Anal. Chem. 1998, 70, 2481-2486
Stepper Motor Controlled Micro Gas Valve Inlet System for Gas Chromatography Mark Nowak, Anita Gorsuch,† Heather Smith, and Richard Sacks*
Department of Chemistry, University of Michigan, Ann Arbor, Michigan 48109
A novel gas valve inlet system for gas chromatography is described. The device incorporates a capillary sample gas delivery tube containing a small orifice in its side from which sample vapor continuously flows. A precision micro stepper motor is used to translate the sample delivery tube parallel to its axis so that the orifice passes by the end of the fused silica separation column. The inlet end of the column and the sample delivery tube are housed in a pressurized injection port which also contains purge flows to vent sample between injections. Two operating modes are described. In the sweep mode, the orifice sweeps past the column end at a constant, adjustable velocity. In the park mode, the orifice is parked in front of the column end for a software selectable time. Injection sample size and bandwidth are adjustable. Bandwidths (σ) as small as a few milliseconds make the inlet suitable for high-speed gas chromatography as well as conventional GC. Electronically controlled sample injection systems are finding numerous applications in gas chromatography (GC). For highspeed separations, narrow injection bandwidth and a precise zerotime reference are important.1-3 Precisely controlled injection time and sample size also are important for multiple input chromatographic analysis.4,5 Most inlet system designs for highspeed GC use either valves and sample loops or focusing devices. For environmental monitoring applications and in many laboratory applications, focusing inlet devices, which function as sample preconcentrators, are often required in order to achieve adequate powers of detection.2,3,6 A limitation of these inlet systems is the need for cooling fluids for very volatile compounds. Gas valves with sample loops are relatively simple and are extensively used for monitoring chemical manufacturing processes. They have also been applied to high-speed GC, and with special techniques, can generate sample vapor injection plug widths of a few milliseconds or less.7-9 Angell et al.7 developed a micromachined inlet using an etched channel as a sample loop † Present address: Chromatofast, Inc., 912 N. Main St., Suite 14, Ann Arbor, MI 48104. (1) Gaspar, G.; Arpino, P.; Guiochon, G. J. J. Chromatogr. Sci. 1977, 15, 256. (2) Klemp, M.; Peters, A.; Sacks, R. Environ. Sci. Technol. 1994, 28, 369A. (3) van Es, A.; Janssen, J.; Cramers, C.; Rijks, J. J. High Resolut. Chromatogr. 1988, 11, 852. (4) Phillips, J. B.; Luu, D.; Pawliszyn, J. Anal. Chem. 1985, 57, 2777. (5) Annino, R. J. Chromatogr. Sci. 1976, 14, 265. (6) Klemp, M.; Akard, M.; Sacks, R. Anal. Chem. 1993, 65, 2516. (7) Angell, J. B.; Terry, S. C.; Barth, P. W. Sci. Amer. 1983, 248 (4), 44.
