V O L U M E 22, N O . 9, S E P T E M B E R 1 9 5 0 tartrate electrolyte of pH 5 to 6 with the potential of the platinum cathode held constant by ail automatic potentiostat (2) a t -0.32 volt versus the saturated calomel electrode. When 50- to 60-mg. quantities of copper were used, the current decreased exponentially from about 500 to 2 ma. over a period of 20 to 30 minutes. In three trials the average calibration factor of the integrator, computed from the weight of silver deposited in thp silver coulometer, was 0.0529 coulomb per count with an average deviation from the mean of +=0.0003 or +0.6%. This value agrees very well with the calibration factor (0.0530) determined with known constant voltages, The particular ball and disk unit used was small, the disk having a radius of only 1.6 cm., and a higher degree of precision probably could be obtained by employing a larger integrating unit. Precision ball and disk units with disk radii of 3.2 and 6.3 cm. (1.25 and 2.50 inches) are available from the Ford Instrument Company, Long Island City, N. Y. It is evident that the inherent precision of the recording potentiometer itself, rather than niechanical limitations, is the factor that determines the ultimate precision and accuracy attainable with this type of integrator. Lag in the recording potentiometer is a possible inherent source of error in cases where the current varies rapidly with time. This can be minimized or eliminated altogether by employing a recording or indicating potentiometer whose speed of response is great compared to the time rate of change of the current. The time required for full scale deflection of the particular Brown Elek-
1221
tronik recorder used in.this study’was 10 seconds. This is amply fast for purposes of coulometric analysis. Although designed specifically as a coulometer, the integrator has other possible uses. It will integrate any curve which the recording potentiometer is capable of drawing. One interesting possibility is the integration of polarographic current-voltage curves recorded over a precisely regulated time interval corresponding to an applied voltage range which includes the polarographic wave. SUMMARY
A mechanical ball and disk integrator controlled by the pendrive mechanism of an ordinary recording potentiometer serves as a convenient electromechanical integrator. The instrument integrates any curve which the recording potentiometer can draw, the integral (area under the recorded curve) appearing as a reading on a revolution counter. The precision and accuracy are better than ~ 1 % .The instrument was designed to replace the more cumbersome classical types of chemical coulometer in coulometric analysis. LITERATURE CITED
(1) (2)
Lingane, J. J., ANAL. CHEM.,21, 497 (1949). Lingane, J. J.. J . Am. Chem. Soc., 67, 1916 (1945): Anal. Chini. Acta, 2, 584 (1948).
R K E I V E D November 19, 1949.
Automatically and Continuously Recording Flow Refractometer GEORGE R. THOMAS’, CHESTER T. O’KONSKI*, AND CHARLES D. HURD iVorthwestern University, Evanston, I l l . S 1946 Claesson ( 1 ) described an instrument for. the con-
1 tinuous analysis of the eluate from an adsorption column,
using the change in refractive index as an indication of change in composition. The essential features were as follows: Light is conducted through a lens system and the emerging beam is passed through two solutions ( A and B) separated from each other in a refractometer cell by a thin glass plate set a t an angle of 45” to the beam of light. Solution B is a liquid of known, constant refractive index, and solution A is the eluate from an adsorption column. As the beam of light passes from A through B, any change in refractive index in A will deflect the beam. This emergent beam is then divided into two beams by passing through a hexagonal glass prism, one edge of which is pointed toward the refractometer cell. At the start the two beams may be equal, but any change in .A will make them unequal. Each beam is received by a photovoltaic cell. The change in the amount of light received by each cell is measured by means of a bridge circuit and a recording galvanometer. This change is directly related to the change in refractive index occurring in A. In connection with some studies to be reported separately, it became necessary to use methods of adsorption analysis. Claesson’s procedure seemed attractive, but it soon became a p parent that his apparatus was needlessly elaborate for the authors’ purpose and prohibitively expensive. The present paper describes a much simpler and cheaper instrument which, however, is comparable in sensitivity and utility to Claesson’s. Attention is called to another continuous recording refractometer (3)designed and built for industrial control use, the action of which depends on intensity of internal reflection, near the critical angle, in a prism in contact with the stream of sample. The 1 Present address, Department of Chemistry, Boston University, Boston. Mass. :Present address, Department of Chemistry, University of California, Rerkdey, Calif.
design of another recording refractometer was also recently reported by Zaukelies and Frost (4) of this laboratory. Their design, an important feature of which was the use of a twincathode photocell, was conceived and developed independently of the authors’ but both came as a result of a talk on frontal analysis given by Claesson a t Sorthwestern University in 1947. The present paper is submitted in view of the importance and potentialities of this analytical approach, and because of the modifications in the optical systems and electrical circuits. The major change (suggested by C. T. O., July 1947) was the use of sensitive phototubes in a balanced circuit to provide high voltage sensitivity and a single vacuum tube in a stable circuit which provides sufficient current to operate a rugged milliammeter, in place of Claesson’s relatively insensitive circuit using photovoltaic cells. This permitted a much more compact instrument because the optical lever arm was reduced from 100 to 10 cm. Another important change was the elimination of the weight-recording device. Claesson’s apparatus recorded the weight of the eluate as a function of the refractive index by means of a spring scale attached to the galvanometer mirror. This has been eliminated by recording the refractive index on a motor-driven chart which moves a t a constant rate. A General Electric photoelectric recording milliammeter was used. A more insensitive Esterline-Angus instrument was tested also in the preliminary work, but only the General Electric instrument was used in the final setup. Inasmuch as constant flow rate is maintained through the column, the chart can be calibrated to read the volume of eluate. Claesson found it necessary to incorporate a complex constant pressure device in his apparatus because with the adsorbent selected-namely, carbon-the flow of solution was too slow. The adsorbents used in the present work, silica gel and alumina, allowed a rapid psssage of the solution and at the same time maintained the necessary selective adsorption characteristics.
1222
ANALYTICAL CHEMISTRY
h minor difference was the use of a stainless steel, righeangle reflecting prism instead of the hexagonal glass prism. DESCRIPTION OF INSTRUMENT
The optical parts essential to the instrument are shown in Figure 1. These include the light sourcc, the refractometer cell, the prism, and the phototubes. Figure 2 portrays the electronic circuit.
t-4I
-
SCALE INCM. 0
5
Figure 1. Refractometer Assembly a. Tungntea lamp
b. Heat-absorbing Elass platCondensing lenses Brass plate for base e. Focusing lens f. Cell 1 g. Cell 2 h. Track E. Reflecting prism j . Phototubes TI, T1 k. Micrometer e. d.
The light source was a 50-candlepower, 6- t o 8-volt, automobile headlight bulb used in conjunction with a set of achromatic condensing lenses (obtained from the Edmund Salvage Co., Audubon, S . J., Stock KO.6179; focal length, 78 mm.; diameter, 46 mm.) and a focusing lens (focal length, 25.4 mm.; diameter, 12.5 mm.). The light source was adjusted in such a way that it focused in the middle of the refractometer cell. The bulb was operated at 6.3 volts from a transformer fed by a Sola constant voltage regulator operating from a 115-volt alternating current line. Although higher operating voltages increascd the sensitivity of thc instrument, the bulbs burned out more rapid1 The refractometer cell was constructed from a brass blocc as shown in Figure 3. The volume of cell 1 was 0.7 ml. The angle between cell 1 and cell 2 was set arbitrarily a t 45O, but could be changed if greater sensitivity were desired. The cells were separated from each other by microscope cover glasses, which were used also as windows a t the ends of the cells. Seepage of liquid around the cover glasses was prevented by use of leadfoil washers. Cell 1 was connected to the vertical adsorption column by means of a brass-to-glass ground joint. The eluate entered through G and made its exit through H. Cell 2 contained the solvent used in the adsorption column.
.
The highly polished stainless-steel, right-angle prism was 2 cm. on each edge. This prism divided the beam into two parts, each of which was received by a phototube. The prism and the two phototubes were housed in a light-tight box, which could be moved in a track perpendicular to the beam by means of a depth micrometer. The circuit, Figure 2, was essentially a cathode-follower circuit in which the grid potential was controlled by the potential existing between the two phototubes connected in series. Two RCA029 phototubes, 2'1 and 7'2, shown in Figure 2, were chosen for the balanced circuit because of their high sensitivity to light from a tungsten lamp. In this type of circuit a large voltage change is produced by a small unbalance of phototube current because of the high dynamic impedance of the vacuumtype phototube. In fact, for the authors' purposes the sensitivity was decreased by inserting two matched 10-megohm resistors, R, and Rj, across the phototubes because stability was the limit on useful sensitivity. By increasing the stability of the optical system and selecting the phototubes and amplifier tube, a ten- or twentyfold increase in sensitivity can be obtained. The single GSJ7 vacuum tube, T,,connected as a triode, served as a current amplifier or an impedance match between the phototubes and the Fecording meter. The circuit was a cathode-follower type u hich proved to be very stable and relatively free from drift. To balance this circuit, switch SI was turned to the position a t which the control grid was connected directly to the junction between B, and &, which were 45-volt B batteries. Then the balancing control, R2, was adjusted until no potential existed across the scnsitivity control, R3,as indicated by a null on the meter, A l . Optical balance was then obtained with liquids in both sections of the refractometer cell. Starting with the sensitivity control near zero, S , was turned to connect the control grid to the phototube circuit and the micrometer screw was adjusted until a null was observed on the meter. Then the sensitivity was increased to any desired value and subsequent changes in the refractive indes of the solution were clearly presented on the recorder tape by suitablv adjusting the micrometer control. Suitable values for k,, ItP, and R3 were found to be 50,000, 5000, and 5000 ohms, respectively. KO thermostat was used, because temperature was not found to be of prime consideration in the present investigation. The concentration of the solutions used was of the order of 1% SO that the change of refractive index of the solution wlth tem-
V O L U M E 2 2 , NO. 9, S E P T E M B E R 1 9 5 0
+
erature approximated the corresponding change for the solvent. ariations in room temperature were transmitted to both cells concurrently through the brass block. Heating by the light source was reduced by inserting- two heat-absorbing_ glass _ plates in the optical system.. Stability of the optical path was of utmost importance, for a distortion of 0.06 mm. would cause full-scale deflection on the milliammeter. In order to acquire this optical stability, the parts of the instrument were secured to a a/,-inch brass plate which,in turn rested on a three-point mount. SENSITIVITY AND CALIBRATION OF INSTRUMENT
1223 sulfide and l,S-dichloro-3,6-dithiaoctane. Straight-line curve8 were obtained for each mixture by plotting percentage of the more strongly adsorbed component (the diethem or the dithia compounds) against LL/&. Ll and L, are the lengths of the lines on the chart of the recording G.E. milliammeter from zero volume to the volumes (VI and V2)a t which points the first and second components, respectively, are just ready to appear in the effluent liquid. For analysis, therefore, one prepares such a curve from known mixtures, then obtains the Ll/L* value for an unknown, and reads off the percentage of A from the curve.
In Figure 4 is plotted the meter reading when the reflecting prism was moved across the beam by means of the depth micrometer mentioned above, Curve A demonstrated that a 0.1-mm. shift in the prism gave full-scale deflection (about 0.2 ma.) on the G.E. recorder at full sensitivity setting ( R sat 5000 ohms) and that the meter reading was a linear function of the prism position. At the lowest sensitivity used (Rg a t about 500 ohms), a 0.23-mm. shift gave full-scale deflection. Curve D indicates that a 1.2-mm. shift in prism position corresponded to full-scale deflection ( 1 ma.) on an Esterline-Angus recording meter. The relationship was not completely linear but could be made so by minor changes in the circuit. With optimum adjustment of the optical path, a shift of 0.06 mm. gave full-scalc deflection on the G.E. meter. By placing solutions of known' refractive index in the cell, it was determined that a change of 2 X 10-3 unit in refractive index corresponded to a 0.2-mm. shift of the prism. Thus, with the instrument as shown, the full-scale sensitivity was conveniently adjustable in the range from 6 X 1 0 - 4 to 1.2 X 10-2 refractive index unit. In the work to be reported ( b ) the G.E. meter was used a t a full-scale sensitivity of 2 X 10-3 unit, which was ample. In a room free from drafts, over-all stability during the rourse of a typical experiment was better than 10-6 unit. Accordingly, a change in refractivebindex of 2 X 10-6 unit waa easily detectable.
Scale lor Curve D 0.9
0.6
1.o
1.4
90
15
10
P
f
5
-10
-15
H
G
- 90
0.1
A
0.7
0.3 0.5 Minimum shmg.r in Piirm Position SCALES FOR CURVES A, 8, and C
Figure 4.
Meter Readings at Various Sensitivities
General Electric meter used for curves A , E , C. Fullsoale deflection equals 50 units o n Y-axis. A . Maximum sensitivity (setting a t 90). RI 5000ohms E . Intermediate gensitivity (setting a t SO) , , C. Lowest sensitivity used (setting a t IO), Ri about 500 ohm# Eaterline-Angus meter used for curve D . with setting at 90. F u l l - e d e deflection eauals 25 units (not 501 on Y-
-
6 Figure 3.
-
axi..
Refractometer Cells SUMMARY
A. Vertical crass section
E . Horizontal cross section C. Microscope cover glasD. Cell 1 E. Cell 2 F. Focusing lens G. Eluate entering cell H . Eluate leaving cell
Experience showed that thermostating the cell alone decreased stability, probably because of temperature differences between the column and cell. Careful jacketing of the entire apparatus would increase the optical stability. With other rcfinemente an ultimate usable sensitivity between 10- and lo-' refractive index unit is considered feasible for. this type of instrument. Minor changes in the circuit also extend the lower range of sensitivity. Utility of this instrument in separating various mixtures of ethers and sulfides by adsorption methods is described in an article ( 2 ) to appear elsewhere. Specifically, these mixtures were studied using alumina and silica gel as adsorbenta with cyclohexane as solvent; butyl ether and 1,sdiethoxyethane; propyl sulfide and lJ!2-bis(ethylrnercapto)ethane; Zchloroethyl
The design and construction of an automatically recording
flow refractometer useful in adsorption analyses are described. Full-scale sensitivity of this unit was conveniently variable between 1.2 X 10-2 and 6 X IO-' refractive index unit. Overall stability corresponded to less than unit, which was the lower limit of sensitivity for the apparatus in the described form. An ultimate sensitivity between 10- and 10-7 refractive index unit is regarded as feasible for this type of instrument. ACKNOWLEDGMENT
The authors are grateful to John Kamper for help in part of the machine work. LITERATURE CITED
(1) Claemon, S.,Arkiu Kemi Mineral. Ceol., 23A,No. 1. 1 (1946). (2)Hurd, Thomas, and Frost, J. A m . Chem. SOC.,72,3733 (1950). (3) Jones, Ashman, and Stahly, ANAL.CHEM.,21, 1470 (1949). (4) Zaukelies and Frost, Ibid., 21,743 (1949). RWEIVEDNovember 1. 1949.