by the presence of the precolumn, and part by the finite time required for the release of the gas. Dissolved gas could only be measured quantitatively by measuring the area under the peaks. The use of a nonautomatic integrator is preferred because the recorder pen does not always return to the same base line after injection or peak tracing. I n such a case a n experienced operator can use his judgment to obtain a t least approximate integrals for areas which might otherwise be meaningless. T o operate at maximum effectiveness the instrument requires certain maintenance. The CaSOI precolumn should be changed after injection of approximately 6 ml. of liquid; the molecular sieve column once for every two changeF, of the CaSOp column. The insulation resistance of the katharometer should be measured periodically; if this decreases below 5 MQ, water vapor h a s entered and the unit must be dried with
hot air. The instr ument may be operated a t room tc mperature, and thermostating is the refore not essential, although some form of draft shield is recommended. Particular care must also be taken in mounting the unit to ensure that it is I lot affected by shocks a n d vibrations. The gas chromatograph was primarily used in a research probh m to follow the transient rates of a bsorption of hydrogen, helium, and cai .bo3 dioxide into water. The dissolved g s s concentration was determinr:d a t a series of times, and checks were I made for traces of other peaks which w( luld indicate the presence of air. The response of the instrument to dissolved gases is governed by the solubility a nd conductivity of the gas in questic m. For example helium-saturated wati 3r gives rise to i t peak area 31% of t h a t for hydrogen-saturated water. Whc ,n gases other t h a n Hz or He are being , determined, a gas of high conductivit. y (such
as Hz or He) should be used as the carrier. ACKNOWLEDGMENT
The authors thank J. H. Purnell of the University Department of Physical Chemistry for initial advice and supply of the katharometer design. LITERATURE CITED
(1) Bovijn, L., et al., Gas Chromatography
Symposium, Butterworths & Co., Ltd., Amsterdam, 1958. (2) Elsey, P. G., AXAL. CHEM.31, 869 (1959). ( 3 ) Kilner, A. A , , Ph.D. Thesis, University of Cambridge, England, 1963. (4) , . Kilner. A. A , , British Chem. Enq. (in press). (5) Petrocelli, J. A,, Lichtenfells, D. H., AN'AL.CHEM.31, 2017 (1959). (6) Ramsey, L. H., Science 129, 900 (1959). (7) Wiseman, W. A , , Ann. S. Y.Acad. S c i . 7 2 , 685 (1959). RECEIVED for review December 31, 1963. Accepted March 24, 1964. Work supported i n part by a research studentship from the Gas Council.
Continuous Automated,, Buretless ' Titrator with Direct Readout W. J. BLAEDEL and R. H. LAESSIG Chemistry Department, University of Wisconsin, Madison, Wis.
b An automated continuous titrator gives direct readout of the concentration of a sample after calibration with a standard sample solution. The measured quantity is the pumping rate of a reagent solution stream, when it is just equivalent chemically to the sample stream, which is pumped a t a constant rate. The characteristics of the titrator are studied for the titration of Fe(ll) samples with Ce(lV). In the concentration range 0.0007 to 0.06M, about 6 minutes are required per sample to titrate 15- to 30-iml. portions of Fe(ll) samples. Relative standard deviations range from 0 . 2 to o.5yO, depending on concentration.
A
titration devices have been known since 1914 (10). Many types have been classified tind reviewed by Phillips (8). For the purposes of this article, titrators may be classified as batch or continuous. -4 batch titrator is one that handles a measured amount of sample and measures the volume of standard solutio11 required to react vgith the whole sample. Most of the automatic titrators described in the literature are of the batch type, with degrees of automation varying from high to low for different versions. However, beUTOMATED
cause of the batchwise nature of ti heir operations, direct readout of the analytical result is generally n o t ea sy, and therefore complete automa tion of the batch titration process t)econ tes difficult. I n any case, a coniplete ly automated batch titrator (from samp le introduction through readout) has nc i t been described within the author' s knowledp e. A continuous titrator is one that meters a sample stream at a constant flow ratc:, and corltrols and measures the flow rate of the reagent sixearn to keep it chemically equivalent to the sample. The reagent stream flow rate is taken to be proportional to the concentration of the sought-for sctbstance in the sample stream. I n principle, at least, a standard sample might be used to calibrate the instrument, making direct readout possible for subsequent samples ( 2 ) . P h i l l i p has reviewed several continuous titrators. One system has been applied to the determination of mercaptans in gasoline by Hallikainen and Pompeo ( 5 ) . The sample and reagent streams flow into a titration vessel equipped with sensing electrodes and an overflow system. The vessel contents are maintained near the equivalence point by controlling the
rate at which the reagent stream is pumped into the titration vessel. The pumping rate is measured and taken as proportional t'o the mercaptan content of the sample. The Milton-Roy Co. has developed a somewhat similar instrument (6). X titrator developed by the Hach Co. is used to determine t,he alkalinity of water samples ( 4 ) . Sample, reagent, and indicator streams flow under gravity through a photometer cell. The reagent stream flow rate is controlled to keep the resulting stream at an absorbance corresponding to the equivalence point. .It equivalence, the reagent flow ratme is proportional to the alkalinity of the sample. Xicholson has described a continuous differential titrator ( 7 ) . The sample and reagent stream are split and pumped into two titration compartments, one of which lags behind the other in concentration. The voltage difference between two indicator electrodes in the two compartments is kept a t a constan't E :elect,ed value by automatically reguli iting the tit'rant flow with a needle alve. The titrant flow rate is read fi om a flowmeter in the titrant stream a td is taken as a measure of sample co ncentration.
1 Zecently the tubular platinum e let %rode (TPE) has been presented as a ice for making electrochemical measNarer nents in flowing solutions (3)
v(
>L. 36, NO. 8, JULY 1964
1617
CONTROL SYSTEM B F L O M f f i STREAM
d t M . F T R o c LINK N E W N K A L ANDrW ELECTRICAL LINK)
Figure 1 . Block diagram of continuous titrator
and applied to the determination of glucose using a differential amperometric technique ( I ) . The low holdup (0.01 to 0.02 ml.) of the TPE makes i t particularly suitable for monitoring or titration systems. This paper describes a continuous titrator that utilizes the TPE as a potentiometric indicator electrode. In conjunction with a reference electrode, a voltage is obtained that deper~dson the concentration of the electroactive qpecies a t the electrode and that may be used to control the pumping rate of the reagent solution. The titrator is unique in that a standard titrant is not required; instead a standard sample is used for calibration. Also, the result is presented digitally, a significant difference from other titrators, in which a computation is necessary to transform the instrumental response into an analytical result. DESIGN OF TITRATOR
Outline and Block Diagram. T h e sample stream is pumped a t a constant rate and the reagent stream is pumped a t a variable rate. After mixing, the resultant stream passes through the T P E and reference electrode. I3y adjusting the speed of t h e reagent pump, the mixed stream can be made to attain the equivalence point composition, as indicated by the potential of the TPE. At equivalence, the reagent pumping rate is a measure of the sample concentration. Since the reagent pumping rate is measurable digitally, the readout of sample concentration can also be obtained digitally. Figure 1 is a block diagram showing the comnonents of the' titrator. Chemical System. The importan components of the chemical systen 1 are the pumps, mixer, and indicatc )r and reference electrodes. The samp le pump i\ a con3tant-speed peristalt ic tylle (Model P.\-56, 56 r.p.m., Nf ,w 13runswick Scientific Co., New BrL Inwick, N. J.). With various si zes of Tygon tubing, flow rates of 15 t 0 0.5 ml. per minute are obtaina bit:.
1618
ANALYTICAL C H E M I ~ T R Y
The reagent purr ,p is a vasiable-speed peristaltic type (. Model 500~-1200,Harvard Apparatus Co., Dover, Mass.) with a feedback control cii-cuit in the power supply t o maintain the pumli a t any set spec :d. The pumping rate of the reagent solution detwmines the ratio of reagen' i to sample in the mixed stream a d h ence the indicator electrode voltage. When the reagent stream is stoichiomei rically equivalent to the sample strea m, as indicated by the electrode vol tage, the reagent pumping speed 1:s d,et .ermined by measuring the rotational speed of the pump shaft. T h e mi.xer is a small Teflon-encased m&gnetic stii-rer sealed in a Teflon chamber, as shown in Figure 2. Entrance a nd exit holes are drilled to receive 7 rygon tubing which is pressfitted in slightly undersized holes after bei ng wetted with cycloliexanone. The tot a1 holdup volume is about 0.7 ml. The stirrer magnet, is driven at 1700 to 2400 r.p.m. from below by a magneti c stirrer (Sargent Co., Chicago, Ill., cat ,alog S o . 576490). T'he mixer is high] ,y efficient and is necessary in order to obtain a stable electrode voltage . W t h o u t the mixer, the electrode 1 :oltage fluctuates greatly in the region of the equivalence point. The indicator electrode is a platinum tube (0.040-inch i.d. x 0.15 inch long), sealed into soft glass capillary a t ea .ch end and connectecl to the flowir ig system by means of Tygon tubing. Th e calomel reference elertrode is also a flowthrough type, p'iaced immedi ately downstream from the indicato, : electrode. Figure 3 shows its cons truction. The electrodtb has a mac hined Teflon body, fitted with a vert ic,al porous glass tube (7-mm. o.d., Cor nirig Glass Works, Corning, X. Y . ) . Thc 3 saturated KC1 solution and the call ,me1 paste are located around the POI 'ous glass tube through which the mr Lin stream of the system flows. C( mt,act with the mercury pool in the b( ittom of the electrode is made through Tygon-enclosed platinum n.ire which ; sealetl into the Teflon. T h e porous F ;lass is sealed into the bottom of the ' dectrode with a Tygon sleeve. The electrode is fitted with a Luvite cover which retards evaporative losses of electrolyte. Leakage through the porous g h s s is negligible. The electrode needs filling about once a month. Indicator System. The indicator system is a potentiometric recorder (Sargent Model SR recorder, E. H. Sargent Co., Chicago, 111.) coupled to the electrode system through a high impedance operational jtmplifier used as a cathode follower (SK2-T' Type, Geo. ,I.Philbrick Re?,earches, Inc., Boston, Mass.). The high impedance coupling system prevents excessive current drain a t the electrodes when the recorder is unbalanced. The recorder is equipped with a 10-turn 10,000-ohm potentiometer used as a potential divider and is adjusted to exhibit the equivalence point potential for the system under study at approximately midscale. Servo-Control System. The servo-
+
Figure 2.
CUTLET
Magnetic stirrer unit
control system monitors the potential of the indicator electrode, compares this potential to a selected potential (chosen to be equal to the indicator electrode potential a t the end point), and operates the reagent pump speed control so that the indicat,or electrode potential is made to correspond to t h e :selected potential. In design of the servo, an important consideration arises from the time required for a change in pump speed to be seen as a change in potential-Le., stream composition-at the indicator electrode. For a fast reaction, like the Fe(I1)-Ce(IV) reaction, the time lag is due principally t o holdup of the solution in the lines and stirrer. Lag due to holdup is independent of concentration. On the other hand, lag may also result from chemical reactions, surface adsorption in the syst'em, or oxide formation or dissolution a t the electrodes; such lag is concent,ration-dependent and was considered of minor importance in the work of t,his report. .It present, more demanding types of titrations are bting studied in which these lags are of greater importance. The adso,rption and oxide format'ion are ccmentration dependent, while the holdup volume is not. IT'hen the system is far from the equivalence point, the servo adjustment may be made coarsely arid continuoudg, so that a minimum amount of time is required to reach the end point. When the system is near t h e end point, the servo adjustment
Figure 3. Flow-through calomel reference electrode
-J
41
I OUTER REGION
OUTER REGION CCONTINWS ADJUSTMENT 1
REAGENT PUMP SPEED Figure 5.
I
Control band concept
Sample of constant composition
’ Figure 4.
0
u’
0
ON OFF ON
Servo amplifier circuit and associated relays
must be made int,erniittently to prevent overshooting of the end point and consequent “hunting” or oscillation. Figure 4 is a scheniatic of the servowhich utilizes the ine potential to activate a shaded pole motor (Figure 6), which in turn adjusts the reagent pump’s motor speed control. The current drain by the SK2-1- input (in the indicator system, Figure 1) is only about ampere. The follower out’put signal is bucked with the selected potential, applied with a mercury cell through a 1000-ohm potentiometer, and the difference volt,age is inverted, amplified, and powerboost,ed by a SKS-S’, SK2-B power amplifier combinatmionin a circuit using a negative feedback loop. The gain of the amplifier-booster stage is approximately 10, though it can be varied by the 1.5-megohm potentiometer in the feedback loop. The bias is in the grounded noninverting input of the SKP-I-, and is adjusted by a 10-turn potentiometer to permit precise adjustment of the bias level. A conventional voltage-regulated power supply (lrodel 510-13, Kepco Labs, Flushing> S.Y.) is used to drive the operational amplifiers. The amplified difference signal varies from positive to negative around zero, depending on whether the indicator electrode voltage lies above or below lected bucking volt,age. The amdifference signal is passed through the activating coils of two high-sensitivity relays in seriw (llicropositioners .IYL%-7304-100, Ihrber-Colman Co., Rockford, Ill.). The 1licrol)ositioner is a poiarity-+ensitive, null-seeking relay, mounted on an 8-pin base. The input pins are 1 and 3. When pin 1 is sufficiently positive wit,h respect, to pin 3, pull-in occurs and 1)in 8 (common)
closes to pin 6. When pin 1 is sufficiently negative with r q i e c t to pin 3, pin 8 closes to pin 2. If the signal is insufficient to cause pull-in, the relay remains open. The 7304 Nicropositioner is rated at, 0.1-volt8pull-in with a power consumption of 14.6 p w . , which corresiionds to a riull-in current of 0.146 m a . The outer microuositioner in Figure 4 is adjusted so tha’t it pulls in to short the shading coils of the shaded pole motor when the amplified difference signal is in t,he outer region (Figure 5 ) , above or below critical values that can be selected by adjustment of the 100ohm potentiometer across pins 1 and 3. \Then the difference signal is within the critical values, the outer micropositioner remains open and the shading coils of the shaded pole motor are not shorted. Then, the status of the inner micropositioner controls the shading coil circuit, The inner micropositioner acts in the same way as the outer, except that the critical values for pull-in are smaller. .Use, the inner micropositioner does not short the motor shading coils directly but instead through an intermittent relay. The intermittent relay has a,n “on” period of only about I second in 12. The “on” time is variable and can be made as long as 3 seconds. Thus, when the system is close to equivalence point, and when the amplified difference signal is small, the shaded pole motor operates for only a short time to make a small change in pump speed and then remains off long enough to allow the indicator e!ectrode to reqiond to the change in concentration. The intermitt,ent relay prevents overshooting and subsequent, hunting about the end point. Difference signals smaller than the critical values required to pull-in the inner relay are within the end point
band, where t,he shading coils of the motor remain unshorted and where the reagent pump speed remains unchanged. The critical values for pull-in of the inner micropositioner are kept small, so that the pumping rate anywhere within the end point band will be within a feu- tenths of 1y0 of the equivalence point pumping rate. The various control bands are depicted in a plot of signal us. pumping rate (for a sample stream of constant composition) shown in Figure 5 . The limits of these bands can be set by adjustment of the amplifier gain and the potentiometers that are in parallel with the micropositioners. Theoretically, of course, the narrowness of the equivalence band is limited only by the gain of the amplifier and the power requirement of the relays. The boundaries of the various bands are selected to fit the characteristics and concentration levels of the samples being titrated. The Harvard variable speed reagent pump required considerable modification for use in the titrator. The original speed control circuit’ built into the pump contained two-stacked, oneturn, wire-wound 5000-ohm potentiometers; these were replaced with two 10-turn precision potentiometers (Model 2201-B, Borg Equipment Division, A\mphenol-Borg Electronics, Janesville, Ris.), which were stacked by connecting t’heir shafts. The 10-turn potentiometers permit>ted very fine control of pump motor speed. The 10-t’urn stacked potent’iometers are turned by the shaded pole servomotor which is controlled by the indicat,or electrode potential. The mechanical linkages be-, tween the motor and potentiometers are shown schematically in Figure 6. The gearing is such that the 10-turn potentiometers can be turned through t,heir range in about 3 minutes. Safety microswitches are used to prevent passing the ends of travel of the 10-turn potentiometers. hlso, a switch is included to pernit manual operation of the shaded pole motor, independent, of the servo amplifier system. Readout System. When the pump rate has been adjusted to maintain the indicator electrode in t’he end VOL. 3 6 , NO. 8, JULY 1964
1619
I Figure 6. motor
Servomotor and potentiometer controls for pump
1. 2.
Shaded pole motor, geared 2 3 6 to 1 G e a r train, net reduction 3.5 to 1 3. Ten-turn 5000-ohm potentiometers, located in pump speed control unit A. Microrwitcher, normally closed 5. Screw-driven carriage, to open microtwitches at ends of potentiometer ranges
point band, the pumping rate is proportional to the sample concentration. T o measure the pumping rate accurately, it was decided to use a digital method, rather than a meter or recorder. T h e method chosen was to convert the reagent p u m p shaft rdtation into electrical pulses and to count these over a precisely timed interval. The shaft of the pump's drive motor (which turns 10-fold faster t'han the pump shaft) was fitted with an aluminum disk 5 inches in diameter and having four 3//32-in~h holes located 90" apart, near the edge (Figure 7 ) , so t,hat' they pass between a photoconductive cell and a 6-volt incandescent bulb located on opposite sides of the disk. n'hen one of the holes in the disk allows light to fall on t,he photocell, an electrical pulse of short duration is produced. The photocell circuit is shown in Figure 8. To secure direct readout, the time interval over which the pulses are counted must be adjustable. Then the pulses obtained from a standard sample can be made numerically equal to the sample concent'ration. The basic pulse-counting syst,em utilizes two preset counters (hiodel
Figure 7. Photoelectric tachometer for reagent pump
1620
ANALYTICAL CHEMISTRY
5424, Berkeley Division, Beckman Instrument Co., Richmond, Calif.). These instruments are basically scalers which can be programmed to count in special sequences. Figure 9 shows the counting circuit schematically. The timing scaler counts line frequency (60 c.p.s.) and is used to count a given number of pulses, after which it closes an internal relay. When the internal relay closes, it shorts the input circuit of the counting
8
0
TO C W M E R
Figure 8.
* Clarex
Photocell circuit
photoconductive cell, CL-403 S
scaler, cutting off t'he input pulses from the photocell. Thus, the function of the timing scaler is to control the interval over which the count'ing scaler accumulates photocell pulses. This interval is precisely adjustable to the nearest MI second. Both scalers are reset through a four-pole normally closed relay (Model KO.L T 334618 a r m a h r e relay, Union Switch and Signal Co., Pittsburgh, Pa.) shown schematically in the normal position in Figure 9. The scalers reset when their circuits are momentarily opened. The four-pole relay is in turn controlled through a cycling relay which resets both counters to zero and then starts them simultaneously. The resetting also opens the internal relay of the timing scaler. The relay system is programmed as follows. (1) The cycling relay causes the four-pole relay to fall out, starting both scalers. ( 2 ) The t,iming scaler reaches the preset' number of counts (of line frequency), a t which time the
internal relay in the timing scaler shuts off the counting scaler, which now indicates the number of photocell pulses that have been accumulated over the timed interval. (3) The cycling relay resets both scalers to zero by opening their reset circuits. (4) Both scalers restart on the fall-out of the four-pole relay, to begin the next cycle. During the titration, the register accumulates counts for a preset time, after which the counts remain fixed in the register for a short t,ime, until the cycling relay resets the system to begin the next cycle. -1s long as the reagent pumli speed is not changed, the same number of counts will repeat' on successive relay cycles. CHARACTERISTICS OF TITRATOR
Precision of Timing Interval. The timing scaler can be preset to the nearest 1,'60 second, which corresponds to 1 cycle of line frequency. Therefore, the variation in a serie;; of 30-second timing intervals d l be less than 1 part in 1800. Over longer intervals, corresponding1~-higher precision could be achieved. Constancy of Pumping Motor. At a particular setting, the ability of the Harvard pump motor to maintain a fixed speed over a period of time was investigated. For pump shaft speeds of 86 and 162 r.p.m.. corresponding to 682 and 1246 count.; per interval of 700 cycles (or 700,60 seconds), 20 consecutive observations a t each speed showed a range of only two counts,
i
D
C K L l f f i RELAY
Figure 9. Cycling readout system
I
U
U
W AV. IO443 COUNTS REL. STD. DEV.=O.I7%
z Figure 10.
0.4d I
I
I
1
Volume delivered per pump revolution
* Read from dial tachometer and roughly proportional to pump speed ** Inversely proportional to volume delivered per pump revolution ( 4 0 counts equal one pump revolution) Conditions: pump tubing 0.073-inch i.d. Silastic tubing used more than 10 hours. Numbers represent order (roughly proportional to time) in which d a t a were taken over 4-hour period indicating the exceptional stability of the Harvard pump motor. The stability over the whole range of pump speeds, 79 t'o 216 r.p.m., was observed to be approximately the same. hpparently the feedback control of the pump power supply is an effective one. The voltage supplied to the pumli was taken from a constant voltage t,ransformer (Model 6115, Raytheon Corp., Waltham, Mass.). Without the transformer, surges in line voltage tripled the above range of observations to about 6 counts per interval. Volume Delivered per Revolution of Pump. T o obtain direct' readout of sample concentrations, i t is essential t h a t the volume of solution delivered per revolution of t'he p u m p shaft' remain constant over the whole range of p u m p speeds. This characteristic of the pump was evaluated by connecting the outlet tube of the pump to a clamped onehole rubber stopper, into which the tip of a volumetric pipet could be inserted. With the pump operating a t steady state, the pipet was inserted and the counts required to fill the pipet between two calibration marks on the upper and lower stems were observed. This was repeated 8 to 10 times for a particular pump speed, over an interval of a few minutes, and the average number of counts was recorded. So long as the pump speed was not altered, the relative standard deviation of a single determination of the counts required to deliver a given volume was very low, only about 0.05%. Figure 10 shows that all determinations are within a relative range of about 0.5%) with a relative standard deviation of 0.1 7y0. Further inspection of Figure 10 indicates a slight dependence of volume delivered per revolution upon pumping rate, and also perhaps upon the operations involved in changing the pump
speed among the different sets of 8 to 10 determinations. This dependence is not very reproducible, as shown by the standard deviations in Table I. At the 1% error level, however, this dependence is unimportant, and it was not investigated further. (The large deviation observed with the 0.030-inch Tygon is unexplained. However, it was noted in taking the data corresponding to Figure 10 that the volume per revolution drifted with time.) Data like those in Figure 10 were taken for pump tubings of various types, sizes, and ages (lengths of prior use), with the results shown in Table I. Tubings of large diameter deliver more reproducibly than smaller tubing. Also, a break-in period of several hours is required for reproducibility. There appears to be no great difference among the different types of tubing. EVALUATION I N TITRATION
T o study the characteristics of the titrator, the well known titration of Fe(I1) with Ce(1V) in sulfuric acid
I
500
I
I
I
I
600 700 800 900 COUNTS PER INTERVAL
Figure 1 1 , Manual titration of Fe(ll) stream with Ce(lV) stream Conditions: 3.4 X 1 O-3M F e M pumped a t constant rate of approximately 5.0 ml./min. 7 . 5 X 1 O-4M Ce(lV1 pumped a t rates varying from 1 A to 31 ml./min. Electrode potential measured against saturated calomel electrode
medium was chosen. The very steep titration curve about the end point makes this system excellent for evaluation of the stability of the titrator and electrodes, because small changes in concentration cause large changes in potential. Reagents. A master stock solution containing ferrous ammonium sulfate (reagent grade, Merck & Co., Rahway N. J.) in 1M sulfuric acid was prepared. Dilutions of this master stock solution were used to prepare subsequent stock solutions from which the individual samples were prepared by weight dilution with l d l sulfuric acid. A similar master stock solution of ceric ammonium nitrate (G. F. Smith Chemical Co., Columbus, Ohio) was prepared in 1.12' HzS04. Reagent solutions a t various concentrations were prepared by dilution with 1111 H2S04. Manual Titration Curve. SELECTION OF TITRATION CONDITLONS. Figure 11 is a plot of electrode voltage (as read on the recorder) against cerate reagent flow rate (read a s counts accumulated over a fixed interval of 7.5 qeconds). For this plot, a 3.4 x 10-3JI Fe(I1) sample
Table I. Volume Delivered per Pump Revolution Relative Tubing specifications standard Number I.D., Prior use, Average counts deviation, of Branda inch hours per pipetfulb 7% trials Silastic 0.073 2 10810 (50 ml.) 0 45 41 10699 (50 ml.) 0 073 10 0 10 55 0 073 >10 10443 (50 ml.) 0 17 68 0 040 8 6000 (10 ml.) 0 39 95 Acid flex 0,083 10 9364 (50 ml.) 0.25 68 0.056 5 8926 (25 ml.) 0.52 127 0 056 >10 8282 (25 ml.) 0 12 95 Tygon 0 030 2 0 21 10 14103 (10 ml.) a Silastic (Technicon Corp., Chauncy, N . Y . ) is a silicone rubber tribing that is more pliable than Acid-Flex or Tygon. Acid-Flex (Technicon Corp.) is designed to withstand concentrated sulfuric acid. Tygon (Formulation S22-1, U. S. Stoneware Co., Akron, Ohio) is an acid-resistant type b Inversely proportional to volume delivered per revolution of pump.
VOL. 3 6 , NO. 8, JULY 1964
1621
I MIN
c _ _
h
F
E
n w
v
Figure 12.
D
I"
I/
Record of servo-controlled titration
Sample stream contains 4.3 X 1 O-3M Fe(ll1 pumped a t about 5.0 ml./min. Reagent is 7.5 X 10-4M Ce(lV) pumped a t 2 8 rnl./min. a t end point. Width of end point band 0 . 0 8 volt, corresponding to 4 counts or obout 0.5%. Counting interval approximately 7.5 seconds
solution was pumped a t about 5 ml. per minute through a fixed-speed New Brunswick peristaltic pump and a 7.5 X 1 0 - ~ mCe(IY) reagent solution was pumped a t manually selected speeds, giving flow rates ranging from 14 to 31 ml. per minute. I n taking the data, the 1000-ohm potentiometer across the bucking cell was set to zero and the switch (S1, Figure 4) was closed to ground. This inactivated the micropositioners and t,he servomotor. Data like those presented in Figure 11 are necessary to select the titration conditions for a particular type of titration. The end point potential, selected a t the steepest slope, is 0.70 volt arid represents the value to which the bucking potential must be set. The end point band and the inner band limits must be selected for the particular titration under study. The outer limits of the inner band must be set aide enough to keep the syst.em from overshootiiig and oscillating between the two limits, yet narrow enough t'o keep the system from going into intermittent control too early, thereby consuming too much time in reaching the end point. The end point band limits must be set wide enough to prevent overshoot and oscillation between the band limits, but narrow enough for high precision. I n short, the band limits must be set to optimum values that depend upon the steepness of the titration curve-i.e., the solution concentrations-and upon the precision desired. The limits may be determined by a simple trial and error adjustment. By using the titration curve for the system under study (Figure l l ) , values of potential can be chosen for the band limits. To set the upper limit of the end point band a t 0.76 volt, 0.06 volt above the end point, S1 is closed to ground, and the 1000-ohm bucking potentiometer is increased t'o make the voltmeter ( V ,Figure 4) read 0.76 - 0 . 7 0 or 0 . 0 6 volt. With this signal applied to the amplifier input, the 250-ohm potentiometer is adjusted to make the inner micropositioner just pull-in. The lower limit of the end point band is automatically fixed by this procedure also, and falls at a value of 0.68 volt shown in Figure 11. (Positive and negative signals required for the pull1622
ANALYTICAL CHEMISTRY
in of the micropositioner are not identical.) The upper limit of the inner band, corresponding to continuous control, is set in a similar way. The value is not critical, and is usually chosen to fall near the beginning of the steep portion of the titration curve. After adjustment of the band limits, the 1000-ohm bucking potentiometer is set to make the voltmeter ( V , Figure 4) read 0.70 volt (the end point potential), and S: is closed to the follower output. The system is then ready for servo-controlled titrations. d selector switch (SpI Figure 4) a t the follower input permits the operator to select either the electrode potential or a variable standby voltage. The standby source is used when warming up the amplifiers, clearing the system of bubbles, flushing the system with water, etc. The gain of the amplifier was made variable for experimental purposes. For the work of this report, however, the gain was fixed a t 10. An alternative method of setting the band limits is to calibrate the potentiometers which are in parallel with the micropositioners for a givena mplifier gain. The selection of band boundaries can then be made from these calibrations. Performance of Servo-Controlled Titrator. Figure 12 shows a recorder chart record for the automatic titration of 4.3 X 10-3J2' Fe(1I) with 7.5 x 10-4JF Ce(1T'). The bucking potential and the band limits were set as shown in Figure 11. At point -4, the sample inlet tube was removed from the previous sample and inserted into the sample under analysis. An air bubble was drawn into the sample line during the transfer, giving a cerate-rich stream and a high potential a t the electrode, region B. During the cerate-rich interval, the servo directed a continuous reduction in the cerate flow rate, giving a ferrous-rich stream, region C, after normal sample flow resumed. I n region C, a continuous increase in cerate flow rate was directed for a short time, followed by an intermittent increase, resulting in a slight overshoot in region D. After correction, the recorder potential remained in the end point band with only two small adjustments in regions E and F . Steady state in the end point band was reached within about 5 minutes after sample introduc-
tion. The values of pumping rate that were read out are transcribed onto the figure to show the steadiness of the readout. Once the system had achieved steady state (beyond point D), the time interval over which the counts were accumulated was adjusted on the timing scaler so that the counter would accumulate a number of counts numerically equal to sample size (or a simple multiple of it), in this case moles of Fe(I1) per liter of sample. This adjustment was made easily and quickly by a simple slide rule calculation or by a trial and error adjustment. Once the system was calibrated in this manner, successive sa-ple concentrations could be read out directly in Fe(I1) molarity. The accuracy of using this direct readout procedure depends upon the linearity of the relationship between pump speed and concentration of the samples and also upon the scatter of individual points about the linear value. Figure 13 shows the linearity of the response and the scatter. For Figure 13, the calibration was made by adjusting the counting interval to give 850 counts for a 4.25 X 10-3All Fe(I1) solution (readout numerically equal to twice the molarity of the sample), with the calibration sample designated by the hollow point on the diagram. The straight line from the calibration point to the origin is the assumed working curve when the direct readout procedure is used. The solid points represent titrations of various Fe(I1) solutions ranging from 0.0024 to 0.0049Jf. The error of the direct readout procedure is expressable as the standard deviation between the experimental points and the straightline working curve. For the data in Figure 13, the relative standard deviation is 0.43%. The largest error is 0.950/,, while the average error is 0.37%. These errors are approximately in agreement with the tolerance alloi%ed in setting the end point band limits
r
25
I
I
3.0
35
MOLARITY Fa
I
40 (1) x IO3
I
45
Figure 13. Dependence of response on sample size Fe(ll) samples 0 . 0 0 2 4 to 0.0049M, pumped a t 5.0 ml./min. Ce(lV) reagent concentration, 7.5 X 1 0-4M. Counting interval 4 5 4 cycles, about 7.5 seconds
T o evaluate the titrator for various sample concentrations, a series of reagent Ce(IV) solutions ranging from 0.0100 to 0.0001Jf was prepared (Table 11). Each reagent Ce(IV) solution was used to titrate a series of Fe(I1) samples that ranged over approximately a twofold range of concentrations. Series 1 to 8 were made with the large diameter Silastic pumping tubes (0.073 inch) a t Ce(IV) reagent flow rates ranging from 14 to 31 ml. per minute. The large reagent flow rates were used to keep the pumping error to a minimum. With smaller pumping tubes, the Ce(IV) reagent flow rates may be decreased. Series 9 and 10 were performed with 0.040-inch Silastic tubing in the Harvard reagent pump and 0.030-inch Tygon tubing in the New Brunswick sample pump. For these lower reagent flow rates (6 to 13 ml. per minute), the record corresponding to Figure 12 showed larger fluctuations and more frequent excursions from the end point band, which explains the lower precision. I n series 11, the same low reagent and sample flow rates were used as in series 9 and 10, but a third stream of supporting electrolyte (1M HPSOI) was pumped into the system just before the mixer. For series 11, the electrode potential records looked much like that of Figure 12. The data in Table I1 illustrate the value of working a t high flow rates. The improved precision is explained by the smaller pumping error associated with the large diameter tubing (Table I), and also by the greater electrode stability at high flow rates, giving fewer excursions from the end point band. If a great premium is attached to low reagent and sample flow rates, they may be used without much loss in precision, providing that a third stream of inert electrolyte is used to maintain a high flow, rate in the floivirig system. Fluctuations at Equivalence Point.
