Chemical characterisation and fabrication of

chitosansilica hybrid scaolds with

3-glycidoxypropyl trimethoxysilane
Louise S. Connell,
a
Frederik Romer,
b
Marta Su

arez,
cd
Esther M. Valliant,
a
Ziyu Zhang,
a
Peter D. Lee,
e
Mark E. Smith,
bf
John V. Hanna
b
and Julian R. Jones
*
a
Chitosan has been explored as a potential component of biomaterials and scaolds for many tissue
engineering applications. Hybrid materials, where organic and inorganic networks interpenetrate at the
molecular level, have been a particular focus of interest using 3-glycidoxypropyl trimethoxysilane
(GPTMS) as a covalent crosslinker between the networks in a solgel process. GPTMS contains both an
epoxide ring that can undergo a ring opening reaction with the primary amine of chitosan and a
trimethoxysilane group that can co-condense with silica precursors to form a silica network. While many
researchers have exploited this ring-opening reaction, it is not yet fully understood and thus the nal
product is still a matter of some dispute. Here, a detailed study of the reaction of GPTMS with chitosan
under dierent pH conditions was carried out using a combination of solution state and solid state MAS
NMR techniques. The reaction of GPTMS with chitosan at the primary amine to form a secondary amine
was conrmed and the rate was found to increase at lower pH. However, a side-reaction was identied
between GPTMS and water producing a diol species. The relative amounts of diol and chitosanGPTMS
species were 80 and 20% respectively and this ratio did not vary with pH. The functionalisation pH had
an eect on the mechanical properties of 65 wt% organic monoliths where the properties of the organic
component became more dominant. Scaolds were fabricated by freeze drying and had pore diameters
in excess of 140 mm, and tailorable by altering freezing temperature, which were suitable for tissue
engineering applications. In both monoliths and scaolds, increasing the organic content disrupted the
inorganic network, leading to an increase in silica dissolution in SBF. However, the dissolution of silica
and chitosan was congruent up to 4 weeks in SBF, illustrating the true hybrid nature resulting from
covalent bonding between the networks.
Introduction
As the world's aging population increases, the number of indi-
viduals aected by bone and cartilage disorders is also
increasing.
1
To treat these debilitating conditions and address
the scarcity of natural gra material, it is necessary for synthetic
materials to be developed that mimic the physical properties of
natural tissue while also stimulating regeneration and remod-
elling by the body.
24
Bioactive glass scaolds have many of the
properties required for tissue engineering scaolds, but they
are too brittle to be used in cyclically loaded applications.
5
A
soer, tougher material is required.
6,7
Solgel derived hybrid
materials, where interpenetrating networks (IPNs) of organic
polymers and inorganic components are covalently bonded,
8
have shown promise as biodegradable tissue regeneration
scaolds. Natural polymers, such as chitosan,
914
poly-glutamic
acid
1518
and gelatin,
19,20
are covalently bonded to a silica
network via graing a silane containing coupling agent to
functional groups along the polymer. These hybrid materials
have been formed into scaolds by various methods, including
foaming,
20
freeze-drying,
11,19
and electrospinning.
21
The poten-
tial benets of hybrids over conventional composites is that
mechanical properties and dissolution rates can be varied by
a
Department of Materials, Imperial College London, South Kensington Campus, SW7
2AZ, UK. E-mail: julian.r.jones@imperial.ac.uk; Tel: +44 (0) 2075946749
b
Department of Physics, University of Warwick, Gibbet Hill Rd., Coventry CV4 7AL, UK
c
Fundacion ITMA, Parque Technologico de Asturias, 33428 Llanera, Asturias, Spain
d
Department of Nanostructured Material, Centro de Investigacion en Nanomateriales y
Nanotecnologa, Principado de Asturias, Parque Tecnologico de Asturias, 33428
Llanera, Spain
e
School of Materials, The University of Manchester, Oxford Rd., M13 9PL, UK
f
The Vice-Chancellor's Oce, University House, Lancaster University, Lancaster LA1
4YW, UK
Electronic supplementary information (ESI) available:
1
H NMR of chitosan and
GPTMS in D
2
O/DCl at pH 4 over time,
1
H
13
C HSQC spectra of fully hydrolysed
GPTMS aer 72 h in D
2
O/DCl at pH 2,
29
Si MAS NMR spectra of functionalised
chitosan at pH 2 and 4 with table of quantication, and thermogravimetric
analysis of 50 wt% organic chitosansilica scaolds aer 0, 72, 168, and 672 h
in SBF. See DOI: 10.1039/c3tb21507e
Cite this: J. Mater. Chem. B, 2014, 2,
668
Received 25th October 2013
Accepted 2nd December 2013
DOI: 10.1039/c3tb21507e
www.rsc.org/MaterialsB
668 | J. Mater. Chem. B, 2014, 2, 668680 This journal is The Royal Society of Chemistry 2014
Journal of
Materials Chemistry B
PAPER
P
u
b
l
i
s
h
e
d

o
n

0
3

D
e
c
e
m
b
e
r

2
0
1
3
.

D
o
w
n
l
o
a
d
e
d

b
y

U
n
i
v
e
r
s
i
d
a
d
e

F
e
d
e
r
a
l

d
a

P
a
r
a
i
b
a

o
n

2
3
/
0
5
/
2
0
1
4

1
6
:
3
6
:
1
4
.

View Article Online
View Journal | View Issue
altering the degree of coupling between the networks and the
overall organic content of the hybrids.
15,20
Chitosan, a polysaccharide derived from the deacetylation of
chitin from crustacean shells and fungi (Fig. 1), has interesting
applications in biomaterials due to its solubility in weakly acidic
solutions,
2224
biocompatibility, wound healing capabilities,
antimicrobial properties, and because it can be degraded by
lysozymes in vivo.
25
The presence of both primary amine and
hydroxyl groups allow for functionalisation, which can increase
polymer solubility or provide coupling points to covalently bond
to a silica network.
921,2630
3-Glycidoxypropyl trimethoxysilane
(GPTMS) can be used as a coupling agent as it contains an
epoxide ring which can undergo acid catalysed nucleophilic
attack, e.g. by a nucleophilic group on a polymer chain, to open
the epoxide ring.
31
This creates a polymer chain with trime-
thoxysilane functional groups which become hydrolysed under
acidic conditions to give silanol groups which condense in a
solgel process (either with or without another silica precursor)
to produce a silica network covalently bonded to the polymer.
Many groups have utilised this reaction in chitosan, but
none have fully characterised the functionalisation step. There
are several mechanisms proposed for the coupling between
chitosan and GPTMS (Fig. 2):
Covalent coupling of the epoxide ring to the primary amine
to form a secondary amine.
10,28,29
Ionic bonding between the positively charged amine
groups and negatively charged silanes.
10,32
Covalent coupling between the hydroxyl groups of the
chitosan via the epoxide ring.
30
Condensation of silane groups with hydroxyl groups.
33
Ionic bonds between protonated amines and negatively
charged oxygen resulting from the epoxide group.
14
Hydrogen bonding between amine, amide and hydroxyl
groups (exclusively or in combination with the above).
14
Qualitative FTIR data has been presented by a number of
researchers who used the reduction of the band at 1550 cm
1
(bending of the primary amine) as evidence of a reaction
occurring
28,29
however, either the same reduction was not
observed or it was not used as evidence for a reaction.
10,30
In an
attempt to quantify the degree of coupling, Shirosaki et al. used
a ninhydrin assay to show that 80% primary amine groups were
converted to secondary amines regardless of the amount of
GPTMS.
10
This was the case even when the ratio of GPTMS to
amine groups was as low as 0.1. However, the accuracy of the
assay is in doubt as the relative reactivity of the dierent chi-
tosans (reacted and unreacted) to the assay was not consid-
ered.
34
Calibration curves are required for each dierent type of
chitosan to quantify each form, which would not be feasible for
this system. Other groups have hypothesised that since the
chitosan is only soluble in weak acid, when the amine groups
are protonated, their nucleophilic behaviour would decrease,
rendering them unable to react with the epoxide ring.
30
Instead,
the ring would be hydrolysed to form diol species. Varghese
et al. hypothesised that the epoxide ring would react with the
OHgroups of the chitosan,
30
although this is unlikely given the
weak nucleophilic nature of the hydroxyl groups and the large
number of water molecules in solution that are more likely to
react. Although intuition suggests that higher pH values would
increase the nucleophility of the amine group, chitosan is only
soluble in weakly acidic solutions and so precludes the use of
neutral or basic reaction conditions.
Gabrielli et al. investigated the reaction between simple
small molecules and GPTMS by NMR and mass spectroscopy as
a model for the synthesis of organicinorganic hybrids.
35
The
simple amine, propylamine, catalysed silica condensation at all
pH values so that solid GPTMS products formed before the
reaction with the epoxide ring could occur. The carboxylic acid,
propanoic acid, reacted with the epoxide ring to form an ester
group, which was detected within 3 h by mass spectroscopy and
within 48 h by NMR.
35
If the amine catalysed-condensation of
the silanol groups of GPTMS in acidic solutions can be delayed
long enough to allow all components to remain soluble, then it
is possible that the amine group may react with the epoxide
ring. Despite the uncertainty surrounding the reaction mecha-
nism, chitosansilica hybrid monoliths, membranes and
particles have been synthesized either with GPTMS as both the
coupling agent and silica source,
10,12,28,29
or with the addition of
tetraethyl orthosilicate (TEOS),
30,36,37
which allows for indepen-
dent control of organicinorganic coupling and nal silica
content.
The primary aim of this work was to fully characterise the
functionalisation reaction of chitosan with GPTMS under
dierent pH conditions and to establish the importance of pH
on the properties of covalently bonded chitosansilica hybrid
Fig. 1 Structure of chitosan (left) derived by deacetylation of chitin
(right).
Fig. 2 Schematic illustrating the dierent reactions proposed to occur
between chitosan and GPTMS. Where diol groups are illustrated,
unopened epoxide rings may also be present.
This journal is The Royal Society of Chemistry 2014 J. Mater. Chem. B, 2014, 2, 668680 | 669
Paper Journal of Materials Chemistry B
P
u
b
l
i
s
h
e
d

