Anal. Chem. 1983, 55, 1615-1617
GM IC-952
1615
1 I'?
l/F
End V i a w
Sido View
Pr e c i s i on I= l o w Me a su r e me n t 1)evi c e
Figure 1. Precision flow measurement device with circuitry.
Flow
irni"/rnlI
Flgure 3. Calibration of flow monitor using soap-bubble flow meter showing flow rate at the column inlet, f h, and column outlet, f wt, both corrected to 20.00 O C , and the average flow rate on the column, F , and corrected flow rate at the column outlet, f,, both measured at the column temperature.
Measurement Device Output
I
Y
temperature in the accepted manner (I). If the inlet pressure is accurately known, flow rates at the column inlet as well as the average flow rate on the column may also be evaluated readily (Figure 3). Inlet pressures were measured for this application with a Texas Instruments Model 144-01 quartz precision-pressure gauge. Three positions along the T / C output curve were considered in making the time difference measurements. These are the difference in time between peak maximum and minimum, the difference in time between the inflection points on the leading edge of the positive and negative curves, and the difference in time between an arbitrarily chosen threshold point on the leading edge of the two peaks near the peak bases somewhat greater than twice the peak-to-peak noise level. Little advantage is found in reproducibility of one method over the others; indeed, all are capable of measurement to better than the constancy of the high-precision flow controller used, which is ca. f0.15%. In the case of the peak maximumlminimum method, two to four values of equal magnitude are generally measured at the maximum and minimum owing to the rapid rate of data acquisition. The computer program used must choose a consistent value, perhaps the middle one, from which to determine the time difference. Thus some advantage may be present in using one of the other two approaches since the output changes more rapidly in these regions.
LITERATURE CITED (1) Dai Nogare, S.;Juvet,
R. 8 . "Gas-Liquid Chromatography: Theory and Practice"; Wiiey-Interscience: New York, 1962; pp 73-77, 162-163.
Figure 2. Recorded flow-monitor output with computer injection of helium every 40 s.
measurements. Calibration curves for the flow rate a t the column outlet, corrected to both 20.00 "C and to the column temperature, were const,ructed by using a soap-bubble flow meter, correcting for wat.er vapor, atmosplheric pressure, and
RECEIVED for review February 22, 1983. Accepted April 11, 1983. R. S. Juvet acknowledges support provided in part by the National Science Foundation under Grants CHE 73-0521.2 and GP-40830X. This work was performed at Ecole Polytechnique while the senior author was on sabbatical leave.
Hydraulic Filter for a Ternary-Solvent Gradlent-Elution Device Robert M. Gershey" and Roger Gueviremont
'
Atlantic Research Laboratory, National Research Council, 14 7 1 Oxford St., Halifax, Nova Scotia, Canada B3H 321
The use and developmient of analytical methods based on liquid chromatography can make large demands on the in'NRC No. 21254.
strumentation budget of any laboratory. Recently, the availability of a variety of commercial microprocessor boards has allowed the d.evelopment of "homemade" gradient elution devices a t low cost ( I , 2 ) . Both of these systems, however,
0003-2700~/83/0355-1615$01.50/0 0 1983 American Chemical Soclety
1616
*
ANALYTICAL CHEMISTRY, VOL. 55, NO. 9, AUGUST 1983
.
I
ALTEX PUMP
D
BASIC PROGRAM
INITIALIZATION
Flgure 1. Block diagram of gradient elution system: (A) solenold valves, (B) low-pass fllter, (C) solvent pump, (D) solvent line to chromatograph, (E) microprocessor, and ( F ) solvent reservoirs.
required two solvent pumps and more savings could be realized with a system that uses only one pump. A solvent gradient can be formed with a single pump that is fitted with a number of solenoid valves, each connected to a different solvent reservoir. The valves are rapidly cycled with each being held open for a time proportional to the concentration of the corresponding solvent in the mixture. Such systems have been described for binary solvent mixtures ( 3 , 4 ) . One problem with this type of system involves proper timing of the solenoid valves with respect t o the motion of the pump piston. This problem may be overcome through use of mixing chambers ( 4 ) or to a certain extent by manipulating the timing signal sent to the solenoid valves. Alternatively, the fluid flow characteristics caused by the pump stroke can be modified to meet the requirements of the solenoid valves. We describe here an hydraulic low-pass filter designed to smooth the flow on the inlet side of solvent delivery pumps. When used in conjunction with a single pump, solenoid valve gradient elution device, this filter eliminates the need for a mixing chamber as described by Saunders ( 4 ) and allows the system to be easily adapted to a variety of reciprocating pumps with widely differing stroke characteristics.
INPUT PORT
INPUT STATUS
IU PUMP STROKES
PATTERN TO VALVES
I
Flgure 2.
I
Flow chart of machine language program.
EXPERIMENTAL SECTION The gradient forming device used in this work consists of a single piston high-pressure pump fitted with three solenoid valves, the filter, and a microprocessor controller (Figure 1). A commercially available set of solenoid valves with a control unit (Autochrom solvent selector, $650) is attached t o the inlet line of a solvent delivery pump (Beckman/Altex llOA). Any valve may be opened manually or by placing the appropriate two-bit signal (‘‘1’’= 5 V, “0” = 0 V) on the input terminals of the control unit. A microprocessor with a single &bit output port (Motorola MEK6802D3, $200) is used to sense the beginning of the refill stroke of the pump and subsequently output the appropriate signals to the solvent selector, A flow diagram of the machine code program for the microprocessor is given in Figure 2, The input parameters necessary for running the gradient are generated by a BASIC program on our laboratory’sApple 11-plus computer. The user provides the initial and final compositions of the solvent mixture and the total volume required for the gradient. Complete listings of both the machine code program for the microprocessor and the BASIC program are available upon request from the authors. The hydraulic filter between the pump and solenoid valves lengthens the pulse in the solvent delivery line from 200 ms to about 5 s, thus allowing time for four cycles of the valves between pump strokes. It must contain both capacitative and resistive elements. The resistance of our filter (Figure 3) is provided by a 25 cm length of Q,5 mm i.d. Teflon tubing and the capacitor is made of a 1-mL syringe barrel closed off at one end and partially filled with solvent. The intake stroke of the pump piston withdraws solvent from the syringe barrel causing reduced pressure within that chamber. Solvent then flows into the barrel from the reservoirs through the solenoid valves and the resistance element.
