A New Mechanisms for Cement Hydration Inhibition

8/12/2019 A New Mechanisms for Cement Hydration Inhibition

http://slidepdf.com/reader/full/a-new-mechanisms-for-cement-hydration-inhibition 1/16

Subscriber access provided by RICE UNIV

Chemistry of Materials is published by the American Chemical Society. 1155 SixteenthStreet N.W., Washington, DC 20036

Article

A New Mechanism for Cement Hydration Inhibition: Solid-State

Chemistry of Calcium Nitrilotris(methylene)triphosphonateMaximilienne Bishop, Simon G. Bott, and Andrew R. Barron

Chem. Mater., 2003, 15 (16), 3074-3088 • DOI: 10.1021/cm0302431

Downloaded from http://pubs.acs.org on January 12, 2009

More About This Article

Additional resources and features associated with this article are available within the HTML version:

• Supporting Information• Links to the 3 articles that cite this article, as of the time of this article download

• Access to high resolution figures• Links to articles and content related to this article

• Copyright permission to reproduce figures and/or text from this article

http://pubs.acs.org/doi/full/10.1021/cm0302431



A New Mechanism for Cement Hydration Inhibition:Solid-State Chemistry of Calcium

Nitrilotris(methylene)triphosphonate

Maximilienne Bishop,† S imo n G. B o t t ,‡ and Andr e w R . B a r r on*,†

Department of Chemistry and Center for Nanoscale Science and Technology, Rice University,

Houston, Texas 77005, and Department of Chemistry, University of Houston,H ouston, Texas 77204

Received F ebru ar y 28, 2003. Revised M anuscript Received M ay 27, 2003

The reaction of the cement retarder nitrilo-tris(methylene)phosphonic acid, N[CH 2P O-(OH)2]3 (H 6ntm p) w ith ca lcium oxide, tricalcium silicat e (C3S), trica lcium alum ina te (C3A),and tetracalcium aluminoferrite (C4AF) has been studied individually, and in the case ofC3A in the presence of gypsum, to gain an understanding of the effect on the individualminerals prior to studying a typical sample of Portland cement. The reaction of H 6n t m pwit h ca lciu m o x ide re su lt s in t h e in i t ia l fo rma t io n o f so lu ble [ Ca (H n ntmp)](4-n )-, wh ichprecipitat es over time a s [Ca (H 4ntmp)(H 2O)]

∞, w hose sheetlike stru cture ha s been confirmed

by single-crystal X-ray diffraction. The study of the hydration of C3S in the presence ofH 6ntmp indicat es that n o C -S -H forms, a nd th e surfa ce chan ges from silicon-rich to ca lcium-

rich associat ed w ith the format ion of va rious calcium phosphonates. The hydr at ion of C3Ais severely inhibit ed in t he presence of H 6ntmp, w ith t he phosphonic acid reacting primar ilyw ith calcium as opposed to aluminum to form a Ca -P-rich layer at the surface of C3A. TheH 6ntmp enhances calcium solubility, promoting the dissolution of calcium from C3A andpromoting, in the presence of gypsum, the formation of et tringite. In the presence of H 6-n t mp t h e su rfa ce of h ydra t e d P ort la n d ceme n t g ra in s is r ich in ca lciu m a n d p h osp h oru sa n d d ef ici en t i n s il icon a n d a l u m in u m , con s is t en t w i t h t h e f or m a t i on of a ca l ci umphosphonate coating spectroscopically related to [Ca(H 4ntmp)(H 2O)]

∞. We have proposed a

new mechanism by w hich phosphonic a cids inhibit cement h ydra tion. Dissolution, of calciumby e xt ra ct ion wit h t h e p h osp h on ic a cid , e xp ose s t h e a lu min u m-rich su rfa ce t o e n h a n ceh ydra t ion , fol lowe d by p recip it a t ion of a la ye red ca lciu m p h osp h on a t e t h a t bin ds t o t h esurface of the cement grains, inhibit ing further hydra tion by acting as a diffusion barrier towa t e r a s w e ll a s a n u clea t ion in h ibit o r . Sa mp le s we re ch a ra ct e rize d by 31P , 27Al, and 29S iMAS NMR sp ect roscop y, sca n n in g e le ct ro n micro scop y, X-ra y di ffra ct ion , a n d X-ra y

photoelectron spectroscopy.

