ( nl,n2 --> ) [ , 1:65535] Delays until micro clock reads n1 minutes and n2 seconds. Allows access to commands listed below while waiting by hitting a key once to get the event controllers attention and a second time to issue the command. If the clock is not updating, the event controller is probably waiting for a comnand. Displays current time and time of next event in upper right corner of screen. Prints report from peakfinder when needed. Ex: Delay until 5 minutes and 30 seconds from injection, turn valve 2 CW and turn it back to CCW 7 ninutes and 45 seconds from injection. INJECT 5 30 EVENT 2 CW 7 45 EVENT 2 CCW 249 Table B2. Event Controller Commands H 0 ’< Wong; Function ABORTRUN Abort run, return control to term. handler TAKEDATA Begin taking data. CLR Clear screen. SETTIME Set delay for peak finder. ENDDATA End a data taking run. DOPEAKS False peak, decrenent peak count. EHELP Lists event controller commands. IOPEAKS Increment peak count. ZOPEAKS Kill (zero) peak count. LMARE Sets length of minute marker for recorder. MOREDATA Restarts data taking after a SUSPEND. NEWREST New data taking interval. NXP lSTOP Turns off recorder driver task. ?PEAK Turns on peak finder task. 4STOP Turns off peak finder task. DIGREC Turns on recorder driver task. ?SYS Displays system status. 2THRESH Sets peak finder thresholds. SUSPEND Suspend data taking temporarily. VIEW List a block on the terminal. TOGGLE Stop/Restart the microcomputer clock. ZERO Sets recorder zero for recorder driver task. Rgent Cogtrol Uging_Retector Output +A/D ( u --) ) [ , 0:4094] Delays until the detector output is greater than u ADC units. Ex: Turn valve 2 CW when the detector output exceeds 3400 ADC units. 3400 +A/D 2 CW -A/D ( u --> ) [ , 1:4095] Delays until the detector output is less than u ADC units. A/DCHG ( n --> ) [ , -4094:+4094] Delays until the detector output changes by the specified nunber of ADC units. The direction is controlled by making n positive for an increase and negative for a decrease. Ex: Turn valve 2 CW after a decrease in detector output of 30 ABC units. 250 ‘7"““““‘—‘—‘r—*f-§1 -30 A/DCHG 2 CW +SFS ( u --> ) [ , 0:99] Delays until the detector output exceeds u percent of full scale on the chart recorder. Ex: Turn valve 2 CW at 558 of recorder full scale. 55 +SFS 2 CW -EFS ( u --> ) [ , 1:100] Delays until the detector output is less than u percent of the chart recorder full scale. SFSCHG ( n --> ) [ , -99:+99] Delays until the detector output changes by n percent of the chart recorder full scale. Direction can be specified as in A/DCHG. SAVE ( n --) ) [ , 0:9] Reads the current detector output and stores the resulting number of ADC units in register n. RECALL ( n --> n2 ) [ , 0:9] Recalls the ADC units which were stored in register n and places them on the stack for use with +A/D or -A/D. Data Acquisition and Storage TAKEDATA ( --> ) [ , : ] Initiates a data taking run. Sanpling rate controlled by the variable ’REST. Sampling delay is performed by AM9513 STC. SUSPEND ( --> ) [ , : ] Tenporarily suspends data taking. Buffers are not flushed. Parameters are not altered. MOREDATA ( --> ) [ , : ] Re-starts data taking after a suspend. Puts a flag in the data file consisting of a -1 followed by the current value of ’REST. 251 ‘2‘}? ' ‘ " ”M.SMJW ENDDATA ( —-> ) [ , : ] Terminates a data taking run. Buffers are flushed. Disk specification is changed back to MICRO.FTH. User is prompted for a file name, and an FPIP command line is sent to transfer the number of blocks acquired from the default data file (DFLTDAT.FTH) which is a FORTH emulator file, to an RSX-compatible file (’usernane’.RSX). ’REST ( variable ) [12,l:65535] Data taking delay in 100ths of seconds. The value returned by this variable is used by the word REST, which does the delay. Although values as small as 1 can be used, duplicate times will be obtained with values of less than 10, since (TFI) returns times in tenths of seconds. LEN ( variable ) [20,1:32767] Expected chromatogram length in minutes. Used by RATE? to set ’REST. RATE? ( n--) ) [ , 1:32767] Takes an argument in minutes and uses it to set ’REST to a rate which will use 40 blocks of data in the time specified. This is 808 of the default file space, and represents about 10,200 points. LENGTH ( --) ) [ , 1:999] Prompts for expected chromatogram length and then does RATE?. NOWBLK ( variable ) [ , l: ] Returns the current block of data being filled. CPNTR ( variable ) [ , 1:1024] Returns the current offset into the data buffer. Represents the number of data points taken times 4. 