The pseudomorphic replacement of marble by apatite:
The role of fluid composition
Elisabete Trindade Pedrosa, Christine V. Putnis, Andrew Putnis
To cite this version:
Elisabete Trindade Pedrosa, Christine V. Putnis, Andrew Putnis. The pseudomorphic replacement of marble by apatite: The role of fluid composition. Chemical Geology, Elsevier, 2016,
425, pp.1-11. .
HAL Id: hal-01299331
https://hal.archives-ouvertes.fr/hal-01299331
Submitted on 7 Apr 2016
HAL is a multi-disciplinary open access
archive for the deposit and dissemination of scientific research documents, whether they are published or not. The documents may come from
teaching and research institutions in France or
abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est
destinée au dépôt et à la diffusion de documents
scientifiques de niveau recherche, publiés ou non,
émanant des établissements d’enseignement et de
recherche français ou étrangers, des laboratoires
publics ou privés.
1
The pseudomorphic replacement of marble by apatite: the role of fluid composition
2
Elisabete Trindade Pedrosa1, Christine V. Putnis1,2 and Andrew Putnis1,3
3
1
Institut für Mineralogie, University of Münster, Corrensstrasse 24, 48149 Münster, Germany
4
2
Nanochemistry Research Institute, Department of Chemistry, Curtin University, Perth, 6845
5
Australia
6
3
7
Abstract
8
The replacement of a natural carbonate rock (Carrara marble) by apatite was used as a model to
9
study the role of fluid chemistry in replacement reactions, focusing on the mineralogy, chemical
10
composition, and porosity of the replacementproduct.Carrara marble was reacted with
11
ammonium phosphate solutions ((NH4)2HPO4), in the presence and absence of four salt solutions
12
(NH4Cl, NaCl, NH4F, and NaF) at different ionic strengths, at 200°C and autogenous pressure.
13
The replacement products were analysed usingpowder X-ray diffraction,Scanning electron
14
microscopy (SEM), electron microprobe analysis(EMPA), and Raman spectroscopy. The reaction
15
in all samples resulted in pseudomorphic replacements and shared the characteristics of an
16
interface-coupled dissolution-precipitation mechanism. Increasing theionic strength of the
17
phosphatefluid increased the replacement rates. With a fixedconcentration of phosphate,
18
replacement rates were reduced with the addition of NH4Cl and NaCland increased significantly
19
with the addition of NaF and NH4F. The addition of different salts resulted in specific porosity
20
structures resulting from the formation of different phosphate phases. Chlorine-containing fluids
21
showed a higher degree of fluid percolation through grain boundaries. This study illustrates the
22
significant impact that small differences in solvent composition can have in the progress of
23
replacement reactions, the nature of the products and the resultant porosity.
24
Keywords
25
Calcite, apatite, fluoride, replacement, pseudomorphism, dissolution-precipitation, porosity
26
1. Introduction
27
Fluid mediated re-equilibration of solids is of great importance in many geological systems (e.g.
28
Putnis, 2002, 2009 and references therein), and industrial processes (e.g. Xia et al., 2008, 2009a).
The Institute for Geoscience Research (TIGeR), Curtin University, Perth, 6845 Australia
1
29
Of particular interest are mineral replacement reactions that result in the formation of
30
pseudomorphs of the parent phase (preservation of the external shape and total volume). In the
31
presence of fluids, the mechanism of pseudomorphic replacement has been explained as the re-
32
equilibration process between fluid and minerals involving an interface-coupled dissolution-
33
precipitation mechanism occurring at the mineral-fluid interface (Putnis and Putnis, 2007; Ruiz-
34
Agudo et al., 2014). In this mechanism the solid-fluid interface is of most importance because as
35
soon as a small amount of solid material dissolves, it generates a boundary layer of fluid that may
36
become supersaturated with a new solid phase that immediately precipitates (Ruiz-Agudo et al.,
37
2012). These reactions are characterized by sharp reaction fronts between parent and product
38
phases, and the development of intracrystalline porosity in the product phase (Putnis and Mezger,
39
2004; Putnis et al, 2005).
40
The formation of porosity is critical because it allows continuous percolation of fluid to the
41
reaction front and ultimately the complete replacement of one phase by another. Porosity
42
generation also has a significant impact on the mechanical behaviour of rocks, an important issue
43
when considering injecting fluids into the upper crust. (Nermoen et al., 2015). Moreover, the
44
relevance of these replacement reactions has been reported in many varying scenarios, such as:
45
CO2 sequestration (Lackner et al., 2002); the durability of nuclear waste disposal (Geisler et al,
46
2015); synthesis of new materials (Ben-Nissan et al., 2003; Xia et al., 2008, 2009a); remediation
47
of contaminated soil (Wang et al., 2012) and groundwater (Turner et al., 2005); new wastewater
48
treatments (Yang et al., 1999); and the preservation of monument surfaces (Sassoni et al., 2011).
49
In this work, we investigate in detail the role of the bulk fluid composition on fluid-mineral
50
reactions using as a model system, the reaction of a calcium carbonate rock (Carrara marble) with
51
various phosphate-containing fluids. Carrara marble is an ideal experimental material because it
52
is a relatively pure calcium carbonate (calcite) rock with regular grain size (average 200 μm), and
53
the pseudomorphic replacement of calcite by apatite has been previously demonstrated (e.g.
54
Ames, 1959, 1961; Kasioptas et al., 2008, 2010, 2011; Jonas et al., 2013, 2014; Borg et al.,
55
2014).
56
Of particular interest is to understand the role of the composition of fluids in controlling the
57
progress of reactive fluid flow. In this work, two main questions were posed:
2
58
(i)
59
60
How does fluid composition influence the evolution of microstructure and porosity
generation of the product phase(s)?
(ii)
61
How does fluid chemistry influence the chemical composition of the product phase(s),
and overall mass transfer?
62
These questions are relevant to natural Earth environments whenever aqueous fluids are in
63
contact with crustal rocks. When rocks are out of equilibrium with the surrounding fluids
64
reactions continuously take place in order to reach a more stable thermodynamic state. The
65
volume of fluid can be quite small for mineral replacement reactions to occur (Milke et al., 2012)
66
and the results of the re-equilibration processes are the formation of new minerals (reaction
67
product phases) as well as a fluid charged with new elements released from the mineral-fluid
68
reaction. In this way elements may be sequestered or released to solution and carried away in the
69
fluid phase (Xia et al., 2008).
70
2. Materials and Methods
71
2.1 Starting material
72
2.1.1 Solids
73
Carrara marble (from Carrara, Italy) was chosen as the parent material due to its high purity of
74
calcium carbonate (CaCO3> 99.7 %), confirmed from X-ray diffraction (XRD) and Raman
75
spectroscopy analyses, and its facility to be replaced by calcium phosphate when reacted with
76
phosphate-bearing solutions (Kasioptas et al., 2008, 2011). The rock is polycrystalline, composed
77
of equant grains of approximately 200 μm in size, and contains 0.3(1)wt.% of
78
Mg,homogeneously distributed. Two sets of small marble cubes were cut from the white parent
79
rock: small cubes of around3mm3, and larger cubes of around 15 mm3. The initial weight of the
80
samples was on average 8(1) mg and 40(1) mg,respectively. The individual cubes were cut and
81
then polished to the desired sizes (1.5 × 1.5 × 1.5 mm for the small cubes and 3 × 3 × 3 mm for
82
the larger cubes) using a digital caliper.
83
2.1.2 Fluids
84
Pure solid ammonium phosphate (NH4)2HPO4(Acrōs Organics powder >99.0 %) was used to
85
make a 2M stock solution from which all other solutions (referredto as phosphate solution) were
86
diluted using deionised water MillQ (resistivity> 18 MΩ.cm).Five sets of experiments were
87
carried
out.
Set
1
were
experiments
with
various
concentrations
of
ammonium
3
88
phosphatesolutions (0.1 0.5, 1, 2 M). Set 2 to 5 used 1 M ammonium phosphate to which varying
89
concentrations of other salts were added. The preparation of these stock solutions was made by
90
dissolving ammonium chloride (T.H. Geyer GmbH 99.8 %), sodium chloride (E. Merck 99.5 %),
91
ammonium fluoride (Alfa Aesar GmbH 98.0 %), and sodium fluoride (E. Merck 99.0 %)
92
powders with MillQ water. The molar concentration of each solution used in the experiments is
93
shown in Table 1.
