Cement & Concrete Composites ELSEVIER Cement and Concrete Composites 21 (1999) 117-121

Thaumasite - background and nature in deterioration of cements, mortars and concretes

John Bensted

Department of Crystallography, Birkbeck College, University of London, Malet Street, London WCIE ?HX, UK

Abstract

Thaumasite has been shown to form at low temperatures, particularly 0-S’C, as a non-binding carbonate sulphate hydrate under conditions of destructive sulphate attack. Formation of thaumasite arises generally from C-S-H and Ca2+, CO:-, SO:-, CO2 and water, or from in the presence of C-S-H, carbonate and/or dioxide and water. It basically resembles a carbonated ettringite, with which it has often been confused in the past. Conversion of the main cementitious binder C-S-H into the non-binder thaumasite is a destructive form of sulphate attack. Greater awareness of the potential problems that thaumasite can cause has arisen with the increased use of limestone fillers in cements, the common employment of limestone aggregates in concrete and the introduction of Portland limestone cements, together with the realisation that structural foundations of buildings are, on average, below ambient temperature and (more often than not with on- and above-ground construction) are within the optimum temperature range for thaumasite to be formed. Instances have been found in specific studies of large quantities of thaumasite being formed in foundation concretes with no evidence for any structural damage above ground level. It is important to be aware of the propensity of thaumasite to form at low temperatures for mix designs, so as not to encourage any destructive sulphate attack by thaumasite to arise. This means utilising low water/ cement ratios for workable mortars and concretes, so as to give reduced permeability. This will prevent, or at least suitably hinder, ingress of destructive ions and water, so that the potential for destructive sulphate attack by formation of thaumasite is not actually realised in practice. 0 1999 Elsevier Science Ltd. All rights reserved.

Keywords: Thaumasite; Cement; Mortar; Concrete

1. Introduction structures in which its appearance as a deterioration Thaumasite has long been known as a naturally product had already been identified. occurring mineral of the type CaSiO,CaSO,. CaC03.15Hz0, which has been found in meta- morphosed rocks that have undergone hydrothermal 2. Structural considerations changes with the passage of time [l-3]. It often arises in thin veinlets in association with both ettringite Controversy reigned over the actual structure of 3CaO~A1203~3CaS04~31-32Hz0 and CaCO, thaumasite, in particular, the precise co-ordination of [2,3]. In construction, thaumasite has been found as a by (hydroxyl), for a number of years. product of deterioration of cements, mortars and The mineral thaumasite was originally classified as concretes in a wide number of instances. Occurrences having a hexagonal pyramidal type of structure of thaumasite in deteriorated concrete were discussed containing SiO, tetrahedra and CO3 triangular planes. at length in a detailed report in the 1960s [4]. At that The formulation Ca6(C03)2(S04)2(Si03)z~30H~0 was time, although the mineral thaumasite was thought to used as a basis by Bni et al. for postulating a plausible have formed at a very low temperature and also at a structural layout [6,7]. An X-ray crystallographic very late stage of mineralisation [5], neither its method investigation by Welin [8] led to the structure of synthesis nor its precise structure were known. 2{Ca3Hz(C03/S04/Si03)*13H20} being proposed. This Controversies over whether the silicon in thaumasite was considered to be of the type Ca6H4(Si04)2(C03)2- was 4- or 6-co-ordinated by oxygen (hydroxyl) needed (S0&26H20 with a CjV, based upon to be resolved in order to understand the nature of columns made from Ca” and SiO!-- ions and Hz0 thaumasite, its formation and the implications for molecules, between which occurred CO:- and SOi-

