Column liquid chromatography - Analytical Chemistry (ACS

---

Anal. Chem. 1908, 60,387 R-435 R (C24) Slavh A. J.; Underhill, P. R.; Young, M. B. J . Vac. Scl. Technol. 1986, A4, 118-122. (C25) Evans, Stephen; Hiorns. Anthony G. Surf. Interface Anal. 1986, 8 , 71-73. (C28) Si&. R. J . €/electron Spectrosc. Relat. Phenom. 1987. 42, 107-125. (C27) van Veen, Nlcolaas, J. Anal. Chem. 1987, 59, 2088-2091. (C28) Werthelm, G. K. Appl. F'hys. 1986, A41, 75-81. C29) Andera. V. Czech. J . Phvs. 1087.. 837.. 625-837. iC30j Egelhoff, William F., Jr. 6bys. Rev. Left. 1987, 59, 559-562. ~~

~~

~~

~~

(C31) Mizokawa, Yusuke; Mlyasato, Tatsuro; Nakamura, Shogo; Gelb, Kent M.; Wllmsen, Carl W. Surf. Sci. 1987, 182, 431-438. (C32) Sluda, R. Surf. Scl. 1986, 177, L1011-1014. (C33) Surf. Interface Anal. 1987, 10, 58-61. (C34) Tomich, D. H.; Grazulls. L.;.Grant, J. T. Surf. Interface Anal. 1987, IO, 87-91. (C35) Erlckson, N. E.; Powell, C. J. J . Vac. Sci. Technol. 1988, A4, 1551-1556. . ... . -.

(C36) Erickson, N. E.; Powell, C. J. Surf. Interface Anal. 1986, 9 , 111-119.

Column Liquid Chromatography Howard G. Barth*J and William E. Barber Hercules Incorporated, Research Center, Wilmington, Delaware 19894

Charles H. Lochmuller Department of Chemistry, Duke University, Durham, North Carolina 27705

Ronald E. Majors E M Science, 111 Woodcrest Road, Cherry Hill, New Jersey 08034

F. E. Regnier Department of Biochemistry, Purdue University, West Lafayette, Indiana 47907

INTRODUCTION This review covers the fundamental developments in the field of column liquid chromatography (LC) during the period of 1986-1987. In an attempt to conserve space, we have excluded patents and theses and included only articles published in English, French, German, and Russian with some exceptions. This review is not a comprehensive coverage of all LC literature. We have attempted to critically select only those references that reflect fundamental developments in LC theory, methodology,and instrumentation. Your suggestions and comments are most welcome and should be sent to the senior author (H.G.B.). Our main data base for this review was CA Selects (HPLC and GPC) from December 1985 to December 1987. In addition, each author used other search routines to augment coverage from CA Selects.

BOOKS AND SYMPOSIA PROCEEDINGS General books on HPLC have been written by Engelhardt ( A l ) ,Katz (A2), Miller (A3),Schoenmakers (A4),and Styskin (A5). Engelhardt and Hupe (A6) also published a volume entitled Coupling Methods i n HPLC. Berridge (A7) has a volume on LC optimization. Kiselev and Jasin (A8) issued a book on gas and liquid adsorption chromatography. Gliickner (A9)has written a comprehensive text on separation of polymers with emphasis on synthetic polymers. Volume 25 of the popular series, Advances i n Chromatography, was published by Giddings et al. (AIO). There have been a number of books published on specific HPLC topics and applications: affinity chromatography(A11 , A12), detectors (A13-A16),ion chromatography (A17-A19), Present address: E. I. du Pont de Nemours & Co., Inc., Central Research & Development Department, Experimental Station, Bldg 228, Wilmington, DE 19898.

preparative LC (A20-A22), HPLC of small molecules (A23), biopolymers (A24-A26), biochemicals (A27-A29), inorganics (A30),and synthetic polymers (A31);and clinical (A32) and pharmaceutical applications (A33). Symposia proceedings have been published for the International Symposium on HPLC of Proteins, Peptides, and Polynucleotides (A34,A X ) , the International Symposium on Column Liquid Chromatography (A36,A37), the Symposium on LC-MS and MS-MS ( A B ) ,the Advances in LC Conference (A39),the Report and Discussion Session on HPLC (A40), the A. J. P. Martin Honorary Symposium (A41), and the Seminar on HPLC of Organohalogen Compounds in Water (A42). An ACS Symposium on SEC (A43)and a symposium proceedings of the International GPC Symposium '87 (A44) have also been recently published.

REVIEWS A listing of reviews on selected applications is given in the reference "B" section and are arranged by topic. This is not a comprehensive compilation since the focus of this review is on fundamental studies of HPLC. Papers dealing with selected applications can also be found under specific chromatographic techniques in other sections.

THEORY AND OPTIMIZATION There are still a number of problems which must be solved before more progress can be made toward establishing a theoretical basis for liquid chromatography. There is a real need for fresh combinations of theory and computational methods to assist in the prediction ab initio of optimal conditions of speed and selectivity based on known or suspected structures and/or a minimum number of direct measurements. A variety of approaches for the latter have appeared which are elaborations of previous methodology and which appear to give improved results. Perhaps the most vexing problem in liquid chromatography is the question of the determination of the "true" void volume.

0003-2700/88/0360-387R$06.50~0 0 1988 American Chemical Society

387 R

COLUMN LIQUID CHROMATOGRAPHY Howmd Q. 8 M h has recenny pined the staff 01 the A n a w l Mvislm 01 the cenbal Research 8 DBvebpment Department at Du Pont Exparlmental Station. Wilmingtm, DE. He Was prevlcurly a Research Scientist and Oroup Leader st Mcuias R e m r c h cater. He received his B.A. (1969) and F'h.0. (1973) in analytical chemistry horn N 6 eastern University. Betwe joining Hercules Inc. in 1974. he was a postdoctoral fellow in Clinical chemistry at Hahnemann Medical College in Philadelphia. He is a frequent k t u r e r at Short Course6 SpOnsWed b; the ACS Polymeric Materials Science and Enginmrina Division. His soecialtles include . powmer characterization. barticle size anaiysis. size eiciuslon chromamgraphy. and high-performanceliquid chromatography. He has published over 40 papers in these areas and has also edited a bo&. Modem &mods of Pa& cle Size Analysis. published by Wiley. Howard Is also senlor author 01 "Applicatbns Review of Particle Size Analysis" published biennially in AnaMica1 Chemislv. Banh is on the 1n~humentatk.n Advisory Panel 01 Ans!~ifcal Chemishy and is Associate Editor of the Jownal of A p p l M PolySclence. He is Cofounder of the International Symposium on Polymer Analysis and Characterization and is presently Chairman of the Delaware Section of the ACS. Dr. Barih is a member 01 the divisions 01 Analytical Chemistry. Polymer Chemistry. and Polymeric Materials Science and Engineering of the ACS. ASTM. AAAS. and the Delaware Valley Chromatography Forum.

WIIHam E. Barber is R e m r c h Chemist with the A n a w l Mviskm 01 M c u k Research Center in Wilmington. DE. He is group leader of the liquid chromatography. polymer analysis. and particle size analysis seckms. t. He recsivBd his B.A. in chemistry st the University 01 Vermont in 1977. Before at! tending graduate school. he Spent 6 months .I at American Cyanamid In Stamford. CT. He received his Ph.0. in analytical chemistry at the University of Minnesota in 1983 under the direction 01 P. W. Carr. Bill joined Hev Cubs in 1983. He Spent 9 months during 1986 as an LC specialist with Hewlen-Packard. He repined Hercules in 1987 where he is invoived in HPLC methods development and coordinating analyticai eflolts lor new prOduCt development. He has Several publications in the areas of chromatographic peak shape analysis. detector lineariw, indirect UV detection. and paired-ion chromatography. Bill is a member Of the Divlsion of AnaWical Chemistry Of the ACS and the Chromatography F m m 01 the Mware Valley.

'--.

~

~~~

~

~

H. Lockmo(lw Is Prolessor 01 Chemistry and 01 Biochemical Engin&np at Duke Universny (Assistant (1969-1974) Associate (1975-1979)). He came to Duke aner two years ptdoctorai at p u d w university (1967-1969) with L. B. Rogers and graduate w w k (M.S. 1965. W.D. 1968) at Fwdham University with Michael Celoia. A Fdbw of the Royal Chemical Society and of the American Institute 01 Chemists. he served as Chairman of the Anahnical DivC I sion of the American Chemicil Society (1983-1984). He is a member of the Corn minm 01 Revision. United States pharmaco pelal Convention (1985-1990) and has been an advisor to the USEPA on Environmental Assessment Programs Snd is a member 01 several review paneis inchding the Integrated Air Can- Reject. He was appointed by the National Research Councll to SBNB ~1 mS AnalyK caI Chemktry Panel WhlCh adviser and evalwtes the programs d me Center 01 Analytical Chemistry-National Bureau Of Standards (1987-1990). He Serves on the Editorial B08rds of the Journal Of Chromalogrsphlc S c h c e . the JOwMl of ChemOmeMn. and PreparaNve Chmmsfcgraphy. the Editorial Advisory Board tar Chemically Modlfed Surfaces, and the Advlsory Board 01 m e Handboo* of Trace Substanms. I n approximately 100 published papem. hls interests have ranged to include such diverse areas as anaty4cal robotics. proton-induced Xqay emision anaysis. mass spectmmetry. and nuclear magnetic resonance but his main efforts are in the area 01 separation SCC snce. especially in the mlecuiar basis for selectivity in chromatography. I n 1985 he received the Pioneer in Laboratory Robotics Award sponsored by me International Symposium on Laboratory Robotics and in 1967 the Arne6 can Chemical Society Award in Chromatography. He consuits with industry in the area of chromatographic instrument design. bonded-phase chemistry. and large-scale separations. He is the auther of three chapters on gas ChrDmatography in instrumental analysis texts and a chapter on bonded phase Chemistry in a handbwk on liquid chromatography and has taught numerous courses on separation methods including many invited Iectu~eson various advanced topics. He has S B N ~on the organizing CommMee of and has chaired numerous symposia sponsored by a varlety 01 learned societies. I n addition. he has Offered lectures in both the Soviet Union and the People's Republic of China under the sponsorship of the Academies 01 Science of those nations.

Chad..

388R

ANALYTICAL CHEMISTRY, VOL. 60, NO. 12. JUNE 15, 1988

Rm*d E. Worn is Qenmai Uannw. matography 01 EM SCIENCE in C k m y Hill. N.I. He received his B.S. in Chemiatry at California State University. F r o m . in 1983 and his F'h.0. at Purdw Unbersity In 1968 under the dlmction of L. B. Rogers. He then pried Ceianese Research in Summit. W. where he supervisad the Separations Lab iw zV2 years. returned to cainmia in 1971 with Varian Associates In Walnut Creek and in hls 18 years at Varian had several assignments in R&D. applications. and ma&eting in the U.S. and Europe. h. M a w has over 100 publications in HPLC. GC. and surtsce chemistry. He has served as Special Editor 01 the Jownal of ChromalDgraphic Science for issues d b voted to LC Columns and is currentiv Editm' of a monthlv feature "Column Watch in LCIGC ~ ~ r l lor n which e he Is also on the Editorial Board. He has served as an aUlhor for lhe Fundamental ReviewS on Liquid Chromatog raphy in AnaMiwI Chembhy lor the last six years. He is a member of the Chromatography Society. Chromatography Forum of Delaware Valley. the ACS. Analytical Dlv.. and c ~ r e n t l yis Chairman 01 the Sub-Division of Chmmatwaphy and Separatbns Chemistry. He also serves as a member of Analfllcal Chemlsfv's Adviswy Board on Instrumentation. Or. Majors S B N ~8s Chairman of HPLC '86 in Sa" Francisco. ~

~~

~~

~

~

