Poly(γ-glutamic acid)/silica hybrids with calcium incorporated in the silica network by use of a calcium alkoxide precursor

Article abstract of DOI:10.1002/chem.201304013

Current materials used for bone regeneration are usually bioactive ceramics or glasses. Although they bond to bone, they are brittle. There is a need for new materials that can combine bioactivity with toughness and controlled biodegradation. Sol-gel hybrids have the potential to do this through their nanoscale interpenetrating networks (IPN) of inorganic and organic components. Poly(γ-glutamic acid) (γ-PGA) was introduced into the sol-gel process to produce a hybrid of γ-PGA and bioactive silica. Calcium is an important element for bone regeneration but calcium sources that are used traditionally in the sol-gel process, such as Ca salts, do not allow Ca incorporation into the silicate network during low-temperature processing. The hypothesis for this study was that using calcium methoxyethoxide (CME) as the Ca source would allow Ca incorporation into the silicate component of the hybrid at room temperature. The produced hybrids would have improved mechanical properties and controlled degradation compared with hybrids of calcium chloride (CaCl₂), in which the Ca is not incorporated into the silicate network. Class II hybrids, with covalent bonds between the inorganic and organic species, were synthesised by using organosilane. Calcium incorporation in both the organic and inorganic IPNs of the hybrid was improved when CME was used. This was clearly observed by using FTIR and solid-state NMR spectroscopy, which showed ionic cross-linking of γ-PGA by Ca and a lower degree of condensation of the Si species compared with the hybrids made with CaCl ₂ as the Ca source. The ionic cross-linking of γ-PGA by Ca resulted in excellent compressive strength and reduced elastic modulus as measured by compressive testing and nanoindentation, respectively. All hybrids showed bioactivity as hydroxyapatite (HA) was formed after immersion in simulated body fluid (SBF). Organic/inorganic hybrid materials: Calcium methoxyethoxide (CME) was found to be an effective precursor for successfully introducing calcium into organic/inorganic hybrid materials (see figure). The Ca²⁺ ions were found to successfully cross-link the polymer chains, thereby avoiding a burst release of polymer and Ca²⁺ in buffered media. CME was found to distribute within the inorganic silica network thanks to the high reactivity of CME with hydroxyl groups.

Full text of DOI:10.1002/chem.201304013

DOI: 10.1002/chem.201304013
Full Paper
&
Hybrid Materials
Poly(g-glutamic acid)/Silica Hybrids with Calcium Incorporated in
the Silica Network by Use of a Calcium Alkoxide Precursor
Gowsihan Poologasundarampillai,*[a] Bobo Yu,[a] Olga Tsigkou,[a, b] Daming Wang,[a]
Frederik Romer,[c] Vineet Bhakhri,[a] Finn Giuliani,[a] Molly M. Stevens,[a, b] David S. McPhail,[a]
Mark E. Smith,[c, d] John V. Hanna,[c] and Julian R. Jones*[a]
Abstract: Current materials used for bone regeneration are
usually bioactive ceramics or glasses. Although they bond to
bone, they are brittle. There is a need for new materials that
can combine bioactivity with toughness and controlled bio-
degradation. Sol-gel hybrids have the potential to do this
through their nanoscale interpenetrating networks (IPN) of
inorganic and organic components. Poly(g-glutamic acid) (g-
PGA) was introduced into the sol-gel process to produce
a hybrid of g-PGA and bioactive silica. Calcium is an impor-
tant element for bone regeneration but calcium sources that
are used traditionally in the sol-gel process, such as Ca salts,
do not allow Ca incorporation into the silicate network
during low-temperature processing. The hypothesis for this
study was that using calcium methoxyethoxide (CME) as the
Ca source would allow Ca incorporation into the silicate
component of the hybrid at room temperature. The pro-
duced hybrids would have improved mechanical properties
and controlled degradation compared with hybrids of calci-
um chloride (CaCl2), in which the Ca is not incorporated into
the silicate network. Class II hybrids, with covalent bonds be-
tween the inorganic and organic species, were synthesised
by using organosilane. Calcium incorporation in both the or-
ganic and inorganic IPNs of the hybrid was improved when
CME was used. This was clearly observed by using FTIR and
solid-state NMR spectroscopy, which showed ionic cross-link-
ing of g-PGA by Ca and a lower degree of condensation of
the Si species compared with the hybrids made with CaCl2
as the Ca source. The ionic cross-linking of g-PGA by Ca re-
sulted in excellent compressive strength and reduced elastic
modulus as measured by compressive testing and nanoin-
dentation, respectively. All hybrids showed bioactivity as hy-
droxyapatite (HA) was formed after immersion in simulated
body fluid (SBF).
form.[1] The ability to bond with bone is thought to be due to
the formation of a hydroxycarbonate apatite (HCA) surface
layer on reaction with body fluid.[2] Using the sol-gel foaming
process, scaffolds have been developed that have an intercon-
nected pore structure suitable for vascularised bone ingrowth
with compressive strength in excess of 2 MPa.[3] However, as is
the case for other porous bioceramics, the brittle nature of bio-
active glasses limits their application to non-load-bearing bone
defects. The synthesis of organic/inorganic hybrids is one strat-
egy to mimic the natural nanocomposite structure of bone.
The aim is to create porous scaffold materials that are bioac-
tive, can share load with the bone and have tailored congruent
degradation. Hybrids are synthesised by incorporating a poly-
mer into the sol-gel process in an early (sol) stage of the pro-
cess, as the silica network is being formed, so that the inorgan-
ic and organic species interact at the nanoscale. Several hy-
brids have been created by sol-gel routes in the past 15 to
20 years.[4]
Introduction
Bioactive glasses bond to bone, have been shown to stimulate
osteogenesis and are available commercially in a particle
[a] Dr. G. Poologasundarampillai, Dr. B. Yu, Dr. O. Tsigkou, Dr. D. Wang,
Dr. V. Bhakhri, Dr. F. Giuliani, Prof. M. M. Stevens, Dr. D. S. McPhail,
Dr. J. R. Jones
Department of Materials, Imperial College London
South Kensington, London, SW7 2AZ (UK)
E-mail: gowshp@gmail.com
julian.r.jones@imperial.ac.uk
[b] Dr. O. Tsigkou, Prof. M. M. Stevens
Department of Bioengineering, Imperial College London
South Kensington, London, SW7 2AZ (UK)
[c] Dr. F. Romer, Prof. M. E. Smith, Prof. J. V. Hanna
Department of Physics, University of Warwick
Coventry, CV4 7Al (UK)
[d] Prof. M. E. Smith
Lancaster University
Vice-Chancellor’s Office, Lancaster, LA1 4YW (UK)
Hybrids are classified as class I or II according to the interfa-
cial bonding present between the organic and inorganic com-
ponents. Class I hybrids only have weak van der Waals forces,
hydrogen bonding and/or ionic bonding, whereas class II hy-
brids have strong covalent bonding,[4b] which is normally ach-
ieved by modifying the polymer so it can bond to the inorgan-
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/chem.201304013.
ꢀ 2014 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA.
This is an open access article under the terms of the Creative Commons At-
tribution License, which permits use, distribution and reproduction in any
medium, provided the original work is properly cited.
Chem. Eur. J. 2014, 20, 8149 – 8160
8149
ꢀ 2014 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
ic phase. Recently, organic/inorganic hybrids have been devel-
oped by using poly-g-glutamic acid (g-PGA) as the organic
component.[4p,5] A class II g-PGA/silica hybrid was achieved by
initially functionalising the g-PGA with glycidoxypropyltrime-
thoxy silane (GPTMS) before the polymer was added to the
silica sol. The epoxy ring of the GPTMS was hypothesised to
open and react with the carboxylic acid group of the g-PGA
(nucleophilic attack),[6] leaving the g-PGA functionalised with
the methoxysilane groups, which hydrolyse and undergo con-
densation with the SiꢀOH groups in the tetraethylorthosilicate
synthesis of class II calcium silicate/g-PGA hybrids. The proper-
ties of the hybrids synthesised with CME are compared with
samples made with calcium chloride.
