Sei sulla pagina 1di 13

Materials Research Bulletin 43 (2008) 68–80

www.elsevier.com/locate/matresbu

Polymeric crystallization and condensation of calcium


polyphosphate glass
Sidney Omelon a,b, Andrew Baer c, Tom Coyle a, Robert M. Pilliar a,
Rita Kandel d, Marc Grynpas a,b,d,*
a
Materials Science and Engineering, University of Toronto, 184 College Street, Toronto, Canada
b
Samuel Lunenfeld Research Institute of Mt. Sinai Hospital, 600 University Avenue, Toronto, Canada
c
Department of Chemistry, University of Toronto, Toronto, Canada
d
Department of Laboratory Medicine and Pathobiology, University of Toronto, Toronto, Canada
Received 12 October 2006; received in revised form 26 January 2007; accepted 7 February 2007
Available online 11 February 2007

Abstract
Calcium polyphosphate (Ca(PO3)2)n (CPP) is under investigation as a resorbable bone biomaterial. Sintering CPP glass particles
in humid air produces a porous, interconnected, degradable, crystalline construct that is suitable for connective tissue engineering
applications. Porous CPP constructs sintered at 585 8C dissolved in water more rapidly than those sintered at 950 8C. FTIR, 31P
NMR, powder XRD, and density data for CPP glass and fully crystalline fibers were compared with data for the as-sintered and
partially dissolved constructs sintered at 585 or 950 8C. The results suggest that condensation continues during sintering, and CPP
glass crystallizes in a process analogous to the crystallization of linear organic polymers. During sintering, water vapor caused
hydrolytic degradation of the surface polyphosphate chains, forming a surface layer with different dissolution properties than the
particle interior. Thus, sintering CPP glass results in a heterogeneous crystalline product that impacts the dissolution rate of CPP as a
degradable biomaterial.
# 2007 Elsevier Ltd. All rights reserved.

Keywords: A. Ceramics; B. Crystal growth; C. Infrared spectroscopy; C. Nuclear magnetic resonance; C. X-ray diffraction

1. Introduction

Over 50 years ago, phosphate ions were observed to condense into poly(P) anions (PO3)n when dehydrated in the
melt phase [1]. Poly(P)s, also known as condensed phosphates, form linear poly(P) chains, poly(P) rings
(metaphosphates), or branched poly(P) structures (ultraphosphates) [2].
The Qn terminology of silicates also applies to phosphates. A phosphate ion connected with one bridging oxygen to
one other phosphate ion is termed a Q1 phosphate. A phosphate ion that is connected to two other phosphate ions is a
Q2 phosphate, as Q3 phosphates are connected to three other phosphates. Hydrolytic degradation is the reaction of a
water molecule with a P–O–P bond that cleaves the P–O–P bond. One of the Q3 P–O–P bonds is very susceptible to
hydrolytic degradation, therefore the branched ultraphosphates are stable in only water-free environments [3] and

* Corresponding author at: Samuel Lunenfeld Research Institute of Mt. Sinai Hospital, 600 University Avenue, Toronto, Canada.
Tel.: +1 416 586 4800x4464; fax: +1 416 586 1554.
E-mail address: grynpas@mshri.on.ca (M. Grynpas).

0025-5408/$ – see front matter # 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.materresbull.2007.02.015
S. Omelon et al. / Materials Research Bulletin 43 (2008) 68–80 69

were not considered in this study. Q2 and Q1 P–O–P bonds also react with water, eventually producing ortho-phosphate
(Q0) ions. However, the kinetics of Q2 and Q1 hydrolysis are not as rapid as the scission of the third oxygen-bridging Q3
bond. It is the depolymerization reaction with water that suggests that polyphosphates would be candidate resorbable
biomaterials.
Calcium polyphosphates (CPP) hydrolytically degrade into ortho-phosphate and calcium ions, which are major
components of bone mineral. This makes CPP a candidate for a resorbable bone biomaterial. CPP is also an effective
substrate for cartilage tissue engineering [4,5]. Previous in vitro [6–7] and in vivo [8] studies of porous, crystalline CPP
showed that the dissolution rate of crystalline CPP is slower than amorphous CPP [7]. The desirable degradation rate of
a bone biomaterial is in the order of a few months. It was reported that porous CPP sintered above 950 8C degrades
more slowly than desired for a bone tissue application [6].
Previous work indicated that the degradation rate of a porous, crystalline, degradable CPP construct can be
increased by sintering the construct at a lower temperature (585 8C versus 950 8C) [7]. The same crystal phase was
identified in both samples. The purpose of this study was to use other analytical methods to examine the effect of the
different sintering processes on the CPP product and its dissolution rate. This paper describes the mechanism of CPP
glass devitrification by comparing the density of CPP glass and crystalline CPP fibers seeded and grown from the melt
with porous CPP glass constructs sintered at 585 or 950 8C. The poly(P) chain length, water content, surface features,
and fracture surface features of CPP glass and the two sintered CPP products, before and after partial dissolution in
water, were measured and compared.