252 D M'a‘bfln ODAT ( --> n ) [ , : ] Puts one datum on the stack. RUN ( --> ) [ , : ] Makes a running display of the current ADC output on the terninal. Peak Finder ?PEAK ( --> ) [ , : ] Activates the peak finding task. Makes a ssall (5 ADC units) positive spike when the start of a peak is sensed. 4STOP ( --> ) [ , : ] Deactivates peak finding task. OPEAKS ( variable ) [ , 0:32767] Returns the nusber of peaks sensed since injection or since the variable was last reset to zero. Can be used to trigger an event. Ex: Turn valve 2 counter-clockwise 0.2 seconds after the 25th peak is sensed: BEGIN PAUSE OPEAKS O 25 = END 20 WAIT 2 CCW In the above example, ,the word PAUSE is used to allow background tasks to execute while the BEGIN-END loop is executing. ZOPEAKS ( --> ) [ , : ] Sets OPEAKS to zero. IOPEAKS ( -—> ) [ , : ] Increnents OPEAKS by l. DOPEAKS ( --> ) [ , : ] Decrements OPEAKS by l. 1THRESH ( variable ) [20,1:32767] Baseline threshold used for sensing the start of a peak. Smaller values are more sensitive (ADC units). 2THRESH ( variable ) [40,l:32767] Peak maximum threshold used for sensing the top of the peak. Smaller values are 253 .39.; _ more sensitive (ADC units). 2THRESH ( --> ) [ , : ] Prompts for values of 1THRESH and 2THRESH. ’TIME ( variable ) [50,1:32767] Sampling delay in 100ths of seconds for 1THRESH and 2THRESH. SETTIME ( --> ) [ , : ] Prompts for a value of ’TIME. Recorder Cgptrol DIGREC ( --> ) [ , : ] Activates the digital recorder driver. ISTOP ( --> ) [ , : ] Deactivates the recorder driver. OFSET ( variable ) [1638,0:4095] Number of ADC units subtracted before the conversion to DAC units is made. Set with the word ZERO. RANGE ( variable ) [2458,0z4096] Range of ADC units converted to DAC units. Set with the word ZERO. ZERO ( --> ) [20,0:100] Prompts for the percentage of the negative range of the ADC to be displayed on the recorder. Normally the detector signal is slightly less than zero volts, so the default is 203. MARK ( variable ) [10,0:4095] Contains the length of the minute mark used in ADC units. LMARE ( --> ) [ , : ] Prompts for the length of the minute marker. EMARK ( variable ) [20,0:4095] Contains the length (in ADC units) used for marking the initiation of the word EVENT. 254 fi MMARK ( n -~> ) [ , 0:255] Makes a positive mark n DAC units on the recorder output. If the mark would cause the DAC to exceed its upper limit, the mark is made in the negative direction. Useful for marking events. ON ( --> ) [ , : ] Turns the recorder chart drive on by opening a nechanical relay in the micro. See wiring description below. OFF ( --> ) [ , : ] Turns the recorder chart drive off by closing the mechanical relay. REGRAPH ( --) ) [ , : ] Prompts for a ’.RSX’ file to be plotted on the chart recorder. The vertical axis is set by inputting the minimum and maximum ADC values to graph. If this is not known, the entire range should be plotted first by responding with 0 for the mininum and 4095 for the maximum. Scale expansion is achieved by inputting larger mini-a and smaller maxina. Scale reduction is achieved by inputting 0 for the minima and a value larger than 4095 for the maxima (but less than 32767). The horizontal scale is set by changing the chart drive speed. Miscellaneous Non-Standard FORTH Worgg HELP ( --> ) [ , : ] Displays MHPLC help screen. EHELP ( --> ) [ , : ] Displays MHPLC event controller help screen. ?DISK ( -—> ) [ , : ] Prints current disk filespec at terminal. 255 11RESET ( --> ) [ , : ] Sets disk filespec to MICRO.FTH. SHOW ( nl,n2 --> ) [ , : ] Useful for listing a range of blocks on the terminal. Argunents are beginning block and ending block. Control Z terminates listing, carriage return lists next block without delay. REVERSE ( n --) ) [ , 0:255] Reverses the order of the top n items on the stack. A 0 argunent reverses the entire stack. NOBRO ( --) ) [ , : ] Sets the terninal line to a ”no-broadcast" state, so anyone trying to broadcast a message to the terminal will not mess up micro/ll-23 comnunication. BRO ( --> ) [ , : ] Allows terminal to accept broadcast TALK ( --> ) [ , : ] Allows direct communication with 11/23 as if micro were not there. Terminated with the ”BREAK” key. Should be used once before downloading to initialize USARTS. Also does a BRO. GETCHR ( --> ) [ , : ] Gets one character fro. the 11/23 USART and clears the USART to receive the next character. FREE ( --) ) [ , : ] Displays the nusber of unused bytes of menory. Additional Editor Commands HT ( --> ) [ , : ] Displays the entire block and shows the current cursor position. VV ( n --> ) [0,0:15] Similar to TECO verify command. 