Results and Discussion
Release of soluble Si and Ca ions from bioactive glass are ben-
eficial to the regeneration of bone.[9] Scaffold materials for
bone regeneration should be able to release these elements at
a controlled rate. Previously, Poologasundarampillai et al. elec-
(TEOS)-based sol, linking the g-PGA to the silica network by Siꢀ trospun silicate/poly-l-lactic acid hybrid material fibre mats by
OꢀSi bonds.
using CME as the calcium source.[10] There, the CME was mixed
with the electrospinning solution in the last seconds of the
procedure to avoid the premature gelation that is due to the
high reactivity of CME with water and hydroxyl groups. Here,
the challenge is to produce g-PGA/silica hybrids by using CME
as the calcium source when the silica content is ~50 wt% and
so that the calcium is incorporated homogeneously within the
organic and inorganic networks. Owing to the high reactivity
of CME with water in the sol-gel process, its incorporation was
not trivial; hence, this will be first explained in more detail.
Inclusion of calcium in the hybrid system is important as it is
vital to the formation of the hydroxycarbonated apatite (HCA)
layer. Together with Si, calcium is also thought to act as a sig-
nalling agent to stimulate differentiation of osteogenic cells to
bone cells as it is released from bioactive glass.[3a] However, tra-
ditional sol-gel calcium sources are not suitable for low-tem-
perature hybrid materials. Calcium nitrate is the most popular
calcium source for bioactive sol-gel glasses but a gel must be
heated to over 4008C to remove the nitrate byproducts.[7] As
biodegradable polymers will be degraded at such high tem-
peratures, an ideal calcium source would allow calcium incor-
poration at room temperature. Calcium chloride is another salt
precursor of calcium that was previously used to synthesise or-
ganic/inorganic hybrids.[4p,5b,8] Yu et al. showed that when calci-
um salts were used, Ca was not incorporated into the silica
network until a temperature of 4008C was reached.[7b] Poologa-
sundarampillai et al. also demonstrated that a moisture-assist-
ed drying was required to allow Ca to enter the organic net-
work (chelate) when using CaCl2 as the calcium source.[5b] One
alternative route is to introduce the calcium in the polymer
component,[5c] then introduce the calcium salt form of the
polymer into the sol. The advantage of this method over using
the free acid form of g-PGA was that the calcium salt g-PGA is
soluble in the sol, so no other solvent was needed. Drawbacks
were that the amount of calcium that could be incorporated
was low and the calcium was associated with the carboxylic
acid groups, which reduced the number of functional groups
available for covalent coupling, and that Ca was not incorpo-
rated with the inorganic silica network. Therefore, a new calci-
um source is required that does not release toxic byproducts
and can be incorporated into the hybrid system, both organic
and inorganic, at low temperature (e.g., lower than 608C). One
such calcium source is calcium methoxyethoxide (CME). Previ-
ously, we showed that for a sol-gel bioactive glass made with
CME as the calcium source, the Ca was incorporated into the
silica network at room temperature.[7b] The glass structure was
studied as a function of heat-treatment temperature by using
X-ray diffraction, which showed the diffraction patterns for the
glasses to be identical at temperatures from ambient to 6008C
for the glasses with CME as the calcium source. Ion dissolution
from the bioactive glasses also was observed not to change
after heat treatment at different temperatures up to 6008C,
suggesting that low-temperature incorporation of calcium was
successful when CME was the calcium source. Here, the aim is
to investigate the use of CME as the calcium source for the
Determination of hybrid synthesis procedure
Several systematic experiments were conducted to devise
a viable method of incorporating CME into the g-PGA/silica
hybrid. Several important observations were made. CME was
found to react with functionalised polymer solution (g-PGA re-
acted with GPTMS) and also with hydrolysed TEOS to cause
premature and heterogeneous gelation. This occurred immedi-
ately when CME was mixed with the functionalised polymer
solution (g-PGA+GPTMS) as calcium alkoxides have high reac-
tivity with carboxylic acid groups,[11] which are present on g-
PGA. When a reduced R ratio (R=moles of H2O/moles of TEOS)
of 2:1 was used for the hydrolysis of TEOS, gelation could be
slowed to several minutes (<20 min) depending on the speed
of addition of CME and the rate of mixing of the solution.
Therefore, when making Class II hybrids with CME, as shown in
Scheme 1, g-PGA was first functionalised with GPTMS in DMSO,
this was then added to hydrolysed TEOS to form the hybrid
precursor solution. CME was then added slowly; however, this
was also found to lead to rapid gelation, but by diluting CME
with DMSO at a 1:1 volume ratio and by adding the diluted
CME solution slowly to the hybrid precursor solution, the ho-
mogeneity of the reaction and gelation was improved. The ge-
lation time for these hybrids was 5 min. In contrast, the hybrid
systems made with CaCl2 as the calcium source gelled after
3 days (with HF also added). The faster gelation of the CME
hybrid system was due to the rapid hydrolysis and condensa-
tion of CME. The use of HF in the CME hybrid system did not
affect the gelation, an observation that might be due to HF
being inhibited by the presence of CME. Magic-angle spinning
(MAS) NMR spectrometry, X-ray diffraction, FTIR spectroscopy,
secondary ion mass spectrometry (SIMS) and SEM were used
to study the chemical, nano- and macro-structures of the hy-
brids produced with this synthesis route. The hybrids are de-
noted as 2ECCaME, where 2 corresponds to two moles of gluta-
Chem. Eur. J. 2014, 20, 8149 – 8160
www.chemeurj.org
8150
ꢀ 2014 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Scheme 1. Schematic of the 3-pot synthesis of Class II g-PGA/bioactive silica hybrids using calcium methoxyethyoxide (CME) as the calcium source.
mic acid (E) to GPTMS (C), with calcium (Ca) from either CME
(ME) or CaCl2 (Cl). Samples made without the addition of HF
are denoted as 2ECCaME*. Calcium-free hybrids are denoted as
2EC and were produced following the same procedure as that
for the synthesis of 2ECCaME, but no calcium was added.
(OSi)n(R)3ꢀn and Si(OSi)n(R)4ꢀn, respectively, where R is a non-
bridging oxygen).[12] The results of the Gaussian fitting of Tn
and Qn distributions to the data of all the 29Si MAS NMR spectra
are summarised in Table 1. The ratio of T3/Ttot and Q4/Qtot in
Table 1 show the degree of condensation of the GPTMS and
silica network respectively, and Dc represents the degree of
condensation of all SiꢀO species in the hybrid. Large variation
between the two ratios and Dc for the different samples sug-
gests that the hybrids had different atomic-scale structures,
a fact that could reveal information about the reaction and ge-
Atomic-scale structure
XRD patterns (Figure 1a) show that 2EC and 2ECCaME* were
amorphous but 2ECCaCl and 2ECCaME contained some low-in-
tensity crystalline peaks that
were identified as CaF2. This indi-
cates that addition of HF to the
calcium-containing samples led
to the formation of CaF2 crystals.
This was previously shown to
occur in 2ECCaCl by XRD and
TEM.[5b] It seems that CaF2 also
forms in the case of the 2ECCaME
sample, but the peaks were
broader and much more intense,
indicating that there were larger
numbers of smaller CaF2 crystals
present in the 2ECCaME sample
compared with 2ECCaCl. The for-
Table 1. Chemical shifts and relative proportions of Tn and Qn species in Class II g-PGA/bioactive silica hybrids
synthesised with calcium chloride (2ECCaCl) or calcium methoxyethoxide (2ECCaME) or without calcium (2EC)
with processing temperatures of 608C. Samples where HF was not added are marked with *. The errors in the
chemical shift (d) is ꢁ0.5 ppm, in the intensity is ꢁ1%, in the ratios is ꢁ0.05 and in Dc is ꢁ3%.
