10
CERAMICS INTERNATIONAL Available online at www.sciencedirect.com Ceramics International 40 (2014) 1323113240 Hydroxyapatite and chloroapatite derived from sardine by-products C. Piccirillo a , R.C. Pullar b , D.M. Tobaldi b , P.M. L. Castro a , M.M. E. Pintado a,n a Centro de Biotecnologia e Química Fina Laboratório Associado, Escola Superior de Biotecnologia, Universidade Católica Portuguesa, Porto, Portugal b Departamento Engenharia de Materiais e Cerâmica / CiCeCO, Universidade de Aveiro, Aveiro, Portugal Received 31 March 2014; received in revised form 5 May 2014; accepted 9 May 2014 Available online 16 May 2014 Abstract In this paper, phosphate-based compounds used in biomedicine were extracted from bones and scales of European sardines (Sardina pilchardus); this is the rst time that different parts of the same sh are used for the extraction of these kinds of materials. The bones and scales behave very differently with processing, producing different materials when annealed between 600 and 1000 1C. The bones formed a mixture of hydroxyapatite (Ca 10 (PO 4 ) 6 (OH) 2 , HAp) and β-tri-calcium phosphate (β-Ca 3 (PO 4 ) 2 , β-TCP), with a higher content of β-TCP obtained with increasing temperature. This bi-phasic material has a high added value, as it is employed as a bioceramic; in fact HAp has good biocompatibility while β-TCP has better resorbability than HAp, despite being less biocompatible. With scales, on the other hand, either a HAp-based material or a chlorine-substitute HAp containing material (chloroapatite (Ca 10 (PO 4 ) 6 Cl 2 , ClAp) were produced. HAp-based material was obtained with a simple annealing process; for ClAp, on the other hand, a combined washingannealing process was used. ClAp is also used in biomedicine, due to its improved resorption, mechanical properties and bioactivity. This is the rst time ClAp of marine origin was produced. & 2014 Elsevier Ltd and Techna Group S.r.l. All rights reserved. Keywords: Bone implants; Bones and scales sh by-products; Calcium phosphate; Chloroapatite; Hydroxyapatite 1. Introduction Calcium phosphate-based compounds are used in several technological applications. Many of these compounds show very high biocompatibility; because of this, they are employed as biomaterials, in particular to fabricate bone implants, bone cements and dental pastes [1,2]. Their use in this eld is mainly due to the compositional similarities that these com- pounds have with human bones and teeth. Hydroxyapatite (Ca 10 (PO 4 ) 6 (OH) 2 , HAp) is probably the most used compound in this eld. It is the main component of human bones and, therefore, it is the species which best mimics its mineral composition and behaviour. Tri-calcium phosphate (Ca 3 (PO 4 ) 2 , TCP) is also employed in this eld. It can exist in two possible forms, α and β, and normally the β form gets converted into the α with annealing at temperatures higher than 12501300 1C [3]. For many biomaterial applications, a mixture of HAp and TCP is often considered [4]; this is because TCP has better resorbability than HAp, despite being less biocompatible. Apart from these applications in biomedicine, HAp has other uses too, especially for environmental remediation. HAp can be employed to remove heavy metals from contaminated waters and/or soils [5,6]; moreover, some forms of HAp show photocatalytic activity and can be used to decompose organic contaminants [7,8]. Chlorine-substituted HAp is also a compound with interesting potential. Chloroapatite (Ca 10 (PO 4 ) 6 Cl 2 , ClAp), where hydroxyl groups are completely substituted by chlorine, can be used in electronics as a phosphor material, due to its uorescence [9]. Compounds with only partial substitution, however, have been investigated for possible applications in biomedicine; the presence of some chlorine in the HAp lattice can actually improve its resorption, as well as mechanical properties [10]. Literature data also show that, in some cases, HAp with high chlorine substitution showed greater bioactivity than non-substituted HAp [11] . www.elsevier.com/locate/ceramint http://dx.doi.org/10.1016/j.ceramint.2014.05.030 0272-8842/& 2014 Elsevier Ltd and Techna Group S.r.l. All rights reserved. n Corresponding author. E-mail address: [email protected] (M.M. E. Pintado).

Hydroxyapatite and chloroapatite derived from sardine by-products

Embed Size (px)

Citation preview

CERAMICSINTERNATIONAL

Available online at www.sciencedirect.com

Ceramics International 40 (2014) 13231–13240

Hydroxyapatite and chloroapatite derived from sardine by-products

C. Piccirilloa, R.C. Pullarb, D.M. Tobaldib, P.M. L. Castroa, M.M. E. Pintadoa,n

aCentro de Biotecnologia e Química Fina – Laboratório Associado, Escola Superior de Biotecnologia, Universidade Católica Portuguesa, Porto, PortugalbDepartamento Engenharia de Materiais e Cerâmica / CiCeCO, Universidade de Aveiro, Aveiro, Portugal

Received 31 March 2014; received in revised form 5 May 2014; accepted 9 May 2014

Available online 16 May 2014

Abstract

In this paper, phosphate-based compounds used in biomedicine were extracted from bones and scales of European sardines (Sardina

pilchardus); this is the first time that different parts of the same fish are used for the extraction of these kinds of materials. The bones and scales

behave very differently with processing, producing different materials when annealed between 600 and 1000 1C.

The bones formed a mixture of hydroxyapatite (Ca10(PO4)6(OH)2, HAp) and β-tri-calcium phosphate (β-Ca3(PO4)2, β-TCP), with a higher

content of β-TCP obtained with increasing temperature. This bi-phasic material has a high added value, as it is employed as a bioceramic; in fact

HAp has good biocompatibility while β-TCP has better resorbability than HAp, despite being less biocompatible.

With scales, on the other hand, either a HAp-based material or a chlorine-substitute HAp containing material (chloroapatite (Ca10(PO4)6Cl2,

ClAp) were produced. HAp-based material was obtained with a simple annealing process; for ClAp, on the other hand, a combined washing–

annealing process was used. ClAp is also used in biomedicine, due to its improved resorption, mechanical properties and bioactivity. This is the

first time ClAp of marine origin was produced.

& 2014 Elsevier Ltd and Techna Group S.r.l. All rights reserved.