0 VV 256 or just VV types the current line and shows cursor position. DD ( n -—> ) [1,1:64] Similar to TECO delete command. Accepts negative arguments. CC ( n --> ) [1,1:?] Sisilar to TECO character move command. Moves the cursor n characters. Accepts negative arguments. LL ( n --> ) [l,l:l5] Similar to TECO line move command. Moves the cursor n lines. Accepts negative arguments. -_.1_ !~ JP ( n --> ) [0,0:1024] Similar to TECO jump command. Moves to the absolute cursor position specified. _._. ...—._ ——.. ' I v TF ( --> ) [ , : ] Types the contents of the "find” buffer. _: TI ( --> ) [ , : ] Types the contents of the "insert” b buffer. K ( --> ) [ , : ] Switches the contents of the find and insert buffers. CUT xxx Cuts up to and including the specified text into the insert buffer. B. Startup Procedure The MHPLC consists of the following conponents: 1. SP8700 solvent delivery system 2. Chronatronix 220 absorbance detector 3. 8085-based microcomputer 4. Heath H-19 terninal 5. Heath EU-205-11 strip chart recorder 6 Two 6-port, 2-position Rheodyne valves with actuators 257 7. Two valve control modules 8. One 5-volt power supply 9. Nitrogen or air tank with regulator The keypad enulator interface has been installed inside the SP8700. Connections are made to the modules as follows: 1. Connect the PIO port on the micro to the keypad emulator on the back of the SP8700. A grey cable with D-connectors on both ends is used for this purpose. 2. Connect pin 17 (bit 0 port C) of the nicro D-connector to valve controller 1 "TTL.” Connect the ”+5” input of the valve controller to the +5 V power supply. 3. Connect pin 21 (bit 4 port 0) of the nicro D-connector to valve 1 sensing switch. Connect the other side of the switch to the +5 V power supply. 4. Connect pin 18 (bit 1 port 0) of the micro D-connector to valve controller 2 "TTL." Connect the ”+5” input of the valve controller to the +5 V power supply. 5. Connect pin 22 (bit 5 port C) of the micro D-connector to valve 2 sensing switch. Connect the other side of the switch to the +5 V power supply. 6. Connect pin 24 (bit 7 port 0) of the nicro D-connector to the injector valve sensing switch. Connect the other side of the switch to the +5 V power supply. 7. Connect pin 25 (cam narker) of the SP8700 D-connector to 258 .- .3“ CW2 A” I’m—-9- the banana plug on the micro labeled ”RST 6.5.” 8. Connect pin 24 (flow ready) of the SP8700 D—connector to the banana plug on the sicro labeled ”RST 5.5." 9. Put detector output switch labeled "RECORDER SPAN” on the 1 mA setting. Short the positive detector output lead to the detector return line using a 2.5 kn resistor. Connect the positive lead to the ADC positive input on the micro. Let the return lead float. Connect the case ground (thin grey wire) of the detector to the analog ground on the nicro. 10. Connect one input of the micro relay to the chart recorder case ground. Connect the other side of the relay to the chart recorder auxiliary connector number 4, labeled ”remote chart control.” 11. Connect the DAC ground to the recorder ground. Connect the positive DAC output (green jack) to the positive recorder input (labeled 1V/10 inches). 12. Connect the +5 V power supply ground to the nicro digital ground (the DAC ground is suitable). 259 1... “41.-Oi... LIE-TAMfiw For reference purposes the SP8700 D—connector pinout is as follows: 1. Disable 74150 2. Select 74150 D Select 74150 C Select 74150 8 Select 74150 A Select 74155 B Select 74155 A QQQOIDOD Acknowledge pulse (from monostable) 12. Flow ready state 13. Cam marker state 25. Ground Pins not listed are not connected. The nicroconputer D-connector pinout is: 1-7: Bits 0-6, port A 8: RST 7.5 9-16: Bits 0-7, port B 17-24: Bits 0-7, port 0 25: Digital ground When the above connections have been nade the software is loaded as follows: 1. Turn on the micro and terminal. If an "ok" does not appear on the terninal, hit a carriage return. If an ”ok” still doesn’t appear nake sure the communication switch on the front of the micro is in the "micro” position. 