Samples
T2
T3
Q2
Q3
Q4
T3/Ttot Q4/Qtot Dc
d
I
d
I
d
I
d
I
d
I
[ppm] [%] [ppm] [%]
[ppm] [%]
[ppm]
[%]
[ppm]
[%]
2EC
2ECCaCl
2ECCaME
–
0.0 ꢀ65.4 11.7
–
0.0
ꢀ103.0 18.1 ꢀ112.4 70.2 1.0
ꢀ101.3 29.2 ꢀ109.8 44.6 0.6
ꢀ102.5 30.5 ꢀ111.4 49.7 0.9
0.8
0.6
0.6
0.4
95
87
90
83
ꢀ58.0 8.0 ꢀ65.1 13.0 ꢀ93.1 5.2
ꢀ59.0 2.0 ꢀ66.9 13.9 ꢀ94.4 3.8
2ECCaME
*
ꢀ61.4 6.9 ꢀ68.2 9.9
ꢀ95.0 13.4 ꢀ103.0 33.2 ꢀ112.2 36.5 0.6
mation of CaF2 by the reaction between HF and CME explains
why HF does not accelerate gelling when it is added to the
hybrid precursor solution as the fluoride is consumed to form
CaF2.
lation mechanism. The 2ECCaME and 2EC samples had a high
T3/Ttot ratio (>0.9) compared with the 2ECCaCl and 2ECCaME
*
samples (0.6), indicating that the GPTMS in 2ECCaME and 2EC
underwent condensation with hydrolysed TEOS or with itself
to a greater extent. For the TEOS-derived silica network, the
Q4/Qtot ratio was high for 2EC (0.8), the ratio was lower for
2ECCaCl and 2ECCaME (0.6) and finally 2ECCaME* had the lowest
value of 0.4. The 2EC sample has the highest T3/Ttot, Q4/Qtot
and Dc (95%) values, which suggests that the silica species in
2EC were highly condensed. The calcium-containing, HF-free
Silica-bonding structure
Figure 1b shows the 29Si MAS NMR spectra for the four sam-
ples. All the samples showed the presence of both Tn and Qn
species (Tn and Qn correspond to the structures of CSi-
Chem. Eur. J. 2014, 20, 8149 – 8160
www.chemeurj.org
8151
ꢀ 2014 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
a higher proportion of Q4 species,[13] but with the addition of
CME the precursor solution gels, trapping HF and all constitu-
ents in the gel. However, another reason for the increased Dc
of 2ECCaME may be due to the formation of CaF2. Calcium from
CME could enter the silicate network if HF was not added, but
instead reacted with Fꢀ to form CaF2, a process that reduced
the number of network modifiers in the silicate network, in-
creasing its connectivity. Therefore, it has the second highest
degree of condensation, that is, second only to 2EC, which has
no Ca.
Hybrid sample 2ECCaCl, which was produced with calcium
chloride as the calcium source, had a similar Dc to that of
2ECCaME, although not as much CaF2 is formed as in 2ECCaME
.
This suggests that the Ca from CaCl2 stayed in the precursor
solution and it had not entered the silica network. This is in
agreement with observations by Yu et al.[7b]
Chemical structure
Figure 1c shows the FTIR spectra of the hybrids. The bands at
1045 and 790 cmꢀ1, corresponding to SiꢀOꢀSi, and that at
947 cmꢀ1, corresponding to SiꢀO from the inorganic compo-
nent of the hybrids, were observed in all spectra. In addition,
no bands corresponding to the oxirane ring were observed, in-
dicating that ring opening was successful in all the hybrids. All
the samples exhibited the band corresponding to amide I and
II from g-PGA, whereas the band corresponding to the free car-
boxylic acid at 1725 cmꢀ1 (C=O) was only visible in the calci-
um-free 2EC sample. However, in the calcium-containing sam-
ples, the band corresponding to the carboxylate anion
(ꢀCOOꢀ) at 1580 cmꢀ1 was present. This indicates that the cal-
cium cross-linked the g-PGA at the carboxylic acid groups,
forming a Ca-g-PGA salt. The ionic cross-linking of the g-PGA
by Ca2+ could be beneficial in the hybrid as it could lead to
stronger hybrid materials.[5c] Qualitatively, the FTIR spectra also
show that a higher proportion of the carboxylic acids in g-PGA
Figure 1. Characterisation of Class II g-PGA/bioactive silica hybrids synthes-
ised with calcium chloride (2ECCaCl) or calcium methoxyethoxide (2ECCaME
or without calcium (2EC) with molar ratios of g-PGA/GPTMS of 2: (a) XRD
(b) 29Si MAS NMR spectra and (c) FTIR spectra. Samples where HF was not
added are marked with *.
)
were converted to carboxylate anions in the HF-free 2ECCaME
*
hybrid compared with 2ECCaCl and HF-containing 2ECCaME
.
This could be due to the HF added to the 2ECCaCl and 2ECCaME
instead using Ca2+ ions to form CaF2 as shown by XRD, that is,
the Fꢀ and COOH groups compete for the Ca2+ during the hy-
bridisation reactions. 2ECCaME* was therefore selected for SIMS,
SEM, compression mechanical testing and cell culture testing.
CME hybrid, 2ECCaME*, had the lowest T3/Ttot, Q4/Qtot and Dc
(83%) values of all the samples, indicating that a polymeric
chain-like silica structure or a network composed of primary
silica particles may have been formed. In the case of the 2EC
sample, a rather coarse particle-like sol-gel structure formed.
Compositional imaging
The rapid reaction of CME with water means that the 2ECCaME
*
Figure 2 shows the time-of-flight (TOF) SIMS surface images of
the 2ECCaME* hybrid. Figure 2a and b show the Ca+ and Si+
mapping, respectively. Total count for Ca+ was 4.8ꢁ107, which
was higher than that of Si+ (9.0ꢁ106); this was due to the Ca+
being easily ionised compared to Si+. The elemental distribu-
tion images show that at the macroscale the Ca and Si were
not homogeneous in 2ECCaME*. The Ca count was found to be
low in regions where Si was high. However, Figure 2c and d,
of SiꢀOꢀCa+ and OꢀSiꢀO+ molecules from the inorganic silica
network show that the Ca was homogeneously distributed in
the silica network, which was also homogeneous. This sug-
and 2ECCaME samples gelled rapidly in comparison to 2EC and
2ECCaCl, therefore the CME-derived samples had a highly dis-
rupted, low condensed silica network, an observation that is in
agreement with the 29Si NMR data. However, the 2ECCaME
sample had a higher degree of condensation (Dc =90%) than
expected. This could be due to the addition of HF prior to the
addition of CME to the hybrid precursor sol. HF was added to
the hydrolysed TEOS and functionalised g-PGA mixture and al-
lowed to react for 3 min. Here, HF catalyses the condensation
of hydrolysed TEOS and GPTMS, forming silica particles with
Chem. Eur. J. 2014, 20, 8149 – 8160
www.chemeurj.org
8152
ꢀ 2014 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Morphology of hybrids
Figure 3 shows SEM images of 2ECCaME* before (Figure 3a) and
after soaking in tris(hydroxymethyl)aminomethane (TRIS) buffer
solution (Figure 3b). The as-made hybrid fracture surface had
a smooth surface topography (inset of Figure 3a) and high-
magnification images showed that it is composed of the typi-
cal sol-gel silica morphology of bonded silica nanoparticles[7a]
but with polymer packed in between the particles. The parti-
cles were estimated from the image to be <50 nm in size.
Similar nanostructures were found for the 2ECCaCl and 2ECCaME
hybrids (not shown).
Figure 2. TOF SIMS images of 2ECCaME*: (a) Ca+, (b) Si+, (c) SiCaO+,
+
+
(d) Si2O+, (e) CNOCa2 and (f) Si2CHO3
.
Figure 3. SEM images of 2ECCaME* before (a and inset) and after (b) soaking
in TRIS buffer solution.
gests that the CME reacts with the hydrolysed TEOS and enters
the Si network at room temperature as observed in silica gels
by Yu et al.[7b] A continuous Si network was formed in the
hybrid and Ca was homogeneously distributed within this net-
work. Finally, Figure 2e and f show the molecular distribution
profile of CNCa2O+ from calcium bonded to the g-PGA and
that of CSi2O3H+ from condensed GPTMS, respectively. These
show identical profiles to that of Ca+ and Si+. Therefore, the
inhomogeneity seen in Figure 2a and b was due to the phase
separation of Ca-cross-linked polymer and condensed GPTMS
molecules.
Mechanical testing of hybrid monoliths
Figure 4 shows the typical nanoindentation loading and un-
loading curves obtained for the hybrids and 100 wt% inorgan-
ic glass samples (70S30CME*: 70 mol% SiO2, 30 mol% CaO). The
testing procedure adopted in this study resulted in crack-free
deformation of the hybrid and 70S30CME* samples, an observa-
tion that was proven by the typical force versus depth curve
observed with a Berkovich indenter (Figure 4). The organic
content in the hybrid material could give rise to viscoelastic ef-
fects that would be observed on the unloading curve as
To summarise, during functionalisation of g-PGA, the epoxy
ring of the GPTMS is expected to react with ꢀCOOH groups of
the g-PGA, forming covalent bonds between GPTMS and g-
PGA, whereas for some GTPMS molecules, the epoxide ring is
expected to open without nucleophilic attack, forming a diol,
and a proportion of GPTMS may condense with itself, forming
dimers or trimers.[6] Once hydrolysed TEOS is added, the SiꢀOH
groups on hydrolysed GPTMS molecules are expected to com-
plete coupling to the silica network through polycondensation,
but additional silica condensation on the GPTMS also took
place. On addition of CME, the CME reacted with all the re-
maining free ꢀCOOH sites of the g-PGA, which then cross-
linked with Ca. At the same time, CME also condensed with
the hydrolysed TEOS and gelation occurred, locking in place
the self-condensed GPTMS and Ca-cross-linked g-PGA. This
leads to a structure where the Si network derived from TEOS is
homogeneous, Ca that has reacted with and entered this inor-
ganic network is also homogeneous, whereas the GPTMS that
has reacted with itself and the free carboxylic acid groups of g-
PGA, which is cross-linked by Ca, are not homogeneous in the
hybrid.
