- Email: [email protected]
Physical chemical characterization of percoll. I. Particle weight of the colloid
Physical Chemical Characterization of Percoll I. Particle Weight of the Colloid TORVARD C. LAURENT, H/~KAN PERTOFT, AND OLAV NORDLI Institute o f Medical and Physiological Chemistry, University o f Uppsala, Biomedical Center, Box 575, S-751 23 Uppsala, Sweden Received May 30, 1979; accepted October 10, 1979 Percoll is a solution of colloid silica in which the particles have been coated with polyvinylpyrrolidone. It is used to purify cells and cell components. Its particle weight, determined by light scattering and turbidimetry in water and in 0.15 M sodium chloride, pH 7.4, is 6.5 x 106. From electron microscopy one can estimate a weight-average particle weight of 5.6 x 106 and a number-average of 4.9 x 106. Osmometry in 0.15 M sodium chloride gives a numberaverage of 6 × 105, indicating the presence in the solution of a low-molecular-weight component, presumably free polyvinylpyrrolidone. Data on Percoll from light scattering and osmometry in distilled water show a typical macroion behavior. Percoll is stable at physiological ionic strength and pH while a pure silica colloid aggregates. In dilute solutions (<0.1% colloid) and pH 6-7 one can register a slow aggregation of the colloidal particles in 0.15 M NaCI. This can be prevented by addition of low concentrations of polyvinylpyrrolidone, indicating that the presence of a small amount of free polymer stabilizes the colloid.
at low ionic strength. The present report deals with the particle weight of Percoll and subsequent papers discuss its particle size and interaction (3) and its sodium binding (4). Extensive studies have been performed on pure silica colloids. Iler has published several reviews on the subject (5-7). We have for comparison performed measurements in parallel on Percoll and Ludox HS, the parent silica colloid of Percoll.
INTRODUCTION
Percoll is a silica colloid in which the particles have been covered by a layer of polyvinylpyrrolidone (PVP). The PVP content is about 11-12% of the dry weight (1). The coating stabilizes the colloid and makes it nontoxic. Percoll can therefore be used as a density gradient medium for the separation of cells and other biological particles (2). A few physical chemical parameters of the colloid have been reported (1). The particles are close to spherical and have a mean diameter in the electron microscope of 17.3 nm. We now report a more extensive characterization of the material. Percoll is used for practical purposes both at physiological ionic strength and in salt-free media (2) and the measurements presented here have therefore been carried out at both conditions in spite of the theoretical difficulties in interpreting the data obtained
MATERIALS AND METHODS
Materials. Percoll (Batch 96544D) was kindly supplied by Pharmacia Fine Chemicals, Uppsala, Sweden. The colloid dissolved in water has a concentration of 26 g/100 ml. A pure silica colloid, Ludox HS, which is the starting material in the production of Percoll, was obtained from Du Pont de Nemours, Wilmington, Delaware. It has a concentration of about 50 g/100
124 0021-9797/80/070124-09502.00/0 Copyright© 1980by AcademicPress, Inc. All rightsof reproductionin any formreserved.
Journalof Colloidand Interface Science, Vol. 76, No. 1, July 1980
125
P H Y S I C A L C H E M I C A L C H A R A C T E R I Z A T I O N OF P E R C O L L TABLE I The Specific Refractive Index I n c r e m e n t at 546 n m for Percoll and L u d o x HS Refractive index increment (ml/g) Solvent
Percoll
Ludox HS
0.15 M NaC1 H~O
0.071 -+ 0.002 0.073 _+ 0.001
0.061 - 0.0003
ml. The colloids were adjusted with hydrochloric acid to the desired pH (usually pH 7.4) and dialyzed for at least 48 hr against repeated changes of distilled water. They were then, when appropriate, dialyzed an equal time against 0.15 M sodium chloride. A concentrated sample of Percoll (0.55 g/ml) was recovered from the bottom of a sedimentation equilibrium run (3). Polyvinylpyrrolidone (PVP) (Plasdone C-15; MW 10,000) was obtained from GAF Corporation, New York, New York. Determination of concentration. The concentration of stock solutions was determined by dry-weight measurements. The solutions and corresponding solvents were dried overnight at 95°C in vacuo and the concentration was taken as the difference of weights of dry residues from solution and solvent. Refractive index increment. A Rayleigh interference refractometer (Hilger & Watts, Ltd., London, England) was used to determine the refractive index increment at 546 nm. The temperature was 20.3°C and 3-4 determinations were made in the concentration interval 7-36 mg/ml for each solute/solvent system. Turbidity. The optical density was measured at room temperature in 5-cm glass cuvettes in a Zeiss spectrophotometer. The turbidity (~-) is, when light absorption can be neglected, defined as 2.303 "optical density/path length. All colloid solutions, obtained by dilution of stock solutions with the appropriate solvents, were cleared from dust by filtration through Millipore filters (Type GS, pore diameter 0.22 /zm) placed
in Swinnex-25 filter holders (Millipore Corp., Bedford, Mass.). The presence of light absorption in 0.1 g/ml Percoll solutions was checked by plotting the optical density at wavelengths between 400 and 800 nm versus t h e r e c i p r o c a l of the fourth power of the wavelength (see below). This gave a linear relationship at wavelengths above 500 nm. There was a significant deviation from linearity at lower wavelengths due to the absorbance by PVP. Further measurements were performed at 546 nm. Light scattering. Light scattering was measured at 546 nm in cylindrical cells in a Sofica PGD 42 000 M photometer (Sofica, Le Mesnil-Saint Denise, France) at 20 +_ 0.1°C. All solutions were freed from dust as described above. Most measurements were performed at an angle of 90° with respect to the direction of the incident beam. When necessary the angular de10 O
6 x
,
I 2
I
0.05
I
0.10
CONCENTRATION(g/mr) FIG. 1. Turbidity m e a s u r e m e n t s performed on Percoll in 0.15 M NaCI (O) and in distilled water (©) and on L u d o x HS in 0.15 M NaC1 (A) and in water (A). T h e refractive i n c r e m e n t u s e d to calculate the c o n s t a n t H (see text) for L u d o x HS in 0.15 M NaC1 was t h e s a m e as that m e a s u r e d in water (Table I). Journal of Colloid and Interface Science, Vol. 76, No. 1, July 1980
126
LAURENT, PERTOFT, AND NORDLI TABLE II Molecular Weights and Viral Coefficients for PercoU and Ludox HS Molecular weight x 10-6 Sample
Method
Mw
0.15 M NaC1 0.15 M NaC1 0.15 M N a C 1
Turbidity Light scattering Osmometry
6.6 6.4
Percoll from sedimentation equilibrium
0.15 M NaC1
Osmometry
Percoll
