CEMENT and CONCRETERESEARCH. Vol. I0, pp. 223-230, 1980. Printed in the USA. 0008-8846/80/020223-08502.00/0 Copyright (c) 1980 Pergamon Press, Ltd.
STUDY of the EARLY HYDRATION of Ca3SiO 5 by X-RAY PHOTOELECTRON + SPECTROMETRY M. REGOURD :~, J.H. THOMASSIN::::, P. BAILLIF ;:~: and J.C. TOURAY:::: :: D~partement Microstructures, C.E.R.I.L.H. 23, rue de Cronstadt 75015 PARIS (France)
:::: Laboratoire de G~ologie, G~ochimie et Min~ralogie Appliqu~es, Universit~ d'Orl~ans,
45045 ORLEANS
(France)
(Communicated by H.F.W. Taylor) (Received Dec. 12, 1979) ABSTRACT C3S pastes (W/S = 0.5), have been studied from 5 seconds to 4 hours by X-ray Photoelectron Spectrometry. XPS reveals surfa,(c transformations of C3S grair~from very early ages of hydration. The modifications have been evidenced by a change in the environment of Si atoms and a variation of theCa/Si ratio. A Primary Hydrate (C/S << 3), a Secondary Hydrate (C/S ~2) and a Tertiary Hydrate (C-S-H type I) have been identified. XPS is a s~nsitive and reproducible method for the study of surface C3S hydration. Des p~tes de C3S ~ rapport E/S = 0,5 ont ~t~ ~tudi~es de 5 secondes ~ 4 heures par la spectrom~trie de photoelectrons. La m~thode XPS r~v~le des transformations ~ la surface des grains de C3S d~s les premiers instants de l'hydratation. Les modifications ont ~t~ mises en ~vidence par un changement dans l'environnement des atomes de Si et une variation du rapport Ca/Si. Un Hydrate Primaire (C/S << 3), un Hydrate Secondaire (C/S 2) et un Hydrate Tertiaire (C-S-H type I) ont ~t~ identifies. La specTrom~trie XPS est une mgthode sensible et reproductible pour l'~tude de l'hydratation superficielle de C3S.
Introduction The initial hydration of C3S has not yet been completely clarified. The nature of the hydrates that form from the beginning of the evolution of the C3S + H20 system through to the dormant period continue to divide researchers, some being partisans of dissolution-precipitation theory, others of topochemical reaction. In a recent synthesis, Skalny, Jawed and Taylor (1) have recalled the results of numerous studies of C3S hydration and their interpretation related to the theory of a protective layer and delayed nucleation. # Presented by M. REGOURD at the meeting of the French Society of Mineralogy and Crystallography on the Dissolution of Minerals : thermodynamic and kinetic aspects, mechanism, applications, 8 March 1979, PARIS (France). 223
224
Vol. I0, No. 2 M. Regourd, et al.
This article develops the results of our study by X-ray photoelectron spectrometry of the surface of the C3S crystals from the moment of their contact with water. Our previous data indicated a very rapid surface development of the hydrated phases with a C/S << 3 (2). Experimental X-Ray Photoelectron
Spectrometry,
XPS
XPS spectrometry is a rather new technique in the cement field. It has been applied essentially to the study of organic surfaces, catalysis and chemisorption on metals. The study of silicate materials is being developed at present. In this field, the most rewarding research has been in the use of this method for determining the surface composition of materials from the intensities of the peaks of different chemical elemen~ (3, 4). X-ray Photoelectron Spectrometry,XPS, also called ESCA, Electron Spectroscopy for Chemical Analysis, is a ~chniqueof measuring the kinetic energy of electrons photoejected from a solid surface, using soft X-rays. The kinetic energy E k of the photoelectrons can be related to the binding energy Eb which these electrons had originally in the solid. The apparatus utilized was an AEI ES 200 Photoelectron Spectrometer fitted width a magnesium anode (~Ig}~ = 1253.6 eV) under an irradiation power of 240 W. The XPS peaks studied were : Ca2p 3/2 (E B = 346.9 eV), Si2p (E B = 101.2 eV), Ols (E B = 530.9 eV) and Cls (E B = 285.0 eV). The samples were fixed on adhesive conducting tape. The analyzed area was about 25 mm2. The characteristics of the method are as follows: [I] the depth of investigation into the oxides and silicates is between 50 and I00 A. This depth has been chosen equal to 3 % (% is the mean free path). It corresponds to 95 % of the analyzed electrons. The other 5 % come from deeper zones (5). It is sufficient to provide data in volume while still allowing surface analysis ; [2] measurements of the intensity of peaks are easily reproduced and are usually of a quality superior to + I0 %, providing that a ratio of concentrations in a homogeneous material is determined [3]. It is a quantitative method providing it can be standardized on fresh fracture surfaces of homogeneous materials of known composition; [4] the samples do not require any particular preparation. Cleavage surfaces, grains, powders and fibers may be analyzed in the same way. Material The tricalcium silicate was synthesized at 1600°C by solid state reaction of a mixture of amorphous silica and reagant grade calcium carbonate in a closed platinum crucible. The product of this synthesis was air quenched and analyzed by X-ray diffraction, atomic absorption and electron probe microanalysis. The polymorphic form was T I. After grinding to a specific surface of 3000 cm2/g, a series of pastes was prepared with distilled water (water/solid = 0.5). At selected intervals, between 5 seconds and 4 hours, the hydration was stopped by the addition of acetone. Calibration
Curve
In order to determine the Ca/Si ratios in hydrated silicates a standard curve was drawn (Fig. I). The ratio of the area of the XPS peaks (Ca2p/Si2p) was compared to the atomic ratios Ca/Si in materials of well determzned com-
Vol. I0, No. 2
225 XPS, C3S, EARLY HYDRATION
Ca :p/Si2p
CaIsi 3
Ca !~/Si2.
1o! .
+
J
2
~.A~"
'J
2
,
,CVsi
3 atomic
FIG. i
FIG. 2
Calibration curve of Ca~ /Si~ intenL Z sity ratio versus Ca/Si Patom~c ratio.
versus time Variation of Ca2pSi of hydration. 2p
position : synthetic glasses (Na20 - CaO - MgO - Si02 system)kindly furnished by Saint-Gobain Industries, natural wollastonite(Ca3SiO 5),dicalcium silicate (Ca2SiO4)and tricalcium silicate(Ca3SiO5).The variation of the XPS results determined on several samples of the same anhydrous Ca3SiO 5 preparation is less than 5 %. Results The results of the photoelectron spectrometry are presented in the form of a kinetic curve (F~.2) and a table giving the position of the XPS peaks Ca2p 3/2, Si2p, Ols (Table I). Table I Difference between the binding energies of the XPS peaks, Ca2p 3/2, Si2p , Ols and width at mid-height of the Si2p peak
(Ca2p 3/2-Si2p) 6 (Si2p - 01s) 16 (Ca2p 3 / 2 - 0 1 s ) Time
eV + 0.2
eV+
0.2
eV + 0 . 2
Width at mid height of the Si2p peak eV + 0 . 1
5,10,15,30 1,3,5,15
245.7
429.7
184.1
2.5
245. 1
429.3
184.2
2.8
245.1
429.1
184.O
2.4
sec min
1,2 hours
4 hours
226
Vol. I 0 , No. 2 M. Regourd, et al.
The variation of the Ca2p/Si2p ratio (Fig. 2) from approximately 5 seconds to 5 minutes proves the importance-of the perturbations on the surface of the g~ains. These results show that the value of the Ca2p/Si2p ratio is always lower in hydrated samples than in C3S , but the binding energy and the shape of the Si2p peak show that the surface of the hydrated particles begins to differ significantly from that of the anhydrous C3S within a very short period. In particular, it is noticed that the differences of the binding energies Ca2p 3/2 - O]s tend toward the constant while those relative to Ca2p 3/2 - Si2p and Si2p Ols vary significantly between the anhydrous sample and the hydrated samples. These last results demonstrate a variation in the binding energy of the Si2p electrons from 5 seconds on. Interpretation and Discussion The XPS data relating to C3S can be interpreted in the framework of a model suggested by the XPS studies of g l a s s - aqueous solution interactions (6). With this model, it would be possible to explain the form of the XPS kinetic curve of Ca3SiO 5 hydration, the evolution of the dissolution of calcium and silicate ions, the existence of a dormant period and the setting of the paste (Fig. 3): i) At the beginning of the C3S-H20 reaction and after protonation of silicate and oxygen ions, a congruent dissolution of Ca3SiO 5 is assumed (7, 8). We emphasize in this regard, that in an open medium with a high rate of percolation, the interaction can limit itself to this first stage because of the lack of accumulation of the dissolution products. ii) The second stage detectable in the first few seconds (5 seconds on the curvejFig. 2) corresponds to the formation of a Primary Hydrate with a ratio C/S << 3 on the grains of the anhydrous silicate. The cloud of points shows that the surface, between 5 seconds and I minute, is a site of continuous exchange and not in a state of equilibrium. The quantitative interpretation of the XPS data has been carried out according to the following formula : I = f~ dI o
[I]
I is the area of the photoelectron peak corresponding
to a given energy
dI represents the contribution of a thickness dx to the depth x Reasoning from the analysis of a unit surface, dl = O.F.S.C.e -x/~. dx
F C x S
[2]
= cross-section of the sub-layer irradiated by the X-ray beams, = X-ray flux, = concentration within the solid, = depth in the solid, = mean free path of electrons, = function of the apparatus.
In the case where C does not vary with the depth, Equation [2] from 0 to ~ gives Equation [3] : I = o.F.S.C.~
the integration of [3]
Considering a model of two layers (C3S and hydrate) (9) and an average Ca/Si ratio close to I (7, 8, IO) in the primary hydrate, the calculations give :
Vol. I 0, No. 2
227 XPS, C3S, EARLY HYDRATION
Ica
=
CIca OCa F SCa { j~' o
Isi = ~S i F Ssi { I : Hydrate,
S' o
CI
e
Si e
dx
-x/%
~ C2 dx + fl' sie
2 : C3S (d = 3.14 gcm
-3
f £~,
-x/% dx}
-x/~
+
C2Ca e
-xI% dx}
, M = 228)
-3 We assume (i) the hydrate to have H20/Si = I, d = 2.2 gcm , M = |34 (ll), (ii) the thickness of the hydrate to be uniform, and (iii) the electrons Ca2p and Si2p to have the same mean free path, %Ca = >,Si = 20 A in C3S and in the hydrate. Dividing
ICa by Isi, we have
Ica
2.2 -£'/% ) + ~ ( I - e
oCa SCa% --
Isi
X
OS i SSiX
Using the standard
curve,
2.2 ( I - e 134
0.01642
Isi
0.01642 - 0.00265 ICa Isi
+ 0.02490
= 2.25
9.42
e-
l'
/%
-1'/% ) + -3.14 e_l, /X ~
the ratio becomes
ICa
For a value of
:
:
e-I'/% e-l'/ %
( t = lO s ), the computed
thickness
of the Pri-
o
mary Hydrate is 8 A. It should be noticed that this thickness is comparable to the thickness of the layer of C3S dissolved after one minute in the experiments of Fuji and Kondo ( 7 ). This value is 7 A for a w/s = 0.7 and a specific surface of 3800 cm2/g. (iii) The third stage manifes~ itself by an increase of the ICa/Isi ratio in which the maximum is attained at the end of one minute of hydration. This situation can be explained by the formation of a Secondary [lydrate having a Ca/si ratio higher that of the Primary Hydrate but lower than the observed maximum value of 2.7 because the latter includes a contribution from underlying C3S. This reaction is compatible with a dual origin of Ca2+ : the solution (process a) and the inner part of the grain (process b). These latter ions are a part of the ions released by the solid during the advancement of the interface hydrate-C3S. In this regard, the values of the intermediate Ca/Si, between 2 and 2.7 (Fig. 2), would confirm the hypothesis of the chemisorption of Ca 2+ ions on a surface enriched in silicon, a hypothesis deduced from measurements of the ~ potential, giving a positive charge to the Ca3SiO 5 particles in the first minutes of hydration (12) o That the values of the Ca/Si ratios are superior to 2 is explained by the contribution to the XPS signal of the subjacent anhydrous C3S , the amount of the Secondary Hydrate still being very low (Fig. 3). In this case,
the ratio of intensities
ICa Isi
4.4 e-l/X) 190 (l -
oCa. SCa.% -
is : 9.42 + --~
e
-11%
x
osi. Ssi %
2.2 (1 - e-l/%) + 3.14 190 228
e-l/~
228
Vol. I0, No. 2 M. Regourd, et al.
(',¢ S
]
t
t
~ ~ s i o 4-
(;3 S
C3S
=E
Congruent dissolution after ion exchange (protonation)
t = lO sec Formation of the Primary Hydrate (C/S - I)
%s
~...
%s ~
0
OH-
J.
%s
=
Anhydrous
I
I!II ill 0
Ca2+ lO sec < t < I min Transformation of the Primary hydrate into a Secondary Hydrate ( C / S ~ 2 )
OH-
"
I min < t < 4 hours Growth of the Secondary Hydrate
Ca:'+ H÷ H-
,:~~
t ~ 4 hours End of the dormant period after reorganization of the Secondary Hydrate. Congruent dissolution of C~S after the breaking of Si-O bonds
t > 4 hours Precipitation of Ca(OH)2 in hexagonal plates and C-S-H in foils or fibers
Fig. 3 Schematic model of suggested mechanism of C3S hydration
up to 4 hours
Vol. I0, No. 2
229 XPS, C3S, EARLY HYDRATION
M
--
190, d = 2.2 gcm
-3
for the hydrate
Ica
0.02316 + 0.01816 e-£/%
Isi
0.01158 + 0.00219 e -I/%
(H20/Si = I)
This calculation may be applied to determine the thickness of the Secondary Hydrate at the maximum of the kinetic curve, t = I minute. The ratio Ica -2.7 o Isi gives £ = 8 A. This value being the same that the computed £' value of the Primary Hydrate strongly suggests that during the first stages of hydration only the fixation of Ca 2+ from the solution occurs (Fig. 3). Finally, the proposed model assumes that a layer of the Secondary Hydrate with C/S -~ 2 grows on a substratum of C3S. This assumption is strongly suggested by the value C/S -~ 2 obtained when this layer is thicker than 60 A at t ~ 15 minutes. Subsequently, the Ca/Si ratio indicated on the kinetic curve slowly decreases and tends asymptotically towards a value close to 2 in about 15 minutes. At this moment, the thickness of the layer of the Secondary Hydrate has sufficiently increased (via the process b also) so that the contribution of the C3S to the XPS signal is negligible. This thickness is equal to 3 times the mean free path of the Ca2p and Si2p electrons, that is to say approximately 60 A. The formation of the Secondary Hydrate could also bepartly explained by precipitation from solution. This mechanism explains why silica concentrations in solutions drop drastically after some minutes of hydration (7, 8). In our model, this process cannot be distinguished from the process a and probably occurs simul t aneous ly. The Secondary Hydrate develops and persists to the supersaturation of the solution~
through the dormant period up
In slowing down the transfers between the solution and the reactional interface, we could consider that Secondary Hydrate plays the role of a barrier which accounts for the low activity of the induction period revealed by the calorimetric curves. However the rate of C3S hydration and the nucleation of hydrates are governed by the concentration of Ca 2+ and OH- in the solution. This was shown by Barret, M~n~trier and Bertrandie (8), Young and Berger (13) : either removal of Ca 2÷ and OH- from the solution reactivated the hydration of C3S or, on the contrary, a supersaturated solution used as mixing water strongly delayed the dissolution of C3S crystals. iiii) After four hours, important reorganizations take place in the interior of the Secondary Hydrate as evidenced by the change in the shape of the Si2p peak. Apparently, this modification indicated the end of the dormant period. The end of the dormant period manifests itself by the massive nucleation of Ca(OH) 2 and the Tertiary Hydrate (C-S-H Typel) with dimeric silicate ions(Si207)6-. One hypothesis to explain this new stage is that of the existence of a not very permeable, superficial barrier that slows down the exchanges between the solution and the solid. In this case, the end of the dormant period correspond to a nmssive increase in the porosity of that barrier, by the formation of cracks, for example at the grain boundaries or at dislocations emergency. It is this hypothesis that we have shown in figure 3 in supposing that the reorganizations observed in the Secondary Hydrate are the cause of the massive
230
Vol. I0, No. 2 M. Regourd, et a l .
increase in its porosity. This scheme shows that the C3S , protected till that moment, can be placed in contact with a very alkaline solution susceptible of leading to a large congruent dissolution. The silica, thus liberated, precipitates at the same time in the form of the Tertiary Hydrate.
Since the presentation of these results at the meeting of the French Society of Mineralogy and Crystallography on the Dissolution of Minerals (8 March 1979), qualitatively similar XPS data have been reported by M~n~trier et al (14) and the existence of a superficial hydroxylation step in the dissolution of C3A has been presented by Barret (15). This step would take place in our figure 3, just before the congruent dissolution at t = E. References I. J. Skalny, I. Jawed and H.F.W. Taylor. World Cement Technology, Sept 1978.
183,
2. J.H. Thomassin, M. Regourd, P. Baillif et J.C. Touray. C.R. Acad. Sc. Paris, 288, Serie C. 93, Jan 1979. 3. J.H. Thomassin. Th~se 3~ cycle. Sp~cialit~ G~ochimie, Orl~ans, 28 Sept 1977. 4. R. Petrovic, R.A. Bernerand and M. B. Golhaber. Geoch. Cosmoch. Acta, 40, 537 (1976). 5. L. Colombin, Thesis, Namur (Belgium) (1979). 6. J.H. Thomassin, S.Scherrer, P. Baillif, J.C. Touray and F. Naudin, Bull. Soc. Fr. Miner. Crist. 102, 319 (1979). 7. K. Fuji and W. Kondo, Yogyo. Kyokai Shi, 83, 214 (1975). 8. P. Barret, D. M~n~trierand D. Bertrandie, Rev. Int. Htes. Temp. et R~fract. 14, 127 (1977). 9. M. Regourd, J.H. Thomassin, P. Baillifand J.C. Touray, Meeting on Dissolution of Minerals, Paris 8 March 1979. Society of Mineralogy and Cristallography. Suppl~ment Tome |02, 2, 36 (|979). |0. L.S. Dent Glasser, E.E. Lachowski, K. Mohan and H.F.W. Taylor, Cement Concr. Res. 8, 733 (|978). I|. H.F.W. Taylor. The Chemistry of Cements, Academic Press, London and New York, Vol 2, 372 (1964). |2. M. Tadros, J. Skalny and R. Kalyoncu, J. Amer. Ceram. Soc. 13. J.F. Young, H. Tong and R.L. Berger, ibid. |4. D. M~n~trier,
59, 344 (1976).
60, 193 (1977).
I. Jawed, T.S. Sun and J. Skalny, Cem. Concr. Res. 2, 473
(1979). 15. P. garret. C.R. Acad. Se. Paris,
288,
461 (Jan 1979).