0958-9465/99/$ - see front matter 0 1999 Elsevier Science Ltd. All rights reserved PII: SO9S8-9465(9X)00076-0 118 J. BenstedlCement and Concrete Composites 21 (I 999) I1 7- I21 ions as well as additional HZ0 molecules; the four Hf The Raman spectrum of thaumasite showed an ions could not be suitably placed and were thought to absence of wavebands which could be associated with be attached to the SO:- groups [8]. However, the the symmetrical SiOi- ion. Furthermore, the pattern orientation of the SOi- ions in this proposed structure observed for the silicate bands [13] was very different was questioned by Font-Altaba [9]. These earlier struc- from the one produced by tetrahedral silicate ions tural postulations all envisaged Si as 4-co-ordinated SiOi~- in solution [19], and also from the spectra of with oxygen in the thaumasite structure. cementititious , like alite and belite [14,20,21], Such a structure containing Si in 4-co-ordination which showed clear evidence of SiOi- tetrahedra, with oxygen was challenged by Moenke on the basis of unlike for thaumasite. an infrared spectroscopic examination [lO,ll]. He In order to check out the ideas of Kirov and suggested that thaumasite actually contained silicon in Poulieff,‘” who had suggested that the infrared vibra- 6-co-ordination with oxygen, because of the similarity tion in thaumasite at 1100 cm-’ was a composite band of the infra-red wavebands in the 765-638 cm ’ and due to both SiOi- and SiOi- ions, chromate-thauma- 935-887 cm-’ regions with the corresponding site was prepared and studied by infrared spectroscopy. wavebands of , a high temperature-high Sulphate wavebands can be characterised on the basis pressure polymorph of silica that was known to contain of the well-known isomorphism of sulphates with silicon in 6-co-ordination with oxygen [12]. Kirov and chromates, where, in the corresponding minerals, the Poulieff [13] strongly disputed Moenke’s interpretation infrared wavebands are largely similar apart from the on the grounds that the IR waveband of thaumasite at relative positions of those wavebands due to SO:- or 1100 cm-’ was a composite due to both SO:- and CrO:-. The CrO:- wavebands arise at lower SiOi- groups. They had examined IR and X-ray wavenumbers than the SiO$-- ones. In this way, the actual positions of SiO:- wavebands can be confirmed diffraction data of thaumasite both at ambient [22,23]. This technique is somewhat analogous to the temperature and upon heating at various temperatures characterisation of H20 and OH- bands in compounds between 200 and 1150°C. In their view [13], because by deuteration, where the corresponding DzO and the vibrational frequencies of SOi- and SiOi~ OD wavebands are shifted to lower wavenumbers, coincided, causing only one absorption maximum at whilst the wavebands of other functional groups remain 1100 cm-‘, these vibrational frequencies would resolve largely unchanged [24,25]. The strong band at upon heating and hence become independently identi- 1100 cm ~’ in thaumasite had moved to 880 cm _’ in fiable. The flaw in this argument is that thaumasite chromate-thaumasite, confirming that the 1100 cm ’ in actually breaks down at 110°C and transforms into a thaumasite is a sulphate waveband. Some minor glassy phase known as thaumasite glass, in which the absorption remaining near 1100 cm-‘, which was randomly distributed silicate groups contain Si revealed in chromate-thaumasite, is due either to an 4-co-ordinated by oxygen [ 141. Si-O-H bending mode, or to any residual calcium Later X-ray crystallographic investigations of silicate hydrate that had not reacted to form chromate- thaumasite suggested that its structure was of the type thaumasite. This absorption was not of the type Ca6[Si(OH),,]2(C03)2(SO&24Hz0 [15-171, similar to expected for an Si04 stretching mode in chromate- that of ettringite Ca6[A1(OH)&(S04)3.26Hz0. The thaumasite and hence in thaumasite. Also, the vibra- stability of thaumasite and its Si(OH), octahedra was tions at 640 and 590 cm-’ in thaumasite are sulphate considered to be due to the other electronegative bands, since they are absent from the spectrum of atoms attached to oxygen, presumably drawing chromate-thaumasite [14]. Therefore, the infrared electrons away from the Si-0 bonds to allow an spectrum of thaumasite shows Si in 6-co-ordination increase in the Si co-ordination number from 4 to 6, with OH, which supports the earlier conclusions of and thus allowing the distorted [Si(OH),]‘- octahedra Moenke [lo]. to become stabilised. *“Si{‘H} cross polarisation magic-angle spinning Further confusion arose from IR investigations by nuclear magnetic resonance has shed more light on the Portnov and Solntsev [18], who assigned wavebands at Si(OH), groups in thaumasite, which has enabled both 510 and 770 cm-’ to H2SiOip ions and a waveband at qualitative and quantitative determinations of thauma- 780 cm ’ to [Si(OH),]*- ions. They thought that site in cementitious materials to be made [26], thaumasite contained silicon in both 4- and Different geometries for the AlO, octahedra in ettrin- 6-co-ordination with oxygen/hydroxyl. gite and for the SiOh octahedra in thaumasite, which The spectroscopic uncertainties about the permit substitution of Si for Al in ettringite to give a co-ordination of silicon were finally resolved by investi- different resonance from that found in thaumasite, gation of the laser Raman spectrum of thaumasite [3] enable clear unambiguous quantification of thaumasite and an infrared examination involving the preparation to be obtained in the presence of ettringite [26], which of the chromate analogue of thaumasite [14]. is not always what happens with other techniques [27]. .I. BenstedlCement and Concrete Composites 21 (1999) I1 7-121 119

