Anal. Chem. 1996, 68, 2717-2719
Solenoid Pumps for Flow Injection Analysis Debra A. Weeks† and Kenneth S. Johnson*
Moss Landing Marine Laboratories and Monterey Bay Aquarium Research Institute, P.O. Box 450, Moss Landing, California 95039
Methods employing flow injection analysis (FIA), particularly for in situ seawater techniques, would benefit from reduction in pump size and power requirement, longer maintenance intervals, and the ability to incorporate microprocessor control of each reagent and sample flow stream. In this work, the peristaltic pump of a conventional FIA system was replaced by three solenoid-driven diaphragm pumps with integral Viton check valves, and the system was tested by performing the simple nitrite analysis, which has well-defined FIA performance characteristics. Sixty injections per hour were possible with flow rates of 0.5 mL/min for reagents and sample. The coefficient of variation was 1% for 10 µM NO2- concentrations, and the detection limit was less than 0.1 µM NO2-. These values match the reported performance for this method using peristaltic pumps. Typical FIA systems employ peristaltic pumps, which give a continuous delivery rate for all reagents, with sample and reagent delivery ratios controlled by varying pump tubing size. These pumps are large, and pump tubing changes may be required daily for some reagents. Many applications, including in situ seawater analysis,1,2 would benefit from a reduction in pump size, weight, maintenance, and power requirements. The flexibility of individual microprocessor control of each flow stream would allow variable flow rates (including no flow) for each reagent, sample, and standard. This would facilitate assay optimization and make it possible to access several different analyses on the same FIA system. Further, this computer-controlled and variable reactant flow rate access may provide a useful tool in solution kinetics analysis. Solenoid minipumps from the Lee Co. are much smaller and lighter than peristaltic pumps (approximate volume for one minipump is 20 cm3, with a weight of 45 g), and they can be individually controlled. Average power consumption at a flow rate of 1 mL/min is nominally 0.14 W for each pump. Four solenoid pumps use about 1/20 the power used by a peristaltic pump. Each pump is rated for ten million cycles, which corresponds to 1 year of continuous operation at 1 mL/min. The pumps function by a solenoid-driven piston pressing or releasing a diaphragm which is integrated with a dual check valve. These solenoid pumps do not expose the reagents and carrier to pump tubing (typically Tygon or silicon), which is not as inert as the Teflon tubing used in most of the system. The † Current address: Chemistry Board, University of California, Santa Cruz, CA 95064. (1) Andrew, K. N.; Blundell, N. J.; Price, D.; Worsford, P. J. Anal. Chem. 1994, 66, 917A-922A. (2) Johnson, K. S.; Coale, K. H.; Jannasch, H. W. Anal. Chem. 1992, 64, 1065A-1075A.
S0003-2700(96)00040-6 CCC: $12.00
© 1996 American Chemical Society
standard wetted surfaces within the pump are made of PEEK and Viton. Solenoid pumps have been used for injecting standards into the sample stream of a continuous flow analyzer designed for longterm, submerged deployment.3 A solenoid pump was also used in another in situ application, where it was employed to renew pH indicator for measuring pCO2 in seawater.4 While apparently well-suited to in situ applications, the solenoid pumps have a functional drawback when used to deliver reagents. Whereas peristaltic pumps provide a continuous and relatively smooth flow of reagents, each solenoid pump provides a 50 µL pulse of fluid per stroke. At a flow rate of 1 mL/min, this amounts to one pump stroke every 3 s, which results in a pulsating flow. An even mixing of sample and reagents is important in FIA, as the reaction is generally not taken to its end point.5 In this paper, the quantitative performance characteristics of the solenoid pumps were evaluated to assess whether useful results could be obtained in a FIA system. The colorimetric determination of nitrite as an azo dye6 is a simple and reliable method already adapted to FIA, with well-defined performance characteristics.7 Therefore, it was selected as the analyte for validating FIA performance with the new pumps. EXPERIMENTAL SECTION Three Lee Co. (Westbrook, CT) solenoid pumps, Model LPLA0520350L, were used (Figure 1). They delivered carrier (either Millipore Milli-Q water or low-nutrient seawater which was allowed to sit at room temperature in the dark for 4 weeks so that any remaining nitrite would oxidize), reagent 1 (10.0 g of sulfanilamide and 100 mL of concentrated HCl per liter), and reagent 2 (1.0 g/L N-(1-naphthyl)ethylenediamine dihydrochloride, NED). A peristaltic pump was used to load sample or standard into the injection loop; a solenoid pump would have the same performance characteristics here, as injection loop loading does not affect the flow characteristics of the analysis. The sample is introduced into the system by the carrier pumping it out of the injection loop when the valve is in the inject position. The pneumatic actuator for the rotary injection valve (Rheodyne), the solenoid pumps, and data capture from the colorimeter were controlled by an IBM-compatible personal computer with a Keithley Metrabyte DAS8 interface card running a QuickBasic program written for these purposes. A standard 5 V power supply (3) Jannasch, H. W.; Johnson, K. S.; Sakamoto, C. M. Anal. Chem. 1994, 66, 3352-3361. (4) DeGrandpre, M. D.; Hammar, T. R.; Smith, S. P.; Sayles, F. L. Limnol. Oceanogr. 1995, 40, 969-975. (5) Ruzicka, J.; Hansen, E. H. Flow Injection Analysis; John Wiley & Sons: New York, 1988. (6) Strickland, J. D. H.; Parsons, T. R. A Practical Handbook for Analysis of Seawater; Fisheries Board of Canada: Ottawa, 1972. (7) Johnson, K. S.; Petty, R. L. Limnol. Oceanogr. 1983, 28, 1260-1266.
Analytical Chemistry, Vol. 68, No. 15, August 1, 1996 2717
Figure 1. Schematic of the FIA system.
Figure 3. Detector response for analysis of 20 µM nitrite standards in Milli-Q water using a 1 cm × 0.2 cm flow cell. Flow rates for the carrier and reagents are all 0.5 mL/min. Reaction coil length is 1 m.
Figure 2. Detector response for analysis of 20 µM nitrite standards in Milli-Q water using a 1 cm × 0.2 cm flow cell. Flow rates for the carrier and each reagent are 1 and 0.5 mL/min, respectively. Reaction coil length is 1 m.
and optically isolated relays were used to power the pumps and actuator. All tubing was 0.8 mm i.d. Teflon. Teflon tubing with 0.5 mm i.d. was tested initially, but it interfered with the flow rate, ostensibly due to high back-pressure. A Hewlett-Packard 8452A diode array spectrophotometer with a 1 cm × 0.2 cm flow cell and a Lachat QuickChem colorimeter with a 2 cm × 0.3 cm flow cell were used as the detectors. The HP spectrophotometer was operated at 540 nm, while a 560 nm filter was used in the Lachat colorimeter.
Figure 4. Detector response for analysis of 10, 0.5, and 0 µM nitrite standards in seawater using a 2 cm × 0.3 cm flow cell. Flow rates for the carrier and reagents are all 0.5 mL/min. Reaction coil length is 1 m.
RESULTS AND DISCUSSION The flow cell volume and the ratio of the sample to the reagent flow rates were critical in reducing detector signal oscillations and optimizing sensitivity. Large oscillations in detector signal (Figure 2) were observed when the system was operated with the smallvolume (30 µL) flow cell in the Hewlett-Packard detector and with unequal flow rates for the sample and reagents (1 mL/min for sample and 0.5 mL/min for each reagent). These oscillations resulted from the incomplete mixing of the sample and reagents. The ratio of the carrier flow rate to those of reagent 1 and reagent 2 has a large effect on mixing in the reaction manifold. This ratio was varied incrementally from 3:1:1 to 1:1:1. This kept the ratio of reagent 1 to reagent 2 constant. The standard method7 is generally performed at flow rates of 2:1:1. The smallest detector
signal oscillations were obtained using the 1:1:1 sample to reagents ratio. The highest peak absorbances were found at the lowest flow rates: 0.5 mL/min for each pump (Figure 3). The lower flow rates give more time for the azo dye to form and are more desirable for in situ applications where reagents will be limiting. Flow rates lower than 0.5 mL/min did not produce acceptable sample throughput. However, adjusting flow rates did not completely eliminate detector signal oscillations. The oscillations in detector signal could be further reduced by using the larger flow cell (140 µL) in the Lachat detector, which averaged the signal from a larger volume of solution (Figure 4). The peak-to-peak variations in detector response with this system were about 0.0009 absorbance, which approaches the stability of the detector electronics. Much of the remaining noise can be
2718 Analytical Chemistry, Vol. 68, No. 15, August 1, 1996
Figure 5. Detector response for analysis of four replicates of 10 µM nitrite samples on two different days, separated by nearly 7 months. The same 2 cm × 0.3 cm flow cell was used, with flow rates for the carrier and reagents all 0.5 mL/min. Reaction coil length is 1 m.