H~O H20
Turbidity Light scattering
0.15 M NaCI 0.15 M NaC1 H20 H20
Turbidity Light scattering Turbidity Light scattering
Percoll
Ludox HS
Solvent
M~
A2 (mole g-2 cm~ × 106)
0.6
1.3 1.8
2.3
1.8
6 (?) 6.4
530
20a 23a 6.5 6.5
a These values should not be regarded as definite molecular weights but as a sign of a progressive aggregation of the Ludox HS particles. p e n d e n c e of the scattering was m e a s u r e d between 30 and 150 ° and the values extrapolated to 0 and 180 °. All readings were corrected for solvent scattering, polarization, and the g e o m e t r y of the scattering volume. The results h a v e been e x p r e s s e d in terms of the Rayleigh ratios, Ro (=io'r2/ I0), where 0 is the angle of observation, io the intensity of scattered light, I0 the intensity of the incident b e a m , and r the distance f r o m the cell to the recorder. Osmometry. Readings were m a d e in an electronic high-speed m e m b r a n e o s m o m e t e r ( H e w l e t t - P a c k a r d Model 501, F & M Scientific Corp., Avondale, Penn.) at 30°C. Dialysis tubing (Union Carbide Co., Chicago, Ill.) was used as m e m b r a n e . The solutions were dialyzed for at least 48 hr against the solvent before filling of the o s m o m e t e r . Each sample of a specified concentration was measured two to four times. The readings were recorded and equilibrium was always reached within 30 rain. In the calculation of i~he ~osmotic pressure f r o m column height~'~ppropriate corrections were made for the densities of the colloid solutions. Journal of Colloid and Interface Science, Vol, 76, No. 1, July 1980
RESULTS
Specific Refractive Index Increment The refractive index increment for Percoll was determined in distilled water and in 0.15 M sodium chloride and for L u d o x HS in water. The results are given in Table I. The value for L u d o x H S in w a t e r is in good a g r e e m e n t with the value of 0.062 ml/g given by Jennings and Jerrard (8).
Turbidity According to theory (see, e.g., Tanford (9)) the turbidity (T) is related to the weight-average molecular weight (Mw) by the equation
HC/T
=
1/Mw + 2A~c + . . . ,
[1]
where c is the concentration of solute (g/ml), A~ is the second virial coefficient, and H is a constant of the f o r m
H=
3 2 zt3n2(dn/dc )2 3Nh 4
where n is the refractive index of the medium (1.34), dn/dc the refractive index
127
PHYSICAL CHEMICAL CHARACTERIZATION OF PERCOLL 10
0 0 0
0
0
0
x
0 0
The turbidity of Ludox HS in distilled water did not show the same extreme concentration dependence as Percoll in water and the molecular weight obtained by extrapolation agreed with that of Percoll. Ludox HS measured in 0,15 M NaC1 displayed a much higher molecular weight. The pure silica colloid did apparently aggregate at the higher ionic strength. The molecular parameters obtained in the measurements are displayed in Table II. Light Scattering
i
I
5 10 CONCENTRATION ( g / m i x l O 3)
I
15
FIG. 2. Light-scattering measurements performed on Percoll in 0.15 M NaC1 (O), 0.15 M NaC1 + 10-3 g/ml of polyvinylpyrrolidone (11), and water (©) and on Ludox HS in water (A).
increment of the solute (see Table I), N is Avogadro's number (6.02 × 1023), and h is the wavelength of the light (5.46 x 10-~ cm). (It follows from the equation that the turbidity should be proportional to 1/h4. The deviation of the optical density for Percoll from this proportionality below 500 nm is a sign of light absorption (see Materials and Methods). The turbidity of Percoll and Ludox HS was measured in distilled water and in 0.15 M NaC1 in the concentration range 0.01-0.12 g/ml and the results are plotted in Fig. 1. The molecular weight of Percoll in 0.15 M NaCI calculated from the intercept on the ordinate is 6.6 x 106. The experimental points obtained in distilled water cannot be extrapolated with any certainty but are consistent with this value. There is apparently a strong increase in the interaction between the Percoll particles when the salt is removed. The limiting slope of the curves, which gives the second virial coefficient and which is a measure of interaction, is manyfold larger in water than in 0.15 M salt.
As light scattering and turbidity are related phenomena they will give the same information. Light scattering is, however, measured at a higher dilution as compared to those required in the turbidity measurements. When colloidal particles are small compared to the wavelength of the light the following relationship holds (9), [2]
Kc/Ro = 1/Mw + 2A2c " ' ,
where Ro is the Rayleigh ratio usually measured perpendicular to the incident light beam (Rg0) and K is a constant with the value K = 27r2n2(dn/dc)2 Nh 4
In particles of larger dimensions internal
1.0
O
.9
m~
.7
tx .5
I 0.005
I 0.01
i 0.01
CONCENTRATION [g/ml)
FIG. 3. The dissymmetry of the light scattering (R45/R135) displayed by Percoll (©) and Ludox HS (A) in water as a function of concentration. Journal of Colloid and Interface Science, V o l . 76, N o . 1, J u l y 1980
128
LAURENT, PERTOFT, AND NORDLI
interference causes Ro to vary with the angle and necessitates measurements of Ro over a large interval and extrapolation to zero angle (R0) for a correct determination of molecular weight. Measurements in 0.15 M NaCl. These were performed on Percoll in the interval 0.001 to 0.01 g/ml. There was no angular dependence of Ro and the molecular weight obtained from the values given in Fig. 2 was 6.4 x 10°. The measurements were performed both in the presence and absence of 0.1% PVP with identical results. The addition of PVP was due to an observation described below that PVP prevents aggregation of Percoll in very dilute solutions. The value of A2 obtained from the slope of the line is given in Table II. Measurements on Ludox HS in 0.15 M NaC1 revealed an angular dependence of Ro and the solution did apparently contain large aggregates. Extrapolation of the
5
t~
3
o
x
x
x
x
x t.a
">IQ:
2
1
0
I
I
I
5
t0
15
CONCENTRATION (g/mix10 ¢) FIG, 4. Extrapolation to zero concentration of
Kc/Ro (©), Kc/R9o (x), and Kc/R18o(0) for Percoll in water. The angular dependence of the light scattering disappears at infinite dilution and the extrapolation corresponds to a molecular weight of 6.4 x 106. Note that Fig. 4 displays a 10-fold lower concentration range than Fig. 3. Journal o f Colloid and Interface Science, V o l . 76, N o . 1, J u l y 1980
1.5
1.0
X
,vl,0.S a
o
~
ols Sin 2 0/2
FIo. 5. The change in angular light-scattering pattern at 20°C of 10-4 g/ml of the Percoll colloid in 0.15 M NaC1, pH 6.22. The recordings were made 3 (Q), 24 (O), 48 (A), and 240 (A) hr after dilution from a 0.24 g/ml solution. The decrease in Kc/Ro especially at low angles shows a time-dependent aggregation. The solution was kept in the same measuring cell during the whole time period.
