Hydrodynamic chromatography - Analytical Chemistry (ACS

---

ReDort

Hydrodynamic Chromatography A chromatographic technique discovered about 13years ago at The Dow Chemical Company and developed there hy a number of scientists now appears to be coming of age. Hydrodynamic chromatography (HDC), as the method is called, is widely used within Dow, and its use should expand as commercial instruments based on it become available in the near future. As with many new instrument developments, HDC was conceived and developed to satisfy a perceived need, in this case the need to reduce the time involved in measuring the size of, and thus better understand and control, colloids. Matter of colloidal dimensions has many forms and impinges in many ways on our daily lives. Clays, viruses, photographic emulsions, paints, and blood are some of the many things that comprise or embody within them this state of matter that in size lies roughly between several nanometers and a few micrometers (Figure 1).The size of such materials is obviously of fundamental interest in our efforts to

I

looa 0.01

low A 0.1

understand their behavior or to put them to use. Until recently, however, methods available for size measurement-particularly size distribution analysissuffered from serious limitations. High cost, complexity of instrumentation, slowness, and often the plain inability to obtain information without compromising the accuracy of the measurement were typical of the difficulties. HDC appears to have overcome many of these limitations. It is fast (recent embodiments of the technique can deliver a complete size distribution in as little as 10 min), the equipment is relatively inexpensive, and, most importantly, since it is in many respects a typically chromatographic technique, it does not call for uncommonly high skill in either ohtaining or interpreting data. Though it is fairly conventional from an operating point of view, HDC is from a chromatographic viewpoint quite unusual in a t least a couple of ways. In the first place it breaks with one of the often-cited prerequisites for

10 000 A 1

Figure 1. Particle size ranges 882 A

ANALYTICAL CHEMISTRY, VOL. 54, NO. 8, JULY 1982

1wmA 10 rrm

I

rn

I

successful chromatography, namely, that the species of analytical interest should he soluble in the mobile phase. The “solutes” of HDC are in most cases quite insoluble in the mobile phase. They are in fact in suspension, and a t times they are large enough to be seen in a light microscope. Not surprisingly, HDC reveals phenomena and problems not commonly encountered in chromatography. I t is in the manner of its separation that HDC is perhaps most unusual and departs radically from conventional practice. Normally chromatography employs two phases-a stationary phase and a mobile phase. The state of the mobile phase, whether it he a gas or a liquid, dictates to which broad class of chromatography a method belongs. Since it employs a liquid as mobile phase, HDC is a branch of liquid chromatography. HDC is unusual, however, in that it does not employ a stationary phase, at least not in the usual sense. Thus the heart of the HDC device is a column packed with spherical particles; but at the same time the interior of these particles is almost invariably quite inaccessible to the “solute” particles being separated. Therefore, any separation that does take place is brought about by phenomena operating exclusively within the void volume of the packed bed. This apparent noninvolvement of a stationary phase has led one author to declare recently (I) that HDC is “not truly chromatographic.” One of our objectives in this REPORT will he to show how, from a somewhat broader perspective, HDC should he considered chromatographic in nature. But first some background on how the method came about and how it operates. HDC began as a response to a rather specific request. One of us (H.S.) was asked to develop a rapid method 0003-2700/82/0351-892A$01.OO/O @ 1982 American Chemical Society

Hamish Small Martin A. Langhorst Dow Chemical USA

Midland, Mich. 48640

7 Pressure Gauge

O.l-prn

Filter

I

Injection Valve

Ell

2- n

Fie

I

Detector

gued, lead to unacceptable peak broadening and poor resolution. As an alternative to size exclusion a mode of separation was proposed that was based on analogies drawn between sorptionldesorption of molecularsized species and flocculationldeflocculation of their colloidally dimensioned counterparts. Accordingly, experiments got under way wherein short beds of cation exchange resin and a deionized water mobile phase were used in an attempt to separate polystyrene latex particles on the basis of size. Initially the separations obtained were slight and rather unpromising. However, most significant was the observation that large particles were eluting ahead of smaller particles, which was the opposite of what was to be expected if a flocculation/ deflocculation mechanism prevailed. Eventually, with a better uuderstanding of the various interactions involved in transporting colloids through packed beds, it became apparent that the chromatographic conditions that were first employedprincipally the low ionic strengthdid not favor the flocculation mechanism. In the meantime we set about exploring and defining the very interesting phenomena associated with the particle transport, concurrently developing the technique that we later called hydrodynamic chromatography

&!*= ID

Computer

(2).

