. 1?. $7.... 3.312.... .. _ .-"-VI-I'I o 0 185A H w'y‘vi“"‘. "';'V o 'P - a E _‘ cH-Rmme' .v v " ...--~«o-v~-~‘ v 9" V'V."W Y‘I a .‘ u'd-f”~10-O."".’. 9.. Q unPW*“"” ‘ ' u C ' ”fly-In o fir 2M”~1‘“9 S . .2. u........;:..o‘. . . . o :3. 78. .a o. ..p.r. J. p . v . «$1....fi.3.... 31... u!- <. ... J: wax 3... .P .71 “at? {J .., ~... I .\ 1 o 0“ ._ . . _, E . . .t .. . v . A . a . n . . n0... . _ o . . . ..... ,. .I . . . . . . . . . . 0. p O. . .v . o . 2.. o .u .a . . ,- o '5. § . ..._ . o _. . y.. . . c . .. . . . . . : .. .. . o. . ... ... . c . nE ‘0 t o I \ . _. . - u c . E... . . >1 . . . .. o . o. . . o .. . o. .‘ g 0. no . . .3 . . . . 1v... . .. “ I .y u. n o n .o . I ' .0. . . ,. . ¢ .- . . .2 .0 I .2 o. . n a . . . .. . .. ... . .. - . ¢ . _ 7. t u c. . .. o ‘ an? M #6 Eé v o . . .A ”1...”. . 4|.on $5.3, J’4r. \ m Er - I95. .uv__.c\.‘. . . ..‘Tu.a..2i'..)..:o._.;.... ...:_ ... 791?...2 ~u...o....:.m.§n.:.f . . o . . 9 t . . . . o 0,. ‘3.000 0‘ . . . n . co". 0.. .\Ql-~ .n...‘..... “2" Int-Wow. ..q|-\\..~‘.v I Hid“! “elite... 5‘ .3.-kfl90 -L L‘Vlwo - ‘0 VC. I... a“ o....v5'l.0h.§ c.0a- ..,¢ l.....l'.. .Ico. a..4.ofi.vov.~. (5.4a.-‘o .00? a Onovcoodcan 0J4 .IV50004 .- o. .4. v .t. Q Q T... .,.o.o..a:aa¢.ans ..o.acoao odocr .‘.vu3,a4.. 3.4 0‘00 I .‘O... I..J.D\' 00"! 0.- onoo'Oraulfi~'o— .4. .uao..o£ ~000.l. . Jl»ooo. ‘91,. o... ...co.¢.. 0;!”a co.-. 00......foo5 u 0.0 .dlop.'o .Ao)...04‘ ‘1’v390090 . a..‘. 34m tint-pto. 0..) . . $014.” I _. .. . III. .? oo o c. p. . E '.ooo.l o .. q 4. o c [a ‘ .v:luob.4ow.rr?o . o o., o...n;v.(. Cbctllv 3.0. (I .a.oA.vc.-. loo-‘3' 0- 01 o I-QOJAO é“ ..a. Jcaa...oo.a...4.).o.4. 0...... aoouoo.... . .p.-.o.. Jolioipo 4 v ’C”.OO.I~ O . 040.011.. 4" c... .1041. aco’t...bs.’. o .ioo.o‘i v.1ofo‘t . 0“..." .. ..¢...c.-..F . anlEld'.’ 4.o.oao(;o.oo .0... l... o.v.3v...¢\soloo$vo a)... .o. 3.3.7.1. .cCOOdO. - o .o..¢0i¢o H. :1 43.7.».1... 33.. "£57. - o lo... a o . {Oln’lv OI. o » I van .EMOA’II... 90.1 .oa.:..*....3.'n.;a.o .o.novn.v14.ra.. u....o‘ao -. .ou.. ;.Qc¢..vpl¢lo 4 . . ' .4...I.V4.ua.afé.d .91. .‘f.m ..ao 11310;...Jiv-v-arr . qv. 1-.n.I‘ o..l."o¢'.‘ ...v0r?a,o1.u .r cr.o.s'!l v_. ., . .4...~ .r .‘Olln... . O. 4 — . o.vlv.~r'oY/oaq7! wOArJoV' c_ 0... II .olOo.l -,..0 ¢_ ‘ ’v'.‘ .. ......'J. 4.)... (3.... «co. .4-- .0. ..0v Oo..'7c..‘ 3. Old... 4 .. .o‘u..l.ooa.o.. ,.oo.c 0’ illirt!‘ .'o..0o.cl . .oo..4 ..¢V...o.. .. claud.fl.lat.li 23.34.; '7 .4 v‘ . ..1 .. o]. v A 0’ .....a.. ."o' f . .....o...:.d...l. if . (+9; - . o no. , . ..Ip,O—’J047’A(4'.'q.c I. . .u .n.. . ¢..1 4.... 0‘60“. D'.OI.vv.o(' 0' o. 1.... .....a .20.cv. . ...a 04.4I'IO'J I..’Q“"r~’.)o .- ..-.Oooo...f..ofilo.;o.Jv.-.a-Ao.'lu o n”) .D/ 11.4. - . 7. .0... . .:~ _ I .(l....oo.l. v f'L v ........a.._;.u‘ao.4v.‘.c' E ..o o 0.0, 1.:o111040c 0.... . .o 1.4. . .. '01.... , .co.a‘.o.. . . v‘c .. .o.. C..l 4... . .oJlo.’.o. v ..I .ol.rIo- [I u 4.... 1.. ..... I... . ..v.-oo. Ala-aalrff'o (c.1070: . ¢ COLO'I.f'.l-, '0 -._4. _ o..a.".ol J‘ Av. .. . .. .rp'fsff’ for. c . . ..5 n1 .. 'l'ilr..m(\f‘ ...l. .. . It. - . - 0V. ' Luignnnrfia an 0'07). :O.a’—t .(l_' .. ..-a .v. 1.0....OAC . . »- .- ..oc..f. . ‘. . .n l . ..lo. 4 .... . .00....I- JOI~..1.’\.O'II.VII . .. a. a. . . .. ..oo. , V ..A (\ouVIO’w’OI... . II I . n.. . E . I I I 'tlo I. n . o. a. .. 9 to . . o.' . c o . _ .. o .o.. ; J o v . . ..E ' h . 0... art- .a .- . I. . .‘n o o to o . .. o . v . . .anlvv.rlv 4 . . A... I I .I o u . I o .IrJoO‘cnlY. C {1' v . ...... o.;...0;o c.t. . ... . v . V 'O.’o.f . a o. I . .v. our-(Iv ~ a o. ..I t, . . . t....:r p.00 ..0 a . |~A . ..'r.o 0.! o v r... o . . . I A on. o n J I 4 .01 ' l . .1 o. . a..v o.vI'InOl I - lo 11.. I l I . 4 u u . n '0 U. ' o. I .o. 9 a I o .. O- o c t Y.. all p. . .. n . I; .I 01/. . . . .4! .v J—I.OOo.-II. .c. ' .0. on: . . .. , ..4 .1);o.f O i I l I. .II . .. I n to v. v a... u . . o no.4! 0 a . . I .o«v.ooooJI . . II . .00 - Vo.’ - . I... ... .o 0 .aa,. 1 n . at . 002‘ u . . v. A. :I If .. . . v n .0 . a. . 0.: . o . . c 1.1 .f at o . ... . .no' _ o . I! .00 0, Cl. 0.. I... .. o.. . n o. a. l... ‘ a c . 9! ll.- yl ' O . .l .. .OIIJQE . a oI . .. ....l . .o . .o a. . (o oorl . . . - . a o u . II .. '0 0 t O . I i I f: . s-.. o...’o. a o . .. s..'vlo. . o . .. . . .. .p. o o . E o . E... a a I . ... . . .o ..u- a . a. . . v .0. . . u . I. v 0 .’ . I o... no. at. . . .: p o n .vv‘ . I . . . . .. Iva! . .I . . . . . n . . . I . our... .1 o I I O 1‘0 a O — ... .. .; .. . . ¢ n.1v .: O. . ¢ . n . . .) A 5 II. . .C I u. I. 1. . ,.IaO‘ob.‘.o00‘45’.0J_v»‘\"-O'901 . ‘ai \tn ’0'} u. 3‘}. Ili>| - . o LIBRARY Michigan State University ANION EXCHANGE CHROMATOGRAPHIC ANALYSIS WITH CONDUCTOMETRIC DETECTION By Hugh Franklin Hussey AN ABSTRACT OF A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE Department of Chemistry l97l ABSTRACT ANION EXCHANGE CHROMATOGRAPHIC ANALYSIS WITH CONDUCTOMETRIC DETECTION By Hugh Franklin Hussey The investigation of conductance detection in ion exchange chromato- graphy was pursued with the use of a bipolar pulse conductance instrument. The instrument is quite sensitive to small changes in conductance and is able to quantitatively determine the number of equivalents of a given ion in a sample from the area of the peak it produces as a result of the difference in the equivalent ion conductances of the counter ions of the eluent and those of the sample with an accuracy of :_5%. By employing 2.8 X l50 mm microbore columns packed with 200-400 mesh Dowex l X 8 anion exchange resin and eluting at flow rates of from 3 to 5 ml cm.2 min’], the system of F' and IO3 eluted with 0.015 N KN03 and the system of P043' and P2074" eluted with 0.20 M KCl buffered at a pH of 5.0 were separated in 35 and 25 minutes respectively with complete resolution. An investigation of the F'-IO3' system was conducted with respect to the variation in resolution (Rs) and elution volume (Vmax) as a function of the size of the sample injected, the eluent flow rate employed and the eluent concentration used in a separation. For the range studied, Rs proved to be nearly independent of sample size .i Hugh Franklin Hussey while Vmax increased slightly in a linear manner with increasing sample size. While Vmax proved to be nearly constant and increased only slightly with increasing flow rate, Vmax was found to be proportional to the reciprocal of the eluent concentration (l/Ce) over a ten-fold range of concentration. With some reservation, there has proven to be no advantage in obtaining equal reductions in analysis time either by the use of higher flow rates or more concentrated eluents since both methods result in equal losses in resolution. The data gathered in the study of Vmax as a function of eluent concentration was found to yield a method of determining and estimating selectivity and molar distribution coefficients under actual operating conditions by a graphical procedure which also gives the void volume (V0) of a packed column. Not only are these values more accurate and practical than those obtained by classical methods but in addition they can be determined in less than half the time. ii ANION EXCHANGE CHROMATOGRAPHIC ANALYSIS WITH CONDUCTOMETRIC DETECTION By Hugh Franklin Hussey A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE Department of Chemistry 197T ACKNOWLEDGMENT The author wishes to express his appreciation to Professor Christie G. Enke for his guidance and encouragement throughout this study. Special appreciation is also expressed to Miss Carol A. McCain for her understanding in the typing and proof-reading of the drafts of this work. Page I. INTRODUCTION ....................... l A. Theory of Small Column Ion Exchange Techniques . . . . 3 1. Relationships between resolution and basic column parameters ................ 3 2. Ion exchange chromatography ........... 4 B. Explanation of the Bipolar Pulse Conductance Technique and Comparison with Other Bridge Techniques ..... 9 l. A.C. bridges with the phase-angle voltmeter . . . 9 2. The bipolar pulse conductance technique ..... ll 3. A comparison of the bipolar pulse and phase- angle voltmeter techniques ............ 12 C. A Comparison of Conductance Techniques with Other Methods of Detection ................. 18 II. EXPERIMENTAL ........................ 21 A. Chromatographic Apparatus .............. 21 B. Pumps and Related Apparatus ............. 25 l. A pulse-free gas operated pump .......... 25 2. A positive pressure reciprocating eluent pump . . 28 C. Instrumentation .................... 31 0. Reagents and Solutions ................ 33 III. RESULTS .......................... 35 A. Separation of F- and 103- .............. 35 1. Quantitative analysis of F- and 10 ' ....... 35 2. Variance of elution volume and resglution with sample size ................. 40 3. Effect of flow rate on resolution and elution volume .................. 45 4. Effect of eluent concentration on elution volume and resolution .............. 50 TABLE OF CONTENTS iv B. Separation of Phosphates .................. 59 l. Quantitative-qualitative analysis ........... 60 2. Comparison with other chromatographic and nonchromatographic methods of analysis ......... 65 IV. CONCLUSION ........................... 70 BIBLIOGRAPHY .......................... 72 LIST OF TABLES Table Page 1. Typical peak area data from a series of injections of a solution 0.l0 N in F' and 0.18 N in 103’ ...... 37 2. Data from the study of elution volume as a function of sample size for the fluoride-iodate system ...... 42 3. Data from the study of elution volume as a function of flow rate for the fluoride-iodate system ....... 48 4. Data from the study of elution volume as a function of eluent concentration for the fluoride-iodate system . . . Sl 5. The average peak area data from a series of injections of a solution 0.10 M in both P043“ and 92074- ........ 61 vi LIST OF FIGURES Figure Page —I 0 Block diagram of the bipolar pulse conductance instrument. .17 2 Schematic of the chromatographic apparatus ......... 22 3 Sample injection tee .................... 23 4 Microflow conductivity cell ................ 23 5 Diagram of the pulse-free gas operated pump ........ 26 6. Diagram of the Milton Roy motor driven pump ........ 29 7 Circuit diagram of the voltage follower RC filter ...... 32 8 Peak area y§_sample size for F' and 103' .......... 38 9 Elution volume and resolution as a function of sample size .41 10 Plot of Vmax vs sample size for F' and 103' ......... 43 11. Elution time and resolution as a function of flow rate . . .46 12. Plot of tmax !§_1/flow rate ................. 47 13. Plot of Vmax v§_flow rate .................. 49 14. Plot of Vmax XE-Ce ..................... 52 15. Plot of Vmax v§_l/Ce .................... 54 16. Elution time and resolution as a function of eluent concentration ....................... 55 17. Peak area !§_sample size for PO43‘ and P2 74' ........ 