experimental data to zero angle and zero concentration according to Zimm (10) gave an approximate molecular weight of 23 x 106. Measurements in water. Light scattering of Percoll and Ludox HS in water showed an unusual angular dependence. The dissymmetry, i.e., the ratio R45/R135, became less than unity within an intermediate concentration interval (Fig. 3). The angular dependence of the scattering disappears at infinite dilution and the extrapolation of Kc/R9o should give the correct molecular weight. Such extrapolations are shown in Fig. 2 for both Percoll and Ludox HS and both give molecular weights of the same order as the measurements of Percoll at high ionic strength (Table II). That the angular dependence disappears at zero concentration is further shown in Fig. 4 where R0 and R180, obtained by extrapolation from measurements at different angles, were used for plots of Kc/Ro and Kc/Ra8o. The slope of the Kc/Ro plot in Fig. 4 was used to obtain a value
PHYSICAL CHEMICAL CHARACTERIZATIONOF PERCOLL
I
.
^A
Z~
Osmometry
A
The osmotic pressure of Percoll solutions, extensively dialyzed against 0.15 M NaC1 or water, is shown in Fig. 7. A highly concentrated Percoll fraction recovered from the bottom of a sedimentation equilibrium run (3) was also dialyzed against 0.15 M NaC1 and measured at different dilutions. This fraction displayed a considerably lower osmotic pressure than the original Percoll solution (Fig. 7). The osmotic pressure (Tr) is related to the molecular weight by the equation (9)
0.5
0
129
I
,I
100
I
I
I
rc/RTc = 1/Mn + A2c + " " ,
200
Hours after dilution
FIG. 6. The change with time of Kc/Rgo in Percoll solutions containing 0.15M NaC1 with ( - - - ) and without ( ) the presence of 5 x 10-4 g/ml of PVP. The colloid concentrations were 2 x 10-4 (O), 4 x 10-4 (A), 6 x 10-4 (O) and 8 x 10 4(A) g/ml. The pH values in the different Percoll solutions were 6.2, 6.4, 6.6, and 6.8, respectively. of Az for Percoll in distilled water. It is given in Table II. M e a s u r e m e n t s on dilute Percoll solutions in salt. The Percoll solutions demonstrated an instability at high dilutions in 0.15 M sodium chloride and p H 6 to 7. This was evident from a time-dependent decrease in the Kc/Ro, especially at low angles, indicating an aggregation (Fig. 5). This effect disappeared at higher concentrations of the colloid and at 0.01 g/ml no aggregation could be detected. As a working hypothesis it was assumed that the PVP coat started to dissociate from the Percoll surface at high dilutions. It should be possible to prevent this by adding PVP and Fig. 6 demonstrates that PVP does prevent aggregation. As a matter of fact the addition of PVP seemed to cause a small initial deaggregation of the Percoll sample. H o w e v e r , also in the samples containing PVP there was eventually a certain formation of aggregates noticeable from the scattering at low angles.
[3]
where R is the gas constant, T the absolute temperature, and Mn the numberaverage molecular weight. A plot of 7r/RTc versus concentration is given in Fig. 8 for the Percoll solutions in 0.15 M NaC1. The intercepts correspond to number-average molecular weights of 6 x l0 s and 2.3 x 106, respectively. The slope of the lines and thus the value of Az is the same for
1.5
== 1.0 =
N
11.5
0
0.1 0.Z 0.3 CONCENTRATION (g/at)
0.4
FIG. 7. The osmotic pressure of Percoll solutions. The measurements were made in distilled water (O) and 0.15 M NaC1 (0). A fraction of Percoll was also recovered from the bottom of a sedimentation equilibrium run performed in 0.15 M NaC1 (3) and the osmotic pressure analyzed in the same medium (A). Journal of Colloid and Interface Science, V o l . 76~ N o . 1, J u l y 1980
130
LAURENT, PERTOFT, AND NORDLI
~a
I 0.1
i 0.2
I 0.3
I O.t,
EONEENTRATION (g/mL)
FIG. 8. The reduced osmotic pressure of Percoll measured in 0.15 M NaC1 (0) and of the Percoll fraction obtained from the bottom of a sedimentation equilibrium run (A). The intercepts on the ordinate correspond to molecular weights of 6 x l0s and 2.3 × 108,respectively. both samples (Table II). The osmometric data obtained in distilled water have not been used to calculate any molecular parameters. The experimental data in Fig. 7 are given because of the practical interest to know the osmotic pressure of Percoll in salt-free solutions (2). DISCUSSION
Molecular Weight Both turbidimetry and light-scattering measurements gave weight-average molecular weights of Percoll o f about 6.5 x 106 irrespective of ionic strength. The same value for L u d o x HS at low ionic strength confirms that this is a correct estimate. One should expect a 10% higher value for Percoll than for L u d o x HS due to the PVP coat. That this is not apparent in the measurements is most probably due to variation in the different batches of the colloids obtained from the manufacturers. Our results for L u d o x HS are in general agreement with numerous reports (for references see Jennings and Jerrard (8)). The number-average molecular weight of Percoll as measured by osmometry is Journal of Colloid and Interface Science, Vol. 76, No. 1, July 1980
considerably lower than the weight-average. This could of course be explained by a large polydispersity of the colloid and such an explanation is in agreement with the observation that a fraction recovered from the bottom layer of a sedimentation equilibrium run has a much higher Mn value. T h e r e is, however, an alternative explanation which is equally plausible. The Percoll may contain small amounts of free PVP. If Percoll has a molecular weight of 6.5 x 106 and if we assume a molecular weight of 104 for PVP then free PVP to an amount of 1.5% of the total dry weight would lead to a number-average molecular weight of 6 x l0 s. In the sedimentation equilibrium runs described in the following paper (3) the distribution of free PVP would hardly be affected by gravity while the concentration of Percoll in the bottom fraction (g Percoll/g H20) is about three times higher than in the original Percoll solution. With a relative concentration of free PVP which is only a third of that assumed above for the original solution one can calculate an Mn value in the order of 2 x 100. That the turbidimetric and light-scattering techniques give identical molecular weights for Percoll in 0.15 M sodium chloride after reliable extrapolations makes the simple turbidimetric technique attractive for routine work on Percoll in spite of the fact that the turbidimetric measurements must be performed at concentrations one or two magnitudes higher than the lightscattering studies.