~~~~

~~~

es in analytical biochemistry and blochemistry. Dr. Regnier Is me aumOr 01 more than 110 papers and reviews on llquid chromatography and has Rve patents on Chromatographic packing materials. His re. search interests are in the area of hlah-m. larmance liquid Chromatography of 'biipolymers. He is a member of the American Chemical Society. the American Society 01 Bioicgical Chemists. and Slgma Xi. Dr. Regnier WweI on the editorial board ot Analufical BioChemishy. A m W I Chemfshy. and Liquid Chromafepphy Magazine.

The inability to make such a measurement prevents the establishment of a true retention index system (although that is not the only reason), the use of liquid chromatography for thermodynamic measurements, and the development of physicochemically based approaches to optimization and retention prediction. There has been mounting evidence that no single void-volume marker can serve as is the case in gas chromatography because every molecule may have its own characteristic void volume. In reversed-phase as well as in normal-phase chromatography with mixed mobile phases the retention and mass transfer mechanisms may be complimted by multiple phase equilibria, pore exclusion, and even interparticle resistance to mass transport which depends on the molar volume of the solute and its chemical character. If this is true, then methods for void volume determination based on the nonlinear regression of the Martin rule (In (t, - to)/@!) vs carbon number) may he of little value, for example. It IS likely that more evidence supporting the complicated nature of the measurement will appear in the near future. Recent papers in this area have been published by Knox and Kalizan ( C l ) ,h u b and Madden (CZ),Niedhart et al. (CSr), Grushka et al. (C4). and Smith et al. (CS). Helfferich (C6) reviewed the theory of multicomponent chromatography. Meyer (C7j discussed liquid chromatcgraphy theory with emphasis on practical applications. Adsorption kinetics were studied by El'tekov et al. (CS) and Marshall and co-workers (C9). Roias et al. (CIO)described a computer program used for the ch;omatographicsimulation of ion exchange. "Optimizat;onm continues to be a loosely defined term in liquid chromatography as many authors seek to put forth a new optimi7ation strategy or methodology. Few papers make clear what an optimum is considered to he: (a1 A method that works. (hi A method that works best in a global sense. (c) A method that combines adequate resolution with throughput (shortest maximum retention time). (d) A method which combines (c) with lowest detection limits or best linearity. Optimally. the chromatography audience would be better served if, at this stage of development in the field, authors ~

~

~~~~~

COLUMN LIQUID CHROMATOGRAPHY

would clearly state what can be optimized by their approach and how rugged the approach is. A method that only works for samples containing equal concentration of analytes and matrix elements is not very rugged. The same is true of methods which locate only local rather than global optima. Several reviews of current “optimization strategies” have appeared (C11-CI 7). Chemometric methods (multivariate approaches including principal component, factor analysis, and pattern recognition) may provide significant advances toward a generalized optimization strategy. A survey of current approaches appeared (C18). The venerable simplex method appears in various forms in studies of optimization. Given the complexity and number of local maxima in resolution one can expect in a ternar mobile phase resolution surface, this method is best applied: if a t all, to HPLC separations where the region of optimal conditions is already known. A randomly cast simplex approach is likely to find any local maxima. Readers are directed toward several examples which illustrate the scope and difficulty of brute-force simplex methods (C19, C20). Isocratic methods of elution remain most desirable because of their simplicity, the elimination of analysis time devoted to column equilibration, steady output to detectors, etc. Gradient methods can produce faster actual analysis times (time to produce separation vs time to analyze a subsequent sample) and better detection limits when transparent solvents are used, and, in some cases, complex gradients are the sole solution to achieving adequate resolution of all analytes and detectable matrix components. Methods for the optimization of both types of approaches continue to proliferate. Computer simulation or combination of simulation with minimal actual data collection has been reported and discussed. Snyder and co-workers (C21, C22) present what appears to be a relatively rugged approach involving a minimum number of gradient runs which allows the prediction of precise isocratic retention values. Hartwick and others (C23) have dealt with the question of gradient elution in coupled columns. Other approaches have been reported which promise varying degrees of universality (C24426). Can we hope for a system which, when presented with a new sample in a new matrix and a few desiderata, will produce a time/detectability/etc., optimized method and results? The question of the “fully automated HPLC” fantasy was discussed by Berridge (C27). Schoenmakers’ book (C28) presents a good deal of what is known about selectivity manipulation and optimal method development and supports the current need for skilled-human/instrument interaction. Control of secondary equilibria coupled with solvent selectivity provides an extra measure of flexibility in controlling and optimizing separations. Foley and May (C29, C30) present thorough examples of how this can be done methodically. A review of optimization approaches in ion chromatography (in German) has appeared (C31). One of the difficulties in optimizing HPLC separations lies in the compatibility of suitable or intended detectors with the best conditions for separation and the tendency to find optima for detectors which may be far from the resolution/speed optima. Where possible, it would be very useful if authors could suggest how far from the separations optima the compromise must be made to get optimal detector, injector, or transfer device response (C32, C33). De Galan (C34) presents an interesting discussion of mobile-phase optimization in RPC involving photodiode-array detection. The separation of high-molecular-weight oligomers by gradient elution in the absence of steric exclusion has been demonstrated in recent times. Whether this observation can be treated by conventional theories of elution or is a special case of “critical solution” (not Flory’s critical solubility), continues to be discussed (C35, C36). While clear evidence has been presented from both sides and both theoretical approaches seem to work, the case seems unproven as to whether elaboration of the traditional theory is sufficient to reject the “critical solubility approach” (or whether these are materially different in the limit). (See SEC section).

COLUMNS Column advances in this review period have focused on new selective bonded- and coated-phase columns (such as highperformance affinity, metal-loaded, ion chromatography), a

further reduction in particle size to 1.5 pm @ I ) , columns for the separation of enantiomers (02),and many more silica- and polymeric-based columns for the biochromatography of proteins, peptides, and nucleic acid constituents. As the competition for commercial columns heats up, prices for “generic” (e.g., octadecylsilane (ODs), amino, silica gel) columns continue to decrease aided by improvements in the design of cartridge systems with relatively low priced replacement cartridge columns. Microbore columns of 1.0 mm i.d., a few years ago, thought to be the panacea of HPLC column technology, have achieved minimal success while short, “fast” LC columns continue to see greater usage, especially in the quality control/routine area. Research on micro- and capillary LC columns of 0.5 mm i.d. and below goes on but commercial instruments to handle them still lag behind. A better understanding of the silica surface and bonded-phase properties has helped to design better, more stable columns but reproducibility of HPLC columns is still the number one concern of liquid chromatographers. General reviews on trends in stationary phases (01, 0 3 ) , bonded silicas (04,05), preparative columns ( 0 6 ) ,and new HPLC resins (07)were published. A general review by Engelhardt (08) on the role of the column included coverage on dispersion and retention mechanisms. Japan has always been a leader in the developmentof gel packing materials and these developments were surveyed in a general manner by Hatano (09) and in a more limited coverage by Nakamura (010). The trend in HPLC is to spherical packing particles but a short review of nonspherical packings showed some application examples (011). An interesting survey (012) of HPLC experts on the future of HPLC column technology provided the following observations: (1) preparative and process columns will continue to grow; (2) affinity and other specialized stationary phases will continue to be developed; (3) alternative stationary phases to silica such as polymeric materials will come onto the market; (4) microbore (1 and 2 mm i.d.) columns will probably be superceded by micro- and capillary columns but interfaced with low-flow-rate “information-rich” detectors such as MS or FT-IR; (5) fast LC columns will see increased use in routine analysis; (6) column-to-column reproducibility and stability are the most serious problems in LC columns today. New columns introduced at the Pittsburgh Conferences were the subject of a series of review papers (013-016). Major advances in commercial analytical and preparative columns for all HPLC modes were covered as well as for column accessories, guard columns, sample preparation devices, and gas chromatography columns. Columns for biomolecules were the subject of a review by Floyd and Hartwick (017). The authors discussed selectivity enhancement by the use of multiple retention mechanisms particularly with the separation of nucleic acids. The separation mechanisms in reversed phase, ion exchange, size exclusion, and hydrophobic interaction modes for large biomolecules were also discussed by Snyder and Stadalius (018). Column Packlng Technlques

Modern HPLC columns are generally packed by highpressure slurry techniques. Verzele and co-workers (019) have concluded that (1) for the production of the best columns below 10 pm, spherical particles are preferred, (2) for columns above 10 pm, irregular particles are preferred, (3) the smaller the particIe the more difficult to pack efficiently and the higher the packing pressure needed, (4) for particle sizes below 10 pm a low-viscosity packing solvent is preferred, (5) slurry concentrationdoes not appear to be critical, and (6) a different packing method is required for practically every packing. His observations are based on months of production of commercial slurry packed columns. In another paper, a low-viscosity “up-tube”(upward) packing procedure was claimed to provide a dense and uniform structure for sub-l0-M.mparticles (020). For 10-pm polar particles, downward packing by a nonbalanced density method was suggested. Other methodsbalanced and nonbalanced density, viscosity slurry, chemical stabilization, and mechanical stabilization-were also discussed. Other techniques on filling columns with silica gel materials were published by Maisch and Grom ( 0 2 1 ) and Hosko and Barths ( 0 2 2 ) for large scale columns. ANALYTICAL CHEMISTRY, VOL. 60, NO. 12, JUNE 15, 1988

389R

COLUMN L I Q U I D CHROMATOGRAPHY

Testing and Characterization

Silica gel is the dominant packing material in adsorption chromatography in HPLC. Microparticulate silica is mechanically stable, relatively inexpensive, and reactive due to silanol functionality. Especially in its chemically modified forms, silica finds use in all LC modes. Often, in the chemical bonding of phases to silica, unreacted silanol groups are removed by silylation with a small organosilane such as trimethylchlorosilane (TMS). Marshall and co-workers (023) studied the chemical reaction of silica with various amounts of TMS prior to octadecylsilylation. They found by FT-IR measurements that the stronger associated silanols (5% of the total) reacted more quickly and an ODS phase prepared under their conditions gave the most efficient phase. Using a variety of spectroscopic and chemical techniques, Koehler et al. (024) undertook a comprehensive characterization of some silicabased stationary phases. They found that unfavorable adsorption of basic compounds and low hydrolytic stability of alkyl bonded phases could be attributed to the existence of isolated, non-H-bridged, highly acidic Si-OH groups. Contrary to popular opinion, these workers also discovered that, for the lowest absorptivity for basic compounds, a silica gel support should have the highest concentration of homogeneously distributed associated or bonded Si-OH groups to ensure a minimum concentration of the highly acidic isolated silanols. To measure total silanol groups, several workers (025) developed an isotopic exchange method with the silanol protons using deuterium. The characterization of bonded-phase silicas has been the subject of several studies. The most comprehensive study of three well-characterized synthesized silica gels of loo-, 200-, and 250-A pore diameter was undertaken by Sands et al. (026). After measurement of surface area, pore volume, and mean pore diameter, the silicas were reacted with C-1, C-4, C-8, and C-18 dimethylmonochlorosilane. They observed the following: (1)the concentration of bonded phase increases as the pore diameter increases; (2) the level of bonding decreases linearly as the chain length of the bonded phase increases; (3) surface area and pore volume are reduced the greatest for smaller diameter bonded silica, especially as carbon-chain length increases. In testing the phases with small nonpolar molecules, they found that k’ and theoretical plates slightly decreased as a function of pore diameter and pore volume and increased as a function of the surface area of the bare silica gel. For proteins, resolution increased as the pore diameter and pore volume increased, the increase being the greatest for the higher molecular weight proteins. A new method for testing HPLC packed columns is based on the experimentally found pressure dependence of preferential sorption of mixed eluent components on the column packing (027). When column pressure is suddenly changed, for example due to back flushing, a change in composition of the liquid sorbed on the sorbent surface occurs and therefore the effluent composition is altered. The shape of the eigenzone of the new composition leaving the column observed with a refractive index detector) can be used to judge sorbent bed uniformity and efficiency of the column packing procedures used. Back flushing makes heavy demands on the sorbent bed, and poorly packed columns were destroyed. Well-packed columns, however, were unaffected and the author proposed the column back flushing method to measure column stability. In the use and testing of HPLC columns, the extra-column band broadening contributions from the HPLC instrument is of considerable importance in order to preserve the column contribution itself. Another procedure for measuring these extra-column effects was the subject of a paper of Claessens and co-workers (028). Merely replacing the column with a union or capillary tube is insufficient. Instead, the linear extrapolation method based on the relative influence of instrument variance with test solutes of different k’ values seemed to work best provided certain conditions were met. In the measurement of k’, the determination of the void volume (holdup time) is of utmost importance. A review on this subject was published by Smith and co-workers (029). Polymeric-based packings have grown in importance, mainly due to their higher pH stability. Stuurman et al. (030) investigated polystyrene-divinylbenzene (PS-DVB) copolymers recommended for use in reversed-phase chromatography (RPC) and found that PS-DVR from two manufacturers are 390R

