Journal of Chromatography A, 1218 (2011) 77967803

Contents lists available at SciVerse ScienceDirect

Journal of Chromatography A
journal homepage: www.elsevier.com/locate/chroma

Energetics of protein adsorption on amine-functionalized mesostructured

cellular foam silica
Jungseung Kim, Rebecca J. Desch, Stephen W. Thiel , Vadim V. Guliants, Neville G. Pinto
School of Energy, Environmental, Biological and Medical Engineering, University of Cincinnati, Cincinnati, OH 45221-0012, USA

a r t i c l e

i n f o

Article history:
Received 18 July 2011
Received in revised form 25 August 2011
Accepted 26 August 2011
Available online 3 September 2011
Keywords:
Mesostructured cellular foam (MCF) silica
Aminopropyl group (APTES)
Adsorption
Energetics
Flow microcalorimetry (FMC)
Sodium sulfate
Lysozyme

a b s t r a c t
The energetics of lysozyme adsorption on aminopropyl-grafted MCF silica (MCF-NH2) are compared to
the trends observed during lysozyme adsorption on native MCF silica using ow microcalorimetry (FMC).
Surface modication on MCF silica affects adsorption energetics signicantly. All thermograms consist of
two initial exothermic peaks and one later endothermic peak, but the heat signal trends of MCF-NH2 are
opposite from those observed for adsorption onto native MCF silica in salt solutions of sodium acetate
and sodium sulfate. At low ionic strength (0.01 M), LYS adsorption onto MCF-NH2 was accompanied by
a large exotherm followed by a desorption endotherm. With increasing ionic strength (0.1 and 3.01 M),
the magnitude of the thermal signal decreased and the total process became less exothermic. Also a
higher protein loading of 14 mol g1 was obtained at low ionic strength in batch adsorption isotherm
measurements. Taken together, the FMC thermograms and batch adsorption isotherms reveal that MCFNH2 has the nature of an ion exchange adsorbent, even though lysozyme and the aminopropyl ligands
have like net charges at the adsorption pH. Reduced electrostatic interaction, reduced Debye length, and
increased adsorption-site competition attenuate exothermicity at higher ionic strengths. Thermograms
from ow microcalorimetry (FMC) give rich insight into the mechanisms of protein adsorption. A twostep adsorption mechanism is proposed in which negatively charged surface amino acid side chains on the
lysozyme surface make an initial attachment to surface aminopropyl ligands by electrostatic interaction
(low ionic strength) or van der Waals interaction (high ionic strength). Secondary attachments take place
between protruding amino acid side chains and silanol groups on the silica surface. The reduced secondary
adsorption heat is attributed to the inhibitory effect of the enhanced steric barrier of aminopropyl group
on MCF silica.
2011 Elsevier B.V. All rights reserved.

1. Introduction
Mesoporous silica surfaces can be modied with functional
groups with a charged nature (such as amine, sulfonic, and carboxyl groups) or a hydrophobic nature (such as octyl and octadecyl
groups) [1]. These groups can be used to control biomolecule
loading and release in drug delivery systems [24], protein adsorption and desorption for bioseparations [5,6], and immobilized
enzyme technology for biocatalysis [7]. Amine-terminated groups
are widely used to modify mesoporous silica surfaces because
these groups are easily grafted to the silica surface at a high surface density [8]. This surface modication changes electrochemical
properties, such as surface charge density and surface functionality,
as compared to those of native mesoporous silica [9]. Amineterminated silica can be used as a support for enzymes immobilized
for adsorption [9] and hydrolysis reactions [10] in aqueous solution

Corresponding author. Tel.: +1 513 556 4130, fax: +1 513 556 3474.
E-mail address: [email protected] (S.W. Thiel).
0021-9673/$ see front matter 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.chroma.2011.08.083

as well as for esterication [11] and transesterication reactions

[12] in nonaqueous solution.
Amine-terminated functional groups have been used to control adsorption onto silica in a number of systems. Adsorbed
uorescent dye was observed to be released more slowly from
amine-functionalized mesoporous silica than from native mesoporous silica nanoparticles at pH 5; this difference was explained
by changes in the electrostatic interactions between the molecule
and the surface due to the amine group [13]. Aminopropyl functionalization was used to control protein sequestration and release
on SBA-15 silica and MCF silica, enabling incorporation of the
ion-exchange properties of the aminopropyl functionality on the
negatively charged molecular sieve silica substrate at pH 7 [14].
A previous study showed that silanol-rich MCF silica does not
have ion-exchange properties [15]. The adsorption capacities of
bovine serum albumin and cellulase were reported to be higher for
amine-functionalized mesoporous FDU-12 silica than for the corresponding native silica due to stronger electrostatic interactions
[16]. It has also been observed that immobilized enzymes on aminefunctionalized mesoporous silica can show improved biocatalytic

