Advertisement

Effect of additives on the performance of Dyckerhoff cement, Class G, submitted to simulated hydrothermal curing

  • Eva Kuzielová
  • Matúš Žemlička
  • Jiří Másilko
  • Martin T. Palou
Article

Abstract

Stability of Dyckerhoff cement Class G partially substituted (15 mass%) by metakaolin (MK), silica fume (SF) and ground granulated blast-furnace slag (BFS) was investigated after 7 days of curing under standard and two different autoclaving conditions. Mercury intrusion porosimetry, X-ray diffraction analysis and combined thermogravimetric–differential scanning calorimetry were used to evaluate pore structure development, compressive strength and their dependence on the type of additives in relation to the particular phase composition. Hydrothermal curing led to the formation of α-C2SH and jaffeite, mostly in the case of referential samples and compositions with addition of slowly reacting BFS. Whilst modest hydrothermal curing (0.6 MPa, 165 °C) favoured formation of α-C2SH, larger amounts of jaffeite were determined after curing at the highest used pressure and temperature (2.0 MPa, 220 °C). Undesired transformation of primary hydration products was prevented especially by addition of highly reactive and very fine SF. Particular composition attained the best pore structure characteristics and compressive strength after curing at 0.6 MPa and 165 °C. Formation of more stable phases with C/S ratio close to 1 was proved by wollastonite formation during DSC analyses. More severe conditions of curing, however, led to the significant deterioration of microstructure and strength of corresponding sample, probably due to the formation of trabzonite, killalaite and zoisite. Considering the values of hydraulic permeability coefficient and compressive strength, replacement of cement by MK improved significantly the properties of cement when compared with the referential as well as with other blended compositions under the mentioned curing conditions.

Keywords

Dyckerhoff cement Hydrothermal curing Silica fume Ground granulated blast-furnace slag Metakaolin 

List of symbols

C

CaO

S

SiO2

A

Al2O3

F

Fe2O3

H

H2O

M

MgO

\(\overline{\text{C}}\)

CO2

\(\overline{\text{S}}\)

SO3

Introduction

Oil well cements, originally developed for the usage in oil and/or gas wells, are sometimes used also in other kinds of wells, such as water wells or thermal recovery wells, where they are submitted to the conditions essentially different from the one they were designed to. Just considering various temperatures and pressures, they affect hydration chemistry and kinetics of cements significantly. Above 100 °C, poorly crystalline/amorphous hydration products convert to the crystalline structures. As it has been proved by many researchers [1, 2, 3], C–S–H (II) possessing high compressive strength and low permeability transforms to much more denser α-C2SH (Ca2(SiO3OH)(OH)) and jaffeite (Ca6(Si2O7)(OH)6, C6S2H3) with low compressive strength and high permeability. According to Barnes and Bensted [4], corresponding strength retrogression is not normally too severe to impair the maintenance of the casings and/or liners in the annular spaces between the borehole walls and the casings/liners. However, the increase in permeability of the hardened cementitious products from around 0.1 md for C–S–H (II) to circa 10–100 md for α-C2SH and jaffeite is sufficient to decrease the durability of the cement placed in the annular space.

Undesirable transformation can be suppressed by additions of silica fume, silica flour, silica sand or fly ash [5, 6]. In such a manner, CaO-to-SiO2 ratio (C/S) is reduced to the value nearly equals 1 or lower and formation of phases, such as tobermorite (idealized formula Ca5(H2Si6O18)·4H2O, C5S6H5), with high compressive strength and low permeability is preferred [7, 8, 9]. Above ~ 150 °C, tobermorite gradually changes into xonotlite (Ca6(Si6O17)(OH)2, C6S6H) and partly to gyrolite (Ca8(Si4O10)3(OH)4·6H2O, C8S12H8), and above ~ 250 °C, truscottite (Ca7(Si4O10)(Si8O19)(OH)4·H2O, C7S12H2) forms from gyrolite and from any residual tobermorite. Although the values of compressive strength and permeability of these phases are declined in comparison with tobermorite, strength retrogression and durability problems are still avoided under normal circumstances [9].

Another way how to improve stability of cementing in geothermal wells is to ensure formation of gehlenite hydrate (Ca2Al2SiO7·8H2O, C2ASH8) and other C–A–S–H phases that can demonstrate even higher stability under hydrothermal conditions. Formation of these phases may be attained by the addition of ground granulated blast-furnace slag.

Among the classes specified by the American Petroleum Institute (API) [10], cements of Classes G and H are the most widely used in the wells at the global scale. Besides the above mentioned, hydrothermal conditions applied to these cement classes may eventuate to some interesting phenomena, such as longer than expected thickening time at around 70–90 °C [11], anomalous thickening activity with silica flour and silica sand [12, 13], S-curve effects during retardation with certain retarders (higher level of retarders lengthens the thickening time in the expected manner at first, then shortens it and later lengthens it again in the usual way) [14] and S-curve effects during compressive strength development [15]. Pozzolanic additives used to stabilize compositions of cements usually totally consume calcium hydroxide. Also only very little ferrite or aluminate phases are present under hydrothermal conditions as they largely finish up in solid solution within the hydrothermal calcium silicate hydrates. In the temperature range 100–200 °C, some hydrogarnet (general formula Ca3Al2(SiO4)3−x (OH)4x ) and strätlingite (also known as gehlenite hydrate) used to be observed [9]. Effective upper temperature limit for usage of Class G and Class H cements under normal downhole conditions is 400 °C. Above this temperature, a disintegration of hardened cement takes place as a consequence of high shrinkage coupled with the decomposition of truscottite and xonotlite as the most stable high-temperature binders formed by these cements.

In comparison with temperature, effects of higher pressures have not been so extensively studied. In general, increases in pressure also accelerate the rate of hydration [9], which results in earlier compressive strength development. Also higher ultimate compressive strengths are observed owing to the high applied pressure.