Keywords: Bone implants; Bones and scales fish by-products; Calcium phosphate; Chloroapatite; Hydroxyapatite

1. Introduction

Calcium phosphate-based compounds are used in several

technological applications. Many of these compounds show

very high biocompatibility; because of this, they are employed

as biomaterials, in particular to fabricate bone implants, bone

cements and dental pastes [1,2]. Their use in this field is

mainly due to the compositional similarities that these com-

pounds have with human bones and teeth.

Hydroxyapatite (Ca10(PO4)6(OH)2, HAp) is probably the most

used compound in this field. It is the main component of human

bones and, therefore, it is the species which best mimics its mineral

composition and behaviour. Tri-calcium phosphate (Ca3(PO4)2,

TCP) is also employed in this field. It can exist in two possible

forms, α and β, and normally the β form gets converted into the

α with annealing at temperatures higher than 1250–1300 1C [3].

For many biomaterial applications, a mixture of HAp and TCP is

often considered [4]; this is because TCP has better resorbability

than HAp, despite being less biocompatible.

Apart from these applications in biomedicine, HAp has

other uses too, especially for environmental remediation. HAp

can be employed to remove heavy metals from contaminated

waters and/or soils [5,6]; moreover, some forms of HAp show

photocatalytic activity and can be used to decompose organic

contaminants [7,8].

Chlorine-substituted HAp is also a compound with interesting

potential. Chloroapatite (Ca10(PO4)6Cl2, ClAp), where hydroxyl

groups are completely substituted by chlorine, can be used in

electronics as a phosphor material, due to its fluorescence [9].

Compounds with only partial substitution, however, have been

investigated for possible applications in biomedicine; the presence

of some chlorine in the HAp lattice can actually improve its

resorption, as well as mechanical properties [10]. Literature data

also show that, in some cases, HAp with high chlorine substitution

showed greater bioactivity than non-substituted HAp [11].

www.elsevier.com/locate/ceramint

http://dx.doi.org/10.1016/j.ceramint.2014.05.030

0272-8842/& 2014 Elsevier Ltd and Techna Group S.r.l. All rights reserved.

nCorresponding author.

E-mail address: [email protected] (M.M. E. Pintado).

The majority of HAp used today is synthetic, prepared with

various techniques [1]. However, HAp and other calcium

phosphate materials can also be prepared from natural sources

and/or wastes and by-products. The use of agri-food by-

products, in particular, has attracted more and more interest

in recent years. This is due to the production of increasing

amounts of waste and/or by-products, which have to be

disposed of with environmental impact. Extracting and/or

obtaining compounds which have high value is, therefore, a

way of addressing this problem while valorising such waste.

HAp can be obtained from by-products of the food industry,

such as animal or fish bones; in fact, HAp is the main

component of the bones themselves, with the remaining part

being made of organic matter (i.e. collagen). Several studies

have been published on this topic: bovine and pig bones, for

instance, were used to extract HAp [12,13], and several kinds

of fish bones have also been used, including cod fish,

swordfish and tuna [14–17].

Fish scales, made of HAp and collagen, can also be

employed to extract HAp. Literature reports exist of HAp

extracted from fish scales for applications in both biomedical

and environmental fields [18–22].

In this paper, we report a study on the use of wastes from

European sardines (Sardina pilchardus) to extract phosphate-

based compounds; both bones and scales were considered. It is

the first time that different parts of the same fish were used for

the extraction of this kind of compound. The materials obtained

were characterised by several techniques (X-ray diffraction,

thermogravimetry, IR spectroscopy and SEM micrography) to

determine their characteristics, and address the possible differ-

ences according to the sources.

2. Materials and methods

2.1. Sample storage and pre-annealing treatment

Sardine bones and scales were provided by A Poveira

(Povoa de Varzim, Portugal). After collection, they were

stored at �15 1C. Before treatment, the bones were defrosted

and cleaned manually with hot water to remove impurities (i.e.

fragments of meat, skin, etc.). They were then dried at 40 1C

overnight. The scales were defrosted and dried at 40 1C

overnight. For selected experiments, scales were soaked in

water prior to, or after, the annealing, to remove sodium

chloride. The process was performed by placing 1 g of dried

scales in 40 ml of water, and leaving them for 24 h. Then the

water was removed and 40 ml of clean distilled water was

added to the same scales; they were then left in water for

another 24 h. The scales were then removed from the water

and dried at 40 1C overnight.

2.2. Annealing and post-annealing treatment

Both bones and scales were annealed in a Nabertherm muffle

furnace. The heating rate was 5 1C/min and the annealing time

was 1 h. Several annealing temperatures between 600 and

1000 1C were used. For some of the scales calcined at 700 1C,

a post-annealing treatment was also performed, as the calcined

scales (in powder form) were washed in water by the same

process described above.

2.3. Analysis of the annealed samples

The phases present in the samples were initially determined

by X-ray diffraction (XRD) with a Rigaku Geigerflex D/max

C-series, with Cu Kα radiation. For multiphasic samples, a

quantitative phase analysis (QPA) of the crystalline phases was

assessed by the Rietveld method [23]. XRD data for the QPA

were collected on a θ/θ PANalytical X'Pert Pro diffractometer,

equipped with a fast RTMS detector (PIXcel-1D), and with Cu

Kα radiation (generated at 40 kV and 40 mA, 10–8012θ range,

a virtual step scan of 0.016712θ and virtual time per step of

50 s). The Rietveld refinements were assessed using the GSAS

software package with its graphical interface EXPGUI [24,25].

The staring atomic parameters of HAP, ClAp, and β-TCP were

taken from previous work by the authors [26], and those of

NaCl from Walker et al. [27]. The following parameters were

refined: scale-factors, background (fitted adopting a 6th order

shifted Chebyshev function), zero-point, unit cell parameters

and profile coefficients – one Gaussian (GW) and two

Lorentzian terms (LX and LY) – and preferred orientation

along the [010] axis of HAp, modelled using the March–

Dollase formalism [28].

SEM micrographs were taken using a Hitachi S-4100 at

25 kV; samples were sputtered with gold prior to the analysis.

Thermogravimetric (TGA) and differential thermal analyses

(DTA) were performed using a Setaram Labsys S/101 TG-DTA

with flowing air.

Infrared transmission spectra were acquired using a FT-IR

PerkinElmer 100 instrument. The spectra were taken on discs

made from 5 mg of each sample mixed with 200 mg of KBr in

transmittance mode.

