Journal of Chromatography A, 1028 (2004) 189–195

Modeling ion-exchange adsorption of proteins in a spherical particle

Stuart R. Gallant∗
Cell Genesys Inc., 500 Forbes Boulevard, South San Francisco, CA 94080, USA

Received 13 October 2003; received in revised form 12 December 2003; accepted 15 December 2003

Abstract

This paper presents a simple model of single- and multicomponent protein adsorption in a spherical particle. The model includes radial
diffusion of salt and protein in the liquid phase coupled to adsorption by an ion-exchange mechanism described by the steric mass action
isotherm. The molecular diffusivities of the protein and salt are reduced in the model by a factor which accounts for the tortuous nature of
the pores and pore constrictions. The model parameters are selected from published values in the chromatographic literature. Of particular
interest are the observations of induced salt gradients during protein adsorption and of multicomponent displacement when more that one
protein is adsorbed simultaneously. These results cannot be predicted on the basis of the traditional Langmuir isotherm or other currently
available descriptions of adsorption. The use of such a model during stationary phase design is discussed.
© 2004 Elsevier B.V. All rights reserved.

Keywords: Adsorption isotherms; Steric mass action; Mathematical modeling; Ion-exchange adsorption; Stationary phases, LC; Confocal microscopy; Proteins

1. Introduction throughout individual beads. The rate and pattern of adsorp-

tion maybe studied as a function of adsorption conditions
In preparative or large-scale purification of proteins, a (protein concentration, pH, ion strength, and other variables)
protein solution is contacted with a chromatographic column and as a function of time by repeatedly sampling beads from
under conditions in which adsorption to the stationary phase the batch adsorption vessel. A number of groups have done
is strong and essentially irreversible. Loading is continued elegant work examining these phenomena [1–4].
until the stationary phase is close to its maximum binding Several important physical phenomena dictate the behav-
capacity for the product. Subsequently, the stationary phase ior of proteins during adsorption in ion-exchange. The first
is washed and the product is eluted. is equilibrium adsorption capacity of the resin for the pro-
The process of adsorption in ion-exchange chromatogra- teins present in the load. This data may be represented by
phy is a multi-step process in which proteins diffuse into the a multicomponent isotherm [5–7] which attempts to fit the
pores of the stationary phase and exchange with ions already physical data with a mathematical expression and mimic the
present in the pores. Displacement of low affinity proteins dependence of the bound protein capacity on such variables
by high affinity proteins may occur, and multiple waves of as the protein concentration and salt concentration. The sec-
proteins moving into the stationary phase and ions moving ond important set of phenomena controlling adsorption are
out of the stationary phase are to be expected. mass transport relationships which control the rate of dif-
Some of these phenomena can be observed experimen- fusion of proteins in free solution and inside the chromato-
tally. Confocal scanning laser microscopy has been used for graphic bead [8–10]. Until recently, mass transport has been
a number of years to study the time-dependent course of envisioned primarily as a phenomena limited by the tortuous
adsorption of proteins in chromatographic beads [1,2]. In nature of the internal pores of the chromatographic beads
this technique, protein adsorption is studied by batch incu- and by pore constrictions. Recent research has also shown
bation of the beads with varying amounts of fluorescently that the rate of internal diffusion with the porous bead can
labeled protein followed by observation of the fluorescence be enhanced by an electrokinetic contribution to transport.
In this paper, a relatively simple model of protein adsorp-
tion will be employed. This model relates the equilibrium
∗ Tel.: +1-650-266-2938; fax: +1-650-266-2910. binding capacities of multiple proteins and a single counter-
E-mail address: [email protected] (S.R. Gallant). ion to the composition of the liquid in the stationary phase

0021-9673/$ – see front matter © 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.chroma.2003.12.042
190 S.R. Gallant / J. Chromatogr. A 1028 (2004) 189–195

pores. Transport within the pores is modeled by simple dif- system in which induced salt gradients play a particularly
fusion of the adsorbed species adjusted for the tortuosity of important role. In addition, SMA has been used in models
the pores. Though the transport model is quite simple, the of preparative ion-exchange separations [21–23].
predictions of the model are realistic and represent phenom-
ena not show previously in other communications within the 2.2. Field equations and boundary conditions
literature. Of particular, interest is the observation of dis-
placement of low affinity protein and salt components by a Transport within the chromatographic bead is described
higher-affinity protein. by the following system of partial differential equations:


∂Ci 1 ∂Qi Di 1 ∂ 2 ∂Ci
+ = r ,
2. Theory ∂t β ∂t θ r 2 ∂r ∂r
i = 1, . . . , NC (5)
2.1. Equilibrium adsorption
where β is a phase ratio, Di a molecular diffusion coeffi-
cient, θ a tortuosity factor, r the radial position, and t the
Ion-exchange can be modeled using a stoichiometric ap-
time. For a non-adsorbing system, these equations reduce
proach in which adsorption of a quantity of charged species
to the equations for diffusion in a sphere [24]. Under con-
by the stationary phase requires release of an equal quantity
ditions in which proteins are adsorbed, the multicomponent
of charge of another species. This approach has been em-
isotherm defines the relationship between the bound protein
ployed successfully by a number of researchers [11–17]. One
concentration Qi and the concentration of protein within the
example of this approach is the steric mass action (SMA)
pores Ci .
formalism, a three parameter model of ion-exchange which
The boundary conditions are given by the feed concen-
predicts the multicomponent adsorption of protein under di-
trations of protein and salt:
lute and concentrated conditions based on single-component
adsorption data [7]. Ci (t, R) = Ci,f (6)
SMA represents the adsorption process as a stoichiometric
where R refers to the particle radius. The initial conditions
exchange of mobile phase protein and bound counterions:
are:
Ci + νi Q̄1 ⇔ Qi + νi C1 , i = 2, . . . , NC (1) Ci (0, r) = C1,init (salt, i = 1) (7)
where Ci and C1 refer to the mobile phase concentration of Ci (0, r) = 0 (proteins, i > 1) (8)
protein and salt, Qi refers to bound protein concentration,
Q̄1 refers to the concentration of bound salt available for where the subscript 2 refers to the protein and τ f represents
exchange, νi stands for the proteins characteristic charge, the duration of the feed pulse in dimensionless units.
and NC refers to the number of components present in mo-
bile phase. The equilibrium constant of the reaction may be 2.3. Solving the model
written:



Qi C1 νi The model can be solved using the finite difference
K1i = , i = 2, . . . , NC (2) method [24]. The following finite difference equations were
Ci Q̄1
employed to approximate the diffusion term of the partial
Each protein molecule may sterically shield some salt coun- differential equations listed earlier:
terions on the adsorptive surface. The quantity of salt coun-

1 ∂ 2 ∂C 1
terions blocked by a particular protein will be proportional ρ = [(m + 1)Cm+1,n
to the concentration of that protein on the surface: ρ2 ∂ρ ∂ρ m(δρ)2
− (2m)Cm,n + (m − 1)Cm−1,n ] (9)
Q̄1i = σi Qi , i = 2, . . . , NC (3)

1 ∂ ∂C 6
Electroneutrality requires that: ρ2 = (Cm+1,n − Cm,n ), m = 0 (10)
ρ ∂ρ
2 ∂ρ (δρ)2

NC
where ρ is non-dimensional radius, m the index of radial el-
Λ = Q̄1 + (νi + σi )Qi (4)
ements in the finite difference grid, and n the index of time
i=2
elements in the finite difference grid. Using these approxi-
SMA has been shown to represent single-component mations, the solution can be obtained by marching across a
isotherms of proteins under varying salt concentration finite difference grid from the initial conditions forward in
[18,19]. Of particular importance is its ability to predict the time using the equation:
amount of salt released from the stationary phase during 

1 ∂ 2 ∂C
adsorption of multicomponent protein mixtures. This pre- Gn+1 = Gn + τ 2 ρ (11)
ρ ∂ρ ∂ρ
dictive ability made it possible for a model based on SMA
to predict displacement development [20], a separation where G = C + Q/β and τ is dimensionless time.
 S.R. Gallant / J. Chromatogr. A 1028 (2004) 189–195 191

Table 1
Parameters employed in simulation
Solute Characteristic charge (νi ) Equilibrium constant (K1i ) Steric factor (σ i ) Diffusion coefficient (Di )

␣-Chymotrypsinogen A 5.03 7.96 × 10−2 7.43 8.5 × 10−7

Cytochrome c 5.67 3.52 × 10−1 27.4 9.1 × 10−7
Particle diameter: 15 ␮m; stationary phase capacity (Λ): 525 mM; phase ratio (β): 1.74; tortuosity factor (θ): 2.3; NaCl diffusion coefficient (D1 ):
1.6 × 10−5 cm2 /s.