94
2.2 Hydrothermal experiments
95
For the hydrothermal replacement experimentseach marble cube was placed into a Teflon®-lined
96
reactor together with 2 mL of the corresponding fluid. The Teflon® reactor was placed into a steel
97
autoclave and tightly sealed to avoid any fluid loss during reaction. The temperature for all
98
experiments was 200°C, at autogenous pressure.After reaction, the autoclaves were removed
99
from the furnace and quickly cooled to room temperature (22°C) in a stream of compressed air.
100
The samples were washed with distilled water,left to dry overnight at 40°C, and then weighed.
101
The pH of the fluids was also measured before and after reaction. Magnesium and calcium
102
contents in the final fluids were measured using inductively-coupled plasma optical emission
103
spectrometry (ICP-OES).
104
The smaller cube samples were reacted for 4 days, in which the only parameter changed was the
105
composition of the fluid.One small Carrara marble cube was used for each experiment, and five
106
sets of experiments were made. Each set was composed of four experiments (see Table 1).
107
Because these experiments used natural samples, it is possible that each one differed slightly in
108
terms of grain boundary distribution. To estimate the potential variation that this difference could
109
have on the results, at least two replicasof each experiment were made.
110
To be able to better identify the product phases (for XRD analysis), the larger marble cubes were
111
reacted with the five different types of fluids for 8 days.
112
2.3 Analytical methods
113
2.3.1 PHREEQC simulations
114
The computer program PHREEQC (Parkhurst and Appelo, 1999) was used to determine the ionic
115
strength (IS) of each of the bulk fluids (without the marble) used in the experiments at 200°C
116
(Table 1) and to define the solution speciation and the saturation states of the solutions relative to
117
all potential phases involved in the reactions at theinitial stage when the first layer of marble
4
118
dissolves into the fluid. As the supersaturation of the fluid is confined to the dissolving mineral
119
surface for the replacement reaction to take place (Pollok et al., 2011; Ruiz-Agudo et al, 2015)
120
calculations of supersaturation were made for a thin a layer of fluid at the mineral-fluid interface
121
(boundary layer) from which the product phase(s) precipitates. For the purposes of estimating
122
supersaturation at the interface, a 0.3 nm thick monolayer of the marble cube (1.5×1.5×1.5 mm)
123
was considered to be dissolving into a thin boundary layer that contained each of the initial fluids
124
(Table 1) used in the experimentsat 200°C. The concentration of dissolved carbonate species
125
iscalculated by the program after adding the initial conditions. The choice of width of the
126
boundary layer is somewhat arbitrary, but serves to indicate the evolution of supersaturation at
127
the mineral-fluid interface with respect to likely precipitating phases. Two boundary layer
128
thicknesses were used to check for consistency of results, 0.1 and 1 µm (fluid volumes of 1.35 ×
129
10-9 and 1.35 × 10-8cm3, respectively).The solution boundary layer also included magnesium in
130
the equivalent amount present in the marble (0.3wt.%).
131
Saturation state is given as SI (saturation index) where SI = log Ω = log (IAP/Ksp), where Ω is the
132
supersaturation, IAP is the ion activity product and Ksp is the solubility product. If SI = 0, the
133
mineral and solution are in equilibrium, if SI < 0 the solution is undersaturated and if SI > 0 the
134
solution is supersaturated. Mineral phases that appear with SI > 1 will theoretically be
135
supersaturated and thus possibly precipitate.The simulations are presented in tables S1 and S2 of
136
the supplementary material. From these simulations we find that the solutions are most highly
137
supersaturated with respect tohydroxyapatite (Ca5(PO4)3OH) in sets 1, 2 and 3, and fluorapatite in
138
experiments 4 and 5. Tricalcium phosphate (Ca3(PO4)2) is second most supersaturated in set 1
139
and third in all other sets. Chlorapatite (Ca5(PO4)3Cl) is second in sets 2 and 3, and
140
hydroxyapatite in sets 4 and 5.
141
2.3.2 X-ray Diffraction
142
To identify the mineral product phases using powder X-ray diffraction (XRD), the larger reacted
143
samples were dried and powdered in an agate mortar. An X’Pert PW 3040 PANalytical
144
diffractometer and the X’Pert Data Collector software were used to collect the diffraction data in
145
the range 5–75° 2 Theta using CuKα1 radiation. The step size was 0.014°2Theta. The primary
146
monochromator used was a Johannson monochromator with a Ge crystal cut on plane (111). The
147
measurement time for each sample was 66 min. An initial analysis of the results was made with
148
the X’Pert High Score and X’Pert Data Viewer.Following the protocol from Jonas et al., (2014),
5
149
a beta tricalcium phosphate (TCP) powder (Sigma-Aldrich ≥ 98%) was measured for comparison
150
with the results.
151
2.3.3 Scanning electron microscopy (SEM)
152
For qualitative chemical analysis and visualization of the replacement textures and porosity
153
microstructures, the smaller reacted cubes were mounted in epoxy resin, sectioned through the
154
centre, polished and C-coatedready forimaging in a SEM(JEOL JSM-6610LV) equipped with
155
energy-dispersive X-ray analysis(EDX) and an electron back-scattered detector.
156
2.3.4 Electron microprobe analysis (EMPA)
157
For quantitativechemicalanalysis and element mapping, small polished cross-sections of the
158
sampleswereanalysed using an electron microprobe (JEOL JXA 8900 Superprobe), equipped
159
with a wavelength dispersive X-ray spectrometer (WDS) and an energy dispersive X-ray
160
spectrometer (EDS).The point analysis was made using 15 kV accelerating voltage, an electron
161
beam current of 4 nA, and a probe diameter of 3 µm.A minimum of 20 points were analysed in
162
each sample. The mapping was made using15kV, a beam current of 50 nA, and 0.5 µm
163
diameter.The major components analysed were those present in the solution that could potentially
164
be incorporated in the replacement product (i.e. CaO, P2O5, NaO, MgO, Cl, and F). Molar
165
proportions of the elements analysed were calculated on the basis of 13 oxygens for apatite
166
(Ca5(PO4)3(OH,F,Cl)). The amount of water in apatite (as OH) was calculated assuming ideal
167
mixing of Cl, F, and OH on the anion position.
168
2.3.5 Raman spectroscopy
169
For better characterization of the crystalphases involved, micro-Raman spectroscopic analysiswas
170
performed on one of each of the samples of the five experiment sets,with a high-resolution
171
Horiba Jobin YVON Xplora Raman system equipped with a confocal microscope (Olympus
172
BX51),using anear infra-red 785 nm laser as excitation source. For searching the OH vibrational
173
band,a higher energy laser was necessary (green, 532 nm). The spectrum of the TCP powder used
174
in XRD analysis was also measured for comparison.The Raman light was collected in 180°
175
backscattering geometry and dispersed by a grating of 1200 grooves/mm to allow the recording
176
of the frequency ranges between 100 and 1200 cm-1 (to include the main phosphate vibrational
177
modes), and between at 3000 and 3600 cm−1 (searching for OH stretching vibrations). A x100
178
objective lens yielded a theoretical spot diameter of 1.1 µm. The spectrometer was calibrated
6
179
against the first order Raman band of silicon located at 520.7 cm-1.Spectra bands were fitted with
180
the Fityk (0.9.8) software (Wojdyr, 2010) using a Lorentzian function.
181
3. Results
182
3.1 Replacement products (solids)
183
3.1.1 X-ray diffraction analysis
184
The XRD powder patterns of the samples from experiment set 1 (reacted with the phosphate
185
fluid) and experiment sets 2 and 3 (reacted with phosphate plus chlorine-containing fluids)
186
showed the same diffraction peaks. These were compared with the powder patterns of the
187
expected mineral products. Specifically, calcite (CaCO3), tricalcium phosphate (TCP -
188
Ca3(PO4)2), hydroxyapatite (HAP - Ca5(PO4)3OH), chloroapatite (ClAp - Ca5(PO4)3Cl), and
189
fluorapatite (FAP - Ca5(PO4)3F). All patterns could be found in the (X’Pert) database with the
190
exception of TCP that was measured in this study. In experiment sets 1, 2 and 3 calcite, HAP and
191
TCP peaks are present (e.g. Fig. 1a) and in experiment sets 4 and 5 (reacted with fluoride) only
192
FAP was present (e.g. Fig. 1b).
193
3.1.2 Extent of replacement and texture
194
After reaction, the external dimensions (~1.5 × 1.5 × 1.5 mm) and the cubic morphology of all
195
samples were preserved. BSE images and EDX analyses of cross-sections of the reacted samples
196
revealed that after four days of reaction some samples were partially replaced while others were
197
fully replaced (details in Table 1). The BSE images also show that the internal microstructure of
198
the marble was reproduced in the new mineral, including the shape and position of grain
199
boundaries, and many fractures of the original marble. Importantly, the mineral replacement
200
product contained newly formed porosity.