o
n

0
3

D
e
c
e
m
b
e
r

2
0
1
3
.

D
o
w
n
l
o
a
d
e
d

b
y

U
n
i
v
e
r
s
i
d
a
d
e

F
e
d
e
r
a
l

d
a

P
a
r
a
i
b
a

o
n

2
3
/
0
5
/
2
0
1
4

1
6
:
3
6
:
1
4
.

View Article Online
monoliths. For the rst time, a variety of solution and solid state
NMR techniques were employed to characterise and quantify
the extent of reaction. The secondary aim was to apply the
knowledge obtained from the characterisation to establish an
optimum pH environment for fabricating freeze-dried porous
scaolds and to determine their suitability for use as tissue
engineering scaolds. Understanding these mechanisms is
critical for the development of chitosansilica hybrids as scaf-
folds for bone or cartilage regeneration or as any medical
device.
Materials and methods
Materials and sample preparation
Unless stated, all chemicals were analytical grade purchased
from Sigma Aldrich, UK and used as received.
Samples used to investigate the functionalisation reaction by
FTIR and solid state NMR were prepared by stirring chitosan
(0.5 g, M
w
: 50150 kDa, degree of deacetylation of 75%) in
deionised water at a concentration of 17 mg mL
1
. 2 M HCl was
added dropwise to solubilise the chitosan and to adjust the pH
of the solutions to either pH 2 or pH 4. 3-Glycidoxypropyl tri-
methoxysilane (GPTMS, 0.183 g, 0.8 mmol) was added dropwise
to the solutions to give a chitosan monomer : GPTMS ratio of
4 : 1. Aliquots of 2 mL were taken at time points of 5 min, 4 h
and 24 h, put into 7 mL polystyrene moulds and immediately
quenched in liquid nitrogen for 10 min to halt the reaction. The
frozen samples were transferred to a CoolSafe 100-4 freeze-drier
tted to a Vacuubrand RZ6 vacuum pump operating at 110

C
and an ultimate total vacuum of 1 10
2
mbar for 2 days until
dry.
For solution NMR experiments, 75 mg chitosan was dis-
solved in 5 mL D
2
O and 2 M DCl (40% DCl in D
2
O diluted
further in D
2
O. All with a deuteration degree of 99.99%) was
added drop-wise to adjust the pH to pH 6, pH 4 and pH 2.
0.7 mL aliquots of solution were taken at 5 min, 4 h, 24 h and 72
h and placed into NMR tubes. All solution NMR spectra were
recorded immediately.
Chemical characterisation of functionalisation reaction
One dimensional (1-D) spectra were acquired at 295.8 K using a
Bruker Avance III 400 HD spectrometer (400.32 MHz
1
H
frequency) equipped with a bbfo/5 mm probe running TopSpin
3.2 soware. All
1
H spectra were measured with 16 384 complex
points and a spectral width of 8006.4 Hz (20 ppm centred at
6 ppm). Experiments used a 30

pulse angle and a 1 s relaxation

delay. The number of transients co-added was 16. An expo-
nential multiplication apodization function of 0.3 Hz and 1
order of zero lling were applied prior to Fourier trans-
formation. All
13
C spectra were acquired with 32 768 complex
points and a spectral width of 26 315.8 Hz (260 ppm centred at
90 ppm). Experiments used a 30

pulse angle, a relaxation delay

of 1 s and
1
H decoupling throughout. The number of transients
co-added was 480. An exponential multiplication apodization
function of 2 Hz was applied prior to Fourier transform.
Quantitative HSQC experiments based on a previously
described method,
38
were acquired at 298 K using a Bruker
Avance II 500 spectrometer (500.13 MHz
1
H frequency) equip-
ped with a bbo/5 mm probe and running TopSpin 2.1 soware.
Spectra were collected with spectral widths of 3001.2 Hz (6 ppm
centred at 3 ppm) and 12 576.41 Hz (100 ppm centred at 50
ppm) for the
1
H- and
13
C-dimensions, respectively. The number
of acquired complex points was 2048 for the
1
H dimension. A
recycle delay of 5.34 s (5 s relaxation delay and 0.34 s acquisition
time) was employed. Long and short delays of 5.92 ms and 2.94
ms were used for the suppression of J-dependence of polariza-
tion transfer as recommended. The number of transients co-
added was 64, and 100 time increments were recorded in the
13
C-dimension. A squared cosine bell apodization function was
applied to both dimensions. Prior to Fourier transformation the
1
H-dimension had 1 order of zero lling and the
13
C-dimension
was linearly predicted to 1024 points.
One pulse
29
Si Magic Angle Spinning (MAS) NMR spectra
were acquired at natural abundance on a 7.05 T Chemagnetics
InnityPlus spectrometer operating at 56.59 MHz. The samples
were spun at 5 kHz using a Bruker 7 mm HX double channel
probe. A p/4 pulse length of 6 ms and a recycle time of 240 s were
used. The spectra were referenced to TMS using a secondary
reference kaolinite with a resonance at
29
Si d: 92 ppm. The
acquisition time was typically 24 hours. Peak tting was used to
calculate the relative abundance of silica T
n
and Q
n
species.
From this, the degree of condensation (D
c
) was calculated as
follows:
D
c

4Q
4
3Q
3
2Q
2
4

3T
3
2T
2
3

100%
where Q
n
is the abundance of Q
n
species and T
n
is the abun-
dance of the appropriate T
n
species.
1
H
15
N Cross Polarisation (CP)-MAS NMR spectra were
acquired at natural abundance on a 11.75 T Bruker Avance-III
500 operating at 500.10 MHz for
1
H and 50.68 MHz for
15
N. The
samples were spun at 10 kHz using a Bruker 4 mm HXY triple
channel probe operating in double channel mode. The
1
H p/2
pulse length was 2.5 ms and the contact time was 500 ms while
the recycle delay was 3 s. The spectra are referenced to MeNO
2
in
CDCl
3
using histidine as a secondary reference with peaks at
15
N d: 333.1, 204.3 and 191.0 ppm. The acquisition time
was typically 24 hours.
Fourier transform infrared spectroscopy (FTIR) was carried
out to investigate the structure of the functionalised chitosan
and hybrids using a Thermo Scientic Nicolet iS10 Spectrom-
eter operating in attenuated reectance (ATR) mode. All
samples were prepared for analysis by grinding to a ne powder.
Spectra were taken in the range 4504000 cm
1
at a resolution
of 2 cm
1
and averaged over 16 scans.
Synthesis of chitosansilica hybrid monoliths
Chitosan solutions were prepared by dissolving 1.5 g of chitosan
in 87 mL deionised water with the pH adjusted to either pH 2 or
pH 4 by adding 2 M HCl drop-wise. GPTMS (0.55 g, 2.3 mmol)
was added drop-wise to the solution to give a 4 : 1 molar ratio of
670 | J. Mater. Chem. B, 2014, 2, 668680 This journal is The Royal Society of Chemistry 2014
Journal of Materials Chemistry B Paper
P
u
b
l
i
s
h
e
d

o
n

0
3

D
e
c
e
m
b
e
r

2
0
1
3
.