For elemental analysis, phosphorus concentration was

determined by a spectrophotometric method: P-containing

samples were reacted with the Merck Spectroquant phosphorus

reagent kit, containing an acidified solution of NH4VO3 and

(NH4)6Mo7O24 � 4H2O to form an orange–yellow coloured

compound of H4PMo11VO40 (molybdovanadophosphoric

acid). This compound was analysed spectrophotometrically,

using a PerkinElmer Lambda 25 spectrometer, at 400 nm,

against a calibration curve of standard KH2PO4 made in ultra-

pure water. Calcium and sodium concentrations were measured

using flame atomic absorption spectroscopy (FA-AAS) in an

UNICAM 960s spectrophotometer (Waltham, USA). Chlorine

and fluorine concentrations were determined using Crison

selective ion electrodes for chloride and fluoride ions, respec-

tively. The samples were measured against calibration curves

of NaCl or NaF.

For elemental composition determination, three experiments

were performed; an average value and the corresponding standard

deviation were calculated. Statistical analysis (ANOVA test) was

performed to compare the composition of some samples (95%

confidence, po0.05).

C. Piccirillo et al. / Ceramics International 40 (2014) 13231–1324013232

3. Results and discussion

3.1. Annealing and characterisation of the bones

Fig. 1 shows the XRD patterns of the bones annealed at

different temperatures between 600 and 1000 1C. It can be seen

that a second phase as well as HAp is detected in all samples,

which is β-TCP—the peaks corresponding to β-TCP are marked

with a β in Fig. 1. The presence of this compound was

previously reported in HAp samples of marine origin [15,17].

The relative intensities of the peaks belonging to the two

phases change depending on the annealing temperature; this

indicates a change in the relative content of the two phases

during the calcination process. Table 1 reports the percentage

weight composition at different annealing temperatures, calcu-

lated from the XRD patterns. It can be seen that the amount of

HAp decreases, while that of β-TCP increases, as temperature

increases.

Bones annealed at 1000 1C (sample B5) were analysed to

determine their elemental composition to have a more com-

plete picture of this material (Table 2). It can be seen that the

molar ratio between calcium and phosphorus is only 1.46; this

value is smaller than the stoichiometric ratios of both HAp

and β-TCP – 1.67 and 1.50 respectively. This explains the

progressive conversion of HAp to β-TCP with increasing

temperature, as β-TCP is favoured under these conditions

[3]. This value may seem too small for a material with this

composition; HAp-based materials with non-stoichiometric

amounts of calcium and phosphorus, however, were already

observed in literature. Zhang, for instance, reports a calcium-

deficient HAp with a Ca/P ratio as low as 1.49 [29]. Moreover

biphasic materials of marine origin made of HAp and β-TCP

showed Ca/P ratios both higher and lower than the theoretical

stoichiometric values, i.e. 1.49 [17] and 1.87 [14].

Samples calcined at other temperatures showed similar

elemental composition (data not shown), with values which

were not statistically different and which did not indicate a

significant variation in composition with calcination

temperature.

Table 2 also shows the presence of other elements in the

sample as minor components; most of the values are compar-

able with the other HAp-based materials of natural origin

[17,30]. The only exception is sodium, whose concentration is

higher than that normally observed in this kind of samples. The

presence of all of these minor elements can be beneficial for

properties such as biocompatibility and bioactivity as already

mentioned above; for instance, chlorine improved HAp resorp-

tion. Sodium, magnesium and fluorine, on the other hand, can

have positive effects on osteoporosis, bone calcination and

osteointegration, respectively [4,31].

Fig. 2(a) shows the TGA spectrum for the bones heated to

1100 1C. A total mass loss slightly greater than 60% can be

observed. To better understand each step taking place during

20 30 40 50 60 700

2000

4000

6000 B5

B4

B3

B2

B1

ββ

β

Inte

nsity (

a.u

.)

2 θ (degrees)

Fig. 1. XRD patterns for sardine bones calcined at different temperatures

(600–1000 1C). The letter β indicates peaks belonging to β-TCP.

Table 1

Phase composition (wt%) of the bones annealed at different temperatures.

Sample Annealing temperature (1C) HAp β-TCP

B1 600 85.8 14.2

B2 700 77.9 22.1

B3 800 73.7 26.3

B4 900 65.0 35.0

B5 1000 56.2 43.8

Table 2

Element composition for bone sample B5 calcined at 1000 1C.

Element Ca P Na Cl F Ca/P (mol/mol)

Concentration (wt%) 32.7 17.4 1.74 0.37 0.108 1.46

-60

-45

-30

-15

0

we

igh

t lo

ss (

%)

Temperature (°C)

Temperature (°C)

0 200 400 600 800 1000

0 200 400 600 800 1000

-0.20

-0.15

-0.10

-0.05

0.00

0.05

∆m

/∆T

(a.u

.)

Fig. 2. (a) TGA spectrum and (b) its first derivative for sardine bones.

C. Piccirillo et al. / Ceramics International 40 (2014) 13231–13240 13233

the calcination, the first derivative of the curve was considered

(Fig. 2(b)). A first loss (about 15%) can be observed for

temperatures lower than 200 1C, corresponding to the release

of the water adsorbed on the surface of the bones. A second

greater loss (a further 30%) is seen between 200 and 600 1C,

and from the derivative of the curve it can be seen that this loss

involves two steps, the first at 200 1CoTo390 1C, and the

second one at 390 1CoTo600 1C. According to literature,

these two peaks are associated with the loss of organic

molecules present in the bones (i.e. collagen), and further

water trapped in the porous bones’ structure [32]. A further

loss (410%) can be seen between 800 1C and 1000 1C; this

could be due to the release of carbon dioxide, originated from

the carbonate present in the bones [33].

The total mass loss observed in these experiments is higher

than what was reported for other samples of similar origin;

bovine bones and bones from fish such as Japanese sea bream

or Atlantic cod fish, for instance, all showed a weight loss of

about 40% [33]. The data shown here indicate that sardine

bones have a higher content of organic matter and/or carbonate

ions in the HAp lattice.

Infrared spectra of bone samples B2 and B5 are shown in

Fig. 3. For both samples, signals due to phosphate ions from

both HAp and β-TCP were detected; in both spectra it is

possible to see peaks at 1090, 1034, 984, 962, 631, 603, 565

and 474 cm�1 (HAp phosphate) and at 1122 cm�1 (β-TCP

phosphate) [13]. For the sample calcined at 700 1C (B2, Fig. 3(a))

it is also possible to see a peak due to the presence of carbonate

ions, in the interval between 1410 and 1470 cm�1 [34]. These

peaks are not present in sample B5 (Fig. 3(b)—the intensity of

the peak is comparable to the noise), indicating a complete release

of carbonate ions by 1000 1C. These data are in agreement with

the TGA results shown in Fig. 2.