Figure 4. Loading and unloading nanoindentation curves of the hybrids syn-
thesised with calcium methoxyethoxide (2ECCaME), calcium chloride (2ECCaCl)
and without calcium (2EC). Traditional 70S30C bioactive glass synthesised
with calcium nitrate sintered at 7008C and 70S30CME* processed at 608C are
included for comparison.
Chem. Eur. J. 2014, 20, 8149 – 8160
www.chemeurj.org
8153
ꢀ 2014 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
a “nose”;[14] this was not observed in the case of these hybrids
under the testing conditions used. The dwell at 50 mN for 20 s
allowed for complete creep relaxation, hence a “nose” was not
observed.
Mammeri study. However, the calcium-containing 2ECCaME*
had a higher modulus compared with PMMA/SiO2. This sug-
gests that careful choice of the organic component and how it
is cross-linked in the hybrid have a great influence on the over-
all mechanical properties.
The reduced elastic modulus (Er) and hardness values from
the force–depth curves[15] and the creep parameter A/d(0)[14]
calculated from the dwell data at 50 mN are summarised in
Table 2. The reduced modulus of the 70S30C glass was at least
Table 2 also shows that the relative modulus of 2ECCaME was
similar to 2ECCaME*. This indicates that the addition of HF did
not influence the mechanical properties of the hybrids, an ob-
servation that supports the finding that addition of HF did not
change the gelation behaviour of the hybrid.
Beake[17] showed that by fitting the logarithmic equation,
Table 2. Organic content, reduced modulus, hardness and creep parame-
ter A/d(0) of 70S30C made with Ca(NO3)2 (70S30C) and with calcium me-
thoxyethoxide (70S30CME) and Class II g-PGA/bioactive silica hybrids syn-
thesised with calcium methoxyethoxide (2ECCaME), calcium chloride
(2ECCaCl) and without (2EC) with a processing temperature of 608C. Sam-
ples where HF was not added are marked with *.
Increase in depth ¼ A lnðBt þ 1Þ
to the 20 s dwell data, at the maximum load of 50 mN, the
extent and rate of inelastic deformation can be obtained,
where t is the dwell time, A is the extent term and B is the rate
term. Beake showed that the normalised A/d(0) term, where
d(0) is the initial deformation, can be used to compare the
extent of inelastic deformation of different materials. The A/
d(0) values for the hybrids and 70S30C samples are given in
Table 2. Sample 2EC and 70S30C heated to 7008C have the
smallest A/d(0) values, suggesting that there was no significant
creep in the 2EC sample compared with other samples includ-
Sample
Organic Reduced modulus Hardness
Creep parameter
A/d(0)
[wt%]
[GPa]
[GPa]
2EC
47
47
47
50
0
4.04ꢁ0.24
8.99ꢁ0.84
7.99ꢁ2.01
3.51ꢁ1.19
0.47ꢁ0.07
24.40ꢁ0.60
0.45ꢁ0.04 0.020ꢁ0.001
0.64ꢁ0.04 0.035ꢁ0.002
0.52ꢁ0.06 0.040ꢁ0.006
0.16ꢁ0.04 0.071ꢁ0.013
0.02ꢁ0.01 0.034ꢁ0.005
1.39ꢁ0.06 0.018ꢁ0.001
2ECCaME
2ECCaME
2ECCaCl
70S30CME
70S30C
*
*
0
ed in this work. The A/d(0) value for the 2ECCaME and 2ECCaME
*
2.5 times greater than that of the hybrids. This was expected
as the 70S30C was heated to 7008C, which leads to a highly
condensed silicate structure, whereas the hybrids were only
heated to 608C. However, all the hybrid samples have much
higher modulus (ꢁ8) than that of the 100 wt% inorganic,
70S30CME*, sample synthesised with CME as the calcium source
and processed at 608C. This shows that the hybridisation with
an organic polymer was very effective at increasing the me-
chanical properties of this inorganic system.
hybrids were 0.035 and 0.040, respectively, whereas for the
2ECCaCl hybrid it was 0.071. This suggests that the CaCl2-de-
rived hybrid was much more viscoelastic compared with the
CME-derived hybrids. Therefore the structure of 2ECCaCl could
be one where the organic phase has more freedom of move-
ment than that in the 2ECCaME samples. Hence, in the 2ECCaME
hybrids, the organic component was much more cross-linked
than in 2ECCaCl. As previously discussed, the increased cross-
linking could possibly be through the Ca2+ to g-PGA and/or to
silica.
The reduced modulus values of 2EC and 2ECCaME*, 2ECCaME
and 2ECCaCl were calculated to be 4.04, 8.99, 7.99 and
3.51 GPa, respectively. The addition of Ca in the hybrids using
CME as the source more than doubled the modulus, whereas
CaCl2 reduced it. This suggests that CME as the calcium source
was effective in incorporation of Ca in the hybrid. During the
synthesis of the 2ECCaME hybrids, as soon as CME was added it
reacted with the inorganic silica network and the free carboxyl-
ic acid groups of g-PGA. This ionic cross-linking by Ca2+ of the
polymer was also observed when CaCl2 was used as the calci-
um source, hence it seems that the improved mechanical
properties of the 2ECCaME hybrids could be due to Ca cross-
linking both the organic and the inorganic components sepa-
rately as well as the to each other, thereby forming a true in-
terpenetrating network of organic and inorganic constituents.
This conclusion is fair as the 2ECCaME* has a higher Er than the
organic-free 70S30CME* and the Ca-free 2EC samples. Mammeri
et al. performed nanoindentation studies on class II poly(meth-
yl methacrylate) (PMMA) /SiO2 hybrids and found a modulus of
6.60 GPa and hardness of 0.54 GPa.[16] The class II g-PGA/SiO2
hybrids, 2EC, synthesised in this study had a lower relative
modulus, perhaps owing to the inherent stiffness of the PMMA
and from the higher curing temperature of 1008C used in the
Compression tests were performed on cylindrical hybrid
samples of 2ECCaME*. Figure S1 in the Supporting Information
shows the stress–strain curves of 2ECCaME* samples with three
different height/width ratios (H/W), as well as those for 70S30C
stabilised to 8008C and 2ECCaCl. 70S30C was included as previ-
ous work has shown bioactive glass scaffolds (70 mol% SiO2,
30 mol% CaO) to have optimal compressive strength when
the glass was sintered to 8008C.[3b] Therefore, monoliths of this
glass and a 2ECCaCl sample were used as a comparison. When
the height-to-width ratios were 1:2 and 1:1, compression tests
showed that the 2ECCaME* samples had higher strength and
elastic modulus compared with the 2ECCaCl hybrid and that
they had comparable strength to 70S30C even though the hy-
brids had ~50 wt% inorganic material. Hybrid sample
2ECCaME* with a H/W ratio of 2:1 failed at a lower strength
than the other 2ECCaME* samples, owing to the large height-
to-width ratio, which made it fail at lower loads from higher in-
herent defects within. Hybrid 2ECCaME* with a ratio of 1:2 had
the highest strain-to-failure value (>70%) owing to the
sample being wider than it was high. This shows that the com-
pressive test is very sensitive to the sample dimensions and
Chem. Eur. J. 2014, 20, 8149 – 8160
www.chemeurj.org
8154
ꢀ 2014 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
that it could lead to false interpretations of strength in hybrid
materials. Figure S1b in the Supporting Information shows the
yield stress as a function of H/W of the 2ECCaME* hybrids. This
clearly shows that the yield stress decreased with increasing H/
W, hence samples with smaller height compared with width
(2ECCaME*, H/W of 1:2) show higher strength and strain. The
hybrid samples failed by cracking along the direction of the
applied force. Once a crack formed, it travelled through the
samples in the direction parallel to the force, resulting in
a chipped sample. The effective cross-sectional area of the re-
maining sample was then reduced and hence the samples
failed at lower loads before reaching its maximum failure load.