Flgure 2. Block diagram of high-speed computerized HDC. Reprinted with permission from J. Colloid Interface Sci.

for determining the size of a commercial "plastisol" resin whose particles were around 1pm in diameter. A method was developed which, though fast, was so narrowly applicable that it brought home the need for some better method of particle size analysis in the submicrometer range.

The idea of extending size-exclusion chromatography to colloids was considered hut discarded in view of the low diffusivity (D a 10-9 cm2 s-1) of the I-pm "solute" particles that were of interest a t that time. Poor mass transfer of colloid into particles of the requisite porosity would, it was ar-

HDC hardware has much in common with conventional liquid chromatography. It employs packed columns and liquid eluents that are usually aqueous solutions of salts and surfactant. A pump capable of very steady and preferably pulseless delivery a t moderate pressures, a sample injection device, a colloid detector (turhidimeter), and various means of data collection and processing complete the

ANALYTICAL CHEMISTRY, VOL. 54, NO.

8, JULY 1982

893P

equipment (Figure 2). The rolumn. the heart of the HDC device, is packed with usually spherical particles. A variety of packing materials have been used-ion exrhange resin beads, nonfunctionalized polymer beads, and glass spheres-hut the resin has proved to be the best from a practical point of view. The resolving power of the packed columns is rririrally dependent on the size of the spheres and the manner of their packing. Beads of 15-20 fim in diameter are commonly used. Early HDC devices used 3-5 m of column (2). but with improvements in packing procedures augmented with romputerized data proressing, the performance of columns just 0.5 m long is now adequate for solving a variety of particle size prohlems. A typical eluent for HDC will contain a buffer, e.g., sodium phosphate \atuiut 0.01 M )and a surfactant such as sodium lauryl sulfate at about

I Volume

1 g/L.

Polystyrene latices of very narrow and well-charas:.erized particle size have heen extensively employed as model colloids in exploring HDC elution behavior. In a typical experiment a small molecular marker \sodium dichromate) is injected to art as an internal monitor of tlow variations. Figure 3 is a chromatogram from injection of a mixture of two monodisperse latices and a simultaneously injected marker. The mixture was eluted through 3 m of20-fimration exchange resin beads in the sodium form. Most noteworthy, of course, is the elution of large particles ahead of the smaller ones. both orecedine ” the marker species. The discoverv of this size-sortine effect, in addition to pointing the way toward a size analysis method. also

the rate of transport of colloid to that of the marker. In most cases the marker is confined for various reasons to travel in the void space of the column, and its elution rate is therefore a measure of the flow rate of the eluent through the packed bed. We have seen how Rr is larger the larger the colloid particle but, in addition, &is invariably greater than unity. In other words, the particles move through the bed with a higher mean velocity than the fluid carrying them. In our initial attempt to explain these unusual observations, we considered the crevice region where beads contacted each other as being a sizediscriminating region that admitted species more or less readily depending on their size. However, calculations of this volume for a typical bed showed it to be an insignificant effect in accounting for the separations observed. A problem in devising a mechanism involved the fact that only a single phase, the eluent, was involved since the beads were quite impermeable to the “solute”-the latex particles. Definitions of chromatography, on the other hand, invariably spoke of two phases, one stationary, the other mobile. But was this two-phase condition a prerequisite for separation? In fact, it is not, and HDC is a good example to illustrate the claim that the prerequisites for separation are: relative motion between two contiguous phases or regions of a single phase, and unequal distribution of solutes between these ohases or reeions of a single phase. Commonlv in LC. the condition of relative motion is very obvious-one phase is stationary with respect to the column while the other, the liquid, is not. In HDC tl relative motion

Figure 3. Chromatogram of mixture of two monodisperse latices and dichromate marker. Reprinted with permission from J. Colloid Interface Sci. provoked a search for a plausible explanation for the unusual phenomenon. It is perhaps appropriate at this stage to take up the question of mecbanism and whether HDC can indeed be classified as a chromatographic method. The Basis of HDC I t is customary in HDC to describe particle elution in terms of the & number, which is simply the ratio of

