ANALYTICAL CHEMISTRY
1236 approaches the theoretical minimum for the instrument. In the course of most analytical procedures it is necessary to remove the cell carrier from the instrument, and refill the individual cells many times. In such cases it is necessary to keep these mechanical factors in mind if reasonably high precision is to be attained. After considering the nature and size of the variations observable within the Beckman spectrophotometer, it seems allowable to draw certain conclusions concerning its use in chemical semimicro procedures ( 7 , 1 7 ) .
If the instrument is to be used only to yield relative valuesLe., to compare a series of unknowns with a set of comparable standards as in many routine analyses-very good and consistent results can be expected, as far as instrumental error is concerned. A variation of the order of u = 0.57, is reasonable, and in certain cases lower values may be reported. If extinction coefficients are to be used as a basis for calculation, as is common in vitamin h methods, for example, or if the final result is to be expressed in absolute terms, considerable caution must be used. Re eatability and error are far from identical in such a case. One slould be slow to attribute significance to discrepancies of 5% or less in the values reported by different laboratories or even between different sets of absolute values obtained within the same laboratory a t markedly different times. In studying apparent deviations from Beer’s laiv it must be definitely established that the nonlinearity observed in the work does not arise from the instrument. In all careful work special precautions should be taken to rule out the possible effect of turbidity. ACKXOWLEDGMENT
The author wishes to express his thanks to Howard Alexander, Mathematics Department, Adrian College, and Warren Gilleran, radio technician, for their comments, and to Olaf Mickelsen, chief chemist of the Division of Chronic Disease of the U. S. Public
Health Service, for his suggestions and encouragement in this work. LITERATURE CITED
(1) Alexander, H., and Caster?W. O., in preparation. ( 2 ) Anderson. R. L.. J . Am. Statist. Assoc.. 42., 612 ,(1947). , Ayers, G.’H., ANAL.CHEM,, 21, 652 (1949). Bastian, R., Ibid., 21, 972 (1949). Bastian, R., Weberling, R., and Palilla, F., Ibid., 22, 160 (1950). Beckman Bull. 91-D, Kational Technical Laboratories. Bessey, 0. A,, Lowry, 0. H., Brock, J. J., and Lopez, J. -4., J . Biol. Chem., 166, 177. Cary, H. H., and Beckman, A. O., J . Optical SOC.Am., 31, 682 (1941); Beckman Bull. 144. Caster, W.O., and Mickelsen, O., Federation Proc., 8, 190 (1949). Ewing, G. E., and Parsons, T., AXAL.CHEY.,20, 423 (1948).
Hanze, A. R., Conger, T. W.,Wise, E. C., and Weisblat, D. I., J . Am. Chem. Soc., 68, 1389 (1946).
Hawes, R. C., National Technical Laboratories, personal communication. Hiskey, C. F., - 2 s . k ~CHEY., . 21, 1440 (1949). Hogness, T. R., Zscheile, F. P., J r . , and Sidwell, A. E., Jr., J . Phgs. Chem., 4 1 , 3 7 9 (1937). Kemmerer, A. R., J . Assoc. Ofic.d g r . Chemists, 29, 18 (1946).
Kincaid, G., Kational Technical Laboratories, personal communication. Lowry, 0. H., Lopez, J. d.,and Bessey, 0. A., J . BioZ. Chem., 160,609 (1945).
Muller, R. H., IND. EKG.CHEM., Oroshink, W., J . Am. Chem. SOC.,67, 1627 (1945). Rawlings, H. IT., and Wait, G. H., Oil and Soup, 23, 83 (1946). Snedecor, G. W.,“Statistical hlethods,” 4th ed., Ames, Iowa State College Press, 1946. Twyman, F., and Lothian, G. F . , Proc. Phys. SOC. (London), 45, 643 (1933).
Vandenbelt, J. hl., Forsyth, J., and Garrett, A . , IXD. ENQ. CHEM., A4NAL.ED.,17, 235 (1945). Wilke, J. B., J . Assoc. Ofic.Agr. Chemists, 30, 382 (1947). RECEIVED March 30, 1950.
Automatic Countercurrent Distribution Equipment LYMAN C. CRAIG, WERNER HAUSMANN, EDWARD H. AHRENS, JR.,AND ELIZABETH J. K4RFENIST The Rockefeller Institute for Medical Research, New York, iV. Y. For several years it has been obvious that an extraction column capable of applying several hundred equilibrium stages or their equivalent would be a very useful tool, particularly for separation of the complex mixtures of rather poor stability so often encountered in biochemistry. Attempts to develop such an apparatus have culminated in the equipment described here. The apparatus is a strictly
T
RUE progress in elucidation of the structures of the more
complicated substances of biological interest can be made in the sense of classical organic chemistry only when experimental means are provided for the isolation of single chemical individuals irrespective of their complexity. KO less a problem is the means at hand for recognizing whether or not a single individual compound actually has been isolated. Even though certain physical measurements, such as those involving phase rule study, are very helpful, if not decisive, in this regard, final evidence of purity depends to a great extent on strenuous fractionation attempts with the most favorable separation procedure and failure to achieve further resolution. It is therefore imperative that constant efforts be made to improve the separation tools available and that as many as possible different separation methods of high resolving power be brought to bear on a single problem. At the outset, it is granted that purity in the absolute senst can never be proved or achieved (8). In fact, a few per cent of an
discontinuous extraction train containing 220 glass equilibration cells. The train is operated by automatic equipment. In 24 hours 800 equilibrium stages (roughly 150,000 extractions) are obtainable. As a “recycling” feature permits several thousand equilibrium stages to be applied, the question of “purity” can be decided with a considerable degree of certainty.
impurity, especially when it is known to be present, may have little bearing on the problem of structure. Sonetheless, it is always desirable to decrease the likelihood of significant impurity as far as is experimentally possible. Because of the steady progress in the improvement of methods for separation of mixtures, this requires continual re-examination of the ”purity” question. Bmong the available separation methods, countercurrent distribution offers a method unique in the sense that in actual practice for most solutes it can be operated so that theorv and practice coincide to a high degree. I t is strictly a discontinuous process. Other countercurrent processes of high resolving power are continuous in nature and are generally interpreted in terms of the ideal discontinuous process. Our understanding of them is in theory only, is partly based on analogy, and involves certain basic assumptions which may or may not hold in actual practice. Such an uncertainty implies, as far as the problem of purity is concerned, that the sample may not have had as high a number of theoretical stages applied (or perhaps higher) as calculated.