62 18. Effluent peaks resulting from 10 ul injections of various phosphate solutions .................... 64 vii I. Introduction Due to a great amount of research and development in recent years, ion exchange chromatography has become a powerful analytical tool. The revival of interest in liquid chromatography has been primarily due to the development of the hardware necessary for high speed, high performance work and innovations in instrumentation. Ion exchange chromatography is found to be ever more widely applied to analysis problems as a result of the elimination of its greatest drawback, a lack of speed. With the aid of a wealth of chromato- graphic theory accumulated over several years (mainly from research in gas chromatography) it has been possible through the rapid transfer of analogous theory and technique to refine ion exchange procedures and methods to the point of being as rapid and efficient as separations carried out by gas chromatography. There are now available pumps that deliver constant flow rates, stronger columns and associated hardware that are leakproof under high pressure and a wide variety of ion exchange resins. The precision and sensitivity of in-stream analysis is a result of the development of better general and selective detectors and related instrumentation which make possible the analysis of many types of ionic systems. The experiments described in this thesis were carried out primarily to investigate the feasibility of using a bipolar pulse conductance technique for the detection and quantitative analysis of components 1 2 as they are separated by microbore ion exchange columns. It will be shown that the number of equivalents in micromolar size samples can be accurately determined if there is a slight difference between the equivalent conductances of the sample and counter ions which are sorbed by an ion-exchange resin. A comparison is made between the results obtained by the bipolar pulse method and other conventional but more sophisticated methods of column effluent detection. 'The relative merits and limitations of these methods and the bipolar pulse conductance technique are discussed. Since ion exchange is largely an instrument and hardware oriented technique, the types of errors resulting from electronic and mechanical problems are brought to light as attempts were made to reduce or eliminate them. Foremost among these are the errors stemming from the pumps used to deliver the eluent to the inlet of the column under pressure. The early experiments were conducted with the aid of a pulseless pressure pump of this author's design which was both inexpensive to build and simple to operate. The design and perfection of this pump resulted only after some time was spent on a previous model and some research was done to find a satisfactory means of delivering an eluent. It was designed with the intent of using it to conduct ion-exchange separations with the accuracy of mechanical pumps but without their limitations. Near the end of this study a reciprocating mechanical pump was obtained and used in the collection of the majority of data presented. However this commercial pump is not without its problems and a lengthy investigation was conducted in order to find methods of alleviating or eliminating them. 3 With the apparatus assembled and Operating satisfactorily, the theory of high speed ion exchange theory was investigated as well as the analysis of the ions of various systems. The effect of several variables on resolution and elution volume were studied. This resulted in the development of a better technique for the determination of selectivity and molar distribution coefficients which can be accurately used to estimate the elution times and volumes of the components of ionic systems over a wide range of eluent concentrations. A. Theory of Small Column Ion Exchange Techniques. 1. Relationships between resolution and basic column_parameters. The resolution and speed of ion exchange chromatography can be increased by optimizing the parameters of column length, eluent flow rate and concentration, resin particle size, column diameter and sample size. Understanding the relationship between these parameters and resolution theory simplifies the task of optimization. Resolution is achieved by maximizing the selectivity of the column or its ability to keep two component zones from spreading into wide bands (9). Resolution is increased by increasing the length of the column. This is evident from the relationship between resolution, zone width,. and zone center separation (5). _ 9%. Rs - w (1) R5 is the resolution, Ax is the zone separation or the distance between two adjacent peaks and w is the zone width or the width of a peak measured at the baseline. Ax is a measure of the selectivity'of the resin in the column: the degree to which it distinguishes between the species causing the two adjacent peaks. W is the measure of the efficiency of 4 the column: the degree to which it keeps zones from spreading. The zone separation is proportional to the column length (L) and the zone width is proportional to the square root of the column length. COUpled with the definition of resolution in Equation (1), one finds the following relationship: Rs = é—-a L -—— = /E' (2) w /L Due to resolution being proportional to the square root of the column length, there is a diminishing return in resolution from successive equal increases in column length. There are also other limitations to increasing resolution exclusively through increases in column length. Extremely long columns packed with fine resin particles may require such high pressures to maintain a convenient flow rate that the pumping capacity of the pump or the breaking point of the hardware is exceeded. Even if a pump and hardware is available which will supply a convenient flow rate at high pressure without leaking, the sample components may be too dilute to detect or the separation may be too time consuming to be practical. Increased resolution obtained through the use of a longer column means the components spend more time on the column during which zone broadening processes take place. They may proceed to such an extent that the concentration of the separated component is below the detection limit of the instrument used. 2. Ion exchange chromatography. Ion exchange chromatography is a rapid analytical method due to the high flow rates that are used. But the high flow rates used are 5 generally much greater than the optimum flow velocity at which zone spreading is a minimum. The precise value of w for a given zone in a given column is a function of many variables. All of these variables can be lumped together into one parameter, H: w = JLH' (3) H or H.E.T.P., the height equivalent to a theoretical plate, is an efficiency parameter which characterizes zone spreading. Expressing the above equation in a different form: H=L (4) H expresses the degree of spreading which occurs as the component zone travels along the column and thus should be minimized. At the flow rates generally used, the greatest contribution to H is due to nonequilibrium between the mobile and stationary phases. This contribution is directly proportional to the flow rate (v) while that due to flow inequalities is independent of v and the contribution due to longitudinal diffusion is inversely proportional to v (10). The time required for separation (T) to occur is directly proportional to H and inversely proportional to v: T « “g3 Since H does not increase as fast as v, the separation time decreases with increasing flow rate. Also an increase in the length of the column will have no effect on the separation time, but will simply increase the time needed for each component to reach the end of the column. Using a slow flow rate, the complete resolution of two components may be achieved at the moment the less strongly retained component passes out of a short column. In order to achieve the same resolution in a shorter time using a higher flow rate a longer column is needed. The fastest separation is therefore achieved by using the longest practical column which will presumably proVide excess resolution and increasing the flow velocity until no more resolution can be sacrificed. As was stated above, at the high flow rates used for fast separations, the major contribution to the total plate height (H) is that due to nonequilibrium of the ion between the eluent (mobile phase) and resin (stationary phase). The result of adsorption and desorption is a random back and forth motion with respect to the zone center which leads to Gaussian spreading. An efficient column requires small distances in the resin across which the ions must diffuse in order to reach the exchange sites. This suggests the use of resin particles of a small diameter. Resins of 200-400 mesh (74-37 microns in diameter) and finer must be used for high resolution chromatography. Small diameter particles eliminate the possibility of deep pools of stagnant eluent and prevent column efficiency losses due to slow distribution of ions between the eluent and resin. While traveling through a column, ions travel back and forth many times between the mobile and stationary phases. With large particles, deep pools of immobile eluent exist in the porous polymer and ions spend more time diffusing randomly in these pools. This results in a greater probability that all the molecules will not spend an equal time in the immobile eluent. At the faster flow rates required for high speed chromatography, ions loitering in these pools lag even further behind those that cross back and forth promptly. This effect creates wide component peaks and the efficiency of such separations is poor. 7 There are some additional advantages of using small particles. As the particle size of an ion exchange resin bead is decreased, the following effects are observed: the time required to reach equilibrium decreases, the efficiency of a given column increases, flow inequalities due to channels in the resin bed are reduced, small bore columns are easier to pack uniformly and the effective ion exchange capacity increases in cases where the active sights inside large particles are unavailable to ions. This last effect manifests itself as an increased selectivity. However some problems arise when using small particles. The back pressure and resistance to flow in a given column increases, the settling rate of the resin decreases and the problems due to clogging of the porous glass or Teflon resin bed support increases. Most ion exchange resins consist of functional or ionizable exchange groups attached to a polystyrene and divinylbenzene (DVB) copolymer matrix. The properties of these resins vary with the nature of the ionizable group, the cross linkage of the matrix (% DVB by weight in the copolymer) and, as previously mentioned, the particle size. Cation resins are primarily