The other difference between the two spectra is the intensity

of the peak at 3572 cm�1, which belongs to the OH group

present in the HAp lattice. It can be seen that this signal is

clearly present in sample B2 annealed at 700 1C (Fig. 3(a)), but

is much less intense for B5 calcined at 1000 1C (Fig. 3(b)).

This difference can be explained considering the different

relative compositions of the two samples (see Table 1); the

calcination at 1000 1C causes a decrease in HAp and an

increase in β-TCP content. Therefore, it is reasonable that a

peak associated with HAp, but not with β-TCP, is less intense

in sample B5.

Fig. 4 shows some SEM micrographs for samples B2 and

B5. It can be seen that after annealing at 700 1C, material B2

consists of very small regularly shaped grains around 50–

100 nm in diameter and slightly larger elongated grains around

100–200 nm in length (Fig. 4(a)). The elongated grains appear

more abundant, and it is presumed that they are HAp, which

typically forms elongated crystals, while the smaller grains are

β-TCP. This is a relatively fine-grained material, bordering on

the upper limits of size (100 nm) for a nanopowder, consider-

ing it is produced by a simple process from a waste material.

After heating to 1000 1C, sample B5 shows that considerable

increase of grain size has occurred. It consists of an approxi-

mately even mixture of rounded submicron grains (β-TCP) and

larger, elongated, straight-sided grains up to 1 μm in length,

30

40

50

60

70

C

OH

β

Tra

nsm

itta

nce

(%

)

Wavenumber (cm-1)

Wavenumber (cm-1)

3500 1500 1000 500

3500 1500 1000 50035

40

45

50

55

OH

β

Tra

nsm

itta

nce

(%

)

Fig. 3. IR spectra for sardine bones samples (a) B2 (700 1C) and (b) B5

(1000 1C).

Fig. 4. SEM micrographs of samples (a) B2, (b) B5. Note that the magnifica-

tion of (a), at 50,000� , is five-times greater than that of b), at 10,000� .

C. Piccirillo et al. / Ceramics International 40 (2014) 13231–1324013234

which appear to have a smooth cross-section at the ends

perpendicular to the sides (Fig. 4(b)), as has been observed in

HAp crystals previously [17]. These observations support the

ratios of HAP and β-TCP summarised in Table 1.

Overall these results show that sardine bones can be easily

converted into HAp-based compounds, which have potential as

biomaterials, as HAp–TCP biphasic materials are often used in

biomedicine. The relative concentration of each phase and the

sample morphology can be tailored by varying calcination

temperature.

3.2. Annealing and characterisation of the scales

Fig. 5(a) shows the XRD patterns for the sardine scales calcined

at 600 1C (S1), 700 1C (S2), 800 1C (S3) and 1000 1C (S4), and

Table 3 summarises the relative phase compositions calculated

from the QPA of XRD diffraction patterns. In S1 the major

phosphate phase present is HAp but, unlike with the bones, no β-

TCP is present. In S2 and S3 the amount of HAp decreased with

temperature, as observed in the bones, while the β-TCP concentra-

tion is comparable for both samples (around 13.5%). Further to

these compounds, however, peaks corresponding to NaCl and

ClAp can also be detected for the samples calcined between 600–

800 1C (S1, S2 and S3).

The NaCl (100) reflection at 2θ¼31.691 is partially over-

lapped with the most intense HAp reflection at 31.771.

However, the NaCl (220) reflection does not have any

interference/ overlapping at 45.451, and can be clearly

observed in samples S1 and S2. The presence of NaCl in

sardine scales is due to the salting process, which is one of the

steps of the fish canning process [35]. As the calcination

temperature increases from 600 to 800 1C, the NaCl peaks

decrease progressively in intensity while, at the same time,

peaks corresponding to chloroapatite (Ca10(PO4)6Cl2, ClAp)

increase (S1–S3). This indicates a reaction between NaCl and

HAp, with the chloride ion replacing the hydroxyl, and in

sample S3 there is actually higher content (wt%) of ClAp than

HAp (Table 3).

The sample calcined at the highest temperature (1000 1C,

S4), however, does not contain either NaCl or ClAp, indicating

the loss of chloride from the material. In fact sample S4

contains just HAp and β-TCP in proportion comparable to the

sample obtained from the bones calcined at the same tempera-

ture (B5).

To better understand the changes taking place in the

material, the elemental composition of the samples was

monitored as a function of the calcination temperature, and

the results of the analysis are shown in Table 4. For the

chlorine ion, a progressive decrease in the concentration can be

observed with increasing temperature. This indicates that,

although some chlorine reacts with HAp to form ClAp, some

Cl� ions are also increasingly released from the material as an

effect of temperature. The original chlorine concentration in

the scales was much greater than the stoichiometric value

needed to form ClAp, so this loss of chlorine did not affect the

formation of increasing amounts of ClAp up to 800 1C.

However, heating to 1000 1C (S4) led to an almost complete

loss of chlorine, and consequently the sample only consisted of

HAp and β-TCP, with no ClAp anymore. Interestingly, at this

temperature the decomposition of ClAp obviously favours the

crystallisation of β-TCP over HAp, as the ratio of these

two phases is close to being 1:1, as seen in the bones as well.

0

5000

10000

15000

20000

25000

30000 S4

S3

S1

Cl

H

Na

S2

β

Inte

nsity (

a.u

.)

2 θ (degrees)

0

1500

3000

4500

6000

S6

S7

S5

Inte

nsity (

a.u

.)

20 30 40 50 60 70

2 θ (degrees)

20 30 40 50 60 70

2 θ (degrees)

20 30 40 50 60 700

3000

6000

9000

12000

15000

Na

S8

S2

H

β

Inte

nsity (

a.u

.)

Fig. 5. XRD patterns for: (a) Sardine scales calcined at different temperatures.

(b) Sardine scales washed in water and then calcined at different temperatures.

(c) sardine scales calcined at 700 1C (S2) and then washed in water (S8).

H¼HAp, β¼β-TCP, Cl¼ClAp, Na¼NaCl.

Table 3

Phase composition (wt%) of the scales annealed at different temperatures.

