Journal of Molecular Structure 704 (2004) 53–58

www.elsevier.com/locate/molstruc

Collagen – hydroxyapatite – water interactions investigated by XRD,

piezogravimetry, infrared and Raman spectroscopy
L.F. Sukhoduba,*, C. Mosekeb, L.B. Sukhoduba, B. Sulkio-Cleffb, V.Ya. Maleevc,
M.A. Semenovc, E.G. Bereznyakc, T.V. Bolbukhc
a
Institute of Applied Physics, National Academy of Sciences of Ukraine, Petropavlovskaya str.58, Sumy 40030, Ukraine
b
Institute of Nuclear Physics, University of Muenster, Wilhelm-Klemm-Str.9, Muenster 48149, Germany
c
Institute of Radiophysics and Electronics, NAS of Ukraine, 12 Acad. Proskura Str., Kharkov 61085, Ukraine
Received 16 November 2003; revised 18 December 2003; accepted 18 December 2003

Abstract
Influence of the mineralization process of collagen (Cg) by hydroxyapatite (HA) on interactions with water molecules was investigated
using infrared and Raman spectroscopy and piezogravimetry methods. Hydroxyapatite was synthesized using a simplified basic reaction with
and without Ti ions, followed by phase analysis with X-ray diffraction spectroscopy (XRD). Cg– HA film samples have been prepared by a
process of slow evaporation of water from the solution at 4 8C. It was found that in the presence of HA some Cg hydration sites, probably
corresponding to its gap region, are closed, at least partly, for direct interactions with water molecules from the Cg hydration shell. A model
illustrating early mineralization process of two Cg layers in the presence of water molecules are proposed for explaining of obtained
experimental data.
q 2004 Elsevier B.V. All rights reserved.
Keywords: Collagen; Hydroxyapatite; Hydration; X-ray diffraction spectroscopy; Infrared; Piezogravimetry; Mineralization

1. Introduction
Collagen proteins are the main bioorganic part of bone.
This macromolecule spontaneously forms triple helices
fibrils of great tensile strength and thermostability with
holes within the 3D-structure where tiny hydroxyapatite
(HA) nanocrystals can grow [1]. The last phenomenon is
referred to as mineralization of collagen [2]. Understanding
of the molecular mechanisms underlying the biomineraliza-
tion (until now not completely understood) is extremely
important for the fundamental problem of the hard tissue
growth as well as for developing of new biomaterials, in
particular 3D, composite HA – collagen, biodegradable bone
substitutes and for the tissue engineering [3 – 5].
To truly understand the mineralization, we need to
understand the influence of water molecules (sometimes
called ‘biological water’ because of its specific structure
near biomolecules surfaces [5]) on the collagen – HA
interactions. The reason lies in two factors: the first is
the critical influence of the hydration shell on the triple helix
collagen structure and its stability [6,7] and the second is
that P– O and Ca2þ groups of HA molecules are rather
hydrophilous. However, up to date no systematic (Cg þ
HA þ water) studies like the ones for pure collagen or other
proteins and polynucleotide macromolecules and their
components [5,8] have been performed.
Thus, in the present paper, attempts are made to clarify
the molecular interactions in the system (Cg þ HA) and
[Cg þ (HA þ Ti)] in dry conditions as well as at various
relative humidities (RH), i.e. for the system (Cg þ HA þ
water) using: the IR and Raman spectroscopy as well as the
piezogravimetry and X-ray diffraction spectroscopy (XRD)
techniques.
2. Materials and methods
2.1. HA synthesis

* Corresponding author. Fax: þ 380-542-223760. A simplified model system was applied to produce
E-mail address: l_sukhodub@yahoo.com (L.F. Sukhodub). synthetic hydroxyapatite in vitro. The formation of HA in
0022-2860/$ - see front matter q 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.molstruc.2003.12.061
54 L.F. Sukhodub et al. / Journal of Molecular Structure 704 (2004) 53–58
different model systems, supported by automatic pH
stabilization was examined by Blumenthal et al. [9] earlier.