Comparison with Electron Microscopy The distribution of particle diameters in Percoll was reported from electron microscopy to be 17.3 + 2.3 nm (1). Nearly the same value was found for L u d o x HS. A negative contrast technique with uranyl acetate was employed in this investigation to visualize the particles. We may use the data with certain assumptions to calculate a molecular weight for the Percoll par-
131
P H Y S I C A L C H E M I C A L C H A R A C T E R I Z A T I O N OF P E R C O L L
ticles. For example, we may assume that the uranyl acetate penetrates the PVP coat and that the electron microscopic pictures essentially show the silica nucleus. This is reasonable in view of the close to identical diameters estimated for Percoll and Ludox HS. The radius (r) reported is a numerical average. The volume and the molecular weight are proportional to the third power of the radius and we have therefore calculated the frequency distribution of r 3 from the data presented in Ref. (1). For comparison of the results with those from osmometry and light scattering we must obtain the number- and weight-averages of r 3 from this distribution. These values can then be used to calculate the numberand weight-average volumes of the silica nucleus, which are converted to the corresponding molecular weights with the aid of the partial specific volume of the silica, 0.45 g/ml (3). To this value should be added the weight of the PVP coat. We can assume that this is 10% of the total weight allowing for 1.5% free PVP in the Percoll solution. The calculated molecular weight values are tabulated in Table III. The weight-average molecular weight calculated for Percoll is about 14% lower than that obtained by light scattering but it must still be regarded as a very good agreement in view of the fact that it corresponds to only a 5% difference in the expected radius. The error in the electron microscopy is probably at least of this order. It is also apparent from the calculation that the colloid is rather homogenous and that the molecular weights obtained by osmometry must be affected by some other component than the electron microscopically visible colloidal particles.
Scattering at Low Ionic Strength Colloidal silica has been employed for calibration of light-scattering photometers for many years. Earlier authors have reported erratic results when they used Ludox
T A B L E III Mol e c ul a r Weights for Percoll C a l c ul a t e d from E l e c t r o n M i c r o s c o p y Radius (nm) of silica nucleus Number-average Weight-average
9.22 9.65
Molecular weight of silica 4.4 × 106 5.0 × 106
Molecular weight of Percoll 4.9 × 106 5.6 × 106
N o t e . F or calculations see text.
diluted in distilled water instead of dilute salt (11, 12). Their turbidity functions for Ludox in water were very similar to those shown in Fig. 1. They reported difficulties in the extrapolations to zero concentrations. De~elic and Kratohvil (12) discussed the results in terms of an increasing second virial coefficient for Ludox with decreasing salt concentration. Light scattering allows measurements at much lower concentrations which facilitates the extrapolation procedure (Fig. 2) and an accurate determination of the molecular weight. The light scattering revealed, however, both a strong concentration dependence of Kc/Ro and a considerable angular dependence. Dissymmetries below unity were recorded in an intermediate concentration interval (Fig. 3). Such a behavior has been described by Doty and Steiner (13, 14) for serum albumin in salt-free solution and should be expected from the electrostatic repulsions between the macroions. The angular dependence of the scattering is due to external interference arising from a nonrandom distribution of particles repelling each other. The strong initial increase in Kc/Rgo (Fig. 2) is due to the large effective volume of a charged particle. The leveling off of the Kc/Rgo versus c function (Fig. 2) at higher values of c is due to a decrease in the effective volume with increasing concentration of counterions. A further discussion of the particle interaction in Percoll can be found in the following paper (3). Journal of Colloid and Interface Science, Vol. 76, No. 1, July 1980
132
LAURENT, PERTOFT, AND NORDLI Stability o f P e r c o l l
ACKNOWLEDGMENTS
Experiments on L u d o x HS indicate that This project was supported by grants from the this colloid is unstable at physiological -Swedish Medical Research Council (13x-4) and ionic conditions and forms aggregates. Pharmacia Fine Chemicals AB. We thank Dr. L. K~tgedal for valuable discussions. This is in accordance with earlier investigations (6) which have pointed to the aggregating effects of electrolytes in the REFERENCES neutral region. Percoll seems much more stable and there were no signs of aggrega1. Pertoft, H., Laurent, T. C., L~t~s, T., and tion at high concentrations. At dilute conK~gedal, L., Anal. Biochem. 88, 271 (1978). 2. Pertoft, H., and Laurent, T. C., in "Methods centrations in the p H range 6 - 7 there of Cell Separation" (N. Catsimpoolas, Ed.), was, on the other hand, a slow aggregaVol. 1, p. 25. Plenum, New York, 1977. tion of the colloid which could be pre3. Laurent, T. C., Ogston, A. G., Pertoft, H., and vented by addition of free PVP. It is posCarlsson, B., J. Colloid Interface Sci. 76, sible that the low concentration of free PVP 133 (1980). 4. Laurent, T. C., and Pertoft, H., J. Colloid in the solution is required to keep the Interface Sci. 76, 142 (1980). surface of the colloidal particles saturated 5. Iler, R. K., "The Colloid Chemistry of Silica with a PVP layer. At high dilutions when and Silicates." Cornell Univ. Press, Ithaca, the free PVP concentration falls below a N. Y., 1955. critical value, the PVP on the silica sur6. Iler, R. K., in "Surface and Colloid Science" (E. Matijevi6, Ed.), Vol. 6, p. 1. Wile~, New York, face starts to dissociate off. This creates 1973. a possibility for polymer molecules to act 7. Iler, R. K., in "Biochemistry of Silicon and Reas cross-bridges between silica particles lated Problems" (G. Bendz and I. Lindqvist, and cause aggregation. The flocculation of Eds.), p. 53. Plenum, New York, 1977. 8. Jennings, B. R., and Jerrard, H. G., J. Polym. silica with polymers is a well-known Sci. Part A 2, 2025 (1964). p h e n o m e n o n (6, 7). The stabilizing ef9. Tanford, C., "Physical Chemistry of Macromolefects of low concentrations of PVP on Percules." Wiley, New York, 1961. coil may be of practical importance, e.g., "10. Zimm, B. H., J. Chem. Phys. 16, 1099 (1948). in biological work. When Percoll is washed 11. Goring, D. A. R., Senez, M., Melanson, B., and Huque, M. M., J. Colloid Sci. 12, 412 (1957). from a biological sample its concentration becomes very dilute. If the PVP dissociates 12. De~elic, G., and Kratohvil, J. B., J. Phys. Chem. 66~ 1377 (1962). from the silica the colloid will adsorb more 13. Doty, P., and Steiner, R. F., J. Chem. Phys. 17, easily to biological membranes. Removal of 743 (1949). Percoll from biological specimens m a y b e s t 14. Doty, t¢., and Steiner, R. F., J. Chem. Phys. 20, be accomplished with dilute PVP solutions. 85 (1951).