Sample Annealing

temperature (1C)

HAp β-TCP ClAp NaCl

S1 600 63.8 0 2.5 33.6

S2 700 45.8 13.5 14.1 26.6

S3 800 36.9 13.6 42.3 7.2

S4 1000 54.2 45.8 0 0

C. Piccirillo et al. / Ceramics International 40 (2014) 13231–13240 13235

The levels of sodium closely mirror those of chlorine in the

material. This suggests that the sodium originating from NaCl

is mainly released and evaporated, and it can replace the

calcium in the HAp lattice only to a very small extent.

Considering the calcium and phosphate values, some

differences with the temperature can also be observed.

Comparing sample S1 with S2, a substantial increase can be

seen in the concentration of both elements. This is reasonable,

since at 600 1C the scales still contain some organic matter and

NaCl— the phosphate component hence represents a smaller

fraction of the whole material. With increasing temperature

both the organic part and the NaCl disappear, with correspond-

ingly higher calcium and phosphorus content. It is interesting

to note that the Ca/P ratio decreases with increasing tempera-

ture, this data being in agreement with the higher β-TCP

content at higher temperatures.

As for the fluorine concentration, a very small increase was

observed for temperatures up to 1000 1C; this can be explained

considering the loss of organic matter and NaCl, as explained

for Ca and P (see above), and the values showed only minimal

differences, which are not statistically significant.

Fig. 6(a) shows the TGA spectrum for the scales, for

temperatures up to 1100 1C (curve i); the total mass loss

observed here was slightly smaller (about 55%) than that

observed for the bones, corresponding to a smaller content of

organic matter. Literature data indicate that the organic content

in fish scales can vary greatly, depending on the kind of fish;

the value observed here, for instance, is smaller than that of

other fish previously studied (Labeo rohita) [36,37].

The first derivative of the curve (Fig. 6(b), curve i) shows

weight losses for To200 1C, 200oTo600 1C (two steps)

and T4800 1C; as already stated for the bones, these

correspond to the loss of adsorbed water, organic matter and

carbon dioxide, respectively. The relative intensities of these

peaks, however, are different. The one corresponding to the

release of water, for instance, is more intense; this indicates a

higher quantity of adsorbed water, probably due to a more

hydrophilic surface. Those related to the loss of organic matter,

on the other hand, are smaller. In addition to these peaks,

however, another one can be observed between 700oTo800 1

C, which was not detected when the bones were calcined; this

can be due to the weight loss associated with the release

of NaCl.

Fig. 7(a) and (b) shows the IR spectra for samples S2 and

S4. S2 shows the same features already observed in the FT-IR

spectra of the bones (Fig. 3(a)). The signals corresponding to

the carbonate ion (1410–1470 cm-1), however, are less intense,

indicating a lower carbonate content in the sample. This is in

agreement with the TGA data, as the peak corresponding to the

carbonate ion loss is less intense for the scales than for the

bones. As for the S4 spectrum, the same peaks as in sample S2

are detected, but differences in the intensities of the OH and

β-TCP peaks can be seen, as they are smaller and larger,

respectively. Moreover, the carbonate ion related signals are at

noise level, as already observed for the bones.

SEM micrographs for samples S2 and S4 are shown in

Fig. 8(a) and (b) respectively, and are very different to those

seen for the bones. S2, heated to 700 1C, consisted of large,

angular very crystalline grains over 1 μm in dimension, which

seem to have long parallel sides, and in many cases a clearly

hexagonal cross-section, suggesting they are HAp (Fig. 8(a)).

Table 4

Elemental composition (wt%) for scales calcined at different temperatures.

Sample Ca P Na Cl F Ca/P (mol/mol)

S1 29.0770.30a 11.1670.15a 14.5270.11ª 14.7870.12a 0.02570.002ª 2.0270.08ª

S2 35.6070.41b 14.1970.19b 9.9770.09b 10.1170.09b 0.02870.002b 1.9470.07ª

S3 36.0570.44 b 21.8970.21c 6.1170.09c 5.9670.03c 0.02970.001b 1.6570.07b

S4 33.1170.39c 22.5370.20d 0.9970.02d 0.5870.02d 0.03170.002b 1.4770.05c

Note: different letters in the same column indicate that data are statistically different.

-50

-40

-30

-20

-10

0

(i)

(ii)

We

igh

t lo

ss (

%)

Temperature (°C)

Temperature (°C)

0 200 400 600 800 1000

0 200 400 600 800 1000-0.4

-0.3

-0.2

-0.1

0.0

(ii)

(i)

∆m

/∆T

(a.u

.)

Fig. 6. (a) TGA spectrum and (b) its first derivative for sardine scales (i) as

received, not washed in water, (ii) washed in water. (For interpretation of the

references to colour in this figure, the reader is referred to the web version of

this article.)

C. Piccirillo et al. / Ceramics International 40 (2014) 13231–1324013236

These are covered with much smaller equiaxed crystals, up to

100 nm in diameter, which resemble the β-TCP nanocrystals

observed in the SEM images of B1. The elongated hexagonal

HAp grains observed here are even larger, and appear much

more crystalline, than the HAp grains formed at 1000 1C from

the bones (B5). This evidence could be due to the influence of

ClAp, which resulted in larger micron scale grains when pure

ClAp was made from cod bones [17], or the action of sodium

migrating to grain boundaries as it is lost, encouraging

sintering and grain growth. When heated to 1000 1C, sample

S4 consisted of very large fused grains tens of microns

in length, which showed evidence of a liquid phase in

sintering (probably from the sodium as it was lost) and a

macroscale porosity between grains, with a covering of

smaller, polyhedral grains of β-TCP up to 1 μm in diameter on

their surface (Fig. 8(b)).

To remove sodium chloride, some scales were washed in

water, as described in the experimental section. After washing

and drying, a mass loss of about 40% was observed, due to the

washing process. This clearly indicates the solubilisation of

some compounds, such as sodium chloride, present in the

scales. The experiments showed that all mass loss was

observed during the first 24 h of soaking; no further change

was observed on the second day of the treatment.

Fig. 5(b) shows the XRD patterns for the scales washed

prior to the calcination and subsequently annealed at 600, 700

and 900 1C (S5, S6 and S7, respectively). It can be seen that,

in this case, the only compounds present in the pattern for the

sample calcined at 600 1C (S5) are HAp and β-TCP—no peak

corresponding to NaCl can be detected. This indicates that the

salt was removed in the washing process, with its concentra-

tion being below the XRD detection limit. The more accurate

QPA analysis also indicated there was no crystalline NaCl

present in any of the washed samples (Table 5).