S0003-2700(97)01259-6 CCC: $15.00 Published on Web 05/09/1998
© 1998 American Chemical Society
and a micromachined injection valve. This inlet system is used in some field-portable GC instruments. Typically, the valve is opened for 10 ms and about 80 nL of sample gas is injected into the carrier flow. van Es et al.8 used a pneumatically actuated sixport valve with a capillary-dimension splitter to introduce millisecond-wide sample pulses into a microbore capillary column. Rankin and Sacks9 used an electrically actuated six-port valve with a sample loop to introduce precise sample volumes into a focusing inlet for high-speed GC. Despite important successes, valves with sample loops lack flexibility, since sample size can be changed only by changing the sample loop volume. In addition, sample vapor must pass through the valves and thus may be subject to alteration and contamination. These valves also may be maintenance intensive. Tijssen et al.10 described a novel gas valve inlet device which uses a needle with a side hole, driven by a gas-pressurized piston, to deliver a stream of sample vapor into a pressurized injection block. To inject a sample, the needle is translated parallel to its axis so that the side hole sweeps past the end of the capillary separation column. Purge flows of carrier gas are used to vent sample vapor between injections. Sample pulses with variances less than 1 ms2 were reported. Peters et al.11,12 used a similar gas valve inlet device with Al2O3 and porous polymer PLOT columns for the high-speed analysis of low molecular weight organic compounds. This report describes a gas valve inlet system which uses a micro stepper motor to translate a sample delivery tube containing a side orifice. A linear step size of 2.5 µm provides very high spatial resolution and thus precise positioning of the orifice along the column axis. A sweep mode is described where the orifice sweeps across the column axis at a rate determined by the motor speed. A park mode is also described where the orifice is parked on the column axis for a software-selected time. EXPERIMENTAL SECTION Apparatus. The inlet system consists of two principal componentssthe injection block and the stepper-motor drive system. Figure 1a shows a section view of the aluminum injection block B. The block is machined in two parts which are connected (8) van Es, A.; Janssen, J.; Bally, R.; Cramers, C.; Rijks, J. A. J. High Resolut. Chromatogr. 1987, 10, 273. (9) Rankin, C.; Sacks, R. LC-GC 1991, 9, 428. (10) Tijssen, R.; van den Hoed, N.; van Kreveld, M. E. Anal. Chem. 1987, 59, 1007. (11) Peters, A.; Sacks, R. J. Chromatogr. 1991, 29, 403. (12) Peters, A.; Klemp, M.; Puig, L.; Rankin, C.; Sacks, R. Analyst 1991, 116, 1313.
Analytical Chemistry, Vol. 70, No. 13, July 1, 1998 2481
Figure 2. Side view of stepper-motor drive system. I, injector block; S, sample delivery tube; L, low-friction slider; M, micro stepper motor; C, motor controller; D, radial drive arm; G, pressurized sample gas source. The inset shows a top view of the optoelectric trigger consisting of a light source and detector P connected to the base and a slit T connected to the low-friction slider.
Figure 1. Section view of injector block (a) and enlarged view of injection cell (b). B, injector block; S, sample delivery tube; C, capillary column; BPR, back-pressure regulator; I, carrier gas inlet; O, carrier gas outlet. Arrows in (b) indicate carrier gas flows.
with an O-ring seal. This provides access to the interior injection cell and facilitates alignment of components. The block contains the sample delivery tube S, the end of the capillary separation column C, and the carrier gas inlet and outlet ports labeled I and O, respectively. A groove is machined in the right side of the block to accommodate the column. A back-pressure regulator BPR (Model 80486, 0-30 psi, Porter Instruments Co., Hatfield, PA) controls the block pressure and thus the column head pressure. The stainless steel sample delivery tube is 0.030 in. (0.76 mm) o.d. and contains a 0.25 mm diameter hole in the tube wall. The tube passes through a pair of 0.031 in. (0.79 mm) diameter holes in the block so that the tube can be translated parallel to its axis. Figure 1b shows an enlarged view of the column and sample delivery tube in the injection cell. The orifice in the side of the sample delivery tube is shown along the column axis. In this position, sample is entrained in the carrier gas flow into the column. The small spaces between the sample delivery tube and the holes in the block provide continuous flows of carrier gas which