-

Figure 4. (a)Hydrodynamic effect. Colloid particles are excluded from the interface where the fluid velocity Is lowest. The larg er the particle is, the greater its mean velocity, (b) Electrostatic effect. Colloid particles are repelled by the charged interface. Thickness (6)of the excluded zone increases with decreasing ionic strength. R, increases with decreasing ionic strength 894 A

ANALYTICAL CHEMISTRY, VOC. 54, NO. 8. JULY 1982

THE ECONOMIC EDG IN ION ANALYSIS The Dionex Series 2OOOi Ion Chromatographs-. . .accurate ion analysis without complicated and costly sample preparation. Save 75 percent or more by slashing preparation time and ~iminatingcostly reagents. Simply dilute your sample and inject. It's hat easy. h a l y z e a wide variety ' If multiple ion samples, several hundred per lay. At the rate of an ion per minute. Get unsurpassed efficienc: The first totally inei high-performance lon Chromatographswstanu up to even the most

corrosive samples and solvents. Ion chromatography resins, available only from Dionex, provide the most stable and durable column packings available. Thus, no pH limitations on your samples. Dionex will keep you on the leading edge of ion anafysis. The Series 20dOi is specially designed to accept expanding technologies as we continue to develop them. For more information about theDionex Edge in ion analysis, call us in California (408-737-0700], New Jersey (609-596-0600), 'Ilinois (312-860-1030), Texas (713-463-6697) r Canada (416-675-3870). Or mail the coupon today.

. I *

SIMPLE, ECONOMIC & EFFICIENT

I

~"-"'-"-""""'---

'.

"

SERIES 2000i

'qN CHROMATOGRAPHS" want to learn about the Dionex Edge.

''" "' D'oNEXm ,

w ~ailto;Dionex,1228Tifa"' Dionex (UK)Lid.. Flrst I Surrey CUI6 5HY El

wr"-.-.--

GG

Sunnyvale.CK he Parade, Frim , D o n e x GmbF if

Germany

-~

El.

: I I I 1

comes about as a result of viscous forces in the flowing eluent that cause it to move more sluggishly the closer it is to the packing-eluent interface. The liquid flow in the interstitial void space is conceptually very similar to the very familiar Poiseuille flow in a capillary, and it is convenient to consider the complex column void space as a system of such capillaries (Figure 4). When this source of relative flow is coupled with the size-discriminating influence of the interfacial region, which allows more ready access of particles the smaller they are, we have satisfied the two prerequisites for separation to take place. Techniques such as field flow fractionation (1)also take advantage of unequal flow within a single phase but to obtain the unequal distribution of solutes they must exert an additional influence, namely, the “field.” In HDC no such extra influence is required, and in this regard it is more akin to chromatography than is field flow fractionation. If therefore one uses our more broadened concept of the necessary prerequisites for chromatographic separations it is evident that HDC may quite logically be considered a form of chromatography. The early recognition that particle separation was intimately connected with phenomena of fluid flow prompted the use of the name hydrodynamic chromatography. However, the hydrodynamics of fluid flowing in the column void space and the purely mechanical intrusion of the packing-eluent interface are not the only factors that influence Rf. Since both the packing and colloid have associated double layers, there is a strong electrostatic interaction between the two that is in turn greatly influenced by the ionic strength of the eluent. This feature of HDC is amply documented and discussed elsewhere (2), so suffice it to say here that Rf increases with decreasing ionic strength of the eluent. At sufficiently high ionic strength the dependence of Rf on particle size reverses, and large particles have been observed to elute later than smaller ones. This is a manifestation of the influence of van der Waals forces that dominate when electrostatic repulsion forces become sufficiently suppressed by the ionic environment. It is obvious that the transport of colloidal particles through packed beds is a complex and very intriguing phenomenon, and it is therefore not surprising that much has been done to define and to provide a theoretical basis for HDC elution behavior. For additional background on these matters the reader is referred to several original publications (3-5). We devote the remainder of this REPORT to applications of HDC. 896A