Therefore, a sample with a high H/W ratio would fail at lower
loads and a sample with a low H/W ratio would withstand the
loading for longer before complete failure. Therefore, care
should be taken when interpreting the results of the compres-
sive test data of these hybrid monoliths. However, in the case
of a porous scaffold for bone regeneration, for which this
hybrid material is being developed, compressive testing is im-
portant and should be performed. Following the international
standard ISO 640:2003, the recommended test specimen di-
mensions of H/W>1 gave an elastic modulus of 2.0ꢁ0.3 GPa
and compressive strength of 64.2ꢁ20 MPa for the 2ECCaME
*
sample. If we assume that the mechanical properties of the
produced porous scaffolds follow the cellular solids theory,[18]
a 70% porous scaffold is predicted to have a strength of
3.17 MPa. This is in the range of the strength of human trabec-
ular bone, which has 2–12 MPa[19] compressive strength.
Figure 5. Release profile curves of (a) polymer, (b) Ca2+ ions and (c) soluble
Si species from hybrids soaked in TRIS buffer solution for 2 weeks.
Hybrid degradation and polymer release
Soaking hybrids in TRIS buffer and analysing the polymer and
ions released can reveal details about the structure and bond-
ing present within the hybrid organic and inorganic constitu-
ents. Figure 5 shows the polymer release profile and Si species
and Ca ion dissolution profiles of the class II hybrid samples
soaked in TRIS buffer for 2 weeks. The concentration of the
species released at a particular point in time was weighted
against the total nominal concentration of that particular spe-
cies in the sample. For example, [g-PGA]/[g-PGATotal] (%) means
the concentration of g-PGA in TRIS buffer solution at that
point in time divided by the total concentration of g-PGA in
150 mg of hybrid sample. Therefore, Figure 5a shows the per-
centage of polymer released from the sample as a function of
soaking time. Figure 5b and c show the percentage of Ca and
Si released from the sample as a function of soaking time, re-
spectively. The calcium-free 2EC hybrid has a steady and slow
release of g-PGA. After 2 weeks of soaking, up to 20% of g-
PGA had been released into the solution. The polymer release
profile of the 2ECCaCl hybrid was more complex; it shows
a burst of release of g-PGA (>12%) in the first hour of soaking,
increasing to 40% at 24 h. Then, a large reduction in concen-
tration (18%) was observed at 1 week. The initial burst of re-
lease indicates that a large portion of g-PGA was loosely
bound in the 2ECCaCl hybrid. The reduction in g-PGA concen-
tration in the TRIS solution after 24 h could be a result of g-
PGA cross-linking to itself (ionic and/or through GPTMS), lead-
ing to insolubility and precipitation. The calcium ion release
profile of the 2ECCaCl hybrid is very similar to its polymer re-
lease profile. The majority of the calcium ion release took place
in the first 2 h of soaking, accounting for the total ion released
from the hybrid in the 2 weeks of soaking. This shows that the
Ca2+ ion was also loosely bound in the hybrid and could be
closely associated with the polymer.
The 2ECCaME hybrids with and without HF addition showed
similar g-PGA release profiles, where a steady increase in g-
PGA release was observed. The 2ECCaME* sample deviated
from this at 24 h soaking time, when a large amount of g-PGA
was released into solution from the 2ECCaME* hybrid. This
could be due to a physical breakup of the sample in solution;
however, a corresponding Si release was not observed at this
point in time. After 2 weeks of soaking, around 60% of g-PGA
had been released from the 2ECCaME hybrid samples. The
2ECCaME* sample showed a steady release of Ca2+ ions in the
first 8 h of soaking, after which a rapid increase in Ca2+ con-
centration from 22% to 53% was observed, followed by a pla-
teau in Ca2+ concentration up to 2 weeks. The profiles of both
the release of g-PGA and the dissolution of Ca2+ ions from the
2ECCaME* hybrid were similar, indicating the calcium released
into TRIS buffer was associated with g-PGA. 2ECCaME showed
a very slow calcium release for the first 24 h of soaking, during
Chem. Eur. J. 2014, 20, 8149 – 8160
www.chemeurj.org
8155
ꢀ 2014 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
which less than 10% of calcium was released. After 2 weeks,
around 30% of Ca2+ ions had been released. This was just
under half of that released from the HF-free 2ECCaME*. The dif-
ference in Ca2+ release between the CME samples was due to
the addition of HF, which caused formation of CaF2. CaF2 is
sparingly soluble in TRIS, therefore as a result of this Ca2+ was
locked in the hybrid.
Finally, the release of soluble Si species from the hybrid is
shown in Figure 5c. Up until 8 h of soaking, very small
amounts of Si species were released into the TRIS buffer solu-
tion from all of the samples. Following this, each of the sam-
ples showed a steady increase to more than 20% Si release
after soaking for up to 336 h. Both CME-derived samples,
2ECCaME and 2ECCaME*, had identical Si release for the whole
duration of testing.
To summarise, the dissolution study in TRIS buffer solution
suggests that the 2ECCaCl samples were poorly hybridised as
a burst of release of g-PGA and Ca and slow release of Si were
observed in the first few hours of soaking. This indicates that
the Si network was strongly bonded whereas the organic net-
work was loosely bonded in the hybrid. In the case of the
CME-derived hybrids, steady release of all constituents was ob-
served with instances of uncontrolled release behaviour. It also
appears that higher proportions of Ca from CME were bound
to g-PGA owing to the reactivity of calcium alkoxide to carbox-
ylic acid groups. Release of ~60% of g-PGA from the CME sam-
ples within 336 h of soaking indicates that the g-PGA was
poorly cross-linked to the silica network. Better cross-linking of
g-PGA to silica is required before this material could be used in
regeneration applications requiring scaffold support for the
long-term (typically more than 6 months).
Figure 6. (a) FTIR spectroscopy and (b) TGA-DSC of the hybrids after 4 weeks
in TRIS buffer solution.
Figure 6 shows FTIR spectra and thermogravimetric analysis–
differential scanning calorimetry (TGA-DSC) plots for the sam-
ples after soaking in TRIS buffer for 4 weeks. All the FTIR spec-
tra had clear bands corresponding to the SiꢀOꢀSi and SiꢀOH
vibrations. The amide and CꢀO vibrations from the polymer
were also clearly visible in the spectrum of the calcium-free
2EC sample. The calcium-containing hybrids have very weak
bands at the amide, OH and CH vibrations, indicating the pres-
ence of some organic material. The TGA-DSC curve (Figure 6b)
shows that the organic content (g-PGA+organic of GPTMS) of
the 2EC sample was >50 wt% after soaking in TRIS buffer for
4 weeks. This is a slight increase in total organic content of the
hybrid from the initial 46 wt% before soaking, a fact that
means that higher proportions of inorganic (silica) than organic
material had been released from the hybrid into the solution.
This is supported by the Si and polymer dissolution rates in
Figure 5. The calcium-containing samples 2ECCaCl, 2ECCaME and
2ECCaME* have similar inorganic content (about 70 wt%) after
soaking in TRIS buffer for 4 weeks. Before soaking, the calcium-
containing samples had just over 50 wt% inorganic content,
this means that after soaking, a larger proportion of organic
material was released. The opposite was observed for the 2EC
sample, this indicates that the presence of Ca increased the
dissolution of organic species in the hybrid. This could be due
to the formation of calcium-polyglutamate, which is more solu-
ble in aqueous media than the free acid g-PGA.[20]
Bioactivity
Figure 7a shows the concentration of phosphorus in simulated
body fluid (SBF) after soaking the calcium-containing hybrid
samples 2ECCaME*, 2ECCaME and 2ECCaCl. All three samples
showed a rapid decrease in P concentration from 1 day to
3 days and a further drop from 3 days to 7 days. This indicates
the deposition of phosphorus on the samples, which could
correspond to the formation of a calcium phosphate layer on
the sample. Figure 7b shows the XRD patterns of the hybrids
before and after soaking in SBF for 3, 7 and 28 days. Mature
crystalline hydroxyapatite (HA) was found on all the calcium-
containing 2ECCa hybrids. Diffraction patterns of the 2ECCaME
*
(HF-free) hybrid before soaking and after soaking for 3, 7 and
28 days are also shown in Figure 7b. It shows that after 3 days
of soaking, CaCO3 was formed on the hybrid in SBF, a product
that disappeared by 7 days as HA began to form. HA detection
was supported by the FTIR curves (Figure 7c), which show the
hybrid sample 2ECCaME* before and after soaking for 3 and
7 days in SBF. Bands corresponding to PꢀO vibrations (Fig-
ure 7c) were observed after 3 days, bands that become intense
and sharp after 7 days of soaking. This suggests a calcium or-
thophosphate phase was deposited on the sample by 3 days,
the phase then matured to HA by 7 days (XRD) and grew into
a more crystalline material by 28 days. FTIR also shows evi-
Chem. Eur. J. 2014, 20, 8149 – 8160
www.chemeurj.org
8156
ꢀ 2014 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Figure 8. Human bone marrow derived mesenchymal stem cells (hMSCs) via-
bility on 2ECCaME*. (a) LIVE/DEADTM assay of the hMSCs cultured on the hy-
brids’ surface for seven days (dead cells circled). (b) Metabolic activity as
measured with the Alamar Blue assay after four and seven days in culture.