ANALYTICAL CHEMISTRY, VOL. 60, NO. 12, JUNE 15, 1988

similar IR spectroscopically but surprisingly they were not homo eneous nor stable during use. A variety of solutes were testedg. For nonpolar solutes, T H F is recommended as an organic modifier. Uncharged acids and anions presented no difficulty but amine cations should be chromatographed at low pH on these polymeric phases. Elution order on a PSDVB column for the drug aztreonan and its metabolites followed conventional RPC retention and suggested a similar mechanism (031). Dawkins and co-workers (032) compared polystyrene (PS)-based and polyacrylamide (PA)-based packings for RPC applications. Upper pressure limits were lower than for silicas, 5000 psi for PS and 3500 psi for PA. In RPC, retention behavior was different for the two packings. In contrast to most forms of LC, plate height was reduced at higher temperatures, interpreted in terms of a reduction in solute dispersion due to improved mass transfer in the stationary phase. Nevejans and Verzele (033) described the swelling propensity factor as a measure of swelling of semirigid packing materials used in HPLC. The swelling is measured by plotting a pressure drop across the column during a solvent gradient. Commercial and homemade polystyrenes were tested as well as a nonswelling ODS packing. Chloromethylation of the materials diminishes the swelling by a cross-linking reaction. Column Hardware and Design

Most columns in HPLC are constructed of stainless steel. Other materials which have been used include glass, fused silica, titanium, and glass-lined tubing. Highly elastic polymeric materials have been investigated (34) and have been applied to the analysis of vitamins (035). The design and production of silicone LC columns was the subject of a Japanese patent (036). Practical aspects of compact glass-cartridge columns were covered by Kehr et al. (037). Topics covered included a Teflon-guard column design, intracolumn sample injector, replacement method for contaminated packing, procedure for shortening the column length, and a water-jacket design. The internal surface of stainless steel tubing in HPLC column was the subject of a brief review (038). Methods of processing stainless steel tubing to produce “mirrorlike” finishes were covered. Data from several studies were cited but any evidence of the presence of a “wall effect” for any of these studies was inconclusive. For small i.d. columns ( < 2 mm), where the ratio of column wall area to the column cross-sectional area is larger, the wall effect seems to have some bearing on performance. In such columns, however, other experimental factors causing band broadening such as extra-column effects and packing reproducibility tend to overshadow these subtle effects. Cartridge columns have gained in popularity in HPLC. Among their advantages are lower replacement costs, integral guard column designs, easy connection and disconnection, and, in general, convenience. A new cartridge system was described and compared to standard hardware columns (039). A disposable plastic cartridge, usually used for sample preparation, was adapted with commercially available fittings and used for preparative separations (040). A new column design with reuseable, hand-tightened end fittings was described in a patent (041). The design has movable ends which allow pressure to be exerted on the packing material to ensure a tight packing structure or to correct any voids or channels which may develop after use. Column Maintenance and Troubleshooting

HPLC columns, especially silica-based, usually have finite lifetimes. With periodic maintenance and the use of guard columns, column lifetimes can be extended. Column problems reviewed by Majors (042) included back pressure, plugged frits, poor reproducibility, sample recovery, resolution loss, instability, voids, and leaks, among others. Possible reasons for on-column sample loss and suggestions on ways to avoid these problems were offered by Abbott (043). Factors which change column retention time were covered by Abbott and Simpson (044). Among those factors discussed in detail were gradual coverage of active sites by irreversibly adsorbed material, bonded-phase cleavage, gradual reaction of column surface, and silanol-catalyzed partial hydrolysis of the bonded

COLUMN L I Q U I D CHROMATOGRAPHY

phase. Flow reversal after packing a column void with pellicular packing, matching the stationary phase of the analytical column, was proposed by Vendrell and Aviles (020)for RPC columns. If the column was “topped off“ and the flow not reversed, resolution was not as good and the column lasted for a shorter period of time. The degradation of silica-based RPC columns using trifluoroacetic acid containing mobile phases for protein and peptide separations was the subject of a study by Glajch and co-workers (045).Monomeric bonded phases are especially susceptible to cleavage. Degradation occurs most rapidly with fresh, totally covered materials. Surprisingly, certain columns, despite the losa of over half of the bonded phase, still exhibited adequate performance. However, reproducibility was affected. For reversed-phase columns used in the separation of peptides, on-column rederivatization to reduce exposed silanols was investigated for C-8 and (2-18 phases (046). The C-8 (C-18) column was regenerated by passing a solution of octyldimethylchlorosilane (or odadecyl-) in dichloromethanethrough the deteriorated column. Improved peak shapes and increase in retention was evident for both types of bonded phases studied. Extending column lifetime by the use of guard and scavenger columns was covered by Dong and co-workers (047). Characteristics of dimensions, packings, dispersion, convenience, and sample capacities of these protection devices were discussed and useful examples shown. Columns which were stored for long periods of time (2.4-5.7 years) on the shelf can undergo considerable decreases in efficiency and changes in selectivity (048).

I NSTRUM ENTATI ON Instruments and components are now becoming more reliable with better performance at more attractive prices. For example, complete multiple solvent, single-pump gradient systems which have a wide dynamic flow range and flow precision of 15 000 plates/m allow these materials to be used in HPLC. Pentyl-, octyl-, and poly(ethy1ene glycol) derivatized agarose media have proven effective in the separation of both model proteins and serum proteins ( S I , S2). A high-capacity hydrophobic adsorbent has also been developed for the removal of human serum albumin from Cohen fractionated serum fractions (S3). New resin-based HIC media with phenyl-, and oligo(ethy1ene glycol) stationary phases have been described for protein separations (S4, S5) that were produced by introduction of hydrophobic stationary phases into the Toyo Soda G5000 PW size exclusion matrix. Silica-based supports are also widely used in HIC. Apparently the largest difference between resin and silica-based materials is stability under basic conditions. Polar-bonded phases with diol, alkyl ethylene glycol (ether), or amide functionalities prepared by silylation of silica have been found to function as either SEC or HIC sorbents depending on the mobile phase (SS-SS). The dynamic loading capacity of a 300-A pore diameter ether-phase column was typically 15 mg/mL (S8). Negative gradients starting with 2 M ammonium sulfate have been effective with both of these materials in protein separations. Acylated polyamine (S9, SIO) and poly(alky1aspartamide) (S11)silica matrices have been equally effective in HIC separations when operated under these conditions. Retention studies on a series of six ligands at equal density showed elution increasing in the order of hydroxypropyl < methyl < propyl < butyl < pentyl < benzyl (SIO). The very slow rate of diffusion of biopolymers into and out of porous media has been a problem in HPLC. Nonporous monodisperse 1.5-pm silica media have been introduce as a potential solution for this problem. High-resolution separations on both the ether and amide functionalized 1 . 5 - ~ m sorbents were achieved in 3 min or less with high recovery of biological activity (S12). Selectivity of these nonporous materials appears to be identical with that obtained with porous media. Retention Mechanisms

Retention of macromolecules in HIC is a function of the nature of the stationary phase (SI-S12), the composition of the mobile phase, and the three-dimensional structure of the macromolecule. It is known that proteins in solution interact preferentially with high concentrations of salts with a net increase in free energy. Sufficiently high concentrations of salt can even cause precipitation of proteins. This salting-out process is driven thermodynamically by the intermolecular interaction of hydrophobic groups on the surface a protein (SI3). Intermolecular association of hydrophobic groups minimizes the increase in free energy by decreasing the hydrophobic contact area of the protein with the polar solvent medium. When a weakly hydrophobic material, such as an HIC column packing material, of greater surface hydrophobicity than the protein is introduced into this system, proteins interact with the sorbent. As in the case of salting-out, adsorption at the sorbent surface minimizes the hydrophobic contact area of the protein and sorbent with the polar solvent medium and produces the minium increase in free energy (SI3). This also explains why proteins may be eluted from HIC media by descending salt gradients.

COLUMN LIQUID CHROMATOGRAPHY

This model implies that some portions of the surface of a protein may be more hydrophobic and likely to dominate the interaction with the surface of an HIC sorbent. This apparently is true. It has been shown with lysozyme that one side of the molecule dominates retention in HIC (5'14). Both the HIC sorbent and solutes are surrounded by a solvent cage of water, ions, and any organic additives that are used in the mobile phase (S15,S16). During adsorption, at least a portion of the structurally ordered solvent in the contact surface area between the sorbent and solute must be displaced. Although displacement of water and ions in the chromatographic contact region will be accompanied by changes in both the enthalpy and entropy of the system, entropic changes are expected to dominate (S17). Moreover, the entropy change will be greatest at elevated temperature. The conformational state of a macromolecule can also contribute to its chromatographic behavior. SEC, IEC, and HIC all detected urea-induced conformational changes in globular proteins such as bovine serum albumin, lysozyme, and trypsin (S18). On the basis of spectroscopic data it has been possible to correlate alterations in the three-dimensional structure of a series of proteins with changes in their chromatographic behavior (SI 7, 5'18). Occasionally multiple conformational states of a protein may coexist in a column and given multiple chromatographic peaks (SIB). Thermally induced conformational changes altered both chromatographic retention, peak width, and 2 number. These differences in the free energy of association also correlated well with changes in the 2 number. Since the temperature at which conformational alterations occurred with cytochrome c varied between columns, it has been concluded that the column plays a role in the structural alteration of proteins (S19). This presents a dilemma in the analysis of conformational structure. One never knows whether the conformer ratio recorded by the LC detector represents the true ratio of conformersin the sample or is the product of alterations induced by the column. Applications

A large portion of the separations in biochemistry and biotechnology has been directed toward protein purification. As a consequence, HIC has been used both to monitor protein purification and to purify proteins. Monitoring ammonium sulfate precipitation by HIC is an interesting example of an analytical application of the technique (S20). The large amount of salt in the sample does not disturb the HIC separation. Since selectivity is quite different in HIC than IEC, SEC, metal chelate, and bioaffinity chromatography, it provides a valuable analytical tool to monitor purification in these other modes. One of the problems of HIC in the preparative mode is the poor solubility of many proteins in the initial mobile phase (S21). Precipitation of large amounts of protein at the column inlet will obviously cause serious problems. One solution to this problem is to use columns of higher hydrophobicitywhich require lower initial salt concentrations. Less hydrophobic columns were shown to separate membrane proteins by a combination of salt and detergent gradients. Use of HIC in the purification scheme for phenylalanine ammonia lyase (S22),ribulose-1,5-bisphosphatecarboxylase (S23),glucose oxidase (S24), catalase (824), human placental glucocerebrosidase (S25),and malate dehydrogenase isoenzymes (S26) all provide examples of the utility of the technique in the preparative mode. HIC was found to be an effective method for the separation of both wheat germ and soybean lectins into isolectin fractions (S27). The use of Ca2+as a structural modifier has been effective in the purification of calmodulin. In the presence of Ca2+, a number of calcium-sensitive proteins can be induced to bind to phenyl-Sepharose. The fact that the hydro hobic chromatographic contact region between calmodu in and phenyl-Sepharose is much less sensitive to monovalent cations than the other adsorbed proteins can be exploited in calmodulin purification (S28). The use of structural-modifying agents to alter selectivity is likely to increase in the future. Even very hydrophobic proteins have been fractionated by HIC. An investigation of the surface hydrophobicity of gliadins by octyl-Sepharose, phenyl-Sepharose, and octadecylsilane derivatized silica showed that these proteins could be eluted from all of the columns. CY-, @, ?-Gliadins showed

1"

a preference for both aromatic and aliphatic column ligands whereas the w-gliadin showed a preference for aromatic ligands. HIC has also been used in the fractionation of membrane (S30, S31) and ribosomal (S32) proteins. Hydrophobicity of the stationary phase can also play a role in the utility of HIC columns (S33, S34). HIC with weakly hydrophobic columns permits the rapid separation of labile isoforms of estrogen receptor (ER) proteins. When a propyl stationary phase was used, organic solvent was required in the mobile phase for complete elution of the ER proteins. In contrast, ER proteins were completely eluted from a polyether column without the use of organic solvents in the mobile phase. ER, purified by HIC, retained steroid binding capacity and protein kinase activity after elution from columns. Molybdate-stabilizedER proteins from rabbit uterine cytosol have been se mated into two major fractions by HIC (S35). PI- and &a&energic receptors have also been fractionated by HIC (S36). Negative salt gradients on several types of HIC bonded phases fractionated bulk tRNA into a variety of amino acid accepting species (S37). Selectivity was noted to vary slightly between stationary phases. In a study on octyl-Sepharose, it was even possible to fractionate phenylalanyl tRNA into two isoreceptor species with a negative ammonium sulfate gradient ranging from 2 to 0.4 M at pH 5.0 (S38).