J. Kim et al. / J. Chromatogr. A 1218 (2011) 77967803

activity compared to those immobilized on the native mesoporous

silica, indicating possible protein orientation control [17].
The characteristics associated with a particular surface functionality can be inuenced by solution properties such as salt
concentration, salt type, and pH. Neutral salts can modulate protein adsorption and desorption in processes such as hydrophobic
interaction chromatography (HIC) and ion exchange chromatography (IEC) [18], by affecting the surface charges of both the protein
[19,20] and the solid surface [21]. In HIC, high salt concentration
increases hydrophobic interactions and adsorption while low salt
concentration is used for protein elution. The opposite salt effect
is used for IEC [18]. Lysozyme adsorption on amine-functionalized
SBA-15 depends on salt type and salt concentration, indicating specic ion effects [9].
The current understanding of the fundamental interactions
between amine-functionalized mesoporous silica and proteins during immobilization in salt solutions is limited. This information
would be useful in designing highly efcient immobilized enzyme
systems and controlling protein immobilization. Under non-ideal
solution and surface conditions, such as at high protein concentrations [22], ow microcalorimetry (FMC) is an invaluable tool
for measuring heat effects to reveal adsorption thermodynamics
and improve the understanding of the complex phenomena of protein adsorption. A previous FMC study [15] of lysozyme adsorption
on mesostructured cellular foam demonstrated that the thermodynamics of protein adsorption on native silica can be modulated by
changing the salt concentration in solution. As the salt concentration increased, decreasing electrostatic interactions with reduced
Debye length and increasing van der Waals interactions between
protein and surface contributed to larger exothermic heats of
adsorption. Multiple heat events were observed in these experiments, suggesting that multiple points or modes of attachment are
possible during protein adsorption on a mesoporous silica surface.
In the study reported here, the thermodynamics of lysozyme
adsorption on amine-functionalized mesostructured cellular foam
silica (MCF-NH2) are examined using ow microcalorimetry (FMC)
under different salt concentrations. Lysozyme adsorption thermodynamics on the functionalized silica are compared those on
native silica surfaces. This report demonstrates the effects of salt
concentration and surface modication for protein adsorption thermodynamics on functionalized mesoporous silica and establishes
the ion exchange nature of MCF-NH2.
2. Experimental
2.1. Materials
Tetraethyl orthosilicate (TEOS, 98%), (3-aminopropyl) triethoxysilane (APTES, 99%) and lysozyme from chicken egg white
(LYS, 90%) were obtained from SigmaAldrich (St. Louis, MO). Acetic
acid and hydrochloric acid were obtained from Fisher Scientic
and Pharmco-Aaper (Brookeld, CT), respectively. Sodium acetate
anhydrous and sodium azide (Biotech research grade) was obtained
from Fisher Scientic (Fair Lawn, NJ). MCF silica was synthesized
using the procedure reported previously [15].
2.2. Surface modication of MCF silica
The synthesized MCF silica material was acid-washed using
0.2 M HCl solution for 2 h at 80 C to remove carbon residues
and maximize the surface silanol concentration. (3-Aminopropyl)
triethoxysilane (APTES) was grafted onto the acid-washed silica
surface [8] by mixing 1 g of acid-washed silica with 5 mL of APTES
in 100 mL of ethanol at 7080 C for 45 h; this material is referred
to below as MCF-NH2. The MCF-NH2 particles were recovered by

7797

vacuum ltration, washed with fresh ethanol, and dried at room

temperature. An initial APTES concentration of approximately
0.27 M was selected to produce maximum coverage on the MCF
silica surface [8].
2.3. Characterization of MCF silicas
Nitrogen adsorptiondesorption measurements (Tristar 3000,
Micromeritics, Norcross, GA) were performed at 77 K to determine
silica pore size, pore volume and BET surface area. All samples
were outgassed for 2 h at 150 C before the measurements. The pore
dimensions of cell and window structure were determined from the
nitrogen adsorption and desorption branches, respectively, using
the BdB (Broekhoff-de Boer)-FHH (Frenkel-Halsey-Hill) method
[23]. Surface modication was characterized by thermogravimetric analysis in air (TGA, SDT Q 600, TA Instruments, New Castle,
Delaware) to conrm the introduction of surface functional groups
and determine their surface coverage. The surface coverage, NS , was
estimated using Eq. (1) [24] and the weight loss measured by TGA.
Ns =

molligand molmolecule 1000 mmol

Wloss

100 g adsorbent Mligand

molligand
mol

(1)

The functional group density, D, was calculated using Eq. (2)

[25].
D = NA

NS
SBET

(2)

In Eqs. (1) and (2), Wloss is the weight loss from TGA
between 250 C and 400 C; Mligand is the molecular weight of
ligand (58 g aminopropyl ligand1 ); NA is the Avogadro constant
(6.02 1023 mol1 ); and SBET is the BET surface area of MCF silica.
2.4. Flow microcalorimetry (FMC)
Thermograms for lysozyme adsorption were collected at room
temperature using ow microcalorimetry (Microscal FMC 3 Vi,
Gilson Instruments, Westerville, OH, USA) following previously
reported methods [15,2729]. The sample loop volume was
1.36 mL, the mobile phase ow rate was 1.9 mL h1 , the bed volume
was 0.17 mL, and the adsorbent sample mass was 37.8 3.5 mg. The
mobile phase consisted of aqueous sodium acetate buffer at pH 5.2
with varying concentrations: 0.01 M sodium acetate, 0.1 M sodium
acetate, and 0.01 M sodium acetate with 1 M sodium sulfate. Sample
solutions were prepared by dissolving lysozyme in samples of each
mobile phase. The integral heat of adsorption was determined from
the area under deconvoluted asymmetric Gaussian peaks using the
PeakFIT software package (Systat Software Inc., San Jose, CA, USA);
calibration factors were calculated for each mobile and solid phase
combination as described previously [15]. Peaks were reproducible
with a peak area standard error of 17%.
2.5. Adsorption isotherm
A study of lysozyme adsorption was completed to investigate
the effect of mobile phase composition on adsorption behavior
using MCF-NH2 silica for the three mobile phases used in the
FMC experiments. The detailed procedure for the measurement
of batch adsorption isotherms is reported in a recent publication
[15]. The batch adsorption was done for 1 h, which was the same
time scale as the FMC experiment. A Type I isotherm equation was
used to t the experimental data by non-linear regression using the
Mathematica software package (Wolfram Research, Champaign,
IL, USA). The reported precision of the regression parameters represent a 95% condence interval around the parameter
estimates.