201
Partially replaced samples included all samples from sets 1, 2 and 3 (e.g. Fig. 2a, b, c). The extent
202
of reaction and the shape of the reaction front was different in each case. Partially replaced
203
samples had sharp compositional reaction interfaces between the parent and the product material
204
that evolved heterogeneously (at different rates) from the surface to the core of the samples,
205
resulting in an irregular and scallop-shaped reaction rim (e.g. Fig. 2a). Inside the reacted rim,
206
sections (up to 10 μm) of unreacted marble were “left behind” by the main reaction front (e.g.
207
Fig. 2a). Even in the samples in which the reaction reached the core, these unreacted pieces of
208
marble were mostly still present (e.g. Fig. 2b). Sometimes these unreacted parts are located at the
7
209
centre of individual grains, but there are many areas where these are crossed by grain boundaries
210
and fractures. Moreover, these unreacted parts typicallyoccur closer to the surface of the samples,
211
although in the highly concentrated fluid (2 M of ammonium phosphate) they appear randomly
212
distributed and throughout the sample (Fig. 2b).
213
The samples reacted with a small amount of Cl-containing fluids (0.01 M) were almost fully
214
reacted, having only a small quantity of unreacted marble near the surface. Increasing
215
concentrations of ammonium chloride and sodium chloride added to the phosphate fluids resulted
216
in a dramatic decrease of the extent of reaction (e.g. Fig. 2c). The reaction rim decreased from an
217
average width of ~500 µm in the samples reacted with 0.1 M NH4Cl to only ~50 µm width in the
218
samples reacted with 0.5 and 1 M NH4Cl. With NaCl, the negative impact on the replacement
219
rate was even greater. With 0.1 M NaCl the reaction rim was only ~25 µm thick, with 0.5 M it
220
decreased to ~10 µm, and with 1 M almost no reaction occurred (< 10 µm rims). The thin
221
reaction rims surrounded the edge of the marble grains, even in grains located at the inner part
222
and core of the samples (e.g. Fig. 2c). In sets 4 and 5, almost all samples were fully reacted (e.g.
223
Fig. 2e, f) with the exception of the samples reacted with 0.01 M of F-containing fluids (e.g. Fig.
224
2d), and one of the samples reacted with 0.25 M NaF that still contained unreacted marble at its
225
core.
226
3.1.3 Porosity structures
227
Although the unreacted Carrara marble contained a number of fractures as well as grain
228
boundaries that conferred some initial porosity and permeability to the samples, the porosity
229
described here is that generated by the replacement reactions.
230
Using BSE imageswe investigated the porous structure at a submicron scale. To help to interpret
231
the experimental results, the newly formed porosity is classified as:
232
o Type 1 – Very fine (<1µm) and round pores (examples in Fig. 3)
233
o Type 2 – Coarser pores, more or less elongated, of 1 to 10 µm diameter (examples
234
in Fig. 3)
235
A summary of the porosity types found in each reacted sample is shown in Table 1.
236
In some samples a small gap of less than 1 µm (Fig. 3a) was present at the interface between
237
unreacted and reacted minerals, but in most samples there was no visible gap (from SEM
8
238
images). Quenching effects, as the samples were quickly cooled and then exposed to the
239
atmosphere, cannot be ruled out as a cause of slight sample changes.
240
In experimental sets 1, 2, and 3,Type 1 pores were mainly located in the interior of grains and
241
Type 2 pores were located at the edge of reacted grains adjacent to grain boundaries (Fig. 3b).The
242
structures that contain Type 2 pores seem to have formed at initial stages of the replacement
243
reaction that proceeded from the grain boundaries into the grains. However, in samples reacted
244
with 0.5 and 1.0 M NaCl the opposite was observed: Type 1 pores were at the edge of the grains
245
and Type 2 in the mineral product inside the grain, indicating that the replacement product
246
evolved from the less porous (Type 1) grain surface (right side of Fig. 3c) to a coarser porosity
247
(Type 2 on the left side of Fig. 3c).
248
In experiments 4 and 5, thetype of porosity appears to correlate with fluoride concentrations. In
249
samples reacted with a low F concentration (0.01 M) the replacement at the surface of the sample
250
contains Type 2 porosity and no unreacted marble. This surface replacement is underlain by
251
another replacement microstructure resembling that of experiment set 1 (Fig. 3d). Porosity in the
252
samples reacted with slightly higher F concentrations (between 0.1 and 0.25 M) was quite
253
homogeneous, mostly round-shaped, and of Type 1. With 0.5MF concentration the porosity was
254
of Type 1, but at the core of the samples a geometrically differentiated microstructure of slightly
255
coarser (Type 2) porosity formed (Fig. 3e, f).
256
3.1.4 Element distribution and its relationship to porosity
257
The average chemical composition of the replacement products of all samples calculated from
258
EMPA point measurements is presented in supplementary material in Tables S3 and S4. EMPA
259
maps of the element distribution of the samples from sets 1, 2, and 3 show that in a number of
260
samples calcium and phosphorous were not homogeneously distributed in the replaced phase(s),
261
varying as a function of the different porosity structures present.
262
In sets 1, 2 and 3, the mineral phase with Type 1 porosity contained lower Ca (Fig 4a, d) counts
263
and higher P counts (Fig. 4b, e, g). Conversely, in Type 2 porosity there were higher Ca counts
264
and lower P (same corresponding figures). This indicates the possibility of different mineral
265
phases. This is further supported by Raman analysis. Raman intensity maps (Fig. 5) of a sample
266
from set 1 show that in the same type of porosity (Type 1 or Type 2) located next to grain
267
boundaries (“Y” shape seen in the Raman maps) the hydroxyapatite band at 961 cm-1 is very
9
268
intense and loses intensity with increasing distance where Type 1 porosity is seen. These maps
269
also show that the tricalcium phosphate (TCP) band at 405 cm-1 occurs in between the grain
270
boundaries and the unreacted marble characterized by porosity Type 1. Calcite bands (154, 281,
271
1087 cm-1) occur where the unreacted marble is located. An example of a spectrum taken at each
272
one of these locations is given in Fig. 6a (unreacted marble), b (porosity Type 1), and d (porosity
273
Type 2). A spectrum of a commercial TCP powder is shown in Fig. 6c for comparison with the
274
spectrum of porosity Type 1 phase. In summary, hydroxyapatite is located at the edge of the
275
grains where the porosity is coarser (Type 2), and TCP (or a very similar mineral) formed in the
276
interior of the grains where the porosity is finer (Type 1). Hydroxyapatite location (next to the
277
grain boundaries) is also marked by the absence of Mg seen in the microprobe map of Fig. 4c. A
278
list of the Raman vibrational bands of the spectra shown in Fig. 6 is presented in Table S5 of
279
supplementary material.
280
In the three first sets, the OH-1 stretching vibration band was found at 3573(1) cm−1 both in the
281
hydroxyapatite location (next to the grain boundaries) and, unexpectedly, at 3569(1) cm−1 at the
282
interior of grains of porosity Type 1 where the vibrational band of TCP (405 cm-1) is present (in
283
its pure state OH is not present in this mineral phase). In sets 2 and 3 (reacted in the presence of
284
chlorine salts), EMPA showed that chlorine was present in trace amounts and only in three
285
samples (reacted with 1 M NH4Cl, 0.5 M and 1 M NaCl in the fluid) at not more than 0.2(1)
286
wt.%. In set 3 (reacted with NaCl), Na was present in all samples and reached 2.1(2) wt.%,
287
although it is unclear if this Na was present as dried fluid in the grain boundaries or in the pores.