ION-EXCHANGE CHROMATOGRAPHY The history and use of ion-exchange chromatography (IEC) dates back more than a half century, during which time it has been used broadly in the separation of both small and large molecules. Since IEC can be carried out near physiological conditions, it has been an important technique in the purification of sensitive biological macromolecules, such as proteins, over the past three decades. With the advent of biotechnology, the development of high-performance media and systems for the analysis and purification of genetically engineered proteins by IEC has accelerated to the point that recent advances in IEC are dominated by life science applications. Ion-Exchange Medla

Developments in ion-exchangesorbents in the past several years have come in the areas of support composition, stationary phase composition, and the introduction of nonporous sorbents. At the present time silica (T1-T3), alumina (T4), agarose (5%)) polymethacrylate (T6, T7),and poly(styrenedivinylbenzene) (2'8) have all been used as supports in ionexchange chromatography (IEC). There are large differences in the ion-exchange properties of macromolecules, such as proteins, on sorbents based on these supports. With the use of so many different support matrices the question of whether the support plays a role in separations becomes very important. It has been demonstrated that retention of standard proteins on supports of very different chemical composition varies less than 5% between columns with the same anionexchanging stationary phase (7%). These differences are probably due more to differences in chemical composition of the stationary phase (2'9) and physical properties of the sorbent matrix than the chemical nature of the support. Chemical stability is another consideration. In the case of silica, solubility increases dramaticallyabove pH 8. It has been shown that the leaching of silica supports by alkaline mobile phases is dramatically reduced by zirconium cladding ( T 3 ) , organic surface coatings, a low concentration of silica in the mobile phase, or some combination of the above (2'10). Alumina-based sorbents provide still another solution to operation of IEC columns at elevated pH (2'4). Polymethacrylate matrices are reported to be stable to pH 12 (T6, 7'7) while agarose and poly(styrenediviny1benzene) matrices are stable to pH 14 (T8). Ligand density, surface ,topography,and the ratio of charged to hydrophobic groups in the stationary phase can vary greatly between commercial columns and play a role in determining selectivity (2'10-TI2).These variations in the chemical and physical properties of stationary phases are related to differences in the synthetic scheme used to create the stationary phase. Although the synthetic procedures used in the preparation of most IEC sorbents are proprietary, they appear to ANALYTICAL CHEMISTRY, VOL. 60, NO. 12, JUNE 15, 1988

411 R

COLUMN L I Q U I D CHROMATOGRAPHY

fall into three broad areas as discussed below. With sorbents such as silica and alumina, stationary phases are applied as thin films at the surface of the support. One technique for preparing silica-based IEC media is surface derivatization with ionic organosilanes. Through the use of a mixture of ionic and nonionic organosilanes, it has been possible to adjust ligand density (T2). Adsorption is a second technique for creating and bonding stationary phases to inorganic supports. In this procedure, small polymers are adsorbed to the surface of supports and subsequently crosslinked into a continuous film. The polymer layer is adsorbed to the surface at so many sites that it can not be removed under conditions that do not chemically alter the polymer. Adsorption techniques have been used to prepare silica (2'131, alumina (T4),and poly(styrene-divinylbenzene) (2'8) based ion-exchangemedia. Ideally, surface films of ion-exchanging groups should be thin and easily accessible so that they will respond quickly to changes in mobile phase ionic strength. A third route for preparing IEC media is to derivatize the whole sorbent matrix. Derivatization may be achieved either after the matrix has been formed, as in the case of the natural polysaccharide based media (T5),or during polymerization, as in the case of synthetic polymer based media. A key difference between surface-modified supports and these media is that ion-exchanginggroups are distributed throughout the support in the later case. It is doubtful1 that all ionic groups in the matrix can participate in electrostatic interactions with solutes because of steric restrictions, particularly with macromolecular solutes. The introduction of nonporous IEC media has been an exciting new advance in ion-exchange technology. Because stagnate mobile phase mass transfer is such a problem with macromolecules in porous media, eliminating sorbent porosity substantially increases the rate at which macromolecules may be separated (TI4). Separations of protein mixtures have been achieved in less than 60 s with nonporous media. The principal limitation of nonporous media is that they are of low surface area and therefore have very low loading capacity. Samples of 10-50 wg can overload these columns. Retention Mechanisms

It has generally been assumed that pressure plays no role in the separation of macromolecules. Although this is true in most cases, it has been noted that when different structural forms of a molecule vary in specific molar volume that the species of smaller volume will be favored at high pressure (T15). The role of three-dimensionalstructure in chromatographic retention has been analyzed by the stoichiometric displaceThrough the number of ions ment model of retention (T16). displaced when a substance adsorbs to a column, it was possible to mess the area of contact of macromolecular solutes with an ion-exchange column. Chromatographic contact reion was found to be closely related to chromatographic be[avior in oligonucleotides. Changes in the chromatographic contact area were found to change when a protein was denatured (2'17, T18). It has been demonstrated previously that carboxymethyldextran (CMD) displacers fractionated on an ion-exchange column can be used for the resolution and high capacity displacement chromatography of proteins on high-performance columns (2'19). Recent studies indicate that fractionation of the CMD is unnecessary. High resolution and high capacity are obtainable by using unfractionated CMD displacers. Up to 25 mg of a mixture of ovalbumin, a-lactoalbumin, and soybean trypsin inhibitor along with many impurities were separated on a 3.3-m IEC column using three unfractionated CMD displacers and a final CMD displacer (T20). The CMD displacers were removed from the protein with a hydrophobic interaction chromatography step. Ampholytes used in isoelectric focusing have also been used as displacers (7'21). Tandem column sets have been used in protein fractionation. By use of a cation- and an anion-exchange column in series, mixtures of acidic and basic proteins were separated in a single chromatographic run with an increasing salt gradient at pH 7.0 (2'22). The serial order of the columns affected the chromatographic results. The effect was attributed to alterations of the salt gradient profile upon traversing the first ion-exchange column. Single columns, packed with a binary 412R

ANALYTICAL CHEMISTRY, VOL. 60, NO. 12, JUNE 15, 1988

mixture of a cation and an anion-exchange sorbent gave similar chromatographic results. Applications

Analytical applications of IEC span the range from amino acids, organic acids, and simple carbohydrates to biopolymers. Some typical examples described in the past few years have been the separation of sugar anomers on a calcium-loaded strong cation-exchange column (T23),determination of xylitol in human serum and saliva by anion-exchange chromatography in 0.15 N base (T24),the separation of underivatized gangliosides on an alkylamine derivatized silica column (T25), quantitation of N-acetylneuraminic acid by cation-exchange chromatography (T26),determine of serum taurocyamine by cation-exchange chromatography (T27J,determination of organic acids (T28-T30), amino acid analysis (T31-T33), peptide mapping (T34), determination of protein-bound metals with directly coupled flame atomic absorption spectrometry (2'35), conformational analysis of proteins (T36), protein subunit heterogeneity (T37),and the study of protein reactivity with thiol reagents (2'38). Early preparative applications of IEC were generally on soft gel columns. The higher resolution, throughput, and loading capacity of high-performance preparative columns have stimulated their use, particularly in the life sciences. Sample loading capacities for proteins on 150 X 21.5 and 200 X 55 mm columns have been reported to be 40-200 and 240-1000 mg, respectively (2'39). Representative enzymes that have been purified are glucoamylase (T40),ribulose 1,5-bisphosphate carboxylase (T41), pepsin isoenzymes (T42), heavy chain isoenzymes of myosin subfragments (T43), pepsinogen A isoenzymes (T44),lactate dehydrogenase isoenzymes (T45), glutathione S-transferase isoenzymes (T46), cytochrome P-450 isoenzymes (T47), protein kinase C (2'48),cellulases (T49, 2'501, adenosine deaminase (7'511, lipase (2'521, and @-gala$tosidase (2'53). It should be noted that anion-exchange is much more widely used than cation-exchange chromatography in the purification of proteins. This is because so many proteins are anionic in the pH range from 4 to 8. The fractionation of antibodies is generally achieved with anion-exchange chromatography. For example, immunoglobulin G (IgG), immunoglobulin A (IgA), and immunoglobulin M (IgM) from human serum can be separated by a strong anion-exchange column (T54, T55). The technique has also been used in the purification of mouse monoclonal IgM produced by hybridoma cells in vitro (T56). Purification of monoclonal antibodies from ascites fluid is slightly more difficult. Although IgG from ascites fluid may be fractionated into at least three subclasses (T57),some of the antibodies are contaminated by transferrin and albumin from the host. Transferrin and albumin are easily eliminated from IgG by a strong cation-exchange fractionation step (2'58). It has been shown that initial fractionation of monoclonal antibody preparations on a quaternary amine Zetaprep cartridge followed by cation-exchangechromatography produced antibodies of greater than 99% purity (2'59). Fractionation of membrane proteins that function either as structural units, transport mediators, receptors, catalysts, or tumor markers (2'60) is of great interest. Through the use of detergents to solubilize integral membrane proteins from human red cells, it was possible to resolve glucose transport proteins into one major and two minor fractions by anionexchange chromatography (2'61). Cytochrome P-450 provides another example. The concentration of detergent, glycerol, buffer, andd salt in the mobile phase all influenced retention and band spreading (T62).A nonionic detergent at 0.4% was the most suitable on both anion and cation-exchange columns; 20% glycerol was more effective than 10%. The fact that band spreading was greatest at low concentration of detergent and glycerol indicates that these reagents overcome hydrophobic interactions with the column. Zwitterionicdetergents also affect mobile phase additives in IEC of membrane proteins as has been demonstrated with NADH-cytochrome b, reductase (T63). Other classes of proteins in which IEC has been effective are microsomal cytochrome P-450 (T64-2'66), proteins of the eye (2'67-T69), bacterially expressed fusion proteins (T70), ribosomal proteins (T71, T72), neurotoxins (T73), ribonucleoproteins (T74-T76), muscle proteins (T77),and milk

COLUMN L I Q U I D CHROMATOGRAPHY

proteins (T78, T79). None of these examples required anything special in terms of elution conditions. High-performance ion-exchange chromatography still remains the dominant technique for the determination of hemoglobins (T80)and hemoglobin variant (2'81). Nonenzymatic glycosylation of hemoglobin in vivo has been found to correlate well with glucose regulation in diabetics. The concentration of glycated hemoglobin gives the equivalent of a time average of serum glucose level. Cation-exchange chromatography has been particularly effective in monitoring glycated hemoglobin (2'82, 2'83). Since blood is the only currently available source of blood clotting factors, they must be obtained by the factionation of plasma. Anion-exchange chromatography is used in the isolation of factors VIII, IX,X, and platelet activating factor (T84-2'87). More than 100000 L of plasma a year are fractionated in the isolation of clotting factors. Purification and characterization of glycoproteins and Carbohydrates are areas of increasing interest in biochemistry. IEC is proving to be surprisingly useful for this purpose even though oligosaccharides are not very ionic. One of the reasons is that either when carbohydrates are attached to a polypeptide, they give the molecule a small amount of charge that complements existing charge in the polypeptide or when acid groups of the polypeptide are involved in coupling, they reduce the charge of the polypeptide (T88-T92). In the case of oligosaccharides, the presence of carboxyl or amino groups in the molecule also assists in their separation by IEC (T93). When the oligosaccharide is completely neutral, then separation must be achieved through the formation of some type of complex with the column or the induction of a charge in the molecule. The separation of carbohydrates on metalloaded cation-exchange columns is an example of the former (T23, T94, 2'95) while cation-exchange chromatography of carbohydrates in strong base is an example of the later (T24). Chromatographicseparation of large polynucleotides began 20 years ago on an ion-exchange material referred to as RPC-5. This sorbent was prepared by coating a hydrophobic plastic bead with trioctylmethylamine. Obviously, this material functioned in a mixed mode. It has been shown in new high-performance media that mixed-mode interactions are desirable in the separation of nucleic acids, particularly oligonucleotides (T96-T99). Mixed-mode interaction have been achieved both by synthesizingmixed phases and by preparing phases that have hydrophobic and ionic character in the same ligand. The advantage of the mixed phases is that selectivity can often be controlled by varying the ratio of hydrophobic to ionic groups on the surface of the sorbent. Macroporosity is also important in the separation of high molecular weight species. Media of 4000-Apore diameter have been shown to be effective in the separation of tRNA, rRNA and DNA from crude cell lysates in the presence of denaturant (7'100). Large polynucleotides are adsorbed on columns at so many sites that gradient elution is essential (TIOI). There is also some difference in the resolving power of various commercial columns, particularly with oligonucleotides (T102). In addition, appreciable base pair specificity was detected with fragments rich in deoxyadenosine and thymidine. One of the more exciting aspects of the new columns is that they could be used in the purification of supercoiled DNA (T103). The IEC method has been used successfully to purify plasmids up to 45 kilobases in length and to isolate rare circular replication intermediates from an overwhelming excess of contaminating linear species in retrovirus-infected cells.

