Early hydration of white Portland cement in the presence of bismuth ...

Early hydration of white Portland cement in the presence of bismuth oxide Q. Li* and N. J. Coleman Mineral trioxide aggregate (MTA) is a clinical product comprising a mixture of 80 wt-% Portland cement and 20 wt-% bismuth oxide, which is used as a root-filling material in dentistry. The influence of bismuth oxide on the hydration reactions of Portland cements is not well understood. In this study, the impact of 20 wt-% replacement of bismuth oxide on the hydration of white Portland cement was monitored by powder X-ray diffraction (XRD), 29Si and 27Al magic angle spinning nuclear magnetic resonance spectroscopy (MAS NMR) and transmission electron microscopy (TEM). The findings of this research have confirmed that bismuth oxide is an inert additive in white Portland cement, which does not participate in the hydration reactions. Keywords: Mineral trioxide aggregate, Portland cement, Bismuth oxide, Nuclear magnetic resonance spectroscopy, Hydration

This paper is part of a special issue on Non-Conventional Cementitious Bindersrete Science 2011

Introduction Mineral trioxide aggregate (MTA) is a proprietary dental restorative comprising a mixture of Portland cement and bismuth oxide (Bi2O3), which has been used as a root-filling material since 1995.1,2 Mineral trioxide aggregate, as specified in the original patent of Torabinejad and White,3 has been approved by the US Federal Drug Administration for use in surgical endodontic procedures and is produced commercially in two forms as ‘grey’ and ‘tooth-coloured’ ProRoot MTA (Tulsa Dental Products, USA). In both products, the principal constituent is Portland cement blended with 20 wt-% bismuth oxide to confer radiopacity (i.e. to enhance contrast on medical radiographs). Mineral trioxide aggregate is presented as a powder that is mixed manually with supplied sterile water. The setting reactions of ProRoot MTA and Portland cements in the presence of bismuth oxide have been studied by a variety of techniques, including powder Xray diffraction (XRD), scanning electron microscopy, energy dispersive X-ray analysis and X-ray photoelectron spectroscopy.4–8 Despite this increasing body of information, the impact of bismuth oxide on the hydration chemistry of Portland cement remains a subject of dispute. For example, Camilleri reports that bismuth oxide alters the hydration mechanism of white Portland cement (WPC) such that bismuth is substituted for silicon in the calcium silicate hydrate gel (C–S–H) phase.4,5 Conversely, in their recent review, Darvell and Wu strongly contest this claim, although do not offer any evidence to the contrary.2 The principal objective of this study is to investigate the impact of 20 wt-% bismuth oxide on the early hydration School of Science, University of Greenwich, Chatham Maritime, Kent ME4 4TB, UK *Corresponding author, email [email protected]

ß 2013 Institute of Materials, Minerals and Mining Published by Maney on behalf of the Institute Received 30 March 2012; accepted 17 August 2012 DOI 10.1179/1743676112Y.0000000058

chemistry of WPC. Bi2O3 blended and unblended cement paste samples were hydrated for 6, 24 and 168 h before analysis by 29Si and 27Al magic angle spinning nuclear magnetic resonance spectroscopy (MAS NMR), powder XRD and transmission electron microscopy (TEM). A novel method was used to analyse resonances arising from the early Q0(H) and Q1 hydration products in the 29 Si MAS NMR spectra, which are typically obscured by that of residual unreacted alite.