V O L U M E 23, NO. 9, S E P T E M B E R 1 9 5 1 Thus far the chief disadvantage of countercurrent distribution has been the labor involved in the application of high numbers of transfers to a given separation. Recently this disadvantage has been overcome by the development of fully automatic equipment and a type of equilibration cell which can be assembled to form an extraction train of any length desired.
1237
r i t e rtt position B. Upon t & i g further to position C, the upper phase decants through c to d. The length of the lower part of the cell is adjusted a t the time of construction, so that with a 10-ml. lower phase the meniscus separating the two phases only reaches point a. When the cell is tilted hack to position A, the contents of d flow throuEh e into the next adjoining cell of the series.
decan&ion tube will permit, hut the former is placed sightly to the rear, so that the stoppers closing the lower tubes are readily accessible
Figure 1.
Complete Extractor
The apparatus has been in satisfactory operation for nearly a year and has proved very useful, although certain small details have been improved from time to time. It permits several thousand acturtl equilibrium stages (220 extractions per stage) to be brought to hear on the problem of the purity of a given preparation. At the same time it has sufficient capacity to serve as a preparative tool for small-scale structural investigations. DESCRIPTION OF THE APPARATUS
The glass case shown in Figure 1 houses an extraction train containing 220 equilibration cells joined glass to glass with no rubber or plastic connections. The case, which has a stainless steel frame, protects the glass train. Automatic filling devices supply solvents to the train and an automatic fraction collector collects the effluent phases, The rear end view in Figure 2 shows these parts. A mechanical robot energizes the driving motors for the movements of the cells and fraction collector and thus controls the entire operation.
Figure 2.
Filling En,d of Extractor
3 . , 1 The cells are held on the metal bar ~y means 01, , tnm mew straps of stainless steel attached to small bolts. The straps pass over the glass tubes and press $hem against the Duralumin bar. The bolts pass through the hitr to nuts OF the other side. Notches have been cut in the bar spaced to receive each cell and hold it in position
Eqliilibration Cells. The cell shown in Figure 3 is made from 15-mm. glms tubing and is approximately 32 ern. in length. The t,hree t,ubes attttached to it are made from 7-mm. alas8 tubing. ~~~. The overflow tube, e, is 10 em. in length. It passesxhrough tiie lower wall of the decantation compartment, d , a t a ring seal and ~
~~~
~~
~~
next eauilihration cal of the s’eries or ioined t o it by a flat aroundglass jdint. Access to each equilibration cell is permitted through b, which is a b t ground-glass stopper designed to close f. The opening tube. f. is 4 om. in length leneth on the Dresent tube, f, present aDuaratus. apparatus. However. However, exmrience exm%ne;lce now suggests sugges& that a length of 8 cb. em. would be prefer-
~~~~~” ~~~r~ ’ time is required’fiithe opening or closing operation. The two uhases in the equilibration cell are brought to equilibrium by rocking from poiition A to B and back again. Each ......
Figure 3. Equilibration Cell Design I n the apparatus currently in use all the cells are sealed toaether through each overflow tube, e, except that a t every tenth Fell they are-joined by flat glass joints. The cells can therefore he handled and removed from the bar in hanks of 10. Less o p partunity for leakage is thus presented and it is necessary to use
ANALYTICAL CHEMISTRY
1238 only two metal straps to support the ten tubes to the Duralumin bar. On more recent commercial models each cell carries a joint' and can easily be replaced in case of breakage. The cells are numbered 0,1,2,3, . . . 109left to right beginning with the upper row. This is the direction of flow of the upper phase. The exit tube from the decantation compartment of the last cell in this row is extended so that the solution flows into the cell of the lower aeries directly below it. The exit tubes of the lox-er decantation compartments are turned so that the flow of
.
,..
,
I
Figure 5.
Figure 4.
When the apparatus is tipped to the decantation position as indicated by the dashed lines, I t fills throu h the opening, b. The contents of the dipper then empty throu& d into a chamber, e , when the apparatus tips to the transfer position. The solvent remains in e until the apparatus again reaches the decantation position. At this point it is hi her than the train and flows into a decantation compartment p7aced in front of cell 0. On the next transfer it empties into cell 0. The di per fills again as e empties. An air vent, c, is inserted to avoif air being trapped. During the equilibration period the empty dipper simply oscillates above the level of the solvent. The constant level trough, 8, is made from a flat bottle approximately 10 x 5 cm. in width and breadth. A hole is ground through one side for the exit tube from the dipper to pass through. The bottle is cut in two p a r k at h and the cut surfaces are ground flat. The upper half of the bottle serves as a covering to prevent evaporation of volatile solvents from the trough. The liquid in g is maintained at a constant level by means of a siphon through the top of Cocurrent Dipper the bottle which leads from a aliter roundDesign bottomed flask serving as a reservoir. The reservoir carries a straight g l w tube through its rubber stopper in addition fa the siphon.
Filling Dipper Design
the upper phases is from right to left in this series, Hence the lower cells are numbered 110, 111, 112. . . 219 right to left and cell 219 is directly below cell 0. The decantation chamber from cell 219 is placed on the upper bar alongside that of cell 0. For this cell the exit t u b e must be much longer than for the others, but the arrangement permits the upper phases finally to flow back to their starting point over cell 0. Here they can either be d r a m off through an exit tube leading to the fraction collector or allon-ed to flow back into cell 0 for another circuit through the series. The two Duralumin bars are joined to crombars at each end. A shaft which passes through a bearing is attached to each crossbar at a central adjustable position, so that the load of the Duralumin bars and attached cells is n e a r l y c o u n t e r balanced. Adjustable weights are attached at each end t o rods extendiiig from the crossbars. These weight. can be moved for final counterbalancing when the apparatus is in operation
Filling Device. At each transfer a portion of the upper phaae is required for cell 0. It is supplied by a simple dipper of the desired size operating off the end of the bearing shaft, as shopn schematically in Figure 4.