Present article continues to study the performance of commercial G-oil well cements under hydrothermal conditions simulated in laboratory autoclave. Referring to our previous results [8, 16, 17], pozzolanic and latently hydraulic additives, namely silica fume, metakaolin and ground granulated blast-furnace slag, as well as their combination, were used to allow formation of more thermal stable phases and thus enhance structure and strength characteristics of final materials necessary for durability and satisfactory function of cementing in wells. Development of pore structure, its relation to phase composition and compressive strength are studied by means of mercury intrusion porosimetry (MIP), X-ray diffraction analysis (XRD) and combined thermogravimetric–differential scanning calorimetry (TGA/DSC) technique.

Materials and methods

Samples with composition depicted in Table 1 were prepared from high sulphate-resistant Dyckerhoff cement Class G (Dyckerhoff GmbH, Germany), metakaolin L05 (MK; Mefisto, České lupkové závody a.s., Czech Republic), silica fume (SF; Oravské ferozliatinárske závody, a.s., Slovak Republic) and ground granulated blast-furnace slag (BFS; Kotouč Štramberk, spol. s r.o., Czech Republic). Oxide composition of all the input materials is presented in Table 2.
Table 1

Composition of prepared samples and used water-to-binder ratio (w/b)

Sample

w PC/mass%

w MK/mass%

w SF/mass%

w BFS/mass%

w/b

D0

100

0.23

D1

85

15

0.24

D2

85

15

0.24

D3

85

15

0.24

D4

85

5

5

5

0.24

Table 2

Oxide composition and specific surface of the used Class G Dyckerhoff cement and supplementary cementitious materials

 

Oxide composition/mass%

Specific surface/m2 kg−1

CaO

SiO2

Al2O3

Fe2O3

MgO

SO3

Dyckerhoff

62.73

20.12

4.46

5.05

0.95

2.21

Blaine

326.5 ± 0.1

SF

0.50

97.10

0.21

0.40

BET

15,000

MK

0.24

49.70

42.36

0.79

0.22

0.08

Blaine

2586 ± 38

BFS

36.53

35.76

9.39

0.24

14.0

0.03

Blaine

469.9 ± 0.5

All depicted values of specific surface areas determined using Blaine represent the average of three measurements

Subsequently, referential as well as blended cements were mixed with adequate quantity of water and homogenized pastes were poured into 160 × 40 × 40 mm moulds, in which they were let to hydrate at 105 °C for 1.5 h. Still hot samples were demoulded and quickly put in autoclave (High-Pressure Autoclave Testing Bluhm & Feuerherdt GmbH) where they were exposed for 7 days to hydrothermal curing at 0.6 MPa and 165 °C or at 2.0 MPa and 220 °C. At the same time, samples of identical compositions were stored under the tap water and submitted to standard curing conditions at atmospheric pressure (0.1 MPa) and laboratory temperature of 25 °C for 7 days.

Compressive strengths of hardened composites were determined using WPM VEB-Werkstoffprufmaschinen Leipzig (up to 300 kN). Each displayed value of the compressive strength represents arithmetic mean of 6 experimental measurements.

Mercury intrusion porosimetry was performed on Quantachrome Poremaster 60GT (Quantachrome UK Ltd.). Several pieces taken from different parts of each dried sample with diameter less than 10 mm and total mass max. 2 g were used for the tests. The maximum applied pressure of mercury was 414 MPa, equivalent to a Washburn pore radius of 1.8 nm.

Phase changes taking place in the samples were monitored by TGA/DSC technique (TGA/DSC − 1, STARe software 9.30, Mettler Toledo). The 50.00 (± 0.03) mg of powdered samples was heated in the open platinum crucibles up to 1000 °C at the heating rate of 10 °C min−1 in the atmosphere of synthetic air (purity 5.0).

Crystalline phases in the samples were detected by XRD analysis (Diffractometer system EMPYREAN, PANanalytical, the Netherlands; CuKα radiation, λ = 0.1540598 nm, operating at 40 kV and 30 mA). The percentage of identified crystalline phases was determined by semi-quantitative analysis based on reference intensity ratio (RIR) method using HighScore Plus software (PANanalytical). Principle of this method is described in more details in [17].

Results and discussion

Results of mercury porosimetry

Durability of cementitious materials is usually assessed by means of their permeability that depends on porosity but mainly on pore characteristics, such as pore size distribution, pores orientation, their surface roughness and connectivity that are commonly lumped together and called tortuosity. Despite known and reasonable doubts regarding the suitability of MIP for evaluation of pores characteristics of these materials, numerous studies confirmed that discussed method can serve at least for estimation and comparison of related samples submitted to the same pre-treatment and measurement conditions.

Portion of pores falling according to their diameters within the particular areas of size is depicted in Table 3. As referred pore sizes are interpreted only in the terms of mercury intrusion context, the term “pore sizes” relates the percolation size of pores corresponding to the intruded pressure only. The following classification of pores proposed by Mehta and Monteirio [18] was used: gel pores (< 4.5 nm), mesopores (4.5–50 nm), middle capillary pores (50–102 nm), large capillaries (102–104 nm), macropores (> 104 nm).
Table 3

“Pore size distribution” (portion of pores in the given area of pore diameter) and other pore structure characteristics of samples