260 2. Type TALK and a carriage return. 3. The 11/23 should respond with a ">" pronpt. 4. Log into the HPLC account. See system manager for password. 5. After login messages have stopped, hit the break key. The terminal should beep and a carriage return should produce an "ok" prompt. 6. Type ”9 LOAD” and return. The middle green light on the micro should flash, indicating a block is being loaded. 7. When the ”0k” prospt is displayed (approximately 15 seconds) type ”LC” and return. 8. The MHPLC software should be fully loaded in about 4 minutes. Block nusbers should appear on the terminal as they are processed. C. Troggleghootigg Most problems with the MHPLC have very simple solutions. Listed below are a few conmon symptoms which represent problems which have occured in the past. Most are caused by simple forgetfulness, and are easily corrected. During the last year of operation, the instrument was very dependable, and with a little care should stay that way. §P8700 Cosmunication Symptom: Microcomputer responds to communication efforts with ”xx -- Not Acknowledged.” 261 Causes/Corrections: 1. Toggle switch on back of SP8700 in disable position. Switch should be up for communication. 2. SP8700 refused to accept command. Make sure key codes are correct and in the proper sequence. Check manual for correct parameter entry procedure. The command ”EDRU PRESS” should always be acknowledged. 3. PIO may have been reprogram-ed. SETUP will restore the chip to the correct state. 4. Cable discontinuous. Make sure plugs are secure. Check for continuity. Valve Control Symptom: Microcomputer responds to valve switch attempt with ”VALVE FAILURE -- RUN ABORTED.” Causes/Corrections: 1. (Valve does not turn.) Five volt power supply not turned on. 2. (Valve does not turn.) PIO may have been accidentally reprogranmed. Issue a SETUP conmand and try again. 3. (Valve does not turn.) Try manual switches. If manual positions work, then either the sensing switch/wire is bad, or the wire to the controller is discontinuous. To test the sensing switch, nanually turn the valve and use ?SYS to see if it reports the correct valve position. 4. (Valve turns, but too slow.) Try increasing nitrogen 262 pressure. If 75 psi does not produce a fast enough valve switch, the valve needs to be loosened. See valve instructions to make turning easier. 5. (Valve turns too slow when using autoinjector.) This is normal. Use the special valve-turning words SCCW and SCW in block 38. §oftwgge Loading Symptom: Blocks cannot be downloaded. Causes/Corrections: 1. USARTS not initialized. TALK first, check for MCR pro-pt, then try again. 2. Not logged in. 3. MICRO.FTH file is locked. Unlock the file with the PIP command: PIP MICRO.FTH/UN The file will lock if FPIP is aborted. 4. (Block will list, but not load) Check for a zero byte in the block. 5. (Block will list, but not load) Bug in the code causes the compiler to enter warp drive. Correct the code. D. Adding a Third Vglve The original MHPLC design made provision for control and sensing of up to 4 valves through PIO port 0. As presently configured, bits 0 and l are used to control 263 valves 1 and 2, and bits 4 and 5 are used for sensing of these valves. Bit 7 is used to sense the position of the injector valve. Since this valve is not automated, however, bit 3 is unused. Bits 2 and 6 were originally meant to be used for control and sensing of the third valve. When control of the strip chart recorder drive was added, a judgement call was made as to whether it would be best to use bit 2 for recorder control and leave port B entirely unused, or use one bit of port B in the event that a third valve would be added. Since two valves seemed adequate for most MHPLC experinents, port B was left unused so it could be devoted to some other purpose. To add the third valve, it is therefore necessary to first move the recorder control to port B. The words ON and OFF would then have to be re-written to change the bit in port B that is controlling the relay. These words are fairly simple and can easily be altered. Control of the third valve is added by connecting bit 2 of port 0 to the new valve controller nodule TTL input. Sensing of valve three is accomplished by connecting bit 6 of port 0 to the sensing switch. The words CW and CCW should then be able to control the new valve when given an argument of 3. 