40

50

60

70

C

OH βTra

nsm

itta

nce (

%)

Wavenumber (cm-1)

40

50

60

OH

β

Tra

nsm

itta

nce (

%)

3500 1500 1000 500

Wavenumber (cm-1)

Wavenumber (cm-1)

3500 1500 1000 500

4000 3500 1500 1000 50040

50

60

70

β

OH

Tra

nsm

itta

nce

(%

)

Fig. 7. IR spectra for sardine scales: (a) sample S2, (b) sample S4, and

(c) sample S6.

Fig. 8. SEM micrographs for (a) sample S2, (b) sample S4, and (c) sample S6.

C. Piccirillo et al. / Ceramics International 40 (2014) 13231–13240 13237

After annealing at 700–900 1C, a further conversion of HAp

into β-TCP was observed, with approximately 40 wt% β-TCP

present in samples S6 and S7 (Table 5); this could be due to

the Ca/P ratio being much smaller than the stoichiometric

value (1.19 vs. 1.67, see Table 6). Such a low value indicates

that, during soaking in water, the amounts of calcium and

phosphate ions dissolved were not in the stoichiometric ratio,

and that disproportionately more Ca2þ was lost to the solution.

This resulted in formation of a much greater amount of the

β-TCP phase at only 700 1C. Literature data show that the

dissolution of the ions in HAp rarely follows HAp stoichio-

metry; on the contrary, this can be affected by several

variables, which include the HAp source and/or preparation

method, the pH of the solution and the possible presence of

other ions, both in the solution and in the HAp lattice [38].

These results, therefore, confirm what has been previously

reported.

Furthermore, although not obvious in the XRD patterns,

QPA revealed that some ClAp has also been formed, despite

the absence of crystalline NaCl from the scales after washing

(Table 5). This was attributed to the presence of some residual

chlorine ions in the sample, in a non-crystalline form, probably

adsorbed on the surface or trapped in pores. As shown in

Table 6, elemental analysis of sample S7 detected a small

amount of chlorine atoms to be present (o0.5 wt%) even after

washing. This was enough to substitute some hydroxyl in HAp

and produce up to 9 wt% ClAp in the washed scales. As

summarised in Table 5 it can also be seen that the sodium

concentration is much lower than for the unwashed scales

(0.25 wt%, vs. 0.99 wt% in sample S4), while for fluorine

there is no significant difference.

The peaks of the XRD patterns of the washed scales are

much less sharp, indicating a lower level of crystallinity of the

sample. This evidence supports the observations made above

that sodium ions may act as a liquid phase on grain boundaries,

encouraging crystallisation and grain growth in the HAp. The

SEM image of sample S6 (Fig. 8(c)) shows no sign of the

liquid phase sintering and massive grain growth observed in

sample S4, and it also appears to be denser, without the large

macropores measuring several microns that have formed

between the massive grains in a rapid, run-away grain growth

process, which can typically result in a loss of porosity as such

macropores are formed. The use of NaCl as a porogen agent, to

create macroporous ceramics, has in fact already been reported

for several materials, including calcium phosphate-based

materials for biomedical use [39]. Without the presence of

sodium, the sample appears to be much denser, with small

submicron pores, and with no evidence of run-away grain

growth having occurred. The relative lack of ClAp may have

also limited the grain growth, which has occurred in this

sample, compared to S4.

The TGA spectrum for the scales washed in water (Fig. 6(a),

curve ii) shows a smaller mass loss (less than 40%) in

comparison to the non-washed scales. As well as the loss of

NaCl, a large difference can be observed between 200 and

600 1C, which corresponds to the loss of organic matter. This

may be due to a partial solubilisation of organic components in

the washing process as well, together with the sodium chloride.

Indeed, the colour change observed in the scales after the

washing (from yellow/pale-brown to white/transparent) con-

firms the solubilisation of some organic molecules. From the

first derivative of the TGA pattern (Fig. 6(b), curve ii), it can

be seen that the two mass losses observed in the unwashed

scales for T4600 1C (carbonate and sodium chloride releases)

are not present here. As most of the sodium chloride was

dissolved in the washing water; it is reasonable not to have a

peak associated with it. It is likely that the carbonate ions were

also dissolved due to their polarity and hence water solubility,

which also partially accounted for the large difference in

weight loss above 600 1C.

The presence of very low carbonate content in the washed

scales was also confirmed by the IR measurements (Fig. 7(c));

in fact the spectra of sample S6 shows almost no carbonate ion

peak (1410–1470 cm�1 interval, signal at the same level of

noise).

A further washing experiment was performed with the

scales; in this case the scales were first calcined at 700 1C

and then washed in water. The temperature value was chosen

to be the same as that of sample S2, which contained HAp and

β-TCP in proportions of 3:1 weight (within experimental

error); the combination of these two materials with this ratio

is often used in commercial biphasic biomaterials [1].

Table 5

Phase composition (wt%) of the scales washed in water and then annealed at different temperatures.

Sample Annealing temperature (1C) HAp β-TCP ClAp NaCl

S5 600 88.8 11.2 0 0

S6 700 50.5 40.2 9.3 0

S7 900 54.4 37.6 7.9 0

Table 6

Elemental composition (wt%) scale sample S7 calcined at 900 1C.

Element Ca P Na Cl F Ca/P (mol/mol)

Concentration (wt%) 37.9370.39 24.7570.28 0.2570.03 0.4770.03 0.03070.002 1.1970.05

C. Piccirillo et al. / Ceramics International 40 (2014) 13231–1324013238

Moreover, ClAp was also present in this sample at a

concentration value (within experimental error) which, accord-

ing to literature, corresponds to an improvement of the

mechanical properties of the material [10]. The combination

of these three phases could, therefore, lead to a biomaterial

with better performance.

After calcination at 700 1C, a weight loss of about 30% was

observed in sample S8 with the washing process. In the XRD

patterns for this sample, the phase and the elemental composi-

tions can be seen in Fig. 5(c), and are summarised in Tables 7

and 8, with sample S2 included in the tables for comparison.