Journal of Colloid and Interface Science, Vol. 76, No. 1, July 1980
at low ionic strength. The present report deals with the particle weight of Percoll and subsequent papers discuss its particle size and interaction (3) and its sodium binding (4). Extensive studies have been performed on pure silica colloids. Iler has published several reviews on the subject (5-7). We have for comparison performed measurements in parallel on Percoll and Ludox HS, the parent silica colloid of Percoll.
INTRODUCTION
Percoll is a silica colloid in which the particles have been covered by a layer of polyvinylpyrrolidone (PVP). The PVP content is about 11-12% of the dry weight (1). The coating stabilizes the colloid and makes it nontoxic. Percoll can therefore be used as a density gradient medium for the separation of cells and other biological particles (2). A few physical chemical parameters of the colloid have been reported (1). The particles are close to spherical and have a mean diameter in the electron microscope of 17.3 nm. We now report a more extensive characterization of the material. Percoll is used for practical purposes both at physiological ionic strength and in salt-free media (2) and the measurements presented here have therefore been carried out at both conditions in spite of the theoretical difficulties in interpreting the data obtained
MATERIALS AND METHODS
Materials. Percoll (Batch 96544D) was kindly supplied by Pharmacia Fine Chemicals, Uppsala, Sweden. The colloid dissolved in water has a concentration of 26 g/100 ml. A pure silica colloid, Ludox HS, which is the starting material in the production of Percoll, was obtained from Du Pont de Nemours, Wilmington, Delaware. It has a concentration of about 50 g/100
124 0021-9797/80/070124-09502.00/0 Copyright© 1980by AcademicPress, Inc. All rightsof reproductionin any formreserved.
Journalof Colloidand Interface Science, Vol. 76, No. 1, July 1980
125
P H Y S I C A L C H E M I C A L C H A R A C T E R I Z A T I O N OF P E R C O L L TABLE I The Specific Refractive Index I n c r e m e n t at 546 n m for Percoll and L u d o x HS Refractive index increment (ml/g) Solvent
Percoll
Ludox HS
0.15 M NaC1 H~O
0.071 -+ 0.002 0.073 _+ 0.001
0.061 - 0.0003
ml. The colloids were adjusted with hydrochloric acid to the desired pH (usually pH 7.4) and dialyzed for at least 48 hr against repeated changes of distilled water. They were then, when appropriate, dialyzed an equal time against 0.15 M sodium chloride. A concentrated sample of Percoll (0.55 g/ml) was recovered from the bottom of a sedimentation equilibrium run (3). Polyvinylpyrrolidone (PVP) (Plasdone C-15; MW 10,000) was obtained from GAF Corporation, New York, New York. Determination of concentration. The concentration of stock solutions was determined by dry-weight measurements. The solutions and corresponding solvents were dried overnight at 95°C in vacuo and the concentration was taken as the difference of weights of dry residues from solution and solvent. Refractive index increment. A Rayleigh interference refractometer (Hilger & Watts, Ltd., London, England) was used to determine the refractive index increment at 546 nm. The temperature was 20.3°C and 3-4 determinations were made in the concentration interval 7-36 mg/ml for each solute/solvent system. Turbidity. The optical density was measured at room temperature in 5-cm glass cuvettes in a Zeiss spectrophotometer. The turbidity (~-) is, when light absorption can be neglected, defined as 2.303 "optical density/path length. All colloid solutions, obtained by dilution of stock solutions with the appropriate solvents, were cleared from dust by filtration through Millipore filters (Type GS, pore diameter 0.22 /zm) placed
in Swinnex-25 filter holders (Millipore Corp., Bedford, Mass.). The presence of light absorption in 0.1 g/ml Percoll solutions was checked by plotting the optical density at wavelengths between 400 and 800 nm versus t h e r e c i p r o c a l of the fourth power of the wavelength (see below). This gave a linear relationship at wavelengths above 500 nm. There was a significant deviation from linearity at lower wavelengths due to the absorbance by PVP. Further measurements were performed at 546 nm. Light scattering. Light scattering was measured at 546 nm in cylindrical cells in a Sofica PGD 42 000 M photometer (Sofica, Le Mesnil-Saint Denise, France) at 20 +_ 0.1°C. All solutions were freed from dust as described above. Most measurements were performed at an angle of 90° with respect to the direction of the incident beam. When necessary the angular de10 O
6 x
,
I 2
I
0.05
I
0.10
CONCENTRATION(g/mr) FIG. 1. Turbidity m e a s u r e m e n t s performed on Percoll in 0.15 M NaCI (O) and in distilled water (©) and on L u d o x HS in 0.15 M NaC1 (A) and in water (A). T h e refractive i n c r e m e n t u s e d to calculate the c o n s t a n t H (see text) for L u d o x HS in 0.15 M NaC1 was t h e s a m e as that m e a s u r e d in water (Table I). Journal of Colloid and Interface Science, Vol. 76, No. 1, July 1980
126
LAURENT, PERTOFT, AND NORDLI TABLE II Molecular Weights and Viral Coefficients for PercoU and Ludox HS Molecular weight x 10-6 Sample
Method
Mw
0.15 M NaC1 0.15 M NaC1 0.15 M N a C 1
Turbidity Light scattering Osmometry
6.6 6.4
Percoll from sedimentation equilibrium
0.15 M NaC1
Osmometry
Percoll
H~O H20
Turbidity Light scattering
0.15 M NaCI 0.15 M NaC1 H20 H20
Turbidity Light scattering Turbidity Light scattering
Percoll
Ludox HS
Solvent
M~
A2 (mole g-2 cm~ × 106)
0.6
1.3 1.8
2.3
1.8
6 (?) 6.4
530
20a 23a 6.5 6.5
a These values should not be regarded as definite molecular weights but as a sign of a progressive aggregation of the Ludox HS particles. p e n d e n c e of the scattering was m e a s u r e d between 30 and 150 ° and the values extrapolated to 0 and 180 °. All readings were corrected for solvent scattering, polarization, and the g e o m e t r y of the scattering volume. The results h a v e been e x p r e s s e d in terms of the Rayleigh ratios, Ro (=io'r2/ I0), where 0 is the angle of observation, io the intensity of scattered light, I0 the intensity of the incident b e a m , and r the distance f r o m the cell to the recorder. Osmometry. Readings were m a d e in an electronic high-speed m e m b r a n e o s m o m e t e r ( H e w l e t t - P a c k a r d Model 501, F & M Scientific Corp., Avondale, Penn.) at 30°C. Dialysis tubing (Union Carbide Co., Chicago, Ill.) was used as m e m b r a n e . The solutions were dialyzed for at least 48 hr against the solvent before filling of the o s m o m e t e r . Each sample of a specified concentration was measured two to four times. The readings were recorded and equilibrium was always reached within 30 rain. In the calculation of i~he ~osmotic pressure f r o m column height~'~ppropriate corrections were made for the densities of the colloid solutions. Journal of Colloid and Interface Science, Vol, 76, No. 1, July 1980
RESULTS
Specific Refractive Index Increment The refractive index increment for Percoll was determined in distilled water and in 0.15 M sodium chloride and for L u d o x HS in water. The results are given in Table I. The value for L u d o x H S in w a t e r is in good a g r e e m e n t with the value of 0.062 ml/g given by Jennings and Jerrard (8).
