/
1‘O
TIME (ms)
Figure 2. Upper curve: Oscilloscope trace. ( 0 )Experimental points obtained from upper curve. (-) Fitted curve: distance vs. time. (- -) Velocity vs. time
-
The most common method of determining dead times is the extrapolation method ( I ) . It consists of carrying out a chemical reaction in a stopped-flow apparatus, recording the absorbance of the reaction mixture as a function of time, and extrapolating the resulting curve back to zero absorbance. The elapsed time between the time of zero absorbance and the stopping time of the syringes (the intersection of the extrapolated line and the experimental curve) is td’. The quantity t d ’ includes the effects of mixing and stopping times as well as temperature, cavitation, and other artifacts which may degrade the performance of stopped-flow mixing systems. Stewart ( I ) has shown that td’ approaches t d in properly operating stopped-flow systems as the reaction half life becomes long with respect to td. Under conditions similar to those used to obtain the datashown in Figure 2, the extrapolation method and the flow velocity method were simultaneously used to acquire data for eight separate pushes of our stopped-flow instrument. The precision of the flow velocity method depends primarily on the precision of the measurement of the period between two peaks in the oscilloscope trace since the dimensions of the stopped-flow unit and the transducer are relatively constant during a push. Replicate determinations of the period of the last two cycles of the transducer (distance = 0.102 cm) before the stop yielded a mean value of 5.25 f 0.09 ms (-+ 1S),and thus a mean value for u,, of 19.3 f 0.5 cm/s. The mean dead time is then 5.08 f 0.09 ms. The precision of the extrapolation method of determining
td‘ depends on how precisely the reaction rate curve may be extrapolated to zero absorbance. The mean value of the dead time obtained by this method in the comparative study was 5.4 f 0.4 ms. Thus the precision of the extrapolation method was f8%, about four times worse than that of the flow velocity method. The determinant error in the flow velocity method depends upon the accuracies of the time base of the oscilloscope, the dead volume, the ruling, and the cross sectional area of the drive syringes. By estimating tolerances for these quantities and applying the usual propagation of error procedures, we are able to calculate an upper bound of f11% for the range of the accuracy of our method of obtaining td. This value could be improved by using a more accurate time standard such as a crystal controlled period meter for measuring the time between pulses of the transducer or by determining the dimensions of the stopped-flow system more accurately. The ability to assign an upper limit to the determinant error in a dead time measurement is a definite advantage of the flow velocity method over methods requiring calibration procedures such as the initial absorbance method (I), the vane method of Gibson (6),and the potentiometer method of Chance (7). These last two techniques, however, give absolute Grection and magnitude indication, and may provide higher resolution. An advantage of optical coupling is that nothing is attached to the drive mechanism of the stopped-flow apparatus that would impede its motion. It should be emphasized that the extrapolation method and the flow velocity method are complementary diagnostic aids in the development and maintenance of stopped-flow mixing systems. A comparison of the times obtained by both methods may reveal mixing, stopping, temperature, or cavitation effects and facilitate their elimination. This flow velocity detector is extremely simple to attach to the drive system, either temporarily or permanently, and it requires only modest and common equipment for implementation.
LITERATURE CITED (1) J. E. Stewart, “Durrum Application Notes No. 4”, Flow Deadtime in Stopped-Flow Measurements, Durrum Instrument Corp., Palo Alto, Calif. 94303. (2) J. M. Sturtevant, in “Rapid Mixingand Sampling Techniques in Biochemistry”, Britton Chance, et al., Ed., Academic Press, New York, N.Y., 1964. (3) Q. H. Gibson and L. Milnes, Biochem. J., 91, 161 (1964). (4) R. M. Reich, Anal. Chern., 43 (12), 85A (1971). (5) P. M. Beckwith and S. R. Crouch, Anal. Chern., 44, 221 (1972). (6) Q . H. Gibson, Discuss. Faraday Soc., 17, 137 (1954). (7) B. Chance, J. Franklin lnst., 229, 455 (1940).
RECEIVEDfor review February 13,1976. Accepted April 23, 1976. This work was supported in part by NSF Grant No. MPS75-03650.