Series

Pore size distribution/vol.%

Total porosity/%

Pore tortuosity

Hydraulic permeability coefficient × 10−13/m s−1

< 4.5 nm

4.5–50 nm

50–102 nm

102–104 nm

> 104 nm

D0-25

1.29

40.01

23.21

14.27

21.22

22.20

1.98

8.29

D1-25

0.00

50.26

15.78

11.79

22.17

19.70

2.01

7.34

D2-25

4.50

39.81

13.85

12.01

29.83

20.90

1.99

3.84

D3-25

1.94

38.68

27.36

12.56

19.46

22.58

1.97

7.63

D4-25

4.12

44.48

17.96

13.91

19.53

19.01

2.02

3.09

D0-165

0.25

44.59

16.43

16.51

22.22

26.43

1.93

13.92

D1-165

1.11

77.8

1.50

4.82

14.77

29.88

1.89

2.77

D2-165

14.52

59.26

0.82

2.34

23.06

21.06

1.99

0.45

D3-165

0.89

41.96

20.98

17.94

18.23

27.92

1.91

10.28

D4-165

3.50

75.25

1.01

2.84

17.40

31.58

1.87

1.36

D0-220

0.00

6.06

8.82

72.08

13.04

32.25

1.87

327.18

D1-220

0.01

12.24

11.38

63.63

12.74

35.62

1.83

121.08

D2-220

0.00

4.83

7.59

72.61

14.97

39.63

1.78

484.71

D3-220

0.00

7.97

8.89

70.95

12.19

34.84

1.84

259.62

D4-220

0.00

8.42

8.22

64.08

19.28

40.58

1.77

267.17

Majority of pores in 7-day samples belonged to mesopores. Middle capillary pores had the second large abundance in the case of referential sample (D0-25) and sample with BFS addition (D3-25), whilst macropores were the second most numerous in the rest of the samples.

Partial replacement of cement by very fine SF (D2-25) resulted in the highest portion of gel pores. Metakaolin addition (D1-25) led to the highest percentage of mesopores. The synergic effect of both highly reactive pozzolans was manifested by advantageous pore size distribution also in the case of quaternary D4-25. Moreover, samples containing MK and SF demonstrated the most convenient values of total porosity, tortuosity and relating hydraulic permeability coefficient. Slowly reacting BFS together with dilution effect in D3-25 resulted in the highest portion of pores above 50 nm and correspondingly in the highest value of total porosity.

Pore size distribution of samples after softer hydrothermal curing (0.6 MPa, 165 °C) confirmed relatively good predisposition of the used cements for the use under investigated hydrothermal conditions. Although small decrease in gel pores portion in the case of referential and BFS containing samples was documented, fraction of pores with size below 50 nm increased in general and that of pores larger than 50 nm decreased when compared with standardly cured samples. Whereas total porosity and tortuosity of all the prepared samples increased, hydraulic permeability coefficient of the samples with MK and SF additions decreased.

Except for D1 composition prepared with addition of MK, gel pores disappeared in all the rest samples submitted to more severe conditions of curing. Also the portion of mesopores decreased rapidly, whilst portion of large capillaries increased for more than 4 times. Surprisingly, unfavourable changes in pore structure parameters were the most pronounced in the sample blended with SF. Despite the significant changes also in the pore structure of composition D1, this sample demonstrated the most convenient pore size distribution and the lowest hydraulic permeability coefficient.

Thermal analysis

DTG curves of samples cured under normal conditions (0.1 MPa, 25 °C) show characteristics of standard cement slurries (Fig. 1). Two peaks lie up to the temperature of 200 °C, followed by small one between approximately 300–400 °C. Whilst the first one between 70 and 130 °C corresponds mostly with the release of water due to the decomposition of nearly amorphous hydrates (mainly C–S–H) [19, 20], the second one can be attributed to the decomposition of gehlenite hydrate and the third one to the hydrogarnet series [21]. Then, the decomposition of portlandite (Ca(OH)2, CH) can be documented by the presence of effect at about 450 °C. Peaks at higher temperatures are caused by the decomposition of different carbonates, such as calcite and other polymorph modifications of calcium carbonate [22].
Fig. 1

DTG curves of samples after 7 days of curing under standard conditions

Portlandite as one from the primary hydration products is consumed in reactions with the used pozzolans and latent hydraulic BFS leading to the formation of other binder phases. As a result, intensity of peak belonging to portlandite decreases, whereas those of C–S–H and C–A–S–H phases increase in order reflecting the well-known reactivity of the used additives. The same is clearly demonstrated by corresponding mass losses depicted in Table 4.
Table 4

Mass loss of 7-day samples in different temperature intervals corresponding with particular hydration products and their total mass loss

Sample

D0

D1

D2

D3

D4

Temperature interval/°C

Mass loss/%

25 °C

 0–400 (C–S–H, C–A–S–H)

8.59

9.97

8.97

9.12

9.24

 400–480 (CH)

2.49

1.82

1.55

2.32

1.62

 490–560

0.33

0.42

0.30

0.34

0.33

 560–1000 (carbonates)

2.74

2.36

2.90

2.84

2.76

 Total mass loss/%

14.20

14.64

13.75

14.67

14.00

165 °C

 0–400 (C–S–H, C–A–S–H)

5.08

7.38

7.96

5.39

6.54

 400–480 (CH, α-C2SH)

4.44

2.32

1.11

4.34

1.56

 490–560 (jaffeite)

1.00

1.22

0.62

1.08

0.98

 560–1000 (carbonates)

2.45

1.88

1.64

2.49

1.75

 Total mass loss/%

13.20

13.00

11.44

13.55

10.98

220 °C

 0–400 (C–S–H, C–A–S–H)

2.76

4.43

2.79

3.17

3.14

 400–480 (CH, α-C2SH)

2.79

2.00

0.77

1.79

1.56

 490–560 (jaffeite)

1.56

1.20

0.89

1.44

1.05

 560–1000 (carbonates)

2.58

1.61

2.59

3.03

1.74

 Total mass loss/%

9.81

9.44

7.14

9.58

7.66

Higher curing pressures and temperatures changed the feature of DTG curves belonging to autoclaved samples significantly (Figs. 2, 3). After curing at 0.6 MPa and 165 °C, evaporation of residual acetone, used to stop hydration before analyses, became more evident as previously observed well-developed and sharp peaks belonging to C–S–H and C–A–S–H phases disappeared and instead of them some wide-spreading peak between 80 and 250 °C appeared. In the case of samples D1-165 and D4-165, also a broad effect with maximum at approximately 400 °C owing to the decomposition of some new aluminium phases can be visible. DTG curves of D0-165 and D3-165 showed in the same temperature area only a shoulder on the strong following peak with maximum at 465 °C. Position of this peak is shifted in comparison with that of portlandite for 14 and 12 °C, respectively. Although the explicit evidence of α-C2SH by means of thermal analysis is not possible since it decomposes in the same temperature interval as portlandite, the shift of maximum together with the increase in its intensity in comparison with standardly cured samples can point out the presence of α-C2SH. The rest of the samples demonstrated in this area only a small effect shifted in less than 6 °C when compared with that after standard curing. A new peak, appearing in the temperature interval 490–560 °C, can be attributed to jaffeite [17]. In comparison with standardly cured samples, the mass losses increased in the discussed area. The smallest mass loss was recorded for the sample prepared with SF addition, D2-165, which demonstrated, as the only sample, even the decrease in mass loss between 400 and 480 °C when compared with corresponding non-autoclaved sample.
Fig. 2