XRD patterns show that no sodium chloride was present,

meaning that the salt, present as a separate phase, was

dissolved in the water even after heating at 700 1C. QPA

proved that there was no crystalline NaCl present in sample S8

(Table 7). Comparing samples S2 and S8 (without and with

washing, respectively) in Table 7, it can be seen that the

relative composition of each phase does not change signifi-

cantly, considering the experimental error associated with

these fittings. Elemental composition (Table 8) shows a

remarkable decrease in sodium and chlorine concentrations

for sample S8, as expected. Moreover, although it had a lower

Ca/P ratio than the unwashed sample S2, sample S8 exhibited

a smaller decrease in the Ca/P ratio than that observed in

sample S7 (1.46 vs. 1.19). This demonstrates that, due to

washing the sample after annealing at 700 1C, less calcium

ions were lost to the solution, while virtually all NaCl was

removed. Washing the annealed powder does not affect its

crystallinity, as the XRD pattern of sample S8 has sharp peaks,

similar to the unwashed sample S2.

To summarise this study of sardine scales, it can be seen that

they can be converted into a HAp-based material, which has

potential as a biomaterial; this was achieved with a simple

annealing process at a temperature high enough to eliminate

the chloride, which is present in the sardine scales from the

salting process.

On the other hand, ClAp-containing materials can also be

obtained from the scales, with a combined process of

annealing and washing. The best results were obtained with

annealing at a relatively low temperature (700 1C), followed by

washing of the powder. In this way, the material was obtained

with high level of crystallinity constituted mainly of HAp and

β-TCP, with some ClAp also present. According to literature

data, such multiphasic material has great potential as a

biomaterial, due to its high biocompatibility and bioactivity,

and improved mechanical properties. To our knowledge, this is

the first time ClAp was extracted from a marine source.

4. Conclusions

Sardine bones and scales were successfully used to extract

apatite- and calcium-phosphate-based materials; it is the first

time that two parts of the same fish are used to extract

compounds presenting high potential as biomaterial.

Results show that bones can be used to obtain materials in

which HAp is the main component, through a calcination

process. With scales, on the other hand, HAp-based material or

ClAp-containing ones can be produced using a simple calcina-

tion and a combined washing–calcination process respectively.

This is the first time ClAp of marine origin is produced.

This study shows how it is possible to valorise byproducts

of the food industries, obtaining high added value products

which can be used in biomedicine and other potential

applications.

Acknowledgements

This work was funded through the ValorPeixe Project

(QREN, AdI 13634). Authors also acknowledge PEst-C/

CTM/LA0011/2013 and PEst-OE/EQB/LA0016/2013 pro-

grammes. C. Piccirillo thanks the FCT for Research Grant

(SFRH/BPD/86483/2012), while R.C. Pullar wishes to thank

the FCT Ciência2008 programme for supporting his work.

References

[1] S.V. Dorozhkin, Bioceramics of calcium orthophosphates, Biomaterials

31 (2010) 1465–1485.

[2] S.V. Dorozhkin, Calcium orthophosphates in dentistry, J. Mater. Sci.:

Mater. Med. 24 (2013) 1335–1363.

[3] J. Peña, M. Vallet-Regí, Hydroxyapatite, tricalcium phosphate and

biphasic materials prepared by a liquid mix technique, J. Eur. Ceram.

Soc. 23 (2003) 1687–1693.

[4] S. Kannan, J.M.F. Ferreira, Synthesis and thermal stability of

hydroxyapatite-β-tricalcium phosphate composites with co-substituted

sodium, magnesium and fluorine, Chem. Mater. 18 (2006) 198–203.

[5] A. Nzihou, P. Sharrock, Role of phosphate in the remediation and reuse

of heavy metal polluted wastes and sites, Waste Biom. Valoriz. 1 (2010)

163–174.

[6] C. Piccirillo, S.A. Pereira, A.P. Marques, R.C. Pullar, D.M. Tobaldi,

M.M. Pintado, P.M.L. Castro, Bacteria immobilization on hydroxyapatite

surface for heavy metals removal, J. Environ. Manag. 121 (2013) 87–95.

[7] H. Tanaka, E. Tsuda, H. Nishikawa, M. Fuji, FTIR studies of adsorption

and photocatalytic decomposition under UV irradiation of dimethyl

sulfide on calcium hydroxyapatite, Adv. Powder Technol. 23 (2012)

115–119.

[8] C. Piccirillo, C.W. Dunnill, R.C. Pullar, D.M. Tobaldi, I.P. Parkin,

J.A. Labrincha, M.M. Pintado, P.M.L. Castro, Calcium phosphate-based

Table 7

Phase composition (wt%) of the scales calcined at 700 1C and then washed

in water.

Washing time (h) HAp β-TCP ClAp NaCl

S2 0 45.8 13.5 14.1 26.6

S8 24 67.6 15.6 16.7 0

Table 8

Elemental composition (wt%) of scales calcined at 700 1C and then washed in

distilled water.

Sample Washing

time (h)

Ca P Na Cl F Ca/P

(mol/mol)

S2 0 35.60 14.19 9.97 10.11 0.028 1.94

S8 24 36.10 19.21 1.30 1.14 0.031 1.46

C. Piccirillo et al. / Ceramics International 40 (2014) 13231–13240 13239

materials of natural origin showing photocatalytic activity, J. Mater.

Chem. 1 (2013) 6452–6461.

[9] W.N. Wang, Y. Kaihatsu, F. Iskandar, K. Okuyaman, Highly luminous

hollow chloroapatite phosphors formed by a template-free aerosol route

for solid-state lighting, Chem. Mater. 21 (2009) 4685–4691.

[10] S. Kannan, A. Rebelo, A.F. Lemos, A. Barba, J.M.F. Ferreira, Synthesis

and mechanical behaviour of chloroapatite and chloroapatite/β-TCP

composites, J. Eur. Ceram. Soc. 27 (2007) 2287–2294.

[11] J.S. Cho, D.S. Yoo, Y.C. Chung, S.H. Rhee, Enhanced bioactivity and

osteoconductivity of hydroxyapatite through chloride substitution, J.

Biomed. Mater. Res. A 102 (2014) 455–469.

[12] A.M. Janus, M. Faryna, K. Haberko, A. Rakowska, T. Panz, Chemical

and microstructural characterization of natural hydroxyapatite derived

from pig bones, Microchim. Acta 161 (2008) 349–353.

[13] N.A.M. Barakat, K.A. Khalil, F.A. Sheikh, A.M. Omran, B. Gaihre,

S.M. Khil, H.Y. Kim, Physiochemical characterization of hydroxyapatite

extracted from bovine bones by three different methods: extraction of

biologically desirable HAp, Mater. Sci. Eng. C 38 (2008) 1381–1387.