3. Formation of thaumasite + A1203~xH20+ 3Ca(OH)z

3.1. Chemical reactions The calcium hydroxide Ca(OH), formed here can readily carbonate: Thaumasite is produced experimentally by the inter- action of starting materials optimally in stoichiometric Ca(OH)2 + C02-+CaC03 + H20. proportions for the formation of thaumasite, preferably The equation for thaumasite formation from ettringite at 0-5°C and certainly below approx. 10°C. The starting is only approximate because some of the aluminium materials should include: can be contained in solid solution within the thauma- l A source of calcium silicate, such as calcium oxide site structure. Also, there is some iron(II1) in solid (or hydroxide) plus silica, tricalcium silicate Ca3SiOs, solution with aluminium(II1) in ettringite, particularly dicalcium silicate Ca,SiO, preferably in the p-form. at late ages when ferrite phase hydration is substan- l A source of calcium sulphate, like tial.*: This would result ultimately in an iron(II1) oxide CaS04*2H20, the hemihydrate (bassanite) gel Fe203.yH20 along with the alumina gel when CaS04.0.5Hz0 or anhydrite CaS04. thaumasite is formed from ettringite. l A source of calcium carbonate, for instance calcite CaC03 or atmospheric carbon dioxide interacting with a source of Ca*’ ions. 3.2. Significance of the 6co-ordination of silicon by l Excess water. hydroxyl in thaumasite