264 Suggestiong;for Aggificial Intelligence Applications Normally, the microcomputer issues commands to the 11/23, but the 11/23 cannot asynchronously send a command to the microcomputer. Establishing two-way communication with the 11/23 would be the first step in configuring the system for closed-loop artificial intelligence schemes. One way of receiving commands from the 11/23 is to program the micro to repetitively check the USARTS for the command. The word to get a single character from the USARTS is GETCHR. The command would then be decoded and acted upon. Probably the easiest way to decode the commands from the 11/23 would be to have the program write the conmands in the FORTH language initially. The new commands would be placed in a block and the only characters which would then need to be manually decoded from the USARTS would be the block number. Control is then tranferred to the new command sequence by LOADing the block. The USARTS can be checked in a semi-asynchronous fashion by conpiling the code that does the checking as a background task. 265 APPENDIX C An Operator’s Guide for the Isolated Droplet Generator APPENDIX C A2 Operator’s Guide for the IDG The purpose of this appendix is to provide a concise procedure for the generation of various droplet streams. External electrical connections are listed for each component. A short troubleshooting section is also furnished. A. Single-droplet Production in the Stand-alone Mode The IDG consists of the following modules: 1. IDG main electronics module 2. Bertran 205A-10P High-voltage power supply 3. Intersil 7226A Counter/timer/frequency neter (CTFM) 4. General Radio 1531AB Stroboscope 5. Bimorph and electrode mounting assembly 6. Liquid delivery device 7. Bimorph/capillary assembly These modules should be assembled and tested as follows: 1. Connect the case of the high voltage power supply to a good ground, such as a water pipe. 2. To deflect charged droplets up, connect the positive 266 aim-mummy. may ~23 mum high voltage to the top deflection electrode. Connect it to the bottom electrode to deflect charged droplets down. 3. Connect the ground wire of the high voltage power supply to the other deflection electrode. 4. Set the output of the high voltage to 4 kV. Put the meter switch to the "I” position to measure the current. Turn the high voltage on and check the current. After an initial deflection, it should quickly drop to zero. If it doesn’t, immediately turn off the power supply and check for a short circuit. 5. If the high voltage appears to be working properly set the meter switch to the "V” position. Turn off the power until the remaining connections are made. 6. Plug in the stroboscope and IDG electronics to grounded outlets. 7. Connect the "Strobe Sync" output on the IDG to the "Input” jack on the stroboscope. 8. Set the strobe dial to one of the "External Input" settings. The low intensity setting can flash faster than the high intensity setting, but droplets are sometimes easier to view with a higher intensity. 9. Plug in the CTFM to the left side of the IDG electronics module using the square keyed plug. 10. Connect the ”Frequency Meter" output on the IDG to the ”A” input on the CTFM. 11. Set the function switch on the CTFM to ”Freq" and the 267 “Twin-N. _" update period to ".1 sec." 12. Set the following switches on the IDG: Source A: up (divider A internal) Latch: down (unlatched) or up (latched) Source B: up (divider B internal) Bimorph Driver Source: up (internal- divider A) Pulse Trigger Source: up (internal- divider B) Bimorph Driver Amplitude: less than about 5.0 Pulse Amplitude: zero volts (fully counterclockwise) Strobe Sync: 100 (divide pulsing frequency by 100) Divider B Adjust: 002 (charge every third drop) Divider A Adjust: 9.9 Frequency Meter Select: Bimorph Driver In 13. Before connecting the bimorph, test the IDG, strobe and CTFM. With the above parameters set, turn on the IDG. The frequency meter should read approximately 10-11 kHz. Adjust the frequency to 10.2 kHz and lock it by placing the latch switch in the up position. The "A/D latched” LED should change from red to green. 