D
o
w
n
l
o
a
d
e
d

b
y

U
n
i
v
e
r
s
i
d
a
d
e

F
e
d
e
r
a
l

d
a

P
a
r
a
i
b
a

o
n

2
3
/
0
5
/
2
0
1
4

1
6
:
3
6
:
1
4
.

View Article Online
chitosan monomer units : GPTMS. The reaction mixture was
stirred at room temperature for 24 h. An appropriate amount of
tetraethoxysilicate (TEOS) to give hybrids with compositions of
either 65 wt% organic or 35 wt% organic was placed in a beaker
with deionised water at pH 2 to give an R ratio of 4 (molar ratio
of water : TEOS). The resultant two-phase solution was vigor-
ously stirred for 1 h to induce hydrolysis. Hydrolysed TEOS was
added to the functionalised chitosan and stirred for 1 h to form
hybrid sols. The sol was cast into Nalgene polymethylpentene
(PMP) containers, sealed and allowed to gel. The gels were then
aged at 40

C for 3 days before the lids were loosened and the
samples were dried at 40

C over the course of two weeks.
Synthesis of chitosansilica hybrid scaolds
Hybrid scaolds were prepared by dissolving 1.5 g chitosan in
87 mL deionised water to give a concentration of 17 mg mL
1
.
To prevent hydrolysis of the chitosan chains but still promote
the functionalisation reaction, the pH was set to pH 4 by adding
2 M HCl dropwise to the solution. GPTMS (0.55 g, 2.3 mmol)
was added drop-wise to the solution and stirred for 24 h at room
temperature. TEOS was hydrolysed as described for monolith
synthesis and added to give hybrid sols of 65 wt% and 50 wt%
organic compositions. 35 wt% organic scaolds could not be
fabricated due to mechanical fragility on handling. 30 mL of the
sol was cast into Nalgene PMP moulds, sealed and allowed to
gel. The gels were aged at 40

C for 3 days before being frozen at
20

C, 80

C or by being submerged in liquid nitrogen. The
frozen gels were transferred to a CoolSafe110-4 freeze-drier and
dried for one week.
Characterisation of the monolith and scaold morphology
The morphology of the surface and the cross-section of the
hybrid monoliths were characterised using a scanning electron
microscope (SEM, Leo 1525) equipped with a GEMINI eld
emission column. The SEM operating voltage was 5 kV with a
working distance between 4 and 6 mm to an Inlens secondary
electron detector (SEI) with a 30 mm aperture for high resolution
imaging. Samples were prepared for analysis by sputter coating
with chromium at 30 mA for 2 min. The pore structure of the
scaolds was imaged using a JEOL JSM 5610 LV microscope in
secondary electron mode with an accelerating voltage of 515 kV
and a working distance of 1012 mm. 1 mm thick sections xed
to stubs with carbon tape were sputter coated with gold for 2
min at 30 mA.
X-ray microtomography (mCT) was performed on the samples
using a Pheonix nanotom (GE Measurement&Control, Wun-
storf, Germany) with 9 mm per voxel resolution. 3D image
analysis methods developed in-house were applied to quantify
the pore and interconnect diameter size distributions.
3941
To validate the mCT data, mercury porosimetry was carried
out on approximately 1 cm
3
samples using a PoreMaster 33
Quantachrome porosimeter in both high and low pressure
modes. Modal pore diameters were calculated from pore size
distributions produced using Poremaster 8.01 soware. Tortu-
osity was also calculated for the scaolds using this soware.
The porosity of the scaolds was calculated from the bulk and
skeletal densities:
% porosity

1
bulk density
skeletal density

100
where bulk density was calculated by mass and dimensional
analysis and skeletal density was measured by helium pycn-
ometry using a Quantachrome Ultrapycnometer 1000.
Mechanical testing
The mechanical behaviour under compression of the monolith
samples was investigated for 5 samples using an Instron 5866
with a load cell of 1 kN and a strain rate of 0.5 mm min
1
.
For the corresponding analysis of scaolds, a Zwick/Roell
Z2.5 machine, tted with a load cell of 2 kN, was used at a strain
rate of 1 mm min
1
. 10 samples were cut into cylinders of 8 mm
diameter and heights of 5 mm using a biopsy punch.
Dissolution testing
Simulated body uid (SBF) was prepared as described by
Kokubo and Takadama.
42
Hybrid monolith samples were
immersed in SBF at a ratio of 75 mg : 50 mL in sealed 125 mL
polyethylene containers. Hybrid scaolds were immersed at a
ratio of 200 mg : 100 mL SBF and placed under vacuum for 5
minutes to ensure that the scaolds were fully immersed in
solution. The solutions were agitated at 120 rpm at 37

C in an
orbital incubator shaker (Newbrunswick Scientic Classic
Series C24). Aer 2, 4, 8, 24, 168, 336 and 672 h, 1 mL of solution
was removed for analysis and replaced with fresh SBF. The
solutions were analysed by inductive coupled plasma optical
emission spectroscopy (ICP-OES, Thermo iCAP 6300) to deter-
mine the Si, Ca and P concentrations. Each sample was
repeated in triplicate. At the end of the study, the solids were
recovered by ltration, rinsed with deionised water and acetone,
and dried overnight at 40

C for further analysis by FTIR and
SEM.
Separate scaolds (200 mg) were immersed in SBF (100 mL)
and removed aer 72, 168 and 672 h. The samples were ltered,
rinsed with deionised water and acetone, and then dried over-
night at 40

C. To determine the organic content remaining in
the scaolds aer immersion, the samples were ground to a ne
powder and approximately 15 mg was placed in a platinum
crucible and heated to 800

C in a NETZSCH STA 449C dual
thermogravimetric analysis-dierential scanning calorimeter. A
blank platinum crucible was used as a reference. A heating rate
of 10

C min
1
was used in continuously owing air.
Results and discussion
Characterisation of functionalisation reaction
The rst step towards synthesising chitosansilica hybrids was
to understand the reaction of GPTMS with chitosan. Initially,
the functionalisation reaction between chitosan and GPTMS
was followed using
1
H NMR in D
2
O adjusted to pH 2, 4 or 6 with
2 M DCl/D
2
O. A methanol peak, resulting from the hydrolysis of
the silane groups was observed (d
1
H 3.19 ppm) even at very
This journal is The Royal Society of Chemistry 2014 J. Mater. Chem. B, 2014, 2, 668680 | 671
Paper Journal of Materials Chemistry B
P
u
b
l
i
s
h
e
d

o
n

0
3

D
e
c
e
m
b
e
r

2
0
1
3
.

D
o
w
n
l
o
a
d
e
d

b
y

U
n
i
v
e
r
s
i
d
a
d
e

F
e
d
e
r
a
l

d
a

P
a
r
a
i
b
a

o
n

2
3
/
0
5
/
2
0
1
4

1
6
:
3
6
:
1
4
.

View Article Online
short time-points, indicating that the hydrolysis of the silane
groups was rapid. However at 5 min in solution at pH 4 and 6,
there were multiple peaks around d
1
H 0.53 ppm due to
incomplete hydrolysis of the silane groups (ESI Fig. S1).
At pH 2, the hydrolysis was so rapid, that no evidence of
partial hydrolysis was observed at 5 min. This is in agreement
with Gabrielli et al. who observed a pHdependence of the rate of
silane hydrolysis in GPTMS.
43
Peaks attributed to f1 and f2
protons of the epoxide ring at d
1
H 2.81 ppm and d
1
H 2.61 ppm
reduced in intensity over time, however, this occurred at a much
slower rate than silane hydrolysis. New peaks were observed in
the d
1
H 3.903.30 ppm region although there was considerable
overlap in the
1
H NMR spectra, making it hard to distinguish
the peaks. Using a combination of
13
C and HSQC (showing
1
H
and
13
C coupling through one bond) experiments allowed the
dierent species to be identied. This was conrmed by
repeating the HSQC experiment for GPTMS alone in D
2
O/DCl
aer 72 h at pH 2 where the epoxide ring was fully opened (ESI
Fig. S2). A fully assigned HSQC spectrum is shown in Fig. 3.
Peaks at (d
1
H 3.73 ppm, d
13
C 63.39 ppm), (d
1
H 3.48 ppm, d
13
C
59.39 ppm) and (d
1
H 3.41 ppm, d
13
C 59.39 ppm) were attributed
to the formation of a diol when epoxide rings are opened by
water in solution.
43
At longer time points, but at all pH values,
other signals were observed at (d
1
H 3.57 ppm, d
13
C 50.96 ppm)
and (d
1
H 3.57 ppm, d
13
C 50.96 ppm) which were attributed to
the reaction of epoxide ring with the primary amine of chitosan
(NH
2
) to form a secondary amine. No other reactions were
identied, suggesting that the only covalent coupling reaction
occurring between chitosan and GPTMS occurred at the primary
amine.
The use of quantitative HSQC experiments showed that the
extent of epoxide opening aer 24 h decreased as pH increased:
9, 68 and 98 mol% epoxide ring remained at pH 2, 4 and 6
respectively (Fig. 4a). This supports the observations of Gabrielli
et al. that the opening of the epoxide ring of GPTMS in water is
acid catalysed and hence slightly acidic conditions are required
for the reaction with nucleophilic species. Gabrielli et al. also
postulated that too much formation of diol would prevent
nucleophilic attack. In contrast with the prediction of Gabrielli
et al., altering the pH did not aect the relative numbers of diol
and secondary amine species formed: the percentage of primary
amines that formed secondary amines remained constant at
around 20% (Fig. 4b).
Analysis of
15
N MAS NMR of chitosan dissolved at pH 4,
quenched in liquid nitrogen and freeze dried showed clearly
that, in pure chitosan, there were two signals due to acetylated
and deacetylated forms of the chitosan monomer (Fig. 5a). Aer
24 h reaction with GPTMS at pH 4, the signal at d
15
N 350 ppm
split into two, indicating a third nitrogen species is present
(Fig. 5b).
This is unequivocal evidence that there was a reaction
between chitosan and GPTMS at the primary amine. It also
shows that the nucleophilic addition between the amine and
Fig. 3 Fully assigned quantitative HSQC NMR spectrum of chitosan
functionalised with GPTMS for 24 h at pH4 with corresponding
1
Hand
13
C 1D spectra showing the potential products and side reactions.
Fig. 4 The quantitative HSQC NMR experiments were used to
calculate, (a) mol% of unopened epoxide, secondary amine product
and diol side-product and (b) relative amounts of secondary amine
product and diol product of the reacted epoxide at pH 2, 4, and 6 for
24 h.
672 | J. Mater. Chem. B, 2014, 2, 668680 This journal is The Royal Society of Chemistry 2014
Journal of Materials Chemistry B Paper
P
u
b
l
i
s
h
e
d

o
n

0
3

D
e
c
e
m
b
e
r

2
0
1
3
.

D
o
w
n
l
o
a
d
e
d

b
y

U
n
i
v
e
r
s
i
d
a
d
e

F
e
d
e
r
a
l

d
a

P
a
r
a
i
b
a

o
n

2
3
/
0
5
/
2
0
1
4

1
6
:
3
6
:
1
4
.

View Article Online
the epoxide ring is the only covalent bonding which occurs
between the amine and GPTMS. However, it should be noted
that this does not rule out the possibility that hydrogen bonding
may occur between amine, amide or hydroxyl species or that all
of the epoxide groups will react.
FTIR spectra of the chitosan functionalised with GPTMS for
24 h at pH 2 and pH 4 (Fig. 6a) are very similar to the pure
chitosan FTIR spectrum. Minor dierences arise at 1507 cm
1
where the secondary amide peak reduced in intensity at pH 4.
This is potentially due to hydrogen bonding of the amine group
in chitosan, which is more prominent at pH 4 because fewer of
the amine groups were converted to secondary amines. There is
no evidence of the epoxide ring remaining at either pH2 or pH4
as the bands for COC stretching of GPTMS would be expected
at 909 cm
1
and 846 cm
1
.
31
This is potentially due to the small
amount of GPTMS used relative to the amount of chitosan and
diol formation which reduces the relative amount of epoxide
ring further. Mahony et al. showed in a silica/gelatin systemthat
the bands corresponding to unopened epoxide ring could not
be distinguished until a molar ratio of GPTMS to gelatin of 1500
was used (at pH 5).
20
Structural characterisation of hybrid monoliths
The chemical structure of the hybrids was characterised in
order to determine the eect of pH and organic content on the
monoliths. FTIR spectra of hybrid monoliths (Fig. 6b), fabri-
cated by combining hydrolysed TEOS with the chitosanGPTMS
solution at pH 4 or pH 2 to give a composition of 65 wt%
organic, show a strong SiOSi stretching band that appeared at
1020 cm
1
. The band at 934 cm
1
was attributed to non-
bridging SiOHbonds and appears more intense at pH2 than at
pH 4, indicating a more condensed network at pH 4. The
primary and secondary amide bands of chitosan were retained
at 1600 cm
1
and 1500 cm
1
. In a similar fashion to the func-
tionalised chitosan, at pH 4, the intensity of the secondary
amine reduced whereas little change was observed at pH 2.
Again, this may be attributed to more prominent hydrogen
bonding at pH 4.
29
Si MAS NMR can be used to quantify the connectivity of a
silica network. The nomenclature Q
n
is used to describe silica
species where the silicon is bonded by n bridging oxygens and 4
n non-bridging oxygens, whereas T
n
is used to describe a
silicon atom bonded to a carbon (as in GPTMS) with n bridging
oxygens with 3 n non-bridging oxygens.
29
Si MAS NMR spectra
showed that the hybrid monoliths had a partially condensed
silica network comprising of distinct T
n
and Q
n
species which
correspond to CSi(OSi)
n
(OH)
3n
and Si(OSi)
n
(OH)
4n
respectively.
44
Peak tting of the one pulse MAS
29
Si NMR spectra allowed
quantication of each of the silicon species present in 65 wt%
organic hybrids (spectra shown in Fig. 7 and calculated
percentage abundance of silicon species in Table 1). In agree-
ment with the FTIR results, the hybrids synthesized at pH 4
were more highly condensed than at pH 2, as indicated by the
higher numbers of Q
4
and T
3
species present. In fact, at pH 4
there were no Q
2
species present whereas there were 5.0 0.4%
Fig. 5
15
NMAS NMR of (a) pure chitosan and (b) chitosan reacted with
GPTMS at pH 4 for 24 h.
Fig. 6 (a) FTIR spectra of pure chitosan and chitosan functionalised
with GPTMS at pH 2 and 4. (b) FTIR spectra of pure chitosan and
chitosansilica hybrid monoliths with 65 wt% organic where the
functionalisation step was carried out at pH 2 and 4.
This journal is The Royal Society of Chemistry 2014 J. Mater. Chem. B, 2014, 2, 668680 | 673
Paper Journal of Materials Chemistry B
P
u
b
l
i
s
h
e
d

o
n

0
3

D
e
c
e
m
b
e
r

2
0
1
3
.

D
o
w
n
l
o
a
d
e
d

b
y

U
n
i
v
e
r
s
i
d
a
d
e

F
e
d
e
r
a
l

d
a

P
a
r
a
i
b
a

o
n

2
3
/
0
5
/
2
0
1
4

1
6
:
3
6
:
1
4
.

View Article Online
present at pH 2. Calculation of the degree of condensation (D
c
)
gave values of 92.7% and 90.2% for pH 4 and 2 respectively. The
more condensed network is due to the fact that at pH <2.2, the
transition state of condensation is stabilised by the ethoxy and
methoxy groups of TEOS and GPTMS. The partially hydrolysed
silica precursor condenses faster, leading to chains of silica
network with a large number of non-bridging oxygens. The
opposite is true at pH >2.2, where fully hydrolysed precursors
condense fastest, leading to highly condensed silica networks
with fewer non-bridging species.
45
Repeating
29
Si MAS NMR for
the functionalised chitosan shows only T
n
species, as expected
as there was no TEOS present (ESI Fig. S3). However, it was
observed that within 5 min, condensation had occurred
between the GPTMS molecules so that at pH 2 up to 60% of the
GPTMS was present in a T
3
form (ESI Table S1). This would
render the molecule unable to condense further when TEOS is
introduced, potentially leading to two distinct silica networks
that do not interpenetrate. The signicance of this is unknown
and further investigation is required to establish the degree of
interaction between the two networks.
SEM images of the fracture surfaces of the monoliths fabri-
cated with 35 and 65 wt%organic at pH4 and pH2 all show that
no macroscale phase separation occurred during hybrid
synthesis at any composition (Fig. 8). Agglomerated particle
morphologies, typical of that formed by the solgel process,
46
were observed. This is due to silica nanoparticles that agglom-
erate and fuse to form a mesoporous silica gel.
46
The apparent
particle diameters were similar for samples made at pH 2 and
pH 4 (compare Fig. 8a with b and 8c with d) but larger particles
are observed as organic content increased. The particle size of
the 35 wt% organic hybrids was more typical for solgel silica
microstructures so the larger particle size is likely due to chi-
tosan polymer coating the surface of the silica particles.
Mechanical and dissolution properties of monoliths
From compression tests, hybrid monoliths containing 35 wt%
organic exhibited brittle behaviour with a strain at fracture of 4
to 8%. Increasing the chitosan content reduced the brittle
character as shown by the deformation prior to fracture for 65
wt% organic monoliths, whereas 35 wt% organic monoliths
failed catastrophically (Fig. 9). The increase in chitosan content
also increased the strain at fracture to around 48%. This had the
eect of reducing the compressive modulus of the monoliths
Table 1 Percentage abundance of silicon species present in 65 wt%
organic hybrids functionalised at pH 4 and 2
pH Q
4
Q
3
Q
2
T
3
T
2
D
c
4 64.2 0.8 22.5 0.7 N/A 8.2 0.6 5.2 0.9 92.7
2 60.0 0.5 25.1 0.4 5.0 0.4 6.9 0.7 3.0 0.4 90.2
Fig. 7
29
Si MAS NMR spectra of 65 wt% organic hybrids synthesized at
(a) pH 4 and (b) pH 2 showing the peak tting used to calculate the
abundance of each silicon species.
Fig. 8 Fracture surfaces of hybrid monoliths imaged by SEM with (a
and b) 35 wt% organic and (c and d) 65 wt% organic contents and
functionalised at (a and c) pH 4 and (b and d) pH2. Aggregated particle
morphologies typical of solgel silica glasses are observed. molecule
unable to condense further when TEOS is introduced, potentially
leading to two distinct silica networks that do not interpenetrate. The
signicance of this is unknown and further investigation is required to
establish the degree of interaction between the two networks.
674 | J. Mater. Chem. B, 2014, 2, 668680 This journal is The Royal Society of Chemistry 2014
Journal of Materials Chemistry B Paper
P
u
b
l
i
s
h
e
d

o
n

0
3

D
e
c
e
m
b
e
r

2
0
1
3
.

D
o
w
n
l
o
a
d
e
d

b
y

U
n
i
v
e
r
s
i
d
a
d
e

F
e
d
e
r
a
l

d
a

P
a
r
a
i
b
a

o
n

2
3
/
0
5
/
2
0
1
4

1
6
:
3
6
:
1
4
.

View Article Online
considerably and reduced the scatter associated with their
failure. Although there was a slight dierence with pH of the
moduli for 65 wt% organic monoliths, there was no dierence
for 35 wt% organic samples. This could be because at low
organic content the silicate network predominates the
mechanical properties masking the eect of the chitosan.
Whereas, at high organic contents, the eect of incorporating
more highly functionalised chitosan, functionalised at pH 2 for
24 h and hence, more highly cross-linked, is to increase the
compressive modulus (Table 2).
Fig. 10a shows the dissolution proles for the 35 and 65 wt%
chitosan hybrids synthesized at pH 2 and pH 4 in SBF as a
function of the time as measured in triplicate by ICP-OES. For
all samples, the concentration of silicon increased with time
indicating the dissolution of the silica network. Samples
synthesized at pH 2 degraded more rapidly than samples
synthesized at pH 4. This is attributed to the higher network
connectivity observed at pH 4 compared with samples synthe-
sized at pH 2. While it might be expected that the higher silica
content would result in faster silica release, it was observed that
the fastest rate of silica release was observed for 65 wt% organic
monoliths. This suggests that a critical inorganic content is
required to produce a fully interconnected silica structure. In
this case, the high organic content disrupts the silica network
and promotes dissolution. There was little change in the
calcium and phosphorus ion concentrations (data not pre-
sented) with time which indicated that an apatite layer was not
formed on the monolithic surface. This was also conrmed by
FTIR and XRD and is to be expected as there was no calcium
present in the hybrid, necessary for bioactivity of the hybrid.
47,48
Chitosansilica hybrid scaold fabrication
Utilizing the information gained about the eect of function-
alisation pH on the chemical, mechanical, and dissolution
properties of chitosansilica hybrids the authors decided to
proceed using a functionalisation pH value of 4. This interme-
diate pH was chosen to prevent excessive diol formation of the
GPTMS or acid catalyzed chain scission of the chitosan polymer,
while maintaining appropriately acidic conditions to catalyse
the nucleophilic reaction and ensure viable reaction times (at
pH 6 the reaction takes several days).
Scaolds were synthesized following the same initial method
as that used for monolith fabrication, where functionalised
chitosan was introduced into a hydrolysed TEOS solution.
However, compositions of 65 wt% and 50 wt% organic were
chosen as it was found that scaolds with lower organic
contents were very fragile and disintegrated to a powder on
handling. The hybrid sols were allowed to gel, frozen at 20

C,
80

C or quenched in liquid nitrogen, and subsequently
Table 2 Mechanical properties of hybrid monoliths
pH Organic (wt%)
Compressive
modulus (MPa)
Strain at fracture
(%)
4 35 1628 375 4 1
2 35 764 325 8 1
4 65 131 30 48 2
2 65 549 201 47 5
(Mean SD, n 5)
Fig. 9 Representative stress strain curves for the monolith hybrids
showing the change in failure mechanism from catastrophic brittle
failure for 35 wt% organic to plastic deformation for 65 wt% organic
monoliths due to the increase in chitosan.
Fig. 10 (a) Dissolution proles of silica monolith hybrids with varying
wt% organic and synthesis pH over 4 weeks in SBF as measured in
triplicate by ICP-OES. (b) Dissolution prole of silicon from 65 wt%
organic and 35 wt% organic hybrid scaolds over 4 weeks in SBF as
measured in triplicate by ICP-OES (mean SD).
This journal is The Royal Society of Chemistry 2014 J. Mater. Chem. B, 2014, 2, 668680 | 675
Paper Journal of Materials Chemistry B
P
u
b
l
i
s
h
e
d

o
n

0
3

D
e
c
e
m
b
e
r

2
0
1
3
.

D
o
w
n
l
o
a
d
e
d

b
y

U
n
i
v
e
r
s
i
d
a
d
e

F
e
d
e
r
a
l

d
a

P
a
r
a
i
b
a

o
n

2
3
/
0
5
/
2
0
1
4

1
6
:
3
6
:
1
4
.