In other words, thaumasite is formed by a general At a first glance the formation of thaumasite seems reaction involving calcium ions, silicate, sulphate, somewhat puzzling, since it only forms at low tempera- carbonate and sufficient water to permit both transport tures. However, once formed, it is stable up to approx. of the potentially reactive species and to form thauma- llO”C, when it decomposes to form the disordered site where these species can interact under the low structure known as thaumasite glass. This stability is to temperature conditions prevailing. This is a very be contrasted with that of its close structural relative important point. ettringite which, although formed at ordinary tempera- The silicates need to be hydraulic so that the main tures, suffers a lattice collapse at approx. 60°C and cementitious binder calcium silicate hydrate C-S-H decomposes at approx. 90°C. Most structures that can be forming, in order that thaumasite can be contain silicon in 6-co-ordination with oxygen are produced. The general chemical reaction can be formed at high pressures and usually high temperatures expressed as follows using the approximate formula [29]. However, a few, like certain silicon phosphates Ca3Si207.3H20 for the non-stoichiometric calcium and thaumasite are formed either at or near silicate hydrate binder C-S-H: atmospheric pressure [29]. “Ca&07.3H20” + 2{CaS04*2H201 + CaC03 + CO2 The reason why thaumasite only forms at low temperatures can be attributed to the need to form a transition state intermediate involving a hexagonal arrangement of OH- ions around a core Si4+ ion long or enough to allow charge delocalisation away from the “Ca3Siz07.3H20” + 2{CaS04*2H20} + CaC03 + 24H20 embryonic [Si(OH),12- groups as they are being formed onto the carbonate groups present; this allows -+Ca6[Si(OH)~]2(C0~)~(S04)2~24H~0) + Ca(OH)2 a stable, configuration of [Si(OH),]‘- octahedra to be The reaction is slow and can often take 6 months to produced. Clearly, as indicated from the appearance of 1 year or more to obtain significant yields of the infrared and Raman spectral bands attributed to thaumasite. Si06 [3,10,14], the symmetry is heavily distorted away Thaumasite can also be formed slowly at low from Oh to lower symmetries. In other words, a low temperatures, as above, from the interaction of ettrin- temperature is essential for producing the transition gite, with which it can also enter into partial solid state intermediate for thaumasite to form. Once solution, with calcium silicate hydrate C-S-H and formed, however with the necessary charge delocalisa- calcite or atmospheric CO2 in the presence of Ca2+ tion for stabilising the [Si(OH)6]2m groups, the thauma- ions: site structure is stable up to approx. 110°C. Unlike for ettringite, there is no lattice collapse within thaumasite Ca,JAl(OH)&(S04)3~26H20 + “Ca&07.3H20” preceding its true decomposition. Structures containing + CaC03 + CO2 +xH,O silicon in 6-co-ordination with oxygen are very rare,29 unlike for germanium [30]. The only known structures +Ca,[Si(OH)6]2(C03)2(S04)2*24H20 + CaS04*2H20 containing [Si(OH),12- groups are thaumasite and 120 J. BenstediCement and Concrete Composites 21 (1999) 117-121 its chromate analogue Ca6[Si(OH)6]2(C03)2(CrO&~ will prevent, or at least seriously frustrate, the trans- 24H20. The hydrated sodium silicate Na2Si03*3H20, port of ions like Ca’+, CO:-, SOi- and sufficient previously thought to be composed of NazSi(OH)6 moisture to the hardened cement surfaces. Good mix units [29], is now known to contain silicon in design is a key consideration in seeking to avoid any 4-co-ordination with oxygen (J. Bensted, unpublished). potential structural problems that might arise if thaumasite is to be formed. 3.3. General considerations