288
Na concentrations are higher at the coarser porosity side (Fig. 4f). Moreover, with an increase of
289
NaCl in the fluids, the total wt.% of elements decreased considerably from 94(3) to 85(5) wt.%
290
probably due to the increasing proximity of the measurement to the grain boundaries.It is worth
291
noting that the increase in NaCl concentration in the fluid inverted the locations of the 2 porosity
292
types compared to all other samples. Here, porosity Type 1 (finer) occurred next to the grain
293
boundaries where Ca counts are lower and P are higher, and the coarser porosity (Type 2)
294
occurred in the interior of grains where opposite Ca and P counts occur (Fig. 4d, e). The two
295
porosity Type locations shown in Fig. 4d and 4e were used for a more detailed EMPA analyses to
296
determine whether they correspond to two different mineral phases (Table 2). In this sample Type
297
1 porosity has a Ca:P wt.% ratio close to that of tricalcium phosphate (1.95), and Type 2 porosity
298
that developed into the grains approximates the ratio of hydroxyapatite (2.16). The opposite
10
299
trend(finer porosity at the centre of core and coarser at the grain boundaries)is seen in all the
300
other samples. The chemical composition of the two distinct porosity areas (Table 2) were
301
recalculated on the basis of a suitable number of anions for both HAP (13 oxygens) and TCP (8
302
oxygens). Finer porosity resulted in areas containing Ca2.9(PO4)2 + 0.2 Na and the coarser
303
porosity for Ca5.3(PO4)3OH + 0.1 Mg + 0.3 Na.
304
The samples reacted in the presence of fluoride gave homogeneous distributions of Ca, P, Mg and
305
Na (when present), with two exceptions - samples reacted with very high F concentration (0.5
306
M), and samples reacted with very low F concentration (0.01 M). The former (high F) showed a
307
high deviation of results at the centre of the cube, registering standard deviations up to 50%,
308
probably due to residue from dried fluid inside this sample because the porosity was slightly
309
coarser at the centre of these samples (Fig. 3e and 3f). In both sets, fluoride average content
310
increased linearly (R2= 99%) in the product samples with increasing F used in the fluids. The
311
second (reacted in very low F concentration) shows that the coarser porosity structure (Type 2)
312
that formed at the surface of the sample is a fluoride-containing replacement rim (Fig. 4h),
313
confirmed by Raman analysis to be fluorapatite (Fig. 6e). In the fluoride-free part, element
314
distribution (comparing Fig. 4b, g, h) and Raman spectra are the same as in the samples reacted
315
with the ammonium phosphate alone. The main interface between the F-containing and F-free
316
phases is marked by a Mg peak (Fig. 4i), and similar to the samples of set 1, is absent around the
317
grain boundary (the location of the grain boundary is marked with a black line in Fig. 8g). In all
318
other samples reacted with fluoride, Mg was homogeneously distributed, but appeared as spots of
319
higher intensity in the replaced areas. In these samples the OH band was not detected in the
320
Raman spectrum.
321
The average chemical formula of fluorapatite samples calculated from the EMPA results were:
322

Set 4: Ca10.1(1) (PO4)5.9(2) F1.8(4) OH0.2(4) ; + 0.1(0) Mg
323

Set 5: Ca10.1(2) (PO4)5.9(1) F1.6(5) OH0.4(5) ; + 0.2(1) Mg + 0.3(1) Na
324
3.1.5Mass changes
325
The relative mass changes before and after each experiment (Table 1) were used to infer the
326
extent of reaction, and in fully reacted samples the mass changes were used to estimate the total
327
porosity.All experimental samples(except the samples reacted in 1 M (NH4)2HPO4 + 0.5 M
328
NH4F) had higher densities (positive weight changes) after reaction.In experiment set 1 the
11
329
relative mass of the samples increased with increasing ammonium phosphate concentration in
330
accordance with the higher degree of replacement.In experiment sets 2 and 3, adding increasing
331
amounts of ammonium chloride resulted in a slight decrease of the relative mass in accordance
332
tothe lower degree of replacement. However, themass of samples reacted with fluoride-
333
containing fluidswas not consistent with the degree of replacement. Samples reacted with 0.1 M
334
F (fully replaced samples) were heavier than those reacted with 0.01M F (partially replaced), but
335
the samples reacted with 0.25 M F (also fully reacted) were lighter than the previous (reacted
336
with 0.1 M F), and with 0.5 M F even lighter (the relative wt.% change is lower, Table 1). This is
337
most likely due to the variations in the porosity in each case.
338
3.2Replacement products (fluids)
339
To identify if some of the parent material (calcium and trace magnesium present in the marble)
340
was transferred to the bulk fluid, the chemical compositions of the initial and final fluids were
341
analysed. The initial fluids contained neither calcium (Ca) nor magnesium (Mg). Parent material
342
losses into the bulk fluid contribute to solid weight losses. In the final fluids very small quantities
343
of Mg and Ca were found. The chemical analyses (ICP-OES) of the fluids after reaction and the
344
respective mass loss, due to element release,are shown in Table 3.There was a slightly higher
345
concentration of Ca and Mg in the fluids after reaction with higher salt concentrations. The
346
highest Ca loss was from the sample reacted with 2 M ammonium phosphate and the minimum
347
with 0.1 M. The highest Mg loss was from the sample reacted with 0.25 M NH4F. The maximum
348
Mg loss from sets 1, 2, and 3 was from the sample reacted with 1 M NaCl. The pH of the bulk
349
fluids did not change during the experiments; initial and final pH of all fluids averaged 7.8(1).
350
3.2.1 Fluid pathways and replacement rates
351
BSE images of the reacted samples showed that the fluid transport pathways depended on the
352
fluid composition. With ammonium phosphate the aqueous fluid moved arbitrarily both through
353
grain boundaries and through the newly formed porosity (Fig. 2a). With the addition of chlorine
354
in the fluids and decreasing replacement rates, fluid transport through grain boundaries was
355
highly preferred (Fig. 2c). With fluoride-containing fluids the replacement rates were very high
356
and transport through the newly formed porosity was preferred. Evidence of this can be seen in
357
almost fully replaced samples that show no replacement around grain boundaries ahead of the
12
358
main reaction front, also evidenced by the lack of any relict unreacted marble within the reaction
359
rims (Figures2e and 2f).
360
4. Discussion
361
All experiments involved the reaction of a calcium carbonate rock (Carrara marble) with different
362
concentrations of phosphate fluids (with and without the addition of chloride and fluoride salts)
363
and resulted in the pseudomorphic replacement of the rock by new phosphate product phases
364
while retaining the external dimensions of the marble parent phase. The conservation of volume
365
results from the high degree of coupling between the dissolution and the precipitation processes
366
within a fluid boundary layer at the parent solid-fluid interface (Putnis and Putnis, 2007). All
367
products were porous, thus allowing progress of the phosphate-bearing solution to the reaction
368
interface within the marble. Such pseudomorphic replacement reactions have been previously
369
reported and described as an interface-coupled dissolution-precipitation mechanism (see Putnis,
370
2009, and references therein).
371
4.1 End products
372
The end products of the experiments performed here resulted in calcium phosphate phases with
373
variable composition. This is not surprising, given the variability of fluid composition and the
374
wide range of element substitutions possible within the apatite crystal structure (Hughes and
375
Rakovan, 2015).
376
Geochemical modeling performed with PHREEQC (Parkhurst and Appelo, 1999) has proven to
377
some extent to be successful for predicting the end products of the reaction of the marble with the
378
five phosphate bearing solutions (adding the two chloride and the two fluoride salts). For set 1,
379
PHREEQC predicted that at initial conditions the mineral with the highest saturation index was
380
hydroxyapatite (HAP) followed by tricalcium phosphate (TCP), and both minerals were found in
381
the reacted samples. For set 2 and 3, the prediction was not as accurate. The SI of chlorapatite
382
(ClAP) was predicted to be the second highly saturated mineral after HAP and followed by TCP,
383
however no measurable ClAP occurred in the reacted samples and only HAP and TCP were
384
present. For sets 3 and 4, the model predicted fluorapatite as being the most highly supersaturated
385
phase and it was indeed the only phase found in the reacted samples of these sets.
386
In a closed system (such as in batch experiments) the replacement reaction results in local
387
changes in the composition of the bulk fluid, but most importantly, inside thenewly-formedpores
13
388
of the sample itself. An evolving solution composition as the reaction interface moves through
389
the reacting crystal, allowing for diffusion of elements within the fluid phase, may also be
390
responsible for the heterogeneity of end-products(such as in experiment sets 1, 2, and 3). Similar
391
examples of compositional and phase variabilityresulting from replacement reactionshas been
392
reported for the replacement of calcite by both dolomite and magnesite (Etschmann et al., 2014)
393
as well as in apatite replacement in the presence of As in solution (Borg et al., 2014). In both
394
cases this was ascribed to variations in composition and hence local equilibrium at a reaction
395
interface.From EMPApoint analyses it was seen that none of the samples matched the
396
stoichiometric Ca:P atomic ratio of apatite (1.67), and the maps showed different distributions of
397
calcium and phosphate, but also the presence of F, and minor Mg, Na, and Cl in the replacement
398
products (dependent on original fluid composition), that suggests possible incorporation of these
399
elements in the apatite structure during replacement.The possibility of incorporating CO3 in the
400
apatite structure is discussed later.