classified by the differences in the nature of the active groups as being strongly, intermediately, or weakly basic. Cation exchanges are similarly grouped according to their acidic nature. The effects of cross linkage on the properties of ion exchange resins are the most important of all. The copolymer matrix of the resins consists of polystyrene chains tied together at intervals by the incorporation of divinylbenzene groups: PI PI PI I1 I I I Styrene DVB — —9— — ¢ — __ —— l1 l1 era—CH2 (EH—CH2 O H O H @ © + © hemmLpeLgade; ——¢-—(':-—C——¢- l2hr380C H I"'| H H CJi==CJi The percentage of cross linkage agent used determines the solubility, swelling characteristics and selectivity as well as other physical and chemical properties for a given type of ion exchanger. As the cross linkage increases, the permeability of the resin decreases and accepts ions of smaller hydrated radii and thus becomes more selective. The tendency for the resin to swell with changes in the ionic form or the ionic strength of the eluent is reduced, resulting in a more constant bed volume. The wet volume exchange capacity increases because highly cross linked particles swell only slightly and will, therefore, contain more exchange sites per unit volume than will a resin of low cross linkage. The selectivity also increases due to the high density of ionic sites in the resin, but the equilibrium exchange rate decreases since ion diffusion through the resin is retarded. For this reason resins of greater than X8 (8% DVB) cross linkage are not widely used and nonequilibrium for X8 resins is not a problem when using fine mesh sizes. The selectivity of the anion exchange resin is reflected in the molar distribution coefficient (Kd) and selectivity coefficient (Excf), for a given ion and exchange resin, as measured under stated conditions. The distribution coefficient is the ratio of the amount of ion on the resin per gram of resin to the amount of ion in the solution per milliliter of solution. However Kd is not a constant for a system being considered, but is equal to ON Eé]-/[C13'] (4). The superscript x refers to a particular ion and the selectivity coefficient is generally expressed relative to chloride ions. Qw is the dry weight capacity of the resin in meq/g. The chloride ion concentration is that of the eluent being used. At equilibrium, the concentrations of the ions on the resin and in the solution are related as follows: 5’5,- = £er [cm/[er]r [x'] (5) where the subscript refers to the concentrations in the resin phase. The collections of selectivity coefficients found in the literature (23) are for a solution containing equal concentrations of chloride and the ion of interest at a stated concentration. The selectivity coefficients are static in nature in that the resin in the chloride form is allowed to equilibrate for several hours with the equimolar solution before the resin and liquid phases are separated and analyzed for chloride ion (26). B. Explanation of the Bipolar Pulse Conductance Technique and Comparison with Other Bridge Technigues. l. A.C. bridges with thegphase-angle voltmeter. The bipolar pulse conductance technique, which has been recently developed by Johnson and Enke, is applicable to many systems in which conventional conductance techniques cannot be used, or can be used only with great difficulty (15). Conventional ac bridge techniques cannot be used in flow systems such as in ion exchange where continuous and instantaneous measurements are required. Some error is always associated with methods using only one variable capacitor in the l0 bridge network due to both the double layer (Cd) and parallel (Cp) capacitances of the cell. Both these sources of capacitance can be balanced out by using variable capacitors in the two arms of the bridge which include the cell and standard variable resistor and precisely balancing the two by adjusting the frequency of the signal source. While this method gives a precise measurement, it is time consuming and not easily automated since three parameters must be balanced. The phase-angle voltmeter technique reduces the tedious capacitance balancing problems of most ac bridge methods since it responds only to the portion of the difference signal that is in phase with the signal source. The bridge can be balanced fairly accurately, without the use of a variable capacitor in the standardization arm of the bridge, by adjusting the standard variable resistor, Rs, until the phase-angle voltmeter shows a zero in—phase signal (15). High frequency signal sources produce a positive conductance deviation due to the shunting capability of Cp. At low frequencies there is a negative deviation in conductance due to the impedance of Cx. Thus to achieve true resistive balance, the signal source frequency must be adjusted to the frequency at which the two deviations exactly cancel. If CD and Cd can be determined for a given cell with a given solvent, and R] and R2, the fixed resistors in the other two arms of the bridge, are known, f the frequency for true exact’ balance, can be calculated (21). _ 1/2 1/2 1/2 fexact - [(1 + R2/R1) - ll /[2n(R2/R1)(Cp Cd) RS) (5) 11 In Equation (6) R5 is a high precision variable resistor or potentio- meter. Regardless of the electrolyte present, (Cp Cd)]/2 should be constant. While this use of a phase-angle voltmeter as a null detector in an ac conductance bridge circuit is an improvement over conventional ac bridge methods, the necessity of estimating or measuring C and Cd, calculating f t’ and adjusting the frequency p exac is a time consuming inconvenience. If fexact is a high frequency, R1 and R2 can no longer be considered pure resistances and additional errors must be considered. Also since an ac signal is continuously applied, heating of the cell and solution occurs causing a temperature control problem. 2. The bipolar pulse conductance technique. In his paper, Johnson concludes that "ac bridge techniques can be used accurately if, and only if, one carefully considers the accuracy required, the cell capacitances, the resistance range, the magnitude and frequency of the signal source, the power dissipation of the cell, and the null detector characteristics, and then designs his apparatus, cell, and experiment accordingly" (15). The bipolar pulse conductance technique overcomes many of the limitations inherent in ac bridge techniques. It is a kind of dc technique which can use a standard conductance cell, is extremely insensitive to cell capacitance, provides essentially continuous readings, and has an accuracy previously associated only with the most sophisticated ac bridge techniques. Consecutive constant voltage pulses of equal magnitude but opposite polarity are applied to a conductance cell and the current is measured at the end of the second pulse, giving the conductance 12 directly. Using pulses as short as 10 usec it has the advantage of high frequency ac methods in that Cd develops little polarization. During the first pulse, Cd will charge slightly at a linear rate causing a drop in the current while Cp quickly charges to a constant potential. During the second pulse the current through C and Rx P (the true resistance) will be enhanced by the capacitive potential accumulated during the first pulse. With this reverse in polarity, Cp will charge to the new applied potential. During the second pulse there will be a net decrease in the cell current since Cp will discharge the same number of coulombs that it stored during the first pulse. At the end of the second pulse no current is flowing through Cp, since it is then at a constant potential. Since Cd is completely discharged, the entire voltage drop across the cell is due to Rx‘ The instantaneous current through the cell at the end of the second pulse is i = d/Rx. The instantaneous current measured at eapplie this time is directly proportional to the conductance and essentially independent of Cp and Cd. 3. A comparison of the bipolar pulse and phase-angle volt meter techniques. The use of a phase-angle voltmeter was discussed above in some detail in order to lay the basis for a comparison between the bipolar pulse conductance instrument and a Chromatronix Conductivity Detector. The Chromatronix Conductivity Meter is chosen for comparison for several reasons. Foremost is the familiarity this author has with the instrument since a Microflow Conductivity Cell designed for use with this meter was used in the experiments performed in this study. The Chromatronix Instrument employs a 4 kHz bridge voltage and a l3 synchronous detector amplifier (phase-angle voltmeter) in order to eliminate the out-of-phase (reactive) component and measure only the in-phase (conductive) component of cell current. Thus it is subject to the difficulties and limitations that are discussed previously. The use of a fixed 4 kHz voltage source does not allow the cancelling out of conductance deviations due to Cp and Cd. Chromatronix Incorporated states in a bulletin that "Due to minor phase shifts in the circuit the synchronous detector cannot completely eliminate the capacitive reactance of the cell impedance" (1). Thus the stated accuracy of even relative measurements is :_l%. The bipolar pulse instrument has an accuracy of :_0.03% over its entire range providing all the operational amplifier circuits are balanced periodically. The sensitivities of the two instruments are comparable. The bipolar pulse instrument has a sensitivity of 0.0001% of the total conductance while the detection limit of the Chromatronix instrument is 0.001 umho. Both are more than sufficient for detecting small dilute samples using narrow bore columns. Also the high precision and low noise level and drift of both instruments make them more than adequate due to much larger errors arising from the chromatography apparatus. Both instruments have direct readout in the form of a panel meter and outputs for a chart recorder. There is no Rs to adjust to null on the Chromatronix instrument since the difference signal as detected by the phase angle voltmeter is read out on the panel in umhos and is quite linear. The bipolar pulse instrument can produce pulses of 10, 100 or 1000 usec duration at a repitition rate of 14 from 0.1 to 104 msec. If the repetition rate of 10 msec (100 Hz) is chosen, the sample and hold amplifier will give a continuous analog readout (15). There are three areas in which the bipolar pulse instrument is superior. These are range, zero suppression, and power output. Both instruments have a measurement range of 106 , but that of the bipolar pulse instrument is more useful. The Chromatronix instrument has a range of from 0.001 to 1000 umho while that of the bipolar pulse instrument is l umho to 105 umhos as listed in Johnson's thesis (13). Actually the range goes much lower as shown by the stated sensitivity of 0.0001%. A test of the bipolar pulse instrument using an Electro Scientific Industries DeKabox was conducted by Johnson and repeated by this author. The box was set at 10 kg and then increased by 0.5 n. The change in conductance could easily be observed by using the sensitivity setting of 1 (maximum gain) and the direct readout to a recorder at a full scale range of 10 mV. This is a difference of 0.005 umhos. A 0.1 9 change was not reproducible due to drift and noise. At the other extreme of the range, the bipolar pulse instrument can measure a conductance as high as 105 umhos(10 9) whereas the Chromatronix instrument is limited to 1000 umhos. While the Chroma- tronix instrument would suffice for the separation of phosphates in which the buffered 0.2 M KCl eluent had a conductance of 327 uMhO when detected by the flow cell designed by Chromatronix for their instrument, it would be over-ranged by the eluents used to separate alkali or transition metals which are 3N to 9N for some separations. This problem could be overcome by the use of a flow cell with a higher 15 cell constant, but this would require either a longer cell path or a smaller cross sectional area. The former would increase the cell volume, causing a greater remixing of the separated components and result in a loss of resolution. The latter would require a smaller bore through the electrodes and spacer which would be limited by the practicability of machining a bore of a diameter less than the 0.031" of the present cell. The range of both instruments is further extended by the use of zero suppression. 