Particle Size Distribution Though in principle HDC can be applied to a great variety of colloid particle size problems, our examples are of necessity chosen from the area of polymeric latices, since that area is one that has been of most interest and direct concern to us. The calculation of particle size distribution is by far the most straightforward application of HDC. In early work, relative particle size distributions were obtained by simply comparing chromatograms. In the case of monodisperse samples, the use of this technique afforded a convenient way of determining the size of a colloid. There were problems, however, in applying this approach to polydisperse colloids. In addition, the analysis was slow. To maximize the utility of this technique, the analysis time had to be reduced significantly and some mathematical technique used to calculate the particle size distribution immediately following the elution of the chromatogram. Obviously the two keys to this approach were the columns and software. The original columns, constructed of glass packed with either copolymer or cation exchange resin, were several meters long. The columns used today are l / ~in. o.d., 10 mm i.d. stainless steel tubing, and are less than l/Z m long. Typical elution time is less than 6 min. The packing material employed today is 15-pm cation exchange resin or copolymer. Typically, the measured efficiencies of present HDC columns vary from 27 000-30 000 plates in a 42-cm length of tubing (63 000-70 000 plates/m). These columns have a useful lifetime of six months of normal use, which corresponds to 2000 samples. Details concerning the preparation and evaluation of these columns can be found elsewhere ( 6 ) . The successful use of an HDC column to measure the particle size distribution requires that we characterize the shape of the column band spreading as closely as possible using some mathematical model. The basic model that we have chosen is a very general one, the Pearson Type VI1 (7). We have modified this function by addition of an asymmetry factor, which is a linear change in the CTof the peak as a function of direction and distance from the peak center and a “snouting factor,” which is an empirically derived foretailing, observed with most HDC columns. In the literature, the fitting of asymmetrical chromatographic peaks has generally been handled using a convolute integral of a Gaussian peak shape and an exponential decay function (8).The linear asymmetry function we have used in our model is much easier to calculate

-

ANALYTICAL CHEMISTRY, VOL. 54, NO. 8, JULY 1982

-

and appears to work as well as the exponential decay function for typical HDC peaks. The modified Pearson Type VI1 distribution can be made to fit most observed HDC peak shapes very accurately. The algorithm for this model is completely defined by six independent variables. Three variables (location, height, and width) describe the size and location of the peak. Three additional variables (shape factor, asymmetry, and snouting factor) describe the shape of the peak. Details concerning this algorithm can be found elsewhere (6). Calculation of Distribution The HDC chromatogram may be written as the convolute integral of several factors with respect to time:

F ( V ) = S[Wtt)G(V,t)K(t)ldt where:

F(V ) = observed chromatogram; W ( t )= distribution of sample; G(V = band spreading of system; K ( t ) = response of detector: and t = time. &( V ) is observed while G( V , t )and K ( t )can be measured using monodisperse latex standards. K ( t ) ,the detector response, is actually measured as a function of particle size, but is converted to a function of time for the calculation. W ( t )can then be calculated by several numerical techniques. The method we have chosen has been developed empirically and independently for HDC and subsequently found to be very similar to a technique developed for GPC by Ishige, et al. (9).

Briefly, the technique involves convoluting an assumed distribution with the measured band-spreading function and detector response curve to calculate the theoretical chromatogram. The ratio of the height of each point on the measured chromatogram to the corresponding point on the calculated chromatogram is then used to correct the original distribution estimate. Other investigators have used mathematical techniques to calculate a particle size distribution from an HDC chromatogram (10). While the technique that we describe here is similar to that used by Silebi and McHugh, there are differences that are detailed elsewhere (6). Accuracy and Precision A critical question in HDC is whether the particle size distribution calculated from the HDC chromatogram is an accurate representation of the true particle size distribution of the sample. This is difficult to prove when sizing small particles, since the standard method, indeed virtually the

Computer Deconvolu Zon Using Equation 1

are obtained when other mixtures of monodisperse latices are analyzed, Le., the same distributions are calculated individually and in mixtures, even when the chromatographic peaks are almost totally overlapped. It is difficult to define the precision of an entire distribution such as that calculated hy HDC. However, the standard deviation of the volume median diameter (50% point on the calculated cumulative volume dstrihution curve) of a typical 2200 A polystyrene latex was found to he *0.85% relative for 15 injections over two days. This precision is typical in HDC and corresponds to an uncertainty in total peak position of -0.1 s or i0.03% relative. The absolute accuracy of the amount of material calculated a t each diameter has not been rigorously defined for our system. However, HDC has been used to measure the relative amounts of mixtures of 850 A and 25M) monodisperse styrenehutadiene latices. Over a ran e of 10 90 to 5050 of the 8502500 laticei, the ahsolute error is consistently less than 1%. For example, the mean ratio and standard deviation for 11consecutive injections of an actual 2080 ratio mixture is 19.5580.45 with a standard deviation of 0.73% (6).