Pressure Control of a Liquid Chromatograph Pump Frank J. Van Lenten’ and L. David Rothman*2 Department of Chemistry, University of Georgia, Athens, Ga. 30602
Supercritical fluid chromatography (SFC) has been shown to be a powerful separation technique, especially for high resolution separations of mixtures of nonvolatile compounds such as polymers (1-3). The mobile phase pumping systems used for these separations are controlled-pressure pumps, since capacity ratios for solutes in SFC are pressure dependent (2). Typical systems used are: fluid displacement with high Present address, Lederle Laboratories, Pearl River, N.Y. Present address, The Dow Chemical Company, Midland, Mich. 48640. 1430
ANALYTICAL CHEMISTRY, VOL. 48, NO. 9, AUGUST 1976
pressure gases ( 4 ) , a pneumatic amplifier pump (5), and a constant volumetric flow pump with column pressure maintained by a valve at the column exit. Pressure programming in SFC has been shown to be analogous to temperature programming in gas chromatography and solvent programming in liquid chromatography (LC) ( 2 ) . In each case, the use of the program allows elution of all components in a reasonable period of time without sacrificing resolution of species that elute early. In systems described previously, pressure programming has been accomplished by
~~~~~
~
PRESSURE TRANSDUCER
Table I. Pressure Stability Av pressure, psi
Std dev, psi
RSD, %
372 446 660
0.4 1.0 2.4
0.12 0.22 0.37
Table 11. Reproducibility of Starting Pressure Pulses Av pressure, psi Std dev, psi RSD, % 89 359.8 2.2 0.61 94 379.7 1.7 0.45 990 404.5 5.2 1.18 99 408.5 3.1 0.76 104 474.6 7.1 1.46 109 500.0 5.2 1.04 The combined RSD for both runs at 99 pulses was 1.13%.
Figure 1. Block diagram of pump controller DEVICE INTERFACE CONTROL PULSE 8
/
/
SN741IU
Q
Table 111. Pressure Program Reproducibility Pulses/min
Av rate, psi/min
Std dev, psi/min
RSD, %
0.5 1.0 2.0
2.69 5.45 10.59
0.18 0.27 0.67
6.5 5.0 6.3
VARIAN PLUG REMOTE
IN271
SN74121
J 102
using a stepping motor to control a gas pressure regulator (2) or a valve at the column exit (5, 6). Our method of p u m p pressure control is applicable to a n y commercial LC p u m p which is driven by a stepping motor. T h e control unit described here is all-electronic.
EXPERIMENTAL A block diagram of the pressure controller is shown in Figure 1.A pressure transducer is used to monitor the pump output pressure. The signal is compared to the reference voltage, V,,f, which may be derived from a potentiometer, a digital-to-analog converter (DAC), or any other source. The voltage difference between the two signals is amplified and used to drive a voltage-to-frequency (V/F) converter, which should have a moderately wide frequency range over the range of possible input voltages. The output of the V/F converter is used to determine the flow rate of the pump by selecting the stepping rate of the pump motor. The flow rate is, therefore, varied as necessary to maintain the pressure set by Vref. The circuit diagram of the controller is shown in Figure 2. The voltage reference selected for our application was an 8-bit DAC (Model No. 19-8-B, Date1 Systems Inc., Canton, Mass.). The DAC output was set by sending a train of pulses into the counter circuitry. The pulses were derived from a general purpose interface (7) to a PDP-11/20 minicomputer (Digital Equipment Corp., Maynard, Mass.). The pressure transducer used for this work was a standard component in the pump (Model 8500 liquid chromatograph, Varian Associates, Palo Alto, Calif.). This transducer has an output voltage of 10 mV/atm. A voltage follower with gain (OA 1)was used to amplify the transducer signal to a more convenient level (0.6 V at 400 psi). The pressure determined from the transducer output was compared to the pressure read on a high quality mechanical gauge (Model No. C-54958, Heise, Newton, Conn.). Disagreement between the two devices over the pressure range of interest was less than 5 psi. The difference between the DAC and amplified transducer-output voltage is amplified by a gain-of-5 difference amplifier (OA 2). The output of OA 2 is then used to control the output frequency of the V/F converter. The V/F converter used in this application was designed and built in this laboratory (8).It provides a logic-level output and frequency range of about 100-3000 Hz with a linear transfer function over the range of input voltages from -0.1 to -3 V dc. A pulse rate of 1 Hz will result in a flow rate of 1 ml/h and the desired range of flow rates is approximately 50-200 ml/h (50-200 Hz). It is necessary to vary the duty cycle of the pulses to make the output of the controller compatible with the remote-control circuitry of the pump. This is accomplished by a one-shot integrated circuit after amplification of the pulses by transistor circuitry. The output of the one-shot is then fed into the remote pump rate input (5102-1).Those pulses drive the stepping motor in the pump, adjusting the pump flow rate to maintain the desired pressure.