14. Turn on the stroboscope. After a brief warm-up period, the lamp should begin to flash. Change the "Strobe Sync” to 1000. The lamp should flash 10 times more slowly. Return the strobe sync to 100. Change the Divider B Adjust to 20. Again, the lamp should flash 10 times more slowly. Change the strobe sync to 10. The lamp should flash at the 268 original rate. 15. Change the frequency neter select to "PT in." The meter should read the droplet pulsing frequency. This should be the bimorph driver frequency divided by 1 plus the number on the thumbwheel. 16. Connect the output labeled "TO BIM" on the IDG to the bimorph clamp. The yellow wire (ground) attaches to the thumbscrew on the back of the Plexiglas. The driver signal (clear insulation) attaches to screw on the clamp situated opposite the knurled adjustment screw. 17. Clamp the bimorph between the rear of the brass block and the non-insulated leg of the clamp. 18. Turn the IDG back on and slowly increase the binorph amplitude. With a frequency of 10.2 kHz, a clear tone should be heard from the bimorph. 19. Connect the pulsing electrode to the red banana jack labeled "TO PULS ELEC.” Turn the instrument back on and slowly increase the pulse amplitude. The meter should appear steady. An oscilloscope can be connected to the pulsing electrode to observe the waveform. 20. The instrument should now be ready to produce droplets in the stand-alone mode according to the procedure given in the applications chapter of this dissertation. If any of the above tests did not work, refer to the troubleshooting section below. 269 R; Pro_gction of Neutral Droplets To produce neutral droplets (i.e., charge the rest of the stream) make the following addition to the above procedure. 1. Connect the "Sync Out B" to the ”Ext In Bimorph Driver” on the IDG. 2. Change the ”Pulse Trigger Source” switch to the external position (down). C. Ug_ of tRe 10 MHz Oscillator The CTFM contains a 10 MHz oscillator which can be used as an alternate input source to divider A. This oscillator offers the advantage of having much higher resolution than the internal 1 MHz oscillator. However, care must be taken in its use, as it is possible to input too high a frequency to the bimorph driver. Use the following procedure. 1. Connect the yellow banana jack on the CTFM (AUX) to the ”Ext In Div A" input on the IDG. 2. Place the bimorph driver source switch in the external (down) position. 3. With the divider A source switch in the internal position, and the frequency selector switch on ”A out” adjust the frequency to a value less than 5 kHz. 4. Change the divider A source switch to the external position. The frequency reading should increase by a factor of 10. 270 5. Adjust the frequency to the exact value desired (do not exceed 50 kHz for long periods). Much better resolution should be available than with the internal 1 MHz oscillator. 6. Return the bimorph driver source switch to the internal position. 7. Return the frequency meter select switch to the bimorph driver in position. D. Operation ip,the Computer-controlled Mode To operate the IDG in the computer-controlled mode, first follow the procedure for the stand-alone mode. If everything appears to be working properly perform these additional steps: 1. Connect the "AM9513-A counter 1" output on the back of the micro to the bimorph driver external input. Be careful not to accidentally connect it to the bimorph driver output! With a T-connector, also connect it to the divider 8 external input. 2. Connect the "AM9513-B counter 3" output on the back of the micro to the pulse trigger external input on the IDG. Also connect the micro digital ground (the DAC ground is suitable) to one of the IDG BNC grounds. 3. Log in to the HPLC account and load block 9. If unsure of how to do this consult the procedure outlined in Appendix B. 4. Type IDG and return to load the IDG software. 271 5. Test the software by saying 1 2 10.20 IDG. The program should respond by showing the two closest frequencies and the one actually used. These parameters are the same as those used in the stand-alone procedure above. The 1 represents 1 droplet charged, and the 2 represents two neutral drops. 6. Place the bimorph driver source toggle to external. 7. Place the pulsing trigger source toggle to external. 8. Place the divider B source toggle to external. 