2. Experimental

2.1. Production of CPP glass

Using the method of Pilliar et al. [6], calcium phosphate monobasic monohydrate (J.T. Baker, Phillipsburg, NJ) was
calcined in a muffle furnace at 500 8C for 10 h to drive off the crystalline water and initiate condensation (Eq. (1)). The
calcined material was melted in an electric furnace at 1100 8C in a platinum crucible and ambient air for 1 h to further
polymerize the phosphates. The melt was quenched in deionized water. The cooled glass frit was milled in an agate
mill, and the products were sieved. The +76 to 105 mm fraction was used to produce the sintered, unprocessed CPP
construct with the desired porosity for bone tissue ingrowth [9].

½CaðPO3 Þ2 n OH2 þ ½CaðPO3 Þ2 m OH2 ! ½CaðPO3 Þ2 ðnþmÞ OH2 þH2 OðvÞ (1)

2.2. Production of CPP fibers

Griffith [10] developed a manufacturing process for crystalline CPP fibers. Briefly, a CPP melt close to the
metaphosphate composition (Ca/P molar ratio of 0.5) was cooled to a temperature between 570 and 970 8C.
Nucleation was induced by seeding the melt. Crystalline fiber growth proceeded by slow cooling of the seeded melt.

2.3. Sintering and crystallization of CPP porous constructs

Sieved CPP glass particles were gravity-sintered in a cylindrical platinum crucible (4 mm I.D.  40 mm H) in a muffle
furnace. The humidity of the ambient air around the furnace was controlled between 30 and 40%. The ‘950 8C-sintered
CPP’ constructs were made by sintering CPP glass particles at 585 8C for 1 h, then further heat treating them at 950 8C for
another hour. The ‘585 8C-sintered CPP’ constructs were produced by arresting the heat treatment after the glass particles
were sintered at 585 8C for 1 h. The sintered CPP rods exhibited a 35  2 volume percent porosity (calculated from the
bulk density) and were cut into disks 2 mm thick with a diamond blade on an Isomet saw (Buehler).

2.4. In vitro aging: partial dissolution study

Samples of both 585 8C- and 950 8C-sintered CPP disks (4 mm D  2 mm H) were individually placed in the wells
of a 24 well polystyrene tissue culture dish (Falcon) with 1 mL of deionized water at 37 8C and approximately 100%
70 S. Omelon et al. / Materials Research Bulletin 43 (2008) 68–80

humidity, simulating the tissue culture process. The deionized water solvent was replaced every two days, simulating a
tissue culture regime. The simulated tissue culture was run for 16 days in total; the samples retrieved from this
experiment will be referred to as ‘aged’ sintered CPP.
After the 16th day, the samples were dried in an oven (60 8C) for at least 2 h before the dry weight was measured in
triplicate (Mettler 1450). The percent change in 585 8C- and 950 8C-sintered, dried disk masses after 16 days was
compared (t-test, Sigmaplot# Version 8.02).

2.5. Characterization by scanning electron microscopy (SEM)

The surfaces and fracture surfaces of CPP disks were examined by SEM with secondary electron imaging (FEI
XL30) after gold coating. Fracture surfaces of CPP samples were generated by breaking the porous disks in tension
using the diametral compression test [11,12] at a loading rate of 0.5 mm/min (Instron). The fracture surfaces of the
sintered products were examined.

2.6. CPP density

A Retsch ball mill (Brinkmann) was used to grind the fibrous CPP samples for the density, the powder X-ray
diffraction, the FTIR and 31P NMR analyses. The as-sintered and aged CPP construct disks were milled by hand for the
same analytical measurements. CPP density was determined by the sink–float method in mixed solutions of toluene,
bromophorm and tetrabromoethane.
The sink–float method involves shaking CPP particles of interest within a ‘balancing solution’. The density
of the balancing solution is modified by changing the relative volumes of toluene (0.86 g/mL), bromophorm
(2.8 g/mL), both from Fisher Scientific and tetrabromoethane (2.967 g/mL) from Sigma–Aldrich until the
sample particles neither sink nor float in the balancing solution. The density of the balancing solutions was calculated
from the average mass of five equal volume aliquots on an AT250 balance (Mettler). The density calculation method
was validated by comparing the measured and reported density of pure tetrabromoethane (2.967 g/mL). Three
specimens were measured for each CPP sample type (amorphous, fully crystalline, 585 8C- and 950 8C-sintered
CPP).