serve to purge sample from the block. The carrier gas flow out the back-pressure regulator also serves to remove sample from the injection block. These purge flows effectively prevent sample vapor from reaching the column except when the orifice in the sample delivery tube is near the column axis. These purge flows are shown by arrows in Figure 1b. The small spaces between the sample delivery tube and the block also provide for low friction translation of the sample delivery tube and serve as motion guides so that the orifice maintains proper alignment with the column axis. The end of the column passes through a small bushing which is fastened to the injection block. This provides positional stability for the column. 2482 Analytical Chemistry, Vol. 70, No. 13, July 1, 1998
As sample emerges from the orifice, a vapor plume is formed. The shape of the plume depends on the diameter of the hole, the block pressure, the sample gas delivery pressure, and the purge flow patterns in the block. It is clear that the end of the column should be very close to the wall of the sample delivery tube in order to achieve minimum injection bandwidth as the hole is translated past the column end. For the work reported here, the column end was always within 0.25 mm of the tube wall. Figure 2 shows a side view of the complete system used to control the position of the sample delivery tube. A micro stepper motor M (Model SX57-51, Compumotor Division, Parker Hannifin Corp., Rohnert Park, CA) connected to a radial drive arm D is used to drive a low-friction slider L. The sample delivery tube S is attached to the slider. The injection block is indicated by I. A pressurized sample gas source G is connected to the sample delivery tube. The motor has 16 selectable resolutions ranging from 200 to 50 800 steps per revolution. Rotational accuracy of (5 arcminutes and repeatability of (5 arcseconds are typical. The motor is controlled by a dedicated microprocessor C (Model SM controller, Compumotor Division). An optoelectric trigger P, consisting of a light-emitting diode and a phototransistor, was used to initiate data collection when the orifice in the sample delivery tube approached the column axis. A top view of this is shown in the inset of Figure 2. A slit assembly T attached to the moving portion of the low-friction slider was used to generate the light pulse. The inlet block was attached to the side of a Varian 3700 GC. The Varian flame ionization detector (FID) was used without change. An amplifier-electrometer circuit, built in house, was used to interface the FID to a 12-bit A/D board (Data Translation, DT2801). For injection bandwidth measurement, all filter capacitance was removed from the amplifier, and the time constant was less than 1 ms. High-speed sample chromatograms were recorded with a 5 ms amplifier time constant. The electrometer was interfaced to a Gateway 2000 386 PC. Data collection was controlled by Labtech Notebook (Laboratory Technologies) software. Data analysis software was written in house. All injection bandwidth measurements were made with a sampling rate of 250 Hz. Materials and Procedures. Hydrogen carrier gas was passed through filters for oxygen, water vapor, and hydrocarbons.
Table 1. Mixtures Used for Chromatograms in Figure 7 peak
compound
boiling point, °C
Impurities in Ethylene -161 -88 -42 -48 -84 -104
A B D E F C
methane ethane propane propylene acetylene ethylene
A B C D E
Chloromethanes in Air methane -161 chlorodifluoromethane -41 methylene chloride 40 carbon tetrachloride 76 chloroform 61 air
concentration 1.0% 1.1% 1.0% 1.0% 101 ppm balance 0.093% 0.10% 0.12% 0.097% 0.10% balance
Table 2. Mixture Used for Chromatograms in Figure 8 peak A B C D E F G H I J K L M N O P Q R S T U V W X
compound isoprene manufacturing stream methane n-butane trans-2-butene 1-butene isobutene cis-2-butene cyclopentane isopentane n-pentane unknown 1,3-butadiene cyclopentene 3-methyl-1-butene trans-2-pentene 2-methyl-2-butene 1-pentene 2-methyl-1-butene cis-2-pentene unknown 1,4-pentadiene 2-butyne 2-methyl-1,3-butadiene cis-1,3-pentadiene trans-1,3-pentadiene
boiling point -161 -0.5 0.9 -6.3 -6.9 3.7 49.2 27.8 36.1 -4.4 44.2 20 36.3 38.6 30 31.2 36.9 26 27 34 42 42