0n the bipolar pulse and Chromatronix instruments, the maximum zero suppression is equal to the largest full scale range being 105 pmhos and l00 pthS respectively. The zero suppression on the bipolar pulse instrument is supplied by a five decade resistance box (0.005% Vishay resistors). The Chromatronix offset consists of two uncalibrated potentiometers for coarse and fine adjustments. An extremely important advantage of the bipolar pulse technique is the small solution heating it produces due to the power output being very low (0.2 “watt/Hz maximum). Since the slowest repetition frequency at which a continuous analog output can be obtained is 100 Hz, this was employed with a pulse width of 10 usec (20 usec total for each measurement). The Chromatronix instrument uses a continuous ac signal and produces a greater solution heating. Solution heating cannot be tolerated due to most salt solutions having a conductance which changes coefficient of about 2% per °C near room temperature. This is generally equal to or greater than the signal or change in conductance produced by an eluted ion band. Both instruments have temperature compensators which operate with the use of a cell containing a thermistor implanted in one electrode. The Chromatronix temperature l6 compensation has an adjustment which can be set to compensate by 0-3% per °C. This requires an estimate of temperature coefficient of the solution used. The bipolar pulse instrument has a little more accurate temperature compensator. It works on the assumption that the temperature coefficients of the thermistor and solution are linear over l°C (the range for which the instrument is designed). With the temperature compensation off and the solution and cell thermostated to the expected operating temperature, the panel meter is nulled. The temperature compensator is turned on and the meter is nulled again with the use of a 10 turn 100 km potentiometer. The temperature is changed by about a degree and a 10 turn 2 k9 potentiometer is adjusted until the panel meter is nulled again. If the temperature changes, a current that is proportional to the product of total conductance times the temperature deviation is supplied to an operational amplifier-which sums this current with the current that is proportional to the total conductance so the output voltage is the same as it would have been had no temperature deviation occured. If the ratio of the temperature coefficients of the solution and the thermistor is known, the setting of the 2 k9 potentiometer can be calculated and adjusted accordingly (14). A block diagram of the bipolar pulse instrument appears in Figure (1). Both of the above temperature compensation methods are time consuming and at best only approximations. In addition, the thermistor is embedded on the outside of one electrode and is more susceptible to ambient temperature fluctuations than to those due to solution heating which takes place primarily near the center of the electrodes l7 timing circuit L precision sampl & hold offset cur rent generator ampl itier xy/lO multiplier m __j genera tor Figure l. Block diagram of the bipolar pulse conductance instrument. 18 or inside the cell compartment. Considering this, the Chromatronix cell is better suited for use with the temperature compensation of the bipolar pulse instrument than the Chromatronix instrument. This is especially true since the former produces little heating and most errors will be due to ambient temperature fluctuations. The simplest method of eliminating this entire problem is to thermostat the cell as was done in the present study. C. A Comparison of Conductance Techniques with Other Methods of Detection. Conductance techniques are ideal for the detection and quantitative determination of inorganic solutes and other ionized substances separated by ion exchange. Conductance is a general detection method and will detect almost all solutes separated by ion exchange. While ultraviolet absorbtion and differential refractive index detectors are widely used in many other types of liquid chromatography one finds only limited application of these two detectors to ion exchange chromatography. With the exception of nucleic and some carboxylic acids most simple ions have little or no ultraviolet absorbtion. The refractive index detector is sometimes considered to be a universal detector. However this is only true when the refractive index of the sample components and carrier are sufficiently different (19). As the effluents from the column performing the separation and an identical dummy column are usually compared, the refractive indices of the aqueous eluent of moderate ionic strength and the effluent of the operative column containing micromole samples are not significantly different. 19 Although polarographic detection in combination with chromato- graphic separation has its greatest value in the analysis of organic compounds (17), the analysis of metal ions (16, 7,21) and inorganic anions were among the first applications of polarography to chromatography by Kemula in 1952. The method is somewhat selective in that the potential of the dropping mercury electrode (DME) in reference to the saturated calomel electrode (SCE) can be set so as to only respond to a limited number of ionic species. This is especially advantageous in the detection of alkalai metals as separated by cation exchange resins and the separation of the transition metal ions as their chloride complexes by anion exchange resins. The potential can be set to detect only the metal ions and not the gradient HCl eluent since hydrogen is very difficult to reduce at a smooth mercury surface due to its high overvoltage. However the positive potential range is limited due to the electrolytic dissolution of mercury which limits the range of anions which can be detected. A platinum microelectrode has some advantages over a dropping mercury electrode in that the useful anodic range is larger and no current oscillations occur since the platinum electrode has a constant area. However the cathodic range is limited to about -1 volt versus an SCE as opposed to -2 volts for a DME which has a higher hydrogen overvoltage (18). Also with a DME at a fixed potential the mean current is constant since the diffusion layer is destroyed by the renewal of the mercury drop. With stationary electrodes such as the platinum microelectrode the current depends on the time of electrolysis. Deposits formed on the electrode greatly influence the properties of the detector and its reproducibility and make it difficult to clean. 20 Very few applications can be found of the platinum microelectrode in the anodic range. Only the anodic voltametry of ascorbic acid has proven to be very successful. Other organic compounds that can be oxidized at a platinum electrode give insoluble oxidation products, precipitating on the electrode surface and resulting in a decreasing sensitivity as was found for oxalic acid and ethylene diamine tetra acetic acid (8,25). The detection of separated components by automated analysis has gained wide spread use during recent years. Commercially available systems such as the Technicon Autoanalyzer have been used in the analysis of condensed phosphates and carboxylic acids (20,27). The analyzer can be programmed to perform an entire spectrophotometric analysis. The column effluent is continually mixed with reagents which either strongly absorb by themselves or form a complex that will absorb in the visible or ultraviolet range. In the former method the absorbing material selectively oxidizes or reduces the sample components and not the eluent material. The absorbance is monitored and the decrease in absorbance is continually recorded to give the elution curve. In the latter method neither the eluent nor the reagent appreciably absorb in the range of the complex formed with the sample components. These automated methods are more costly and sophisticated than the other methods described above, but give highly accurate and reproducible results with a minimum of operator attention. 11. Experimental High speed-high resolution ion exchange chromatography is only possible through the use of equipment and materials which give uniform and reproducible operating conditions. This is especially true of eluent pumps, constant temperature baths and ion exchange resins. The hardware portion of the apparatus used in this study determined the limit of accuracy and reproducibility of the data obtained. Some errors did arise from the detector and instrumentation and had to be corrected. An overall view of the apparatus followed by a detailed description of each component is necessary in understanding the problems that had to be dealt with in this study. A. Chromatographic Apparatus. With the exception of the eluent pumps, the chromatographic systems were assembled from equipment produced by Chromatronix Incorporated. As can be seen from Figure (2) the design of the system depended on the pump. A CAV-203 valve from C.I. was used to shut off the flow from the pump operated on compressed air. Connections from the pump to a valve and from the valve to a 107B25 sample injection tee were made with 0.063" 1.0. X 0.125" 0.0. Teflon tubing and TEF 107 tube end fittings. The tee is illustrated in Figure (3). A shut off valve is not needed with a Milton Roy minipump since turning the motor off quickly stops eluent flow. A 500 ml erlenmeyer 21 22 iniection tee pulse a . :::“:::f./ l 4L ulse—free t—reservoir (A motor pump ‘ . driven compressed m r pump column f—water jacket (9 micoflow ce||—\ -- bipolar pulse voltage conductance follower &RC filter chart recorc'i7 instrument 7 MFA Figure 2. Schematic of the chromatographic apparatus. 