*=p<0.001.
tion (Figure 8b). The Alamar Blue assay confirmed that the hy-
brids were not toxic to the hMSCs and further demonstrated
that the cells grew on the hybrids over the period of 7 days, in-
dicating a favourable cell response. In addition, the dissolution
products of the hybrids were not toxic and also allowed the
hMSCs to grow over time, indicating that there was no hinder-
ing of cell growth (Figure 8b).
Figure 7. (a) Concentration of P in SBF with soaking time, (b) XRD and
(c) FTIR spectra of the hybrids after 4 weeks in SBF solution.
dence of amide I and II and CꢀOꢀC vibrations up to 7 days, an
observation that indicates that majority of the polymer was
still present in the hybrid.
Conclusion
Cell study
Hybrid materials were successfully produced by using calcium
methoxyethoxide (CME) as the Ca precursor. The CME was
found to induce gelation on mixing with the organic/inorganic
mixture. Once controlled, this was beneficial as rapid gelation
is usually achieved by using HF as a catalyst, which is a hazard-
ous reagent. Very little difference was found in terms of gela-
tion time between the class II CME hybrids whether using HF
or not. 29Si MAS NMR showed the presence of both T and Q
species in the hybrids. It also showed that the SiꢀOꢀSi network
connectivity was lower in the CME samples when HF was not
used owing to Ca acting as a network modifier when intro-
duced as CME. However, when HF was used, the SiꢀOꢀSi net-
work connectivity increased. This was due to the combined in-
fluence of HF catalysing the condensation of the Si network
and the formation of CaF2 by the reaction of CME with HF,
To determine whether the hybrid samples were cytotoxic,
human bone marrow derived mesenchymal stem cells (hMSCs)
were cultured on 2ECCaME and the viability was examined with
the LIVE/DEAD assay after 3 days (Figure 8a). The assay re-
vealed that there were numerous live hMSCs on the hybrids
and very few dead cells. Additionally, from the LIVE/DEAD
assay it was observed that the cells were well spread on the
hybrids after 3 days in culture, indicating a favourable cell at-
tachment. To further examine the effect of the hybrids on cell
viability and growth, the Alamar Blue assay was used to mea-
sure the cell metabolic activity after 4 and 7 days. The meta-
bolic activity of the cell monolayer on the tissue culture plastic
around the hybrid samples was also measured to check if re-
lease from the samples resulted in cell death or number reduc-
Chem. Eur. J. 2014, 20, 8149 – 8160
www.chemeurj.org
8157
ꢀ 2014 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
(C), with calcium (Ca) from the CME source. Samples made without
the addition of HF were denoted as 2ECCaME*.
which in effect reduces the Ca available to disrupt the Si net-
work. Importantly, when HF was not used, calcium from CME
was incorporated into the hybrid systems better than when
using CaCl2. Hydroxyapatite (HA) was formed on all class II hy-
brids in SBF, indicating all the samples were bioactive. The rate
of hybrid dissolution was strongly influenced by the type of Ca
source used, CME was successful at cross-linking the polymer;
hence a steady polymer release was observed whereas in the
CaCl2 samples a burst release was observed. Class II hybrids
synthesised with CME showed compressive strength similar to
70S30C glass stabilised to 7008C and the presence of Ca in the
hybrid improved the modulus of the hybrid two-fold, indicat-
ing CME is a promising calcium source for hybrid synthesis.
Class II hybrids made with calcium chloride were synthesised by
using the procedure developed by Poologasundarampillai et al.[4p]
The g-PGA was reacted with GPTMS in DMSO for 12 h under an N2
atmosphere. This was then rotary-vacuum evaporated to partially
remove the DMSO. To this, calcium chloride, which was pre-dis-
solved in water to aid solvation, was added. After mixing for
30 min, the hydrolysed sol was added and mixed for another 1 h.
The hybrid precursor solution was then cast in polymethylpentene
(PMP) moulds and HF was added at a volume ratio of hybrid pre-
cursor sol to HF of 1:17. These samples, denoted as 2ECCaCl, were
aged and dried at 608C.
Calcium-free hybrids (denoted as 2EC) were produced following
the same procedure as that for the synthesis of 2ECCaME, but no
calcium was added. For the synthesis of the 2EC and 2ECCaME sam-
ples, the overall ratio of Si4+/H2O/H+ were 1:2:2, whereas that of
2ECCaCl was 1:12:2.
Experimental Section
Sample preparation
Characterisation
The g-PGA was purchased in the free acid powder form from Natto
Biosciences (Quebec, Canada) with 95 wt% polymer and 5 wt%
water. All other materials were purchased from Sigma–Aldrich and
used as received.
Several analysis techniques were used to study the atomic
(MAS NMR), molecular (FTIR and XRD), nano- (SEM) and macro-
scale (SIMS) structure and composition of the hybrids. All 29Si solid-
state magic-angle spinning nuclear magnetic resonance
(MAS NMR) data were acquired at 7.05 T on a Varian InfinityPlus
Calcium methoxyethoxide (CME) was prepared following the
method in literature.[7b,21] In brief, calcium granules (1 g) were re-
acted with 2-methoxyethanol (24 mL) under an argon atmosphere
at 808C for 24 h. The resultant solution was centrifuged at
4000 rpm for 10 min to remove unreacted calcium metal. The con-
centration of CME in solution was confirmed gravimetrically by
heating to 10508C for 12 h, upon which the solvent evaporated
and the alkoxide converted to CaO.
300 spectrometer operating at
a
29Si Lamor frequency of
59.62 MHz. These measurements were undertaken by using
a Varian 7.5 mm probe spinning at 5 kHz. A 5 ms excitation pulse
(flip angle ~p/4) was applied with a 240 s recycle delay to ensure
complete relaxation was obtained for each measurement. All
29Si MAS NMR data were referenced to the IUPAC recommended
primary reference of TMS (d=0 ppm) by a secondary reference of
kaolinite located at d=ꢀ92 ppm. To quantify the Tn and Qn distri-
bution of the Si speciation, each resonance in the 29Si MAS NMR
data was simulated with a Gaussian peak shape by using the
Origin 8.5 Pro software package (Origin 8.5 Pro, OriginLab, North-
ampton). X-ray diffraction was performed on a Bruker D2 Phaser,
using a step-scanning method with CuKa radiation at 30 kV and
10 mA with a count rate of 0.25 s per step, from 2q values of 108
to 708. Attenuated total reflectance Fourier transform infrared
spectroscopy (ATR-FTIR) was used to study the chemical structures
of the hybrids. Samples were analysed on a Thermo Scientific Nico-
let iS10 FTIR.
Class II calcium-free hybrids and hybrids with either CME or calci-
um chloride as the calcium source were synthesised in this work.
The weight% of organic/inorganic material was 50:50, whereas the
molar ratio of Si/Ca was 70:30 to make the inorganic component
the same composition as 70S30C (70 mol% SiO2, 30 mol% CaO)
bioactive glasses. Poly(g-glutamic acid) (g-PGA) was used as the or-
ganic phase, which was covalently cross-linked to the silica phase
(derived from tetraethylorthosilicate (TEOS)) through the organosi-
lane glycidoxypropyltrimethoxy silane (GPTMS). The total organic
content of the hybrid was derived from g-PGA and the organic
portion from GPTMS and the inorganic content attributed to the
silica derived from TEOS and GPTMS and the CaO from CME.