2.7. Crystallinity index

Powder X-ray diffraction (P-XRD) patterns were generated using a powder diffractometer (Rigaku Multiflex,
0.5 82u/min, Cu Ka, 40 kV, 20 mA, 10–50 82u) and with Jade 6 (Materials Data). Sintered sample XRD patterns
were smoothed with a Savitzky-Golay quartic filter (21 points). As an indicator of the degree of crystallinity, the
crystallinity index was determined by the method of Wakelin [13]. This method compares X-ray intensity
differences between samples of the same crystal structure. Amorphous and fully crystalline samples are required
for this method as it takes advantage of the quantifiable reduction in X-ray diffraction intensity caused by an
amorphous content in the sample. At each 82u, the differences in X-ray intensities of the sample of interest and the
amorphous state (Iu  Ia) are compared to the differences in X-ray intensities of the fibrous, fully crystalline and
amorphous states (Ic  Ia). The crystallinity index is calculated from the intensity difference information by a
graphical or integral method.
By the graphical method, the difference in intensities between the sample and amorphous phases at
each 82u are plotted against the difference in intensities between the crystalline and amorphous phases. The degree
of crystallinity is determined by the slope of the line of best fit if the data points lie along a straight line
passing through the origin. By the integral method, the ratio of the sum of the absolute intensity differences
between the sample and the amorphous state to the sum of the absolute intensity differences between the fully
crystalline and the amorphous state at each 82u represents the relative fraction of crystalline content in the sample
(Eq. (2)).

P
jI u  I a j
crystallinity index ¼ P (2)
jI c  I a j
S. Omelon et al. / Materials Research Bulletin 43 (2008) 68–80 71

2.8. Fourier transform infrared spectroscopy (FTIR)

FTIR spectra were obtained with a Paragon 500 FT-IR Spectrophotometer (Perkin-Elmer). A series of 35 scans
were run from 4000 to 400 cm1, at an interval of 2 cm1 and with a resolution of 4 cm1 for four samples from each
group. Pellets were made by drying and pressing with KBr either amorphous, sintered or aged CPP samples retrieved
from the partial dissolution experiment.
Assuming that the IR absorption between 3100 and 3700 cm1 is due to water/–OH groups, the ratio of IR absorption
area for the water peak to all IR absorption peaks for each CPP sample was calculated with the absorption spectra [14]
using the Spectrum# software. An estimate of the comparative weight percent water values for the CPP samples was
calculated from this data, assuming that (a) there are 2-OH bonds absorbing IR energy for each water molecule, (b) the
areas under the absorption peaks between 2000 and 400 cm1 can be attributed to P–O bond absorbance only, and (c)
there are six P–O bonds absorbing IR energy per Ca(PO3)2 unit. The estimated mole fraction of absorbance due to water
bonds can be translated into an estimate of the weight fraction for the sample with the molecular weights of water and
(PO3)2 for comparison of the estimated water contents of the different CPP sample groups.
31
2.9. Solid state P nuclear magnetic resonance

High-resolution solid state 31P NMR spectra were recorded on an Avance DSX400 (Bruker) using magic angle
spinning (MAS). The NMR spectra were recorded using a standard 4 mm MAS probe at 162 MHz for 31P MAS with
samples spun at 5 and 8 kHz. Chemical shifts were obtained relative to an external standard of NH4(H2PO4). All
spectra were collected with a 908 pulse duration of 5 ms and a relaxation delay of 10 s. The number of transients
collected was typically 5000–6000.
Ortho-phosphate (Q0), end-chain (Q1), mid-chain (Q2), and branched (Q3) phosphorous atoms are easily identified
with 31P NMR by their different chemical shifts [15]. The different 31P chemical shifts in the solid state are clearly
identified with MAS [16], as they vary depending on the cation type [17] and concentration [18]. An estimation of the
poly(P) chain length [15] can be made by comparing the combined areas [19] under the main and sideband signals of
the Q1 and Q2 species.

3. Results and discussion

3.1. As-sintered particle surface examination (SEM)

Fig. 1(a) shows the morphology of the as-made (quenched and milled frit) amorphous CPP. The fracture/surface
features are characteristic of faceted, fractured glass. The fully crystalline CPP fibers grown from the melt displayed an
acicular habit in Fig. 1(b), which likely follows the c-axis of the b-Ca(PO3)2 crystalline structure [20–22] summarized
by Durif [23]. The surface of the 585 8C-sintered CPP construct is shown in Fig. 1(c), with a higher magnification
presented in Fig. 1(d). The particle surface was interspersed with distinct indentations (indicated by the arrows). These
indentations may be the locations of initial surface nucleation [24]. A fractured sinter neck surface is visible in the
bottom right corner of Fig. 1(d).
Fig. 1(e and f) shows the outer surfaces of the 950 8C-sintered CPP construct, that exhibited a slightly different surface
morphology. The initial sintered surface of the 950 8C-sintered CPP sample has a grain-like surface morphology that is
partially covered by a thin (less than 1 mm) smooth layer. The sinter neck fracture site in the lower edge of Fig. 1(f) reveals
the intra-particle micro-pores (less than 5 mm) that are characteristic of the 950 8C-sintered CPP construct.

