Pumping pressure and reagent consumption in continuous flow

Discrete injection segmented flow analysis of nutrients in small-volume water samples. Wayne S. Gardner and ... ALTEX SCIENTIFIC INC. Analytical Chemi...
0 downloads 0 Views 418KB Size
used here purely for demonstration purposes. Indeed, there is good reason to suppose that most thermally stable reactive species can be generated in this manner. In a sense, the present results remove the “hardware” aspects from consideration in reactive gas generation and focus attention on the thermal properties of organic molecules. We also note that the majority of substances that are presently considered hazardous are, in fact, liquid or solids a t ambient conditions and are not particularly unstable in the bulk. For such substances, generation can be effected without the pyrolyzer and obviously our observations with respect to reproducibility, precision, accuracy, and range will hold with equal, if not greater, force. Thus, this instrument may well provide a simple general solution to a difficult analytical problem.

LITERATURE CITED (1) G. 0. Nelson, “Controlled Test Atmospheric, Principles and Techniques”, Ann Arbor Science Publishers, Ann Arbor, Mich., 1971. (2) W. Tsang, J. Res. Natl. Bur. Stand., Sect. A, 78, 157 (1974).

B. Dimitriades, C. F. Ellis, and 0. E. Seizinger. in “Advances in Chrornatography”, Vol. 7, J. C. Giddings and R. A. Keller, Ed., Dekker. New York. 1968. J. Calvert and McQuigg, Int. J. Chern. Kinet., Symposium No. 1 Proc. of the Symposium on Chem. Kin. Data for the Upper and Lower Atmosphere, 1974, p 113. Fed. Regist., 36,No. 157, Aug. 13, 1971. S. W. Benson. and H. E. O’Neal, “Kinetic Data in Gas Phase Unimolecular Reactions”, Natl. Stand. Ref. Data Ser., Natl. Sur. Stand. ( U .S .), 21, 645 pages (Feb. 1970). R. K. Stevens, A. E. O’Keeffe, and G. C. Ortman, unpublished results.

RECEIVEDfor review April 26,1976. Accepted July 22,1976. Supported in part by the Office of Air and Water Measurement and the Center for Fire Research of the National Bureau of Standards. Certain commercial materials and equipment are identified in this paper in order to specify adequately the experimental procedure. In no case does such identification imply recommendation or endorsement by the National Bureau of Standards, nor does it imply that the material or equipment identified is necessarily the best available for the purpose.

Pumping Pressure and Reagent Consumption in Continuous Flow Analysis with Unsegmented Reaction Streams Marvin Margoshes Technicon Instruments Corporation, Tarrytown, N. Y. 1059 1

Three research groups have independentlydeveloped equipment for rapld chemical analysis by continuous flow without segmentatlon by air bubbles. All three systems rely on rapld flow of a reagent stream in narrow tubing to llmlt the spreading of a sample that is injected as a bolus of liquid. Elementary considerations show that an unsegmented continuous flow analysis system will either consume considerably more reagent than a segmented system or else will require very hlgh pumping pressures.

At least three independent groups have recently described systems for chemical analysis based on continuous flow in an unsegmented fluid stream. The system described by White and Fitzgerald ( I , 2 ) employs a photochemical reaction in a flowing stream, with the sample introduced by gravity ( I ) or by injection via a motor-driven syringe ( 2 ) . The reagent stream is pumped by applying a small air pressure (5 psi) to the reagent reservoir. The system of Ruzicka and Hansen ( 3 ) employs a peristaltic pump and rapid, manual injection of the sample into the flowing stream. Stewart, Beecher, and Hare ( 4 ) also used gas pressure pumping, but at a much higher pressure of 400 to 500 psi; they introduced the sample into the reagent stream by means of a liquid chromatography valve. The three systems have similarities. All employ rapid flow in narrow tubing to control peak spreading. All are used for automation of analyses with simple chemical steps and short reaction times. The purpose of this publication is to discuss some limitations of unsegmented continuous flow analytical systems, particularly the relation between pumping pressure and reagent consumption.