All experiments were performed with the following basic
reaction:
5CaðNO3 Þ2 ·4H2 O þ 3ðNH4 Þ2 HPO4 ! Ca5 ðPO4 Þ3 OH
þ 6NH4 NO3 þ 19H2 O þ 4HNO3
Two stock solutions of the reagent salts were prepared,
one containing the nitrate with a concentration of 0.16 M/l,
the other one containing the phosphate (0.1 M/l). Both water
solutions were mixed slowly under shaking to a total
volume of 300 ml. In the case of HA precipitation in the
presence of Ti metal ions, TiCl3 (15% water solution) was
added to the solution of calcium nitrate before mixing both
stock solutions. Then the calcium nitrate solution was added
dropwise to the ammonium hydrogenphosphate solution. To
keep the pH of the model system well in the basic region
(, 12) 20 ml of ammonia (25%) was added. After that the
mixture was heated up to 80 8C for about 10 min. After 24 h
of incubation at 37 8C the solid precipitate was obtained.
2.2. XRD
All XRD measurements were performed with a Siemens
D5000 diffractometer at the Institute of Nuclear Physics of
Muenster University, Germany. In all the cases, Cu Ka
radiation from a Cu X-ray tube (run at 30 mA and 40 kV)
was used. The samples were measured in the 2Q range from
5 to 65 8C with a measuring time of 1 s per 0.018. Various
polycrystalline phases could be distinguished by compari-
son with the diffraction data from the JCPDS database [10].
All samples were ground to fine powders and pressed into
sample holders to tablets with a diameter of 20 mm. Sample
holders were installed in the diffractometer with the surface
of the tablets well in the measurement plane.
2.3. IR spectroscopy
IR spectra of all the samples were measured using a two-
beam spectrophotometer UR-20 (Carl Zeiss, Jena,
Germany) with NaCI and KBr prisms. The spectral slit
width was 4.5 cm21 and the scan speed was 64 cm21/s. The
accuracy of the optical density measurements was no
less than 2% and that of the frequency measurements was
1 –2 cm21. A germanium filter was placed into the global
chamber in order to protect films against heating via
irradiation. All spectra were treated using the base line
correction. For the HA and the Cg spectra, a zero absorption
corresponds to the optical density at 1300 and 1800 cm21
correspondingly.
2.4. Raman spectroscopy
Raman spectra of HA were obtained using a spec-
trometer Z-16 (‘DILOR’, France) with a double
monochromator. An argon laser was applied (‘Innova 70’,
England) with excitation at 514 nm with power up to
100 mW. Scan speed was 50 cm21/min, spectral slit width
was 500 mm. Samples were placed in glass capillaries with
internal diameters of 1 mm. Temperature of the measure-
ments was 20 8C.
2.5. Piezogravimetry
The piezogravimetric method for determining of mass of
sorbed water per gram of a sorbent at various RH was
applied. The resonance frequency of the quartz resonator
depends on the mass of material placed on the surface of a
quartz plate. At first, the sample (thin film) on the
quartz plate was drying in a vacuum chamber. Then the
films were humidified by dosed injections of water
vapor to the chamber. Mass of sorbed water at a given
RH was determined by measuring the resonance
frequency changes between the dry and wet films. The
essential advantages of the method are: small amounts of
investigated materials (1025 g or less) and the extremely
high accuracy of measurements (^ 0.01 g of water per g of
a sample).
2.6. Collagen – HA sample preparation procedure
Pure hydroxyapatite crystals and HA with Ti ions
synthesized both at 37 8C and after sintering at 1000 8C
for 1 h were used. Collagen type I was extracted from the rat
tails at the Biological Department of Kharkov National
University (Ukraine). The collagen concentration in acetic
acid solutions corresponded to 2.3%. The mineralization of
collagen was obtained by adding 1 mg of HA fine powder to
2 ml of the collagen solution. The mixture was kept at 4 8C
for 24 h.
2.7. Collagen film preparation procedure
Film samples were prepared using a process of slow
evaporation of water at 4 8C from a pure collagen solution as
well as from a collagen solution containing HA. The
solutions were placed on horizontal fluorite substrates that
do not absorb irradiation within the investigated spectral
range and also are not soluble in water. Then the substrates
with obtained thin films were placed in a hermetic cell in
order to obtain full drying via vacuum pumping of the cell.