“0
Figure 3. Schematic diagram of (B) solvent line to pump inlet.
hydraulic filter: (A) solenoid valves,
Flow rates were measured photographically. Photographs of the syringe barrel and a stopwatch taken at half-second intervals were analyzed under a low power microscope.
‘RESULTS AND DISCUSSION We tested the system by using solvents of different viscosity and monitoring the composition of the solvent mixture with a refractive index detector. Smooth linear gradients are produced provided the time available between pump strokes is longer than the time required for the solvent to flow through the filter. The time constant of the filter can be easily adjusted to suit solvents of differing viscosity or t o match the duration of solvent flow through the valves with the time available be-
1617
Anal. Chem. 1983, 5 5 , 1617-1619
tween pump strokes. This is done eitber by changing the length of the resistance element or by altering the amount of air in the capacitor. The values for these parameters can be calculated as follows. The resistance to fluid flow in a cylindrical tube is proportional to the length and inversely proportional to the square of the radius of the tube. This is expressed in the HagenPoiseuille equation (5:)as
ar4 8P.L
U=---ap
(1)
where U is the volume flow rate of fluid, r is the radius of the tube, p is the dynamic viscosity, L is the length of the tube, and AP is the pressurle drop across the length of the tube. The capacitance of the fiiter is formulaited by using the ideal gas law. The instantaneous pressure in the syringe is equal to Pi
=
Po vo
d Vi dt
r i l l I
0
2
3 t (SI
I
I
I
I
5
Figure 4. Instantaneous volume vs. time calculated for: (A) pentane, (C) methanol, (D) water, and (E) 2-propanol. Open trlangles represent values determined for water in the present system.
(e) acetonltrlle,
v, + v, - vi
where Vo is the amount of air initially in the syringe a t pressure Po (the capacitance), V, is the volume of the pump stroke, and Vi is the volume of fluid that has been drawn through the resistance tube. Atmospheric pressure less this quantity is the pressure drop AP of eq :l.This substitution allows us to write an expression for the amount of liquid flowing through the fillter a t any time ~
E //
106~r4Vo 1oeTr4 + 8pL( Vo + V, - Vi) 8p.L
of viscosities commonly encountered are included in the figure. Also included is the observed response of the filter with distilled water. It is clear that eq 5 accurately predicts the behavior of the device. In order to design a filter for use with any pump, the user first determines the time between pump strokes of volume V,. The value for p is that of the most viscous solvent used. Of the parameters V, and L , one is chosen and the other calculated from eq 6 or 7.
I-
I-
(3)
Rearrangement and substitution yield
(4) where A = -106ar4/8pL. Integration of eq 4 gives an expression for the time required to pass a given volume of liquid, Vi, through the filter lJ7i
v,
t = l d t = - A f A-[[In
ACKNOWLEDGMENT We thank B. D. Johnson for helpful discussions on the fluid mechanics of the system and W. D. Jamieson, S. Whiteway, E. Lewis, and E. C. V. Butler for helpful comments on the manuscript.
L Pi
P O
AP
(V, - Vi) - In V,]
(5)
r
t
U
Solving for L and Vo gives
Vi VO VP P
and
GLOSSARY length, cm instantaneous pressure, g cm-I s - ~ initial pressure, g cm-l s+ pressure drop, g cm-I s - ~ tube radius, cm time, s volume flow rate of fluid, cm3 s-l instantaneous volume of fluid, cm3 initial volume of air, cm3 volume of pump stroke, cm3 dynamic viscosity, g cm-l s-l LITERATURE CITED
(7) where B = In (V, - Vi) - In Vp. Figure 4 shows the rlesponse of the present filter ( L = 25, V, = 0.139 and Vo = 0.5) as calculated from eq 5. Solvents often used in liquid chromatography representing the range
(1) Bedard, P.; Purdy, W. C. Anal. Lett. 1983, 16, 149-158. (2) Brady, J. E.; Carr, P. W. J. Chem. Educ. 1983, 60,83. (3) Billiet, H. A. 14.; Keehnen, P. D. M.; De Galan, L. J . Chromatogr. 1979, 185, 515-528. (4) Saunders, D. L. J. Chromatogr. Sci. 1977, 75, 129-136. (5) Brodkey, R. S. “The Phenomena of Fluid Motions”, 2nd ed.; Addison-
Wesley: Reading, MA, 1967; Chapter 9.
RECEIVED for review February 23,1983. Accepted May 9,1983.
Ion Chromatographic Deterrnination of Fluorine, Chlorine, Bromine, and Iodine with Sequential Electirochemical and Conductometric Detection Chung-Yu Wang, Scott D. Bunday, alnd James C. Tartar* Department of Chemistry, North Texas State University, Denton, ‘Texas 76203
Ion chromatography (IC) as developed by Small et al. ( I ) has proven to bo a very useful technique for determinatiosn
of inorganic anions at the parts-per-million and sub-partper-million range. Applications and limitations of IC with
00O3-27O0/83/0355-1617$01.50/00 1983 American Chemical Society