Introduction

Cementit ious ma teria ls ha ve been in use for over 2000years and to this day are considered essential to boththe constr uction a nd oil indust ries. Portla nd cement, inpart i c ul ar , has found suc c ess i n a w i de range of ap-plicat ions due to i ts unique hardening process. Thead dition of wa ter to dry cement powder r esults in a th incement slurry tha t can be easi ly man ipulated a nd casti nt o d if fe re nt s h a pe s. I n t i me , t h e s lu r r y s et s a n ddevelops strength through a series of hydration reac-tions.1 Hydration of cement is not linear through time;it proceeds very slowly at first, a llowing t he thin mixtureto be properly placed before hardening. Despite theirmundane and low-tech status, cements and the cemen-t a t i on p roce ss a r e n ot w e ll u n de rs t ood d u e t o t h e

complex nature of the number and interdependence ofthe chemical reactions.

P ortland Cement is manufa ctured by heating calciumcarbonate and clay or shale to 1450 °C. The calciumcarbona te is converted t o calcium oxide, or l ime, a ndthe clay minerals y ield dicalcium si l icat e (Ca 2S iO 4,C2S)2 and other inorganic oxides such as aluminatesand ferri tes. Further heating melts the aluminate andferrite phases. The lime reacts with dicalcium silicatet o f rom t r i cal ci um si li cat e (C a 3S iO 5, C 3S ). As t h emixture is cooled, tricalcium alumina te (Ca 3Al2O6, C3A),and tetracalcium aluminoferrite (Ca 4Aln Fe(2-n )O7, C4AF)crysta llize from the melt a nd t ricalcium silicat e and theremaining dicalcium silicate undergo phase transitions.1

These four m inerals comprise th e bulk of most cementmixtures.2

* To w hom correspondence should be a ddressed. E-mai l : [email protected]. URL: ww w.rice.edu/barr on.

† Rice University.‡ University of Houston.(1) Taylor, H. F. W. Cement Chemistry, 2 nd E d .; Academic Press:

London, 1997.

(2) Each of the minerals, Ca 2S iO 4, C a 3S iO 5, C a 3Al2O6, a n d C a 4Aln -F e(2-n)O7, can be conceptual ly broken down into the basic calc ium,si l icon, aluminum, and i ron oxides (CaO, SiO 2, Al2O3, a n d F e2O3,respectively) an d hence the tra ditional cement nomenclatu re abbrevi-ates each oxide, C2S, C3S, C3A, and C4AF. For consistency with theindustry nomenclature, we will use the component formulation [i .e.,(CaO)3(SiO2)] rather than the chemical formula (i .e., Ca 3S iO 5).

3074 Chem. M ater. 2003, 15 , 3074-3088

10.1021/cm0302431 CC C: $25.00 © 2003 American C hemical S ocietyP ubl ish ed on Web 07/09/2003



In the oil industry, bore holes of ever-increasingdepths a re supported by P ortland cement. This a pplica-tion requires a high degree of control over the settingkinetics to al low the cement to be pumped down in al iquid form. A number of chemicals are employed todelay the sett ing t ime. Recent work has concentratedon a series of organic phosphonic acids, in particular,nitr ilo-tr is(meth ylene)phosphonic a cid, N[CH2PO(OH)2]3

(I , H 6ntmp).3 The phosphonic acids have been termed

“super reta rders” d ue to their increased effect on cementhydra tion relat ive to t he effect of conventional retar d-ers.1,3 The a bility of phosphonic acids t o reta rd cementsett ing by lengthening the induction period, but notslowing down the time it takes for setting to occur (oncethe acceleratory period has begun), makes them espe-

cial ly interesting candidates for study.4

The chemicalreac t i ons t hat c ause t he del ay i n hardeni ng are notcompletely understood; however, they are cri t ical todeveloping a ra t ional methodology for the control ofcement setting.