401
Considering our results, thegeneral chemical equations of the replacement reactions can be
402
written for sets 1, 2, and 3 as,
403
5 CaCO3 (s) + 3 HPO42- (aq) + H2O → Ca5(PO4)3OH (s) + 5 CO32- (aq) + 4 H+(aq),
404
and
405
3 CaCO3 (s) + 2 HPO42- (aq) → Ca3(PO4)2 (s) + 3 CO32- (aq) + 2 H+ (aq),
406
with trace amounts of Mg2+ and Na+ (when present) substituting for Ca2+, and possible CO32-
407
substitutionfor OH-and/or PO43-.
408
For sets 4 and 5 as,
409
5 CaCO3 (s) + 3 HPO42- (aq) + F- → Ca5(PO4)3F (s) +5 CO32- (aq) + 3 H+ (aq),
410
With possible minor trace amounts of Na+ substituting for Ca2+, and CO32- substituting for OH-
411
and/or PO43-.
412
4.2 Porosity generation
413
Results showed that the different porosity microstructures are directly linked to the chemical and
414
mineralogical changes in the reacted samples. The precipitation of different phases depends on
415
the saturation state of the fluid, and the different solubilities, molar volumes and morphologies of
416
the precipitating phases will determine the porosity (Pollok et al., 2011). Thus ultimately fluid
14
417
chemistry determines porosity and the composition of the fluid at the reaction interface depends
418
on the fluid and element transport through the reaction rims. A more homogeneous porosity
419
distribution was associated with a homogeneous distribution of elements (Ca, P, F, Mg), such as
420
in samples reacted with moderate F contents (0.1 and 0.25 M). The distinct porosity Type 2 that
421
was present at the edge of replaced grains was identified as being a magnesium-depleted area
422
(from EMPA maps) and Raman analysis showed that the product was hydroxyapatite. Where the
423
porosity was finer (Type 1) Raman spectra significantly changed, identifying other mineral
424
phase(s) present (TCP, or a hydroxyapatite solid solution).
425
Assuming that the end product of the first three experiment sets was a mixture of tricalcium
426
phosphate (TCP) and hydroxyapatite (HAP), and in the last two sets fluorapatite (FAP), the
427
theoretical molar volume changes that would result from these replacement reactions is -8.7% for
428
TCP, -13.8% for HA, and -14.7% for FAP. On this basis alone, it is expected that the replacement
429
of calcite by FAP results in the highest porosity, followed by HAP and TCP. However, the
430
solubility differences between parent and product phases in the solution at the interface are not
431
known precisely. Furthermore as the fluid composition evolves during the replacement reaction,
432
so do the relative solubilities of the associated phases in the evolving fluid at the reaction
433
interface. As fluorapatite is the least soluble of the apatite phases this would also add to an
434
increased volume deficit and hence higher porosity for the case of FAP. Although a more detailed
435
measurement of total porosity would be needed (such as by use of microtomography), results
436
showed, as expected, that samples replaced by FAP effectively contained larger pores than the
437
product mixture of TCP and HAP.
438
From SEM observations, the samples reacted with NaF have coarser porosity than those reacted
439
with NH4F although, Table 4 shows that the samples reacted with the same concentration of
440
NH4F have higher total porosity. The different scale of the porosity in each case may reflect the
441
effect of fluid composition on the nucleation density of the apatite, with the presence of NaF
442
resulting in lower nucleation density, larger crystals and coarser porosity while the presence of
443
NH4F results in high nucleation density, finer crystals and porosity on a much finer scale. Such
444
results have implications for the design of porous apatite materials using the replacement route.
445
For example, for the conservation of cultural heritage (Sassoni et al., 2011), the replacement of
446
limestone by a less soluble material such as apatite might be more durable, however, due to high
447
porosity it might break easily losing its suitability.The replacement of calcite by apatite could
15
448
represent a route for the development of a ceramic for medical application (Vallet-Regí,
449
2001).There is a clinical need for new bone replacement materials that combine long implant life
450
with compatibility and appropriate mechanical properties. Dissolution-precipitation is a process
451
that has been proposed for the synthesis of porous biocompatible material (apatite) for bone
452
implants (Heness and Ben-Nissan, 2004).
453
4.2.1Estimating porosity by examining weight changes
454
The weight percentage (wt.%) change between unreacted and reacted samples depends not only
455
on the progression of the reaction, but also on the density differences between the parent and
456
product phase(s), and on the total porosity formed. Moreover, the total porosity depends on the
457
molar volume changes and the solubility differences between the parent and product (Pollok et
458
al., 2011). The generation of porosity is a key factor for the progression of reaction because it
459
creates the pathway for fluid infiltration towards an ongoing reaction front. The simple fact that
460
we obtained fully reacted samples indicates that this pathway (or permeability) existed at least
461
until full replacement of the samples occurred.
462
For fully-reacted cubes that did not change size after reaction (pseudomorphed), the total volume
463
of the pores formed can be determined from the difference in expected weight (if there was no
464
porosity) and the actual weight measured after the sample is dried. Using the initial sample
465
(Carrara marble microprobe analysis, calcite 99.7%) density calculated as pure calcite (2.71
466
g/cm3) and the density of fluorapatite (FAP) as 3.20 g/cm3, the porosity of most of the samples
467
replaced by fluorapatite could be calculated (Table 4).The relative molar volume difference
468
between calcite and fluorapatite is -14.7 %. This matched the amount of porosity formed in the
469
samples reacted with 0.5 M NaF in the phosphate fluid (Table 4). Samples reacted with higher F-
470
content in the fluid had higher porosity (Table 4) and higher but more variable F counts in the
471
EMPA data. This suggests that samples reacted with fluids containing a higher F content
472
approximate to end-member fluorapatite, while samples reacted with lower F content are possibly
473
replaced by hydroxyfluorapatite (Ca10(PO4)6OHxF2-x) which would be slightly less porous
474
because the relative volume difference between calcite and HAP (-13.4 %) is lower than between
475
calcite and FAP ( -14.7 %).
476
4.3 The role of fluoride in the replacement reaction
16
477
Fluoride salts were added to the phosphate fluids in experiment sets 4 and 5. Fluorapatite is the
478
least
479
(Ca5(PO4)3(OH,Cl,F),(Dorozhkin, 2007). This was also the main replacement product in our
480
experiments when F was added to the fluids, as was confirmed by EDX and Raman analyses.F-
481
containing fluidsalso significantly enhanced replacement rates. Fluoride added as NH4F resulted
482
in even higher replacement rates than with NaF. The replacement products contained overall
483
homogeneous porosity structures, no unreacted marble, and less chemical variability in the
484
fluorapatite.
485
During the replacement of calcite by fluorapatite the “preferred” fluid migration pathway was
486
through the newly-formed porosity rather than along the pre-existing grain boundaries and
487
fractures in the marble.The replacement reaction evolved from the surface into the core of the
488
sample with almost noobservable influence of the presence of grain boundaries. This was
489
different fromthe case for the samples reacted with chloride salts where the reaction took place
490
first along grain boundaries. This suggests different wetting properties of fluids of different
491
composition may influence the preferred reaction path taken,but may also be related to the much
492
faster reaction rate in F-bearing fluids and the formation of fluorapatite.
493
4.4 The role of carbonates in the replacement reaction
494
During the replacement of the calcite by calcium phosphates the carbonate may be released to the
495
fluid phase or reincorporated in the new phases. The fact that the pH of the fluids at the end of the
496
experiments did not increase, as would be expected if significant carbonate ions were
497
present,suggests that carbonate released during the dissolution of calcite may have been
498
incorporated into the structure of the product phases. To some extent, this is supported by Raman
499
analysis.
500
Carbonate apatites are classified in three different types depending on the crystal lattice position
501
of CO32-: A-type, when it occupies the OH− site; B-type, when it occupies the PO4−3 site; and AB-
502
type, when it occupies both PO43− and OH− sites. In a Raman study by Awonusi et al, (2007) for
503
heavily carbonated apatites (up to 10%) of B-type, a combination of the υ1 vibrational mode of
504
carbonate with the υ3 phosphate mode was assigned to a broad band at the region between 1070-
505
1076 cm-1. For the same type of substitution, Penel et al, (1998) assigned the band at 1070 cm -1,
506
and Antonakos et al., (2007) assigned the bands 1072, 1074, 1075, and 1080 cm-1. Besides this
soluble
compound
among
the
three
common
calcium
orthophosphates
17
507
band, all three studies reported two more bands present: one at the υ3 mode of PO43- between
508
1026 cm-1 (highly carbonated) and 1030 cm-1 (less carbonated); and the other between 1045 and
509
1047 cm-1, also from lower to highly carbonated apatite. In the previously cited studies the υ1-
510
PO43- band position of B-type carbonate apatites was located between 959 and 961 cm-1.
511
The Raman signal of the samples from our experiments may have similarities to those inprevious
512
studies, specifically, the υ1-PO43- band position at around 961-962 cm-1, and the variable υ1-PO43-
513
bands at around 1028(1), 1044(2), and 1075(1) cm-1(Table S5 supplementary material). Hence,
514
the replacement of calcite by a carbonated apatite seems probable.
515
4.5 The role of magnesium in the replacement reaction
516
Magnesium was transported out of the calcite during the replacement reactions. In the F-free
517
fluids the transport of magnesium was localized specifically to the edge of replaced grains and
518
corresponded to the location of hydroxyapatite. The Raman detection of the OHgroup in the Mg-
519
containing areas might be an indicator that instead of TCP, a hydroxyapatite that incorporated
520
minor Mg and carbonate, such as (Ca,Mg)5(PO4)3(OH,CO3)may be present.
521
The amount of magnesium in the product fluids of samples replaced by fluorapatite was only
522
slightly higher than in the free-fluorine fluids, but the fact that it appears as spots in the samples
523
might indicate that it was trapped in the pores and the reaction time was not enough to permit its
524
fullescape into the bulk fluid. Magnesium is most probably not incorporated into the fluorapatite.
525
4.6 Fluid composition and replacement kinetics
526
In this study, higher concentrations of (NH4)2HPO4 resulted in faster replacement rates. The
527
coupling between dissolution and precipitation that results in pseudomorphic replacement
528
suggests that the rate controlling step is dissolution (Xia et al, 2009b) and that whatever
529
influences the dissolution step would have a similar effect on the replacement kinetics. However,
530
atomic force microscopy (AFM) experiments by Klasa et al. (2013) showed that the dissolution
531
rate of calcite (CaCO3) is significantly reduced in the presence of (NH4)2HPO4 in solution, but
532
not in the presence of Na2HPO4 (both solutions pH ~ 8)compared to that observed in pure
533
deionized water, suggesting that the NH4+ group may inhibit calcite dissolution. Moreover, the
534
experiments performed here showed that phosphate fluids containing Cl had a negative kinetic
535
effect on the replacement rates, both when adding Cl as NaCl and NH4Cl, with the addition of
536
NaCl having a more significant effect in decreasing the reaction rate than NH4Cl. An AFM study
18
537
by Ruiz-Agudo et al., (2010) showed that in the presence of small amounts of NaCl (1 mM),
538
calcite dissolution is slightly higher than in deionized water, and at higher concentrations of NaCl
539
the dissolution rate is enhanced. A further AFM study by Wang et al., (2012) on the replacement
540
of calcite by apatite showed that in the presence of high concentrations (0.5 M) of NaCl, the
541
calcite surface displayed a greater density of calcium phosphate nuclei, indicating that the
542
precipitation of apatite was enhanced in the presence of NaCl. Both studies suggest that the
543
presence of NaCl would increase the replacement rates, because it increases both dissolution of
544
calcite and precipitation of calcium phosphates. However, in this study the presence of NaCl did
545
not increase the replacement rates suggesting that some factors other than dissolution rate and
546
precipitation rate may affect the overall reaction rate. On the other hand, the AFM experiments
547
were carried out under quite different conditions of fluid flow over a free cleaved calcite surface
548
and so the chemical composition of the fluid was maintained constant, not allowing for fluid re-
549
equilibration. In our experiments it is possible that the higher NaCl concentrations together with
550
PO4-2 in solution effectively increased the threshold for supersaturation of the precipitating phase
551
within the boundary layer.
552
4.7Fluid transport
553
During a replacement process, natural porosity of rocks, such as grain boundaries and
554
fractures,play an important role in allowing fluid access within the rock. This enhances the
555
available internal surfaceswhere replacement reactionsare initiated. Fluid movement through
556
newly formed pores plays an important role in the advancement of a reaction front. If fluid
557
movement was limited, the reaction front may be arrested.In this case the replacement process
558
would stop. Fluid movement could also be limited by the loss of permeability. Permeability
559
depends on the interconnection of pores formed in the product phase. When a sample is fully
560
replaced, such as occurred in several samples of sets 3 and 4,this indicates that it was, at least
561
temporarily, permeable. The importance that interfacial fluid composition plays on replacement
562
reactions is extensively reported in Putnis (2009).
563
Our results indicate that there are two types of fluid preferred pathways: along grain boundaries
564
and fractures and through newly-formed connected pores. We hypothesize that when replacement
565
rates are very low, the absence of newly formed porosity results in the motionof fluid along the
566
grain boundaries, resulting in the replacement of interior grains (both surface and interior grains
567
contained very small replacement rims), as seen with the samples reacted with high chloride
19
568
content solutions. On the other hand, when replacement rates are high, the fluid moves through
569
the newly formed poresas seen in the samples reacted in the presence of fluoride. The movement
570
of the fluid preferentially through newly-formed poresrather than fractures and grain boundaries
571
may be related to wettability. The product and parent phases have different critical surface
572
tensions that are determined by the chemistry of each solid phase, and therefore have different
573
surface tension or wetting properties. However, in the absence of data on the wetting properties
574
of different fluid compositions on calcite at the temperature of the experiments this hypothesis
575
could not be tested.
576
Conclusions
577
This study demonstrates that fluid composition of a reacting fluid controls the chemical
578
composition, mineralogy, and porosity generated in a product during a pseudomorphic
579
replacement reaction. All experiments on the replacement of calcite by Ca-phosphate shared the
580
typical characteristics of an interface-coupled dissolution-reprecipitation mechanism.
581
Carrara marble (calcite) can be replaced by hydroxyapatite and another phase that shares many
582
similarities to tricalcium phosphate upon reaction witha phosphate solution ((NH4)2HPO4).The
583
addition of NH4Cl and NaCl to this solution resulted in the same product phases. However, in the
584
presence of fluorideas dissolved NH4F and NaF,the mineral product was fluorapatite.
585
Reaction rates were affected by both the composition and ionic strength of the bulk fluids.An
586
increase inthe phosphate concentrations in the fluidincreased the replacement rates. For the same
587
concentration of phosphate, the addition of increased amounts of chloride (Cl) lowered the
588
replacement rates. The inhibitory effect of chloride added as NaCl was higher than when added
589
as NH4Cl, indicating that the cations Na+ and NH4+may play an important role in the kinetics of
590
this replacement reaction. In the presence of fluoride the reaction rates were greatly increased and
591
the product phases were uniform in composition.All samples reacted with 0.1 M or more of
592
fluoride ions were fully reacted within the reaction time.In the presence of a small amount of
593
fluoride (0.1M) the reaction resultedfirst in the generation of a fluorapatite rim and proceeded
594
inwardsfrom the surface to the core of the marble cube.
595
Porosity within the replacement products varied with the chemical composition of the fluids, and
596
depended mostly on the mineral phase that was precipitating.Porosity conferred permeability to
597
the rock during replacement, permitting mass transfer between the bulk fluid and the reaction
20
598
interface. The presence of grain boundaries also conferred permeability to the rock samples. Both
599
mass transfer pathways were significant for the replacement using phosphate fluids alone, but
600
with the addition of chloride salts the replacement rates decreased significantly and the mass
601
transfer through grain boundaries became increasingly dominant. With fluoride-containing fluids
602
grain boundary mass transfer was not as important because the rapid advance of the reaction front
603
resulted in preferred mass transfer through the newly formed porosity.
604
This study demonstrates the important impact that small differences in fluid composition can
605
have on the progression of solid-liquid reactions. Small fluid compositional changes can change
606
the nature of the products, the microstructure and texture of rocks and the resultant porosity and
607
permeability.Thisis significant for the interpretation of microstructures in rocks as well as for
608
thesynthesis of new materials.
609
Acknowledgements
610
This project is supported by the Marie Curie Initial Training Network, FlowTrans (grant
611
agreement n° 316889) within the EU Seventh Framework Programme.The authors thank François
612
Renard, Liene Spruzeniece and Teresita Moraila-Martínez for valuable discussions. The authors
613
also thank Jasper Berndt-Gerdes andBeate Schmitte for help with electron microprobe analyses,
614
Peter Schmidt-Beurmann for help with XRD, Helen E. King and Martina Menneken for
615
instruction for Raman use, and Veronika Rapelius for ICP-OES analyses.
616
References
617
Ames, Jr., L.L., 1959. The genesis of carbonate apatites. Economic Geology. 54(5), 829–841.
618
Ames JR., L.L., 1961. Anion metasomatic replacement reactions. Economic Geology, 56(3), 521-
619
532.
620
Antonakos, A., Liarokapis, E., Leventouri, T., 2007. Micro-Raman and FTIR studies of synthetic
621
and natural apatites. Biomaterials, 28, 3043–3054.
622
Awonusi, A., Morris, M.D., Tecklenburg, M.M.J., 2007. Carbonate assignment and calibration in
623
the Raman spectrum of apatite. Calcified Tissue International, 81, 46–52.
624
Ben-Nissan, B., 2003. Natural bioceramics: from coral to bone and beyond. Current Opinion in
625
Solid State and Materials Science, 7, 283–288
21
626
Borg, S., Liu, W., Pearce, M., Cleverley, J., MacRae, C., 2014. Complex mineral zoning patterns
627
caused by ultra-local equilibrium at reaction interfaces. Geology,42(5), 415-418.
628
Dorozhkin, S. V. (2011). Calcium orthophosphates: occurrence, properties, biomineralization,
629
pathological calcification and biomimetic applications. Biomatter, 1(2), 121–164.
630
Etschmann, B., Brugger, J., Pearce, M.A., Ta,C., Brautigan, D., Jung, M., Pring, A., 2014. Grain
631
boundaries as microreactors during reactive fluid flow: experimental dolomitization of a calcite
632
marble. Contributions to Mineralogy Petrology, 168-1045.
633
Geisler, T., Nagel, T., Kilburn, M. R., Janssen, A., Icenhower, J. P., Fonseca, R. O. C., Grange,
634
M., Nemchin, A., 2015. The mechanism of borosilicate glass corrosion revisited. Geochimica et
635
Cosmochimica Acta, http://doi.org/10.1016/j.gca.2015.02.039.
636
Heness, G., Ben-Nissan, B., 2004. Innovative bioceramics. Mater. Forum, 27, 104–114.
637
Hughes, J.M., Rakovan, J.F., 2015. Apatite: A Mineral for All Seasons: Structurally Robust,
638
Chemically Diverse: Apatite and Apatite Supergroup Minerals. Elements, 11(3), 165-170.
639
Jonas, L., John, T., King, H.E., Geisler, T., Putnis, A., 2014. The role of grain boundaries and
640
transient porosity in rocks as fluid pathways for reaction front propagation. Earth and Planetary
641
Science Letters, http://doi.org/10.1016/j.epsl.2013.10.050.
642
Jonas, L., John, T., Putnis, A., 2013. Influence of temperature and Cl on the hydrothermal re-
643
placement of calcite by apatite and the development of porous microstructures: The American
644
Mineralogist, v. 98, p. 1516–1525.
645
Kasioptas, A., Geisler, T., Perdikouri, C., Trepmann, C., Gussone, N., Putnis, A., 2011.
646
Polycrystalline apatite synthesized by hydrothermal replacement of calcium carbonates.
647
Geochimica et Cosmochimica Acta, http://doi.org/10.1016/j.gca.2011.03.027.
648
Kasioptas, A., Geisler, T., Putnis, C.V, Perdikouri, C., Putnis, A., 2010. Crystal growth of apatite
649
by
650
http://doi.org/10.1016/j.jcrysgro.2010.05.014.
651
Kasioptas, A., Perdikouri, C., Putnis, C.V., Putnis, A. 2008. Pseudomorphic replacement of
652
single calcium carbonate crystals by polycrystalline apatite. Mineralogical Magazine,
653
http://doi.org/10.1180/minmag.2008.072.1.77.
replacement
of
an
aragonite
precursor.
Journal
of
Crystal
Growth,
22
654
Klasa, J., Ruiz-Agudo, E., Wang, L.J., Putnis, C.V., Valsami-Jones, E., Menneken, M., Putnis,
655
A., 2013. An atomic force microscopy study of the dissolution of calcite in the presence of
656
phosphate ions. Geochimica et Cosmochimica Acta, http://doi.org/10.1016/j.gca.2013.03.025.
657
Lackner, K.S., 2002. Carbonate Chemistry for Sequestering Fossil Carbon. Annual Review of
658
Energy and the Environment, 27(1), 193–232.
659
Milke, R., Neusser, G., Kolzer, K., Wunder, B., 2013. Very little water is necessary to make a dry
660
solid silicate system wet. Geology, http://doi.org/10.1130/G33674.1.
661
Nermoen, A., Korsnes, R.I., Hiorth, A., Madland, M.V., 2015. Porosity and permeability
662
development in compacting chalks during flooding of non-equilibrium brines - insights from long
663
term experiment. Journal of Geophysical Research: Solid Earth,
664
http://doi.org/10.1002/2014JB011631.
665
Parkhurst, D.L. and Appelo, C.A.J., 1999. User's guide to PHREEQC (version 2): A computer
666
program for speciation, batch-reaction, one-dimensional transport, and inverse geochemical
667
calculations: U.S. Geological Survey Water-Resources Investigations Report 99-4259, 312 p.
668
Penel G., Leroy G., Rey C., Bres E., 1998. MicroRaman spectral study of the PO 4 and CO3
669
vibrational modes in synthetic and biological apatites. Calcified Tissue International, 63:475–
670
481.
671
Pollok K., Putnis C.V. and Putnis A., 2011. Mineral replacement reactions in solid solution-
672
aqueous solution systems: Volume changes, reaction paths and end points using the example of
673
model salt systems. American Journal of Science, 311, 211-236.
674
Putnis A., 2002. Mineral replacement reactions: from macroscopic observations to microscopic
675
mechanisms. Mineralogical Magazine, 66, 689-708.
676
Putnis, A. 2009. Mineral replacement reactions. Reviews in Mineralogy and Geochemistry, 70,
677
87-124.
678
Putnis, A., Putnis C.V., 2007. The mechanism of reequilibration of solids in the presence of a
679
fluid phase. Journal of Solid State Chemistry, 180, 1783-1786.
680
Putnis, C.V., Tsukamoto K., Nishimura Y., 2005. Direct observations of pseudomorphism:
681
compositional and textural evolution at a fluid-solid interface. American Mineralogist, 90, 1909-
682
1912.
23
683
Putnis, C.V., Mezger K., 2004. Isotope tracing of a mineral replacement reaction: The KCl-KBr-
684
H2O system as a model example. Geochimica Cosmochimica Acta, 66, A618.
685
Ruiz-Agudo, E., Putnis, C.V., Hövelmann, J., Álvarez-Lloret, P., Ibáñez-Velasco, A., Putnis, A.,
686
2015. Experimental study of the replacement of calcite by calcium sulphates. Geochimica et
687
Cosmochimica Acta, http://doi.org/10.1016/j.gca.2015.02.012.
688
Ruiz-Agudo, E., Putnis, C.V., Putnis, A., 2014. Coupled dissolution and precipitation at mineral–
689
fluid interfaces. Chemical Geology, 383, 132–146.
690
Ruiz-Agudo, E., Putnis, C.V., Rodriguez-Navarro, C., Putnis A., 2012. The mechanism of
691
leached layer formation during chemical weathering of silicate minerals. Geology 40, 947-950.
692
Ruiz-Agudo, E., Kowacz, M., Putnis, C.V., Putnis, A., 2010. The role of background electrolytes
693
on the kinetics and mechanism of calcite dissolution. Geochimica et Cosmochimica Acta, 74(4),
694
1256–1267.
695
Sassoni, E., Naidu, S., Scherer, G.W., 2011. The use of hydroxyapatite as a new inorganic
696
consolidant for damaged carbonate stones. Journal of Cultural Heritage, 12(4), 346–355.
697
Turner, B.D., Binning, P., Stipp, S.L.S., 2005. Fluoride removal by calcite: evidence for fluorite
698
precipitation and surface adsorption. Environmental Science & Technology, 39(24), 9561–8.
699
Vallet-Regí, M., 2001. Ceramics for medical applications. Journal of the Chemical Society,
700
Dalton Transactions 97–108.
701
Wang, L., Ruiz-Agudo, E., Putnis, C.V, Menneken, M., Putnis, A., 2012. Kinetics of calcium
702
phosphate nucleation and growth on calcite: implications for predicting the fate of dissolved
703
phosphate species in alkaline soils. Environmental Science & Technology, 46(2), 834–42.
704
Wojdyr, M., 2010. Fityk: a general-purpose peak fitting program. Journal of Applied
705
Crystallography, Volume 43, pages 1126-1128
706
Xia, F., Brugger, J., Ngothai, Y., O’Neill, B., Chen, G., Pring, A., 2009a. Three-dimensional
707
ordered arrays of zeolite nanocrystals with uniform size and orientation by a pseudomorphic
708
coupled dissolution -reprecipitation replacement route. Crystal Growth & Design 9, 4902–4906.
24
709
Xia, F., Chen, G., Ngothai, Y., O’Neill, B., Putnis, A., Pring, A., 2009b. Mechanism and kinetics
710
of pseudomorphic mineral replacement reactions : A case study of the replacement of pentlandite
711
by violarite. Geochimica et Cosmochimica Acta 73, 1945–1969
712
Xia, F., Zhou, J., Brugger, J., Ngothai, Y., O’Neill, B., Chen, G.R, Pring, A., 2008. Novel Route
713
To Synthesize Complex Metal Sulfides: Hydrothermal Coupled Dissolution - Reprecipitation
714
Replacement Reactions. Chemistry of Materials 20, 2809–2817.
715
Yang, M.I.N., Hashimoto, T., Hoshi, N., Myoga, H. 1999. Fluoride removal in a fixed bed
716
packed with granular calcite. Water Research 33(16), 3395-3402.
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
25
735
736
737
Figure captions
738
Figure 1 Example of an XRD powder pattern representative of: (a) samples from experiment sets
739
1, 2 and 3 (reacted with ammonium phosphate and with chloride salts). The peaks matching
740
calcite (CaCO3), hydroxyapatite (HAP), and tricalcium phosphate (TCP) patterns are indicated;
741
(b) samples from experiment sets 4 and 5 (reacted with ammonium phosphate plus fluoride salts)
742
that matched fluorapatite.
743
Figure 2 BSE images of cross sections of Carrara marble cubes reacted for 4 days at 200°C, using
744
different solutions: (a) 0.5 M (NH4)2HPO4; (b) 2 M (NH4)2HPO4; (c) 1 M (NH4)2HPO4 + 0.5 M
745
NaCl (d); 1 M (NH4)2HPO4 + 0.01 M NaF (similar result with 0.01 M NH4F); (e) 1 M
746
(NH4)2HPO4 + 0.5 M NH4F; (f) 1 M (NH4)2HPO4 + 0.5 M NaF. Darker grey areas correspond to
747
the unreacted marble and brighter grey areas to the replacement product. The squares indicate the
748
sites of the zoom areas as shown in Figure 3.
749
Figure 3 Detail BSE images of the mineral and porosity textures of areas assigned in Figure 2a, b,
750
c, d, e and f, respectively. Darker grey areas correspond to the unreacted marble and brighter grey
751
areas to the replacement product. Black areas correspond to pores.
752
Figure 4 Electron microprobe element maps (100×100 µm) of samples reacted with: (a), (b), and
753
(c) 1 M (NH4)2HPO4; (d), (e) and (f) 1 M (NH4)2HPO4 plus 0.5 M NaCl; (g), (h), and (i) 1 M
754
(NH4)2HPO4 plus 0.01 NaF.
755
Figure 5 Example of a Raman spectra measured at: (a) unreacted marble (b) core of a fully
756
replaced grain of set 1 (reacted with ammonium phosphate and finer porosity - Type 1) (c)
757
commercial (beta) tricalcium phosphate powder; (d) surface of a replaced grain right next to the
758
grain boundary (reacted with ammonium phosphate and coarser porosity - Type 2); (e) samples
759
reacted with fluoride-containing fluids (sets 4 and 5).
760
Figure 6 Intensity maps (85 × 85 µm) of Raman vibrational bands in the area assigned in Fig. 2b
761
(the maps are mirror images of Fig. 3b). The sample was reacted with 2 M (NH4)2HPO4 at 200°C
762
for 4 days. Each map corresponds to one vibrational band.
763
26
764
765
766
Table 1 Summary of experimental and calculated results
Experim
ent
Set 1
n (NH4)2HPO4
Set 2
(NH4)2HPO4 + n
NH4Cl
Set 3
(NH4)2HPO4 + n
NaCl
Set 4
Set 5
Solid samples
Initial
IS
Mass
(200°C) mass
change
(mg)
(%)
Replacement
extent
0.2
0.9
2.1
4.7
2.1
2.2
2.3
2.6
2.1
2.2
2.6
3.0
8.8 ± 0.5
9.3 ± 0.0
8.7 ± 0.8
8.5 ± 0.9
8.4 ± 1.2
8.8 ± 0.7
8.9 ± 0.7
7.9 ± 2.1
9.1 ± 0.0
9.2 ± 0.3
9.2 ± 0.3
7.6 ± 1.8
0.9 ± 0.2
2.7 ± 0.7
4.2 ± 0.2
4.5 ± 0.3
3.7 ± 0.0
3.3 ± 0.4
2.8 ± 0.7
3.0 ± 0.2
3.8 ± 0.8
4.0 ± 1.3
2.2 ± 1.4
0.7 ± 0.8
Partial
Partial
Partial
Partial
Partial
Partial
Partial
Partial
Partial
Partial
Partial
Partial
0.01 2.1
0.1 2.2
0.25 2.3
9.2 ± 0.0
9.5 ± 0.4
8.8 ± 0.6
Partial
Complete
Complete
0.5
2.6
8.8 ± 0.8
3.1 ± 0.7
3.1 ± 0.5
1.8 ± 0.3
-0.2 ±
0.1
0.01 2.1
0.1 2.2
0.25 2.3
9.4 ± 0.0
8.2 ± 0.2
6.9 ± 1.1
4.0 ± 0.0
5.6 ± 0.3
4.0 ± 0.6
Partial
Complete
Complete
0.5
8.7 ± 0.7
0.8 ± 0.1
Complete
Fluid solution
(NH4)2HPO4 + n
NH4F
(NH4)2HPO4 + n
NaF
n
(M)
0.1
0.5
1
2
0.01
0.1
0.5
1
0.01
0.1
0.5
1
2.6
Complete
Porosity Type
(code)
1,2
1,2
1,2
1,2
1,2
1,2
1,2
1
1,2
1,2
1,2
1
2 (surface); 1, 2
(core)
1
1
1 (surface),2
2 (surface) 1,2
(core)
1
1
2 (1, inside the
crystal)
767
768
Table 2 Electron microprobe results of the replacement product of a sample reacted with 1 M
769
ammonium phosphate plus 0.5 M sodium chloride, showing the contrast between areas of finer
770
and coarser porosity. Values normalized to 100 wt.%.
wt.%
Type 1 porosity (Ca/P =
1.86)
Type 2 porosity (Ca/P =
Na2O
1.78 ±
0.22
2.31 ±
CaO
52.49±
1.32
56.39 ±
Cl
0.04 ±
0.03
0.11 ±
MgO
0.6 ±
0.1
0.6 ±
P2O5
46.1±
1.5
40.6 ±
Total
100.01 ±
0.01
100.01 ±
27
2.31)
0.16
0.8
0.04
0.1
0.9
0.01
771
772
Table 3 Ca and Mg concentrations (measured by ICP-OES) in the bulk fluid after reaction and
773
the respective mass lost by the samples. Both Ca and Mg were measured using a minimum of two
774
wavelengths, and standard deviations were lower than 0.06 ppm for Ca and 0.02 ppm for Mg.
sets 1, 2, 3
Min (mg/L)
Max (mg/L)
Average (mg/L)
Mass loss (wt.%)
Ca
0.1
3.6
1.1 ± 0.5
0.02 ± 0.01
Mg
0.0
2.7
0.9 ± 0.8
0.02 ± 0.02
sets 4, 5
Ca
0.1
1.7
0.7 ± 0.5
0.01 ± 0.01
Mg
0.3
2.9
1.5 ± 0.9
0.03 ± 0.02
775
776
Table 4 Average porosity (%) of samples fully replaced by fluorapatite.
Fluid composition
NH4F
NaF
1 M (NH4)2HPO4 + 0.1 M
12.7 ± 0.4
10.6 ± 0.3
1 M (NH4)2HPO4 + 0.25 M
13.8 ± 0.2
12.3
1 M (NH4)2HPO4 + 0.5 M
15.5 ± 0.1
14.7 ± 0.0
777
28