~
~
Figure 6. I.
11.
~~~~
Side view of driving wheel view of wheel
Top
~~
~-
~
Schematic Drawing of Part of Robot III. IV.
Side view of rack and clutch Top
view of rack and clutch
V O L U M E 23, NO. 9, S E P T E M B E R 1 9 5 1
The Robot. The niovements which operate the extractor are relatively simple tipping movements designed to place the cells in the positions shown in Figure 3. Nonetheless, the design no^ in use is somewhat complicated, in order to give versatility and perinit each movement and interval to be adjusted independently of t,he others. Detailed drawings of the various parts of the robot are given in Figures 6 and 7. Each figure gives several views. In Figure 6 two side views, I and 111,and two top views, I1 and IV, are given. .4n end view, I, is given in Figure 7. -4 drawing of tbe “coinbiner” i. given in 11, Figure 7, and I11 and IV are side ‘and erid views, respcctively, of the timer.
,?f
I
Figure 7.
f4
Schematic Drawing of Sections of Robot
I. End view 11. combiner
1239
111. IV.
Side view of t i m e r End view of t i m e r
The glass tube permits entrance of air and the height of its lower end controls the level of the liquid in 9. It is thup the well-knon-n constant-level device. Provision in the filling device is also made for feeding in a sinall amount of the heavy phase to correct for phase distortions and any small loss of that phase throughout the train. This feature amounts to another dipper which delivers a much smaller volume of the heavier phase each time a transfer is made. Usually oiily a few drops of the heavy phase at each transfer are required to maintain the decantation levels throughout the train. Therefore a large rewrvoir is not necessary.
-4500-mI. round-bottomed flask, A of Figure 5 , is mounted above cell 0 with its neck in a horizontal position parallel with the aluminum bar wpporting the cells. In this position 150 ml. of the solvent vi11 not spill from the open mouth of the flask, even though the liquid is thrown back and forth during the equilibration. A plug of cotton in the mouth reduces eva A bent, s m n l l glass tube, B , passes through t r k % h of the round-bottomed flask and leads to the decantation compartment of cell 0. kea^ its inner opening is blown a bubble, C, of the volume dwired. Just outside the opening of the flask, the glass tube carries one of the fIat interchangeable joints, D,attached eo that the height and position of the dipper can be adjusted for filling and emptying at the desired position. When a larger volume of mcurrent is required, a tube with a larger bubble can be interchanged at the glass jomt. The straight portion of the tube, which passes through the opening of the flask and carries the joint, is set at an angle so that it will drain into the decantation Compartment of cell 0 in the decantation position. The bubble fills again in the transfer position.
1 solid iron wheel, I, 22 em. in diameter and 1.7 cm. thick. is niounted on the drive shaft, 2, of the reduction gear of t,he electric motor. Two circular tracks, A and B , which merge a t one place, are cut into the wheel. Each track is 12 mm. wide and 9 mm. deep. The circle of the larger track. A, is 17.2 em. in dimieter; B is 8.6 em. in diameter. .hi :tluniinum bar, 3 approximately 30 O I ~ .in length and 2.5 cni. in width and breadth, acts as a. lever u r n to drive the rack, 4. The effertive length of the arm and hence of the stroke is adjustablo by virtue of the bolt and sliding wction near the rack bearing. On its o t h ~ end, r t,he bar carries a side L extension, 3a, in vienI1 of Figure 6, so that the bar can be attached by two bearings to a rigid hroad piece of aluminum metal, 5 , 18cm. in height, 10 em. thick, :tnd 2.5 em. wide. Part 5 is in t,urn attached by screws to the b a ~ plate, r 6, of the robot. The base, 6, is a n aluminum plate 1.3cm. thick, 46 em. long, and 23 cni. wide. -%tn~:trlythe central part of the arm, 3, is attached a pin held in the ;trni by two roller bearings. The larger end of tho pin or shaft c.st,erids into the track in 1 and rolls on the sides of the track as wheel 1revolves. This drives the arm up snd down. The two bearings by which 3 and 3a are attached to 5 prevent thc arm from moving away from the wheel. The roller remains in the smaller track, B , for the equilibration period. It is shifted to the larger triick, A, in order to reach the higher position required for the dccaiitation, C‘ of Figure 2. This pmition is reached whcn wheel 1 is stopped 180’ from the position shown. The settling position, B of Figure 2, is that shown in view I of Figure 6. The transfw psition, d of Figure 2, is reached when the rollvr is in the smaller track, B, and the wheel is stopped 180” froin t h e position shown in I. The movement of the arm is transmitted througb the ruck, ?, to the spur gear, 7, shown in view I11 of Figure 6. This gear ls mountid directly on the rod, 8, which wrvw aa the bearing on n-hich the aluminum bars carrying the extraction cells :ire mounted. Thc rack, 4,can be disengaged from the Lipur gear, 7, in order to permit hand operation of the cells. It is held against the spur gear by :t broad brass piece, 9, which extends to the o p p s i t t . edge of the spur gear and on both sidw of it, as ahown in view IV. The shaft, 8, passes through a short slot set at an arigle in 9. Part 9 carries two roller bearings, 10, which press against the hack of the rack and normally keep the teeth engaged. Part I 1 together with 9 forms the bearing around 8. The forinel, is attached to 9 itt its right h m d end by means of a short h ~ l t so . that it is movable. The other left-hand end of 11 is held iii Imition a t the lower edge of part 9 by a flat spring on 11 and :I pin which maps into the hole shon-n in 9. S e a r the upper left-h:tiid corner of 9 another hole is provided, so that the arm of 11 c a n be moved to it. This movement forrrP 9 :tlong its slot and to the right a. sufficient distance to disengage the rack from the .spur gear. The riuinber o f strokes per equilibrutic,n is adjustable froiii 5 to 50 by in(:ans of the small spur gear, 12, which is mountcvl on t,he aluminum piece 5. It is 8 cm. in diameter. The L cstension of t,lielever arm, 3a, extends beyond the bearing on 5, so that a push rod, 13, can be attached. The push rod moves a lnetal part, 14, which is held against’ 12 by n spring and is adjust:ilde by a smew in 13. Each time a dowin-ard stroke of 13 is niade during the equilibration 12 is advanced one tooth; 12 is spring loaded but is held in position during the upward stroke of 13 by the ratchet part, 15, which is pressed into position by a s rhiy The spur gear returns to its original position when &e decatitation is made. At this point the push rod, 13, travels far enough so that 14 disengages itself. At the same time 15 is pushed out by the pin on 12 and is held out by a spring ratchet attachment, which is released again when the spur gem returns its starting point. The spur gear, 12, carries an adjustable outer arm mounted on its shaft, which carriee a pin and determines the distalice the
ANALYTICAL CHEMISTRY