SIZE EXCLUSION CHROMATOGRAPHY Size exclusion chromatography (SEC) continues to be the premier technique for determining the molecular weight distribution of polymers. Over the past two years, there has been a dramatic increase on the use of SEC to characterize biopolymers. Whereas in the past, most investigators, especially those in the life sciences, relied heavily on conventional SEC technology, we are now beginning to see a change over to high-performancesystems. Another development that has occurred is an increased use of multiple detectors, most notably low-anglelaser light scattering and viscosity detectors, to characterize the compositional heterogeneity of polymers. Finally, there has also been renewed interest in the use of reversed-phase and adsorption chromatography for charac-

terization of synthetic polymers. Books and Reviews

Gloeckner (A9)has published a book on LC polymers which include all aspects of polymer separation including SEC. Parvez et al. ( U l ) authored a book on SEC and ion-exchange chromatography of proteins and peptides. In 1987 Waters Division of Millipore held a GPC symposium and the proceedings have been published (A44). An ACS symposium proceedings on SEC has also been issued (A43). Mori (U2) reviewed developments in SEC during 1983-1985. Brief reviews of SEC and its applications were published by Heisz (U3, U4) and Kuo and Provder (U5). Fischer (U6) wrote a detailed review on the use of SEC in food analysis. A comprehensivereview of the analysis of compositional and structural heterogeneity of polymers was written by Gloeckner (U7). Balke (U8) reviewed the theory and application of orthogonal chromatography for the characterization of complex polymers. Balke (US)also reviewed nonlinear regression, graphics, and error propagation analysis as applied to SEC. Barth ( UlO) discussed numerous nonsize-exclusion effects that can occur in high-performance SEC. Theory

Gorbunov and Skvorbov (UII, U12) developed expressions for the distribution coefficients of macromoleculesand their separation efficiencies as a function of chromatographic conditions. Weak adsorption was suggested for improved separations of linear and cyclic polymers. These authors (U13) also described an equation for the separation of polymers that was independent of polymer structure, solvent, adsorbent, and temperature. Su and Mou (U14) developed a new calibration method and SEC separation model. Cheng (UI5) presented a concept that relates SEC with hydrodynamic chromatography in which the separation is based on both surface-exclusionand geometric-exclusioneffects. These mechanisms are a function of solute size, total accessible surface area, and surface geometry. Tejero et al. (U16) developed theoretical expressions for the SEC distribution coefficient and the preferential sorption coefficient. Lecourtier and Chauveteau (U17) proposed a model based on surface exclusion chromatography to explain the rate of diffusion of xanthan through porous media as a function of solute size and packing porosity. Column Resolution

Snyder and co-workers (U18, U19) developed a general model to predict plate numbers of large biopolymers during SEC. Sablonniere et al. (U20) compared the resolution and efficiency of TSK-SW columns to those of open columns for the analysis of glucocorticoid receptor. Engelhardt and Schoen (U21) discussed optimization strategies for SEC of proteins on modified silica columns. Yamamoto et al. (U22)measured plate heights of proteins as a function of mobile phase velocity, particle diameter, gel type, and temperature. Grinshpun and Rudin (U23) presented a method for evaluating SEC column performance. Chen (U24) used the uDz parameter for evaluating column performance. Synovec and Yeung (U25) found that ultrasonication of SEC columns results in an increase in band broadening. The magnitude of this effect was correlated with the diffusion coefficients of a series of standard compounds. Band Broadening

Costa et al. (U26)described in detail the mathematical basis of SEC peak broadening by using the generalized form of Tung's integral equation. Also treated are the fundamental aspects of modeling SEC according to linear chromatography theory and the experimental measurement of the dispersion function parameters. In a subsequent paper (U27),these authors proposed a method for band spreading correction. Alba and Meira (U28)presented a method for the calculation of band broadening that uses a recycle technique. These authors also reported on a stochastic matrix approach based on Wiener filtering theory (U29). ANALYTICAL CHEMISTRY, VOL. 60, NO. 12, JUNE 15, 1988

413R

COLUMN LIQUID CHROMATOGRAPHY

Cheng and co-workers (U30, U31) developed an equation that relates column dispersion to elution volume and also studied factors influencing band broadening. Eggink et al. (U32)described a general expression for band broadening with a LALLS detector. Lederer et al. (U33)presented a modified method for simultaneously calibrating and correcting for axial dispersion using an online LALLS photometer. Hua and co-workers (U34) presented several approaches for establishing a poly(viny1butyral) calibration curve including a method for correcting band broadening. Cheng et al. (U35) described a procedure for simultaneous calibration and correcting for band broadening. Schroder and Ebert (U36)developed a personal computer program for correcting axial dispersion. These authors also evaluated a modified broad molecular weight standard Calibration method. Chen et al. (U37)presented a peak spreading correction by means of Wesslau and Tung’s molecular weight distribution functions. Callbration

Universal. Potschka (U38) evaluated a series of well-defined monodisperse calibrants of various shapes, e.g., proteins, DNA, and tobacco mosiac virus, in terms of universal Calibration based on the calibrant’s viscosity and Stokes radii, molecular weight, mean linear projected length, contour length, etc. They found that universai calibration was valid if based on viscosity radius of gyration. Squire (U39) discussed the use of hydrodynamic volumes of proteins and random-coil polymers to construct a universal calibration curve. Fishman and co-workers (U40) investigated the use of the radius of gyration of dextran and pullulan to establish a universal calibration for pectins. Hennick and co-workers (U41)reported that the universal calibration based on dextrans and poly(ethy1ene glycols) cannot be used for SEC of heparins. These authors found that an online LALLS detector gave reliable results for heparins. Guven (U42) examined the universal calibration of poly(ethylene glycol) using the Flory-Fox and the Ptitsyn-Eisner viscosity equations. Sestrienkova et al. (U43) also reported difficulties when using poly(ethy1ene glycol) to construct a universal calibration curve. Biagini et al. (U44)showed that a polystyrene universal calibration curve was not applicable to N-trifluoroacetylated poly(t-caprolactam) which gave very peculiar elution behavior. Senak et al. (U45)found that the universal calibration held for poly(vinylpyrro1idone)when chromatographed with a 50% aqueous methanol solution containing 0.1 M LiN03 and using a TSK-PW column. Results were compared with online LALLS data and osmometric results. Boimirzaev et al. (U46) achieved a universal calibration curve for poly(viny1pyrrolidone) and polystyrene in dimethylformamide. Styring et al. (U47) used a series of well-characterized latex particles to study the validity of universal calibration as a function of mobile phase ionic strength and surfactant type. Deckers and co-workers (U48) evaluated the universal Calibration approach by using pullulan standards and an iterative procedure based on polydisperse samples of known intrinsic viscosities in combination with monodisperse standards to characterize pectins. Wang et al. (U49)used a trial-and-error procedure for obtaining Mark-Houwink parameters from a universal Calibration curve. This approach was applied to polystyrene, ethylene-propene copolymer, polypropylene, and poly(viny1 chloride) at 90 and 135 OC in o-dichlorobenzene. Zhang (U50, U5I) reported on a procedure for determining the molecular weight of polymers of unknown Mark-Houwink constants. Deniz and Gueven (U52) reported on the errors introduced when calculating average molecular weights when using incorrect Mark-Houwink constants with the universal calibration curve. Ito et al. (Lr53) took into account the dependency of the Mark Houwink constants on molecular weight when usin the universal calibration approach. Brauer et al. (U547 evaluated a universal calibration curve for norborneneethylene copolymers by comparing the intrinsic viscosity data calculated by using SEC data with those from direct measurements. Broad-Molecular-Weight Standards. Goetz et al. (U55) evaluated the use of broad-molecular-weight distribution 414R

ANALYTICAL CHEMISTRY, VOL. 60, NO. 12, JUNE 15, 1988

standards of polycarbonates and nitrocellulose. Xu and coreported on a broad-molecular-weight standard workers (U56) approach in which only the number-average molecular weight is known. Chen (U57) described a broad-molecular-weight standard method in which any type of average molecular weight can be used. Szewczyk (U58, U59) presented an approach for determining both a calibration curve and Gaussian spreading function parameter using two broad-molecularweight standards. Standards. LeMaire et al. (U60) presented a list of water-soluble globular proteins and detergent-soluble membrane proteins of known Stokes radii to be used as SEC standards. Kobata et al. (U61)published a table of effective size of oligosaccharides, expressed in glucose units. Ohta and co-workers (U62) used fractionated blue dextran to calibrate columns. Baltisberger and Jones (U63, U64)prepared and evaluated oligomeric aromatic ethers of diary1 and furan types as SEC calibrants for asphaltene and preasphaltene analyses. Swanson et al. (U65)compared polystyrene and polyisoprene SEC standards for u8e in rubber analysis. Munir and Goethals (U66) evaluated the use of -poly(N-tert-butylaziridine) as an SEC .internal standard. Ollivon et al. (U67) calibrated TSK-PW columns by using vesicles of known Stokes radii for subseauent analvsis of liposomes and viruses. Ohno et al. (U68) derived SEC calibration relationship for cartilage proteoglycan subunit by using the hydrodynamic radius and weight-average molecular weight. Helm et al. (U69) performed SEC of tricarbanilate derivates of starches and calibrated the chromatographic system by using carbanilate-derivatized pullulan standards. LALLS Calibration. (Also see LALLS and viscometers in the Miscellaneous Detectors section.) Prochazka and Kratochvil (U70) presented a detailed analysis of the effect of errors associated with concentration and LALLS detectors on the accuracy of the molecular weight distribution. He et al. (U71) evaluated a LALLS detector by using narrow-molecular weight distribution polystyrene standards. Bailly et al. (U72) evaluated calibration methods including the use of an online LALLS detector for the molecular weight determination of bisphenol-A polycarbonate. Froment and Revillon (U73) demonstrated that the number-average molecular weight and molecular weight distribution of a polymer using online LALLs is strongly dependent on axial dispersion for a Schulz-Flory distribution. Miscellaneous. Jakes and Saudek (U74) presented a detailed study on the evaluation of empirical molecular weight distribution functions as determined from SEC. Sorokin (U75) developed an absolute method for determining molecular weight averages based on stoichiometric relationships and using labeled polymer molecules. El’tekov (U76)proposed a generalized calibration procedure and studied its dependence on mobile phase composition, surface chemistry, pore distribution, and temperature. Adler et al. (U77) related the Wiener number, a structural description for polycyclic aromatic hydrocarbons (PAH), to the elution volume of a series of PAHs. Kolegove et al. (U78) established a procedure for determining the molecular weight of poly(methy1 methacrylate) based on a polystyrene Calibration. Dai (U79)derived equations for calculating average molecular weights using a Wesslau distribution and Lansing-Kraemer distribution functions. Nonsire-Excluslon Effects

Shear Degradation. We and Cai (U80)showed severe degradation of high-molecular-weight poly(2-vinylpyridine) (>2.8 X lo6) through an SEC column. Concentration Effects. Chiantore and Guarta (U81) presented an approach of evaluating the relative contributions on viscosity and hydrodynamic volume contraction effects on SEC distribution coefficients. Meyer and Reichert (U82) reported on the concentration dependency of polystyrene and its effect on the universal calibration curve. Zhang et al. (U83) studied the effects of polystyrene concentration and molecular weight on elution volume. Tejero and co-workers (U84-USS) studied the concentration effects in theta binary and ternary polymeric systems and derived a model for quantitatively describing shifts in elution volume. Mingshi and Guixian (U87) proposed a model to