7798

J. Kim et al. / J. Chromatogr. A 1218 (2011) 77967803

Table 1
Physical properties of MCF silica.
Sample

Cell (nm)

Window (nm)

Pore volume
(cm3 g1 )

BET surface
area (m2 g1 )

Surface
coverage (Ns )
(mmol g1 )

Ligand density (D)

(ligand nm2 )

MCF [15]
MCF-NH2

33.0
22.1

16.6
8.2

2.4
1.0

634.3
401.3

0.5

0.54

3. Results and discussion

3.1. Textural properties, ligand density and surface functionality
of aminopropyl-functionalized MCF silica
MCF silica combines low mass transport resistance with high
adsorption capacity and controlled pore size. The open pore structure is composed of relatively narrow windows and larger cell
pores. BdB-FHH analysis of the nitrogen adsorptiondesorption
data showed that the synthesized MCF silica had a cell diameter of 33 nm, a window diameter of 16.6 nm, a total pore volume
of 2.4 cm3 g1 , and a BET surface area of 634 m2 g1 . The aminopropyl groups incorporated during surface modication altered the
silica physical properties, shown in Table 1, resulting in reduced
effective pore diameters [30]. However, the pore window diameter
(8.2 nm) and cell pore diameter (22.1 nm) in the MCF-NH2 functionalized materials are large enough to accommodate lysozyme
(1.9 nm 2.5 nm 4.3 nm) [5] into the pore structure.
The incorporation of aminopropyl groups was conrmed by
DSC, as shown in Fig. 1. The weight loss across broad temperature ranges (such as 200700 C [31] or 300800 C [32]) has been
used to estimate the aminopropyl group coverage on mesoporous
silica to be in the range of 12.4 ligands nm2 [31,32]. Using this
approach, the estimated aminopropyl surface density for the MCFNH2 produced in this study is about 1.1 ligand nm2 on MCF-NH2,
consistent with the earlier reports. However, dehydroxylation can
cause weight loss at temperatures in the range of 450900 C [33].
Previous thermal analysis of aminopropyl grafted-mesoporous silica found that the main aminopropyl decomposition event occurs in
the range of 250400 C [31]. Consequently, in this study the weight
loss from 250 C to 400 C (2.9%) was used to estimate the aminopropyl density and coverage to be 0.5 mmol ligand g1 adsorbent
and 0.54 ligand nm2 .
The overall properties of the aminopropyl-functionalized silica will potentially be inuenced by free silanol groups that
can be available even after the surface has reached maximum

Fig. 1. TGA and DSC results for aminopropyl functionalized mesocellular foam silica
(MCF-NH2). Heating rate: 10 C min1 ; air ow rate: 50 mL min1 .

functionalization [30]. Since amorphous silicas have a silanol surface density of 4.65.7 OH nm2 regardless of origin and structural
differences [34], and since each grafted APTES ligand binds to up
to three silanol groups [1], about 4 silanol groups per APTES ligand
are expected to remain on the functionalized surface.
3.2. Effect of ionic strength and surface modication
Aminopropyl-functionalized SBA-15 (SBA-15-NH2) has been
reported to have a higher surface charge than native SBA-15 across
a wide pH range (pH 310) [9,30]. The surface charge of native
mesoporous silica changes with salt concentration and salt type
due to specic cation effects rather than specic anion effects [21].
Protein loading on SBA-15-NH2 is inuenced by the presence of dissolved salts; for example, at pH 9, the adsorption of lysozyme on
SBA-15-NH2 was increased by increasing the ionic strength using
various salts [9]. These literature results indicate that both the salt
concentration and the surface modication have signicant roles in
modulating the nature of the guesthost interactions that underlie
the protein adsorption process on mesoporous silica.
The batch isotherms obtained in this study for lysozyme adsorption on MCF-NH2 silica at pH 5.2 in different ionic strength
solutions, modulated by sodium sulfate, are presented in Fig. 2.
The curves were obtained from non-linear regression using a
Type I isotherm (r2 0.95); regression parameters are presented
in Table 2. The highest protein loading was obtained at the lowest
ionic strength (0.01 M); the protein loading at an ionic strength
of 0.1 M was higher than that observed at an ionic strength of
3.01 M. In a recent paper the authors reported the opposite trend
for batch adsorption of lysozyme onto MCF silica [15]: protein
loading increased with ionic strength (increasing sodium sulfate concentration) due to attenuated electrostatic interaction and
enhanced van der Waals interaction. Independently measured

Fig. 2. Isotherms for adsorption of lysozyme onto aminopropyl functionalized

mesocellular foam silica (MCF-NH2) at pH 5.2, 25 C, 150 rpm, and 1 h adsorption
period. Initial protein concentration: 1391048 M; (1) ionic strength: 0.01 M; (2)
ionic strength: 0.1 M; (3) ionic strength: 3.01 M.