Three test mixtures were used to evaluate the inlet system. A mixture of low molecular weight hydrocarbon impurities in ethylene was provided by Union Carbide Corp. An isoprene process stream sample was provided by Shell Development Co. A mixture of chlorinated methanes in air was obtained from Scott Specialty Gases. Retention standards were prepared from reagent grade compounds. All separations were performed using 0.32 mm i.d. Al2O3 porous-layer open tubular (PLOT) columns. Tables 1 and 2 summarize the mixtures used in this study. The ethylene and the chloromethane samples were delivered to the sample delivery tube directly from compressed gas tanks through a two-stage regulator. The isoprene sample and standards were vaporized in a sealed, gastight container, and compressed, high-purity nitrogen was used to deliver the headspace vapor to the inlet. For injection plug bandwidth measurements, methane was used as the sample. A 1.0 m long, 0.32 mm i.d. uncoated open tubular column was used so that relatively short transport time to the detector could be achieved at modest injector block
pressures. Since the purge flows as well as column flow increase with increasing block pressure, relatively low block pressures were used to reduce carrier gas consumption. Block pressures used ranged from 1.3 to 2.3 psi above atmospheric pressure. Average carrier gas velocitity ranged from 100 to 300 cm/s, depending in part on column length. Purge flows were not measured but were quite large with the result that the carrier gas tank had to be replaced about once a week. The sample gas delivery pressure typically was about 2.5 psi above atmospheric pressure RESULTS AND DISCUSSION The use of a micro stepper motor to control the position of the orifice in the sample delivery tube allows several software adjustable injection modes and sequences. In the sweep mode the sample delivery tube is translated at nearly constant velocity so that the orifice passes across the end of the column. Injection bandwidth is adjusted by control of the motor speed. For the park mode, the orifice in the side of the sample delivery tube is rapidly moved from a rest position where the sample gas is completely vented to the location of the column axis. The orifice then is parked in front of the column, and sample vapor is entrained in the carrier gas flow into the column. After a softwareselected park time, the hole is translated back to its original position. Precise repetitive injections can be made at a repetition rate which depends on the park time. Sweep-Mode Operation. The stepper-motor drive system is set to deliver sample to the capillary column when the axis of the rotating drive lever (see Figure 2) is orthogonal to the linear translation axis of the low-friction slider supporting the sample delivery tube. For this condition and for relatively small rotation angles from this position, the linear step size is found as the ratio of the circumference of the circle generated by the end of the drive lever to the number of steps per revolution. For the work reported here, a lever length of 10.16 mm and 25 000 steps per revolution were used. This gives a step size of 2.55 µm. From the manufacturer’s stated accuracy of (5 arcminutes and repeatability of (5 arcseconds, the corresponding translational accuracy and repeatability are about (15 µm and less than (1 µm, respectively. The orifice in the sample delivery tube is 250 µm in diameter, and the column is 320 µm in diameter. Thus sample will be delivered to the column over an orifice translation range of 570 µm. This is equivalent to about 224 steps, and corresponds to a motor axis rotation of about 3.2 degrees. Under these conditions, the linear step size is very nearly constant as the orifice sweeps past the end of the column. For a motor rotation rate of 1.0 revolutions per second (rps), the base (4 σ) width of the injection band is expected to be about 8.9 ms. The sweep mode was investigated for motor rotation rates ranging from 0.50 to 10.0 rps. This corresponds to orifice translation rates from 3.1 to 63.8 cm/s, respectively. For these studies, the injection port pressure was 16.7 psi, and the outlet pressure was 14.5 psi. Figure 3 shows sweep-mode methane peaks for orifice translation rates of 63.8 cm/s (a), 12.8 cm/s (b), 9.6 cm/s (c), 6.4 cm/s (d), and 3.8 cm/s (e). In all cases, relatively symmetric peaks are observed. Peak widths (σ) are in the 5-6 ms range for peaks (a)-(d). Peak (e) has a width of about 8 ms. Peak areas increase steadily with decreasing orifice translation rate. Analytical Chemistry, Vol. 70, No. 13, July 1, 1998
2483
Figure 3. Methane peaks for the sweep mode using orifice translation velocities of 63.8 cm/s, (a), 12.8 cm/s (b), 9.6 cm/s (c), 6.4 cm/s (d), and 3.8 cm/s (e).
Figure 4. (a) Sweep-mode plots of peak width (A) and peak area (B) as functions of orifice translation velocity. (b) Plot of peak area as a function of the reciprocal of the orifice velocity.