1240 gear travels on being released and must return again stepwise in order to reach the 0 position. Near the outer ed e of the spur gear, a series of holes numbered 1 to 50 are drillef as shown in Figure 6, one for each tooth of the gear. The pin on the arm snaps into any desired hole and is thus adjustable. Another pin in place of the 0 hole on the spur gear presses against 15 when it reaches the upper position; 15 in turn presses switch 3 and stops the motor for the settling interval. Part 16 is rigidly attached to the shaft of the spur gear, so that it reaches the position shown in I a t the settling position. When the motor starts again, 16 pushes up on 17, which pushes down on the lower tip of 18; 17 and 18 are held by a brass plate attached to the stationary piece, 5 . 18 carries a pin on its upper end which pushes 19, attached to wheel 1, outward and holds it out as the wheel moves forward. Thus the other curved wedgeshaped end of 19 is moved from the outer edge of the track to the inner edge and held there for a time sufficient for the roller on the arm, 3, to enter the outer track, A. 19 is therefore the track switch, which is normally held by a s ring in the position shown. The corresponding part on the rig& is merely held in position by a spring. The roller pin forces it open on passing through.
I
'"'
ROBOT M O T O R
UNIT
___
.._.__.___________.._~.....
I I
I I
I
!
I L
The robot includes a device for combining fractions, which can be set so that the collector turn table is advanced every transfer or only every 2nd, 3rd, . . 6th transfer as desired. This automatically combines adjacent effluent phases and reduces the number of test tubes required for a given fractionation. The device is called a "combiner" and is shown schematically in view I1 of Figure 7.
.
Six V-grooved wheels numbered 1 to 6 are attached rigidly to a hollow shaft, 26. The wheels are rotated the distance of 1 groove each time the arm, 3 moves to the decantation position by a small push rod, 27, and arm attached to one end of a rod extending through 26. On the other end of the rod is attached a ratchet mechanism, 30, which pushes the wheels. The push rod does not move far enough to enga e the ratchet arm on the other movements of the lever arm, 3. $he first wheel has 60 grooves. Each time a groove moves into position switch 9 is allowed to close a t 28 and the fraction collector motor is activated. In the grooved wheel numbered 2 every other groove is omitted. Thus switch 9 closes only every other transfer. Similarly 3, 4, 5 , and 6 have only 1/8, 1/4, and 1/6 the grooves, respectively. 28 can be moved to any desired wheel by means of a screw thread turned by the knob at 29. The combiner is enclosed in a small metal box. The wiring scheme is given in Figure 8. The driving motor available a t the time happened to be a direct current motor; hence both direct current and alternating current power are used. Both sources of power can be turned off simultaneously a t a main switch, S w l , which is a double-pole, double-throw switch.
I
I
the shaft. The relative positions of the metal flaps are adjustable by loosening the nut, 25, which holds the three circular metal plates on the wheel. The wheel is calibrated in minutes and seconds. If the parts should be moved so as to upset the sequence, the next cycle will automatically correct the change.
-
TIMING
UNIT
F i g u r e 8.
i
Schematic W i r i n g Diagram
The timing mechanism is shown in views I11 and IV of Figure 7. It provides for three independently adjustable intervals corresponding to the settling period, a shorter time for draining a t the decantation stage, and a third short time for drainage a t the transfer stage. This mechanism is mounted on the gear box of the electric motor. The time clock, 20, is a Hayden timing meter of 0.2 r.p.m. It is mounted on an aluminum base and two upright pieces. It drives a wheel, 21, by means of a shaft, 22. The wheel, 21, carries three circular plates attached to it. Each plate carries a metal flap, a, b, and e, attached as shown in Figure 7,111. Each flap depresses switch 11 as it passes over. Thus flap a starts the motor a t the end of the settling period, b, after the decantation and c after the transfer interval. The time clock is set in motion when switch 3 is closed by the metal piece, 16, a t the end of the equilibration stage. After c has passed switch 11, the equilibration stage has begun. At the second and third strokes of arm 3, wheel 21 is released from the timer shaft a t a saw tooth clutch, 24. A solenoid, 23, operating around the shaft, 22, and energized by switch 8 disengages the clutch. Wheel 21 is spring-loaded and therefore returns t o its starting point. The power to the solenoid is then turned off and the teeth a t 24 are engaged again by a spring on
The driving motor, M1, is a shunt-wound l/ao-hp. Jannette direct current motor geared down to a range of 20 to 30 r.p.m. The speed within these limits is variable by virtue of R, an adjustable resistance in series with the field. In order to assure a slower, smoother movement when the cells are tipped to the higher position required for decantation, an alternating current relay, RY3, has been inserted which shorts R and slows the motor to the minimum speed until the transfer is completed. Snitches 3, 4, 5, and 6 are motor controlling switches (GE switchettes) whose normal position is closed as shown in Figure 8. Switch 3 is opened by the advancement of the spur gear, 12, of Figure 6, after the required number of strokes of the lever arm, 3. Hoxever, the apparatus must be stopped a t the proper angle for the phases to settle. This is accomplished by switch 4 which is opened by an adjustable plate attached t o the side near the periphery of the wheel, 1. It is opened every time the lever arm reaches the highest point of the equilibration stroke. The latter is the correct position for settling. Both sivitches 3 and 4 must be open before the motor will stop. Switch 5 is opened by the push rod, 13, every time the lowest position of the arm, 3, is reached. Switch 6 is opened when the ratchet arrangement, 15, permits the spur gear, 12, to return to its original position. Switches 5 and 6 both must be open to stop the motor in the transfer position. Switch 7 is opened to stop the motor in the decantation position. Riding pickaback on switches 3, 6, and 7 are three other switchettes whose normal position is open. These act oppositely to their companions and activate the timing mechanism. Switch 8 is the reset switch which energizes the solenoid, L, and disengages the timing motor clutch in order to permit the wheel, 21, to return to the 0 position. At the same time the timing motor is stopped by the opening of switch 6. In case switch 8 fails to act, an emergency manual re-set switch, 12, is activated by a pin on the third circular plate attached to the wheel, 21. The timing motor is started again by w i t c h 3 at the end of the equilibration. RY1 and RY2 are 110-volt direct current relays used to ensure that switch 11 has control of the motor only until sviitches 3 , 4 , 5 , 6, and 7 are all closed again. C is a 4-microfarad 600-volt paper condenser placed across all the switches in the direct current circuit, so that arcing across the switch contacts is minimized. Fraction Collector. The fraction collector is essentially that of Stein and Moore (11). I t is placed under the automatic filling device as shown in Figure 2.