DTG curves of samples after 7 days of curing at 0.6 MPa and 165 °C

Fig. 3

DTG curves of samples after 7 days of curing at 2.0 MPa and 220 °C

After curing at 2.0 MPa and 220 °C, the following changes can be observed on DTG curves. Former C–S–H and C–A–S–H peaks were no longer detected, and further decrease in mass loss in the temperature interval between 0 and 400 °C was recorded. Except D2-220, the decrease in mass loss between 400 and 480 °C confirmed smaller amount of corresponding phases than in the samples cured at 0.6 MPa and 165 °C. On the contrary, intensity of peak belonging to jaffeite increased. The only D1-220, initially containing MK, showed the decrease in mass loss in comparison with the sample of the same composition cured at 0.6 MPa and 165 °C. Determined mass loss decreased in the order: D0, D3, D1, D4 and D2, similarly to that in the previously discussed temperature interval. D2-220 demonstrated the lowest mass loss in the both intervals, the same as in the case of samples autoclaved at lower pressure and temperature. Particular substitution of cement by silica fume thus seems to form the most resistant composition against transformations in the case of both hydrothermal curing.

The transformation of α-C2SH to jaffeite is clearly demonstrated in Figs. 4 and 5. Relevant figures compare DTG curves of D0 (referential) and D3 (prepared with 15 mass% of BFS) samples, the two compositions demonstrating the most significant undesired influences of hydrothermal curing, after all the used curing regimes. It can be seen that whilst the softer autoclave treatment caused the shift of the peak maximum between 400 and 480 °C to higher temperatures, curing under more severe conditions resulted in its movement back to the initial position as well as to the repeated decrease in its intensity. At the same time, next peak attributed to jaffeite increased.
Fig. 4

DTG curves of referential samples cured under different conditions for 7 days

Fig. 5

DTG curves of samples prepared with partial substation of cement by BFS cured under different conditions for 7 days

The composition of D2 samples containing 15 mass% of silica fume demonstrated the lowest total mass loss among the prepared samples regardless of the used curing regime (Table 4). Also the mass loss up to the temperature of 560 °C is the lowest for the same composition. As the majority of hydrated phases is already decomposed at this temperature, it can be supposed that lower amount of water was incorporated in the structure of particular samples. Escalante-Garcia [23] and other authors [24] postulated that the CaO content in the pozzolans plays an important role in the binding of additional water and they explained this by the presence of cementitious Ca-bearing phases that hydrate directly taking in additional water. This can be one from the reasons for lower amount of bonded (“non-evaporable”) water in the samples of silica fume containing D2 series.

Comparing the different curing conditions, the total mass losses of all the samples autoclaved at 0.6 MPa and 165 °C were lower than values determined for standardly cured samples. Conditions of 2.0 MPa and 220 °C resulted in even more evident inhibition of reactions and values of total mass loss decreased for more than 30% for D0, D1 and D3 and for more than 40% for D2 and D4 when compared with those of standardly cured samples. The lowest difference between the values of total mass loss of standardly and hydrothermally cured samples showed referential D0 composition. Besides the dilution effect, D3 (15 mass% of BFS) showed higher total mass loss than referential as well as other blended samples from the first two series. When curing at 2.0 MPa and 220 °C was performed, D0-220 had the highest total mass loss. It is possible that higher curing pressures and temperatures initially accelerate hydration reactions; however, especially in the case of high reactive SF, quickly formed hydration products cover the unreacted grains and inhibit their other reactions. On the contrary, slowly reacting BFS allows more continual progress of reactions.

The influence of modified C/S ratio can be clearly demonstrated by means of exothermic effects observed at higher temperatures on DSC curves (Figs. 68). Standard curing conditions and composition of sample D2 (15 mass% SF), enriched by SiO2, allowed the formation of hydrated phases with C/S ~ 1 [25] that consequently transformed to wollastonite (CaSiO3, CS) [16] during DSC measurement. Besides particular well-developed effect, another exothermic peak appeared on DSC curves of samples D1-165 and D4-165. Existence of this peak spreading at higher temperatures can be attributed to mullite (Al6Si2O13, A3S2) formation and it reflects higher content of aluminous ions mainly from MK addition [26].
Fig. 6

DSC curves of standardly cured samples in the selected temperature interval

Fig. 7

DSC curves of samples cured at 0.6 MPa and 165 °C in the selected temperature interval

Fig. 8

DSC curves of samples cured at 2.0 MPa and 220 °C in the selected temperature interval

Particularly, hydrothermal curing at 2.0 MPa and 220 °C changed the character of samples significantly and the presence of ordinary hydration products was no longer detected by DTG analyses. Observed exothermic effects (Fig. 8) thus arose from the transformation of different phases. Since the used conditions caused the crystalline character of the samples, the occurrence of observed exothermic effects could be considered according to the results of XRD analyses (Sect. 3.3). However, whilst the presence of the second peak can be still attributed to the crystallization of mullite mainly from hibschite (Ca3Al2(SiO4)3−x(OH)4x, x is from 0.2 to 1.5, C–A–S–H) and katoite (Ca3Al2(SiO12)3−x(OH)4x, x is from 1.5 to 3.0, C–A–S–H) transformations, the origin of peak lying at lower temperatures is not so clear and must be the consequence of more complicated high-temperature reactions.