3.2. As-sintered fracture surface examination (SEM)

Fig. 2(a and b) presents the sheet-like fracture surfaces of a 585 8C-sintered CPP particle (arrows). These images
suggest a lamellar structure within the particles.
The fracture surface of the 950 8C-sintered CPP particles shown in Fig. 2(c) also illustrates smooth planes on the
fracture surface (arrow), under a surface layer that appears to have fractured in a faceted manner. Spherical and
irregular micro-sized pores ranging in size from 0.04 to 4.13 mm (size range measured from three specimens and 350
measurements) are visible within the interior of the 950 8C sintered particles.
72 S. Omelon et al. / Materials Research Bulletin 43 (2008) 68–80

Fig. 1. SEM images of CPP Samples: (a) glass particles, (b) fully crystalline fibers, (c and d) 585 8C-sintered, and (e and f) 950 8C-sintered CPP.

The lamellar structure observed in Fig. 2(b) suggests that there are cleavage planes located within the particles. The
development of these cleavage planes by sintering may be explained if an analogy is drawn between the crystallization
mechanism of long-chained, un-branched organic polymers [25] and the crystallization of amorphous, long-chained CPP.
In this lamellar growth model of organic polymers, crystalline and amorphous interlayers may form during
crystallization, forming a semi-crystalline product. The lamellar model of long-chain organic polymer crystallization
was proposed for CPP crystallization from the amorphous state by Abe [26]. The different material properties of the
alternating crystalline and amorphous layers are proposed to account for the observed CPP lamellar fracture patterning.
The internal micro-pores observed on the fracture surface of the 950 8C-sintered CPP was initially attributed to the
volume change due to the transition from the amorphous to the crystalline state [27]. However, the spherical pores may
also have resulted from gas evolved during the sintering process [28,29]. Water evolution is a possible source of intra-
particular vapor if the poly(P) chains condensed during the sintering operation (Eq. (1)) assuming there are P-OH
groups at the ends of poly(P) chains within the glass. The by-product water vapor could coalesce and form the
observed intra-particle micro-pores during the sintering process.

3.3. CPP density

Table 1 compares the densities of the amorphous, fully crystalline CPP fibers, and sintered CPP solids with
literature values. The densities of the CPP glass and fully crystalline phases were comparable to literature values

Fig. 2. SEM images of sintered CPP fracture surfaces. (a and b) 585 8C-sintered CPP, (c) 950 8C-sintered CPP.
S. Omelon et al. / Materials Research Bulletin 43 (2008) 68–80 73

Table 1
CPP densities (n = sample size)
CPP sample type Density (g/cm3) (measured) Density (g/cm3) (literature) n
Glass 2.53  0.02 2.48–2.62 [26] 4
585 8C-sintered CPP 2.78  0.02 N/A 5
950 8C-sintered CPP 2.74 0.02 N/A 4
Crystalline fibers 2.85  0.02 2.85 [30] 3

[26,30]. The density value of the 950 8C-sintered CPP is not reliable, due to the intra-particle pores that are invisible to
the sink–float density measurement technique.
The intermediate density of the 585 8C-sintered CPP can be explained if it is regarded as a semi-crystalline
polymeric solid; the lower density of the 585 8C-sintered CPP may then be attributed to a 20  2% amorphous CPP
content [31] (Eq. (3)).

rc ðrs  ra Þ
%crystallinity ¼  100 (3)
rs ðrc  ra Þ

where rc is the density of fully crystalline (fibrous) CPP, rs the density of the 585 8C-sintered CPP, and ra is the density
of the CPP glass.

3.4. Partial dissolution (aging)

Fig. 3 shows the average percent reduction in the mass of the 585 8C- and 950 8C-sintered CPP constructs after 16
days in deionized water at 37 8C. The significantly higher (t-test, p = 0.001) dissolution rate of the 585 8C-sintered
CPP is attributed to the selective dissolution of an amorphous CPP content.