This scheme will be stable for a sufficiently dense grid in lower affinity protein, a series of simulations were car-
the time dimension and converges rapidly to solution with as ried out. The simulated adsorption conditions are given in
few as 10 spatial elements giving a reliable approximation. Table 2, including the salt concentration, protein concen-
The stability condition dictates that each doubling of radial tration(s), and the adsorption time (i.e., the length of the
grid density requires that the density of elements in the time simulated adsorption experiment).
dimension be increased by a factor of four [25]. The sim- In the first simulation, binding of a solution of cytochrome
ulations listed in Table 2 were carried out with 50 spacial c at a concentration of 0.2 mM with 30 mM salt concentra-
elements and sufficient elements in time to give numerical tion was simulated. These conditions are ones that promote
stability. The model was written and solved using Visual strong binding of cytochrome c. Under these conditions, the
Fortran Professional Edition 5.0.A running under Microsoft isotherm is quite “square” [19]. The expected equilibrium
Windows ME on an Intel Pentium II processor. bound concentration of cytochrome c is 13.6 mM. (For com-
parison, the maximum binding capacity of cytochrome c on
this stationary phase is 15.9.)
3. Model parameters In Fig. 1, a topographic plot of the unbound cytochrome
c in the pores of the chromatographic bead is given for the
The equilibrium adsorption parameters (Λ, β, νi , K1i , first set of simulation conditions (Table 2). The bottom of
and σ i in Table 1) for the proteins ␣-chymotrypsinogen the figure represents the beginning of the simulation (time =
A and cytochrome c were obtained on a 15 ␮m diameter 0). The chromatographic bead is equilibrated with 30 mM
strong cation-exchange resin (Amersham Biosciences) in a sodium, and from the inside of the bead (left side of fig-
previous publication [22]. The system was buffered at pH ure) to the outside of the bead, no protein is present in the
6.0 using sodium monobasic phosphate and sodium dibasic bead. Zero protein concentration is represented by dark blue;
phosphate, and the salt employed was sodium chloride. The 0.2 mM cytochrome c is represented by dark red. From the
characteristic charge (νi ) and equilibrium constant (K1i ) right-hand side of the figure (outside of the bead), the cy-
were obtained from linear elution data taken over a range of tochrome c diffuses toward the core. The velocity of the
salt concentrations. The steric factor (σ i ) was obtained from front of cytochrome c accelerates as it approaches the core,
breakthrough experiments carried out under concentrated and this can be seen by the change in slope of the isolines
protein conditions. of concentration (they bend over indicating an increase in
The diffusion coefficients of the salt and protein and the velocity). This is primarily because the volume of free sta-
tortuosity factor were estimated from published values in tionary phase decreases with the square of the radius. Under
the scientific literature [26–29]. these adsorption conditions, the center of the bead reaches
99% of the feed concentration (0.2 mM) at 10.5 s.
Fig. 2 gives a more quantitative picture of the unbound
4. Results cytochrome c in simulation 1. In this figure, the pore concen-
tration of cytochrome c versus radial position is plotted for
In order to understand the dynamics of the protein 0.0, 2.7, 5.7, 8.7, and 11.7 s of the simulation. Throughout
adsorption process and the interaction of the adsorbed the course of the simulation, it can be seen that the gradi-
higher-affinity protein with induced gradients of salt and ent of cytochrome c is such that cytochrome c concentration