V O L U M E 23, N O , 9, S E P T E M B E R 1 9 5 1
1241
may be quickly removed t)y restoppering and drawing :iir through the train.
I ! METHODS OF OPERATION
The apparatus can t x opertitcd in a variety (:I ways, d e p e n d i n g on the purpose of the distribution, the quantity of t h c s solutr~. the complexity o f t h e m i x t u r e , a n d thr. numtm of t r a n R f e r s r c yuired. Of the niany possible ways only three a1.1~ discussed here. Ts-o h a w hren discussed (7) in conTube No T m n s f e p No. nection with other models.. Figure 9. Distribution Pattern for .imino Acid I < ~ r r l Fundamental Procedure. Experimental Somewhat in excess of tht, Calculated e q u i l i b r a t e d lower phasr. requircd for all the t u l w LVheii the apparatus moves to the transfer position, the upper (2250 nil.) is filled into the a p p a l a t u ~ I)hase, which is in the decantation compartment of cell 219, flows out through a glass tube extending to a point in front of the bearWith the train in the decantation psition 250-nil. purtiorib a r e ing shaft but behind the bend in the tube, d, of the dipper. Here added a t oints approximately 20 cells apart, beginning a lex it empties into a stationary small glass bulb which carries a tube tubes in afvance of those LThich will receive the sample. A short leading to the fraction collector below. funnel with a flat joint which can be attached to openingf, Figure When the apparatus moves to the decantation position, switch 3, is useful for this pur se. All openings are then closed and 9 from the combiner, Figure 7, activates a shaded pole induction the apparatus is move$from the decantation to the transfer motor, M2, of Figure 8, in the fraction collector which advances position a number of times. The loner phase is distributed the turntable a single tube. At the end of one move switch 10 automatically by this maneuver to every cell, and as the disis momentarily closed to energize the 110-volt relay, RY4, and tribution proceeds the small excess finally flows into the fractinn cause its mercury switches to reverse position. This stops the collector in front of the advancing upper phases. motor, M2, and keeps it a t rest until the power to the collector is The phases containing the solute are inserted by means of a interrupted again by switch 9. A11 switches in the fraction syringe into re11 0 or into as many consecutive cells as may be collector are mercury switchelq, so that sparking and the risk of required. This point has been discussed elsewhere ( 4 , Y ) . Gpper explosion is avoided. phase free of solute is placed in approximately 10 cells in front of The holes in the turntable provide for 200 standard 20 X 150 the solute band in order to cnsure conditioning of the lower nim. borosilicate glass test tubes. Thus, when three effluent phases in advance and thus minimize volume distortion. fractions are placed in a single test tube by setting the combiner The automatic filling device is then attached and the apparatus a t t 3, the capacity of the collector is enlarged to 600 effluent is operated with the desired number of strokes, settling interval, phases. When the combiner is set a t 6, the capacity can he exetc., until 220 - b transfers have been applied. Here b is the tended to 1200 by employing only 5-ml. upper phase@. number of cells initially required for the solute. In the event that a shorter distribution gives sufficient resolution, it is not Care and Housing of Apparatus. The glass train, filling necessary to use the full 220 transfers. The analysis of the disdevices, and fraction collector are all enclosed in a glass show case tribution is made before any of the upper phase which has carried solute has had opportunity to leave the train. The data are with sliding doors as shown in Figure 1. The robot, however, is treated as before ( 7 ) for the fundamental qerieb. outside of the glass case, FO that possible sparking from the electrical connections will not ignite inflammable vapors. Single Withdrawal. The maximum t ransfela of the previously The case, of standard stainless steel and glass show case design, amounts to a hood, aa it is connected with the laboratory hood discussed procedure are reached and the extraction is continued vent, The entire apparatus is housed in an air-conditioned room by allowing effluent upper phases to collect in the fraction colleccontrolled to =k 1O C. Howevt>r, the temperature fluctuations tor. In this manner of opmation the method is most nearly analinside the glass case where the train operates are much less than ogous to chromatography. For expression of the result it ita *lo C. The hood vent has an adjustable damper to reduce the best to give each effluent fraction the transfer number n-hich amount of air pulled through the air conditioned room a t times caused it to emerge from the train. .4definite advantage of single when there is little solvent wcaping into the atmosphere. withdrawal lies in the fact that the itnalysis of the effluent fracTwo large stainless steel pans lie beneath the train. An outlet tions involves only one phase. The rewlts are best reprebented from these pans leads to the drain in the floor. In case a break by a plot of the weight pel effluent fraction (or Home figure proportional to it) as ordinate against the tranbfer number a s abshould occur in a glass tube during the night when the machine is operating unattended, flammable solvent would then not wcumuscissa, as shown in Figuie 9 The analysis may show that all the solute has not emerged from late but would run down the drain. The main use of the pans is in washing out the cells a t the conithe train and that satisfactory separation has h e n accomplished. pietion of a run. For this purpose all the stoppers are removed Here expression of the results involves two patterns. The fundafrom the cells and unwanted solvent is dumped into the pans by mental pattern representing the cells is plotted as given in Figure disconnecting the train 0 om the robot and tipping it forward. 9, left. It is the convention of this laboratory to place the effluent Each cell is then flooded with water from a rubber hose attached pattern on the right of the fundamental pattern, with the highest to the tap. After the tubes have been inverted so that therinse transfer number nearest the highest cell number of the fundaRater reaches all parts, it is emptied into the pans. A4wash of mental pattern, as in Figure 9, right This arrangement gives the distilled water is similarly given. Finally, about 10 ml. of acetruest expression of the over-all result, since there is littIe differtone are placed in each cell from a wash bottle for the final rinse. ence betReen the composition of the last effluent and the upper The acetone evaporate? overnight through the open stoppers or it phase of cell 219.