3.5. Aged CPP surface examination (SEM)

The surfaces of the aged 950 8C- and 585 8C-sintered CPP after a few hours and 16 days in deionized water are
demonstrated in Fig. 4(a–f). The smooth surfaces of the as-sintered CPP (Fig. 4(a and d)) were dissolved during the
aging process, exposing new and different surface morphologies (Fig. 4(b–c, e–f)).
The surface of the partially dissolved 950 8C-sintered CPP (Fig. 4(b and c)) shows grain-like surface features that
were not as well-defined before aging. The dissolution of a smooth surface layer of the 585 8C-sintered CPP
(Fig. 4(e)), revealed surface voids at the center of nucleation sites (arrows). Sixteen days of dissolution removed the
smooth surface layer, exposing selective dissolution of what could be laminar amorphous regions of the 585 8C-
sintered CPP (Fig. 4(f)).

Fig. 3. Percent submerged mass lost by 585 8C- and 950 8C-sintered CPP constructs during a 16-day dissolution experiment in deionized water
(n = 3).
74 S. Omelon et al. / Materials Research Bulletin 43 (2008) 68–80

Fig. 4. SEM images of sintered CPP surfaces. (a–c) 905 8C-sintered CPP. (a) Initial condition, after (b) 3 h and (c) 16 days in deionized water. (d–f)
585 8C-sintered CPP, (d) initial condition, after (e) 3 h and (f) 16 days in deionized water.

3.6. Crystallinity index (CI)

All CPP samples matched the CPP phase identified by the International Center for Diffraction Data powder
diffraction file (ICDD-PDF) card #17-0500 (Ca(PO3))2, calcium phosphate) (Fig. 5). No other phases were detected
for the samples analyzed in this study.
The CPP melt hold time has an effect on the glass devitrification temperature; CPP glass that is melted for less than
10 h undergoes anomalous crystallization below the devitrification temperature [32]. As the CPP glass used in this
study was melted for 1 h, it is expected that the anomalous crystallization would occur at the lower sintering
temperature of interest (585 8C).
The 585 8C-sintered CPP samples exhibited lower overall intensities than the fibrous, fully crystalline CPP. The
reduction in X-ray intensities of the sintered samples could be attributed to a reduction in crystallite size and/or to an
amorphous content in the specimen. If the mechanism by which linear organic polymers crystallize is analogous to the
mechanism by which linear poly(P)s crystallize, then the XRD method of Wakelin [13,33] would be a useful means to
measure the amorphous content in CPP samples. Fig. 6 shows a representative result of the graphical method that
compares the crystallinity indices (CI) of (A) as-sintered, 950 8C- and (B) 585 8C-sintered CPP powders. The lines of
best fit indicate an almost 100% CI for the 950 8C, as-sintered CPP, and a lower CI for the 585 8C, as-sintered CPP.
Table 2 summarizes the CI calculated by both graphical and integral methods for the as-sintered and aged 585 8C- and
950 8C-sintered samples (four specimens per group).
An average of the ratio of the 585 8C-sintered:950 8C-sintered CI values for the as-sintered (0.76  0.30) and aged
(0.88  0.29) CPP samples was calculated from the integral and graphical data. As the CI for the 950 8C-sintered is

Table 2
Crystallinity index values calculated by both graphical and integral methods (n = 4)
CPP sample type CI (%, graphical method) CI (%, integral method) Ratio CI (585 8C):CI (950 8C)
585 8C-sintered 69  22 83  18 As-sintered 0.76  0.30
950 8C-sintered 98  13 103  18
Partially dissolved, 585 8C-sintered 81  19 95  23 Aged 0.88  0.29
Partially dissolved, 950 8C-sintered 92  15 108  9
S. Omelon et al. / Materials Research Bulletin 43 (2008) 68–80 75

Fig. 5. P-XRD of CPP glass, 585 8C-sintered and fully crystalline CPP.

approximately 100% before and after aging, the increase in the aged CI ratio is a result of an increase in the 585 8C-
sintered CI. Should amorphous CPP be dissolved from the as-sintered 585 8C-sintered by the aging process, the aged
CI would be expected to increase. The increase in the aged 585 8C-sintered CPP CI supports the hypothesis that the
intermediate density of the as-sintered, 585 8C-sintered CPP may be due to an amorphous content.

Fig. 6. Example results of the graphical method of the crystallinity index from (A) 950 8C- and (B) 585 8C-sintered CPP.
76 S. Omelon et al. / Materials Research Bulletin 43 (2008) 68–80

Fig. 7. FTIR example scans: 585 8C-sintered CPP (A) aged and (B) as-sintered. The circles are a guide to the eye for peaks between 3100 and
2800 cm1.