.f

I1

Particle Size

Distribution

Figure 5. HDC particle size distribution

only availahle reference method, transmission electron microscopy (TEM), has inherent errors that may exceed 5-7% for monodisperse latices (11). An indication of the accuracy of HDC may be obtained, however, hy comparing the size of a standard latex as determined hy T E M and HDC. Such an experiment was conducted in our laboratory in which the volume mean diameter of the standard was reported by TEM to be 2423 8, and hy HDC to he 2504 A. The observed error of -3% relative is less than that inherent in the TEM measurement technique. Our HDC distribution calculation technique has the ability to separate chromatographically unresolved peaks as in Figure 5 into the original monodisperse latices with distributions similar to those obtained when the latices are run individually. Similar results

Particle Aggregation Interactions between latex particles can bring about their aggregation. Sometimes this is inadvertent and undesirable while at other times the aggregation is intentional and heneficial. HDC bas been very effective in elucidating a number of particle aggregation problems. In one case, it was necessary to know if shearing of a particnlar latex had induced aggregation. Electron microscopic examination of the latex before and after shearing gave an amhiguous result since the sample appeared to he agglomerated before shearing and the apparent agglomeration could have been the result of sample preparation. Chromatograms of the sheared and unsheared latex, however, clearly showed the presence of an extra peak in the former that could he attributed to aggregates (2). The unsheared latex showed no extra peak. In another instance the chromatogram of a supposedly monodisperse latex showed pronounced skewing toward high particle size. Subjecting the dilute latex suspension to an ultrasonic treatment eliminated the skewness and gave a chromatogram that was characteristic of a “monodisperse” latex. We concluded that the latex contained a relatively high concentration of loose aggregates that was effectively broken up hy the ultrasonic treatment. These examples illustrate

how in some cases HDC may he very effectively used‘to arrive at important qualitative codclusions without resorting to a complete particle size analysis. The measurement of particle association in latex thickener systems is an example of HDC applied to a case where aggregation is intentional and must he controlled. In many latex applications various water-soluble polymers are added to impart desirable rheology to the final product. The development of thixotropic latex paints that are “thick” and dripless on the applicator hut free flowing when brushed or rolled is one such example. There is some evidence that this thixotropic behavior is due to a loose bridging of latex particles by the polymeric additives, and HDC offered a possible means of substantiating this. Chromatograms run on unmodified latices and latices to which thickeners had been added clearly showed a shift to faster elution times-presumably larger particles-as more thickener was added. Furthermore, the efficacy of various polymeric thickeners correlated nicely with their effect on the Rr of latices to which they had been added (12). Swelling Effects HDC has a unique ability to elucidate particle swelling effects in polymer latices. Swelling can he extremely

4000

3000

045 L

sm

5 5

BOO

0

1

2

3

% Acrylic Acid

4

5

6

in SIBIAA Latex

I Flgure 8. Swelling effects in SBAA latices, indicating the influence of the environment on the effective diameter of the latex particles. Eluent A, pH = 7 , sodium lauryl sulfate 0.5 g1L. Eluent B, pH = 10, sodium lauryl sulfate 0.5 glL. Eluent C, pH = 10. Triton X-100 0.5 glL. Reprinted with permission from A&. Colloid Interface Sci.

ANALYTICAL CHEMISTRY. VOL. 54, NO, 8, JULY 1982

-

897A

REAGENTGRADE SOLVENTS Users of Corco chromatographic, elec tronic and spectrophometric reagent. grade solvents know they can count or Corco for highest quality. Reliability. Anc delivery. In pints. gallons, five-gallons, 01 drums. At Corco, we've been specialists ir reagent-grade solvents since 1953and have built our reputation on meetin< the requirements and specifications o our customers, ACS and ASTM. Corco can also provide reagent grade acids, bases, and specialty chem icals. Write or call regarding your specific solvent requirements and our complett solvent listing in Bulletin 10. Or circle tht number.