The synthesis route was modified to that described in previous
work.[4p] Here, the use of CME added a level of complexity owing
to its high reactivity with water, leading to a three-pot reaction as
shown in Scheme 1. First, g-PGA dissolved in DMSO was reacted
with GPTMS at a molar ratio of GPTMS to monomeric glutamic
units of 1:2 for 12 h to form the Si-functionalised g-PGA in solu-
tion. The inorganic silica sol was prepared separately by hydrolys-
ing TEOS with H2O and HCl (2N) for 30 min. The Si-functionalised
g-PGA and hydrolysed TEOS were added together and mixed for
5 min, followed by addition of the third pot of CME in methoxy
ethanol and DMSO (1:1 v/v). This hybrid precursor mixture was al-
lowed to mix for a further 5 min before casting into Teflon contain-
ers for aging (3 days with the containers sealed) and drying
(10 days) at 608C. Samples were also made with HF to study the
influence of HF on gelling and silica network formation. HF
(5 wt%) was added to the hybrid precursor sol after mixing pots
1 (Si-functionalised g-PGA) and 2 (hydrolysed TEOS) at a hybrid sol/
Compositional homogeneity
Secondary ion mass spectroscopy (SIMS) was performed with
a TOF-SIMS V instrument (ION-TOF GmbH, Germany) to study the
homogeneity of species throughout the hybrids. The samples were
polished to a mirror finish to obtain smooth surfaces for SIMS anal-
yses. An area of 500ꢁ500 mm2 was sputtered with a 10 keV Cþ60
cluster ion source under a current of 1.2 nA for 1 s every scan.
After which, an area of 250ꢁ250 mm2 was analysed by rastering
over by a 25 keV Biþ3 beam with an average current of 0.55 pA. An
image of 256ꢁ256 pixels for a total of 840 scans was collected. A
low-energy electron gun of 20 eV was used for charge compensa-
tion.
Topography
Scanning electron microscopy (SEM) was used to examine the mor-
phological and textural features of chromium-coated hybrids. A
Leo 1525 with Gemini column fitted with a field emission gun at
HF volume ratio of 17:1. These samples were denoted as 2ECCaME
,
where 2 corresponds to two moles of glutamic acid (E) to GPTMS
Chem. Eur. J. 2014, 20, 8149 – 8160
www.chemeurj.org
8158
ꢀ 2014 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
5 kV, working distance between 5–10 mm, with a 30 mm aperture
and In-lens secondary electron detector (SEI) was used. All the
samples were coated with chromium at 70 mA for 1 min.
4, 8, 24, 72, 168, 336 and 672 h) for analysis with ICP-OES and 1 mL
of fresh SBF was added to keep the solution volume of 100 mL. A
major reduction in phosphorus concentration in SBF was observed
after 72 h and 168 h of soaking, therefore to detect HCA layer for-
mation, the samples were filtered (1 mm paper) after 72 and 168 h
of soaking and the monoliths were rinsed with acetone, dried and
then evaluated by XRD and FTIR. XRD was carried out on ground
samples with a PANalytical X’Pert Pro MPD series automated spec-
trometer, using a step-scanning method with CuKa radiation at
40 kV and 40 mA with a 0.0408 2q step and a count rate of 50 s
per step, from 2q values of 58 to 758. Samples for FTIR were
ground with potassium bromide (KBr) with a 1:100 weight ratio.
The resultant powder was pressed into a pellet and FTIR was per-
formed on a Bruker Vector 22 thermal gravimetric infrared spec-
trometer (TGA-IR) through transmission with a wavelength of
Mechanical properties of the hybrids
Uni-axial compression tests were performed on cylindrical hybrid
samples with 5 mm diameter and heights ranging from 2–10 mm.
Tests were performed on a Zwick 1474 fitted with a 100 kN load
cell at 8.33 mmsꢀ1 strain rate until failure or 80% maximum defor-
mation.
Nanoindentation was performed on the hybrids and inorganic sol-
gel glass 70S30C by using a testing procedure similar to that
adopted by Mammeri et al.[22] A loading rate of 5 mNsꢀ1 to maxi-
mum indentation load of 50 mN was applied. A dwell for 20 s at
maximum load was applied before unloading at 10 mNsꢀ1 to
5 mN, when a final dwell for 60 s was also applied to determine
the thermal drift contribution of the indentation system to total
displacement measured by a capacitive transducer. A NanoTest
Vantage (Micro Materials Ltd, UK) was used to perform the nanoin-
dentations by using a Berkovich pyramidal tip.
633 nm and in the range of 400 to 2000 cmꢀ1
.
Cell culture
To examine the effect of the new 2ECCaME* hybrids on cell toxicity
and growth, hMSCs were seeded directly on the surface of the
hybrid monoliths. The hybrid 2ECCaME* monoliths (5ꢁ5 mm2) were
sterilised by immersion in 70% (v/v) ethanol followed by extensive
washing in phosphate buffered saline (PBS) and then incubation
in Dulbecco’s modified Eagle’s medium (DMEM), supplemented
with penicillin (50 UmLꢀ1), streptomycin (50 mgmLꢀ1) and ampho-
tericin B (2.5 mgmLꢀ1), at 378C overnight (preconditioning) prior to
cell culture experiments. hMSCs were cultured in low glucose
phenol red-free DMEM, supplemented with 10% fetal bovine
serum (FBS), L-glutamine (2 mm), penicillin (50 UmLꢀ1), streptomy-
cin (50 mgmLꢀ1) and amphotericin B (2.5 mgmLꢀ1), all from Invitro-
gen, UK, (cell culture medium) in a humidified incubator at 378C
and 5% CO2.
Dissolution in TRIS buffered solution
Hybrid samples were soaked in TRIS buffer to study the dissolution
behaviour over time. The release of polymer and dissolution of Si
species and Ca2+ ions from the hybrids often reveal a large
amount of information on the structure and bonding present be-
tween the constituents of the hybrid. TRIS buffer was prepared fol-
lowing the procedure outlined in ref. [23], where tris (hydroxy-
methyl) aminomethane (15.090 g) was dissolved in deionised water
(2000 mL), to which HCl (1n) was added at 378C to adjust pH to
7.30. Hybrid monoliths weighing 150 mg were soaked in TRIS
(100 mL) and incubated at 378C in an orbital shaker set to
120 rpm. After 1, 2, 4, 8, 24, 168 and 336 h of soaking, 1.5 mL solu-
tion was aliquoted for Si and Ca elemental analysis and g-PGA con-
centration determination. To compensate for the removed 1.5 mL
solution, 1.5 mL of fresh TRIS buffer solution was added to the
original test solution. The soluble silica species and Ca2+ ion con-
centrations in the aliquoted solutions were analysed by inductive
coupled plasma optical emission spectroscopy (ICP-OES, Thermo
iCAP 6300 Duo). g-PGA release into the TRIS buffer solution at
each time point was obtained by performing a Pierce Micro bicin-
choninic acid (BCA) protein assay. Polymer concentration in solu-
tion was determined by reference to a standard curve consisting
of known concentrations of functionalised g-PGA in TRIS buffer so-
lution. The known concentration of standard solutions (1, 50, 100,
200, 300, 500, 750 and 1000 mgmLꢀ1) were prepared by first react-
ing g-PGA with GPTMS (molar ratio of GPTMS to monomeric gluta-
mic molecule was 1:2) in DMSO. Then, appropriate amounts of the
reacted solution were added to TRIS buffer to make up the re-
quired standards. An Anthos 2020 spectrophotometer with mea-
surement filter of 562 nm was used to analyse the absorbance of
each of the standards and samples.
For direct culture on the monoliths, the hMSCs were resuspended
in a small volume (20 mL) of cell culture medium, seeded on the
surface of the hybrids and incubated for 1 h at 378C and 5% CO2.
Then, the appropriate amount of medium was added to cover
each hybrid (i.e., 0.5 mL/well, 48-well plate). The seeding density
used was 1ꢁ104 cellscmꢀ2 for all experiments.
Cytotoxicity was evaluated with the LIVE/DEAD viability/cytotoxici-
ty assay (Molecular Probes, UK), which was performed according to
the manufacturer’s instructions. Stained samples were examined
under an Olympus BX-URA2 fluorescence microscope. Images
were captured by using a Zeiss Axiocam digital camera and ana-
lysed by using KS-300 software (Imaging Associates).
For cell growth, the Alamar Blue assay was used to measure the
metabolic activity and consequently the rate of cell growth after 4
and 12 days of culture. As a positive control, hMSCs were seeded
on tissue culture plastic and cultured in parallel with the hybrids.