BACKGROUND Unsegmented reaction streams have been employed for many years in chromatography. The amino acid analyzer of Spackman, Stein, and Moore (5)incorporated such a method

for the reaction between the column effluent and the ninhydrin reagent. These authors recognized that band spreading in the flowing stream could be controlled by keeping the tube diameter small at all points along its length. They observed that a 0.5-ml aliquot of the effluent (equivalent to 1 min of flow from the column) occupied a length of 150 cm in the narrow tubing they used. It was soon recognized that turbulence is the key to control of band spreading in a fluid stream. This can be seen in publications by Sternberg and Poulson (6),by Giddings (7),and by Pretorius and Smuts (8).In laminar flow, the fluid in the center of the tube moves at twice the mean fluid speed, leading to rapid longitudinal mixing which is seen as peak spreading. In turbulent flow, the fluid velocity is more nearly constant across the diameter of the tube and peak spreading is greatly reduced. Segmented continuous flow analysis, devised by Skeggs (9), employs air bubbles to divide the sample stream into many individual portions, effectively preventing longitudinal mixing. The recent publications on analytical systems that employ unsegmented streams show that continuous flow analysis can be accomplished without air bubbles. The considerations described in the next section indicate that the penalty that is paid in unsegmented flow is either increased reagent consumption or greatly increased pumping pressure.

PUMPING PRESSURE AND FLOW VOLUMES An exact description of longitudinal mixing in unsegmented continuous flow analysis systems is beyond the scope of this publication. Lane and Sirs (IO) have shown how a bolus of an indicator will spread under flow conditions not unlike those in the unsegmented analytical systems. The analytical systems are not exactly described by the equations of Lane and Sirs, which apply only to the spreading of a bolus of dye. There is an extra complexity in the analytical systems, where the color does not form until the bolus of sample mixes with the reagent ANALYTICAL CHEMISTRY, VOL. 49, NO. 1, JANUARY 1977

17

Table I. Reagent Flow Characteristics of Unsegmented Continuous Flow Analysis Systems Q, d, Sampling rate, System Ref. ml/s cm Re sampleslh 2 4

White and Fitzgerald Stewart et al. Ruzicka and Hansen

1.6 0.03 0.3

3

0.6 0.025 0.15

Table 11. Reagent Consumptions in One Manual Method and Two Continuous Flow Methods for the Determination of Phosphate Method Manual Unsegmented flow 225 sampleslh 420 sampledh Segmented flow‘

Reagent, grams per sample Ascorbic acid Molybdate Acid” 0.02

0.05

0.90

0.02 0.01 0.0049

0.03

9.12 0.06 0.031

0.02

0.0028

a Manual method and segmented flow, HzS04. Unsegmented flow, “ 0 3 . Data from Ruzicka and Stewart (11). Ref. 12.

stream and enough time passes for the reaction to take place. Rather than attempting an exact treatment, it will be shown that useful conclusions can be drawn by characterizing the systems in terms of the Reynolds number. The Reynolds number, Re, is given by Re = IpQIrda

(1)

where p is the density of the fluid (g ~ m - ~Q) is , the fluid flow (cm3s-l), d is the diameter of the tube (cm),and 9 is the fluid viscosity (poise). For water and for dilute solutions with a viscosity and density close to the values for water, Re = 127 Qld

(2)

An unsegmented flow is considered to be laminar when Re < 2000 and turbulent when Re 3 3000, with a transition region when Re is between 2000 and 3000. Wall roughness effects, tube curvature, and tube imperfections can all contribute to transition at lower Re. Table I lists the values of Q, d, and Re for the three unsegmented flow analytical systems. The Reynolds numbers are from one-ninth to one-twentieth of the value for transition to turbulent flow, giving evidence that the flow patterq is altered in a favorable way a t Reynolds numbers as low as 150 to 250. Further confirmation can be seen in Figure 4 of the publication by Lane and Sirs (IO), which shows a narrow indicator band at the detector for a Reynolds number of 165, and a peak with extended tailing a t a lower solution flow with a Reynolds number of 2.7. The unsegmented continuous flow systems require some peak spreading to effect mixing of the sample with the reagent. Too much peak spreading reduces the number of samples that can be analyzed per hour, and it also reduces the height of the absorption peak for a given analyte concentration. The length of the tubing must be enough to allow time for the reaction to take place after mixing of the sample and reagent. The data in Table I suggest that the Reynolds number for a successful unsegmented analytical system can be a low as 150. Sternberg and Poulson (6) introduced the concept of “incipient turbulence” to account for their observation that peak spreading is greatly reduced a t Reynolds numbers well below 3000. The systems of White and Fitzgerald and of Ruzicka and Hansen use relatively large amounts of reagent. Table I includes data on reagent use per sample for the three systems, 18