XRD analysis

XRD patterns of the samples as well as the results of semi-quantitative analysis are depicted in Figs. 911 and Table 5, respectively. Determined quantities of particular phases cannot be considered as absolute values and serve us only for approximate comparisons between compositions of samples submitted to more severe hydrothermal curing as not the presence of amorphous phases is expected. Besides this, semi-quantitative analysis of sample D2-220 was not possible due to missing data of trabzonite (Ca4Si3O10·2H2O, C4S3H2) in the used library of patterns. In the case of standardly cured samples and samples submitted to softer autoclaving that contain different amounts of amorphous phases, the results of semi-quantitative analyses can be used only to compare the abundance of crystalline phases within individual samples.
Fig. 9

XRD patterns of standardly cured samples. Abbreviations: A—akermanite (C2MS2), B—brownmillerite (C4AF), C—calcite (C\(\overline{\text{C}}\)), D—dolomite (CM\(\overline{\text{C}}\) 2), E—ettringite (C6A\(\overline{\text{S}}\) 3H32), H—hatrurite (C3S), L—larnite (β-C2S), M—merwinite (MC3S2), Mc—calcium aluminium hydroxide carbonate hydrate (C4A\(\overline{\text{C}}\)H10), P—portlandite (CH), Q—quartz (S), V—vaterite (C\(\overline{\text{C}}\))

Fig. 10

XRD patterns of samples cured at 0.6 MPa and 165 °C. Abbreviations: α—α-dicalcium silicate hydrate (α-C2SH), A—akermanite (C2MS2), B—brownmillerite (C4AF), C—calcite (C\(\overline{\text{C}}\)), D—dolomite (CM\(\overline{\text{C}}\) 2), H—hatrurite (C3S), Hi—hibschite (C–A–S–H), J—jaffeite (C6S2H3), K—katoite (C–A–S–H), L—larnite (β-C2S), M—merwinite (MC3S2), P—portlandite (CH), Q—quartz (S)

Fig. 11

XRD patterns of samples cured at 2.0 MPa and 220 °C. Abbreviations: α—α-dicalcium silicate hydrate (α-C2SH), A—akermanite (C2MS2), B—brownmillerite (C4AF), C—calcite (C\(\overline{\text{C}}\)), D—dolomite (CM\(\overline{\text{C}}\) 2), H—hatrurite (C3S), Hi—hibschite (C–A–S–H), I—hillebrandite (C2SH), J—jaffeite (C6S2H3), K—katoite (C–A–S–H), Ki—killalaite (C–S–H), L—larnite (β-C2S), M—merwinite (MC3S2), P—portlandite (CH), R—calcium silicate hydroxide (C5S2H), Q—quartz (S), S—scawtite (C7S6 \(\overline{\text{C}}\)H2), T—tobermorite (C5S6H5), Tr—trabzonite (C4S3H2), V—vaterite (C\(\overline{\text{C}}\)), Z—zoisite (C–A–S–H)

Table 5

Percentage (mass%) of crystalline phases (CP) determined by semi-quantitative analyses of XRD data

CP

0.1 MPa, 25 °C

0.6 MPa, 165 °C

2.0 MPa, 220 °C

D0

D1

D2

D3

D4

D0

D1

D2

D3

D4

D0

D1

D2

D3

D4

H

39

36

41

32

38

18

18

50

10

41

18

18

+

10

18

L

20

20

21

19

19

13

15

19

16

17

14

15

+

13

17

B

8

10

10

9

10

8

7

9

6

6

5

6

+

4

6

Q

 

2

  

2

    

2

    

1

M

   

7

6

   

3

4

   

3

5

A

   

2

2

   

2

2

   

2

2

D

3

3

3

2

3

3

3

3

2

4

2

4

+

3

5

C

4

5

2

7

3

3

4

4

5

4

4

3

+

8

3

V

4

1

        

3

3

   

P

20

17

21

19

15

23

  

19

 

15

  

3

 

E

2

4

2

2

2

          

T

           

1

   

α

     

22

  

30

    

6

 

I

             

6

 

J

     

6

8

 

3

3

20

8

+

19

6

R

          

14

  

17

 

Hi

      

30

8

 

10

 

27

+

 

21

K

     

4

15

7

4

7

5

15

+

6

6

Tr

            

+

  

Ki

            

+

  

Z

            

+

  

Mc

 

2

 

1

           

S

              

10

Semi-quantitative analyses of phases detected in D2-220 were not possible due to the absence of trabzonite data in the respective databases

α, α-dicalcium silicate hydrate (α-C2SH); A, akermanite (C2MS2); B, brownmillerite (C4AF); C, calcite (C\(\overline{\text{C}}\)); D, dolomite (CM\(\overline{\text{C}}\) 2); E, ettringite (C6A\(\overline{\text{S}}\) 3H32); H, hatrurite (C3S); Hi, hibschite (C–A–S–H); I, hillebrandite (C2SH); J, jaffeite (C6S2H3); K, katoite (C–A–S–H); Ki, killalaite (C–S–H); L, larnite (β-C2S); M, merwinite (MC3S2); Mc, calcium aluminium hydroxide carbonate hydrate (C4A\(\overline{\text{C}}\)H10); P, portlandite (CH); R, calcium silicate hydroxide (C5S2H); Q, quartz (S); S, scawtite (C7S6 \(\overline{\text{C}}\)H2); T, tobermorite (C5S6H5); Tr, trabzonite (C4S3H2); V, vaterite (C\(\overline{\text{C}}\)); Z, zoisite (C–A–S–H)

The amount of unreacted clinker minerals, C3S, C2S and C4AF, decreased when compared the corresponding peaks intensities of standardly cured samples with those submitted to autoclaving. In the case of D2 (15 mass% of SF) and quaternary D4 compositions, the amount of clinker minerals did not decrease considerably until the curing at 2.0 MPa (220 °C) was applied. Regarding the composition D2, intensities of corresponding peaks even point out the higher amount of unreacted minerals in the sample cured under softer hydrothermal conditions than in standardly cured sample.