For each Alamar Blue measurement, the hybrids were moved to
a new plate and incubated at 378C and 5% CO2 for 2 h with
phenol-free DMEM (without added serum) containing 10% (v/v)
Alamar Blue solution (Invitrogen, UK). Then, 100 ml of each sample
was transferred to a new black-with-clear-bottoms 96-well plate
and the fluorescent intensities were measured by using a micro-
plate reader, with fluorescence excitation and emission wave-
lengths of 570 and 610 nm, respectively. To determine the effect of
the dissolution products released over time by the hybrids on
hMSCs, the metabolic activity of the cells on the plate was mea-
sured as well with the Alamar Blue assay. The fluorescence values
obtained were averaged, and the average was expressed as a per-
centage of the cells on cell culture plastic at day 4. Experiments
were performed in triplicate, six hybrids for each experiment. The
FTIR, SEM and TGA-DSC were performed on the samples that were
incubated in TRIS buffer solution for 672 h to study the effect of
dissolution on the hybrid.
In vitro “bioactivity” testing in SBF
The rate of HCA formation in simulated body fluid (SBF) was deter-
mined. Monoliths (150 mg) were immersed in SBF (100 mL) and
placed in an orbital shaker with an agitation rate of 120 rpm at
378C. 1 mL of each sample was collected at each time point (1, 2,
Chem. Eur. J. 2014, 20, 8149 – 8160
www.chemeurj.org
8159
ꢀ 2014 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Funct. Mater. 2010, 20, 3835–3845; p) G. Poologasundarampillai, C. Ion-
escu, O. Tsigkou, M. Muthu, R. G. Hill, M. M. Stevens, J. V. Hanna, M. E.
Smith, J. R. Jones, J. Mater. Chem. 2010, 20, 8952–8961.
[5] a) M. Y. Koh, C. Ohtsuki, T. Miyazaki, J. Biomater. Appl. 2011, 25, 581–
594; b) G. Poologasundarampillai, B. Yu, O. Tsigkou, E. Valliant, S. Yue,
P. D. Lee, R. W. Hamilton, M. M. Stevens, T. Kasuga, J. R. Jones, Soft
Matter 2012, 8, 4822–4832; c) E. M. Valliant, F. Romer, D. Wang, D. S.
McPhail, M. E. Smith, J. V. Hanna, J. R. Jones, Acta Biomater. 2013, 9,
7662–7671.
statistical significance of the obtained data was assessed by one-
way ANOVA variance analysis by using the Sigma stat software
(SigmaStat 4.0, Systat Software, California) and level of significance
was set at p<0.001.
Acknowledgements
The EPSRC is thanked for funding (EP/E057098, EP/E051669
and EP/I020861/1). J.V.H. and M.E.S. thank the EPSRC and the
University of Warwick for partial funding of the solid-state NMR
infrastructure at Warwick, and acknowledge additional support
for this infrastructure obtained through Birmingham Science
City: Innovative Uses for Advanced Materials in the Modern
World (West Midlands Centre for Advanced Materials Projects
1 and 2), with support from Advantage West Midlands (AWM)
and partial funding by the European Regional Development
Fund (ERDF). J.V.H. also thanks James MacDonald for help in
acquiring some NMR data. J.R.J. was a Royal Academy of Engi-
neering/EPSRC Research Fellow and also acknowledges the
Philip Leverhulme Prize for funding.
[6] L. Gabrielli, L. Russo, A. Poveda, J. R. Jones, F. Nicotra, J. Jimꢂnez-Bar-
bero, L. Cipolla, Chem. Eur. J 2013, 19, 7856–7864.
[7] a) S. Lin, C. Ionescu, K. J. Pike, M. E. Smith, J. R. Jones, J. Mater. Chem.
2009, 19, 1276–1282; b) B. Yu, C. Ionescu, R. A. Martin, R. J. Newport,
J. V. Hanna, M. E. Smith, J. R. Jones, Langmuir 2012, 28, 17465–17476.
[8] a) M. M. Pereira, J. R. Jones, L. L. Hench, Adv. Appl. Ceram. 2005, 104,
35–42; b) B. A. Allo, A. S. Rizkalla, K. Mequanint, Langmuir 2010, 26,
18340–18348.
[9] I. D. Xynos, A. J. Edgar, L. D. Buttery, L. L. Hench, J. M. Polak, J. Biomed.
Mater. Res. 2001, 55, 151–157.
[10] G. Poologasundarampillai, B. Yu, J. R. Jones, T. Kasuga, Soft Matter 2011,
7, 10241–10251.
[11] M. Guglielmi, G. Carturan, J. Non-Cryst. Solids 1988, 100, 16–30.
[12] K. J. D. Mackenzie, M. E. Smith, Multinuclear solid state NMR of inorganic
materials, Pergamon, Oxford, 2002, p. 748.
[13] C. J. Brinker, J. Non-Cryst. Solids 1988, 100, 31–50.
[14] F. Mammeri, E. Le Bourhis, L. Rozes, C. Sanchez, A. Huignard, D. Lefevre,
J. Non-Cryst. Solids 2004, 345–346, 610–614.
Keywords: bioactivity · calcium methoxyethoxide · hybrid
materials · poly(g-glutamic acid) · sol-gel process
[15] W. C. Oliver, G. M. Pharr, J. Mater. Res. 1992, 7, 1564–1583.
[16] F. Mammeri, L. Rozes, E. Le Bourhis, C. Sanchez, J. Eur. Ceram. Soc. 2006,
26, 267–272.
[17] B. Beake, J. Phys. D 2006, 39, 4478–4485.
[18] L. J. Gibson, M. F. Ashby, Cellular solids: structure and properties, Cam-
bridge University Press, Cambridge, 1999.
[19] D. R. Carter, W. C. Hayes, Science 1976, 194, 1174–1176.
[20] G.-H. Ho, T.-I. Ho, K.-H. Hsieh, Y.-C. Su, P.-Y. Lin, J. Yang, K.-H. Yang, S.-C.
Yang, J. Chin. Chem. Soc. 2006, 63, 1363–1384.
[1] J. R. Jones, Acta Biomater. 2013, 9, 4457–4486.
[2] L. L. Hench, R. J. Splinter, W. C. Allen, T. K. Greenlee, J. Biomed. Mater.
Res. 1971, 5, 117–141.
[3] a) L. L. Hench, J. M. Polak, Science 2002, 295, 1014–1017; b) J. R. Jones,
L. M. Ehrenfried, L. L. Hench, Biomaterials 2006, 27, 964–973.
[4] a) H. Schmidt, Struct. Bonding (Berlin) 1992, 77, 119–151; b) M. B. Novak,
Adv. Mater. 1993, 5, 422–433; c) P. Judeinstein, P. Oliveira, H. Krug, H.
Schmidt, Chem. Phys. Lett. 1994, 220, 35–39; d) D. A. Loy, K. J. Shea,
Chem. Rev. 1995, 95, 1431–1442; e) U. Schubert, N. Husing, A. Lorenz,
Chem. Mater. 1995, 7, 2010–2027; f) P. Judeinstein, C. Sanchez, J. Mater.
Chem. 1996, 6, 511–525; g) J. Mackenzie, Q. Huang, T. Iwamoto, J. Sol-
Gel Sci. Technol. 1996, 7, 151–161; h) D. Tian, P. Dubois, R. Jꢂrꢃme, J.
Polym. Sci. Part A 1997, 35, 2295–2309; i) M. Laridjani, J. Mater. Sci.
1999, 34, 5945–5953; j) Z. Ahmad, J. E. Mark, Chem. Mater. 2001, 13,
3320–3330; k) G. Kickelbick, Prog. Polym. Sci. 2003, 28, 83–114; l) S.
Medda, J. Non-Cryst. Solids 2003, 318, 149–156; m) T. Ogoshi, Y. Chujo,
J. Mater. Chem. 2005, 15, 315–322; n) M. Vallet-Regi, D. Arcos, Curr.
Nanosci. 2006, 2, 179–189; o) O. Mahony, O. Tsigkou, C. Ionescu, C.
Minelli, L. Ling, R. Hanly, M. E. Smith, M. M. Stevens, J. R. Jones, Adv.
[21] D. M. Pickup, S. P. Valappil, R. M. Moss, H. L. Twyman, P. Guerry, M. E.
Smith, M. Wilson, J. C. Knowles, R. J. Newport, J. Mater. Sci. 2009, 44,
1858–1867.
[22] F. Mammeri, E. L. Bourhis, L. Rozes, C. Sanchez, J. Eur. Ceram. Soc. 2006,
26, 259–266.
[23] D. S. Brauer, M. Mneimne, R. G. Hill, J. Non-Cryst. Solids 2011, 357, 3328–
3333.
Received: October 14, 2013
Revised: April 3, 2014
Published online on May 18, 2014
Chem. Eur. J. 2014, 20, 8149 – 8160
www.chemeurj.org
8160
ꢀ 2014 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim