Introduction
Uniform dispersion of multi walled carbon nanotubes (MWCNT) in a cement matrix is one of the key factors to maximize their effect on the matrix properties [
1]. MWCNT are usually dispersed in the mixing water of a cement/MWCNT composite using sonication as dispersion method. Nevertheless, the effectiveness of sonication has been found to be reversible in time due to reagglomeration phenomena [
2]. Dispersing agents such as surfactants are used to improve the stability of MWCNT dispersions in water. Surfactant-to-MWCNT ratios up to 8:1 by mass have been identified as an effective dispersing agent for MWCNT [
1]. However, it is known that surfactant molecules adsorb onto the surface of anhydrous cement grains [
3], and in high concentrations, lignosulfonates or polyacrilates among others, hamper the hydration reaction of cement due to steric hindrance between the adsorbed molecules of surfactant, retarding the hydration reaction during the first hours [
4]. This work studies how the retardation effect of a commercial dispersion of MWCNT with an anionic surfactant as a dispersing agent modifies the kinetics and amount of hydrates produced in an oil well cementing paste at different temperatures.
Materials and methods
The materials used in the experimental campaign were class G cement and commercial MWCNT pre-dispersed in water with a solids concentration of 3 % by mass. Chemical composition of cement and MWCNT is presented in Table 1. MWCNT were acquired already pre-dispersed in water; an anionic surfactant was used as dispersing agent. The specific type of surfactant was not disclosed by the manufacturer, but knowing that is of anionic nature guarantees that in its structure the surfactant molecule has a hydrophobic end and a hydrophilic end with a negative charge. This type of structure includes a whole range of molecules such as sulfonates, phosphates and carboxylates that are widely used in the cement industry. MWCNT had an average diameter of 9.5 nm and an average length of 1.5 µm according to the manufacturer. MWCNT were characterized using RAMAN, FTIR and energy dispersive X-ray (EDX) spectroscopies, and transmission electron microscopy (TEM). RAMAN spectrum was acquired using a He/Ne laser with a 632.85 nm wavelength and a power output of 3.0 mW. FTIR spectrum was acquired from 400 to 4000 cm-1 with a resolution of 2 cm-1, using a sample of 0.1 mg of MWCNT and 100 mg of KBr. TEM images were obtained from a 5 µl aliquot of the dispersion that was dropped in a #300 copper mesh and left to dry at room temperature.
A reference sample of plain cement was prepared using a water to cement ratio of 0.45, which was hand mixed until a homogeneous paste was obtained. Cement/MWCNT pastes were prepared combining the MWCNT dispersion with the mixing water, maintaining the same water to cement ratio (w/c: 0.45), then adding cement and hand mixing. A portion of 5 g of each paste was sealed in a glass ampoule and used immediately for isothermal calorimetry at 23°C and 65°C. Heat flow and cumulative heat curves by mass of cement were obtained with a TAM Air isothermal calorimeter, using water as reference material. A second portion of the samples was cured in 50 ml plastic airtight containers at room temperature or in a 65°C water bath. When testing age was reached the hydration was stopped using a freeze dryer. Thermogravimetric analyses (TGA) were carried out using a platinum crucible in a N2 inert atmosphere with a gas flow of 100 ml/min, measured up to 900°C with a heating rate of 10°C/min.
Results and discussion
Characterization of the MWCNT
TEM imaging of the MWCNT is presented in Fig. 1. It can be seen that the length of the MWCNT is in the scale of the micrometers and its width in the scale of the nanometers. Chemical composition obtained by EDX is presented in Table 1. The results were corrected by loss on ignition, which corresponded to carbon. Al2O3 was detected in the MWCNT in a proportion that indicates its use as a catalyst during their production. RAMAN spectrum is presented in Fig. 2. D and G absorption bands (ID/IG = 2.04) characteristic of highly graphitized carbon were identified, additionally the radial breathing mode (150 to 300 cm-1) confirmed that the nanotubes were multi wall. FTIR spectrum of the dispersion is presented in Fig. 3. Absorption bands characteristic of C-C bonds from the MWCNT structure were identified; additionally, a group of C-H, -OH and C-O bonds was identified, which correspond to the anionic surfactant used as a dispersing agent. According to the manufacturer the concentration of surfactant was 5.0 % of the mas of dispersion.
Isothermal calorimetry results
Heat flow curves measured at 23°C (Fig. 4(a)) showed that at point 1 the MWCNT dispersion retarded the hydration reaction, prolonging the induction period up to 15 h, and delaying the maximum of the main peak of hydration up to 25 h. This effect was found to be proportional to the amount of MWCNT present in the sample and can be attributed to the surfactant used as a dispersing agent. The amount of surfactant in each sample is proportional to the amount of MWCNT dispersed in water. At point 2, after the surfactant desorbed from the cement grains and the hydration resumed, the heat release from C-S-H and AFt formation at the sulfate depletion peak became enhanced. The enhancement at the C-S-H formation can be attributed to a nucleation effect of the MWCNT [
5], and the enhancement at the AFt formation can attributed to the Al
2O
3 content of the MWCNT dispersion. Additionally, at point 3 it was found that after 50 h of hydration the peak associated with the transformation of AFt into AFm was increased in the samples blended with MWCNT, this as consequence of the higher AFt formation.
Heat flow curves obtained at 65°C (Fig. 4(b)) showed a general acceleration of the hydration reaction due to the effect of temperature, both in the plain cement as in the samples blended with MWCNT. The retardation effect of the surfactant was minimized but still present, indicating that the adsorption/desorption process of the surfactant on the surface of the anhydrous surfaces still occurs, but is accelerated by temperature. This indicates that the elevated down-hole temperature conditions by themselves can partially solve the retardation issue in field operations.
Cumulative heat curves measured at 23°C (Fig. 5(a)) showed that even though the hydration reaction was retarded, the total amount of energy released by the samples blended with MWCNT was higher than the one from plain cement. This indicates that when the surfactant’s desorption from the anhydrous surfaces occurs, the hydration reaction is resumed (end of induction period) and accelerated. This can be attributed to a nucleation effect of the MWCNT [
7] or to the extension of the induction period [
6]. After 140 h of hydration the 0.25%MWCNT sample presented a higher cumulative heat value than the 0.50%MWCNT sample. Further testing is required to confirm if this tendency is maintained after longer hydration times and if this phenomenon is dominated by the nucleation effect of the MWCNT or the extension of the induction period.
Cumulative heat curves at 65°C (Fig. 5(b)) showed that samples blended with 0.25%MWCNT and 0.50%MWCNT presented approximately the same amount of total heat release as the plain cement sample during the 140 h of hydration measured, and also reached a heat release plateau quicker than the samples studied at 23°C. This indicates that the acceleration of the hydration reaction caused by the MWCNT nucleation or the increase of induction period at 23°C is not significant enough at 65°C to generate a visible change in the cumulative heat curves.
Thermogravimetric analyses results
TGA were used to confirm the effect of the MWCNT dispersion over the amount of hydration products precipitated during the hydration of the paste. Three points were chosen in each heat flow curve to stop the hydration of the samples: 1) at the end of the induction period, 2) at the end of the main hydration peak, and 3) after the formation of AFm, when all three samples reached a similar heat flow value. This assured that at each point all three samples were in a similar stage of hydration, and helped in the interpretation by decoupling the influence of the dispersant from the influence of the MWCNT. Localization of the points is presented in Fig. 4.
Two typical TG/DTG curves were obtained, one for samples with hydration stopped before the main hydration peak (point 1) and one for samples with hydration stopped after the main hydration peak (points 2 and 3). Figure 6 presents the TG/DTG curves obtained and the decomposition reactions associated with each event of the MWCNT 0.25% sample. The curves obtained for the rest of the samples presented the same decomposition events and are not presented here. Weight fractions of CaCO3 and Ca(OH)2 were calculated. Due to the non-stoichiometric nature of C-S-H and the impossibility of separating the individual mass loss associated with the dehydration of gypsum, C-S-H, AFt, AFm and aluminate calcium hydrates (ACH), the weight fraction of these components was not calculated but presented as a mass loss of Total Combined Water (TCW). The CaCO3 weight fraction calculations were corrected by the initial carbonate content of the cement before hydration, and the Ca(OH)2 weight fraction calculations were corrected by the CaCO3 content. Results obtained are presented in Figs. 7 to 9.
Results from point 1 (Fig. 7) showed an increase in the amount of Ca(OH)2 both at 23°C and 65°C, proportional to the content of MWCNT dispersion and not affected by the temperature. This can be associated with the increase of the induction period caused by the surfactant. The positive slope of the induction period of the different samples presented Fig. 5 indicates that even though the hydration is retarded, it is not completely hampered, and during this extra time additional Ca(OH)2 is produced. Results from points 2 and 3 (Figs. 8 and 9) showed that the addition of the MWCNT dispersion did not have a significant effect over the amount of Ca(OH)2 and TCW, and that these amounts were not modified by temperature. Additionally, in all three points and at both temperatures an increase of the weight fraction of CaCO3 proportional to the MWCNT content was found. All samples were stored in similar conditions; therefore, there is no reason to suspect of a differentiated carbonation due to environmental causes. Further work is required to identify the cause of this differentiated carbonation.
Conclusions
The commercial MWCT dispersion studied retards the hydration reaction of cement due to an adsorption of anionic surfactant onto the surface of the anhydrous cement grains. The retardation effect of the surfactant did not have a negative impact over the amount of hydration products precipitated, neither at 23°C nor at 65°C. This combined with the fact that the retardation was minimized by the elevated down-hole well temperature, and that the MWCNT maintained its integrity due to a lack of chemical affinity with the cement matrix, allows concluding that the commercial MWCNT dispersion presents a good potential to be applied in oil well cement pastes. The specific type of anionic surfactant used by the manufacturer of the MWCNT dispersion was not disclosed; some of the behaviors found in this research could be better explained if the specific type of dispersant molecule was known. Further work is required to characterize the effect of this MWCNT dispersion over the rheology, microstructure and mechanical properties of the oil well cement pastes.
Higher Education Press and Springer-Verlag Berlin Heidelberg
PDF
(602KB)