All the prepared samples contained two carbonates: dolomite (CaMg(CO3)2, CM\(\overline{\text{C}}\) 2) and calcite (CaCO3, C\(\overline{\text{C}}\)). In addition to calcite, also calcium carbonate in the vaterite polymorph was detected. The presence of vaterite was proved by XRD only in referential D0 and D1 (15 mass% of MK) compositions cured under standard conditions as well as at 2.0 MPa (220 °C).

Curing at 0.1 MPa (25 °C) led in all the samples to the formation of portlandite and ettringite (Ca6Al2(SO4)3(OH)12·26H2O, C6A\(\overline{\text{S}}\) 3H32). The amount of ettringite reflected the content of Al3+ ions; correspondingly, it was the highest in D1-25 prepared with MK. Besides the mentioned, C4A\(\overline{\text{C}}\)H10 (Ca4Al2(OH)12(CO3)·5H2O) was detected in metakaolin containing samples D1-25 and D4-25.

Except referential D0 and D3 (15 mass% of BFS) compositions, portlandite was not further present on XRD patterns of samples submitted to hydrothermal curing. Also ettringite was not detected in all the autoclaved samples. Instead of them, α-C2SH was proved in D0-165 and D3-165, which confirmed the assumptions on the basis of thermal analysis. Jaffeite was found in the samples after both of the used autoclaving regimes except D2-165. In addition to the mentioned, some new and in our previous work [17] not observed crystalline phases were detected. The presence of katoite was proved in all the autoclaved samples, whilst hibschite only in D1, D2 and D4 compositions. Reinhardbraunsite (Ca5(SiO4)2(OH)2, C5S2H) was detected in D0-220 and D3-220, accompanying in D3-220 also with C2SH in the form of hillebrandite (Ca2SiO3(OH)2, C2SH). The pressure of 2.0 MPa and the temperature of 220 °C resulted in the SF containing composition D2 to the formation of trabzonite, killalaite (Ca3.2[H0.6Si2O7](OH), C–S–H) and zoisite (Ca2Al3(Si2O7)(SiO4)O(OH), C–A–S–H). Scawtite (Ca7(Si3O9)2(CO3)·2H2O, C7S6 \(\overline{\text{C}}\)H2) occurred only in quaternary D4-220 and tobermorite in metakaolin containing D1-220, respectively.

In the light of the above-mentioned results, the following explanations of thermal analysis curves are possible. In the case of referential D0-165 and BFS containing D3-165, the shift of maximum in the area of portlandite decomposition to higher temperatures (~ 465 °C) can be really attributed to the simultaneous presence of α-C2SH and portlandite. Amount of portlandite present in the samples D1-165, D2-165 and D4-165, cured under the same conditions, and causing the small effect at little bit lower temperature was not sufficient to be detected by XRD. On the contrary, the following effect with maximum at higher temperature (~ 530 °C) when compared with D0-165 and D3-165 (~ 509 °C) could be caused by the presence of jaffeite and two calcium aluminium silicate hydroxides, hibschite and katoite. Whilst hibschite was proved only in D1-165, D2-165 and D4-165, katoite was detected in all these samples, the same as after more severe hydrothermal curing. Curing at 2.0 MPa (220 °C) did not change the intensities of particular peaks significantly, whereas those of jaffeite increased and the presence of this phase was confirmed also in the SF containing composition D2. The highest intensities of hibschite and katoite peaks were determined in the case of MK containing D1 followed by quaternary D4 in accordance with the amount of Al3+ ions.

Both the mentioned end members of hydrogarnet family were reported to form under hydrothermal conditions also in other studies [27, 28, 29], and they are partially responsible for the strength deterioration in the autoclaved samples (Sect. 3.4) [30]. The authors of [31] dealing with CaO–SiO2–H2O and CaO–SiO2–Al(OH)3–H2O systems proved that the presence of Al3+ ion accelerates the crystal growth of tobermorite to some extent. However, in the case of excessive addition of Al resource, Al3+ takes the place of Si4+ and impedes the formation of tobermorite crystal structure, which results in the change of tobermorite into hibschite.

DTG curves and XRD patterns of compositions D0 and D3 showed similarities after both the autoclaved regimes. Contrary to the corresponding samples cured under modest hydrothermal conditions, in the case of referential D0-220 almost all the α-C2SH (or all) transformed to jaffeite, whilst sample D3-220 showed some decreased amount of α-C2SH together with similar hillebrandite. According to [32], hillebrandite forms through reversible transformation of dellaite (Ca6Si3O11(OH)2, C6S3H) which can form from α-C2SH.

Mechanical properties

Measured values of compressive strength (CS) are depicted in Table 6. The highest value of CS among the standardly cured samples showed D1, prepared with addition of MK, followed by D3 and D4. Surprisingly, sample prepared with addition of SF (D2), as very reactive pozzolan, demonstrated the lowest value of CS among the blended samples. This could be explained by suppressed reactions leading to the lower amount of hydration products (Sects. 3.2, 3.3). The lowest CS after standard curing was determined for referential sample, which confirmed that the dilution effect caused by partial substitution of cement by supplementary cementitious materials was overcome by their reactivity.
Table 6

Compressive strengths (CS) of prepared samples after 7 days of curing

Curing

0.1 MPa, 25 °C

0.6 MPa, 165 °C

2.0 MPa, 220 °C

Samples

CS/MPa

D0

58.8 ± 12.1

47.4 ± 2.2

25.4 ± 3.6

D1

73.3 ± 3.0

63.0 ± 10.2

28.1 ± 3.3

D2

64.5 ± 4.0

95.8 ± 5.2

13.9 ± 1.2

D3

72.1 ± 8.7

39.0 ± 4.3

22.0 ± 1.6

D4

65.0 ± 4.4

57.6 ± 4.1

11.8 ± 1.0

Except D2-165, CS of samples cured at 0.6 MPa and 165 °C decreased significantly. On the contrary, CS of the mentioned D2-165 increased for almost 50% which is in agreement with the previously discussed results of other analyses. Autoclave treatment had the most disadvantageous effect on CS of D3-165 followed by D0-165, D1-165 and D4-165 compositions. The pressure and temperature of 2.0 MPa and 220 °C, respectively, had even worse influence on CS of all the samples. When compared with modest autoclave treatment, the bigger decrease in CS was measured for D2-220. Since the amounts of α-C2SH and jaffeite determined by thermal analysis were the lowest among the corresponding samples (α-C2SH was not detected by XRD), destructive decrease in CS, resulted also from MIP analysis, could be explained by formation of trabzonite, killalaite and zoisite that were not proved in other samples. Particularly, trabzonite appeared in D2-220 as one from the main crystalline phases (Fig. 11).