Manulacturm ot Reagent & Electronic Chemicals Tyburn Road & Cedar Lane Fairless Hills. Pa. iW30 (215) 295-5006 CIRCLE 32 ON READER SERVICE CARD

898 A

ANALYTICAL CHEMISTRY, VOL.

important since it alters the volume fraction occupied by the particles, which in turn is a major factor in controlling rheology. When the polymer phase is of a nonpolar type such as styrene or styrene-butadiene, the volume of the particles will not be affected by altering the pH of the aqueous phase or its ionic strength or the type and amount of surfactant it contains. On the other hand, if the latex contains a sufficient level of an ionic comonomer such as acrylic acid, then the particle size can be profoundly affected by changes in these factors. Figure% illustrates the observations for a series of styrene-butadieneacrylic acid latices of varying acrylic acid content where the environment was changed by altering the pH and surfactant in the HDC eluent. I t is evident how HDC revealed some profound changes in the size of the latex particles in response to the changes in their aqueous environment.

Saunders, who have had a major part in applying and refining the technique. The patient support of Dow management we also gratefully acknowledge. The HDC method described in this article is patented by Dow Chemical Company and licensed to Micromeritics Instrument Corporation for commercial use.

Summary Some of the attributes of HDC that recommend it as a means of particle size analysis are: Itoffers a fast, precise means of analysis in the submicrometer range. I t has very broad scope. In addition to polymer latices it has been used on inorganic colloids such as silver halides, silica, ferric oxide, 4nd titanium oxide, on carbon blacks, and on viruses. While HDC was designed as a tool for colloid studies, recent work has demonstrated its utility in the field of very high molecular weight polymers for which conventional sizeexclusion chromatographic methods are unsuitable (13). The equipment is relatively inexpensive and if carefully maintained has long-term reliability. No special skills have to be developed to practice the method-it requires only normal chromatographic technique. In many applications it is not necessary to know other properties of the colloid such as refractive index or density. Some other techniques that are appropriate to this size range are, on the other hand, density as well as size devendent. which comvlicates their use. HDC offers a means of examininz how colloids shrink, swell, aggregate, or otherwise alter their effective hydrodynamic diameter as their environment changes. To our knowledge this is an ability that is unique to HDC.

(9)

Acknowledgment In bringing this technique to its present state of development many Dow scientists have been involved. We would especially like to mention G. McGowan, R. Pelletier, and F.L. , NO. 8. JULY 1982

References (1) Giddings, J. C. Anal. Chem. 1981.53, I,?ilA

(2) Small, H.J. Colloid Interface Sei. 1974.48,147. (3) Prieve, D. C.; Hoysan, P. M. J. Colloid Interfoee Sei. 1978.64,201. (4) Buffham,B. A. J. Colloid Interface Sei. 1918.67.154. (5) Silebi, C.; McHugh, A. J. AIChE J. 1978,24(2), 204. (6) McGowan, G. R.;Langhorat, M. A. J.

Collord Interface Scr., m press. Veeraragharen,V. G.; Rubin. H.; Winchell, P. G. J. Appl. Crystallogr. 1977,10,66. (8) Grushka, E. Anal. Chem. 1972.44, (7) Hall, Jr., M. M.;

-.

1111

Ishige, T.; Lee, S.I.; Hamielic, A. E.J. Appl. Polym. Sei. 1971.15,1607. (10) Silebi, C. A,; McHugh, A. J. J. Appl. Polym. Sci. 1979.23,1699. (11) Davison, J. A,; Haller, J. S. J. Colloid Interface Sei. 1974,47,459. (12) Small, H.; Saunders,F. L.; Solc, J. Ad". Colloid Interface Sci. 1976,6,231. (13) Prud'homme,R. K.; Froman, G.; Hoagland, D. A. Corbohydr. Res., in press.

Small

Langhorst

Hamish Small, a research scientist with Dour Chemical Companyy,has a BSc and a n MSc from the Queen's Uniuersity of Belfast, Northern Ireland. He is the inuentor of ion chromatography and hydrodynamic chromatography. His research interests include ion exchange (equilibria, kinetics, synthesis, and applications) and liquid chromatography. Martin Langhorst obtained a n AB degree from Centre College (Kentucky) and a n MS a t the Uniuersity of Kentucky. He is a senior research chemist a t Dow's Michigan Division Analytical Laboratories. His current research interests include chromatographic methods of polymer characterization, membrane separations, hydrodynamic chromatography, and particle size analysis.