J. Kim et al. / J. Chromatogr. A 1218 (2011) 77967803

Table 2
Type I isotherm parameters for the adsorption of lysozyme on MCF-NH2 silica at pH
5.2 and 25 C as a function of ionic strength. Adsorption period: 1 h; initial protein
concentration: 1391048 M. Values are presented as mean standard error.
Ionic strength (M)

nm (mol g1 )

K (mM1 )

r2

0.01 SA
0.1 SA
3.01 (0.01 M SA + 1 M SS)

61. 18.
19.8 1.4
19.7 6.0

0.0012 0.0006
0.0033 0.0001
0.0019 0.0011

0.989
0.996
0.959

7799

the surface ligands. Protruding positively charged residues on the

lysozyme surface might then be able to penetrate through the ligand steric barrier to form strong secondary attachments to silanol
groups on the MCF silica surface. The broad peak of the second
exotherm indicates restricted access to the sterically hindered secondary adsorption site.
The largest heat of adsorption on MCF-NH2 silica was observed
at 0.01 M ionic strength, with similarly patterned but attenuated
thermograms observed in 0.1 M and 3.01 M ionic strength buffers.
The magnitude of the heat of adsorption was approximately four
times greater in 0.01 M ionic strength (0.01 M acetate buffer) than
in 3.01 M ionic strength (1.0 M sodium sulfate in 0.01 M sodium
acetate buffer) and the adsorption process became less exothermic
with increasing ionic strength. Together, these trends indicated that
the attractive adsorption interactions weaken with increasing ionic
strength. The FMC results are consistent with the trend of decreased
protein loading at high ionic strength.
The energetic trends of MCF-NH2 with increasing ionic strength
are contrasted to native MCF silica in Fig. 6. Heat signals were normalized to the amount of protein adsorbed at equilibrium to allow
direct energetic comparison between MCF-NH2 and MCF. Although
the MCF-NH2 window pores are smaller than those of MCF (8.2 vs.
16.6 nm), both window sizes allowed lysozyme penetration into
the pore structure. At 0.01 M ionic strength, MCF showed a sharp
initial exotherm followed by a small, at secondary exotherm and
a small endotherm. The MCF-NH2 exotherm was broader with an
enhanced second exotherm and larger endotherm. At 3.01 M ionic
strength, MCF had a broad and strong merged primary and secondary exotherm with a negligible endotherm, while MCF-NH2
had an attenuated initial exotherm followed by a small, broad secondary exotherm and a weak endotherm.
The difference between lysozyme adsorption onto MCF silica
and MCF-NH2 silica is attributed to signicant changes in the silica
surface charge and chemistry, demonstrating that surface modication affects protein adsorption thermodynamics. It has been
reported that terminated surface functional groups have signicant effects on protein adsorption, and that an amine-terminated
surface shows less protein adsorption afnity than surfaces with
hydrophobic termination [38]. Aminopropyl grafting changes the
electrochemical properties of the mesoporous silica surface resulting in positive zeta-potentials and positive surface charge densities
over a broad range of solution pH (pH 37) [9,30]. It was suggested
that aminopropyl-grafted mesoporous silica has the dual characteristics of size exclusion and ion exchange adsorbents [14]. In ion
exchange applications, low salt concentration is used for selective
protein retention on adsorbents and high salt concentration is used
mainly for protein elution [18]. The lysozyme adsorption capacity
and enthalpy of adsorption demonstrate that MCF-NH2 behaves
as an ion-exchanger, showing stronger attraction, less desorption,
and higher lysozyme loading in a lower ionic strength buffer, and
attenuated interactions and loading at higher ionic strength.

SA = sodium acetate; SS = sodium sulfate.

lysozyme loadings for the mobile and stationary phases used

in this study at the FMC sample concentration of 10 mg mL1
LYS are shown in Table 3. These loading ranged from 16 1 to
22 2 mg mL1 . For comparison, the monolayer LYS capacity on
a sulfopropyl agarose adsorbent at 0.02 M NaCl was reported to
be 74 2 mg mL1 [35]; the less repulsive adsorption environment
created by the negatively charged sulfopropyl group could contribute to the higher loading on the agarose adsorbent.
Flow microcalorimetry was used to measure the effect of ionic
strength on the enthalpy of protein adsorption. The thermograms
for lysozyme adsorption on MCF-NH2 silica at the same buffer
and sodium sulfate concentrations used for the isotherm measurements are shown in Figs. 35. In each thermogram, an exotherm
began when the protein sample reached the packed bed 1015 min
after the protein injection and continued until about 70 min when
fresh mobile phase began to replace the protein sample in the bed.
From 70 min to 110 min, an endotherm developed as partial protein
desorption occurred.
Each thermogram was deconvoluted into two exothermic peaks,
attributed to adsorption, and an endothermic peak, attributed to
desorption, as shown in Figs. 35. Heats of adsorption and desorption onto MCF and MCF-NH2 silica are compared in Table 3,
to highlight the effects of surface functionalization on adsorption energy. Previous calorimetric protein adsorption studies
revealed that an exothermic adsorption event occurs when attractive forces between protein and surface dominate; repulsive
lateral proteinprotein interactions, protein restructuring, and solvent release from the surface can cause endothermic adsorption
[22,36]. Although complex thermograms are sometimes attributed
to protein conformation changes [27,37], lysozyme is not prone
to conformational change [36]; no restructuring endotherm is
observed. A previous study of lysozyme adsorption on mesoporous
silica attributed multiple exothermic peaks to a primary adsorption event followed by reorientation or secondary attachment [15].
The second exothermic peak was greater than the initial peak for
all cases of lysozyme adsorption onto MCF-NH2 silica, indicating
that stronger interactions occurred as a result of reorientation or
secondary adsorption. The aminopropyl groups extend from the
silica surface, offering a site for the initial adsorption. The sharp
initial exotherm indicates that the lysozyme has easy access to
the initial adsorption site, presumably the amine termination of