Experimental Materials, preparation and characterisation The WPC used in this study was supplied by Lafarge and is commercially available as ‘Snowcrete’. Its chemical composition and principal constituent phases were provided by the manufacturer and are listed in Table 1. Cement paste specimens were prepared in duplicate by manually mixing WPC with distilled water at a water/cement ratio of 0?375 by mass. The samples were packed into polypropylene tubes, hermetically sealed and cured at 37?5uC (i.e. ‘body temperature’). Specimens WPC-6, WPC-24 and WPC-168 were cured for 6, 24 and 168 h respectively, before hydration was quenched by solvent exchange with propan-2-ol. Samples blended with bismuth oxide, namely, WPCBi-6, WPC-Bi-24 and WPC-Bi-168, were prepared similarly with partial replacement of the WPC by 20 wt-% Bi2O3 (ex. Sigma-Aldrich) at a water/solid ratio of 0?3 (i.e. a water/cement ratio of 0?375 by mass). Powder XRD was performed on all specimens using a ˚ , a step Philips D8 diffractometer with Cu Ka51?5406 A size of 0?019u in the 2h range from 5 to 45u and a measuring time of 141?8 s per step. X-ray diffraction data were analysed using DIFFRAC.EVA software (supplied by Bruker). The TEM images of WPC-6 and WPC-Bi-6 were obtained by dispersing the ground samples in methanol before deposition onto a carbon film grid. Bright field images were acquired using a

Advances in Applied Ceramics

2013

VOL

112

NO

4

207

Li and Coleman

Early hydration of white Portland cement in presence of bismuth oxide

JEOL JEM200CX microscope and a Gata Orius SC200 digital camera. The MAS NMR spectra were collected on a JEOL JNM-ECX 300 MHz spectrometer. Single pulse 27Al MAS NMR spectra were obtained with a pulse delay of 0?5 s, an acquisition time of 0?01024 s and 7000 scans. 1 H-29Si cross-polarisation (CP) MAS NMR spectra were obtained with a contact time of 1023 s, a pulse delay of 5 s, and an acquisition time of 0?0256 s. Scans (68 000, 34 000 and 17 000) were collected for the 1H-29Si CP MAS NMR spectra of the 6, 24 and 168 h specimens respectively. Single pulse 29Si MAS NMR spectra were obtained with a pulse delay of 5 s and an acquisition time of 0?02048 s. The 119 000, 68 000 and 34 000 scans were collected for the 29Si MAS NMR spectra of the 6, 24 and 168 h specimens respectively. 29Si and 27Al chemical shifts were referenced to tetramethylsilane and the aluminium hexaquo ion [Al(H2O)6] respectively. The free induction decay (FID) profiles were processed by Delta software (provided by JEOL) to obtain spectra, which were then analysed and deconvoluted using Igor Pro software. Analysis of early hydration products by

29

Si MAS NMR

29

The use of Si MAS NMR for the characterisation of hydrated systems has been developed by a number of workers over many years. For an accurate interpretation of C-S-H structure, accurate and reproducible curve fitting is imperative. This paper uses a similar approach to that first described by Brough.9 The possible incorporation of aluminium into the C-S-H structure has been accounted for according to the method described in Refs 10-12. However, determination of the structure of C-S-H in the presence of anhydrous cement, necessitates subtraction of the signal arising from the anhydrous material. Such a method was developed by Love13,14 and then extended to the analysis of early age hydration products by Li15; which formed the basis of the approach in this study. In the case of each hydrated cement paste sample, the intensity of the 29Si MAS NMR spectrum of the anhydrous WPC was adjusted such that the Q0 signal of alite was of equal intensity to that of the sample. This WPC background spectrum was then subtracted from that of the sample paste before deconvolution. The chemical shifts of the silicate hydration products of each paste were identified from its 1H-29Si CP MAS NMR spectrum, and this information was used as a first approximation in the deconvolution of the ‘subtracted’ 29 Si MAS NMR spectrum. The application of this method to the analysis of the 29Si MAS NMR spectrum of sample WPC-168 is illustrated in Fig. 1.