3.7. FTIR

The FTIR spectra of all CPP samples was similar to those reported for CPP in literature [34,35]. Fig. 7 compares the
FTIR absorption spectra of (A) aged and (B) as-sintered 585 8C-sintered CPP. The area under the absorption curve for
the water/–OH peak at 3100–3700 cm1 was larger for the aged than the as-made sintered CPP samples. If it is
assumed that absorption at the water/–OH peak is indicative of the water content in the sample, then the percent of the
water/–OH peak area as calculated from all of the peak areas of a given sample would indicate the relative water
content of the CPP glass, as-sintered and aged CPP sample groups (Fig. 8).
The estimated water content of the CPP glass (1.4  0.3 wt%, n = 4) is close to the values reported in poly(P) glass
by Abe (0.6–0.8%) [26]. The higher water content measured in the sintered CPP solids (585 8C sintered: 3  0.3 wt%,
950 8C sintered: 3  0.3 wt%, n = 4) compared to the CPP glass was unexpected, as the only source of water during
the sintering process would have been water vapor in the ambient environment. Another source of water in the sintered
CPP samples is water produced during sintering by polyphosphate condensation in the solid state (Eq. (1)). Gomez
et al. [36] suggested that condensation of P–OH groups was responsible for water production in sintered sodium
polyphosphate and sodium-calcium polyphosphate glasses. Sales [37] reported condensation of Q0 (ortho-phosphates)
to Q1 (linear poly(P)) during devitrification of lead phosphate glasses, and Schneider and Jost [21] condensed b-
Ca(PO3)2n from Ca2[P4O12]4H2O at approximately 550 8C; these condensation processes would also have resulted in
water production within the solid.

Fig. 8. Weight percent estimate of water content (n = 4).


S. Omelon et al. / Materials Research Bulletin 43 (2008) 68–80 77

After aging in water, the water content of the aged, dried samples sintered at either temperature increased over the
initial as-sintered value (585 8C sintered: 10  1 wt%, 950 8C sintered: 8  1 wt%, n = 4). The increase in percent
water content caused by aging the 585 8C-sintered CPP can be explained by the presence of amorphous CPP in the as-
sintered samples. Kasuga et al. [38] reported that amorphous CPP transforms into a hydrogel before it dissolves in
water. Assuming amorphous CPP located in the sintered CPP constructs transforms into a hydrogel before dissolving,
partially dissolved CPP hydrogel would entrap water within the hydrogel matrix when the aged samples were dried
before analysis. The increased water content of the aged, 950 8C-sintered CPP may be due to trapped water in the intra-
particle micro-pores from the aging experiment (Fig. 1(f)) that was not removed by sample drying.
If the aging process caused hydrolytic degradation of poly(P), then new P–OH bonds would be generated where the
P–O–P bonds were broken and an increase in P–OH stretches (3100 and 2800 cm1) [26] would be detected. No
quantitative measurement of the absorbance area representing the energy of the P–OH bonds was attempted because
the peak areas were very small. However, the dashed circles in Fig. 7 indicate small peaks at these locations for the
aged 585 8C-sintered CPP sample (A), while there is no clear peak for the as-sintered CPP sample (B), suggesting that
some hydrolytic degradation of poly(P) could be caused by aging in deionized water.

31
3.8. P NMR

The measured Q1 (11 to 19 ppm) and Q2 (29 to 38 ppm) peak positions were assigned according to the Q1
and Q2 chemical shifts for CPP reported by Fletcher [39] and Witter [40] (Fig. 9). The Q2 peak intensity is much larger
than the Q1 peak intensity. This is expected, as the CPP glass is made with a CPP composition (Ca/P = 0.5) that favors
the formation of long, linear poly(P) chains composed of many Q2 phosphates.
A comparison of the average poly(P) chain lengths for the amorphous, sintered, and partially dissolved CPP
samples can be made by comparing the areas under the Q1 and Q2 peaks. Because the Q1 species are associated with
the ends of linear PP chains, an increase in the percent Q1 area of the total Q1 and Q2 areas is associated with a decrease
in the average poly(P) chain length. Table 3 shows the percent Q1 areas of the CPP samples and the proposed poly(P)
speciation that explains these results.
The sintered CPP constructs would have a higher fraction of Q1 phosphates than the original glass if the water vapor
in the furnace hydrolytically degraded the surface poly(P) chains, generating more end-chain phosphates. The aged,
sintered CPP solids are believed to have a lower Q1 fraction than the original glass for two reasons: the poly(P) chains
inside of the particles condensed during the sintering process, and the shorter surface poly(P) chains were removed by
the dissolution experiment.