ANALYTICAL CHEMISTRY, VOL. 49, NO. 1, JANUARY 1977

340 150 250

60 100 270

Reagent use, ml/sample 96 1.08 4

and Table I1 lists reagent consumption for the determination of phosphate by the manual method, by the method of Ruzicka and Stewart, and by the equivalent segmented flow method. The reagent consumption in the unsegmented continuous flow method is two- to ten-fold larger than for the segmented flow method. The large reagent consumption is not of consequence when the reagents are inexRensive. In many biochemical analyses, the reagents are costly. Reaction times are frequently longer in these instances as well. If the use of unsegmented continuous flow is to become more prevalent, it will be necessary to decrease reagent consumption and perhaps to increase reaction time. The unsegmented systems of White and Fitzgerald and of Ruzicka and Hansen also consume more sample (approx. 0.5 ml) than is acceptable in current practice in clinical chemistry. If the sample size is to be reduced without loss of sensitivity, the reagent flow must be reduced proportionately. Otherwise, the sample dilution in the flowing stream would be excessive and the absorbance readings would be too small..Sample sizes are nqt so limited in many other important applications of continuous flow analysis. Three groups have independently developed unsegmented continuous flow systems that employ flows with Reynolds numbers of 160 to 340, suggesting that this provides adequate mixing of sample with reagent while preventing excessive peak spreading. Ninhydrin reaction streams in amino acid analyzers, such as that of Spackman et al. (5),also have Reynolds numbers in this range, except for those analyzers that employ segmented flow. No successful unsegmented flow ninhydrin system seems to have been developed that operates a t a lower Reynolds number despite the obvious and valuable improvement in reagent consumption if the flow could be reduced. There is therefore considerable empixical evidence to support Sternberg and Poulson’s concept of incipient turbulence, associated with a Reynolds number near 200. If the reagent flow i s t o be reduced withput lowering the Reynolds number, then the tube diameter must be reduced proportionately in order to keep constant tlie ratio Qld. The reaction time is given by VlQ. V is the Golume contained in the tube, and it is proportional to d 2L, where L is the length of the tube. To maintain the same reaction time, therefore, d2LIQ must be kept constant. Since Qld i s also to be kept constant, d L must be kept constant. At Reynolds numbers in the laminar flow region, the pressure drop in the tube is given by Pojseuille’s equation,

P = 128QLq/rd4

(3)

I t is easy to shqw from this that P is proportional to (Qld) ( d L ) (lld4). As both Qld and dL cannot‘hechanged, reducing the reagent consumption by reducing the tube diameter causes an inordinate ’increase in pumping pressure. A factor of ten decrease in reagent consumption causes a factor of 10 000 increase in pumping pressure.

DISCUSSION The considerations presented here are of help in evaluating the place of segmented vs. unsegmented continuous flow analysis. Unsegmented flow requires either a high rate of reagent consumption or a high pumping pressure. The system

of Stewart, Beecher, and Hare ( 4 ) uses a reagent flow of 1.8 ml/min, or 1.08 ml/sample at the rate of 100 samples per hour. However, this moderate sample consumption is accomplished by reducing the tube diameter and increasing the pumping pressure. Stewart et al. ( 4 )pointed out some of the problems associated with high pumping pressure. They have not been able to add a second reagent to the flowing stream. (Ruzicka and Hansen ( 3 )showed that it is possible to add a second reagent in unsegmented flow systems that operate at low pressure.) The reaction time is also limited, both by peak spreading and by excessive pumping pressure. Stewart et al. reported that the maximum reaction time possible with their system is 120 s, as a longer reaction time would require tube lengths too long for the available pumping pressures. Both White and Fitzgerald ( 1 , Z ) and Ruzicka and Hansen ( 3 ) developed systems with low pumping pressure, but the reagent consumption was large in both cases. The method of analysis used by White and Fitzgerald was photochemical, and the reagent can be reused ( I ) . Ruzicka and Hansen employed relatively inexpensive reagents. When the reagents are inexpensive and the sample sizes are not a limitation, unsegmented continuous flow becomes more attractive. Unsegmented continuous flow appears to have some advantages when it can be used. Segmented continuous flow systems take some time to reach a steady state, especially when the reaction times and tube lengths are long. As was

noted by Ruzicka and Hansen ( 3 ) ,the unsegmented streams reach stable flow almost immediately, and the system can start doing analyses virtually as soon as the pump is started.