----
ANALYTICAL CHEMISTRY
1242 The method of operation and the treatment of results are well shown by an actual experiment on an artificial mixthre of 10 amino acids. The mixture chosen contained 300 mg. of each amino acid. At the start the sample u-as dissolved in a mixture of 80 ml. of each phase, sufficient to fill the first 8 celIs. The system was made by equilibrating an equal volume of 5% hydrochloric acid with n-butyl alcohol. Fifteen strokes a t each stage were applied, and 30 seconds were required for the phases to separate. The apparatus was permitted to operate until 780 transfers had been applied, approximately 20 hours. The analysis x-as made by weight (6) as the hydrochloride residue. For convenience, the lower phase alone of the fundamental series was analyzed, as only amino acids of low partition ratio remained in this series. The total aeight per tube can be calculated, if desired, as the partition ratio is known. The identities of the bands from right to left are: tryptophan, phenylalanine, leucine, isoleucine, tyrosine, methionine, valine, a-aminobutyric acid, alanine, and glycine. This mas confirmed by spotting appropriate samples on a broad paper chromatogram (6). All the bands \yere reasonably 7%-ellseparated except the phenylalanine, leucine, and isoleucine triplet, which can be resolved by recycling or by changing to a more favorable system. In the former case effluents 380 to 560 could be introduced into a freshly charged apparatus a t cell 0 in the order in which they emerged and then recycled several times. An actual example of the latter method of resolving the triplet will be given under the “recycling” procedure. Theoretical curves 15 ere calculated for the fundamental pattern as given ( 7 ) by Equation 1. With a slide rule the calculation of a theoretical curve requires approximately 10 minutes.
In a withdrawn series each successive effluent fraction has had one further transfer applied and exact calculation thus becomes very laborious. However, a satisfactory approximation can be developed from Equation 1 by neglecting the esact mechanism by which the distribution is actually reached. An approximation formula for the purpose must involve the partition ratio and the average number of transfers involved in the band. K can be calculated from the position of the maximum, the average number of transfers, n, for a symmetrical curve. The result is given by Equation 2, where u is the number of cells in the train.
K = u / ( n - u)
In expressing the result for this case, cell numbers would not be plotted but instead the transfer number on which the last upper phase left the cell in question would be assigned to the lower phase remaining in the cell. Calculation of a thearetical curve could be made from Equation 3 by substituting 1/K for the value of K as calculated from the fundamental series. Recycling Procedure. If on analysis a t the fundamental stage only overlapping bands of similar partition ratios are revealed, it is wasteful of solvent to use the single withdrawal procedure, as the upper phases in most of the cells would be free of solute, On the other hand, if the phases contain no solute, they can be used over again without further treatment. The decantation compartment exit of cell 219 is therefore connected to cell 0 and the upper phase, which would have been an effluent phase, is thus caused t o enter the system again to begin a second passage around the series. In fact, when the band has been narrowed to only two or three closely related components. several thousand transfers can be applied by continued recycling.
0.8 0.6
0.4 i
8
0.2 100
180
140
:I
300
6’
0.51
0.2
, ’ , ‘
0.1
5,
260
260
220
$I
, , 300
-pa‘
340
’1
,
, ,
‘b.J
380
420
Tube No.
(2)
The particular transfer required to move the maximum of substance from the fundamental series becomes the maximum in the withdrawn series. But because the withdrawn series has no lower phase, it bears the relationship of the fraction K / ( K + l ) to the fundamental series. Thus, incorporating this fraction into Equation 1so that the material balance is maintained, Equation 3 is derived.
(3) Equation 3 has had extensive checking and has given withdrawn curves which agreed with the experimental curve on many occasions. The slight discrepancy on the right-hand curve of Figure 9 for tryptophan is due to the fact that the sample was placed in 8 cells at the start. Agreement would be better with solutes with lower partition ratios, where more transfers are required before the band emerges. If the distribution characteristics of the solutes are known, it may be desirable to interrupt addition of the upper phase a t a given point and permit all the upper phases to flow from the cells as they reach the end cell. This has the advantage that only a lower phase remains in the cells to be dealt with analytically.