9. Change the frequency meter select switch to bimorph driver in. The frequency should be 10.2 kHz. 10. Change the frequency meter select switch to pulse trigger in. The frequency should be 3.55 kHz. 11. To view the droplets produced in this manner, the flashing frequency must be manually adjusted by varying the divider B adjustment thumbwheel. In this case a value of 002 should stabilize the stream. E. Grounding Considerations Several ground connections are sometimes needed to stabilize the droplet stream. The liquid delivery vessel should be grounded to assist in droplet charging. If a ringstand is used to hold the bimorph mount, the ringstand should also be grounded to prevent 60 Hz pickup. Pickup of 60 Hz noise makes the droplet stream appear to travel in a sinusoidal trajectory. 272 If a trap is used to catch charged droplets, the trap should be grounded or else charge will build up on the trap and alter the droplet trajectory. Likewise, any nearby object which can pick up static charges or act as an antenna should be grounded. The IDG electronics case or high voltage case can be used as ground points. F. Troubleshooting Symptom: Frequency meter noisy and .gives incorrect frequency. Correction: 1. Decrease bimorph driver amplitude when measuring frequency. 2. Use frequency meter select on bimorph driver input instead of divider A output. Symptom: Frequency meter reads abnormally high frequencies from divider A. Cause/Correction: l. Instrument was turned off and on too fast with divider A latched. The instrument must be allowed to sit for approximately 3 minutes after being turned off before being turned back on, or divider A must be in the unlatched mode. 2. External 10 MHz oscillator being used. Adjust divider A to lower frequency to less than 50 kHz. Symptom: No output from bimorph driver. Cause/Correction: 273 l. Bimorph driver source toggle on "external” with no signal connected. 2. Bimorph driver was operated at too high of an input frequency. Check transistors on bimorph driver and replace if necessary. 3. Short circuit in bimorph clamp assembly. Check for continuity. Symptom: No pulsing voltage. Cause/Correction: 1. Check transistors on power supply board and replace if necessary. Symptom: Erratic strobe lamp operation. Cause/Correction: l. Lamp being operated at too high a frequency. Reduce frequency. 2. Lamp being operated at too high an intensity. Use lower intensity setting. Symptom: Cannot charge droplets. Cause/Correction: 1. Droplet stream too fast. Use slower flow rate. 2. Solvent cannot be charged. Use different solvent or solvent mix with supporting electrolyte. 3. High voltage power supply not turned on. 4. Solution delivery vessel not grounded. 5. Broken cable to pulsing electrode. Check for continuity. 274 6. Phasing not properly adjusted. Adjust bimorph amplitude in stand-alone mode or use phasing keys in computer controlled mode. Symptom: Droplet strean will not remain stable for long periods. Cause/Correction: 1. Flow rate changing. If constant pressure delivery is being used, then the in-line filter may be dirty. Replace filter element. 2. Frequency not latched. Place latch toggle in up position to lock frequency. 3. Marginal production frequency used. Use better frequency. Suggestionp_for IDG-LIBS Research One of the main considerations in isolated droplet introduction to a laser spark is phasing of the laser trigger with droplet introduction. Fortunately, the IDG can generate the trigger signal in either the stand-alone node or the computer-controlled mode. In the stand-alone mode, the easiest introduction technique is to put the entire droplet stream through the plasma region and use the second frequency divider (divider B), normally used for pulsing droplets, as a trigger signal for the laser. Phasing is then acconplished by physically moving the bimorph mount closer or farther away from the 275 plasma using a micro-positioner. Most likely a signal translation nodule would have to be built to convert the TTL-based divider B output to a signal appropriate for the laser trigger. In the computer-controlled mode, a similar strategy is used. The entire droplet stream is sent through the focal point, and the 9513 counter normally used to trigger the pulsing circuit is used to trigger the laser. Phasing in this case can be accomplished electronically using the phasing keys. Phasing can be adjusted in increments of 1 us under computer control. A signal translation module would still have to be built, as the 9513 output is a TTL-based signal. A serial line to an 11/23 supporting FORTHPIP would also have to be available. 276