The necessary content of water in the films was reached by
keeping the films in the atmospheres created by saturated
solutions of inorganic salts. The films for the piezo-
gravimetric investigations were grown directly onto quartz
plates using the same procedure.
2.8. Pellet preparation procedure
The HA and HA-Ti samples in amounts of 1 –1.2 mg
were grinded to fine powders in an agate mortar together
 L.F. Sukhodub et al. / Journal of Molecular Structure 704 (2004) 53–58
with KBr (400 mg) powder and then the mixtures
were pressed into pellets. A thickness of the pellet was
about 1 mm.
3. Results and discussion
Phase analysis of the XRD patterns of the obtained
samples detected different calcium phosphates in the
diffraction patterns. In the samples without Ti ions, only
pure stoichiometric hydroxyapatite could be identified
(PDF-number 9-0432 [10]). With the increasing amounts
of the metal ions in the model system additional peaks in the
patterns indicated presence of the new phases in the HA
lattice (sintered at 950 8C) consisting mainly of tricalcium
phosphate (TCP). TCP has a Ca/P ratio of 1.5 (in
comparison to 1.67 for HA) and appears in two crystal-
lographic modifications, namely a-TCP and b-TCP (Fig. 1).
The latter phase is a crystallographic modification of a-TCP
with smaller thermostability. The occurrence of b-TCP can
be explained by thermal decomposition of non-stoichio-
metric HA with Ca/P ratios lower 1.67 and followed
possible incorporations of carbonate ions with formation a
so-called basic apatite, BCaP [12]. From the XRD patterns
of HA prepared with different concentrations of Ti ions
appears that the samples contained noticeable amount of
both TCP phases starting at the 2% level (Fig. 1). These two
modifications of TCP were observed in all the heated
samples that were produced in the presence of Ti and other
metal ions (V, Cr, Al), except Co [11].The presence of the
metal oxides in the samples at very high concentration
(20%) of Ti ions (Fig. 1) indicates their incorporation in the
precipitated HA followed by diffusion of these ions to
Fig. 1. XRD patterns of HA prepared with different concentrations of Ti
ions under physiological conditions and sintered at 950 8C. The main peaks
correspond to the HA phase, peaks marked as A, B, R and AN correspond to
a-TCP, b-TCP, Rutile and Anatase phases.
55
Fig. 2. IR spectra of HA samples at 37 8C and sintered at 1000 8C.
the surface of the powders and reaction with oxygen from
the air, caused by the thermal treatment [11].
Fig. 2 represents IR spectra of HA samples obtained at
37 8C (a) and after sintering at 1000 8C (b). The main band
frequencies (cm21) and assignments for HA are collected in
Table 1.
Comparison of the obtained IR spectra shows that the
samples heating leads to some spectral changes and
peculiarities, namely:
† decrease of the half-width for peaks connected to the n3
vibrations of the phosphate groups and, as a consequence
of this, the appearance of the new peak at 1090 cm21;
† high frequency shift of the bands in the 1035 –1050 cm21
range;
† disappearance of the bands connected to the vibrations
of the CO22 3 groups.
These changes show that the sintering of the calcium
phosphate samples leads to a more perfect crystal apatite
phase, free from the carbonate component. This last element
is a very important component of the bone minerals because
it influences several biological processes (caries, bone
resorption). The carbonate bands characteristic for the
B-type carbonate (the CO3 groups replace the PO4 groups)
were observed at around 1410 and 1480 cm21 (Fig. 2).
Thus, the carbonate phase found in the HA samples is of the
B-type. The bands of the carbonate replacing the OH groups
(A-type) at around 1550 cm21 were not detected in
Table 1
The bands frequencies and their assignments for the main HA absorption
bands
Frequency (cm21) Assignment
561, 574, 602 n4 O–P–O bend, triply degenerate
631 OH liberation
962 n1 P–O symmetric stretch
1032, 1046, 1087 n3 P–O antisymmetric stretch, triply degenerate
56 L.F. Sukhodub et al. / Journal of Molecular Structure 704 (2004) 53–58
Fig. 3. IR (a) and Raman (b) spectra of HA and HA þ Ti samples sintered at
1000 8C.
the samples. One of the possible mechanisms of the
carbonate incorporation into the HA lattice may include a
chemical reaction of carbon dioxide (coming from the air)
with the OH2 groups from the solution. After heat treatment
Fig. 4. Hydration isotherms for pure collagen and for collagen mineralized
by HA.