COLUMN LIQUID CHROMATOGRAPHY

predict the effect of polymer concentration on the hydrodynamic volume and peak elution volume. Adsorption and Mobile Phase Selection. Reviews describing nonsize-exclusion effects in SEC of water-soluble UM), Bruessau polymers have been presented by Barth (UIO, (U89),and Cooper (U90). Dubin (U91) reported on electrostatic effects in aqueous SEC. Lundy and Hester (U92) analyzed a number of different water-solublepolymers by using Sephacryl and an online viscometric detector. Gharfeh et al. (U93)presented a method for SEC of high-molecular-weight anionic polymers. Tarvers and Church (U94) evaluated the effect of ionic strength on the elution behavior of proteins by using TSK-SW columns. Ionic interactions of proteins with TSK-SW columns have also been studied by Denner et al. (U95) and Irvine (U96). Meyerson and Abraham (U97) evaluated the use of dihydroxyalkyl-bonded silica for the SEC peptides and proteins. Koehler (U98)evaluated poly(vinylpyrro1idone)-coated silica for protein separation. Mant et al. (U99) described a series of synthetic protein standards used to determine column resolution and for monitoring nonideal elution behavior. Washabaugh and Collins (U100) studied the SEC behavior of a number of solutes, which affect protein stability, using Sephadex columns. The solutes included ethylene glycol, urea, TRIS, guanidine, and anionic salts. Mobile phases containing trifluoroacetic acid (TFA) (U101-U105), TFA and methanol (U106), and TFA and acetonitrile (U107) have been used for SEC of proteins. Bindels et al. (U108)used different urea concentrations with a LALLS detector to study the dissociation of a-crystallin. Aoki et al. (U109) employed 6 M urea to chromatographcasein micelles and 6 M guanidineOHC1was used as a mobile phase for SEC of proteins on agarose (U110, U111) and TSK-SW columns (U112). Kenley (U113) studied the effect of ionic strength in acetonitrile on SEC elution behavior of a decapeptide. Konishi (U114) reviewed the SEC behavior of proteins in SDS, urea, and guanidineHC1. Welling and co-workers (U115) compared several column packings and evaluated the use of nonionic surfactants in the mobile phase for SEC of membrane proteins. McNeill and Donnelly (U116)evaluated the use of controlled-pore glass for SEC of casein micelles. A 0.005 M NaOH mobile phase was used with Sephacryl for the SEC of starch (U117). Fuchs ( U l l 8 , 119) reported on the interactions between organics found in natural waters and TSK packings. The SEC elution behavior of humic acids has been studied by Mori et Lehto et ai. al. (U120), Yonebayashi and Hattori (U121), (U122),and Katayama et al. (U123). Karpukhin et al. (U124) used SEC to study iron compounds in soil and their interactions with humic and fulvic acids. Wonnacott and Patton (U125, U126) used a weak cationic-bonded stationary phase, aminopropyl group, for SEC analysis of cationic polymers. Soponkanaporn and Gehr (U127) used polyamine-coated silica for SEC of high-molecular-weight cationic, anionic, and nonionic acrylamide-based polyelectrolytes. Van der Klashorst (U128)evaluated various mobile phase compositions for the analysis of lignin and found that 90% aqueous methanol containing 2% acetic acid prevented adsorption on KPorasil. Guy et al. (U129) developed an SEC system for the analysis of free and bound metals. These authors had investigated a number of different packings. Dimethylformamide-containing LiBr and H3P04was used for the SEC analysis of poly(amino acids) (U130),phenolcontaining fraction of Kraft liquor (U131),and acrylonitrile copolymers (U132). Cotts (U133)used N-methylpyrrolidone with LiBr as an eluent for poly(amino acids). Dimethylacetamide with LiN03 was reported by Hlaing et al. (U134) and Ekmanis (U135) for the analysis of urea-formaldehyde resins and cellulose, respectively. Bobleter and Schwald (U136, U137) and Cosgrove et al. (U138)reported on the use of cadoxen as the mobile phase for SEC of cellulose. The effect of mobile phase composition on the analysis of coal products has been reported by Bartle (U139),Evans (Ul40),and Haenel (U141). Barrales-Rienda et al. (U142) studied the nonsize-exclusion effects of poly(N-vinyl-3,6-dibromocarbazole)in THF using cross-linked polystyrene as the packing. The equilibrium

distribution of the polymer between the mobile phase and packing was treated thermodynamically as .a problem in preferential sorption in a ternary system. Troeltzsch (U143) investigated the elution behavior of alcohols on BioBeads and Sephadex LH packings. Craven et al. (U144) studied the partitioning effects of a series of oligo(oxyethy1enes) during SEC. Chiantore and Guaita (U145)reported on the effect of mobile phase on retention of oligomers and small molecules. Mabrouk et al. (Ul46) determined the effect of various solvent-temperature combinations on the elution behavior of organic compounds including alkanes and aromatics. Kohn (U147)evaluated different mobile phase composition for the analysis of poly(dimethylsi1oxanes). For the SEC of used a mobile phase consisting Nylon 12, Ogawa et al. (U148) of hexafluoro-2-propanol and toluene. Poly(ary1ether ether ketone) was analyzed by using a 1/1phenol-trichlorobenzene mobile phase (U149, U150). Yoon et al. (U151) used a tetrachloroethane/o-dichlorobenzenemobile phase for SEC of poly(ethy1eneterephthalate). For the analysis of poly(phenylene sulfide), Kinugawa (U152) used a cross-linked polystyrene packing at 220 "C with chloronaphthalene as the solvent. Parks et al. (U153) preconditioned a cross-linked polystyrene packing with an organometallic cation to establish a positive charged surface for the SEC analysis of tin-bearing organometallic polymers. Gharfeh (U154) used a toluenepiperidine step-gradient and Ultragel as the packing for determining a hindered-amine additive in polyethylene. This method was described as a size-exclusion nonaqueous reversed-phase technique. Accuracy/Precislon

Miller (U155) discussed the source and significance of errors affecting the reproducibility of SEC analysis of polymers. Analysis of variance was used to determine and compare variations from different sources. Computer Processing

McCrary (U156) presented a chemometric procedure for the statistical comparison of SEC chromatograms. Long et al. (U157) described an SEC data reduction software package for calculations using a personal computer. Silica Packlngs

Unger (U158) evaluated diol-bonded silica packings for SEC of proteins. Anselme et al. (U159) and Wang et al. (U160) reported on the synthesis and evaluation of glycidoxypropyltrimethoxysilane-bondedsilica for proteins and other water-soluble polymers. Murakami and Mori (U161) obtained a patent for a glycidoxypropyltrimethoxysilane-bondedporous glass packing for SEC. Kat0 et al. (U162) evaluated a new SEC packing, TSKSWxL, for protein separation. Mori described a procedure for slurry packing diol-bonded silica (U163)and porous glass (U164)packings. These packings were also evaluated in terms of efficiency and applications to water-soluble polymers and polystyrene, respectively. Wickramanayake and Aue (U165) described a poly(oxyethylene) stationary phase for use in SEC as well as for GC. Patents were awarded to Karger et al. (U166) for a etherbonded phase, to Regnier and Gupta (U167) for a polyamine-based bonded phase, and to Okamoto and Hatada (U168)for an optically active polymeric stationery phase for use as an SEC packing. Stout and DeStefano (U169) developed a zirconia-clad silica packing (Zorbax GF) for protein separations that can be used with high pH mobile phases. Patents for preparing chemically stable porous glass containing zirconia and other oxides were issued to Eguchi et al. (U170) and Gonzalez et al. (U171). Dawidowicz (U172)produced glass beads of controlled porosity by caustic leaching. This investigator (U173) also studied the influence of boron enrichment on the surface of controlled-pore glass on the elution behavior of polymers. Organic-Based Packings

Hatano (UI 74) reviewed SEC, ion-exchange, and reversed-phase packings made in Japan. Characteristics of these ANALYTICAL CHEMISTRY, VOL. 60, NO. 12, JUNE 15, 1988

415R

COLUMN L I Q U I D CHROMATOGRAPHY

materials as well as applications are given. SEC packings prepared from cellulose acetate (U175), amylopectin (U176),agar (U177),dextran (U178),and pullulan (UI 79) have been reported. Agarose-based packings, Superose, have been evaluated by Johansson and Aahsberg (U180)and Kramlova et al. (U181). A polypeptide packing, prepared from poly(y-methyl-L-glutamate) or poly(ybenzyl-L-glutamate) was synthesized and evaluated by Ihara et al. (U182, U183) and with both aqueous and organic mobile phases. Yang and Verzele (U184) evaluated a hydrophilic functionalized cross-linked polystyrene-divinylbenzene packing (Rogel-P) for protein separations. Chaumont et al. (U185) were awarded a patent for a ethylene oxide grafted polystyrene-divinylbenzene copolymer for aqueous SEC applications. Asahipak, a vinyl alcohol based copolymer, was evaluated for the separation of oligonucleotides (U186) and peptides (U187)and in organic mobile phases (U188). Dawkins et al. (U189)synthesized and evaluated cross-linked polyacrylamide packings for water-soluble polymers. Mixed-bed TSK-PW columns (TSK-GMPW, an acrylate polymer) were applied to water-soluble polymers (U190). Wu et al. (U191) evaluated Sinopak-EGPM, an acrylate copolymer, for the separation of proteins. Patents were awarded to Mitsubishi Chemical Industries (U192) and Hradil et al. (U193) for the preparation of acrylate-based packings for aqueous SEC. Divinylbenzene cross-linked polymers have been prepared by Bian et al. (U194)and Tokunaga and Hashimoto (U195). Sugitani et al. (U196)was granted a patent for a carbon black filled cross-linked styrene-divinylbenzene packing. Selected Applications

Compositional Heterogeneity. The use of multidetectors in SEC was reviewed by Bruessau (U197). Garcia-Rubio (U198) discussed the effect of detector sensitivity and measurement error on data interpretation from multiple detectors in SEC. SEC and online LALLS detection have been used to characterize dextran (U199), polyquinolines (U200), eth lene-propylene copolymers (U201),and block copolymers anBblends of polystyrene and dimethylsiloxanes (U202). UV detection in combination with a refractometer has been used to determine the chemical heterogeneity of isoprenestyrene block copolymers (U203),oligodiene urethanes (U204), lignin-acrylamide graft copolymers (U205-U207), and acrylic polymers (U208). Aiba (U209) determined the degree of N-acetylation of chitosan by using SEC and UV detection of acetamide groups. Trugo and Macrae (U210) characterized roasted coffee with a refractometer-UV detector. Okamoto and Yoshinaga (U211) determined the hydroxyl equivalent and functionality of polyethers by using a refractometer-UV detector. An online & detector was used to determine silicon-phenyl groups in polydimethylsiloxanes (U212)and phenolic hydroxyl groups in phenol-formaldehyde resins (U213). Mori (U214) determined the end groups of poly(ethy1ene sebacate) as a function of molecular weight by derivatizing the hydroxyl and carboxylic end groups with 3,5-dinitrobenzoyl chloride and 0-(pp-nitrohenzy1)-N3’-diisopropylisourea, respectively, and using an IR detector. Gloeckner (U215)presented a review of high-performance precipitation LC for the analysis of copolymers. This approach was used to separate copolymers of styrene and ethyl methacrylate copolymers (U216) and acrylonitrile-styrene and methyl methacrylate-styrene copolymers (U217). Mori et al. (U218) separated styrene-methyl methacrylate copolymers by using liquid adsorption chromatography followed by SEC. Lai et al. (U219)used a chemically bonded phase to separate phenyl-containing vulcanizable silicone and polystyrene. Gorshkov et al. (U220)used liquid-adsorption chromatography to characterize the molecular weight distribution of polybutadiene. Inverse SEC. Jerabek (U221)discussed the feasibility of determining the porosity of polymer particles by using inverse SEC. Jerabek et al. (U222) also used inversed SEC to determine the morphology of glycidyl methacrylate copolymers. Knox and Ritchie (U223) extended and computerized a previously developed equation relating an SEC calibration curve 418R