eliminated by simple digital signal processing techniques. Pulse dampening methods were also tried to further reduce detector signal oscillations without changing performance characteristics, but there was no apparent improvement. Increasing the reaction coil length was tested to see if adequate mixing and elimination of the oscillations in detector signal could be obtained with longer coils. No coil and 0.5, 1.0, 1.5, and 2.5 m coils were used. Reduced detector oscillations and increasing peak absorbance were found with longer coil lengths. However, the 1.5 and 2.5 m coils caused a reduction in flow rate, possibly due to increased back-pressure with length. Since these coils were hand-knit from 0.8 mm i.d. Teflon tubing, a variable amount of kinking could have been introduced, further increasing resistance to flow. This reduction in flow rate with longer coils illustrates an additional weakness of the solenoid pump. They operate with less than 90% efficiency at back-pressures above 0.1 bar. A 1 m reaction coil was selected for the work reported here. The optimal sample loop size would be that which resulted in the largest amount of sample with resolution to the baseline for the highest analyte concentrations. An injection loop of 100 µL was selected, as that volume of sample, with the reagents and subsequent dispersion added, would more than fill the large flow
cell void volume. A 50 µL injection loop resulted in a 25% loss in response. Increasing the loop size to 200 µL increased the peak heights, but the baseline for the 10 µM NO2- standard peaks was not resolved in less than 1 min. Consistent heights for both the 10 and 0.5 µM NO2- standards were obtained with equal flow rates on all pumps and the large volume flow cell (Figure 4). Only a few representative consecutive injections are presented in order to resolve the baseline detector oscillations still present. Peak areas were also consistent, with a coefficient of variation less than 1% for the 10 µM NO2- standard, calculated from three sets of 10 consecutive injections. Milli-Q and seawater blanks, in their respective carrier, give no apparent response. Detection limits were less than 0.1 µM NO2-, as calculated from 3σ of 20 replicates of the 0.5 µM NO2- standard. This is comparable to the values obtained with a FIA system using a peristaltic pump.7 Several replicates of a standard curve consisting of 40, 20, 10, 5, 0.5, and 0 µM NO2- standards were generated; the lowest R2 value obtained was 0.9990, with a slope of 0.093 absorbance/µM and y-intercept of 0.007 AU. To demonstrate reproducibility, 10 replicates of a 10 µM NO2- sample were run nearly 7 months apart on the same system. These compared well with each other, with slight differences in baseline noise (Figure 5). CONCLUSIONS Solenoid pumps can produce acceptable performance characteristics in a typical FIA system, and they have features which could result in better in situ FIA systems. They also have the flexibility to be independently controlled, which would be an advantageous feature of the reagent delivery system for any FIA system. Although the solenoid pumps also have a deficiency, due to pulsation of the flow stream, this work demonstrates that much of this effect can be eliminated. We believe that, with further development, FIA systems based on solenoid pumps would represent a significant improvement over FIA systems using peristaltic pumps. ACKNOWLEDGMENT This work was supported by Office of Naval Research Grants N00014-93-I-0856 and N00014-89-J-1010 and by the Monterey Bay Aquarium Research Institute. Special thanks to Kenneth Coale for his input. Received for review January 16, 1996. Accepted April 24, 1996.X AC960040E X
Abstract published in Advance ACS Abstracts, June 1, 1996.
Analytical Chemistry, Vol. 68, No. 15, August 1, 1996
2719