Turbidity According to theory (see, e.g., Tanford (9)) the turbidity (T) is related to the weight-average molecular weight (Mw) by the equation
HC/T
=
1/Mw + 2A~c + . . . ,
[1]
where c is the concentration of solute (g/ml), A~ is the second virial coefficient, and H is a constant of the f o r m
H=
3 2 zt3n2(dn/dc )2 3Nh 4
where n is the refractive index of the medium (1.34), dn/dc the refractive index
127
PHYSICAL CHEMICAL CHARACTERIZATION OF PERCOLL 10
0 0 0
0
0
0
x
0 0
The turbidity of Ludox HS in distilled water did not show the same extreme concentration dependence as Percoll in water and the molecular weight obtained by extrapolation agreed with that of Percoll. Ludox HS measured in 0,15 M NaC1 displayed a much higher molecular weight. The pure silica colloid did apparently aggregate at the higher ionic strength. The molecular parameters obtained in the measurements are displayed in Table II. Light Scattering
i
I
5 10 CONCENTRATION ( g / m i x l O 3)
I
15
FIG. 2. Light-scattering measurements performed on Percoll in 0.15 M NaC1 (O), 0.15 M NaC1 + 10-3 g/ml of polyvinylpyrrolidone (11), and water (©) and on Ludox HS in water (A).
increment of the solute (see Table I), N is Avogadro's number (6.02 × 1023), and h is the wavelength of the light (5.46 x 10-~ cm). (It follows from the equation that the turbidity should be proportional to 1/h4. The deviation of the optical density for Percoll from this proportionality below 500 nm is a sign of light absorption (see Materials and Methods). The turbidity of Percoll and Ludox HS was measured in distilled water and in 0.15 M NaC1 in the concentration range 0.01-0.12 g/ml and the results are plotted in Fig. 1. The molecular weight of Percoll in 0.15 M NaCI calculated from the intercept on the ordinate is 6.6 x 106. The experimental points obtained in distilled water cannot be extrapolated with any certainty but are consistent with this value. There is apparently a strong increase in the interaction between the Percoll particles when the salt is removed. The limiting slope of the curves, which gives the second virial coefficient and which is a measure of interaction, is manyfold larger in water than in 0.15 M salt.
As light scattering and turbidity are related phenomena they will give the same information. Light scattering is, however, measured at a higher dilution as compared to those required in the turbidity measurements. When colloidal particles are small compared to the wavelength of the light the following relationship holds (9), [2]
Kc/Ro = 1/Mw + 2A2c " ' ,
where Ro is the Rayleigh ratio usually measured perpendicular to the incident light beam (Rg0) and K is a constant with the value K = 27r2n2(dn/dc)2 Nh 4
In particles of larger dimensions internal
1.0
O
.9
m~
.7
tx .5
I 0.005
I 0.01
i 0.01
CONCENTRATION [g/ml)
FIG. 3. The dissymmetry of the light scattering (R45/R135) displayed by Percoll (©) and Ludox HS (A) in water as a function of concentration. Journal of Colloid and Interface Science, V o l . 76, N o . 1, J u l y 1980
128
LAURENT, PERTOFT, AND NORDLI
interference causes Ro to vary with the angle and necessitates measurements of Ro over a large interval and extrapolation to zero angle (R0) for a correct determination of molecular weight. Measurements in 0.15 M NaCl. These were performed on Percoll in the interval 0.001 to 0.01 g/ml. There was no angular dependence of Ro and the molecular weight obtained from the values given in Fig. 2 was 6.4 x 10°. The measurements were performed both in the presence and absence of 0.1% PVP with identical results. The addition of PVP was due to an observation described below that PVP prevents aggregation of Percoll in very dilute solutions. The value of A2 obtained from the slope of the line is given in Table II. Measurements on Ludox HS in 0.15 M NaC1 revealed an angular dependence of Ro and the solution did apparently contain large aggregates. Extrapolation of the
5
t~
3
o
x
x
x
x
x t.a
">IQ:
2
1
0
I
I
I
5
t0
15
CONCENTRATION (g/mix10 ¢) FIG, 4. Extrapolation to zero concentration of
Kc/Ro (©), Kc/R9o (x), and Kc/R18o(0) for Percoll in water. The angular dependence of the light scattering disappears at infinite dilution and the extrapolation corresponds to a molecular weight of 6.4 x 106. Note that Fig. 4 displays a 10-fold lower concentration range than Fig. 3. Journal o f Colloid and Interface Science, V o l . 76, N o . 1, J u l y 1980
1.5
1.0
X
,vl,0.S a
o
~
ols Sin 2 0/2
FIo. 5. The change in angular light-scattering pattern at 20°C of 10-4 g/ml of the Percoll colloid in 0.15 M NaC1, pH 6.22. The recordings were made 3 (Q), 24 (O), 48 (A), and 240 (A) hr after dilution from a 0.24 g/ml solution. The decrease in Kc/Ro especially at low angles shows a time-dependent aggregation. The solution was kept in the same measuring cell during the whole time period.
experimental data to zero angle and zero concentration according to Zimm (10) gave an approximate molecular weight of 23 x 106. Measurements in water. Light scattering of Percoll and Ludox HS in water showed an unusual angular dependence. The dissymmetry, i.e., the ratio R45/R135, became less than unity within an intermediate concentration interval (Fig. 3). The angular dependence of the scattering disappears at infinite dilution and the extrapolation of Kc/R9o should give the correct molecular weight. Such extrapolations are shown in Fig. 2 for both Percoll and Ludox HS and both give molecular weights of the same order as the measurements of Percoll at high ionic strength (Table II). That the angular dependence disappears at zero concentration is further shown in Fig. 4 where R0 and R180, obtained by extrapolation from measurements at different angles, were used for plots of Kc/Ro and Kc/Ra8o. The slope of the Kc/Ro plot in Fig. 4 was used to obtain a value
PHYSICAL CHEMICAL CHARACTERIZATIONOF PERCOLL
I
.