1 Example of novel method for analysis of early hydration products by 29Si MAS NMR

Results and discussion Powder XRD Powder XRD data confirm that the principal constituent phases of the anhydrous WPC used in this study are alite (C3S), belite (C2S) and tricalcium aluminate (C3A) (Fig. 2a). Trace quantities of anhydrite (CaSO4) are also present. The formation of ettringite (AFt) and portlandite (CH) is noted in the diffraction patterns of both the Bi2O3 blended and unblended pastes after 6 h of hydration (Fig. 2e and b respectively). In both cement pastes, the reflections of portlandite develop at the expense of those of C3S as hydration proceeds; however, the rate of production of portlandite in the unblended specimens is greater than that of the Bi2O3 blended pastes. The reflections of Bi2O3 persist in the diffraction patterns of the blended pastes throughout the 7 day hydration period, and there is no evidence for the formation of any other bismuth bearing phases (Fig. 2e–g).

Table 1 Composition of white Portland cement (WPC)

208

Major oxide components

Minor oxide components

Major crystalline phases

Formula

Mass/%

Formula

Mass/%

Formula

Mass/%

CaO SiO2 Al2O3 SO3

69.2 25.0 1.76 2.00

MgO P2O5 Fe2O3 SrO

0.49 0.43 0.33 0.14

Ca3SiO5 Ca2SiO4 Ca3Al2O6 Ca2(Al/Fe)O5 CaSO4

65 22 4.1 1.0 y2


2013

VOL

112

NO

4

Li and Coleman

2 Powder XRD data for a anhydrous WPC, b WPC-6, c WPC-24, d WPC-168, e WPC-Bi-6, f WPC-Bi-24 and g WPC-Bi-168 ( : C3S; #: C2S; e: C3A; %: CaSO4; :

AFt; &: CH; 27

N*

¤

: Bi2O3)

Al MAS NMR

The 27Al MAS NMR spectrum of anhydrous WPC is shown in Fig. 3a and comprises a broad resonance at ,80 ppm, which is assigned to tetrahedrally coordinated aluminium ‘guest’ species in the alite and belite phases.16 In both the presence and absence of Bi2O3, the intensity of this signal diminishes significantly within 6 h of hydration and is replaced by two sharp peaks at y14 and y10 ppm arising from octahedrally coordinated aluminium in ettringite and tetracalcium aluminate hydrate (C4AH13) respectively (Fig. 3b and e).17 As hydration proceeds, a further broad resonance develops at y65 ppm in the spectra of both the Bi2O3 blended and unblended pastes. This is attributed to the substitution of AlO5{ 4 tetrahedra in the bridging positions of the C–S–H gel phase.18 A shoulder at y4 ppm in the spectra of WPC-168 (Fig. 3d) and WPC-Bi-168 (Fig. 3g) is tentatively assigned to the presence of a third calcium aluminate hydrate phase.18,19 There is no evidence to indicate that the presence of bismuth oxide has any impact on the development of the calcium (sulpho)aluminate hydrate phases in WPC. 1

1

3

H-29Si CP MAS NMR

H-29Si CP MAS NMR spectra of the hydrated cement paste samples prepared during this study are shown in Fig. 4. Only hydrated silicate species appear in 1H-29Si CP MAS NMR spectra, as this technique relies on the relaxation of 29Si nuclei via neighbouring protons. These


27

Al MAS NMR spectra for a anhydrous WPC, b WPC-6, c WPC-24, d WPC-168, e WPC-Bi-6, f WPC-Bi-24 and g WPC-Bi-168: spinning side bands are denoted by asterisks

CP spectra were collected to locate the approximate chemical shifts of the early silicate hydration products. After 6 h of hydration, isolated hydrated Q0(H) species are present at approximately 275?5 ppm in the CP spectra of both WPC-6 and WPC-Bi-6 (Fig. 4). The former also contains significant intensity in the Q1 and Q2 regions of the spectrum, which develop into the respective partially resolved peaks at 280 and 286 ppm with increasing time. It should be noted that the signal intensities recorded in CP spectra are functions of relaxation and contact parameters, number of protons and their 1H-29Si internuclear distances; hence, they are not proportional to the relative abundance of the different Qn species. 29