Flgute 2. Schematic diagram of pump controller (Varian 8500) O A l , OA2, and OA3 are SN72741N integrated circuit N1 is an SN7400 NAND gate integrated circuit
operational amplifiers:
RESULTS Supercritical fluid chromatography (SFC) has been performed in both the isobaric and programmed-pressure modes. It is, therefore, important t o demonstrate how this p u m p system will perform in each mode. Tests were first performed t o see how well a constant pressure could be maintained (important for isobaric experiments) and then how well the initial pressure and the rate of change in pressure could be controlled. These tests were performed over the range of pressures most commonly used for SFC in this laboratory (400-800 psi). T h e data obtained when t h e system was set t o deliver a constant pressure are shown in Table I. At each pressure, ten pressure measurements were performed at regular intervals over a period of at least 30 min. As the data show, t h e relative standard deviation (RSD) was in all cases less than 0.5%. Such pressure control is sufficient for most isobaric SFC separations. T h e next test of system performance was a n examination of t h e reproducibility with which the p u m p could be stepped t o a desired pressure. This is of interest, since all experiments in S F C require the selection of a n initial pressure, a n d the reproducibility of that pressure will influence t h e reproducibility of t h e separations obtained. Since most S F C experiments performed in this laboratory begin at pressures of 400-500 psi, experiments were performed over this pressure range. Starting at 0 psi pressure, a set number of counts or pulses were s e n t into the counter a n d the pressure was recorded. T h i s experiment was performed t e n times at each initial pressure tested. T h e data in Table I1 show that the initial pressure could be selected within 1.5%relative standard deviation. The final test concerned t h e reproducibility of linear pressure programs generated with this system. Counts were sent into the controller at rates of 0.5,1.0, and 2.0 counts/min. T w o programs were performed at 0.5 count/min and four programs for 1.0 a n d 2.0 counts/min. T h e average rate of pressure change a n d t h e program-to-program standard deANALYTICAL CHEMISTRY, VOL. 48,
NO. 9, AUGUST 1976 * 1431
viations are shown in Table 111. As the data show, the rates were reproducible to better than 7% RSD.
DISCUSSION The experimental evaluation of the system has been used to demonstrate the stability and reproducibility of the pump controller. The main sources of error are the resolution of the 8-bit DAC (approximately 0.4%)and the ability of the pressure transducer to reset reproducibly (approximately 0.2%). Since there are no commercial pumps designed for pressure programming applications, it is difficult to compare them with our system; however, most syringe pumps including the Varian, claim reproducibility of flow rates of 1%,while some reciprocating pumps, especially with flow or electronic feedback, feature reproducibilities of better than 0.3% (9,10). The only pressure programming in SFC to our knowledge has been by Jentoft and Gouw (2,11,12) and Nieman and Rogers (5) but specifications for pump performance were not mentioned. Using a pneumatic amplifier pump, Nieman and Rogers could achieve reproducibility of desired pressure to approximately 5%, with pulses causing difficulties at low pressures. Jentoft and Gouw (2) mentioned that, with proper modifications, syringe pumps could be used for pressure programming but they did not elaborate further. In a recent article, Guiochon et al. (13)pointed out that syringe pumps will usually achieve a steady-state flow only after 15-60 min of operation, depending on the characteristics of the pump, column resistance, flow rate, and mobile phase viscosity. In practice, Achener et al. (14) showed that steady-state flow could be achieved in less than 3 min using the “fast pump” mode. Once the reservoir of eluent had been
brought up to pressure, the outlet valve of the reservoir closed, a sample injected, and the pump restarted, a steady flow was attained within 30 s. The device that we have described permits a similar procedure to be followed nearly automatically. It not only permits elutions a t a constant flow, but also facilitates pressure programming under computer control in linear or nonlinear modes.