It is clear that the non-binder thaumasite can be 4. Conclusion formed from the binder calcium silicate hydrate C-S-H in the presence of carbonates and/or The background and nature of the formation of atmospheric carbon dioxide, sulphates and moisture at thaumasite, a non-binding calcium carbonate silicate low temperatures, optimally O-SC, where reaction sulphate hydrate, in the deterioration of cements, conditions like relative solubilities are at their most mortars and concretes is discussed. It is known that favourable. It can also be produced from ettringite and thaumasite is readily formed when the hardened C-S-H when carbonate and/or CO2 and sufficient cement binder calcium silicate hydrate C-S-H is in water are present at the low temperatures measured. contact with Ca*+, carbon dioxide, carbonates, In the past, thaumasite formation was not readily sulphates and sufficient moisture at low temperatures recognised, because of its similarity, when measured by below approx. 15°C and particularly at approx. 0-5°C. various techniques, to carbonated ettringite [27]. Such formation is becoming more clearly recognised Thaumasite, when produced in such structures, like now that thaumasite is being properly identified and C-S-H, ettringite etc. is not a pure mineral and not confused with carbonated ettringite. Limestone is contains other cations and anions in solid solution. not chemically inert in hardened cements, mortars and However, now thaumasite is more readily known concretes, but can react, if the conditions are right, to about, it is being located more frequently in structures produce thaumasite. This can arise in limestone aggre- where deterioration is taking place. An example of this gates with the (more reactive) fines content, in is in foundations of buildings where the temperatures Portland limestone cement and masonry cements may commonly average approx. 10°C when the ambient containing ground chalk or limestone, as well as in temperatures are around 20°C. Also, where limestone sulphate-resisting Portland cements. aggregates have been employed in concrete, the fines Indeed, thaumasite has been found in large quanti- content in particular can participate in the thaumasite ties in some foundation concretes examined [31] formation reaction. Limestone ‘fillers’ in cement and without any evidence of structural damage above Portland limestone cement, where up to 35% ground- ground, indicating that similar problems might occur in limestone may be contained, can be susceptible to elsewhere and have so far gone unreported. Formation thaumasite formation if the conditions are right. Also, of thaumasite can be accompanied by swelling or sulphate-resisting Portland cement can be subject to expansion of the affected concrete, and added thaumasite formation, again if the conditions are right. problems of freeze-thaw probably contribute to the The thaumasite formation reaction effectively ceases degradation process [31]. when the temperature rises above approx. WC, but The potential for structural problems with thauma- starts being produced again when the temperature is site may not arise in practice if the mortars and lowered below WC, and especially when it lies concretes are well made, using a low water/cement between 0 and 5°C. In relative small amounts locally it ratio to minimise internal transport of ions like Ca’+, may not be serious if the ratio of binder C-S-H to CO:-, SOi- and of water as the carrier for them. If non-binder thaumasite remains large enough not to this is achieved, when the temperature falls below 15°C affect the basic structural integrity. However, if this such that formation of thaumasite can arise, then the ratio becomes small, then the structure is in danger of relative absence of these ions and of water to transport serious deterioration. Figures for safe or unsafe ratios them will ensure that thaumasite is not formed in such cannot be generalised, as they would depend upon the localities and deterioration from its formation will not precise circumstances and locations for the structures ensue there. Conversely, where mortars or concretes in question. are badly produced at high water/cement ratios, where Although thaumasite formation cannot be prevented internal transport of the aforementioned ions and in toto if the temperatures are low enough, its negative water is facilitated, then, when the temperature is suffi- effects can be minimised by employing water/cement ciently low, thaumasite can readily form [32]. ratios as low as practicable (for good workability Awareness of the dangers that thaumasite can bring considerations) in cements, mortars and concretes. should permit designs of more mortars and concretes This will allow low permeabilities to be obtained, which to minimise its potentially destructive occurrence. J. BenstedlCement and Concrete Composites 21 (1999) I1 7-121 121