Table 3
Heat of adsorption and desorption of lysozyme on MCF-NH2 and MCF [15] silicas at pH 5.2; feed protein concentration: 10 mg mL1 ; sample loop volume: 1.36 mL; ow rate:
1.9 mL h1 ; MCF-NH2 sample size: 37.8 3.5 mg, MCF sample size: 25.2 2.9 mg; temperature: 25 C. Values are presented as mean standard error.
Ionic strength (M)

(mol g
MCF-NH2
0.01 M SA
0.1 M SA
0.01 M SA + 1 M SS (3.01 M)
MCF [15]
0.01 M SA
0.01 M SA + 1 M SS (3.01 M)

Exothermic peaks (kJ mol1 )

Protein loading
)

(mg mL

II

Endothermic peak (kJ mol1 )

II

H

H

H + H

H

III

Net heat of adsorption (kJ mol1 )

HTotal

14.2 1.0
10.9 0.8
10.2 0.7

22.3 1.6
17.1 1.3
16.0 1.1

22.6
4.6
6.1

37.3
15.6
8.6

60. 10.
20.2 3.4
14.7 2.5

37.3
4.1
8.2

22.6 3.8
16.0 2.7
6.5 1.1

10.2 0.7
27.6 0.8

16.0 1.1
43.4 1.3

10.1
25.3

11.7
24.0

21.8 3.7
49.3 8.4

4.5
0

17.3 2.9
49.3 8.4

HTotal = HI +HII + HIII .

SA = sodium acetate; SS = sodium sulfate.

7800

J. Kim et al. / J. Chromatogr. A 1218 (2011) 77967803

Fig. 3. Deconvoluted thermograms for lysozyme adsorption and desorption on MCF-NH2 at pH 5.2 in 0.01 M sodium acetate. Sample loop volume: 1.36 mL; ow rate:
1.9 mL h1 ; adsorbent sample size: 37.8 3.5 mg; temperature: 25 C; feed protein concentration: 10 mg mL1 . Curves shown are for (1) experimental data (black line); (2)
total peak t (purple line); (3) 1st exothermic peak t (red line); (4) 2nd exothermic peak t (blue line); (5) endothermic peak t (olive line). (For interpretation of the
references to color in this gure legend, the reader is referred to the web version of the article.)

Fig. 4. Deconvoluted thermograms for lysozyme adsorption and desorption on MCF-NH2 at pH 5.2 in 0.1 M sodium acetate. Sample loop volume: 1.36 mL; ow rate:
1.9 mL h1 ; adsorbent sample size: 37.8 3.5 mg; temperature: 25 C; feed protein concentration: 10 mg mL1 . Curves shown are for (1) experimental data (black line); (2)
total peak t (purple line); (3) 1st exothermic peak t (red line); (4) 2nd exothermic peak t (blue line); (5) endothermic peak t (olive line). (For interpretation of the
references to color in this gure legend, the reader is referred to the web version of the article.)

Fig. 5. Deconvoluted thermograms for lysozyme adsorption and desorption on MCF-NH2 at pH 5.2 with 1.0 M sodium sulfate in 0.1 M sodium acetate. Sample loop volume:
1.36 mL; ow rate: 1.9 mL h1 ; adsorbent sample size: 37.8 3.5 mg; temperature: 25 C; feed protein concentration: 10 mg mL1 . curves shown are for (1) experimental
data (black line); (2) total peak t (purple line); (3) 1st exothermic peak t (red line); (4) 2nd exothermic peak t (blue line); (5) endothermic peak t (olive line). (For
interpretation of the references to color in this gure legend, the reader is referred to the web version of the article.)

J. Kim et al. / J. Chromatogr. A 1218 (2011) 77967803

7801

Fig. 6. Comparison of thermograms for lysozyme adsorption and desorption on MCF silica and MCF-NH2 silica at pH 5.2. Sample loop volume: 1.36 mL; ow rate: 1.9 mL h1 ;
adsorbent sample size (MCF-NH2 silica): 37.8 3.5 mg; adsorbent sample size (MCF silica): 25.2 2.9 mg; temperature: 25 C; feed protein concentration: 10 mg mL1 . (a)
0.01 M sodium acetate; (b) 1.0 M sodium sulfate in 0.01 M sodium acetate. Curves shown are for (1) MCF silica (black line); (2) MCF-NH2 silica (red line). (For interpretation
of the references to color in this gure legend, the reader is referred to the web version of the article.)