Table 2
Simulation conditions
Simulation Na+ concentration (mM) Cytochrome c ␣-Chymotrypsinogen A Adsorption time (s)
concentration (mM) concentration (mM)
1 30 0.2 – 15
2 200 0.2 – 15
3 200 0.0002 0.2 1500
4 30 0.2 0.2 15
5 30 – 0.2 15
192 S.R. Gallant / J. Chromatogr. A 1028 (2004) 189–195

Fig. 1. Topographic plot of unbound cytochrome c concentration (simulation 1): 0.2 mM cytochrome c diffusion into a strong cation-exchange bead with
30 mM sodium phosphate, pH 6.0.

decreases toward the core of the bead. This allows a driving concentration of cytochrome c versus radial position is plot-
force to exist for continued diffusion of cytochrome c into ted for the same simulation times. In this figure, it can be
the core of the bead. seen that the advancing wave of protein saturates the sta-
Of course, the bound cytochrome c which represents the tionary phase completely before moving on resulting in a
majority of the protein in the bead. In Fig. 3, the bound more “square” advancing front than the would be expected
by simply examining the pore concentration of cytochrome
0.25 c. In fact, this is the profile observed in some confocal mi-
croscopy experiments which have been reported in the liter-
Concentration (mM)

0.20 ature.
0.15 As the cytochrome c is adsorbed, it desorbs an amount
of sodium sufficient to maintain electroneutrality on the
0.10 stationary phase surface. In Fig. 4, the induced salt gra-
0.05
Increasing Time dient caused by cytochrome c adsorption is seen. At 0.0 s
the entire bead has a sodium concentration of 30 mM. (In
0.00 the figure, this line overlays the 14.7 s line, and both of
0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0
which are horizonal with values of 30 mM over their entire
Radius (micron)
length.) As the cytochrome c is adsorbed, desorbed sodium
Fig. 2. Unbound cytochrome c concentration vs. radius (simulation 1):
concentration vs. radius lines are shown at 0.0, 2.7, 5.7, 8.7, and 11.7 s.
30.07
16 30.06
Concentration (mM)

14 30.05 Increasing Time

Concentration (mM)

12 30.04
10 30.03
8 30.02
6 30.01
Increasing Time
4 30.00
2 29.99
0 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0
0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 Radius (micron)
Radius (micron)
Fig. 4. Induced salt gradient due to cytochrome c adsorption (simulation
Fig. 3. Bound cytochrome c concentration vs. radius (simulation 1): 1): concentration vs. radius lines are shown at 2.7, 5.7, 8.7, 11. 7 and
concentration vs. radius lines are shown at 0.0, 2.7, 5.7, 8.7, and 11.7 s. 14.7 s (0.0 and 14.7 s actually overlie each other; see text).
 S.R. Gallant / J. Chromatogr. A 1028 (2004) 189–195 193

increases the sodium concentration throughout the bead dif- 14

fusing to the center of the bead and also diffusing out of the 12
Concentration (mM)

bead. A reverse gradient is quickly formed with the high- 10

est sodium concentration at the center of the bead drive 8
sodium out of the bead efficiently. At the highest value, the
6
induced gradient is only 30.064 mM (i.e., 0.064 mM above Increasing Time
4
the ambient sodium concentration prior to cytochrome c ad-
2
sorption). At the point that the cytochrome c is completely
adsorbed, the sodium returns to its original concentration 0
0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0
of 30 mM (the excess sodium having diffused out of the
Radius (micron)
bead).
In simulation 2 (results not shown), the sodium concen- Fig. 6. Bound cytochrome c concentration vs. radius in competitive ad-
tration was increased to 200 mM. At this concentration the sorption with ␣-chymotrypsinogen A (simulation 4): concentration vs.
radius lines are shown at 0.0, 2.7, 5.7, 8.7, and 11.7 s.
cytochrome c is not as strongly retained and will be eluted
from the column with a retention factor (k ) of 3.6. Since
the bound protein concentration is lower, the cytochrome Simulation 4 addresses the adsorption of two proteins
c can saturate the stationary phase much more rapidly. In ␣-chymotrypsinogen A and cytochrome c on a strong
this simulation, the wave of cytochrome c advances much cation-exchange bead. The protein concentrations are both
more rapidly toward the center of the bead. The unbound 0.2 mM and the sodium concentration is 30 mM. In Fig. 6,
cytochrome c concentration at the center of the bead reaches the bound cytochrome c concentration versus radius is
99% of 0.2 mM by 5.7 s. shown for the same times shown in Fig. 3 so that the cases
In simulation 3, the sodium concentration was held at of cytochrome c with and without ␣-chymotrypsinogen A
200 mM and the concentration of cytochrome c was reduce competition can be compared. What can be seen is that
to 0.0002 mM, a more typical concentration for a dilute feed- for the case of competitive adsorption, the cytochrome c
stream from cell culture supernatant. A similar pattern of advances into the bead somewhat more quickly, without the
advance of unbound cytochrome c in the pores of the bead pronounce square front of the single-component case. Fur-
is observed as was seen in simulations 1 and 2; however, ther, because the ␣-chymotrypsinogen A is not completely
the time required for the adsorption is substantially longer excluded from that stationary phase in the competitive ad-
than in the preceding simulations. Under these adsorption sorption case, the final concentration of bound cytochrome
conditions, 39 s is required for the unbound cytochrome c c is reduced slightly compared to the case of single protein
concentration at the center of the bead to reach 99% of feed adsorption.
concentration at the outside of the bead. (Recall, it only re- Fig. 7 gives the bound of ␣-chymotrypsinogen A concen-
quired 5.7 s the 99% point in simulation 2.) tration versus radius for the same times depicted in Fig. 6.
Fig. 5 shows the concentration of bound cytochrome c As can be seen in the figure, there is a strong displace-
in simulation 3 for 0.0, 6.0, 12.0, 18.0, 24.0, and 30.0 s. ment effect in which adsorption and then desorption of the
The contrast between Figs. 3 and 5 is marked. In Fig. 3, ␣-chymotrypsinogen A in the region before the advancing
the low salt concentration and high protein concentration cytochrome c front leads to a strong overconcentration of
drives an extremely square front of adsorbed protein. In the lower affinity protein ␣-chymotrypsinogen. And as the
contrast, in Fig. 5 with the low salt concentration and low adsorption continues, the cytochrome c largely excludes the
protein concentration, the protein adsorption occurs almost
everywhere in the bead simultaneously.
35

30
Concentration (mM)

0.018 Increasing Time

25
0.016
20
0.014
Concentration (mM)

0.012 15
0.010
10
0.008
Increasing Time 5
0.006
0.004 0
0.002 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0
0.000 Radius (micron)
0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0
Radius (micron)
Fig. 7. Bound ␣-chymotrypsinogen A concentration vs. radius in com-
petitive adsorption with cytochrome c (simulation 4): concentration vs.
Fig. 5. Bound cytochrome c concentration vs. radius (simulation 3): radius lines are shown at 0.0, 2.7, 5.7, 8.7, and 11.7 s. The 0.0 s line is
concentration vs. radius lines are shown at 0.0, 6.0, 12.0, 18.0, 24.0, and visible only as a small line at the bottom right of the graph; at this time,
30.0 s. the ␣-chymotrypsinogen A is just beginning to enter the bead.
194 S.R. Gallant / J. Chromatogr. A 1028 (2004) 189–195

Fig. 8. Topographic plot of unbound ␣-chymotrypsinogen A concentration (simulation 4): 0.2 mM ␣-chymotrypsinogen A and 0.2 mM cytochrome c
diffuse into a strong cation-exchange bead with 30 mM sodium phosphate, pH 6.0. ␣-Chymotrypsinogen A is seen to be displace to center of the bead
and finally excluded.

␣-chymotrypsinogen A from the bead, leading to a final salt actually diffuses into the bead. In order to understand
bound ␣-chymotrypsinogen A much lower than transient this phenomena, it is helpful to recall that the steric factor
wave of highly concentrated protein. of cytochrome c is larger than that of ␣-chymotrypsinogen
Interestingly, much of the ␣-chymotrypsinogen A has only A (see Table 1). What this means in practice is that, to
a transitory presence in the bead; it diffuses in and is ad- desorb ␣-chymotrypsinogen A, both cytochrome c and salt
sorbed, but subsequently, it is desorbed and diffuses back out are required. The cytochrome c drives the desorption, but
of the bead. This can best be observed in Fig. 8 where the sodium must diffuse to the surface to maintain electroneu-
unbound ␣-chymotrypsinogen A concentration is depicted trality. Only at the end of the adsorption process, when both
in topographic form. Up to about 6 s of time, the concentra- proteins have reached their equilibrium concentrations, does
tion of unbound ␣-chymotrypsinogen A decreases toward the sodium concentration return to 30 mM throughout the
the center of the bead, indicating that the protein is diffus- bead.
ing to the center from the outside. Then, at a little more than
6 s, the unbound ␣-chymotrypsinogen A concentration be-
comes greatest near 3.0 ␮m of radius; this indicates a bifur-
30.14
cation of diffusion. ␣-chymotrypsinogen A to the outside of 30.12
3.0 ␮m diffuses out to the bead, while ␣-chymotrypsinogen 30.10
Concentration (mM)

A to the inside of 3.0 ␮m diffuses inward. Subsequently, 30.08 5.7 s

the unbound concentration rises further at the core (at about 30.06 2.7 s
8 s), and the unbound ␣-chymotrypsinogen A diffuses ex- 30.04
clusively to the outside of the bead. Greater than about 30.02 0.0 s
30.00
10 s, the ␣-chymotrypsinogen A because uniform through-
29.98
out the bead. All of these phenomena are driven by the 29.96 8.7 s
wave of cytochrome c following the adsorbed ␣-chymotryp- 29.94
sinogen A. 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0
Finally, the behavior of the salt in simulation 4 is worth Radius (micron)
noting. As seen in Fig. 9, the salt is initially displaced due
to cytochrome c and ␣-chymotrypsinogen A adsorption. It Fig. 9. Unbound sodium concentration vs. radius in competitive adsorption
of ␣-chymotrypsinogen A and cytochrome c (simulation 4): concentration
forms an inverse gradient and begins to diffuse out of the vs. radius lines are shown at 0.0, 2.7, 5.7, and 8.7 s. By 11.7 s, the
bead. However, at 8.7 s, the sodium concentration at the sodium concentration returns to 30 mM throughout the bead. See text for
center of the bead has actually dropped below 30 mM and explanation.
 S.R. Gallant / J. Chromatogr. A 1028 (2004) 189–195 195

5. Conclusions θ tortuosity factor

Λ column capacity (mM)
This paper has presented a simple model of single- and νi characteristic charge
multicomponent adsorption in an ion-exchange bead. The ρ dimensionless radial distance
mass transport is modeled as simple liquid phase diffusion of σi steric factor
unbound protein. The ion-exchange adsorption is modeled τ dimensionless time
as stoichiometric exchange taking into account restrictions
on available surface area for binding due to steric effects. Subscripts
Observations of induced salt gradients and of strong multi- i mobile/stationary phase component number
component displacement effects show the SMA isotherm is (i = 1 designates salt)
uniquely appropriate for modeling these effects. As has been m finite difference index of radial elements
shown previously [19], the traditional Langmuir isotherm is n finite difference index of time elements
not adequate to model the strong salt-dependent changes in
affinity observed in ion-exchange adsorption of proteins.
Clearly, this a quite a simple model in that it does not in- References
clude electrophoretic mechanisms of transport [9]. In spite of
its simplicity, it is able to predict phenomena of single- and [1] A. Ljunglöf, R. Hjorth, J. Chromatogr. A 743 (1996) 75.
multicomponent protein adsorption that have been observed [2] A. Lijunglöf, J. Thommes, J. Chromatogr. A 813 (1998) 387.
experimentally by confocal microscopy [1–4]. It seems log- [3] P. DePhillips, A.M. Lenhoff, J. Chromatogr. A 933 (2001) 57.
[4] T. Linden, A. Ljunglöf, L. Hagel, M. Kula, J. Thömmes, Sep. Sci.
ical that this model and others will used to develop insights
Technol. 37 (2002) 1.
into the behavior of ion-exchange stationary phases. Such a [5] F.G. Helfferich, P.W. Carr, J. Chromatogr. 629 (1993) 97.
model allows the roles of individual components (salt and [6] J.C. Bellot, J.S. Condoret, Process Biochem. 28 (1993) 365.
protein in bound and unbound states) to be analyzed. Com- [7] C.A. Brooks, S.M. Cramer, AIChE J. 38 (1992) 1969.
parison of experimental data to theoretical predictions can be [8] A.I. Liapis, B.A. Grimes, K. Lacki, I. Neretnieks, J. Chromatogr. A
921 (2001) 135.
used to understand what physical phenomena are controlling
[9] B.A. Grimes, A.I. Liapis, J. Colloid Interface Sci. 248 (2002) 504.
protein adsorption in chromatographic separation processes. [10] S.R. Dziennik, E.B. Belcher, G.A. Barker, M.J. DeBergalis, S.E.
In other words, such a model offers a rigorous framework for Fernandez, A.M. Lenhoff, PNAS 100 (2003) 420.
interpretation of experimental observations associated with [11] N.K. Boardman, S.M. Partridge, Biochem. J. 59 (1955) 543.
the design of new stationary phases. Analysis of restrictions [12] W. Kopaciewicz, M.A. Rounds, J. Fausnaugh, F.E. Regnier, J. Chro-
matogr. 266 (1983) 3.
on mass transport and variations in selectivity can be carried
[13] R.R. Drager, F.E. Regnier, J. Chromatogr. 359 (1986) 147.
out most conveniently if there is some standard of compar- [14] A. Velayudhan, Cs. Horváth, J. Chromatogr. 443 (1988) 13.
ison and prediction of what the optimum stationary phase [15] P. Cysewski, A. Jaulmes, R. Lemque, B. Sebille, C. Vidal-Madjar,
might offer in performance is also possible. G. Jilge, J. Chromatogr. 548 (1991) 61.
[16] A. Velayudhan, Cs. Horváth, J. Chromatogr. A 663 (1994) 1.
[17] A. Velayudhan, Studies in nonlinear chromatography, Doctoral dis-
sertation, Yale University, 1990.
6. Nomenclature [18] S.D. Gadam, G. Jayaraman, S.M. Cramer, J. Chromatogr. 630 (1993)
37.
C1,init initial salt concentration (mM) [19] S.R. Gallant, A. Kundu, S.M. Cramer, J. Chromatogr. A 702 (1995)
Ci mobile phase concentration (mM) 125.
Ci,f feed concentration at column inlet (mM) [20] S.D. Gadam, S.R. Gallant, S.M. Cramer, AIChE J. 41 (1995) 1676.
[21] S.R. Gallant, A. Kundu, S.M. Cramer, Biotechnol. Bioeng. 47 (1995)
Di diffusion coefficient (cm2 /s)
355.
Gi Ci + Qi /β [22] S.R. Gallant, S. Vunnum, S.M. Cramer, J. Chromatogr. A 725 (1996)
K1i equilibrium constant 295.
NC number of components present in mobile phase [23] S.R. Gallant, S.M. Cramer, J. Chromatogr. A 771 (1997) 9.
Qi stationary phase concentration (mM) [24] J. Crank, The Mathematics of Diffusion, Oxford University Press,
Oxford, 1975.
Q̄1 bound salt which is not sterically shielded (mM)
[25] D.A. Anderson, J.C. Tannehill, R.H. Pletcher, Computational Fluid
Q̂1 bound salt which is sterically shielded (mM) Mechanics and Heat Transfer, Hemisphere, New York, 1984.
r radial position (cm) [26] E.L. Cussler, Multicomponent Diffusion, Elsevier, Amsterdam,
t time dimension (s) 1976.
[27] E.L. Cussler, Diffusion, Mass Transfer in Fluid Systems, Cambridge
University Press, Cambridge, 1984.
Greek letters [28] M.E. Young, P.A. Carroad, Biotechnol. Bioeng. 22 (1980) 947.
β phase ratio (εp /(1 − εp )) [29] A.M. Athalye, S.J. Gibbs, E.N. Lightfoot, J. Chromatogr. 589 (1992)
εp void fraction in particle 71.