[14] M. Boutinguiza, J. Pou, F. Comesaña, A. Lusquiños, B. de Carlos,

Biological hydroxyapatite obtained from fish bones, Mater. Sci. Eng. C

32 (2012) 478–486.

[15] M. Boutinguiza, A. Lusquiños, R. Riveiro, R. Comesaña, J. Pou,

Hydroxyapatite nanoparticles obtained by fiber laser-induced fracture,

Appl. Surf. Sci. 255 (2009) 5382–5385.

[16] J. Venkatesan, Z.J. Qian, B. Ryu, N.V. Thomas, S.K. Kim, A

comparative study of thermal calcination and an alkaline hydrolysis

method in the isolation of hydroxyapatite from Thunnus obesus bone,

Biomed. Mater. 6 (2011) 035003–035012.

[17] C. Piccirillo, M.F. Silva, R.C. Pullar, I. Braga da Cruz, R. Jorge,

M.E. Pintado, P.M.L. Castro, Extraction and characterization of

apatites-and calcium phosphate-based materials from cod fish bones,

Mater. Sci. Eng. C 33 (2013) 103–110.

[18] S. Kongsri, K. Janpradit, K. Buapa, S. Techawongstein, S. Chanthai,

Nanocrystalline hydroxyapatite from fish scale waste: preparation,

characterization and application for selenium adsorption in aqueous

solution, Chem. Eng. J. 215 (2013) 522–532.

[19] S. Mondal, S. Mahata, S. Kundu, B. Mondal, Processing of natural

resourced hydroxyapatite ceramics from fish scale, Adv. Appl. Ceram.

109 (2010) 234–239.

[20] Y.C. Huang, P.C. Hsiao, H.J Chai, Hydroxyapatite extracted from fish

scale: effects on MG63 osteoblast-like cells, Ceram. Int. 37 (2011)

1825–1831.

[21] N.N. Panda, K. Pramanik, L.B. Sukla, Extraction and characterization of

biocompatible hydroxyapatite from fresh water fish scales for tissue

engineering scaffold. Bioprocess Biosyst. Eng., 37 (2014) 433–440.

[22] K. Zhu, X. Gong, D. He, B. Li, D. Ji, P. Ki, Z. Peng, Y. Luo, Adsorption

of Ponceau 4R solutions using alkali boiled Tilapia fish scales, RCS Adv.

3 (2013) 25221–25230.

[23] R.A. Young, in: The Rietveld Method – International Union of

Crystallography Monographs on Crystal, Oxford University Press, UK,

1995, p. 312.

[24] A.C. Larson, R.B. Von Dreele, General structure analysis system

(GSAS), Los Alamos National Laboratory, Los Alamos, US, 2004 (Los

Alamos National Laboratory Report LAUR 86-748).

[25] B.H. Toby, EXPGUI: a graphical user interface for GSAS, J. Appl.

Crystallogr. 34 (2001) 210–213.

[26] C. Piccirillo, S.I.A. Pereira, A.P.G.C. Marques, R.C. Pullar, D.

M. Tobaldi, M.E. Pintado, P.M.L. Castro, Bacteria immobilisation on

hydroxyapatite surface for heavy metal removal, J. Env. Manag. 121

(2013) 87–95.

[27] D. Walker, P.K. Verma, L.M.D. Cranswick, R.L. Jones, S.M. Clark,

S. Buhre, Halite-sylvite thermoelasticity, Am. Mineral. 89 (2004)

204–210.

[28] W.A. Dollase, Correction of intensities for preferred orientation in

powder diffractometry: application of the March model, J. Appl. Cryst.

19 (1986) 267–272.

[29] H. Zhang, M. Zhang, Characterization and thermal behavior of calcium

deficient hydroxyapatite whiskers with various Ca/P ratios, Mater. Chem.

Phys. 126 (2011) 642–648.

[30] M. Ozawa, M. Suzuki, Microstructural development of natural hydro-

xyapatite originated from fish bone waste through heat treatment, J. Am.

Ceram. Soc. 85 (2002) 1315–1317.

[31] A. Bigi, E. Foresti, R. Gregoriani, A. Ripamonti, N. Roveri, J.S. Shah,

The role of magnesium in the structure of biological apatites, Calcif. Tiss.

Int. 50 (1992) 439–444.

[32] J.J. Lim, Thermogravimetric analysis of human femur bone, J. Biol. Phys.

3 (1975) 111–128.

[33] M. Figueiredo, A. Fernando, G. Martins, J. Fretia, F. Judas,

H. Figueiredo, Effect of the calcination temperature on the composition

and microstructure of hydroxyapatite derived from human and animal

bone, Ceram. Int. 36 (2010) 3283–3293.

[34] K. Haberko, M.M. Bucko, J. Brzezinska-Miecznik, M. Haberko,

W. Mozgawa, T. Panz, A. Pyda, J. Zarebski, Natural hydroxyapatite –

its behaviour during heat treatment, J. Eur. Ceram. Soc. 26 (2006)

537–542.

[35] V. Ferraro, A.P. Carvalho, C. Piccirillo, M.M. Santos, P.M.L. Castro,

M.E. Pintado, Extraction of high added value biological compounds from

sardines, sardine-type fish and mackerel canning residues, Mater. Sci.

Eng. C 33 (2013) 3111–3120.

[36] R. Chakraborty, S. Bepari, A. Baneriee, Application of calcined waste

fish (Labeo rohita) scale as low-cost heterogenous catalyst for biodiesel

synthesis, Bioresour. Technol. 102 (2011) 3610–3618.

[37] S. Mondal, R. Bardhan, B. Mondal, A. Dey, S.S. Mukhopadhayay,

S. Roy, R. Guha, K. Roy, Synthesis, characterization and in vitro

cytotoxicity assessment of hydroxyapatite from different bioresources

for tissue engineering application, Bull. Mater. Sci. 35 (2012) 683–691.

[38] J. Oliva, J. Cama, J.L. Cortina, C. Ayora, J. De Pablo, Biogenic

hydroxyapatite (Apatite IITM) dissolution kinetics and metal removal

from acid mine drainage, J. Hazard. Mater. 213–214 (2012) 7–18.

[39] A.C. Tas, Preparation of porous apatite granules from calcium phosphate

cement, J. Mater. Sci.: Mater. Med. 19 (2008) 2231–2238.

C. Piccirillo et al. / Ceramics International 40 (2014) 13231–1324013240