Because lysozyme and MCF-NH2 both carry a net positive charge

at pH 5.2 [9,39], one might expect electrostatic repulsion interactions to be dominant. The initial exotherms observed at all salt
concentrations showed that there are signicant attractive interactions between lysozyme and MCF-NH2 despite the shared net
positive charge on both the surface and the protein. Shibata and
Lenhoff [40] observed the adsorption of positively charged proteins (lysozyme and chymotrypsinogen A) onto positively charged
quartz-NH2 using total internal reectance uorescence spectroscopy. The adsorption of proteins was attributed to combined
effects of electrostatic interactions, van der Waals interactions
and hydrophobic interactions. Measurements of attenuated interactions between protein and the surface were attributed to
hydrophobic interactions, which are typically associated with
endotherms, acting against the exothermic electrostatic and van
der Waals forces during protein adsorption [40].
Double layer electrostatic interactions and Lifshitz-van der
Waals interactions are the main contributions to the total
proteinsurface interaction [4143]. The authors previously
reported [15] that electrostatic and van der Waals interactions
are both important for lysozyme adsorption on MCF silica at low
ionic strength. At higher salt concentrations, electrostatic interactions decrease and van der Waals interactions increase signicantly
resulting in enhanced exotherms. In this study, electrostatic interactions were attenuated at higher ionic strengths. The Debye length
in each salt solution is calculated with the following equation:


K01

kB T0 r
2NA e2 I

(3)

where kB is the Boltzmann constant; 0 is the permittivity of free

space; T is the temperature, 298 K; NA is the Avogadro constant; e is
the charge of an electron; and I is the ionic strength of the solution.
The relative dielectric constants, 0 , [44] in different salt concentrations were estimated from the reported values. Each parameter
and calculated Debye lengths are summarized in Table 4. The Debye
lengths at 0.1 M and 3.01 M decreased from the value at low ionic
strength by about 68% and 99%, respectively, indicating that electrostatic interactions cannot play a major role in protein adsorption
on MCF-NH2 silica at higher ionic strengths. The effective charge of
lysozyme decreases in higher sodium sulfate solution due to preferential binding of sulfate anions onto positively charged amino
acid groups [39]. The reduced surface charge of lysozyme can

also reduce electrostatic interactions between lysozyme and MCF

silica surface. Although electrostatic repulsion between the likecharged protein and MCF surfaces might be expected to be the
dominant interaction, the initial exotherm derived from attractive interactions, especially in normal buffer (no salt), suggests
that aminopropyl groups change the orientation of the adsorbing
protein to interact with negatively charged groups on the protein
surface.
The lysozyme surface is anisotropic at pH 5.2, with 18 surface
residues (Arg, His, and Lys) that can become positively charged
and 4 surface residues (Asp 18, Asp 87, Asp 119 and Glu 7) that
can become negatively charged [45]. Lysozyme has two binding
sites, region A (1 Lys, 5 Arg, 33 Lys, 114 Arg, 128 Arg) and region
B (1 Lys, 14 Arg, 15 His, 68 Arg) [46]. These binding sites preferentially adsorb onto negative surfaces at pH 5.2, but other binding
sites involving Asp (side chain pKa = 3.65) and Glu (side chain
pKa = 4.25) [20,47] could coordinate with positively charged surface
ligands and create the observed exotherms. The negatively charged
residues are located in close proximity in the lysozyme molecule
[46], suggesting that the orientation of the protein while approaching and adsorbing onto the MCF-NH2 silica is crucial. At higher ionic
strengths, the Debye length (k0 1 ) is shorter [19,45], signicantly
reducing electrostatic interactions (either repulsive or attractive)
[45]. At high Debye lengths (3.0 nm), the protein can be pulled
into a favorable orientation while approaching the pore surface,
allowing coordination of negatively charged amino acids on the
protein and positively charged ligand groups while reducing repulsion between positively charged surface amino acids and positively
charged ligand groups. As the Debye length decreases, protein
molecules increasingly approach the pore surface in random orientations, weakening electrostatic interactions and reducing the
scope for attachment. In addition, as the Debye length decreases,
secondary attachment, which requires positive amino acid side
chains to move toward the silica silanol groups by electrostatic
interaction, becomes less likely.
Table 4
Relative dielectric constant, r and Debye length, k0 1 , as functions of ionic strength.
Ionic strength (M)

k0 1 ()

0.01
0.1
3.01

78
77
58

30
9.5
1.5

7802

J. Kim et al. / J. Chromatogr. A 1218 (2011) 77967803

Short-range van der Waals interactions become the dominant

driving force at higher ionic strengths [42]. The energy change due
to van der Waals interactions (Uvdw ) can be estimated using the
Hamaker equation [48]:
Uvdw (h) =


H 1
6

1
+ ln
h+2

 h 
h+2

(4)