P hosphonic acids a re known to complex ca lcium a ndother M2+ cations,5 poison the nucleation and growthof barium sulfate crystals,6 and inhibi t the hydrationof Fe2O3 and Al2O3 surfaces via direct surface a dsorp-tion.7 These processes fall int o three of the four prima rymodels for cement hyd ra tion inhibition: calcium com-plexation, surface adsorption, nucleation poisoning, andprotective coa ting. 1 Th er e h a v e b ee n a n u m be r ofprevious investigat ions into the act ion of phosphonic

ac ids i n c em ent . Ram ac handran et a l . found t ha t t henumber of PO3 groups on the phosphonic acid influencedthe effectiveness as a retarder.8 The same study alsofound t hat t he phosphoni c ac i ds w ere s l i ght l y m oreeffective tha n t he corresponding phosphona tes; however,no precise mecha nism of inhibition wa s proposed. In aseparate series of studies, a model of the kinetics ofcement hydration in the presence of phosphonic acidsw a s d ev eloped in w h ich it w a s a s su med t h a t t h ephosphonic acid residues inhibited the crystallizationof gelat inous et tringi te by adsorbing onto the surfaceof the grains.9

The hydration of cement is obviously far more com-pl ex t han t he sum of t he hydrat i on reac t i ons of t heindividual minerals. Figure 1 depicts a typical cementgrain, showing the larger C2S and C3S part icles sur-rounded by the much smaller C3A and C4AF matrix.In our present st udy, th e reactions of H 6ntmp with C3Sand C3A were studied individual ly to gain an under-standing of the effects on the individual minerals priort o s t udyi ng a t ypi c al sam pl e of P ort l and c em ent . I na ddition, the product from th e reaction of H 6ntmp withcalcium oxide (and hydroxide) has been characterizedto provide a comparative model system.

As a result of our investigat ion, we propose a new mechan ism for cement inhibition: “dissolution-precipi-tat ion”.

(3) Chi lds, J . D. ; Sut ton, D. L . ; Sabins, F. L . U.S. Patent 4,676,-832, 1987.

(4) Coveney, P.; Davey, R.; Griffin, J . ; Whiting, A. J. Chem. Soc.,Chem. Commun. 1998, 1467.

(5) Sa wa da, K.; Dua n, W.; Ono, M.; Sa toh, K. J. Chem. Soc., Dalton

T r a n s . 2000, 6 , 919.(6) Benton, W.; Col l ins, I . ; Grimsey, I . ; Pa rkinson, G . ; Rodger, S .

Farad ay D iscuss. 1993, 95 , 281.(7) (a) Da vis, G.; Ahear n, J .; Venables, J . J . Vac. Sci. T echnol. 1984,

2 , 763. (b) Davis, G .; Ahearn, J . ; Ma tienzo, L.; Venables, J . J . M a t e r .

Sci. 1985, 20 , 975.(8) Ramachandran, V.; Lowery, M.; Wise, T.; Polomark, G. M a t er .

St r u c t . 1993, 20 , 425.(9) (a) Coveney, P .; Humph ries, W. J . Chem. Soc., Faraday T rans.

1996, 92 , 831. (b) Wattis, J . ; Coveney, P. J. Chem. Phys. 1997, 106 ,9122. (c) Griffin, J . ; C oveney, P .; Whiting, A.; Da vey, R. J . Chem Soc.,

Pe r k i n T r a n s . 1999, 2 , 1973.