Si MAS NMR

The 29Si MAS NMR spectrum of anhydrous WPC used in this study is shown in Fig. 1 (labelled ‘WPC background spectrum’). This spectrum comprises a sharp peak at 272 ppm, which is assigned to Q0 species in belite and a broad signal spanning the region 268 to 278 ppm arising from various crystallographically distinct Q0 species in alite.19,20 The original 29Si MAS NMR spectra of the Bi2O3 blended and unblended hydrated pastes are shown in Fig. 5. In both cases, the signal arising from alite diminishes at the expense of the formation of signals in the Q1 and Q2 regions of the spectrum as hydration proceeds. A qualitative inspection of these 29Si MAS NMR data indicates that the


2013

VOL

112

NO

4

209

Li and Coleman

4


1

H-29Si CP MAS NMR spectra of hydrated paste samples of Bi2O3 blended and unblended WPC

rate of formation of Q1 and Q2 hydration products in the cement paste containing Bi2O3 is initially slower than that of its unblended counterpart. Deconvolution and analysis of ‘subtracted’ spectra

29

Si MAS NMR

The broad signal arising from unhydrated alite obscures the region of the 29Si MAS NMR spectrum in which the early Q0(H) and Q1 hydrated silicate products appear. In order to evaluate the early hydration products more accurately, the residual alite signal in the spectra of the hydrated pastes was removed by subtraction of a suitably adjusted WPC background spectrum before deconvolution (as outlined in the section on ‘Experimental’). The subtracted, deconvoluted and fitted 29 Si MAS NMR spectra, together with the residue of the subtracted and fitted spectra for the unblended WPC pastes, are given in Figs. 1 and 6. Those of the blended WPC-Bi pastes are given in Fig. 7. In each case, in addition to the early hydration products, the subtracted spectrum also contains a proportion of the unhydrated belite resonance at 272 ppm. This signal remains in the subtracted spectrum owing to its relatively slow hydration rate compared with that of alite.21 The small residues (i.e. the differences between the subtracted and fitted spectra) obtained by this method indicate that it is a suitable approach for the deconvolution and analysis of the resonances arising from early hydration products. Further validation for this method derives from the fact that each of the deconvoluted Q0(H) signals possesses

210


2013

VOL

112

NO

4

5

29

Si MAS NMR spectra of hydrated paste samples of Bi2O3 blended and unblended WPC

the same shape and line width, as do the Q1, Q2(1Al) and Q2 signals for a given sample type. The position and number of signals present in the subtracted and deconvoluted 29Si MAS NMR spectra of the pastes blended with Bi2O3 do not differ from those of the unblended pastes. This observation supports the argument that bismuth oxide does not participate in the hydration processes of WPC and that it merely acts as an inert filler. The relative abundance of the various silicate species and the degree of hydration of the WPC and WPC-Bi pastes are listed in Table 2. These data confirm that the presence of 20 wt-% Bi2O3 reduces the initial rate of formation of C–S–H within the first 24 h; however, after 1 week, there is little difference between the extents of hydration of the blended (58%) and unblended (60%) pastes. It is likely that the initial reduction in hydration rate of the Bi2O3 blended paste arises from the increased ‘wetting volume’ of this system. It is considered that the presence of bismuth oxide particles (of diameter y2 mm) in the blended paste will increase the total surface area presented for wetting and will thereby initially reduce the quantity of free water available for hydration. Moreover, the addition of an inert filler would normally lead to a modest acceleration of early age hydration due to the presence of nucleation sites. This may mean that the effect of the increased ‘wetting volume’ is even greater than it initially appears.22 It should be noted that the ability of a radiopacifying agent to reduce the initial rate of hydration of a dental cement is a distinct disadvantage, as a common complaint regarding MTA

Li and Coleman

6 Subtracted, ﬁtted and deconvoluted spectra of unblended WPC pastes


29

Si MAS NMR

is that its setting time (which is reported to vary between 50 min and 2?75 h) is inconveniently long.23 Transmission electron microscopy 29