Conclusions

The effect of partial substitution (15 mass%) of Dyckerhoff cement Class G by three supplementary cementitious materials, namely metakaolin, silica fume and ground granulated blast-furnace slag, was investigated in order to improve its stability under hydrothermal conditions simulated in laboratory autoclave. The following findings can be concluded according to the analysed results from MIP, TGA/DSC and XRD.
  1. 1.

    Both the overall pore structure characteristics and the phase composition of the cement paste matrix affected the durability influencing properties of final materials (permeability and strength).

     
  2. 2.

    Silica fume was proved as the most promising replacement material in the case of curing at 0.6 MPa and 165 °C. The most convenient pore structure characteristics and the highest compressive strength value resulted from its finesse, high pozzolanic reactivity and resistance of developed phases against undesired transformations of primary hydration products into α-C2SH and jaffeite. Formation of phases with C/S ratio ~ 1 was clearly demonstrated by wollastonite formation during DSC analyses.

     
  3. 3.

    Although the same composition submitted to the curing at 2.0 MPa and 220 °C still demonstrated the lowest amounts of α-C2SH and jaffeite, pore structure of the particular sample underwent unfavourable changes leading to the significant decrease of strength. According to the results of XRD analyses, determined strength retrogression could be explained by the formation of trabzonite, killalaite and zoisite.

     
  4. 4.

    Despite the formation of phases from hydrogarnet series, composition prepared with the addition of metakaolin seemed to be the most suitable for the use under more severe hydrothermal conditions.

     
  5. 5.

    Referential composition as well as the sample initially containing slowly reacting ground granulated blast-furnace slag showed the strongest negative impact of high-temperature transformations under softer hydrothermal curing.

     
  6. 6.

    On the contrary, more severe conditions of autoclaving inhibited reactions especially in the samples with silica fume and metakaolin additions, probably due to the formation of some incrustation made up of the quickly formed hydration products around the unreacted grains and inhibiting reactions in later times.

     
  7. 7.

    Modest hydrothermal curing preferred the formation of α-C2SH, whereas higher amounts of jaffeite were determined after curing at 2.0 MPa and 220 °C.

     

Notes

Acknowledgements

This work was supported by courtesy of APVV-15-0631, Slovak Grant Agency VEGA No. 2/0097/17 and by Project Sustainability and Development REG LO1211 addressed to the Materials Research Centre at FCH VUT.