^A
Z~
Osmometry
A
The osmotic pressure of Percoll solutions, extensively dialyzed against 0.15 M NaC1 or water, is shown in Fig. 7. A highly concentrated Percoll fraction recovered from the bottom of a sedimentation equilibrium run (3) was also dialyzed against 0.15 M NaC1 and measured at different dilutions. This fraction displayed a considerably lower osmotic pressure than the original Percoll solution (Fig. 7). The osmotic pressure (Tr) is related to the molecular weight by the equation (9)
0.5
0
129
I
,I
100
I
I
I
rc/RTc = 1/Mn + A2c + " " ,
200
Hours after dilution
FIG. 6. The change with time of Kc/Rgo in Percoll solutions containing 0.15M NaC1 with ( - - - ) and without ( ) the presence of 5 x 10-4 g/ml of PVP. The colloid concentrations were 2 x 10-4 (O), 4 x 10-4 (A), 6 x 10-4 (O) and 8 x 10 4(A) g/ml. The pH values in the different Percoll solutions were 6.2, 6.4, 6.6, and 6.8, respectively. of Az for Percoll in distilled water. It is given in Table II. M e a s u r e m e n t s on dilute Percoll solutions in salt. The Percoll solutions demonstrated an instability at high dilutions in 0.15 M sodium chloride and p H 6 to 7. This was evident from a time-dependent decrease in the Kc/Ro, especially at low angles, indicating an aggregation (Fig. 5). This effect disappeared at higher concentrations of the colloid and at 0.01 g/ml no aggregation could be detected. As a working hypothesis it was assumed that the PVP coat started to dissociate from the Percoll surface at high dilutions. It should be possible to prevent this by adding PVP and Fig. 6 demonstrates that PVP does prevent aggregation. As a matter of fact the addition of PVP seemed to cause a small initial deaggregation of the Percoll sample. H o w e v e r , also in the samples containing PVP there was eventually a certain formation of aggregates noticeable from the scattering at low angles.
[3]
where R is the gas constant, T the absolute temperature, and Mn the numberaverage molecular weight. A plot of 7r/RTc versus concentration is given in Fig. 8 for the Percoll solutions in 0.15 M NaC1. The intercepts correspond to number-average molecular weights of 6 x l0 s and 2.3 x 106, respectively. The slope of the lines and thus the value of Az is the same for
1.5
== 1.0 =
N
11.5
0
0.1 0.Z 0.3 CONCENTRATION (g/at)
0.4
FIG. 7. The osmotic pressure of Percoll solutions. The measurements were made in distilled water (O) and 0.15 M NaC1 (0). A fraction of Percoll was also recovered from the bottom of a sedimentation equilibrium run performed in 0.15 M NaC1 (3) and the osmotic pressure analyzed in the same medium (A). Journal of Colloid and Interface Science, V o l . 76~ N o . 1, J u l y 1980
130
LAURENT, PERTOFT, AND NORDLI
~a
I 0.1
i 0.2
I 0.3
I O.t,
EONEENTRATION (g/mL)
FIG. 8. The reduced osmotic pressure of Percoll measured in 0.15 M NaC1 (0) and of the Percoll fraction obtained from the bottom of a sedimentation equilibrium run (A). The intercepts on the ordinate correspond to molecular weights of 6 x l0s and 2.3 × 108,respectively. both samples (Table II). The osmometric data obtained in distilled water have not been used to calculate any molecular parameters. The experimental data in Fig. 7 are given because of the practical interest to know the osmotic pressure of Percoll in salt-free solutions (2). DISCUSSION
Molecular Weight Both turbidimetry and light-scattering measurements gave weight-average molecular weights of Percoll o f about 6.5 x 106 irrespective of ionic strength. The same value for L u d o x HS at low ionic strength confirms that this is a correct estimate. One should expect a 10% higher value for Percoll than for L u d o x HS due to the PVP coat. That this is not apparent in the measurements is most probably due to variation in the different batches of the colloids obtained from the manufacturers. Our results for L u d o x HS are in general agreement with numerous reports (for references see Jennings and Jerrard (8)). The number-average molecular weight of Percoll as measured by osmometry is Journal of Colloid and Interface Science, Vol. 76, No. 1, July 1980
considerably lower than the weight-average. This could of course be explained by a large polydispersity of the colloid and such an explanation is in agreement with the observation that a fraction recovered from the bottom layer of a sedimentation equilibrium run has a much higher Mn value. T h e r e is, however, an alternative explanation which is equally plausible. The Percoll may contain small amounts of free PVP. If Percoll has a molecular weight of 6.5 x 106 and if we assume a molecular weight of 104 for PVP then free PVP to an amount of 1.5% of the total dry weight would lead to a number-average molecular weight of 6 x l0 s. In the sedimentation equilibrium runs described in the following paper (3) the distribution of free PVP would hardly be affected by gravity while the concentration of Percoll in the bottom fraction (g Percoll/g H20) is about three times higher than in the original Percoll solution. With a relative concentration of free PVP which is only a third of that assumed above for the original solution one can calculate an Mn value in the order of 2 x 100. That the turbidimetric and light-scattering techniques give identical molecular weights for Percoll in 0.15 M sodium chloride after reliable extrapolations makes the simple turbidimetric technique attractive for routine work on Percoll in spite of the fact that the turbidimetric measurements must be performed at concentrations one or two magnitudes higher than the lightscattering studies.