References [I91 Fortnum D, Edwards IO. The Raman spectrum and the struc- ture of the aqueous silicate ion. Journal of Inorganic and Nuclear Chemistry 1956;2:264-265. PI Ramdohr P, Strunz H. Klockmanns Lehrbuch der Mineralogie, 15 Auflage. Stuttgart: Ferdinand Enke, 1967. PO1 Bensted J. Uses of Raman spectroscopy in cement chemistry. PI Hinz W. Silikate-Grundlagen der Silikatwissenschaft und Journal of the American Ceramic Society 1976;59:140-143. Silikattechnik, Band 1. Die Silikate und ihre Untersuchungsme- (211Bensted J. Alcuni studi strumentali sull’idratazione de1 cement0 thoden. Berlin: Verlag fiir Bauwesen, 1970. Portland - Parte 1: spettroscopia Raman. Some instrumental investigations of Portland cement hydration - Part 1: Raman [31Varma SP, Bensted J. Studies of thaumasite. Silicates Indus- triels 1973:38(2):29-32. spectroscopy. 11Cement0 1979;76(2):67-74. [41Erlin B, Stark DC. Identification and occurrence of tbaumasite PI Bensted J. Characterisation of sulphate bands in the IR-spectra in concrete. Highway Research Record 1965;115:108-113. of minerals. Naturwissenschaften 1976;:63:193. [51Vogt T. Thaumasite from Sulitelma, Norway. Norsk Geologisk [231Bensted J. Caratterizzazione delle bande d’onda dei solfati negli Tidsskrift 1938;18:291-303. spettri infrarossi dei minerali di cemento. Characterisation of PI Brfi L, Cuber0 M, PCrez Rodriguez M. Determinacibn de sulphate wavebands in the infrared spectra of cement minerals. estructuras por el mCtodo direct0 mediante la maquina de Eller. 11Cement0 1990;87(3):137-146. Anales de la Real Sociedad Espaiiola de Fisica y Quimica Serie 1241Derkosch J. Methoden der Analyse in der Chemie Band 5. A 1954;50(1-2):105-106. Absorptionsspektralanalyse in ultravioletten, sichtbaren und [71Br6 L, Cuber0 M, tipez A. Funcidn de Patterson tridimen- infraroten Gebiet. Frankfurt am Main: Akademische Verlagsge- sional de la thaumasita. Anales de la Real Sociedad Espafiola sellschaft, 1967. de Fisica y Quimica Serie A 1957;53(1-2):67-76. 1251Nakamoto K. Infrared and Raman spectra of inorganic and PI Welin E. The of thaumasite Ca,H,(CO?/ co-ordination compounds, 4th edition. New York: Wiley-Inters- S0,)Si0,.13Hz0. Arkiv fijr Mineralogi och Geologie, Sverige cience, 1986. 1956;2:137-147. WI Skibsted J, Hjorth L, Jakobsen HJ. Quantification of thaumasite I91Font-Altaba M. A thermal study of thaumasite. Mineralogical in cementitious materials by %{‘H} cross polarisation magic- Magazine 1960;32:567-572. angle spinning NMR spectroscopy. Advances in Cement [lOIMoenke H. Ein weiteres Mineral mit Silizium in Research 1995;7:69-83. 6er-Koordination: Thaumasit. Naturwissenschaften 1964;51:239. v71 Bensted J. Problemi the si verificano nell’identificazione delle Moenke H. Mineralspektren II. Tafel 6.164. Berlin: Akademie- PI thaumasite. Problems arising in the identification of thaumasite. Verlag, 1966. 11Cement0 1977;74(3):81-90. Lyon RJP. Infrared confirmation of 6-fold co-ordination of u21 Bensted J. Fase ferritica - Uno studio spettroscopio all’infrar- silicon in stishovite. Nature (London) 1963;196:266-267. WI osso. The ferrite phase - an infrared spectroscopic study. 11 Kirov GN, Poulieff CN. On the infrared spectrum and thermal t131 Cement0 1976;73(1):45-51. decomposition products of thaumasite CaH2(C0,/ Bensted J, Varma SP. Silicaten met silicium in zes-coordinati& S0,)Si0,.13H20. Mineralogical Magazine 1968; 36:1003-1011. v91 met zuurstof. Klei en Keramiek 1973;23(4):66-69. P41 Bensted J, Prakash Varma S. Studies of thaumasite - Part II. Silicates Industriels 1974;39:11-19. [301Strunz H. Die Beziehungen der lsotypie zwischen Silikaten und Germanaten. Versuch einer Germanat Klassifikation. Naturwis- [151Edge RA, Taylor HFW. Crystal structure of thaumasite, a mineral containing [Si(OH),]‘- groups. Nature (London) senschaften 1960;47(7):194-195. 1969;224:363-364. [311Crammond NJ, Halliwell MA. The thaumasite form of sulphate attack in concretes containing a source of carbonate ions-a WI Laffaille A, Protas J. Nouvelles donntes sur la structure de la thaumasite. Comptes Rendus Hebdomadaires des Sciences de microstructural overview. AC1 SP-154-19. 2nd CANMET/ACI 1’AcadCmie des Sciences, Paris, SCrie D 1970;270:2151-2154. International Symposium for Advances in Concrete Technology, [I71Edge RA, Taylor HFW. Crystal structure of thaumasite Las Vegas, Nevada, 12 June 1995, pp. 357-380. American [Ca,Si(OH),,12H,O](SO,)(CO,). Acta Crystallographica B Concrete Institute, Washington DC (1995). 1971;27:594-601. [321Bensted J. Taumasite - un prodotto di deterioramento delle P31 Portnov AM, Solntsev BP. Thaumasite from the Kola Peninsula strutture di cement0 indurito. Thaumasite - a deterioration (in Russian). Trudy Mineralogicheskovo Muzeya im. A.E. product of hardened cement structures. 11 Cement0 Fersmana, Akademiya Nauk SSR 1971:20:217-220. 1988;85(1):3-10.