Si and 27Al MAS NMR data collected in this study indicate that the early rate of formation of the C–S–H gel is reduced in the presence of 20 wt-% Bi2O3 and that the development of the calcium (sulpho)aluminate phases is not affected by this additive. Representative TEM images of WPC-6 and WPC-Bi-6 are shown in Fig. 8 and confirm these findings. Both paste samples comprise similar lath-like ettringite crystals of ,100 nm in width. In comparison, sample WPC-6 contains a significantly greater proportion of fibrillar outer product C–S–H gel than does WPC-Bi-6. Visual examination of the TEM images also indicates that the bismuth oxide particles remain intact and that bismuth is not

7 Subtracted, ﬁtted and deconvoluted spectra of Bi2O3 blended WPC pastes

29

Si MAS NMR

Table 2 Relative abundance of Qn species and degree of hydration for Bi2O3 blended and unblended WPC pastes Sample

Q0/%

Q0(H)/%

Q1/%

WPC-6 WPC-24 WPC-168

94.3 57.8 40.2

0.34 1.41 4.48

2.61 23.5 29.1

WPC-Bi-6 WPC-Bi-24 WPC-Bi-168

97.6 71.4 42.1

0.72 2.54 3.97

0.72 12.8 26.8

Q2(Al)/%

Q2/%

Degree of hydration/%

0.73 5.87 9.17

2.03 11.4 17.0

5.7 42 60

0.51 7.11 12.5

0.41 6.13 14.6

2.4 29 58


2013

VOL

112

NO

4

211

Li and Coleman


mixture containing the 2 mm bismuth oxide particles. After 1 week, the degree of hydration of the blended (58%) and unblended (60%) pastes were found to be similar.

References

8 Images (TEM) of a WPC-6 and b WPC-Bi-6

transferred to either the ettringite or C–S–H gel product phases.

Conclusion Before this investigation, the impact of bismuth oxide on the hydration chemistry of Portland cement was disputed. This research has demonstrated that 29Si, 1 H-29Si CP and 27Al MAS NMR are powerful techniques for the analysis of Portland cement based dental materials and has provided abundant evidence to support the argument that bismuth oxide plays no appreciable role in the hydration processes of Portland cement. X-ray diffraction, 27Al and 29Si MAS NMR and TEM analyses of WPC paste and that containing 20 wt-% replacement bismuth oxide indicate that this additive does not participate in the cement hydration reactions. The only observable difference between the blended and unblended pastes was a reduction in the initial rate of formation of C–S–H gel in the blended paste within the first 24 h. This was attributed to a relative reduction in free water arising from the increase in surface area of the