Fig. 5. IR spectra obtained at various values of relative humidity (RH) for
pure collagen (a) and for collagen mineralized by HA (b).
of the HA samples, the CO2 molecules are released from the
carbonate ions and as a result the calcium-deficient apatite
(Ca/P , 1.4) transforms into the stoichiometric one (Ca/
P ¼ 1.67) at the recrystalization temperature [13].
IR spectra of HA and HA-Ti samples at 37 8C show no
essential differences. Comparison of the IR spectra (Fig. 3a)
and the Raman spectra (Fig. 3b) of the HA and HA þ Ti
samples sintered at 1000 8C indicates that in the case of the
HA-Ti sample a noticeable redistribution of the band
intensities corresponding to the stretch (950 – 1150 cm21)
and the bend (550 – 620 cm21) vibrations of the phosphate
Table 2
The bands frequencies and their assignments for the main collagen
absorption bands
Frequency (cm21) Assignment
1640–1670 Amide I (CyO bend)
1540–1570 Amide II (N–H stretch, C– N bend)
1420–1460 C –N bend
1230–1250 Amide III (C–N bend, N–H stretch)
 L.F. Sukhodub et al. / Journal of Molecular Structure 704 (2004) 53–58
groups takes place. The frequency shifts of the absorption
bands are not observed in the IR spectra.
Fig. 4 demonstrates the hydration isotherms obtained at
the RH ranging from 0 to 90% for pure collagen and for the
collagen molecules mineralized by HA.
It should be noted that after vacuum drying (at 0% RH)
no less that one a residual water molecule per Gly-X-Y unit
has to remain in the sample. These water molecules stabilize
the triple helix structure of a collagen molecule. This very
lowly hydrated triple helix is native and can be reversibly
converted to the fully hydrated structure by addition of
water [10].
From the behavior of the obtained isotherms it follows
that the level of sorption of water molecules for the pure
collagen is rather higher than that for the Cg þ HA system.
It means that the HA molecules have a high tendency to
close some hydration sites within the collagen structure.
Fig. 5 presents the IR spectra of the pure collagen films
(a) and for the collagen mineralized with HA (without Ti
ions) obtained at various contents of water in the samples.
Fig. 6. A schematic diagram illustrating early mineralization of collagen in the water environment.
57
The assignments for the most important absorption bands in
the IR spectra of collagen are collected in Table 2.
From the obtained spectral data (Fig. 5) appears that the
increase of RH in the investigated system leads to a low
frequency shift and rising in intensity of the Amide I peak
for the pure collagen film (a). Beside this fact, the high
frequency shifts for Amide II and Amide III band are
observed. Probably, it is caused by the water molecules
binding to the CyO and C –N polar groups of the collagen.
At all the RH values Amide I/Amide II intensities ratio is
more than one (Fig. 5a).
For the (Cg þ HA) system, the IR spectra display some
important differences as compared to the pure Cg system:
† Amide I/Amide II ratio for the whole RH range is less
than 1;
† in dry conditions the Amide II frequencies and C –N
bend frequencies are shifted into the higher frequency
range (by 10– 20 cm21) (Fig. 5b).
58 L.F. Sukhodub et al. / Journal of Molecular Structure 704 (2004) 53–58
It follows from the literature that these bands are rather
sensitive to structural changes within collagen molecules
[14,15]. In particular, during its thermal denaturation, the
Amide II peak is shifted into the lower frequency range
(contrary to the one observed by us that shift to higher
frequencies). During the process of the fibril collection the
intensity of the Amide II band considerably increases
[16], but in our experiments we observed the opposite
process. Thus, the mineralization of collagen induces
some structural changes in Cg molecules without
denaturation. Changes of the HA absorption bands are
not observed (Fig. 5b).
In wet conditions, the mineralized Cg samples show an
increase of the intensities and the low frequency shift for
the Amide II band. These effects are caused by hydration
of the peptide groups, as it was observed for the pure
collagen.