View Article Online
freeze-dried. Chitosan has been chosen for scaold synthesis by
freeze-drying as the polymer forms sheets between the ice
crystals as the sol is forced out of the solidifying pure water,
where ultimately the ice crystals form the interconnected pore
structure of the scaolds.
49,50
Hybrid scaold morphology
Investigation of the morphology of the scaolds by SEM
(Fig. 11) showed that reducing freezing temperature reduced
the pore diameters. This can be attributed to the higher degree
of supercooling that occurs at lower freezing temperatures,
hence increasing the nucleation rate of ice crystals. Although,
more ice crystals form, the lower temperatures means that the
growth of the crystals is slower, resulting in many small ice
crystals and hence smaller pores in the nal scaold. The pores
were elongated and angular with a certain degree of direction-
ality as the gels tended to freeze from the outside-in with a
protrusion forming in the centre where the ice forced the gel as
it expanded during freezing.
Pore interconnectivity and interconnect size is oen more
important that pore size. Mercury porosimetry uses a model to
obtain the diameters of pores that constrict the mercury intru-
sion as a function of pressure. Analysis of the modal pore
interconnect diameters by mercury porosimetry conrmed that
the interconnect diameter reduced as the freezing temperature
reduced. The scaolds frozen at 20

C had modal pore
diameters of 178 47 mm and 156 7 mm, 80

C were 150
39 mm and 140 15 mm, and those quenched in liquid nitrogen
were 21 12 mmand 23 20 mmfor 50 wt%and 65 wt%organic
respectively (Fig. 12).
A guide for a suitable interconnect diameter for bone tissue
engineering scaolds is 100 mm.
51
At 20

C and 80

C, the
interconnect diameters were well above 100 mm. Quenching in
liquid nitrogen caused a signicant decrease in pore intercon-
nect diameter. The interconnect diameters of 65 wt% organic
and 50 wt% organic scaolds were similar at each freezing
temperature, however, the total porosity of the scaolds varied
with composition (96.7 0.2% and 97.5 0.2% for 50 wt% and
65 wt% organic respectively, Table 3). This is due to the water
content of the gels prior to freeze-drying. The scaolds with
higher organic content contained relatively more chitosan
solution (17 mg mL
1
) and so also contain more water. When
the water is frozen and removed during freeze-drying, the ulti-
mate result is to increase the porosity of the scaolds.
mCT images of the 65 wt% organic scaolds frozen at 20

C
and 80

C, shown in Fig. 13, illustrate the angular and
Fig. 12 Modal pore interconnect diameters calculated from inter-
connect diameters determined by mercury porosimetry.
Table 3 Percentage porosity of scaolds with organic content and
freezing temperature
Organic content (wt%) Freezing temp. (

C) Porosity (%)
65 20 97.5 0.4
80 97.5 0.1
196 97.5 0.2
50 20 96.9 0.2
80 96.7 0.2
196 96.4 0.1
(Mean SD, n 10)
Fig. 13 X-ray microtomography (mCT) of 65 wt% organic scaold
frozen at (a) 20

C and (b) 80

C, illustrating the elongated and
irregular pore morphology typical of freeze-drying.
Fig. 11 Images of the morphology of 65 wt% organic and 50 wt%
organic hybrid scaolds formed by freeze drying at dierent
temperatures by SEM. The decreasing pore size as the freezing
temperature reduced can be observed clearly.
676 | J. Mater. Chem. B, 2014, 2, 668680 This journal is The Royal Society of Chemistry 2014
Journal of Materials Chemistry B Paper
P
u
b
l
i
s
h
e
d

o
n

0
3

D
e
c
e
m
b
e
r

2
0
1
3
.

D
o
w
n
l
o
a
d
e
d

b
y

U
n
i
v
e
r
s
i
d
a
d
e

F
e
d
e
r
a
l

d
a

P
a
r
a
i
b
a

o
n

2
3
/
0
5
/
2
0
1
4

1
6
:
3
6
:
1
4
.

View Article Online
irregular pore morphologies that are characteristic of scaolds
fabricated via freeze-drying. Applying 3D image analysis tech-
niques, the modal pore diameter of the 20

C 65 wt% organic
scaold was 313 mm and the modal interconnect diameter was
189 mm, which is in good agreement with the mercury poros-
imetry data. The images also showed that the scaolds were well
interconnected, important for tissue ingrowth and vasculariza-
tion. The mean tortuosity of the scaolds, another property
which may be important for successful regeneration of tissue,
was measured by mercury porosimetry as 1.93 0.23, 1.65
0.24, and 1.37 0.31 for 20

C, 80

C, and 196

C scaolds
respectively. This is within the range reported for cancellous
bone by Pakula et al. of 1.1 to 2.8.
52
Mechanical behaviour of the chitosansilica hybrid scaolds
The mechanical properties of the scaolds were investigated
under compression and the data is presented in Table 4.
A slight increase in the compressive modulus was observed
at 50 wt% organic compared with 65 wt% organic, however, due
to the highly porous nature of the scaolds, there was a large
degree of scatter within the data and the dierence was not
statistically signicant. The strain at failure did not vary with
freezing temperature although a small increase in compressive
modulus and compressive strengths was observed for samples
quenched in liquid nitrogen. At 87.5, 69.9 and 143.0 kPa for 20

C, 80

C and liquid nitrogen 50 wt% organic hybrids
respectively and 80.8, 62.0 and 103.0 kPa for 20

C, 80

C and
liquid nitrogen 65 wt% organic hybrid scaolds respectively,
the compressive strengths are far too low for load sharing
applications for bone regeneration as originally intended. This
is due to the very high porosities of the scaolds. The freeze
drying method does not give control of percentage porosity.
Given the promising mechanical properties of the monolith
samples, if the porosity were reduced, then the compressive
strengths may be increased, making them more suitable for
bone regeneration scaolds. Alternatively, these scaolds may
be used in non-load sharing applications such as cartilage
regeneration. These scaolds may be particularly attractive for
cartilage regeneration due to the elongated pore morphologies
and since chitosan has a similar structure to anionic glycos-
aminoglycans found in articular cartilage.
53
Dissolution behaviour of hybrid scaolds
The silicon release in SBF as measured in triplicate by ICP-OES
(Fig. 10b) was very rapid for both the 65 wt% and 50 wt%
organic scaolds. The fastest rate of silicon release was up to 8
h, with the silicon concentration in solution plateauing at
around 80 g L
1
and 90 g L
1
for 50 and 65 wt% organic
respectively aer 24 h. As with the monolith hybrid samples,
greater silicon release was observed for higher organic content
hybrids due to disruption of the silica network by the organic
component. Phosphorus and calcium ion concentrations did
not vary over the timescale of the experiment (data not pre-
sented) and so it can be concluded that no apatite formed on
the sample surfaces, as expected.
FTIR analysis of the remaining solids aer 4 weeks in SBF
(Fig. 14) showed that the amide I and II bands were retained
although there was a signicant reduction in the intensity of the
amide II band. This indicates that there was still chitosan
remaining in the hybrid aer the dissolution study, conrmed
by thermogravimetric analysis (TGA, ESI Fig. S4). The weight
loss by TGA between 200

C and 600

C of the 50 wt% organic
scaold prior to immersion in SBF due to combustion of the
organic component was 38 wt%. Aer 72 h immersion, this
increased to 40 wt%and then remained constant at 1 w and 4 w.
This suggests that there is rapid silica dissolution within the
rst 72 h, as also indicated by the ICP-OES dissolution proles,
Table 4 Table Mechanical properties of freeze-dried hybrid scaolds
Organic
content
(wt%)
Freezing
temp. (

C)
Compress
modulus (MPa)
Failure
stress (kPa)
Strain at
failure (%)
65 20 0.85 0.32 80.8 28.9 11.9 3.9
80 0.73 0.29 62.0 17.6 11.6 4.8
196 1.37 0.64 103.0 45.2 8.7 3.2
50 20 1.06 0.50 87.5 41.9 11.9 6.4
80 0.91 0.40 69.9 21.3 7.8 2.7
196 1.08 0.14 143.0 71.3 14.5 7.5
(Mean SD, n 10)
Fig. 14 FTIR of hybrid scaolds before and after 4 w immersion in SBF
of (a) 65 wt%organic and (b) 50 wt%organic scaolds. The presence of
amide I and II bands indicates chitosan remains in the scaolds after
immersion.
This journal is The Royal Society of Chemistry 2014 J. Mater. Chem. B, 2014, 2, 668680 | 677
Paper Journal of Materials Chemistry B
P
u
b
l
i
s
h
e
d

o
n

0
3

D
e
c
e
m
b
e
r

2
0
1
3
.

D
o
w
n
l
o
a
d
e
d

b
y

U
n
i
v
e
r
s
i
d
a
d
e

F
e
d
e
r
a
l

d
a

P
a
r
a
i
b
a

o
n

2
3
/
0
5
/
2
0
1
4

1
6
:
3
6
:
1
4
.