23 syringe Figure 3. Sample Injection Tee. to the ' d e ectro es conductance ‘5‘ instrument ETC FE1 N [NE . {Winlet tube Achernfistor Figure 4. Microflow Conductivity Cell 24 flask with a ground glass top and side arm with stopcock contained the eluent during deaeration by a Cenco Hyvac 2 pump and serves as a reservoir for the eluent pump. Connections between the pump inlet and reservoir and between the pump outlet and three feet of natural rubber tubing were made with the same C.I. fittings and tubing. The rubber tubing was followed by five yards of 0.031" 1.0. X 0.063" 0.0. Teflon tubing to form a surge-damping network. The other end of this small diameter tubing was connected to the sample injection tee. Samples were injected into the tee with a # 701 10 pl syringe from Hamilton Co. The inlet head is threaded into the coupling assembly of a borosilicate glass column to make a tight seal. The glass columns were contained in a Plexiglass water jacket. The conductance of the column effluent was measured with a Chromatronix MCC-75 Microflow Conductivity Cell which is shown in Figure (4). The column is connected to the cell by a 0.012" bore Teflon tube of 5 ul internal volume. The stainless steel electrodes each have an internal volume of 1.2 ul while that of the Kel-F spacer separating them is 1.5 ul. A 50,000 ohm thermistor is epoxied onto one electrode for temperatureemeasurement. ‘The stainless steel housing was removed and the inner assembly was tightly wrapped in a coil of l/4" 0.0. copper tubing which was surrounded by a one piece Styrofoam block. A constant temperature bath with a pump and a portable cooling bath both from Neslab Instruments, were used to thermostat the conductivity cell and chromatography column. The temperature was maintained at 25.2°C :_0.05°C. 25 B. Pumps and Related Apparatus. In its most elemental form, high pressure liquid chromatography requires only a column and a solvent reservoir which is pressurized with gas. The first pump used in this study was operated on pressurized N2 which pushes down on the surface of a solution in a beaker and forced it out of a tube, the end of which was immersed in the solution. Due to the N2 dissolving in the solution under high pressure and bubbling out in the flow cell near atmospheric pressure, this pump was not applicable to in-stream effluent analysis. A pump was needed that had the capability of delivering a continuous stream of eluent under high pressure without the pulsing problems of most reciprocating pumps. It also had to be of a design that prevented the direct contact of compressed air with the eluent. .A pliable solventreservoir under the pressure of a tank of compressed air was decided to be of the best design since it would be both simple to operate and could be built of readily obtainable materials. 1. A pulse-free gas pperatedppump. The pump shown in Figure (5) was built. The chamber lid and coupling for the polyethylene bag are made of aluminunL A polyethylene Playtex nursing bag is attached to the coupling with an expander. The coupling is of the same dimensions as the top 1 5/8'I of a Playtex nursing bottle. Rubber O—rings make a seal between the bottom side of the lid and the top edge of the chamber and between the lid and coupling in two places. Compressed air is introduced to the chamber by way of a 1/4" copper tube and Swagelok fitting in the lid. The gas can be released by a shut off valve connected to the lid with 1/8" 0.0. Teflon tubing. There are two outlets in the lid. One 26 e ltgent ergpty V'compressed air tobleed—off valve column refill T “—z‘ II 6.8.5 A; '1: _ 4.50"» 4 Figure 5. Diagram of the Pulse—Free Gas Operated Pump. 27 goes to the column and is controlled with a bleed-off valve and the other is used to add more eluent without disassembly and to purge the bag of air before pumping is begun. The outlets are made with 1/8" 0.0. Teflon tubing with flared ends which are recessed in the bottom side of the lid. The lid is coated with Dow-Corning water-proof sealant to avoid contact with the aluminum lid by the eluent. The bag was coated with stopcock grease to reduce permeation of air into the bag. When the pump was carefully assembled with the O-rings and the bag greased and checked to insure there were no leaks and the solvent deaerated, a gas free stream of eluent could be delivered to the column. Once the pressure to the pump was carefully regulated, a flow rate that did not vary by more than 5% per hour could be obtained. Some care was needed to insure a constant and reproducible flow rate. The Teflon column bed support of 10 u pore size had to be prevented from fouling up with fine particles in order to insure that a constant back pressure was presented to the outlet of the pump. This was accomplished by packing the first few millimeters of the column above the bed support with 30-50 mesh resin beads. The most difficult problem was that of asserting a constant and reproducible pressure to the chamber. For a given column which presents a constant back pressure, the flow rate is proportional to the applied pressure. The 15 cm long columns packed with 200-400 mesh resin and porous bed supports only required 0.5 to 10.0 p.s.i. to obtain flow rates of 0.150 to 0.250 ml/min. The delivery pressure gauge registered 0-60 p.s.i. and the scale was not divided into equal increments in 28 the 0 to 5 p.s.i. range. Thus it was difficult to set the regulator at one or two p.s.i. accurately and reproducibly. This problem could be solved by joining a well calibrated, high precision, low range pressure gauge to a connection tee between the pump and a needle valve at the outlet of the regulator. With the bleed-off and needle valves closed, the regulator could be set to a pressure reading slightly higher than that needed for a given flow rate. The needle valve could then be opened and the chamber pressurized to the exact setting needed and the needle valve closed. When tightly secured, the chamber is observed to hold a set pressure for several days after the main valve on the tank has been closed. 2. A positive pressure reciprocating eluent pump. Before a further investment of time and money was made in the pulseless pump described above, a Milton Roy motor driven, controlled volume minipump was purchased. It is a reciprocating positive displacement pump designed to meter liquids in measured volumes against a positive differential pressure between the pump suction and discharge. The pump performs this metering function within a repeatable accuracy of :_l%. The pump, as shown in Figure (6), consists of a drive unit, a plunger and a displacement chamber in which the plunger reciprocates. The pump delivers a controlled volume of liquid with each discharge stroke unless the outlet system becomes obstructed or extremely constricted. The pump capacity is adjusted by changing the plunger stroke length. The desired stroke length is set by means of an adjustment knob and sleeve with a calibrated scale much like a Vernier caliber. The adjustment can accurately be set to 0.2% 29 .8228 cm>wcu copes mom coppwz ms» co Ewcmmwa .o mczmwd iota Sarge 3.0:... acid... tomcat“ HMWWW\ I"'. .fl”VflaaZZZ92&rW MmmWWHV-V J , ,.,._I.\\_.=\g\_ 22:22:: __.I .\\../ _ 111; a. 523. :NNW\§ . NNAU?7& .coEoua_am:u W {I’ll/flail. IL 4 ‘y .. r c: O . 3 h H .— L 4 1|. _ 5 uzccouuo xon _ i scion—60.. c208 Loom 30 of full stroke which is 0.001 inch of stroke adjustment. Once the stroke adjustment is set the pump will deliver a constant flow rate regardless of the column to which it is connected (except in the case of an obstruction in the output line or the attainment of a very high back pressure which activates thermal overload protective devices in the motor starting circuit). While the motor driven pump is easier to set at a given flowrate with repeatable accuracy than the pump previously described, it has a pulsating output which is quite noticable when running a chromatogram. The motor drives the plunger once every two seconds and superimposes a jogging motion on a continuous signal which gives it the appearance of a DME polarograph. A flowrate of 0.21 ml/min is obtained at 6.9% of full stroke. When pumping 0.015 N KNO3 as an eluent, a noise pulse of 2 mV is produced. A 10 ul sample was the upper volume injection limit of this study. When 10 ul of a solution 0.1 N in KF was injected, a peak of about 21 mV was observed. If a correlation between sample size and peak area is desired, this introduces an intolerable source of error. Much time and effort were spent in developing a surge-damping network which would effectively reduce the pulsating output of the pump. Neither Milton Roy nor Chromatronix manufactures such a device. Several networks were constructed by interposing several feet of rubber tubing followed by five to six yards of 0.031" 1.0. Teflon tubing. Many problems arose from the flexible tubing portion of the network. It is difficult to connect flexible tubing to Chromatronix tubing and fittings without leaks developing. Thick walled Tygon 31 tubing was not flexible enough to damp out the surge and thin walled Tygon tubing would inflate and give an erratic and irreproducible flow rate. The final system used consisted of three yards of 1/8" 0.0. natural rubber tubing which was epoxied to specially machined Kel-F fittings with inch long stems, the other end of which threaded into standard Chromatronix couplings. When the bed support was kept free of fine particles this system gave a constant pulseless flow. Although these connections hold longer than those with Tygon tUbing, they do occasionally break. The best compromise between total pulse damping and reliability could most likely be obtained by using a diaphram type pressure gauge in place of the flexible tubing. C. Instrumentation The bipolar pulse conductance instrument used in this study was designed and built by Dr. Donald E. Johnson and is described in a previous section of this thesis. His thesis and publications should be consulted for greater detail (13). The instrument was operated at a pulse repetition frequency of 100 Hz and a pulse width of 100 usec so as to obtain a continuous analog output signal without appreciable solution heating. The precision offset generator was used to bring the panel meter to within a few millivolts of null balance once a steady conductance was established. The use of a voltage follower operational amplifier circuit was required when recdrding the output signal of the bipolar pulse instrument. Both the Omnigraphic 3000 Recorder from Houston Instruments and the EUW-20 Recorder from Heath had low input .me_wa om cm:o__0d mmmu_o> mgu mo Emcmmwo pwsucwu .N mcsmwm 32 J1 .vcmu l .10. .14 EoEatmf 3.2. .2033 ._ 50.; _Ocm_m 15gb 70" > E E 560- .5 40+ 4.: $50- :5’ 30 2404 22’ I 8 30" g 204.. 