Figure 10. Separation of Overlapping Triplet by Recycling Process
Experimental - - - - Calculated
The fractionation of the triplet, phenylalanine, leucine, isoleucine, insufficiently separated in Figure 9 can be taken as an example. The free amino acids were distributed in the system n-butyl alcohol-water, using five cells for the initial charge. Recycling was begun a t 240 transfers. The machine was permitted to run until 1137 transfers had been accomplished. S o solvent was added during this time. Analysis of the lower phases gave the pattern shown in Figure 10 (upper). It was plain that all solute had cleared the cells below 150, but that many of these had been refilled by the advancing phenylalanine band. Therefore, cell 0 becomes 221, 1becomes 222, etc. When the extraction vas interrupted, the phenylalanine band was well resolved, but the leucine-isoleucine band u-as still an overlapping doublet, The former band was accordingly removed and replaced by fresh solvents. The machine was then permitted to operate until 2772 transfers had been reached Analysis now gave the pattern shown in Figure 10 (lower), which is regarded as
V O L U M E 23, NO. 9, S E P T E M B E R 1 9 5 1 a sufficiently good separation, even though further separation only involves permitting the machine to continue. I n the event of a binary mixture with beta values of 1.1 or less, the width of the overlapping bands would extend over more than 220 cells before complete separation had been achieved. But the advancing edge of the band would contain A of satisfactory purity while the trailing edge would contain B of satisfactory purity. Perhaps 20 cells of each might be withdrawn. After filling with fresh solvents, a calculated number of transfers could be made before repeating the double withdrawal. This method of accomplishing alternate or double withdrawal does not require the constant attention of the procedure proposed (7) for the fewer cells in the series. DISCUSSION
The separation and characterization of complex mixtures of organic compounds have always been more or less of an a r t to the organic chemist. Often the reason for a favorable separation has not been well understood because of many interdependent factors which could not be dissociated clearly from each other. Just as mysteriously, the same set of conditions often were found to fail on a different mixture. One of the advantages of countercurrent distribution is that the nature of the process lends itself to experimental analysis of the separate factors involved in the fractionation and in the interpretation of the data. This in turn permits a reproducible fractionation of high efficiency. These considerations may conveniently be grouped under three main headings: The numbers of actual transfers a t equilibrium or otherwise which can rofitably be applied. The stufy of systems and their modifications by specific solutes in order to obtain the required selectivity. Supporting analytical methods. All three of these considerations are interrelated. For instance,
if an automatic apparatus is available whereby high numbers of transfers can be accomplished easily, it may not be necessary to search for a system which is of maximum selectivity for a given separation. On the other hand, with high numbers of transfers, a system must adhere much more closely to the ideal with respect to the partition isotherm. High numbers of transfers also require more analytical work and therefore more rapid methods are desirable. Part of the objective in developing the apparatus described here was trying to find the maximum number of cells which might profitably be employed by a single v,-orker in a laboratory before the whole procedure would become unwieldy. Although that stage would long since have been passed, without proper organization, the experience gained with the present equipment would indicate that even higher numbers of cells than 220 would appear entirely workable. High fractionating power in general requires relatively more solvent. Obviously, assuming a standard sized cell, the volume of solvent required is directly proportional to the number of cells involved. Thus, the 220 cells of the present apparatus require 2.2 liters of each phase to complete the “fundamental” series. “Single withdrawal” might require tenfold this amount where a higher degree of separation is desired. Unless solvent is extremely cheap and pure, it would appear best to develop some procedure for using the same solvent again and again. The “reflux” principle of fractional distillation is brought to mind. In a general way, the effect of the reflux of distillation might be viewed in the following manner. Interchange between vapor and liquid takes place only a t the surface of the liquid. There ia a fixed length of the column to be traveled by the vapor, but because the liquid is constantly flowing back down the column at a rapid rate, the vapors contact a much longer liquid surface than is presented by only the length of the column. The length actually contacted is a function of the speed of back flow, although at higher speeds a point is reached where efficiency falls due to
1243 other factors. The proper state of affairs is conveniently created by the concentrating action of the reflux condenser. It is theoretically possible to simulate an analogous state of affairs with regard to the movement of the individual solutes in an extraction column and likewise provide more opportunity for repeated interchange. Such a procedure involves removal of the solute from the appropriate fractions and movement of it systematically backward in’the column. Obviously, this would be very laborious. However, a different approach accomplishing nearly the same purpose is practical. Fractionation according to the binomial expansion methodically divides and repeatedly subdivides a unit quantity into fractions each becoming smaller as the process continues. The fractions, beyond a certain point on each side of the “band,” become so small that they may be neglected entirely. The solvent in these can therefore be regarded as free of solute and reintroduced into the system again. The “recycling process” does this work automatically. The effect achieved is that of greatly increasing the number of transfers or contacts. When applied to complex mixtures, a preliminary distribution can be made for the purpose of sorting the mixture into groups each containing solutes with similar partition ratios. A distribution is then made on each group by the “recycling” process as outlined in the example with leucine, isoleucine, and phenylalanine. At the end of the distribution, the solute can be recovered in a few simple operations from the dilute solutions comprising the appropriately combined fractions. The net effect here is not greatly different from reflux. Thus, with the 220-tube apparatus, recycling ten times would give nearly the same effect as the “fundamental” procedure in a train with 2200 cells in it. Recycling approaches experimentally the ideal more nearly n-ith large numbers of cells in the train. In the original steel machines of 25 tubes ( 7 ) , recycling to 50 transfers would be the upper limit, since a single solute with a partition ratio of 1 would then yield a significant amount of solute in every tube. There would be little room for throwing off impurity. A significant amount of solute for purposes of discussion may be defined (3) arbitrarily as an amount greater than 1.0% of the solute present in the cell containing maximum material. However, with 220 tubes in the series, 5260 transfers can be applied to the same solute by recycling before significant amounts of solute would be in every tube. The upper phase here has passed through the train 24 times. The concept of band spread with increasing numbers of transfers has been treated elsewhere (3). The band spread, in terms of transfers applied, is a function of the value of K . Thus with a K of 0.2 the band spread would not be 220 tubes until 8000 transfers had been reached and 14,800 transfers Rould be required for a K of 0.1. As pointed out earlier (2, 4),the best separation of a binary mixture in terms of transfers is obtained when the geometric mean of the K’s is 1. However, with automatic equipment a different emphasis is reached, since there is no labor involved in making transfers. K’s with a geometric mean of 0.2 will give a better resolution, even though a somewhat longer time will be required. Even better resolution is obtained with lower K values, but here the migration rate becomes slow. By way of analogy, it is interesting that the high resolutions recently obtained in chromatography (12) and ion exchange (IO,13) require slow migration of the solute through the column and relatively large volumes of effluent. There is thus offered the opportunity for many more interchanges between the phases. With automatic equipment in general there need be less emphasis on the selectivity of the system and the range of K values. Likewise adjustment of relative phase volumes is less important, With the hand-operated equipment, a practical K range appeared to be 0.2 to 5, whereas T5ith the equipment reported here the satisfactory K range is extended to a range of 0.01 to 100. This naturally means that mixtures containing a dozen or more components can be separated in a single run.