Thus, piezogravimetry, IR and Raman spectroscopy are
very sensitive techniques for investigation of collagen–
hydroxyapatite – water interactions with different water
contents in the films. For the explanation of the observed
strong influence of hydroxyapatite on the structure of the
collagen hydration shell the ‘Cg þ HA þ water’ model is
proposed (Fig. 6).
The model consists of the two collagen layers with a
number of the cylindrically shaped molecules (3000 Å in
length and 12 Å in diameter) aggregated parallely. Two
types of hydration water around collagen peptide are
introduced. The first type corresponds to the so-called first
cylinder layer of hydration [6], whereas the second type
corresponds to the water shell that is localized in the gap
area, probably—with stronger interactions. These sites of
hydration in the case of the mineralized collagen molecules
are not available, at least partially, for water from the whole
hydration shell of Cg (Fig. 4). Possible interactions of the
HA crystals localized in the gap regions with their surface
water are not shown in the model.
4. Conclusions
1. Cg þ HA (with Ti ions) model system films have been
developed and used for investigations of some
peculiarities in the influence of water molecules on the
Cg fibrils mineralization both by pure HA and HA with
additives (other calcium phosphate phases and metal
ions-components of biological implants). A set of various
complementary experimental techniques (XRD, piezo-
gravimetry, IR and Raman spectroscopies) was used.
2. Mineralization of the Cg fibrils by the HA crystals
located within the gap area dramatically decreases a
number of water molecules present in the initial
hydration shell. The process causes some structural
changes within the Cg structure, but without denatura-
tion. The model which describes this phenomena has
been proposed.
3. The overall energy balance that governs early mineral-
ization process of the Cg peptides in water environment
should be developed for full understanding of the process
in the Cg þ HA þ water model proposed in this study.
Acknowledgements
This work was supported by the BMBF, WTZ-Ukr.
project, a DFG Merkator’s grant of Prof. L. Sukhodub at the
University of Muenster and by the Ministry of Science and
Education of Ukraine.
Authors are grateful to Prof. Ye. Persky for the samples
of collagen.
References
[1] T. Andrew Taton, Nature 412 (2001) 491–492.
[2] J.-H. Brandt, M. Mertig, A. Teresiak, W. Pompe, Chem. Mater. 11
(1999) 2694–2701.
[3] M.C. Chang, T. Ikoma, M. Kukuchi, J. Tanaka, J. Mater. Sci.: Mater.
Med. 13 (2002) 993–997.
[4] B. Bunneister, M. Gelinsky, H. Domaschke, A. Rosen-Wolf,
Th. Hanke, W. Pompe, Biomaterialien 4 (2) (2003).
[5] B. Kasemo, Surf. Sci. 500 (2002) 656 –677.
[6] J. Bella, M. Eaton, B. Brodsky, H.M. Berman, Science 266 (1994)
75– 81.
[7] S.K. Holmdren, K.M. Taylor, L.E. Bretscher, R.T. Raines, Nature 392
(1998) 666 –667.
[8] V. Maleev, M. Semenov, V. Kashpur, T. Bolbukh, A. Shestopalova,
D. Anishnenko, J. Mol. Struct. 605 (2002) 51–61.
[9] N.C. Blumenthal, A.S. Cosma, J. Biomed. Mat. Res. 23 (1989)
13– 22.
[10] JCPDS—International Center for Diffraction Data Powder diffraction
file (1990) 1–40.
[11] C. Moseke, L.F. Sukhodub, V. Pilipenko, B. Sulkio-Cleff, unpub-
lished.
[12] J.C. Elliott, Structures and Chemistry of the Apatite and other
Calcium Orthophosphates, Elsevier, Amsterdam, 1994.
[13] S. Danilchenko, L.F. Sukhodub, unpublished.
[14] Ch.A. Miles, M. Gelashvili, Biophys. J. 76 (1999) 3243–3252.
[15] H.A.J.S. Susi, R.J. Carroll, J. Am. Leather Chem. Assoc. 66 (11)
(1971) 508 –519.
[16] R.J. Jakobsen, L.L. Brown, T.B. Hutson, D.J. Fink, A. Veis, Science
220 (4603) (1983) 1288–1290.