Figure 4a shows plots of peak width (A) and peak area (B) as functions of orifice velocity. For orifice velocity values less than about 6 cm/s, peak width increases steadily with decreasing velocity. For velocity values greater than 6 cm/s, peak widths are nearly constant at about 6 ms. In addition, for orifice velocities greater than about 6 cm/s, more point scatter is observed in the peak width data. A break also is observed in the peak area data for an orifice velocity of about 7 cm/s. These results suggest that mechanical instability may occur for the higher orifice velocity values. Any movement of the sample delivery tube normal to its translation axis will result in variations of peak area and peak width. Flexing of the tube or movement 2484 Analytical Chemistry, Vol. 70, No. 13, July 1, 1998
Figure 5. Methane peaks for the park mode using park times of 1 ms (a), 5 ms (b), 50 ms (c), and 100 ms (d).
of the tube in the 25-µm gap between the sample delivery tube and the vent holes may be involved. Shot-to-shot variation in peak width typically is in the range (3-8% relative standard deviation for five measurements. Relative standard deviations for peak area measurements are typically in the range (0.5-2.5%. The amount of sample deposited on the column is proportional to the time interval for which the proximity of the orifice allows entrainment of sample into the column. Thus, peak areas should be proportional to the reciprocal of the orifice velocity. This is confirmed by the plot shown in Figure 4b. The linear regression line has a correlation coefficient greater than 0.99. The minimum observed peak width of about 6 ms is the result of other sources of band broadening. Peak width (σ) from broadening on the column was calculated to be about 3 ms. This is based on the Golay equation (13) for the case of an unretained solute. The dead time of flame ionization detectors has been shown to be on the order of 1 ms or less.14 Stray capacitance in the electrometer-amplifier circuit also results in significant band broadening. The data acquisition frequency of 250 Hz used here may contribute significantly to shot-to-shot variation in measured peak width. Park-Mode Operation. Figure 5 shows a series of methane injections using the park mode. Peaks a-d are for park times of 1, 5, 50, and 100 ms, respectively. For the 1-ms park time, the peak width (σ) is about 15 ms. This is the minimum observed injection plug width. For park times greater than 100 ms, the injection plugs become more or less rectangular with no significant change in peak height. For park times greater than about 200 ms, the programmed park time and the measured full width at half-maximum are in good agreement. For all park times, the measured full width at half-maximum is about 35 ms greater than the programmed park time. For this mode of operation, the minimum observed injection plug width and the shape of the rising and falling portions of the peaks are determined largely by the time required for deceleration of the orifice as its center approaches the column axis and (13) Giddings, J. C. Anal. Chem. 1962, 34, 3. (14) Gaspar, G.; Annino, R.; Vidal-Madjar, C.; Guiochon, G. Anal. Chem. 1978, 50, 1512.
Figure 6. Park-mode plots of peak width (A and B) and peak area (C and D) as functions of park time for sample gas pressures of 19.8 psi (A and D) and 16.9 psi (B and C).
acceleration as its center leaves the location of the column axis. The acceleration and deceleration of the motor are under program control. For this study, values of 100 rps2 were used. Larger values will result in reduced minimum injection plug width but at the price of increased mechanical stress on the sample delivery tube and poorer reproducibility for replicate injections. Figure 6 shows plots of peak width at half-height (A and B) and peak area (C and D) vs park time for park times from 1 to 1000 ms. Plots labeled A and D are for sample gas delivery pressures of 19.8 psia and, plots labeled B and C are for a delivery pressure of 16.9 psia. For all plots, the injector block pressure was 16.7 psia. Note that the differences in injection plug width for the different pressure cases are small and may reflect differences in the shape of the vapor plume and the carrier gas flow patterns. The peak width plots are very linear with regression correlation coefficients greater than 0.999. Shot-to-shot relative standard deviations of peak width for five measurements typically are in the range (1-3%. The sample delivery pressure must always be greater than the block pressure to ensure that sample enters the block. The rate of sample entering the block can be controlled by the sample delivery pressure. This is illustrated by comparison of peak area plots C and D. Thus, sample size and injection bandwidth can be controlled independently. The peak area plots are linear with park time and have regression correlation coefficients of 0.9999 and 0.9975 for plots C and D, respectively. Shot-to-shot relative standard deviations in peak area for five injections using the same park time typically are in the range (1-3%. Applications. Figure 7 shows high-speed chromatograms of impurities in ethylene (a) and a mixture of chlorinated methanes (b) (see Table 1) using the gas valve inlet system. The inlet was operated in the sweep mode with an injection bandwidth of less than 10 ms. Both chromatograms were obtained with the use of 0.32-mm i.d. KCl-deactivated Al2O3 PLOT columns. For chromatogram (a), the column length was 4.0 m and the average carrier gas velocity was about 180 cm/s. The column temperature was 60 °C. For chromatogram (b), the column length was 2.0 m, the carrier gas velocity was about 300 cm/s, and the column temperature was 140 °C. The ethylene sample is completely separated with good resolution in less than 20 s. The mixture of chlorinated methanes is completely separated in less than 6 s.