Comparison with Electron Microscopy The distribution of particle diameters in Percoll was reported from electron microscopy to be 17.3 + 2.3 nm (1). Nearly the same value was found for L u d o x HS. A negative contrast technique with uranyl acetate was employed in this investigation to visualize the particles. We may use the data with certain assumptions to calculate a molecular weight for the Percoll par-
131
P H Y S I C A L C H E M I C A L C H A R A C T E R I Z A T I O N OF P E R C O L L
ticles. For example, we may assume that the uranyl acetate penetrates the PVP coat and that the electron microscopic pictures essentially show the silica nucleus. This is reasonable in view of the close to identical diameters estimated for Percoll and Ludox HS. The radius (r) reported is a numerical average. The volume and the molecular weight are proportional to the third power of the radius and we have therefore calculated the frequency distribution of r 3 from the data presented in Ref. (1). For comparison of the results with those from osmometry and light scattering we must obtain the number- and weight-averages of r 3 from this distribution. These values can then be used to calculate the numberand weight-average volumes of the silica nucleus, which are converted to the corresponding molecular weights with the aid of the partial specific volume of the silica, 0.45 g/ml (3). To this value should be added the weight of the PVP coat. We can assume that this is 10% of the total weight allowing for 1.5% free PVP in the Percoll solution. The calculated molecular weight values are tabulated in Table III. The weight-average molecular weight calculated for Percoll is about 14% lower than that obtained by light scattering but it must still be regarded as a very good agreement in view of the fact that it corresponds to only a 5% difference in the expected radius. The error in the electron microscopy is probably at least of this order. It is also apparent from the calculation that the colloid is rather homogenous and that the molecular weights obtained by osmometry must be affected by some other component than the electron microscopically visible colloidal particles.
Scattering at Low Ionic Strength Colloidal silica has been employed for calibration of light-scattering photometers for many years. Earlier authors have reported erratic results when they used Ludox
T A B L E III Mol e c ul a r Weights for Percoll C a l c ul a t e d from E l e c t r o n M i c r o s c o p y Radius (nm) of silica nucleus Number-average Weight-average
9.22 9.65
Molecular weight of silica 4.4 × 106 5.0 × 106
Molecular weight of Percoll 4.9 × 106 5.6 × 106
N o t e . F or calculations see text.
diluted in distilled water instead of dilute salt (11, 12). Their turbidity functions for Ludox in water were very similar to those shown in Fig. 1. They reported difficulties in the extrapolations to zero concentrations. De~elic and Kratohvil (12) discussed the results in terms of an increasing second virial coefficient for Ludox with decreasing salt concentration. Light scattering allows measurements at much lower concentrations which facilitates the extrapolation procedure (Fig. 2) and an accurate determination of the molecular weight. The light scattering revealed, however, both a strong concentration dependence of Kc/Ro and a considerable angular dependence. Dissymmetries below unity were recorded in an intermediate concentration interval (Fig. 3). Such a behavior has been described by Doty and Steiner (13, 14) for serum albumin in salt-free solution and should be expected from the electrostatic repulsions between the macroions. The angular dependence of the scattering is due to external interference arising from a nonrandom distribution of particles repelling each other. The strong initial increase in Kc/Rgo (Fig. 2) is due to the large effective volume of a charged particle. The leveling off of the Kc/Rgo versus c function (Fig. 2) at higher values of c is due to a decrease in the effective volume with increasing concentration of counterions. A further discussion of the particle interaction in Percoll can be found in the following paper (3). Journal of Colloid and Interface Science, Vol. 76, No. 1, July 1980
132
LAURENT, PERTOFT, AND NORDLI Stability o f P e r c o l l
ACKNOWLEDGMENTS
Experiments on L u d o x HS indicate that This project was supported by grants from the this colloid is unstable at physiological -Swedish Medical Research Council (13x-4) and ionic conditions and forms aggregates. Pharmacia Fine Chemicals AB. We thank Dr. L. K~tgedal for valuable discussions. This is in accordance with earlier investigations (6) which have pointed to the aggregating effects of electrolytes in the REFERENCES neutral region. Percoll seems much more stable and there were no signs of aggrega1. Pertoft, H., Laurent, T. C., L~t~s, T., and tion at high concentrations. At dilute conK~gedal, L., Anal. Biochem. 88, 271 (1978). 2. Pertoft, H., and Laurent, T. C., in "Methods centrations in the p H range 6 - 7 there of Cell Separation" (N. Catsimpoolas, Ed.), was, on the other hand, a slow aggregaVol. 1, p. 25. Plenum, New York, 1977. tion of the colloid which could be pre3. Laurent, T. C., Ogston, A. G., Pertoft, H., and vented by addition of free PVP. It is posCarlsson, B., J. Colloid Interface Sci. 76, sible that the low concentration of free PVP 133 (1980). 4. Laurent, T. C., and Pertoft, H., J. Colloid in the solution is required to keep the Interface Sci. 76, 142 (1980). surface of the colloidal particles saturated 5. Iler, R. K., "The Colloid Chemistry of Silica with a PVP layer. At high dilutions when and Silicates." Cornell Univ. Press, Ithaca, the free PVP concentration falls below a N. Y., 1955. critical value, the PVP on the silica sur6. Iler, R. K., in "Surface and Colloid Science" (E. Matijevi6, Ed.), Vol. 6, p. 1. Wile~, New York, face starts to dissociate off. This creates 1973. a possibility for polymer molecules to act 7. Iler, R. K., in "Biochemistry of Silicon and Reas cross-bridges between silica particles lated Problems" (G. Bendz and I. Lindqvist, and cause aggregation. The flocculation of Eds.), p. 53. Plenum, New York, 1977. 8. Jennings, B. R., and Jerrard, H. G., J. Polym. silica with polymers is a well-known Sci. Part A 2, 2025 (1964). p h e n o m e n o n (6, 7). The stabilizing ef9. Tanford, C., "Physical Chemistry of Macromolefects of low concentrations of PVP on Percules." Wiley, New York, 1961. coil may be of practical importance, e.g., "10. Zimm, B. H., J. Chem. Phys. 16, 1099 (1948). in biological work. When Percoll is washed 11. Goring, D. A. R., Senez, M., Melanson, B., and Huque, M. M., J. Colloid Sci. 12, 412 (1957). from a biological sample its concentration becomes very dilute. If the PVP dissociates 12. De~elic, G., and Kratohvil, J. B., J. Phys. Chem. 66~ 1377 (1962). from the silica the colloid will adsorb more 13. Doty, P., and Steiner, R. F., J. Chem. Phys. 17, easily to biological membranes. Removal of 743 (1949). Percoll from biological specimens m a y b e s t 14. Doty, t¢., and Steiner, R. F., J. Chem. Phys. 20, be accomplished with dilute PVP solutions. 85 (1951).
Journal of Colloid and Interface Science, Vol. 76, No. 1, July 1980