ANALYTICAL CHEMISTRY, VOL. 60, NO. 12, JUNE 15, 1988

to the pore-size distribution of rigid, porous materials. Bertoniere et al. (U224) determined the pore structure of liquid ammonia treated cottons by using inverse SEC. Nesterov and Ignatova (U225) measured polymer-solvent interaction parameters of cast films of poly(methy1 methacrylate) by using inverse SEC. LeMaire et al. (U226) demonstrated that the pore-size distribution of a packing can be calculated from a protein calibration curve. Nikolov (U227)presented a strict definition of pore-size distribution and related it to pore-volume distribution based on SEC measurements. Microcolumn SEC. Mori (U228) proposed a procedure for packing 1.5 mm i.d. columns with cross-linked polystyrene by using the balanced-density slurry technique. The evaluation of these columns was reported in a subsequent paper (U229). Hibi et al. (U230) packed and evaluated 2.1 mm i.d. columns for SEC. To increase peak capacity, Takeuchi et al. (U231) evaluated a 2-m long 0.35 mm i.d. microcolumn for protein separation. Physiochemical. Studies involving protein-protein and protein-ligand binding interactions using SEC have increased in the past two years. Kasai (U232) reviewed the use of SEC to determine the dissociation constants of ligand-protein complexes. Cann and Winzor (U233) presented a theoretical and experimental study of frontal SEC to measure complex formation between proteins and other macromolecules. Stevens (U234) described an iterative computer simulation to describe protein-protein interactions. The Hummel-Dreyer method was used to investigate the binding of hemoglobin to polyanionic polymers (U235),benzyl-thiouracil to human serum albumin (U236),and calcium to EGTA (U237). Andreu (U238) discussed the principles, sources of errors, and criteria of application of the Hummel-Dreyer method. Association constants for NAD binding to lactate dihydrogenase was determined by Motorin (U239). Binding of AMP by fructose 1,6-diphosphatase was measured by Kid0 et al. (U240). Mazzini and co-workers (U241) studied the effect of guanidineSHC1 on the self-association behavior of bovine liver glutamate dehydrogenase. Mutter and Altmann (U242) investigated the aggregation behavior of homooligopeptides by using SEC. SEC was used by Tazuma and Holzbach (U243)to study the association between conjugated bilirubin and various biliary lipid particles. Braco et al. (U244) used SEC to study the molecular association between phosphatidylcholine and gramicidin A in THF. Porte and Harricane (U245)reported on the complex formation of plasma gelosin with actin. Sun and Wong (U246) used SEC to determine L-tryptophan-serum albumin binding constants. Zimina et al. (U247) studied protein association in surfactants by SEC. They also reported a procedure for determining the critical concentration of micelle formation in surfactants. Asakawa et al. (U248) used a partition chromatography model to simulate the SEC of a mixed surfactant system in which two kinds of mixed micelles are assumed to coexist. Booth and co-workers (U249, U250) used aqueous SEC to determine the hydrodynamic volumes of micelles of poly(oxyethylene)methyl n-alkyl ethers. Spacek (U251)used SEC to determine the average time for the formation of micelles in Kraton G 1652. Shalongo et al. (U252) used SEC to study the conformational changes of denatured thioredoxin with different concentrations of guanidineSHC1. Adachi (U253)measured the surface hydrophobicity of hemoglobins by using SEC. Garcia et al. (U254) chromatographed glycogen and its hydrolyzed products as a function of mobile phase pH, ionic strength, and temperature in order to calculate thermodynamic parameters of solute-packing (Sephadex) interactions. Cacace et al. (U255)measured the temperature dependence of adsorption of tryptophan, AMP, and (e-dinitropheny1)lysineon Sephadex and BioGel. Van Iersel et al. (U2.56)was able to determine accurate protein extinction coefficients by using SEC coupled with a photodiode array detector. Frank et al. (U257) established the purity of proteins by using SEC and a photodiode array detector. Ito and Ukai (U258) examined various equations for determining Mark-Houwink constants from the universal cal-

COLUMN L I Q U I D CHROMATOGRAPHY

ibration procedure. Samay ( U . 9 )presented an approach for estimating Mark-Houwink constants from SEC data. Mark-Houwink constants have been measured for pol urethane block copolymers ( U260), poly(N-vinyl-3,6- ibromocarbazole) (U261),and cellulose sulfate (U262) by using SEC data. Kubin (U263) has incorporated axial peak spreading corrections in the determination of Mark-Houwink constants. Kever et al. (U264) presented a method for determining Mark-Houwink constants that uses SEC microcolumns. Lloyd and co-workers (U265) determined the phase behavior and interaction parameters of polystyrene and polyisoprene in toluene by using an SEC technique to determine the composition of the phases. Das (U266) studied dilute solution properties of homo- and block copolymers by a technique termed "variable temperature" SEC. Tian (U267) used SEC coupled to a capillary viscometer to determine the unperturbed dimensions of polystyrene and poly(methy1 methacrylate). SEC was used by Tsitsilianis and Dondos (U268) to study temperature-induced polymer transitions of polystyrene and methyl methacrylate-styrene block copolymers. Mencer and Rek (U269) reported on polymersolvent interaction by use of SEC. Zhmakina et al. (U270) determined the effective diffusion coefficient of polystyrene in the pore volume of macroporous glass from peak broadening measurements. Periyasamy and Ford (U271) measured rate constants of exchange for solvents diffusing into and out of cross-linked polystyrene packings. Polymer Branching. Hamielec and Meyer (U272) reviewed the use of online LALLS photometry for determining long-chain branching as well as peak broadening,unperturbed polymer chain dimensions, microgel content, and shear degradation. Shiga and Kat0 (U273) derived a general relationship between molecular weight and number of branched points per molecule from SEC fractions of branched polymer taking into account axial dispersion. Halbwachs and Grubisic-Gallot (U274) discussed data processing using online LALLS and viscometry detectors. Online LALLS was used to determine branching of poly(methyl methacrylate) and star microgels (U275), polyethylenes (U276, U277), branched polystyrene (U278),ringshaped polystyrene (U279), polyoctenamer (U280),cis-1,C polybutadiene (U281),and polysaccharides (U282). Chu and *workers (U283)determined branching in poly(viny1acetate) by using dynamic light scattering and compared these results to data obtained from SEC viscosity and LALLS measurements. Long-chain branching in poly(butadienes) was measured by Jin et al. (U284, U285) using an online viscosity detector. Offline viscometry of SEC fractions was used by Kuge et al. (U.286)to determine branching in dextrans. Mrkviokova and Janca (U287) used an online viscosity detector to characterize polymer branching in terms of the weight-average number of branching sites and the branching index. Coleman and Dawkins (U288)determined long-chain branching of poly(vinyl acetate) by using SEC combined with offline measurement of viscosity. Preparative SEC. Nishimoto et al. (U289) presented methods for scaling UD analvtical SEC to the DreDarative mode. Examples Gin; proteins were given. Regnier et al. (U290) discussed the cost and performance characteristics of preparative SEC of biopolymers. Janson (U291) reviewed preparative chromatography of biopolymers. Sasaki et al. (U292)reviewed the use of TSK packings for industrial-scale purification of biopolymers. Kelley et al. (U293) presented a systematic approach for the design and optimization of preparative SEC columns based on theoretical and empirical equations. Yamamoto et al. (U294) discussed the use of medium-performance SEC to scale-up proteins separations and to reduce analysis time. Selected applications of preparative SEC include RNA (U295),plasmid DNA (U296),plasma fractionation (U297), oligosaccharides (U298),polystyrene and epoxy resin (U299), and acrylic resins (U300). Troeltzsch (U301) applied preparative SEC using a recycling technique for the isolation of acylic oligomers of isoprene. Miscellaneous. Gunderson and Giddings (U302) and Gao et al. (U303) compared the resolving power of thermal fieldflow fractionation to that of SEC. Adachi et al. (U304) presented a method using SEC to estimate the rejection

d

coefficient-molecularweight relationship of an ultrafiltration membrane. Nesterov et al. (U305) discussed the use of SEC to determine the molecular weight distribution of block copolymers. Philip (U306) used a multiloop sample valve to collect SEC fractions of coal liquids and injected them into a GC/MS system.

POSTCOLUMN DER IVATIZATI ON New Designs

Postcolumn derivatization in the traditional sense involves the addition(s) of a reactant(s) to the column eluent into a tee by means of a pump(s); a reaction coil(s) placed between the tee and the detector allows the chemical reaction(s) to occur. Innovations in design include hollow-fiber reactors, solid-phase reactors, and electrochemically generated reactants. General reviews of postcolumn derivatization have been written by Cassidy and Karcher ( V I ) ,Frei et al. (V2),Lillig and Engelhardt (V3),and Schlabach and Weinberger (V4). Krull and co-workers reviewed postcolumn photochemical derivatization (V5)and solid-phase reactors (V6). Postcolumn derivatization techniques for electrochemical detectors have been covered by Krull et al. (V7)and Leroy and Nicolas (V8). Hollow fibers are membrane-devices which allow the permeation of certain species and preclude the permeation of others. Haginaka et al. (V9)used a sulfonated polyethylene hollow fiber suspended in a 2 M NaOH solution to measure @-lactamaseinhibitors in serum and urine. The hydroxide ion causes degradation of the @-lactamaseinhibitors, clavulanic acid and sulbactam, and the degradation products are measured in the UV at 272 and 278 nm, respectively. The same authors (VlO) extended the technology to the analysis of barbiturates by postcolumn pH modification. After separation, the pH of the eluent was raised to 10 by introducing ammonia or ammonium ion and the barbiturates were detected in their ionized form at 240 nm. The performance of annular membranes and screen-tee reactors for the postcolumn reaction detection of lanthanide metal ions was determined by Cassidy and co-workers (VI I). Base-line noise due to mixing homogeneity and pump pulsations as well as effects on the reactor design on column efficiency was studied for three different internal volume membrane reactors. Although some leakage of eluents into the reagent solution occurred at high eluent or reagent flow rate, at normal operating conditions the reactors performed satisfactory. A sulfonated hollow-fiber membrane was used to derivatize amino acids using orthophthaldehyde (OPA) + mercaptoethanol. The fiber was immersed in a solution of the reactants which diffused through the membrane to the flow stream (V12). Solid-phase reactors provide a simple way to achieve postcolumn response. The reactor is placed between the column exit and the detector. Chemical reaction occurs only when an active substance leaves the column. In addition, no pump is required. Finely divided copper was used by Irth et al. (V13) to detect thiram and disulfiram based on a complexation reaction. The complex, Cu(I1) Nfl-dimethyldithiocarbamate has an absorption maximum at 435 nm and sub-ppb determinations were achieved. The oxidation of catecholamines can be accomplished online in a lead oxide or manganese oxide solid-phase reactor which was integrated into the trihydroxyindole (THI) method (V14). These solid-state reactors are simplier than the more common ferricyanide oxidation procedure and the operating cost is lower. A grafted disulfide reactor was used to detect thiol compounds at the 0.05-nmol level (V15). In a slightly different configuration (V16),a solid-state reagent bed consisting of bis(2,4,6-trichlorophenyl)oxalate (TCPO) was situated parallel to the analytical column. This compound is popularly used in the peroxyoxalate chemiluminescence detection of fluorescers and its use requires mobile phase compositionsgreater than 80% acetonitrile in water. If the HPLC conditions were such that this condition was met, a split-flow arrangement was used with no additional reagent pump. If the mobile phase composition is 50% acetonitrile or less, a single pump is needed to bring about the appropriate reaction conditions. A similar solid-state chemiluminescence detection was used by Poulsen et al. (V17, V18) for the detection of quinones. A photochemical reaction generated hydrogen peroxide from ANALYTICAL CHEMISTRY, VOL. 60, NO. 12, JUNE 15, 1988

417R

COLUMN LIQUID CHROMATOGRAPHY

the quinone. The peroxide was detected by using peroxyoxolate energy transfer chemiluminescence by reaction with a fluorphor, 3-aminofluoranthene,bonded to glass oxide beads or silica gel and with an oxalate ester (TCPO) in the solid state. The studies were conducted with both solid-state reactors and pumped postcolumn reagents. The miniaturization of solid-phase reactors for online POstcolumn derivatization of carbamate pesticides was accomplished by using a strong anion exchanger to hydrolyze the carbamate yielding methylamine which could be detected by the OPA fluorescence reaction (V19). Immobilized urease on silica gel was used in a similar manner to determine urea and ammonia. The reador was designed to work with 1.0 mm i.d. analytical columns. Immobilized enzymes have been widely used as solid-phase postcolumn reactors. The subject was reviewed by Potezny (VZO),Dalgaard (VZI),and Bowers ( V . 2 ) . Acetylcholine and choline were determined by interaction with acetylcholinesterase and choline oxidase coimmobilizedby reaction with glutaraldehyde onto alkylamino-bonded silica (V23). The enzymatically released hydrogen peroxide was detected amperometrically. Detection limits were in the low pmol range. Instead of amperometry, using the same system Honda et al. (V24) detected the liberated hydrogen peroxide by the peroxyoxalate chemiluminescencemethod. By use of immobilized glycerol dehydrogenase, glycerol, 1,2-propanediol, and triglycerides (after cleavage with lipase and esterase) were determined by using NADH, produced by the enzymic reaction, as a fluorophor (VW). Meek and Nicoletti (V26) determined organic phosphates such as inositol triphosphates by using an anion-exchange separation column followed by a column of immobilized alkaline phosphatase to hydrolyze the phosphate esters. The stream is then mixed with molybdate solution and the inorganic phosphate determined by the conventional reaction. Electrochemical Derlvatlzatlon