ACKNOWLEDGMENT Assistance in manuscript preparation from L. B. Rogers is gratefully acknowledged, as is the contribution of B. P. Semonian to the V/F converter circuitry. LITERATURE CITED (1) (2) (3) (4) (5) (6) (7) (8) (9)
J. C. Giddings, J. Chromatogr. Sci., 7, 276 (1969). R. E. Jentoft and T. H. Gouw, J. Chromatogr., 68,303 (1972). S.T. Sieand G. W. A. Rijnders, Anal. Chim. Acta, 38, 31 (1967). S.T. Sie and G. W. A. Rijnders, Sep. Sci., I, 459 (1966). J. A. Nieman and L. B. Rogers, Sep. Sci.. I O , 517 (1975). F. J. Van Lenten and L. B. Rogers, unpublished work (1975). J. E. Davis and E. D. Schmidlin, Chem. Instrum., 4, 169 (1973). L. D. Rbthman and B. P. Semonian, unpublishedwork (1975). H. Barth, E. Dalloneier, G. Courtois,’ H. E. Keller, and B. L. Karger, J. Chromatogr., 83, 289 (1973). (IO) K. Asai et al., Shimadzu Seisakusho Ltd., Kyoto, Japan, Paper No. 267. 1975 Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy. (1 1) R. E. Jentoft and T. H. Gouw, J. Chromatogr. Sci., 8, 138 (1970). (12) R. E. Jentoft and T. H. Gouw, Anal. Chem., 44,681 (1972). (13) G. Guiochon et al., J. Chromatogr., 112, 399 (1975). (14) P. Achener, S. R. Abbott, and R. Stevenson, to be published; private communication, F. Baumann, March 1, 1976.
RECEIVEDfor review March 15,1976. Accepted April 29,1976. Partially supported by ERDA Contract No. E(38-1)854.
Supplementary Information for “Reagent Chemicals”, 5th Edition The ACS Committee on Analytical Reagents wishes to announce additional modifications of requirements andlor tests prior to the publication of the first supplement. For other modifications, refer to Vol. 47, No. 7, p 1149 and Vol. 48, No. 4, p 792.
Aluminum Sulfate On page 67 in the requirement for Insoluble matter, raise the limit from 0.005%to 0.01%.
Bromocresol Green On page 136 in the requirement for Visual transition interval and on page 137, line 2, of the corresponding test, change pH 4.0 to pH 3.8.
Oxalic Acid On page 424 in the requirement for Sulfate, raise the limit Yrom 0.002%to 0.005%.
Iron. Determine the iron by the atomic absorption spectrophotometric method described on pages 21-25. Sample Stock Solution. Dissolve 5.0 g in water and dilute with water to 100 ml. Preparation of Solutions. Use an aliquot size of 20 ml of Sample Stock Solution in 50-ml volumetric flasks. Add 0.02 mg of iron (Fe) to Control Solution E. Observe the absorbances of the five solutions using the 248.3-nm iron line. Calculate the net quantity of iron in Sample Solution A. The net quantity should not exceed 0.01 mg* Sulfanilic Acid Insert the following note in the space preceding the Requirements: This reagent is available as both the anhydrous and the hydrous (1H20) salt. The identity should be placed on the label.
Phosphoric Acid, MetaOn page 440, change two of the requirements as cited below: Assay. Noiless than 33.5 nor more than 36.5% HP03. Stabilizer. Not less than 57 nor more than 63% NaP03.
Sodium Chloride On page 560 in the requirement for Sulfate, raise the limit from 0.001%to 0.004%.
Sodium Molybdate On page 582 for the Iron test substitute the test cited below: 1432
ANALYTICAL CHEMISTRY, VOL. 48, NO. 9. AUGUST 1976
CORRECTION Analysis of Hexafluoropropylene/Vinylidene Fluoride Copolymers by High Resolution Continuous Wave and Fourier Transform Nuclear Magnetic Resonance Spectrometry
In this article by E. G. Brame, Jr., and F. W. Yeager, Anal. Chem., 48,709 (1976), lines 4-6 under Results and Discussion should read “from the area of the methylene lines (6 = 2.1-5.0) relative to that of the aromatic proton lines (6 = 7.8) of the internal standard, 2,5-dichlorobenzotrifluoride.The integrals for”