20+ 10' lO~r ' : g , . . 4 i 10 15 20 25 30 2'0 3'0 4T0 Time in minutes Time in minutes Flow rate: 1.180 ml/min Flow rate: 0.200 ml/min .SO-w Flow rate: 0.586 ml/min 1 Flow rate: 0.384 ml/min 60‘- 40‘" 50‘. > > E E .E .C ‘0 4430'“— 4—1 4 .C .C: U) U) :2 :; ao~~ .5220“ x f6 f0 33 83 20‘- lO-r lO-~ L1 1 1 a 1 A I : p = _11 5 1‘0 1'5 2‘0 2‘5 3'0 5 10 15 20 25 Time in minutes Time in minutes All samples are 10 p1 of a solution 0.10 N in F- and 0.18 N in 103 separated with a 0.015 N KNO3 eluent. Figure 11. Elution time and resolution as a function of flow rate. . m' tmax 1n 1nutes 47 <3-103' andlj-F' for 10 p1 samples of a solution (3 0.10 N in 103' and 0.10 N in F' eluted by / 30.- 0.015 N K1103 / / / / / / / / / 201~ / / D C) O / / ./ ./ / ./ / / / / ./ / o ./ / D ]()4+ C] / ./ / ./ / / 13 / o ./ / / / El/ / ./ 1 1 1 1 1 C) 1.0 2!) 313 ‘41) 51) l/Flow rate in min/m1 F1gure 12. Plot of tmax y§_1/flow rate. 48 Table 3. Data from the study of elution volume as a function of flow rate for the fluoride-iodate system.* Flow Rate l/Flow Rate in tmax (elution time) Vmax (elution in ml/min . min/ml in min volume) in ml Fluoride: 0.204 4.90 18.0 3.67 0.206 4.86 18.25 3.76 0.371 2.70 10.5 3.90 0.384 2.61 9.75 3.74 0.586 1.71 6.50 3.81 1.180 0.848 3.50 4.13 19923:: 0.204 4.90 30.75 6.27 0.206 4.86 31.50 6.49 0.371 2.70 17.75 6.58 0.384 2.61 17.00 6.53 0.586 1.71 10.88 6.37 1.180 0.848 5.63 6.64 *The data was collected from the separation of 10 pl samples of a solution 0.10 N in F' and 0.18 N in 10 ' with a 0.015 N KNO3 eluent. 3 .mpmc 30pm m> ll XME > to po_a .m_ meamwu cwHWP\am cw cowpmcpcmucou pcmapm «<0 omo mod. cod vod «Qc o T I ” u L “ 0d 11011 l 11. hp .11 .11 nwz 11 III! ‘I ‘III ‘I IrOQV D el-l .1nvn 9 4 -invo 0‘ III Ill I all. Oil..- Il- Au 11 .11 .1. 11 11. .11 av SEE 08.0 pm 83:. 1.. E z o; 10K new 1moH z m_.o covpzpom 8 mo mmFQEMm F; e co; unungvcm 1moH1O xew A 1111 v.1 50 a function of flow rate, it appears that there is actually a slight increase in Vma with flow rate. Since Vmax for F' increases at a x slightly faster rate with increased flow rate than Vmax for 103-, the volume separation of the two Vmax's decreases. This effect only slightly decreases the resolution as the flow rate increases. The greatest contribution to the decrease in resolution with increasing flow rate is from zone spreading. In order to keep zone spreading to a minimum, a column must operate as close to equilibrium conditions as possible. The deviation from equilibrium conditions manifests itself as a lagging of the center of a given maximum ion concentration in the resin phase behind the corresponding center in the eluent phase. This leads to greater zone spreading as the flow rate increased and is responsible for the subsequent loss of resolution. 4. Effect of eluent concentration on elution volume and resolution. The eluent concentration of KNO3 was varied in order to investigate its effect on the analysis time and the resolution of a separation. The study was carried out by measuring t the time needed to max’ obtain the maximum concentration of an eluted ion in the column effluent, for fluoride and iodate while varying the concentration of KNO3 in the eluent, Ce. Knowing the flow rate of about 0.200 ml/min, Vmax can be determined. The data collected from the injection of 4 p1 samples of a solution 0.10 N in F' and 0.18 in 103' and subsequent separation by eluents of varying concentrations appear in Table (4). As one can see from Figure(14), Vmax increases exponentially as the eluent concentration is decreased. The choice of eluent 51 Table 4. Data from the study of elution volume as a function eluent concentration for the fluoride-iodate system.* Ce (eluent 1/Ce in Flow Rate tmax (elution Vmax (elution conc.) in l/eq in ml/min time) in min volume) in ml eq/l Fluoride: 0.010 100 0.211 25.0 5.28 0.010 100 0.211 24.5 5.17 0.015 66.7 0.196 17.75 3.48 0.015 66.7 0.197 18.0 3.55 0.015 66.7 0.204 17.25 3.52 0.015 66.7 0.204 17.5 3.57 0.030 33.3 0.211 9.25 1.95 0.030 33.3 0.211 9.25 1.95 0.030 33.3 0.211 9.38 1.98 0.030 33.3 0.211 9.25 1.95 0.060 16.7 0.202 5.00 1.01 0.060 16.7 0.202 5.13 1.04 0.060 16.7 0.202 5.38 1.09 0.060 16.7 0.202 5.45 1.10 0.100 10.0 0.212 3.80 0.807 0.100 10.0 0.212 3.90 0.827 0.100 10.0 0.212 3.80 0.807 Iodate: 0.010 100 0.211 40.88 8.32 0.010 100 0.211 42.5 8.97 0.015 66.7 0.196 30.0 5.88 0.015 66.7 0.197 30.0 5.91 0.015 66.7 0.204 29.0 5.92 0.015 66.7 0.204 29.0 5.92 0.030 33.3 0.211 15.13 3.19 0.030 33.3 0.211 14.75 3.11 0.030 33.3 0.211 14.75 3.11 0.030 33.3 0.211 14.75 3.11 0.060 16.7 0.202 7.50 1.52 0.060 16.7 0.202 7.50 1.52 0.060 16.7 0.202 8.00 1.62 0.060 16.7 0.202 8.20 1.66 0.100 10.0 0.212 5.25 1.11 0.100 10.0 0.212 5.40 1.14 0.100 10.0 0.212 5.25 1.11 *Same conditions as in Table 4 but with 4 pl samples. 52 1 o <3-103' andtj-F' for a sample solution 0.18 N in 10 ' and 0.10 N in F' eluted 8 01- 3 ’ at 0.200 ml/min 50m- 0 E] 'E 5- 4.0-- X 1'6 :35 Cl (3 2.0" C] <3 D o E] 1 t i *** i 0 4 6 8 l 0 Eluent concentration of KNO3 X 102 N Figure 14. Plot of Vmax !§_C e. 53 concentration is therefore critical if a fast separation is to be achieved. By replotting the data in Table (4), as Vmax versus the reciprocal of the eluent concentration, the linear relationship in Figure (15) is obtained. This graph lends itself to a more accurate interpolation and estimation of Vmax for a given eluent concentration. The resolution of F- and 103' decreases rapidly as Ce increases. This can be seen from Figure (16). There are several reasons for presenting the data in this fashion as opposed to calculating Rs. There is no standard way of expressing Rs and several different methods are found in the literature (5). It becomes increasingly difficult to measure w or some other zone width parameter as resolution decreases. Finally this method of comparing resolutions has become increasingly common during recent years. An interesting comparison can be made between two separations of the same ions obtained by using different flow rates and eluent concentrations. ' with 0.015 N KNO Eluting F' and 10 at a flow rate of 1.180 ml/min 3 3 the tmax's are 3.50 minutes and 5.63 minutes respectively. Eluting F' and 103' with 0.10 N KNO3 at a flow rate of 0.212 m1/min the tmax's are 3.80 minutes and 5.25 minutes respectively. It will be noted that the data collected for the working curve was obtained from separations run at 0.2 ml/min with a 0.015 N KNO3 solution. A comparison of these two separations must be made guardedly due to the difference in sample size. However, as is stated above in subsection 2, under identical conditions of flow rate and eluent concentration there is no observable difference between the resolutions obtained with 4 pl and 10 p1 samples of a solution 0.10 N in F' and 0.18 N in IO3 If these two chromatographs, which are found in Figures (16) and (11), are compared, 54 00:19. x2; .5 no: .2 23: m_oe\mtmp__ at mU\_ 00F cm 00 CV ON T u + n u \\ \ \ \\ 1 d \m \ \ \ \ \O \\ \\ mu \\ 11n¥N \ \ \ \ \ \ \ \ 1\ \ O \ \1 D \ \ . 1\. \ \ 110V \ \ \ HU\\ \\ \\ .\ no .-0pv \ \ \ -1o.m O \ acm=_a -mozx gap: :_E\_E oom.o pm nana_m L E z 2.0 2; .me E z 2.0 8.538 a .5 8353 _ e .58 1... 2:... .ms. 5.3 [ll] Lil. XPUIA Peak height in mv 55 Peak height in mv Eluent concentration: 8Ck‘ 170H Peak height in mv .3 k) to £5 <3 <3 <3 c: l L 1 l #1 1 Eluent concentration: I 1 1 l 5 Time in minutes Peak height in mv 1'0 1'5 2'0 Time in minutes 0.03 N KNOB 0.01 N KNO r 1604 140‘ o~ on " :3 <3 <3» 55 <3 I 1 l 1 l I I l 4Cki I 2‘0' 2'5 ' 3'0 3'5 ' 4'0 4'5 3 1.3 5 1'0 351- ‘53 15" Eluent concentration: 1% Peak height in mv Time in minutes 10, 1'5 2'0 2'5 3'0 3'5 Time in minutes 1504 1401 5 :5 <3 c: 1 l 80” 3 8 l L N O 1 1 I Eluent concentration: 0.06 N KNO3 l . I 2 Time in minutes 0.015 N KNO3 L 1 2 I 1 468 'Eluent concentration: 0.10 N KNO 3 All 4’pl samples of a solution 0.10 N in F' and 0.18 N in 103 were separated at a flow rate of 0.200 ml/min. Figure 16. Elution time and resolution as a function of eluent concentration. 56 they appear to be identical with respect to resolution. One may conclude that over the range of flow rates and eluent concentrations studied, the attainment of shorter and identical elution times by either increasing the flow rate or eluent concentration results in separations of identical resolution. There would seem to be no advantage in using one method over the other in order to decrease the time of analysis. The separation time, the time between the tmaxls of F’ and 10 ', of the sample run at 1.80 ml/min with a 0.015 N KNO3 eluent is 2.13 minutes. The separation time for the sample run with a 0.10 N KNO3 eluent at 0.212 ml/min is 1.45 minutes. The zone width, w, which was measured by extrapolating lines drawn tangentially to the side of the 103' peak to the extrapolation of the base line, is almost 50% wider for the sample run at the higher flow rate than that run at the lower flow rate using the 0.10 N eluent. It is expected that it would be wider since this was a 10 pl sample whereas the other zone width was produced by a 4 p1 sample. The resolutions of the two separations thus appear to be the same since the sample with the greater separation time also has the greater zone width. If it is assumed that w would have been significantly smaller had a 4p1 sample been used, this would seem to favor the use of a higher flow rate for obtaining a decreased analysis time with a minimum loss of resolution. However, one would expect the separation time to be smaller for a 4 p1 sample. One may see this by observing Figure (10) for which tmax is proportional to vmax' For both F' and 103', tmax decreases with decreasing sample size. Since the tmax of a I03' peak decreases more rapidly with decreasing sample size than that of the F' peak, a smaller separation time would be expected for a 4 pl sample. While no absolute conclusion can be 57 reached without actually running both studies with samples of the same size, it again appears that it is equally detrimental to the resolution of two components whether identical decreases in analysis time are achieved by increasing the eluent concentration or by increasing the flow rate of a separation. 5. Determination of distribution and selectivity coefficients. The graph of Vmax versus l/Ce in Figure (14) can be used to calculate Kd and EXN03' for both F' and 103' over a wide range of eluent concentrations. The values obtained are dynamic values since they are calculated from data obtained under Operating conditions and are of greater practical value. The selectivity coefficients found tabulated in the literature are static in nature as was previously discussed in the section on the theory of high speed ion exchange techniques. Since a selectivity coefficient is not a constant, these values are of limited use. Equations (7) and (8) are used to calculate Kd and ExN0 -. Before these values can be calculated, V0 and Qw must be knogn. In Figure (14) the y intercept of a line drawn through the data points gives Vm = V0. The values of V0 obtained were 0.300 ml and 0.275 ml. These values were checked by repeatedly injecting 10 pl of 1.0 N KNO3 into the columnirt a flow rate of 0.220 ml/min and recording the unsorbed effluent peak at a recorder chart speed of 2.0 in/min. A Vo value of 0.299 :_0.05 ml was obtained. The Dowex 1 X 8 had a nominal exchange capacity of 3.28 meq/dry gram in the chloride form and was calculated to have an exchange capacity in the nitrate form of 2.95 meq/dry gram. 58 Since the plot of Vmax versus 1/Ce is a relatively straight line, one may assume that the selectivity coefficient is nearly constant over the eluent concentration range of 0.01 N to 0.10 N. From the equation: v = v 0w EXNO3- 1/[N03'] + v0 (9) one sees that the product Vo Qw EXN0 - is the slope of the lines in 3 Figure (14). By using this relationship, a selectivity coefficient is obtained which is an average of those that can be calculated from the equation: (v - vo)[No3’]/vo ow = EXNO - (10) mx 3 3 for each of the several sets of data points. The values EFN0 - = ID ' 3 0.0554 eq ml.1 9"1 and E 3 _ = 0.09435 eq m1"1 9'] were calculated from N0 3 the slopes of the lines in Figure (14). Bauman and Wheaton (26) determined EFC]- to be 0.090 eq ml'1 9'] for 20-50 mesh Dowex l X 8 by a classical method which was previously explained in the section on the theory of ion exchange chromatography. Using this value, a selectivity coefficient for F' with respect to N03- of 0.0237 eq ml'1 9"1 was calculated from the relationship: EFN0 - = - N0 - 3 EFCI-lE C1" It is difficult to make a comparison between the two values of EFNO - since the selectivity coefficient depends on the 3 conditions under which the value is determined. The extent to which this method of determining selectivity coefficients can be used in calculating Vmax for a given ion will depend on several factors. The accuracy of Vmax depends on the accuracy of the molar distribution coefficient which in turn depends on that of 59 the selectivity coefficient. The accuracy of the selectivity coefficient is usually limited by that with which the exchange capacity is known. The error in an estimated Vmax will depend on the extent to which the conditions for the determination of the selectivity coefficient differ from those under which a separation is performed. The estimated Vmax will be too small for separations performed on a sample larger than that for which the selectivity coefficient was determined and will be too large for separations performed on samples which are smaller than those used in the selectivity coefficient determination. The size of the error will depend on the concentration of the eluent used. Variations in sample size will not shift the relative proportions of free and adsorbed counter and sample ions as greatly in systems employing more concentrated eluents. Thus the selectivity coefficient and Vmax will be more nearly constant. If the eluent is dilute as in the previous study of vma versus sample size, Vmax will vary greatly x due to the large shifts in the selectivity coefficient. These shifts were caused by the injection of a solution 0.10 N in F' and 0.18 N in 103' into a column being eluted with 0.015 N KN03. Holding all other conditions constant, Vmax will still increase slightly with increases in the flow rate since the selectivity coefficient approaches a constant only near equilibrium conditions. At higher flow rates, the center of maximum ion concentration in the resin phase lags behind that in the eluent phase causing a slight shift towards larger values of Vmax as well as greater zone spreading. B. Separation of Phosphates. Quantitative-qualitative methods for the determination of phosphates have recently become an area of increasing analytical interest. 60 The separation of an orthophosphate-pyrophosphate mixture and subsequent detection by the bipolar pulse method was investigated. The problems and limitations of this method are considered below in comparison with other chromatographic and nonchromatographic means of detection. 1. Quantitative-qualitative analysis. The separation of orthophosphate and pyr0phosphate was carried out on a 0.28 X 15 cm glass column packed with 160-325 actual wet mesh Dowex l X 8 in the chloride form. The resin was pretreated as described previously. The sample injections of a solution, which were 0.1 M both in orthophosphate and pyrophosphate, were eluted with a 0.2 M KCl solution buffered at a pH of 5.0. The areas of the separated peaks, which were resolved at a flow rate of 0.275 ml/min, were determined by the method used in the previous study of 103' and F'. Three to four injections of each sample size were separated. The data in Table (5) is the average of the several values obtained at each sample size. This data appears in Figure (17) plotted in the form of peak area as a function of sample size. The points appear to form straight lines and it would seem that the method could easily be used for accurate phosphate analysis. While this is true for the analysis of orthophosphates, the determination of the amount of pyrophosphate present in a given sample is not so simple. When a steady pumping rate is achieved and the drift in the baseline conductance is small, a single analysis for orthophosphate is accurate to about :_5%. However, the amount of pyrophosphate can be accurately determined only by averaging the results of many runs on each sample. The reasons for this difficulty are discussed in the following section. 61 Table 5. The average peak area data from a series of injections of a 3' and P 0 4‘. solution 0.10 M in both P04 2 7 Sample Size Ion Peak Height Peak Width Peak AEea in pl in cm at 1/2 Peak Ht. in cm in cm 4 9043' 18.5 0.85 6.53* 4- 4 9207 5.0 0.90 4.55 6 9043‘ 10.5 0.90 9.37 6 92074' 6.60 0.90 5.95 8 P043- 13.1 1.05 13.6 4- 8 9207 3.5 2.3 7.95 10 9043’ 14.5 1.05 15.1 4- 10 P207 3.9 2.5 9.80 *The peak area was adjusted to an absolute scale. 2 Peak area in cm 1 51)‘r l 01)“ 51)- I 62 3- andIJ-P 0 4 o-PO in both P04 and P207 eluted with 0.2 M KCl buffered at pH 5.0 eluted at 0.277 ml/min w- «r l 213 ‘41) 151) 813 L J l Sample size in pl 3- Figure 17. Peak area y§_sample size for P04 and P207 4- / / ' for a sample solution 0.1 M ./ O 4_ 1131) 63 The working curve in Figure (17) only extends down to the 2 pl sample size since at and below this sample size, the pyr0phosphate peak was difficult to detect and highly irreproducible. An attempt was made to obtain a smoother working curve by the injection of a solution 0.05 M in orthophosphate and 0.2 M in pyr0phosphate. The same sample sizes were again used with the addition of the 2 pl size. While this resulted in a range over which the pyrophosphate peaks were more easily detectable, it also produced errors resulting from an over- loading of the buffer capacity of the eluent. This is easily seen from Figure (18) which compares the peaks resulting from 10 pl injections of various phosphate solutions. The injection of the solution which was 0.1 M in both phosphates gave two peaks whereas that of the solution 0.05 M in orthophosphate and 0.2 M in pyrophosphate produced three peaks. The additional peak which begins eluting before the injection peak has completely passed out of the column is due to hydroxyl ions. The unbuffered sample solutions are at a pH of 9 to 10. If the sample injected is too concentrated, the buffering capacity of the eluent is exceeded and hydroxyl ions, produced by the hydrolysis of P207'4 ions, pass through the column in an appreciable amount. This explains the relative thinness of the orthophosphate peak (the second peak) produced by the solution which is 0.05 M in orthophosphate and 0.2 M in pyrophosphate in comparison with that produced by 0.05 M orthophosphate alone. Since the orthophosphate peak begins eluting before the hydroxyl peak has passed out of the column, its width is less than that it normally would be. The orthophosphate peak is also taller due to the small second peak evident in the chromatograph of 0.2 M pyrophosphate. This small peak l.l[£ll.llll‘.ii'lllll 64 30011 3001- {l 25018 50.- E E .5 200“- .c 200.- 2 150-- E 50‘r .E‘ .9 g 1001- 3 1001- .¥ .52 (U f0 33 50" 8.’ 50"] W i '11: ‘1 1&4 1 ‘ 1 1 1 41 510152025 5 101520 25 Time in minutes Time in minutes Sample: 0.05 M in PO43' Sample: 0.1 M in P043- . and P 0 4‘ Flow rate: 0.285 ml/m1n 2 4 Flow rate: 0.246 ml/min 2001' 300T ISO'W' 250—11— ) 100-1- > 200-- E E .E 50.1- .5 150-- 22:" E .31 01' .21 1004— 0) 0) .C .C .g -50+t _¥ 504- 5 n E 11 ”100T1 _03 ‘450‘* ._504.pv -‘200ru d —q(x)+ 111- : 1 1 1 ~11 1 1 1 1 4 5 1015 20 25 5 1015 20 25 Time in minutes Time in minutes Sample: 0.20 M in P2 74' Sample: 0.05 M in P04 and . 4- Flow rate: 0.285 m1/min 0'20 M 1" P2 7 Flow rate: 0.272 ml/min Figure 18. Effluent peaks resulting from 10 pl injections of various phosphate solutions. 65 is a result of some of the pyrophosphate not being in the form of H2P207'2. This small peak is only evident when the solution 0.2 M in pyrophosphate alone is injected and has a tmax of 6.25 minutes. The peak produced by orthophosphate alone has a tmax of 4.24 minutes while that due to orth0phosphate produced by the separation of the phosphate mixture is 4.75 minutes. The shift toward a longer tmax is due to the small peak eluting at 6.25 minutes which also makes the height of the orthophosphate peak larger than normal. 2. Comparison with other chromatographic and nonchromatographic methods of analysis. Although the use of conductance detectors in ion exchange chromatography is not new, the use of the bipolar pulse method appears to be superior to most commercially available instruments. The bipolar pulse method also has several advantages over several other chromatographic and non-chromatographic techniques in the analysis of phosphates and other systems of interest. However the bipolar pulse method has some of the same limitations inherent in all conductance methods, but unlike most conventional ones it offers the possibility of overcoming these limitations. In recent years, Javier, Crouch, Malmstadt and Ingle (12,3,11) have developed several rapid and highly accurate kinetic methods for the analysis of phosphate and other anions. These rapid mixing and stopped- flow spectrometric methods ascertain the orthophosphate concentration of a sample by determining a pseudo first order reaction rate. The stepped-flow method (12) and one of the rapid mixing methods (11) determine the orthophosphate concentration by relating it to the rate of reaction of Mo(VI) with P04"3 to form the yellow 12-molybdophosphoric 66 acid (12-MPA). The other method (3) is quite similar except that it follows the formation of molybdenum blue through the reduction by ascorbic acid of previously complexed 12-MPA. In particular, the stopped-flow method of analysis by Javier, Crouch, and Malmstadt (12) can analyze for phosphates at the level of 0.1 ppm with less than 1% relative standard deviation in the average result for a given sample at a rate of one result per second. For the analysis of phosphate in blood serum and other biologicals it is far superior to the method of analysis investigated in this study. About 0.25 ml of a sample having a phosphate concentration of 1 ppm can be analyzed whereas the bipolar pulse-chromatographic method requires a l to 10 p1 of sample 1 to 10'2 N. The sample solution must at a concentration of about 10' be of a concentration of 10"2 N to insure a sufficient number of equivalents of phosphates are introduced by a 10 pl injection. The rapid injection of a volume greater than this into the flowing eluent stream of a column with a void volume of 300 p1 contributes to a significant decrease in resolution through zone spreading. While both methods possess the ability to detect and.analyze for about 10'8 eq of orthophosphate, the kinetic method can do so with about 1% accuracy whereas the conductance method has an accuracy of about 5 to 10%. For systems containing large concentrations of condensed phosphates, the separation of the ions by ion exchange chromatography with subsequent detection by the bipolar pulse method has several advantages over the kinetic methods. Since conductance is a general method of effluent detection, it is sensitive to all types of ions. Although the analysis time for the orthophosphate-pyrophosphate mixture of about 25 minutes is much longer than the time of 1 second required 67 for the determination of orthophosphate by the st0pped-flow method (12) and 30 seconds for the fast mixing method, neither of these kinetic methods can directly analyze a mixture of phosphates. These methods cannot detect the condensed phosphates without their prior conversion to the orthophosphate forms. This requires the use of approximately 6 N sulfuric acid to insure complete degradation to this detectable form. The optimum nitric acid concentration is between 0.2 M and 0.5 M for the stopped-flow method to insure a fast reaction. The concentration is adjusted to around 0.9 M in the fast mixing method to form heteropoly blue in order to insure that the reaction proceeds at a slow enough rate for the initial pseudo first order reaction to prevail for about 30 seconds without consuming more than 5% of the sample. In either case the pH of the degraded phosphate sample must be adjusted before an analysis can be run. This would lead to an apparatus that is much more complicated than the chromatographic system used in this study if an automated analysis of a mixture of condensed phosphates is desired. One of the most accurate automated methods for the analysis of condensed phosphates relies on the reaction of orthophosphate with acidic ammonium molybdate to form 12-MPA and the subsequent reduction by hydrazine sulfate to the heteropoly blue complex (20). A chromatographic separation by gradient elution is monitored spectro— photometrically to determine the orthophosphate concentration by measuring the absolute absorbance due to the blue complex rather than its initial rate of formation. A Technicon Autoanalyzer continuously performs a quantitative acidic degradation of the condensed phosphates and isolates a portion of the resulting orthophosphate by dialysis. 68 This is reacted to form 12-MPA followed by the reduction to the blue complex. The concentration of the complex is colorimetrically determined and recorded to give an elution curve. The most sensitive range of detection for the Autoanalyzer is 10 pg of P205 (20). This is the combined amount of P205 due to each phosphate species in the aliquot. By comparison the detection limit for the bipolar pulse method is 4 pg of P04'3 and about 35 pg of P207'4. While the accuracy of the Autoanalyzer method is generally :_3% for most major components in a sample, the accuracy of the bipolar pulse method is :_5% for the determination of P04'3 and only :_10% for P207'4. Although being a general detectoriin~ all ions gives the bipolar pulse method an advantage over other eluent detectors in the simplicity of the design of the apparatus and its operation, being a general detector limits the applicability of the method in the detection of complex systems such as mixtures of condensed phosphates. The accuracy in the determination of P207'4 is much greater for the Autoanalyzer method since a narrow and more nearly Gaussian peak is obtained through the use of a gradient elution technique. This allows all the condensed phosphates in a mixture to be separated with nearly complete resolution, as was the case in this study, but with a minimum of zone spreading. While the use of an eluent of unchanging concentration will suffice in the separation of two ions having similar selectivity coefficients and the same charge, it will not satisfactorily separate a mixture of ions possessing various charges. The separation of PO 3' 4 2 possible by the movement of H2P2072' being more strongly retarded and P 074’ with complete resolution in 25 minutes was only than that of the less highly charged H2P04' as both ions were eluted 69 through the resin bed. The use of a reproducible concentration gradient, which requires a precision gradient former, is not readily applicable to most conductance methods. As mentioned previously, the change in conductance due to an eluted component is generally less than 2% of the background conductance. The precision offset generator of the conductance instrument could be adjusted such that the recorder signal would be zero volts at the initial concentration of the gradient. However the full scale range of the recorder needed to cover the entire gradient range over which the separation of a complex mixture takes place would be too large to produce component peaks of a size that could be observed,much less integrated. The analysis of the data from the separation of a mixture of ions by a gradient elution technique will be successfully performed only after the data handling has been automated through a system such as a small computer interfaced with the bipolar pulse instrument. CONCLUSION Although the quantitative results obtained in this study are not as accurate and reproducible as those obtained by other methods of chromatographic detection, the bipolar pulse conductance instrument has proven to be a simple and effective, general detector for routine analysis by ion exchange chromatography. This investigation has provided the information needed to overcome the technical problems of the chromatographic apparatus which limited accuracy and reproducibility to a level far below that of the conductance instrument. This is a hardware problem which can be solved mainly by the use of better commercially available equipment. If this is done, the bipolar pulse conductance instrument should prove equal or superior to most present methods of detection. This instrument also offers the advantages of greater versatility in routine analysis than most "selective" methods, great ease of adaption in changing from one system to another and is less expensive to assemble and operate than most autoanalyzers. The bipolar pulse conductance technique will offer other advantages in the future as a result of the present trend toward automated systems operated by small laboratory computers. A much simpler and less expensive instrument is required if the technique is to be computer interfaced since a substantial portion of the present instrument is digital in operation. Most of the digital timing functions are duplicated by most small computers. Thus a computer interfaced bipolar pulse 70 71 instrument would be simpler and less expensive than the present instrument due to the elimination of most digital timing components, while most other methods would become more complicated and expensive. The small computer also offers a means of eliminating a present limitation of the conductance instrument. Gradient elution ion exchange chroma- tography can be performed with the use of even the most sophisticated of nonlinear gradients by recording the conductance of the eluent at regular intervals. The analysis of a sample can then be performed by the use of the same gradient and the elution curve obtained from a point by point subtraction of the gradient conductance data from that of the actual analysis. BIBLIOGRAPHY 11. 12. 13. 14. 15. 16. 17. 18. 0400on BIBLIOGRAPHY Chromatronix Inc., Bulletin MCD 170, Berkeley, California. Condal-Bosch, L., J. Chem. Educ., 41, A235 (1964). Crouch, S. R., and Malmstadt, H. V., Anal. Chem., 33, 1090 (1967). Dean, J. A., "Chemical Separation Methods", Van Nostrand Reinhold Co., New York 1969, p. 103. Giddings, J. C., "Dynamics of Chromatography Part II, Marcel Dekker, Inc., New York 1965, p. 34. 3913,, p. 268. Gierst, L., and Dubru, W., Bull. Soc. Chim. Belges, 33, 379 (1954). Hedenburg, J. F., and Freiser, H., Anal. Chem., 33, 1355 (1953). Helferich, F., "Ion Exchange“, McGraw-Hill, New York, 1962, p. 458. Horvath, C. G., Preiss, B. A., and Lipsky, S. R., Anal. Chem., 32, 1422-28 (1967). Ingle, J. 0. Jr., and Crouch, S. R., Anal. Chem., 33, 7 (1971). Javier, A. C., Crouch, S. R., and Malmstadt, H. V., Anal. Chem., 51, 239 (1969). Johnson, 0. E., Ph.D. thesis, Chemistry Department, Mich. State Univ. (1970). 1913,, p. 82. Johnson, D. E., and Enke, C. 0., Anal. Chem., 43, 329 (1970). Kemula, W., Roczniki,Chem., 33, 281 (1952). Koen, J. G., and Te Maarssen, 6., "Continuous Polarographic Detection In Column Liquid Chromatography", Elseiver Publishing Company, Amsterdam, The Netherlands, 1970, p. 11. Ibid., p. 19. 72 19. 20. 21. 22. 23. 24. 25. 26. 27. 73 Leitch, R. E., and Kirkland, J. J., Ind. Res., 13, 36 (1970). Lundgren, D. P., and Loeb, N. R., Anal. Chem., 33, 366 (1961). Mohnke, M., Schmunk, R., and Schotze, H., Z. Anal. Chem., 313, 137 (1966). Parker, H., J. Amer. Chem. Soc., 33, 411 (1931). Peterson, 5., Ann. N.Y. Acad. Sci., 33, 144 (1954). Reiman, W., and Breyer, A. C., "Chromatography: Columnar Liquid— Solid Ion-Exchange Processes", in I. M. Kolthoff and P. J. Elving (Eds.), Treatise on Analytical Chemistry, Part 1, Volume 3, Chap. 35, Interscience, New York, 1961, p. 1555. Skobets, E. M., and Atamanenko, N. N., Zav. Lab. 2191 (1949). Wheatpn, R. M., and Bauman, W. C., Ind. Eng. Chem., 53, 1088 1951 . Zerfing, R. C., and Veening, H., Anal. Chem., 33, 1313 (1966). MICHIGAN STATE UNIVERSITY I 111 1111111 I Ill 1 3 1193 03083 03 ll 6 ARIES