ANALYTICAL CHEMISTRY
1244
Other considerations ki selecting a system are those of adheience to ideality and capacity. These factors are interrelated, as adherence to ideality is nearly always more closely approached for the more dilute solutions. When many cells are in the series, the solute can be scattered in a larger number of cells at the beginning and thus larger capacity is reached as well as higher separating power. The problem of ,deviations from ideality will be taken up in a separate publication. More analytical work a t the end of a distribution is required for the higher number of transfers in which more components arc> revealed. However, the number of analyses is not in direct proportion to the number of transfers or cells. Only a sufficient number of analyses per band is required to show defuiitely its position, height, and width. Approximately 10 analyses or points on the pattern are adequate per component, whether 100 or 1000 transfers have been applied. This obtains because the concentration changes are more gradual with the broader bands produced by higher numbers of transfers. In cases of overlapping bands, a few more points may be required As far as the quantitative aspects go, agreement between the calculated and the experimental curves in Figure 10 is all that could be expected. The automatic distribution apparatus can therefore be regarded as fully calibrated and perfectly reliable. ACKNOWLEDGMENT
The cells, clamps, ground stoppers, cell supports, etc., were built by Otto Post. The clamps, ground stoppers, and cell
supports q-ere of his design. The authors are also indebted to him for constant advice on all the mechanical features of the apparatus, exclusive of the robot and fraction collector. The robot is a completely original design drown up by Josef Rluni and Richard Janes. I t and the fraction collector were made under their direction in the Rockefeller Institute instrument shop. Much of the credit for the success of the whole undertaking is due them. LITERATURE CITED ‘1)
I3arry, G. T., Rato, T..and Craig, L. C.. J . Bid. Chem., 174,
209 (1948). ( 2 ) Bush, M., and Densen, P., AKAL.CHEY.,20, 121 (1948). (3) Craig, L. C., I b i d . , 22, 1347 (1950). (4) Craig, L. C., and Craig, D., in “Technique of Organic Chemistry,” Vol. 111, p. 200, Xew York, Interscience Publishers. 1950. (5) (’raig, L. C., Gregory, J. D., and Harry, G. T., J . Clin. Incest., 28, 1014 (1949). 16) (’raig, L. C., Hausmann, IT.., Ahreiis, E. H., Jr., and Harfenist, E. J., ANAL.CHEM.,23, 1326 (1951). (7) Craig, L. C., and Post, O., Ibid.. 21, 500 (1949). (8) Eyring, H., Ibid., 20, 98 (1948). (9) Gregory, J. D., and Craig, L. C., Ann. h’. Y . Acad. Sci., in press. (10) Schubert, J., ANAL.CHEM.,22, 1358 (1950). (11) Stein, W. H., and Moore, S., J . Bid. Chem., 176, 337 (1948). (12) Ibid., 178,79 (1949). (13) Tompkins, E. R., A N ~ LCHEM.. . 22, 1352 (1950). RECEITED February 20,1951
Quantitative Application of the Kiliani Reaction VERNON L. FR4MPTON, LUCIA PEEPLES FOLEY, LEL4ND L. SMITH’, AND JANE G. MALONE Basic Cotton Research Laboratory, University of Texas, Austin, Tex. An analytical procedure, which is dependent upon the quantitative addition by aldoses of hydrocyanic acid at pH 8.5, and 39OC., and which is capable of a high degree of precision, is described for aldoses. The nitrile resulting from the addition is hydrolyzed in an alkaline medium to stoichiometric proportions of ammonia, which is trapped and determined titrimetrically. An ammonia blank must be determined with the reagents because of the spontaneous hydrolysis of hydrocyanic acid solutions. The method was applied to several simple sugars, to plant saps and juices, and to cellulose. The procedure may be used satisfactorily in end group determinations with cellulose.
T
HERE have been very few studies devoted to the quantitative aspects of the Kiliani reaction. The more significant studies are the papers by Lippich (3) and by hlilitzer ( 4 ) . Both workers used the Liebig-Denigbs method for the deterniinatiou of unreacted cyanide; Lippich distilled off the excess cyanide, whereas Militzer determined the excess cyanide in the presence of the cyanohydrin. The authors find, however, that more satisfactory results may be obtained if the cyanohydrin is hydrolyzed in an alkaline medium, and the ammonia liberated is determined quantitatively. The addition of hydrocyanic acid to glucose is not quantitative under the conditions employed by Lippich; the time is too brief and the temperature is unfavorable. The precision of the determination involving Militzcr’s procedure is low bccause of the reversal of the reaction Glucose
+ HCS-
++
cyanohydrin
in an alkaline medium, because of the polymerization of cyanide under the conditions employed, and because ammonium hydroxide will not satisfactorily retain hydrocyanic acid under these same conditions. These factors that reduce the precision with 1
Present address, Columbia University, New York. N. Y.
Militzer’s procedure do not affect the stoichiometric recovery of ammonia under the conditions out,lined in this communication. METHOD
A known quantity of glucose (2.0 to 600 mg.) in 3 ml. of water is placed in a 500-ml. round-bottomed flask. Five milliliters of 0.4 N acetic acid solution and 5 ml. of 0.8 N potassium cyanide are added and the flask, stoppered with a T stopper, is placed in a thermostatically controlled bath a t 39” C. After 3 hours the contents of the flask are acidified to the methyl red end point, and the unreacted cyanide is driven off by passing air through the solution. After the odor of hydrocyanic acid can no longer be detected (15 to 30 minutes) the flask is attached to a steam distillation apparatus (see Figure l), 10 mi. of a 20% sodium hydroxide solution are added, and the solution is steam-distilled until a proximately 50 ml. of distillate have been condensed. T f e ammonia evolved is trapped in standard hydrochloric acid, and back-titrated with standard sodium hydroxide solution. Methyl purple is used as the indicator. An ammonia blank is determined in precisely the same manner, except that no substrate is added. Satisfactory results are not obtained unless the pH is in the range 8.5 to 9. The data for glucose in Table I, obtained following the procedure outlined above, are completely satisfactory. [KOdifferences were observed in the four lots of glucose used in this studv.