ACKNOWLEDGMENT Discussions with Robert Stoy have been of great help in the preparation of this publication. LITERATURE CITED V. R. White and J. M. Fitzgerald, Anal. Chem., 44, 1267 (1972). V. R. White and J. M. Fitzgerald, Anal. Chem., 47, 903 (1975). J. Ruzicka and E. H. Hansen, Anal. Chim. Acta, 78, 145 (1975). K. K. Stewart, G. R. Beecher, and P. E. Hare, Anal. Biochem., 70, 167 (1976). D. H. Spackman, W. H. Stein, and S. Moore, Anal. Chem., 30, 1190 (1958). J. C. Sternberg and R. E. Poulson, Anal. Chem., 36, 1492 (1964). J. C. Giddings, “Dynamics of Chromatography, Part I. Principles and Theory”, Marcel Dekker. New York, 1965, pp 219 ff. V. Pretorius and T. W. Smuts, Anal. Chem., 38, 274 (1966). L. T. Skeggs, Am. J. Clin. fathol., 28, 311 (1957). D. A. Lane and J. A. Sirs, J. fhys. E, 7, 51 (1974). J. Ruzicka and J. W. B. Stewart, Anal. Chim. Acta, 79, 79 (1975). Industrial Method No. 94-70W/B, “Orthophosphate in Water and Wastewater”, Revised Jan. 1976, Technicon industrial Systems, Tarrytown, N.Y. 10591.

RECEIVEDfor review August 5,1976. Accepted October 25, 1976.

Analog Storage Register for Fast Transient Recording T. A. Last and C. G. Enke’ Department of Chemistry, Michigan State University, East Lansing, Mich. 48824

A fast transient recorder, employing a temporary analog storage register was developed. This type of transient recorder can provide a significantly better price/performance ratio than the more conventional digital types over a sampling rate range from 0.1 to 10 MHr. The implementationof the analog storage device is dlscussed, and the performance of the transient recording system is demonstrated. Linearity and accuracy of better than 1% were obtained at recording rates up to 10 MHz.

There are many measurements (e.g., kinetics studies by perturbation techniques ( I ) ) which require the ability to record the amplitude vs. time function of a short-lived and rapidly varying signal. The most popular solution to this problem has been the “fast digitizer” (2) which converts the incoming analog signal amplitude into a digital data word at regular intervals and then stores the digital words from successive conversions in a high speed memory. This produces a digital record of the signal amplitude a t a number of sequential points in time. While this digital recording technique can be inexpensively implemented a t moderate data rates (up to about 50 KHz), the cost increases rapidly as the speed requirements for the analog-to-digital converter (ADC) and digital memory become more demanding. One way to eliminate the need for these high speed components while maintaining a high recording rate capability is through the use of a temporary analog storage register which can read data in at a high rate and read data out at a lower rate. A moderately fast data acquisition system is then ade-

quate to handle the readout rate from the analog storage register. In fact, the output rate from the analog storage register can be made slow enough to eliminate the need for any digital storage other than the computer memory in a computer-based acquisition system. These two recording schemes are contrasted in Figure 1.The transient waveform is clocked very rapidly into the analog memory and then clocked out at a rate compatible with an ordinary ADC. The converted data can then be sent to the computer, at software data transfer rates, where it is stored for later analysis. Since the analog storage register is comparatively inexpensive, the cost of the analog storage transient recorder can be significantly less than a digital storage recorder which has the same recording speed. A transient recorder developed in our laboratory, which utilizes a Reticon SAD 100 serial analog delay element is shown in Figure 2. The timing section of the recorder contains a crystal oscillator which is divided to obtain selectable, accurate data clocking rates up to 10 MHz (the analog register’s approximate maximum data rate). This section produces the frequency shifting necessary for changing from the fast clocking rate (while the data are being stored) to the slow rate (while the data are being read out). The Reticon SAD 100 consists of 100 sequentially addressable storage cells contained on a single monolithic integrated circuit. Functionally, this is implemented with 2 banks of 50 cells each, which are alternately accessed to increase the storage rate. Each storage cell consists of a read in switch (FET),a read out switch, and a storage capacitor. As the ring counter in each bank is sequenced through its possible states, the nth cell is addressed for read in while the ( n 1)th cell is

+

ANALYTICAL CHEMISTRY, VOL. 49, NO. 1, JANUARY 1977

19