3.9. Proposed model

The model proposed to explain the results is summarized in Fig. 10. Sintering CPP glass in humid air generates a
population of shorter poly(P) chains on the CPP particle surfaces because high activity water vapor hydrolytically

31
Fig. 9. P MAS NMR spectrum example (950 8C-sintered CPP). Asterisks indicate stationary peaks.
78 S. Omelon et al. / Materials Research Bulletin 43 (2008) 68–80

Table 3
Polyphosphate Q1 population fraction calculated from 31
P MAS NMR
CPP sample type ðQ1area =ðQ1 þ Q2 Þarea Þ100 Comments

Amorphous 3.1
Poly(P) chains formed in the melt

585 8C-sintered 6.0


Surface poly (P) chains hydrolyzed

Interior poly(P) chains condense

950 8C-sintered 4.5


Surface poly(P) chains further hydrolyzed

Interior poly(P) chains continue to condense

585 8C-sintered (aged) 2.7 Shorter surface poly(P) chains dissolved. Longer interior poly(P) chains detected

950 8C-sintered (aged) 2.9 Shorter surface poly(P) chains dissolved. Longer interior poly(P) chains detected

Fig. 10. (a) Amorphous CPP network, (b) 585 8C sinter: thin surface layer of shorter poly(P) chains, interior consists of elongated poly(P) chains
with alternating amorphous and crystalline regions, (c) 950 8C sinter: thicker surface layer of shorter poly(P) chains, crystalline interior composed of
elongated poly(P) chains with internal micro-porosity, (d) partially dissolved, 585 8C-sintered CPP: shorter poly(P) chains on the surface have
dissolved, amorphous content transforms to hydrogel and is preferentially dissolved (e) partially dissolved, 950 8C-sintered CPP: shorter poly(P)
chains on the surface have dissolved, the crystalline interior dissolves very slowly.
S. Omelon et al. / Materials Research Bulletin 43 (2008) 68–80 79

degrades poly(P) (Eq. (1), reverse direction). As no ortho-phosphates were detected by the 31P NMR analysis, it is
assumed that mid-chain scission occurred at the surface, producing smaller poly(P) chains from the longer surface
poly(P) chains.
This surface layer of shorter, more mobile poly(P) resulted in a thin surface layer that could crystallize more rapidly
than the bulk material. In the interior, crystallization proceeded gradually via a mechanism similar to linear inorganic
polymer crystallization (Fig. 10(b)). The interior poly(P) chains condensed (lengthened), reducing the average Q1
population and producing water vapor that formed micro-pores (Fig. 10(c)). The overall Q1 fraction in sintered CPP is
larger than the initial glass because of greater number of end-chain phosphates in the surface layer. Both the aged
585 8C- and 950 8C-sintered samples were composed of a lower fraction of Q1 species than the original CPP glass
because a preferential dissolution of the shorter poly(P) chains on the surfaces removed the shorter surface poly(P),
revealing the longer poly(P) chains formed from the sintered glass. The amorphous CPP in the aged 585 8C-sintered
construct transformed into a hydrogel and dissolved more rapidly than the crystalline CPP.

4. Conclusions

Sintering amorphous CPP solids in humid air does not result in simple sintering, crystallization, and the formation
of a porous construct as expected for other calcium phosphate ceramics. A new model for CPP glass reaction and phase
transformation during sintering in humid air is proposed. This polymeric crystallization mechanism can have an
impact on the dissolution mechanism of sintered CPP in deionized water and possibly other dissolving media.

Acknowledgements

We thank Dan Mathers for FTIR assistance, Dr. Jason Hong for fabricating the CPP glass, Dr. Jian Wang for
performing the diametral compression tests, as well as Drs. James Anson and Ben Hatton for advice and discussion on
the manuscript. The Natural Sciences and Engineering Research Council of Canada (NSERC) is gratefully
acknowledged for funding this research project.

References

[1] J.R. Van Wazer, J. Am. Chem. Soc. 72 (1950) 644–647.