212


2013

VOL

112

NO

4

1. H. W. Roberts, J. M. Toth, D. W. Berzins and D. G. Charlton: ‘Mineral trioxide aggregate material use in endodontic treatment: a review of the literature’, Dent. Mater., 2008, 24, 149–164. 2. B. W. Darvell and R. C. T. Wu: ‘‘MTA’’–an Hydraulic Silicate Cement: review update and setting reaction’, Dent. Mater., 2011, 27, 407–422. 3. M. Torabinejad and D. White: ‘Tooth filling material and use’, US patent no.5,415,547, 1995. 4. J. Camilleri: ‘Hydration mechanisms of mineral trioxide aggregate’, Int. Endod. J., 2007, 40, 462–470. 5. J. Camilleri: ‘Characterization of hydration products of mineral trioxide aggregate’, Int. Endod. J., 2008, 41, 408–417. 6. Y. L. Lee, B. S. Lee, F. H. Lin, A. Y. Lin, W. H. Lan and C. P. Lin: ‘Effects of physiological environments on the hydration behavior of mineral trioxide aggregate’, Biomaterials, 2004, 25, 787–793. 7. T. Dammaschke, H. U. V. Gerth, H. Zu¨chner and E. Scha¨fer: ‘Chemical and physical surface and bulk material characterization of white ProRoot MTA and two Portland cements’, Dent. Mater., 2005, 21, 731–738. 8. J. S. Song, F. K. Mante, W. J. Romanow and S. Kim: ‘Chemical analysis of powder and set forms of Portland cement, gray ProRoot MTA, white ProRoot MTA, and gray MTA-Angelus’, Oral Surg. Oral Med. Oral Pathol. Oral Radiol. Endod., 2006, 102, 809–815. 9. A. R. Brough: ‘Solid state NMR studies of inorganic materials’, DPhil thesis, University of Oxford, 1993. 10. I. G. Richardson, A. R. Brough, R. Brydson, G. W. Groves and C. M. Dobson: ‘The location of aluminum in substituted calcium silicate hydrate (C-S-H) gels as determined by 29Si and 27Al NMR and EELS’, J. Amer. Ceram. Soc., 1993, 76, 2285–2288. 11. I. G. Richardson, A. R. Brough, G. W. Groves and C. M. Dobson: ‘The characterization of hardened alkali-activated blast-furnace slag pastes and the nature of the calcium silicate hydrate (C-S-H) phase, Cem. Concr. Res., 1994, 24, 813–829. 12. I. G. Richardson, G. W. Groves: ‘The structure of the calcium silicate hydrate phases present in hardened pastes of white Portland cement/blast-furnace slag blends’, J. Mater. Sci., 1997, 32, 4793– 4802. 13. C. A. Love: ‘Microstructure and silicate anion structure of hardened blended cement pastes’, PhD thesis, University of Leeds, 2002. 14. C. A. Love, I. G. Richardson and A. R. Brough: ‘Composition and structure of C-S-H in white Portland cement-20% metakaolin pastes hydrated at 25uC’, Cem. Concr. Res., 2007, 37, 109–117. 15. Q. Li: ‘The nature of early age hydration products’, PhD thesis, University of Leeds, 2008. 16. J. Skibsted, H. J. Jakobsen and C. Hall: ‘Direct observation of aluminium guest ions in the silicate phases of cement minerals by 27 Al MAS NMR spectroscopy’, Inorg. Chem., 1993, 32, 1013–1027. 17. J. Skibsted, E. Henderson and H. J. Jakobsen: ‘Characterization of calcium aluminate phases in cements by aluminum-27 MAS NMR Spectroscopy’, Inorg. Chem, 1993, 32, 1013–1027. 18. M. D. Andersen, H. J. Jakobsen and J. Skibsted: ‘Characterization of white Portland cement hydration and the C-S-H structure in the presence of sodium aluminate by 27Al and 29Si MAS NMR spectroscopy’, Cem. Concr. Res., 2004, 34, 857-868. 19. J. Skibsted, H. J. Jakobsen and C. Hall: ‘Quantification of calcium silicate phases in Portland cements by 29Si MAS NMR spectroscopy’, J. Chem. Soc. Faraday Trans., 1995, 91, 4423-4430. 20. H. Krøyer, H. Lindgreen, H. J. Jakobsen and J. Skibsted: ‘Hydration of Portland cement in the presence of clay minerals studied by Si-29 and Al-27 MAS NMR spectroscopy’, Adv. Cem. Res., 2003, 15, 103–112. 21. J. Hjorth, J. Skibsted and H. J. Jakoben: ‘29Si MAS NMR studies of portland cement components and effects of microsilica on the hydration reaction’, Cem. Concr. Res., 1988, 18, 789–798. 22. E. H. Kadri, S. Aggoun, G. De Schutter and K. Ezziane: ‘Combined effect of chemical nature and fineness of mineral powders on Portland cement hydration’, Mater. Struct., 2010, 45, 665–673. 23. B. Czarnecka, N. J. Coleman, H. Shaw and J. W. Nicholson: ‘The use of mineral trioxide aggregate in endodontics – status report’, Dent. Med. Probl., 2008, 45, 5-11.