Small fluctuations in pumping rate a t the end point cause fluctuations in the electrode potential a s shown in Figure 12. At low concentrations, where the titration curve is not steep, t’hese fluctuations penetrate the end point band limits infrequently, and the servo system operates only infrequently after the end poiiit is reached. At higher concentrations, where the titration curve is steeper, the fluctuations increase. Excursions from the equivalence point band occur more frequently, and the servo system operates more often. At reagent Ce (IV) concentrations much above O . O l M , the fluctuations for the previously described system become so large that the system oscillates about the end point hand. At times the fluctuations are
Table II.
Series 1
2 3 4 5” 6 7 8
A
10 11b
Evaluation of Titrator
Ce(1V) reagent molarity ( X lo4)
Range of Ce(1V) flow rates, ml./min.
Range of Fe(I1) sample molarities ( X lo4)
100 50 10 8.8 7.5 5.0 2.5 1.0 100 10 100
14-3 1 14-3 1 14-31 14-31 14-31 14-31 14-31 14-31 6-13 6-13 6-13
270 to 630 130 to 310 27 to 63 23 to 58 21 to 49 13 to 31 6 . 8 to 17 2 . 7 to 6 . 3 190 to 430 19 to 43 190 to 430
~~
~~
Number of Fe(I1) sample Fe(I1) flow rate, samples rnl./min. titrated 5
5 5 5
25 13 12 25 16 21 14 21 26 10 20
Relative standard deviation,
c7c
0.2 0.3 0.4 0.3 0.4 0.5 0.4 0.9 0.4 0.5 0.3
16 titrations shown in Figure 13. In addition, a third stream of 1M H2SO4was introduced just upstream from mixer. A 23 r.p.m. New Brunswick pump was used with 0.081-inch Tygon tubing, giving a flow rate of 7 rnl./min. a
sufficient to penetrate the outer band region because, at high concentrations, small impulses of the servo system can change the potential sufficiently t o traverse the whole inner band. For titrations with concentrated solutions the pumping rates must be decreased, or the ratio of on-to-off time in the intermittent relay must be decreased. The concentration regions above 0.01.Id were not investigated further. At one stage in the work, regular fluctuations with a 6-second period were observed. These were traced to insufficient tension on the yoke of the constant speed New Brunswick sample pump, and they disappeared when the tension was increased. N o other artifacts of this sort were observed. Modifications of Titrator. T h e main purpose of this work is to illustrate and evaluate the application of the continuous method with its direct readout to the titration process. There is no pretense t h a t t h e equipment described is optimally designed in a n y respect, and it is not claimed to be a “practical” titrator. On the contrary, the form that a titrator would take would depend upon the relative importance assigned by a particular user to various noncompatible titration objectives: high precision, small sample size-Le., low flow rate or dilute solutions-speed of titration, and simplicity of equipment and procedure. For the system described was designed with 8. rather complex
automated continuous titrations may be performed at the 0.3% errcr level. On the other hand, if a 1% standard deviation were tolerable, the system for measuring and reading out the pump speed could be greatly simplified, the sample size-Le., pumping rate-could be greatly reduced, or the time of titration diminished, whichever the user considered to be of prime importance. ACKNOWLEDGMENT
Appreciation is due to Carter L . Olson for preliminary discussions leading to the concept and early design. The aid of the technical staff of the university, particularly Robert Schmeltzer and Harland Bright, is gratefully acknowledged. LITERATURE CITED
(1) Blaedel, W. J., Olson, C. L., ANAL CHEM.36, 343 (1964). (2) Blaedel, W. J., Olson, C. L., J . Chem. Educ. 40, A549 (1963). (3) Blaedel, W. J., Olson, C. L., Sharma, L. R., ASAL. CHEM.35, 2100 (1963). (4) Hach Co., Ames, Iowa, “Hach CR
Automatic Analyzers.”
(5) Hallikainen, K . E., Pompeo, D. J., Instruments 25, 335 (1952). ( 6 ) Milton-Roy Co., Philadelphia, Pa., Data Sheet 230. (7) Nicholson, M. M., ANAL.CHEW33, 1328 (1961). ( 8 ) Phillips, J. P., “Automatic Titrators,” Academic Press, New York, 1959, ( 9 ) Ross. J. W.. Shain. I.. . ~ N A L . CAEM.
VOL. 36,
NO. 8, JULY 1964
1623