Here h is the ratio of distance between the protein and surface

and the radius of the protein (2.1 nm for lysozyme). The Hamaker
constant (H) is estimated to be 1.27 1020 J at 298 K [15]. In this
model, the maximum attraction will occur at the closest possible
distance between the protein and surface, typically taken to be
0.1 nm [48]. The corresponding van der Waals interaction energy
is thus estimated to be 18 kJ mol1 , 35 times the heat of the
rst exotherm (HI ) in the highest ionic strength system studied
here (when electrostatic interactions are minimal). Interestingly,
the dispersion energy matches the total enthalpy of adsorption
(HI + HII ) well. The steric barrier of silica surface ligands prevents or retards the closest approach of the protein to the silica
surface. Additionally, salt ions can form hydrogen bonds with ligand groups on the silica surface, suppressing protein binding in
high ionic strength solutions [49] by competing for the silica surface
amine groups.
3.3. Adsorption mechanism
The proposed mechanism [15] for lysozyme adsorption onto
native MCF silica is a random adsorption driven by the combined
interactions between electrostatic and van der Waals interactions
at low salt concentration and primarily by van der Waals interactions at high salt concentration. This primary adsorption is followed
by a reorientation of the protein relative to the silica surface accompanied by additional electrostatic interactions or van der Waals
interactions between surface amino acid groups on the lysozyme
surface and silanol groups on the silica surface. The adsorption is
followed by desorption except at high salt concentration. Lysozyme
adsorption onto native MCF silica increases with increasing ionic
strength because van der Waals interactions become much stronger
as electrostatic interactions descrease with decreased Debye length
at high ionic strength. Likewise, the adsorption of lysozyme onto
MCF-NH2 silica begins with an electrostatic interaction between
the positively charged amine group grafted to the silica surface and
a negative group on the lysozyme surface. Once the lysozyme has
been brought into proximity with the surface, it can form secondary
attachments with silanol groups or interact with the surface via
van der Waals interactions. With increasing salt concentration, the
electrostatic interactions decrease, the Debye length decreases, and
van der Waals interactions tends to increase but might be limited
due to the steric barrier posed by the aminopropyl groups on the
MCF silica. In addition, competition between salt ions and surface
ligand groups may cause the lysozyme to adsorb onto the surface
in more random orientations, resulting in decreased adsorption
enthalpy. Secondary adsorption is driven by van der Waals interactions between the protein and the surface; however, as before the
surface functional ligands may acts as a steric barrier to inhibit the
closest approach of protein to surface for enhanced van der Waals
interactions.
This study has shown that a simple silica surface functionalization reversed the lysozyme adsorption energetic and capacity
trends in the presence of varying sodium sulfate concentrations. Partially replacing silanol groups with aminopropyl groups
changed silica surface properties in terms of charge and chemistry
to impact protein adsorption thermodynamics. Unlike native silica, MCF-NH2 functions as an ion exchanger. The interplay of ionic
strength and surface modication was demonstrated with batch

adsorption isotherms and calorimetric data for protein adsorption

on functionalized mesoporous silica.
4. Conclusions
The adsorption energetics of lysozyme on amine-functionalized
mesoporous silica are reported here. FMC thermograms were used
to identify mechanisms of protein adsorption as a function of
ionic strength and surface conditions. Surface modication by
grafting aminopropyl groups on MCF (MCF-NH2) silica affected
adsorption energetics signicantly as compared to that of acidwashed MCF silica. Although MCF-NH2 and lysozyme had like
charges at pH 5.2, attractive interactions were the main driving
force for protein adsorption, demonstrating that the interactions
between specic amino acid groups and the surface ligand were
an important factor for protein adsorption energetics in this
study.
The total enthalpy of adsorption decreases with increasing ionic
strength for lysozyme adsorption on MCF-NH2 silica. Lower ionic
strength results in stronger exothermic peaks and correspondingly
stronger endothermic desorption events. At high ionic strength, the
short Debye length means that the net protein charge is less important than interactions with specic sites on the external protein
surface. The trend observed in the adsorption isotherm is consistent
with calorimetric results; higher protein loading was observed at
lower ionic strength. Thermograms and batch adsorption isotherms
conrmed that MCF-NH2 silica has ion exchange properties as
proposed in a previous study [14]. Attenuated exothermic peaks
observed with increasing salt concentration were attributed to
decreasing electrostatic interaction and inhibiting van der Waals
interactions due to steric barrier of surface functional groups
(aminopropyl groups) on MCF silica.
Acknowledgments
The authors are grateful for nancial support from the Nanoscale
Interdisciplinary Research Teams (NIRT) program sponsored by
National Science Foundation (CTS-0403897) (JSK), the University
of Cincinnati University Research Council (JSK) and NSF IGERT program (RJD).
References
[1] C.-H. Lee, T.-S. Lin, C.-Y. Mou, Nano Today 4 (2009) 165.
[2] Y. Zhao, J.L. Vivero-Escoto, I.I. Slowing, B.G. Trewyn, V.S.Y. Lin, Expert Opin. Drug
Deliv. 7 (2010) 1013.
[3] J.L. Vivero-Escoto, I.I. Slowing, B.G. Trewyn, V.S.Y. Lin, Small 6 (2010) 1952.
[4] S. Wang, Micropor. Mesopor. Mater. 117 (2008) 1.
[5] M. Hartmann, Chem. Mater. 17 (2005) 4577.
[6] Y. Ma, L. Qi, J. Ma, Y. Wu, O. Liu, H. Cheng, Colloids Surf. A 229 (2003) 1.
[7] H.H.P. Yiu, P.A. Wright, N.P. Botting, J. Mol. Catal. B: Enzym. 15 (2001) 81.
[8] Y.-H. Liu, H.-P. Lin, C.-Y. Mou, Langmuir 20 (2004) 3231.
[9] A. Salis, M.S. Bhattacharyya, M. Monduzzi, J. Phys. Chem. B 114 (2010) 7996.
[10] J. Zhao, Y. Wang, G. Luo, S. Zhu, Bioresour. Technol. 101 (2010) 7211.
[11] S.-E. Park, Sujandi, Curr. Appl. Phys. 8 (2008) 664.
[12] Y. Liu, G. Zhao, W. Zhu, J. Wang, G. Liu, W. Zhang, M. Jia, J. Braz. Chem. Soc. 21
(2010) 2254.
[13] P. DeMuth, M. Hurley, C. Wu, S. Galanie, M.R. Zachariah, P. DeShong, Micropor.
Mesopor. Mater. 141 (2011) 128.
[14] Y.-J. Han, G.D. Stucky, A. Butler, J. Am. Chem. Soc. 121 (1999) 9897.
[15] J. Kim, R.J. Desch, S.W. Thiel, V.V. Guliants, N.G. Pinto, J. Chromatogr. A 1218
(2011) 6697.
[16] S. Hartono, S. Qiao, K. Jack, B.P. Ladewig, Z. Hao, G. Lu, Langmuir 25 (2009) 6413.
[17] K. Szymanska, J. Bryjak, A.B. Jarzebski, Top. Catal. 52 (2009) 1030.
[18] G. Carta, A. Jungbauer, Protein Chromatography, Wiley-VCH, 2010.
[19] M. Bostroem, D.R.M. Williams, B.W. Ninham, Biophys. J. 85 (2003) 686.
[20] D.E. Kuehner, J. Engmann, F. Fergg, M. Wernick, H.W. Blanch, J.M. Prausnitz, J.
Phys. Chem. B 103 (1999) 1368.
[21] A. Salis, D.F. Parsons, M. Bostrom, L. Medda, B. Barse, B.W. Ninham, M. Monduzzi,
Langmuir 26 (2010) 2484.
[22] M.E. Thrash, N.G. Pinto, J. Chromatogr. A 944 (2002) 61.
[23] W.W. Lukens Jr., P. Schmidt-Winkel, D. Zhao, J. Feng, G.D. Stucky, Langmuir 15
(1999) 5403.