View Article Online
whereas chitosan dissolution was slower. However, aer the
rst 72 h, the two components are released at the same rate so
that the relative composition remains constant up to 4 w in SBF.
Congruent dissolution, seen here aer 72 h, is one of the
dening features of a successful hybrid material and so this is a
promising result for the long term mechanical and chemical
stability of the chitosansilica hybrid.
Although the assessment of biological activity is beyond the
scope of this article, similar chitosanGPTMS systems have been
studied previously in vivo and in vitro.
1012,36,37,54
The good prolif-
eration of osteoblastic MG63 cell cultures on chitosansilica
hybrid membranes and freeze dried scaolds with varying
GPTMS and TEOS contents showed that the hybrid materials
were biocompatible.
10,11,37
Compared with pure chitosan scaolds
and membranes, the hybrid materials showed better prolifera-
tion and multilayers of well spread MG63 cells aer 6 days in cell
culture,
10
however, the type of silica species present aected the
behaviour of the cells, with an increase in TEOS promoting
osteodierentiation rather than proliferation as seen in hybrids
with high GPTMS contents but no TEOS.
37
Scaolds freeze dried
at 20

C exhibited cell penetration deep inside the material,
indicating good interconnectivity and permeability.
11
In vivo
studies were carried out in adult female Wistar rats to determine
the biocompatibility of chitosanGPTMS freeze-dried scaolds
and membranes.
54
For each animal, three 2 2 cmsamples were
implanted into 3 cm long dorsal incisions and were recovered
aer 1, 2, 4 and 8 weeks. From the results of these studies, the
authors are condent that the chitosansilica hybrid materials
presented here would be suitable for tissue regeneration appli-
cations, particularly the highly porous freeze dried scaolds.
Conclusions
Summary of eect of pH on monolith hybrids
A combination of solution and solid state NMR techniques
showed a reaction between the epoxide ring of GPTMS and
chitosan at the primary amine. Following the reaction at three
dierent pH values has shown that this reaction was acid
catalyzed, with signicantly more epoxide ring opening at pH 2
than at pH 4 or 6. However, it was also shown that an unwanted
side reaction occurred between water and the epoxide ring
resulting in diol formation and that this was the dominant
reaction at all pH values. Hydrolysis of the methoxysilane
groups of GPTMS was rapid under acidic conditions, however,
condensation occurred simultaneously, so that within 5 min T
3
species are present in GPTMS. Fabricating monolith hybrids
was achieved by introducing the functionalised chitosan into a
sol of hydrolysed TEOS. The silica network of the monoliths was
less condensed when chitosan was functionalised at pH 2
compared with those functionalised at pH 4. This had the eect
of increasing the rate of silica dissolution in SBF for the pH 2
sample. The eect of pHon mechanical properties was minimal
at 35 wt% organic as the brittle nature of the silica phase
appeared to predominate. However, at 65 wt% organic, the
organic phase had a more signicant eect on the mechanical
properties as the elongation at failure was increased from 7 to
40%. The samples fabricated at pH 2, which had a greater
degree of coupling between the chitosan and GPTMS showed a
slight increase in compressive modulus.
Summary of the fabrication and characterisation of hybrid
scaolds
Chitosansilica hybrid scaolds were fabricated by combining
the solgel process with a freeze-drying step. Chitosan was
functionalised using pre-determined optimum pH conditions
and compositions of 50 wt% and 65 wt% organic. Freezing
temperatures had a dramatic eect on the modal pore inter-
connect diameter. Scaolds fabricated by quenching in liquid
nitrogen had interconnect diameters of 2023 mm which is too
small for tissue engineering applications. Scaolds frozen
at 20 and 80

C are suitable as they have pore interconnects
well in excess of 100 mm, the critical value required for tissue
engineering scaolds. The compressive strengths of the scaf-
folds were too low to be used in load-sharing applications,
primarily due to their high porosities of 9697%. Reducing the
porosity will increase the compressive strengths of the scaolds
for alternative applications, such as non-load bearing cartilage
regeneration, may be more appropriate. A 4 weeks dissolution
study in SBF showed that silicon release was rapid within the
rst 24 h, but aer this time the chitosan and silica are released
at the same rate so that the relative composition of the hybrid
remains unchanged aer 72 h up to 4 weeks. This is an
important result that points towards long term mechanical
stability and chemical activity of the scaolds.
Here for the rst time:
A combination of solution and solid state NMR techniques
have been used to probe the functionalisation reaction between
chitosan and GPTMS.
It has been shown that covalent bonding occurs between
the primary amine of chitosan and the epoxide of GPTMS to
form a secondary amine, allowing covalent coupling between
chitosan and a silica network.
The extent of reaction at dierent pH values was quantied
to show that both the reactions of GPTMS, with water and with
chitosan, are acid catalyzed and that the relative amounts of
product and side-product does not depend on pH.
That functionalisation pH was shown to have an impact on
the mechanical properties of hybrids at 65 wt% where the
properties of the organic component become more dominant.
That high organic content was shown to disrupt the silica
network, speeding up the rate of silica dissolution in both
monolith and scaold hybrids.
The interconnect diameters were quantied for freeze-
dried chitosan scaolds and conrmed that 20 and 80

C are
appropriate freezing temperatures for fabricating tissue engi-
neering scaolds.
Chitosan and silicon were shown to be released congru-
ently when immersed in SBF for up to 4 w.
Acknowledgements
The authors would like to thank Mr Peter Haycock, Department
of Chemistry, Imperial College London, for carrying out the
678 | J. Mater. Chem. B, 2014, 2, 668680 This journal is The Royal Society of Chemistry 2014
Journal of Materials Chemistry B Paper
P
u
b
l
i
s
h
e
d

o
n

0
3

D
e
c
e
m
b
e
r

2
0
1
3
.

D
o
w
n
l
o
a
d
e
d

b
y

U
n
i
v
e
r
s
i
d
a
d
e

F
e
d
e
r
a
l

d
a

P
a
r
a
i
b
a

o
n

2
3
/
0
5
/
2
0
1
4

1
6
:
3
6
:
1
4
.