[2] J.R. Van Wazer, Phosphorus and Its Compounds, vol. 1, Interscience Publishers Inc., New York, 1958.
[3] G. Palavit, C. Mercier, L. Montagne, M. Drache, Y. Abe, J. Am. Ceram. Soc. 81 (1998) 1521–1524.
[4] S.D. Waldman, M.D. Grynpas, R.M. Pilliar, R.A. Kandel, J. Biomed. Mater. Res. 62 (2002) 323–330.
[5] C.A. Séguin, M.D. Grynpas, R.M. Pilliar, S.D. Waldman, R.A. Kandel, Spine 29 (2004) 1299–1307.
[6] R.M. Pilliar, M.J. Filiaggi, J.D. Wells, M.D. Grynpas, R.A. Kandel, Biomaterials 22 (2001) 963–972.
[7] N.L. Porter, R.M. Pilliar, M.D. Grynpas, J. Biomed. Mater. Res. 56 (2001) 504–515.
[8] M.D. Grynpas, R.M. Pilliar, R.A. Kandel, R. Renlund, M.J. Filiaggi, M. Dumitriu, Biomaterials 23 (2002) 2063–2070.
[9] R.M. Pilliar, J. Biomed. Mater. Res. 21 (1987) 1–33.
[10] E.J. Griffith, U.S. Patent, 4,360,625 (1982).
[11] M.K. Fahad, J. Mater. Sci. 31 (1996) 3723–3729.
[12] T. Hinoki, T. Lara-Curzio, L.L. Snead, 27th International Cocoa Beach Conference on Advanced Ceramics and Composites, Coca Beach, FL,
(2003), pp. 34–38.
[13] J.H. Wakelin, H.S. Virgin, E. Crystal, J. Appl. Phys. 30 (1959) 1654–1662.
[14] B.C. Smith, Infrared Spectral Interpretation: A Systematic Approach, CRC Press, Boca Raton, 1999, p. 6 (Chapter 1).
[15] J.R. Van Wazer, C.F. Callis, J.N. Shoolery, J. Am. Chem. Soc. 77 (1955) 4945–4946.
[16] T.M. Duncan, D.C. Douglass, Chem. Phys. 87 (1984) 339–349.
[17] R.K. Brow, C.C. Phifer, G.L. Turner, R.J. Kirkpatrick, J. Am. Ceram. Soc. 74 (1991) 1287–1290.
[18] R.K. Brow, R.J. Kirkpatrick, G.L. Turner, J. Non-Cryst. Solids 116 (1990) 39–45.
[19] D. Massiot, F. Fayon, M. Capron, I. King, S. Le Calvé, B. Alonso, J.-O. Durand, B. Bujoli, Z. Gan, G. Hoatson, Magn. Reson. Chem. 40 (2002)
70–76.
[20] W. Rothammel, H. Burzlaff, R. Specht, Acta Crystallogr., Sect. C: Cryst. Sruct. Commun. C45 (1989) 551–553.
[21] M. Schneider, K.-H. Jost, Z. Anorg. Allg. Chem. 527 (1985) 99–104.
[22] A.O. McIntosh, W.L. Jablonski, Anal. Chem. 28 (1956) 1424–1427.
[23] A. Durif, Crystal Chemistry of Condensed Phosphates, Plenum Press, New York, 1995.
[24] W. Höland, G. Beall, Glass-Ceramic Technology, The American Ceramic Society, 2002, p. 67 (Chapter 1.5).
80 S. Omelon et al. / Materials Research Bulletin 43 (2008) 68–80

[25] J.M. Schultz, Polymer Crystallization: The Development of Crystalline Order in Thermoplastic Polymers, Oxford University Press,
Washington, DC, 2001, pp. 1–6 (Chapter 1).
[26] Y. Abe, in: M. Grayson, E.J. Griffith (Eds.), Topics in Phosphorus Chemistry, vol. 11, John Wiley & Sons, Toronto, 1983, pp. 19–67.
[27] A. Wantanabe, Y. Imada, S. Kihara, J. Am. Ceram. Soc. 69 (1986) C-31–C-32.
[28] D.E.J. Talbot, Int. Metall. Rev. 20 (1975) 166–184.
[29] A. Nouruzi-Kharasani, METAL Cast. Surf. Finish. (1996) 34–39.
[30] A. Wantanabe, M. Mitsudou, S. Kihara, Y. Abe, J. Am. Ceram. Soc. 72 (1989) 1499–1500.
[31] W.D. Callister, Materials Science and Engineering: An Introduction, John Wiley and Sons, 2003, p. 470 (Chapter 15).
[32] R.C. Ropp, Inorganic Polymeric Glasses, Studies in Inorganic Chemistry, vol. 15, Elsevier, New York, 1992, pp. 116–185 (Chapter 3).
[33] W.O. Statton, J. Appl. Polym. Sci. 7 (1963) 803–815.
[34] K. Meyer, J. Non-Cryst. Solids 209 (1997) 227–239.
[35] K. Meyer, H. Hobert, A. Barz, D. Stachel, Vib. Spectrosc. 6 (1994) 323–332.
[36] F. Gomez, P. Vast, P. Llewellyn, F. Rouquerol, J. Non-Cryst. Solids 222 (1997) 415–421.
[37] B.C. Sales, J.O. Ramey, L.A. Boatner, Phys. Rev. Lett. 59 (1987) 1718–1721.
[38] T. Kasuga, T. Wakita, M. Nogami, M. Sakurai, M. Wantanabe, Y. Abe, Chem. Lett. 8 (2001) 820–821.
[39] J.P. Fletcher, R.J. Kirkpatrick, D. Howell, S.H. Risbud, J. Chem. Soc., Faraday Trans. 89 (1993) 3297–3299.
[40] R. Witter, P. Hartmann, J. Vogel, C. Jäger, Solid State Nucl. Magn. Reson. 13 (1998) 189–200.

Potrebbero piacerti anche