Electrochemically generated bromine was used to brominate aflatoxinsas they eluted from a reversed-phasecolumn ( V27). The brominated aflatoxins were measured fluorometrically with low ppb detection limits in cattle feed. The reaction of carbohydrates with ethylene sulfate in a weakly alkaline medium at elevated temperature generates electrochemically oxidizable compounds and gave a lower limit of detection for aldoses of approximately 1 pmol (V28). Chromophoric Derlvatlratlon

Visible detection can be used where a postcolumn reaction produces a colored compound or where the spectrum of a dye molecule is shifted by interaction with an analyte. Free formaldehyde in cosmetic samples was determined by Engelhardt and Klinker (V29) using the lutidine method in a reaction detector with knitted open tubes. Detection at 420 nm was used but sensitivity was increased 4-fold, down to 15 ppb, with fluorometry. After rare earths were separated by ion exchange, the addition of Arsenazo I11 solution in chloroacetic acid buffer gave a colored compound detectable at 665 nm (V30). Phytic acid as well as inositol mono-to-penta-phosphates were measured in food a t 500 nm using a reactant of 0.015% Fe(II1) chloride containing 0.15% 5sulfosalicyclic acids (V31). Phenols on the U.S.EPA priority pollutant list could be measured by using a reversed-phase column and phosphoric acid-methanol gradient (V32). Detection was accomplished by the use of 4-aminoantipyrine and K,Fe(CN)3 solution added postcolumn to give a visible chromophore with absorbance maxima at 509 or 470 nm depending on the phenol structure. Reducing sugars were measured in dry wine by postcolumn derivatization with tetrazolium blue (V33) after separation on a cation-exchange column in the lead form. Cysteine was measured in pharmaceutical solutions by the reaction with 5,5'-dithiobis(2-nitrobenzoicacid) (DTNB) which releases a yellow chromophore (V34). A split-flow parallel column ion-exchange procedure for postseparation pH modification was described by Jansen and co-workers (V35). The second column consists of an anionexchange column in the hydroxide form. The mobile phase contains acetate which, after splitting between the analytical 418R

ANALYTICAL CHEMISTRY, VOL. 60, NO. 12, JUNE 15, 1988

column and the second column, when it passes over the anion exchange releases hydroxide ion which is recombined with effluent from the analytical column. This alkaline medium can be used to enhance detection processes as was demonstrated with the enhanced UV detection of barbiturates at 254 nm. By use of a similar approach,the same research group used an anion-exchange resin in the hydroxide form to separate and supply hydroxide needed for postcolumn derivatization with phenanthrenequinone (V36). Diazotized sulfanilic acid was used for the postcolumn detection of phenols using air-segmented continuous flow; highly colored azo compounds gave detection limits of 17 pg/L in water (V37). Kondoh et aL (V38)determined triglycerides by ~ t c o l u m n hydrolysisfollowed by oxidation of glycerin to formaldehyde. The formaldehyde was reacted with acetylacetone and then detected at 410 nm. Penicillins (V39) and amoxicillin (V40) were degraded by the treatment with NaClO in alkaline medium in the presence of methanol. The product has an absorption maximum at 280 nm. Ampicillin and its metabolites were measured in human urine using the alkaline degradation procedure but this time in the presence of mecuric chloride and EDTA disodium salt with absorbance measurement at 300 nm (V12). Cyclodextrins in biological fluids were detected by a vacancy colorimetric detection using postcolumn complexation (V41). The ionexchange separation of phosphonoformate, phosphite, and phosphate followed by addition of bromine (oxidizes all to phosphate) then molbdovanadate reagent gives nanogram detection limits when measured at 340 nm. The excess bromine must be reduced by addition of sulfite prior to detection ( V42). Chemiluminescence and Fluorescence

A review on the use of luminescence techniques covered three aspects of online use: (1) postcolumn derivatization using fluorescencedetection; (2) laser fluorescence detection; and (3) chemiluminescence usin the peroxyoxalate system (V43). N-Nitrosamines were hytrolyzed and the hydrolysis products reacted with Ce(1V) to produce the fluorescent ion Ce(II1) allowed measurement at ppb levels (V44). A wellstudied two-step postcolumn method for the analysis of carbamate pesticides involves the alkaline hydrolysis to form methylamine, then the reaction of this product with OPA giving a strongly fluorescent product. Engelhardt and Lillig (V45) studied reaction conditions and applied a cyclone type mixer in order to detect 20 ppb. The combination of ascorbic acid, hydrogen peroxide, and HC1 is known to induce fluorescence in the digoxin molecule. This postcolumn reaction permitted measurementof digoxin at 0.5 ng/mL plasma levels (V46, V47) after separation by RPC. N-Acetylcysteine, (intact and oxidized forms) in plasma was measured by its rapid reaction of pyrene maleimide which gave a fluorescent product (V48). Chemiluminescence using the well-established peroxyoxalate system has been used for postcolumn detection of a variety of compounds. This reaction was reviewed by Imai (V49). Imai et al. (V50) also studied this reaction in order to simplify the detection system by premixing certain reagents and studying effects of pH, salt, reagent concentration and ratios. This research group also investigated several other oxalates in addition to the PO ular bis(2,4,6-trichlorophenyl) oxalate (TCPO) and bis(2,4-&itrophenyl) oxalate (DNPO) reactants, which were studied as a function of pH and composition of the final solution. DNPO was found to be a more suitable reagent since its half-life is lower and less band broadening occurs. A new chemiluminescent reagent bis[ 4nitro-2-(3,6,9-trioxadecyloxycarbonl)phenyl] oxalate was found to be more stable than TCPJ in the presence of hydrogen peroxide (V51). The miniaturization of the peroxyoxalate system using packed-capillary fused-silica columns of 0.32 mm i.d. was demonstrated by DeJong and co-workers (V52, V53) and by Weber and Grayeski (V54),the latter using a sheath flow of chemiluminescent reagents around the column effluent. A new postcolumn chemiluminescent system based on the reagent tris(2,2'-bipyridine)ruthenium(III) and the analyte aliphatic trialkylamines was suggested (V55). Mattsev et al. (V56) used a postcolumn chemiluminescencereaction to determine heme-containing proteins separated by SEC.

COLUMN LIQUID CHROMATOGRAPHY

PRECOLUMN DERIVATIZATION Developments in precolumn derivatization included more use of solid-phase derivatization, new high-sensitivity fluorescent reagents, and online automated procedures. Similar to postcolumn derivatization, precolumn reagents are available with fluorescent, colorimetric, chemiluminescent, and elec. General articles on precolumn derivatization trochemical include refs%- W4. The area of amino acid precolumn derivatization was well studied in this period as evidence by several reviews (W5- W7). FMOC-C1,9-fluorenylmethylchloroformate, is a fluorescent reagent which can react with both primary and secondary amino acids. Einarsson (W8)modified this procedure by first removing the primary amino acids with OPA-mercaptoethanol reaction, followed by fluorescent labeling of secondary amino acids with FMOC-C1. The FMOC derivatives can be easily separated on a RPC column. Other precolumn reagents for amino acids included 9-anthryldiazomethane (W9, W l O), phenylisothiocyanate (WII-W14), OPA (W15, WlS),dabsyl chloride (WI7, WIB), dansyl chloride (W19),the chemiluminescent tag 4-isothiocyanatophthalhydrazide(W20),2,3naphthalene dicarboxaldehyde and cyanide (W21),and anilinothiazolinone ( W22). Optically active amino acids can be separated with various precolumn derivatization reagents. Einarsson and co-workers (W23) reacted D-and L-amino acids with OPA and optically yvanoside (TATG). active 2,3,4,6-tetra-0-acetylyl-thio-j3-gluco The diastereomers are separated by RP8 and were detected with picogram sensitivity by fluorescence. The molecule (+)-1-(1-naphthy1)ethyliiocyanatereacts with racemic amino acids to form naphthylethyl carbamoyl derivatives (W24). Detection can be either absorbance or fluorometry. Use of OPA combined with chiral mercaptans (Boc-L-cysteine,Nacetyl-L-cysteine, and N-acetyl-D-penicillamine)as reagents resulted in an HPLC method for the separation of enantiomeric amino acids and amino alcohols (W25). Likewise, Lam (W26) resolved D- and L-amino acids by first reacting with OPA in the presence of N-acetyl-L-cysteine. The derivatized amino acid was then separated in an RPC column by using a mobile phase containing L-proline and Cu(1). A study of the mechanism of optical selectivity confirmed that the chiral sulfhydryl reagent was responsible for the formation of a diastereomeric mixed chelate complex and for the resolution of the isomers. Several new group selective precolumn derivatization reagents were noted. For carboxylic acids, 3-bromomethyl6,7-dimethoxy-l-methyl-2( lZ-Z)-quinoxalinoneis a highly selective and sensitive fluorescent reagent (W27, W28). It reacts with fatty acids in acetonitrile in the presence of 18-crown-6 and K2C03to produce fluorescent esters (W28). A popular fluorescent carboxylic acid specific reactant N-(bromomethyl)-7-methoxycoumarinwas used to determine carboxybetaine amphoteric surfactants in household and cosmetic products by RPC (“29) and pyrimidine compounds in serum (W30). New carboxylic acid derivatizing agents based on coumarin were evaluated by usin peroxyoxalate chemiluminescence; one in particular, 7-(iethylamino)-3-[4-((iodoacetyl)amino)phenyl]-4-methylcoumarinrequired only a one-step reaction and gave detection limits in the low femtomole range (W31). Catecholamines are among the more interesting endogenous compounds to be analyzed in body fluids and tissues. Although some catecholaminespossess native fluorescence, they are often in trace amounts and are frequently derivatized. A novel approach that permits derivatization and impurity removal on a solid phase was described by Tsuchiya et al. (W32). The catecholamines are sorbed onto alumina, the amino group not responsible for adsorption was dansylated by a solid-phase reaction, the excess reagent and fluorescent impurities were washed out, and the danyslated catecholamines eluted and were separated by RPC. New reagents for other compound groups were noted for the following: thiols (W32), glucuronides (W34),aldehydes/ketones (W35),certain peptides (W36), alcohols (W37), fatty acids (W38), and long chain n-alkylamines (W39). Solid-phase derivatization is an area where potential advantages can be gained. Selective polymeric and other reagents where the derivatizing agent is incorporated, coated, or bonded have been reviewed by Colgan and co-workers

(W40)for electrochemical detection. The reagents can be used offline or online. Such reagents can be in the form of a replaceable cartridge and can easily be changed when exhausted. Certain solid phase reagents can be configured to only release reactant when a reactive species passes over its surface. A polymeric anhydride containing o-acetylsalicyl as the labeling moiety was used offline by Chou et al. (W41) to derivatize primary and secondary amines which could be monitored by a UV or electrochemical detector. A catalytic Nafion-H precolumn reactor placed between the injector and the analytical column was used by Sadek et al. (W42) promote cyclization reactions. The excess reactant and derivatization products were separated online in the LC column. The procedure ww also used as a micropreparativetechnique. Colgan et al. (W43) derivatized ethylene dibromide with silica-supported silver picrate. An area of considerable promise is the automation of precolumn derivatization reactions enhanced by a new generation of HPLC autosamplers, online sample manipulation devices, and through robotic approaches. The FMOC-C1derivatization of primary and secondary amino acids was automatically carried out in autosampler vials by Cunico and co-workers (W44). The autosampler added reagent to the amino acid solution by using vial-to-vial transfer capabilities and permitted liquid-liquid extraction of reaction byproducts by an automixing feature. A slightly modified FMOC-C1-amino acid procedure was described by Betner and Foeldi (W45) where the residual FMOC-Cl is eliminated by treatment of the reaction mixture with 1-aminoadamantane rather than by extraction as in the above automated procedure. The reaction byproduct eluted late in the chromatogram eliminating one step in the automation sequence but extending analysis time. Autosamplers were used to optimize reaction time for the derivatization of neurotransmitter amino acids (W46),amino sugars (W47),and aspartate and glutonate in CSF (W48). An automated derivatization of amine-containing antibiotics (W49) and 6-acetylmorphine (W50) were reported. Precolumn derivatization procedures have also been reported for alcohols (W51), amino compounds (W52, W53), digoxin (W54),taurine (W55),and thiobenzamide derivatives ( W56).

MICRO LC, MICROBORE LC, AND FAST LC Although the terms micro LC and microbore LC have been used interchangeably for some time, microbore LC has evolved to refer to columns with internal diameters of 0.5-2.2 mm with microbore 2 mm being the most popular. Micro LC generally refers to columns with