References

  1. 1.
    Palou MT, Živica V, Ifka T, Boháč M, Zmrzlý M. Effect of hydrothermal curing on early hydration of G-Oil well cement. J Therm Anal Calorim. 2014;.  https://doi.org/10.1007/s10973-013-3511-7.CrossRefGoogle Scholar
  2. 2.
    Taylor HFW. Cement chemistry. 2nd ed. London: Telford Services Ltd.; 1997.CrossRefGoogle Scholar
  3. 3.
    Arabi N, Jauberthie R, Chelghoum N, Molez L. Formation of C–S–H in calcium hydroxide–blast furnace slag–quartz–water system in autoclaving conditions. Adv Cem Res. 2015;.  https://doi.org/10.1680/adcr.13.00069.CrossRefGoogle Scholar
  4. 4.
    Bensted J, Barnes P. Structure and performance of cements. 2nd ed. London: Spon Press; 2002.Google Scholar
  5. 5.
    Hope BB. Autoclaved concrete containing flyash. Cem Concr Res. 1981;.  https://doi.org/10.1016/0008-8846(81)90064-8.CrossRefGoogle Scholar
  6. 6.
    Morsy MS, Alsayed SH, Aqel M. Effect of elevated temperature on mechanical properties and microstructure of silica flour concrete. Int J Civ Environ Eng. 2010;10:1–5.Google Scholar
  7. 7.
    Palou MT, Kuzielová E, Novotný R, Šoukal F, Žemlička M. Blended cements consisting of Portland cement–slag–silica fume–metakaolin system. J Therm Anal Calorim. 2016;.  https://doi.org/10.1007/s10973-016-5399-5.CrossRefGoogle Scholar
  8. 8.
    Kuzielová E, Žemlička M, Bartoničková E, Palou MT. The correlation between porosity and mechanical properties of multicomponent systems consisting of Portland cement–slag–silica fume–metakaolin. Constr Build Mater. 2017;.  https://doi.org/10.1016/j.conbuildmat.2016.12.105.CrossRefGoogle Scholar
  9. 9.
    Bensted J. Special cements. In: Hewlett PC, editor. Lea’s chemistry of cement and concrete. Amsterdam: Elsevier Ltd.; 1998. p. 783–840.CrossRefGoogle Scholar
  10. 10.
    Lyons WC, Plisga GJ. Standard handbook of petroleum and natural gas engineering. 2nd ed. Oxford: Elsevier Ltd.; 2005.Google Scholar
  11. 11.
    Bensted J. Retardation of cement slurries to 250 F. Petrol Eng Int. 1991;.  https://doi.org/10.2118/23073-MS.CrossRefGoogle Scholar
  12. 12.
    Bensted J. Thickening behaviour of oilwell cement slurries with silica flour and silica sand additions. Chem Ind Lond. 1992;18:702–3.Google Scholar
  13. 13.
    Bensted J. Difficulties in specifying free lime limits for oilwell cements. In: API task group on eastern hemisphere cementing, Wiesbaden; 1995. p. 6.Google Scholar
  14. 14.
    Bensted J. Use S-curve effect in oilwell cement thickening. In: Proceedings of the 15th annual international conference on cement microscopy. Dallas: ICMA; 1993. p. 51–71.Google Scholar
  15. 15.
    Bensted J. S-curve effect in oilwell cement compressive strength development under hydrothermal conditions. Cem Concr Res. 1995;.  https://doi.org/10.1016/0008-8846(95)00002-X.CrossRefGoogle Scholar
  16. 16.
    Palou MT, Šoukal F, Boháč M, Šiler P, Ifka T, Živica V. Performance of G-Oil Well cement exposed to elevated hydrothermal curing conditions. J Therm Anal Calorim. 2014;.  https://doi.org/10.1007/s10973-014-3917-x.CrossRefGoogle Scholar
  17. 17.
    Kuzielová E, Žemlička M, Másilko J, Palou MT. Pore structure development of blended G-oil well cement submitted to hydrothermal curing conditions. Geothermics. 2017;.  https://doi.org/10.1016/j.geothermics.2017.03.001.CrossRefGoogle Scholar
  18. 18.
    Mehta PK, Monterio PJM. Concrete: microstructure, properties and materials. 3rd ed. New York: McGraw-Hill; 2006.Google Scholar
  19. 19.
    Fares H, Remond S, Noumowe A, Cousture A. High temperature behavior of self-consolidating concrete microstructure and physicochemical properties. Cem Concr Res. 2010;.  https://doi.org/10.1016/j.cemconres.2009.10.006.CrossRefGoogle Scholar
  20. 20.
    Ibrahim IA, ElSersy HH, Abadir MF. The use of thermal analysis in the approximate determination of the cement content in concrete. J Therm Anal Calorim. 2004;.  https://doi.org/10.1023/B:JTAN.0000032255.58397.4b.CrossRefGoogle Scholar
  21. 21.
    Silva de Souzab LM, Fairbairn EMR, Filho RDT, Cordeiro GC. Influence of initial CaO/SiO2 ratio on the hydration of rice husk ash-Ca(OH)2 and sugar cane bagasse ash-Ca(OH)2 pastes. Quim Nova. 2014;.  https://doi.org/10.5935/0100-4042.20140258.CrossRefGoogle Scholar
  22. 22.
    Matsushita F, Aono Y, Shibata S. Carbonation degree of autoclaved aerated concrete. Cem Concr Res. 2000;.  https://doi.org/10.1016/S0008-8846(00)00424-5.CrossRefGoogle Scholar
  23. 23.
    Escalante-Garcia JI. Nonevaporable water from neat OPC and replacement materials in composite cements hydrated at different temperatures. Cem Concr Res. 2003;.  https://doi.org/10.1016/S0008-8846(03)00208-4.CrossRefGoogle Scholar
  24. 24.
    Yazıcı H, Deniz E, Bülent Baradan B. The effect of autoclave pressure, temperature and duration time on mechanical properties of reactive powder concrete. Constr Build Mater. 2013;.  https://doi.org/10.1016/j.conbuildmat.2013.01.003.CrossRefGoogle Scholar
  25. 25.
    Kuliffayová M, Krajči Ľ, Janotka I, Šmatko V. Thermal behaviour and characterization of cement composites with burnt kaolin sand. J Therm Anal Calorim. 2012;.  https://doi.org/10.1007/s10973-011-1964-0.CrossRefGoogle Scholar
  26. 26.
    Habert G, Choupay N, Montel JM, Guillaume D, Escadeillas G. Effects of the secondary minerals of the pozzolans on their Pozzolanic activity. Cem Concr Res. 2008;.  https://doi.org/10.1016/j.cemconres.2008.02.005.CrossRefGoogle Scholar
  27. 27.
    Klimesch DS, Ray A. Effects of quartz particle size and kaolin on hydrogarnet formation during autoclaving. Cem Concr Res. 1998;.  https://doi.org/10.1016/S0008-8846(98)00111-2.CrossRefGoogle Scholar
  28. 28.
    Ríos CA, Williams CD, Fullen MA. Hydrothermal synthesis of hydrogarnet and tobermorite at 175 C from kaolinite and metakaolinite in the CaO–Al2O3–SiO2–H2O system: a comparative study. Appl Clay Sci. 2009;.  https://doi.org/10.1016/j.clay.2008.09.014.CrossRefGoogle Scholar
  29. 29.
    Saoût GL, Lécolier É, Rivereau A, Zanni H. Study of oilwell cements by solid-state NMR. CR Chim. 2004;.  https://doi.org/10.1016/j.crci.2003.10.018.CrossRefGoogle Scholar
  30. 30.
    Jing Z, Jin F, Hashida T, Yamasaki N, Ishida H. Hydrothermal solidification of blast furnace slag by formation of tobermorite. J Mater Sci. 2007;.  https://doi.org/10.1007/s10853-007-1726-3.CrossRefGoogle Scholar
  31. 31.
    Liu FM, Chen DP, Ni W, Cao ZY. Effect of Al3+ on tobermorite crystallinity. J Univ Sci Technol B. 2000;7(2):79–81.Google Scholar
  32. 32.
    Tantawy MA, Shatat MR, El-Roudi AM, Taher MA, Abd-El-Hamed M. Low temperature synthesis of belite cement based on silica fume and lime. Int Sch Res Notices. 2014;.  https://doi.org/10.1155/2014/873215.CrossRefPubMedPubMedCentralGoogle Scholar

Copyright information

© Akadémiai Kiadó, Budapest, Hungary 2017

Authors and Affiliations

  1. 1.Institute of Construction and ArchitectureSlovak Academy of SciencesBratislavaSlovak Republic
  2. 2.Faculty of Chemical and Food TechnologySlovak University of TechnologyBratislavaSlovak Republic
  3. 3.Materials Research Centre, Faculty of ChemistryBrno University of TechnologyBrnoCzech Republic

Personalised recommendations