Figure 7. High-speed chromatograms obtained using the sweep mode. The injection band widths (σ) were less than 10 ms. See Table 1 for peak identification. (a) Impurities in ethylene. (b) Chlorinated methanes in air.
Figure 8. Chromatograms of an isoprene process stream sample obtained using the sweep mode. See Table 2 for peak identification. (a) KCl-deactivated Al2O3 PLOT column. (b) Na2SO4-deactivated Al2O3 PLOT column.
Figure 8 shows chromatograms of an isoprene (2-methyl-1,3butadiene) manufacturing process stream. Again, the sweep mode was used with injection bandwidths less than 10 ms. See Table Analytical Chemistry, Vol. 70, No. 13, July 1, 1998
2485
2 for peak identification. The mixture contains mostly C4 and C5 saturated and unsaturated hydrocarbon compounds. Using conventional techniques, chromatograms of comparable quality typically require 10-20 min. The chromatograms were obtained with 5 m long, 0.32 mm i.d. Al2O3 columns using KCl (a) and Na2SO4 (b) deactivation. The columns were operated at 100 °C. The Na2SO4 deactivated phase is somewhat more polar and more selective for alkenes and alkynes. For chromatogram (b), the isoprene (Peak V) elutes in about 120 s and is not shown. Note that in chromatogram (b) peak pairs C-D and O-P are well separated while coelutions occur in chromatogram (a). However, component pair T-U coelutes in (b) and is adequately separated in (a). These chromatograms suggest that the use of parallel columns would be very effective. Since the movement of the sample delivery tube is under program control, the rapid, sequential injection of sample plugs onto two or more parallel columns should be straightforward. CONCLUSIONS The work described in this report used relatively large-bore capillary columns, and the orifice in the sample delivery tube was of comparable size. For these conditions, a stepper motor with lower resolution and precision could be used. However, for microbore columns and with smaller orifice size, the 2.55 µm spatial resolution of the orifice would be very useful and could be improved further using a shorter radial drive arm or greater motor
2486 Analytical Chemistry, Vol. 70, No. 13, July 1, 1998
resolution. Very fast separations on columns at least as small as 5 µm i.d. have been reported.15 In addition, more complex injection sequences for multiple-input chromatography are possible. The inlet system described here has been shown to be useful for high-speed GC with analysis times of a few seconds. The inlet should also be very useful for conventional GC. The inlet should be useful for high-speed on-line monitoring in applications where the use of cryogenic fluids or refrigeration units is undesirable. In addition, the very short cycle times possible with the inlet system could be very useful for the study of more rapid chemical and physical processes. Significant limitations of the inlet system are the relatively high rate of carrier gas consumption and the attendant limitations on injection port pressure. The purge vents were machined to provide a 25 µm (0.001 in.) clearance for the sample delivery tube. This was the capability of the machining facility. For the relatively short vent tube length used here, the pneumatic restriction was low and the flow rate of carrier gas was excessive. Alternative vent designs are under development. Received for review November 17, 1997. Accepted March 31, 1998. AC971259B (15) Phillips, J. B.; Derhsing, L.; Lee, R. J. Chromatogr. Sci. 1986, 24, 396.