View Article Online
quantitative HSQC experiments. This research has been funded
by the EPSRC (EP/E057098/1, EP/E051669/1 and EP/I020861/1)
and the Department of Materials, Imperial College London.
EMV was a Natural Sciences and Engineering Research Council
of Canada (NSERC), Canadian Centennial Scholar, MS was
supported by Ficyt under the Argo program. JVH and MES
acknowledge support for the solid-state NMR facilities at War-
wick used in this research which were funded by EPSRC and the
University of Warwick. NMR was also partially funded through
the Birmingham Science City projects which were supported by
Advantage West Midlands (AWM) and the European Regional
Development Fund (ERDF). JVH and MES acknowledge EPSRC
support for FR via project EP/I004688/1.
Notes and references
1 R. Burge, B. Dawson-Hughes, D. H. Solomon, J. B. Wong,
A. King and A. Tosteson, J. Bone Miner. Res., 2007, 22, 465
475.
2 L. L. Hench and J. M. Polak, Science, 2002, 295, 10141017.
3 R. Langer and D. A. Tirrell, Nature, 2004, 428, 487492.
4 J. R. Jones, J. Eur. Ceram. Soc., 2009, 29, 12751281.
5 M. M. Pereira, J. R. Jones and L. L. Hench, Adv. Appl. Ceram.,
2005, 104, 3542.
6 J. R. Jones, Acta Biomater., 2013, 9, 44574486.
7 E. M. Valliant and J. R. Jones, So Matter, 2011, 7, 50835095.
8 B. M. Novak, Adv. Mater., 1993, 5, 422433.
9 Y. Shirosaki, C. M. Botelho, M. A. Lopes and J. D. Santos, J.
Nanosci. Nanotechnol., 2009, 9, 37143719.
10 Y. Shirosaki, K. Tsuru, S. Hayakawa, A. Osaka, M. Lopes,
J. Santos, M. Costa and M. Fernandes, Acta Biomater.,
2009, 5, 346355.
11 Y. Shirosaki, T. Okayama, K. Tsuru, S. Hayakawa and
A. Osaka, Chem. Eng. J., 2008, 137, 122128.
12 Y. Shirosaki, K. Tsuru, S. Hayakawa, A. Osaka, M. A. Lopes,
J. D. Santos and M. H. Fernandes, Biomaterials, 2005, 26,
485493.
13 M. J. Simoes, A. Gartner, Y. Shirosaki, R. M. Gil da Costa,
P. P. Cortez, F. Gartner, J. D. Santos, M. A. Lopes,
S. Geuna, A. S. Varejao and A. C. Mauricio, Acta Med. Port.,
2011, 24, 4352.
14 G. Toskas, C. Cherif, R.-D. Hund, E. Laourine, B. Mahltig,
A. Fahmi, C. Heinemann and T. Hanke, Carbohydr. Polym.,
2013, 94, 713722.
15 E. M. Valliant, F. Romer, D. Wang, D. S. McPhail,
M. E. Smith, J. V. Hanna and J. R. Jones, Acta Biomater.,
2013, 9, 76627671.
16 G. Poologasundarampillai, C. Ionescu, O. Tsigkou,
M. Murugesan, R. G. Hill, M. M. Stevens, J. V. Hanna,
M. E. Smith and J. R. Jones, J. Mater. Chem., 2010, 20, 8952.
17 G. Poologasundarampillai, B. Yu, O. Tsigkou, E. Valliant,
S. Yue, P. D. Lee, R. W. Hamilton, M. M. Stevens,
T. Kasuga and J. R. Jones, So Matter, 2012, 8, 48224832.
18 M.-Y. Koh, C. Ohtsuki and T. Miyazaki, J. Biomater. Appl.,
2011, 25, 581594.
19 L. Ren, K. Tsuru, S. Hayakawa and A. Osaka, Biomaterials,
2002, 23, 47654773.
20 O. Mahony, O. Tsigkou, C. Ionescu, C. Minelli, L. Ling,
R. Hanly, M. E. Smith, M. M. Stevens and J. R. Jones, Adv.
Funct. Mater., 2010, 20, 38353845.
21 C. Gao, Q. Gao, Y. Li, M. N. Rahaman, A. Teramoto and
K. Abe, J. Appl. Polym. Sci., 2013, 127, 25882599.
22 S. V. Madihally and H. W. T. Matthew, Biomaterials, 1999, 20,
11331142.
23 M. Rinaudo, G. Pavlov and J. Desbri`eres, Polymer, 1999, 40,
70297032.
24 M. Rinaudo, G. Pavlov and J. Desbri`eres, Int. J. Polym. Anal.
Charact., 1999, 5, 267276.
25 S. Minami, M. Morimoto, Y. Okamoto, H. Saimoto and
Y. Shigemasa, in Materials Science of Chitin and Chitosan,
ed. T. Uragami and S. Tokura, Kodansha Ltd., Tokyo, 2006,
ch. 7, pp. 191217.
26 S.-H. Rhee, J.-Y. Choi and H.-M. Kim, Biomaterials, 2002, 23,
49154921.
27 A. Osaka, S. Hayakawa, K. Tsuru, S. Takashima, M. Kubo and
Y. Shirosaki, J. R. Soc., Interface, 2005, 2, 335340.
28 Y. Liu, Y. Su and J. Lai, Polymer, 2004, 45, 68316837.
29 A.-C. Chao, J. Membr. Sci., 2008, 311, 306318.
30 J. G. Varghese, R. S. Karuppannan and M. Y. Kariduraganavar,
J. Chem. Eng. Data, 2010, 55, 20842092.
31 P. Innocenzi, T. Kidchob and T. Yoko, J. Sol-Gel Sci. Technol.,
2005, 35, 225235.
32 S. S. Rashidova, D. S. Shakarova, O. N. Ruzimuradov,
D. T. Satubaldieva, S. V. Zalyalieva, O. A. Shpigun,
V. P. Varlamov and B. D. Kabulov, J. Chromatogr. B: Anal.
Technol. Biomed. Life Sci., 2004, 800, 4953.
33 F. Al-Sagheer and S. Muslim, J. Nanomater., 2010, 2010, 18.
34 S. Prochazkova, K. M. Varum and K. Ostgaard, Carbohydr.
Polym., 1999, 38, 115122.
35 L. Gabrielli, L. S. Connell, L. Russo, J. Jimenez-Barbero,
F. Nicotra, L. Cipolla and J. R. Jones, RSC Adv., 2014, 4,
18411848.
36 Y. Shirosaki, K. Tsuru, H. Moribayashi, S. Hayakawa,
Y. Nakamura, I. R. Gibson and A. Osaka, J. Ceram. Soc.
Jpn., 2010, 118, 989992.
37 Y. Shirosaki, K. Tsuru, S. Hayakawa, Y. Nakamura,
I. R. Gibson and A. Osaka, in Bioceramics Development and
Applications, ed. S. Kim, The Korean Society for
Biomaterials, 2009, vol. 22, pp. 217220.
38 S. Heikkinen, M. M. Toikka, P. T. Karhunen and
I. A. Kilpelainen, J. Am. Chem. Soc., 2003, 125, 43624367.
39 J. R. Jones, G. Poologasundarampillai, R. C. Atwood,
D. Bernard and P. D. Lee, Biomaterials, 2007, 28, 14041413.
40 R. C. Atwood, J. R. Jones, P. D. Lee and L. L. Hench, Scr.
Mater., 2004, 51, 10291033.
41 S. Yue, P. D. Lee, G. Poologasundarampillai and J. R. Jones,
Acta Biomater., 2011, 7, 26372643.
42 T. Kokubo and H. Takadama, Biomaterials, 2006, 27, 29072915.
43 L. Gabrielli, L. Russo, A. Poveda, J. R. Jones, F. Nicotra,
J. Jimenez-Barbero and L. Cipolla, Chemistry, 2013, 19,
78567864.
44 K. J. D. MacKenzie and M. E. Smith, Multinuclear Solid-State
Nuclear Magnetic Resonance of Inorganic Materials, Elsevier
Science, 2002.
This journal is The Royal Society of Chemistry 2014 J. Mater. Chem. B, 2014, 2, 668680 | 679
Paper Journal of Materials Chemistry B
P
u
b
l
i
s
h
e
d

o
n

0
3

D
e
c
e
m
b
e
r

2
0
1
3
.

D
o
w
n
l
o
a
d
e
d

b
y

U
n
i
v
e
r
s
i
d
a
d
e

F
e
d
e
r
a
l

d
a

P
a
r
a
i
b
a

o
n

2
3
/
0
5
/
2
0
1
4

1
6
:
3
6
:
1
4
.

View Article Online
45 J. D. Wright and N. A. J. M. Sommerdijk, Solgel materials:
chemistry and applications, Taylor & Francis Ltd, London, 2000.
46 S. Lin, C. Ionescu, K. J. Pike, M. E. Smith and J. R. Jones, J.
Mater. Chem., 2009, 19, 1276.
47 J. Zhong and D. C. Greenspan, J. Biomed. Mater. Res., 2000,
53, 694701.
48 K. Tsuru, C. Ohtsuki, A. Osaka, T. Iwamoto and
J. D. Mackenzie, J. Mater. Sci.: Mater. Med., 1997, 8, 157161.
49 S. Deville, E. Saiz, R. K. Nalla and A. P. Tomsia, Science, 2006,
311, 515518.
50 S. Deville, Adv. Eng. Mater., 2008, 10, 155169.
51 S. F. Hulbert, S. J. Morrison and J. J. Klawitter, J. Biomed.
Mater. Res., 1972, 6, 347374.
52 M. Pakula, F. Padilla, P. Laugier and M. Kaczmarek, J. Acoust.
Soc. Am, 2008, 123, 24152423.
53 A. Di Martino, M. Sittinger and M. V. Risbud, Biomaterials,
2005, 26, 59835990.
54 S. Amado, M. J. Simoes, P. A. S. Armada da Silva, A. L. Lus,
Y. Shirosaki, M. A. Lopes, J. D. Santos, F. Fregnan,
G. Gambarotta, S. Raimondo, M. Fornaro, A. P. Veloso,
A. S. P. Varejao, A. C. Maurcio and S. Geuna, Biomaterials,
2008, 29, 44094419.
680 | J. Mater. Chem. B, 2014, 2, 668680 This journal is The Royal Society of Chemistry 2014
Journal of Materials Chemistry B Paper
P
u
b
l
i
s
h
e
d

o
n

0
3

D
e
c
e
m
b
e
r

2
0
1
3
.

D
o
w
n
l
o
a
d
e
d

b
y

U
n
i
v
e
r
s
i
d
a
d
e

F
e
d
e
r
a
l

d
a

P
a
r
a
i
b
a

o
n

2
3
/
0
5
/
2
0
1
4

1
6
:
3
6
:
1
4
.

View Article Online