J. Kim et al. / J. Chromatogr. A 1218 (2011) 77967803

[24]
[25]
[27]
[28]
[29]
[30]
[31]
[32]
[33]
[34]
[35]
[36]

M.A. Abu-Daabes, N.G. Pinto, Chem. Eng. Sci. 60 (2005) 1901.

N.L. Dias Filho, Colloids Surf. A 144 (1998) 219.
M.E. Thrash Jr., J.M. Phillips, N.G. Pinto, Adsorption 10 (2005) 299.
A.C. Dias-Cabral, A.S. Ferreira, J. Phillips, J.A. Queiroz, N.G. Pinto, Biomed. Chromatogr. 19 (2005) 606.
J.M. Phillips, N.G. Pinto, J. Chromatogr. A 1036 (2004) 79.
A. Nieto, M. Colilla, F. Balas, M. Vallet-Reg-Reg6, 2004, 79. 79.Pinto.
V. Colilla ZeleZele., F. Balas, M. Vallet-Reg-Reg6, A. Zukal, N. Murafa, G. Goerigk,
Chem. Eng. J. 144 (2008) 336.
P. Shah, N. Sridevi, A. Prabhune, V. Ramaswamy, Micropor. Mesopor. Mater. 116
(2008) 157.
O. Sneh, S.M. George, J. Phys. Chem. 99 (1995) 4639.
L.T. Zhuravlev, Colloids Surf. A 173 (2000) 1.
Q.H. Shi, G.D. Jia, Y. Sun, J. Chromatogr. A 1217 (2010) 5084.
A. Katiyar, S.W. Thiel, V.V. Guliants, N.G. Pinto, J. Chromatogr. A 1217 (2010)
1583.

7803

[37] K.E. Michael, V.N. Vernekar, B.G. Keselowsky, J.C. Meredith, R.A. Latour, A.J.
Garca, Langmuir 19 (2003) 8033.
[38] D.J. Kim, J.M. Lee, J.G. Park, B.G. Chung, Biotechnol. Bioeng. 108 (2011) 1194.
[39] Y.R. Gokarn, R.M. Fesinmeyer, A. Saluja, V. Razinkov, S.F. Chase, T.M. Laue, D.N.
Brems, Protein Sci. 20 (2011) 580.
[40] C.T. Shibata, A.M. Lenhoff, J. Colloid Interface Sci. 148 (1992) 469.
[41] C.M. Roth, A.M. Lenhoff, Langmuir 11 (1995) 3500.
[42] J. Staahlberg, B. Joensson, C. Horvath, Anal. Chem. 64 (1992) 3118.
[43] C.J. Van Oss, M.K. Chaudhury, R.J. Good, Chem. Rev. 88 (1988) 927.
[44] A.S. Lileev, Z.A. Filimonova, A.K. Lyashchenko, J. Mol. Liq. 103-104 (2003) 299.
[45] E. Seyrek, P.L. Dubin, C. Tribet, E.A. Gamble, Biomacromolecules 4 (2003) 273.
[46] W.K. Chung, S.T. Evans, A.S. Freed, J.J. Keba, Z.C. Baer, K. Rege, S.M. Cramer,
Langmuir 26 (2010) 759.
[47] A.L. Lehninger, D.L. Nelson, M.M. Cox, Principles of Biochemistry, 1993.
[48] M.E. Thrash Jr., N.G. Pinto, J. Chromatogr. A 1126 (2006) 304.
[49] T. Yang, C.M. Breneman, S.M. Cramer, J. Chromatogr. A 1175 (2007) 96.