Figure 1. A pictoral representation of a cross section of a cement grain. Adapted from Cem ent M icr oscopy ; Ha lliburton Services:D u nc an, O K.

N ew M ech an ism for Cem en t H yd rat ion I n h ibi ti on Ch em . M at er ., V ol . 15, N o. 16, 2003 3075



Experimental Section

Samples of tr icalcium silicate (C3S) and tricalcium alumi-nat e (C 3A) we r e obt aine d fr om C onst r u ct ion Te chnologyLaboratories, Inc. , Skokie, IL. Class H Portland Cement andcement clinkers w ere provided by Ha lliburton Energy Services,Duncan, OK . All samples were used as received a nd stored ina desiccat or to reduce hydra tion and ca rbonation in a ir. Nitrilo-tris(methylene)phosphonic a cid (H 6ntmp) wa s purchased fromAld r ic h c hemicals as a 50 wt % aq u e ous solu t ion and u se dwithout further purification. A sample of the H 6ntmp conta in-

ing r e t a r d e r, D e q u est 2000, wa s p r ovid ed b y H a ll ib u r t onEnergy Services. Characterization by solution 31P N M R a n d31P MAS NMR spectroscopies showed three major impurities,[RCH 2P(O)(OH)2], H 3P O4, a n d H P ( O)(O H )2. H o w ev er ou rana ly sis r e veale d t hat H 6ntmp is th e most rea ctive species incement pa stes.10 SE M ima ges were observed on a P hilips XL-30 E SE M. The samp le s we r e a t t ac he d t o a g lass s l id e u singcarbon tape and then sputtered with gold to reduce chargingeffects. Powder X-ray diffraction studies were performed on aSie mens d if f r act omet e r . XP S ana ly sis w as p er for me d at t heUnive r si t y of H ou st on on a P H I 5700 E SC A sy st e m wit h a1-mm omni foc us lens ar e a , an a lu minu m a node a t 400 W,27-28 mA, a nd a base pressure of 10-10 Torr.

NMR Spectroscopy. Solid s t a t e 29Si MAS NMR spectrawere observed at 39.8 MHz on a Br uker Avance 200 spectrom-eter using a VTN probe designed for use with a 7-mm o.d. ZrO 2

rotor. A MAS spin rat e of 5 kHz w as employed. Sa mples werestudied using a 5.5- µs pulse based on 90° pulse width calibra-tions with hexamethylcyclotrisiloxane. Chemical shifts werereferenced to hexamethylcyclotrisiloxane at δ ) -10.9 ppm.Spectra were collected with a 30-60-s acquisition delay. FIDswere processed w ith 5-Hz line broadening.

Solid s t a t e 27Al MAS NMR spectra were obtained at 52.1MHz on a Bruker Avance 200 spectrometer equipped with twoprobes designed for use with 7- and 4-mm o.d. ZrO2 rotors.Sa mples were studied using 0.5- and 0.3- µs pulses for the 7-and 4-mm p r ob es, r e sp ect ively , b ase d on 90° p u lse wid t hcalibrations with solid aluminum nitrate. MAS spin rates of5 kHz (7-mm rotor) an d 9-kHz (4-mm rotor) were used. S pectrawere collected w ith a 0.1-1.0-s acquisit ion delay. FIDs wereprocessed with 50-Hz line broadening. Samples were exter-nally r e fe r e nc e d t o t he p e ak maximu m of solid a lu minu m

n i t r a t e a t δ ) -1.9 ppm. The shift of solid alum inum n itra teon this instrument was found by using an external referenceof 1.0 M aqueous aluminum nitra te at δ ) 0.0 ppm. Solid state27Al MAS N MR sp e c t r a a t 130. 3 MH z we r e ob t aine d on aBr uker Avance 500 spectrometer designed for use w ith a probeequipped with a 2.5-mm o.d. ZrO2 rotor. Spectra were studiedusing a 0.3- µs pulse based on 90° pulse width calibrations withsolid a lu minu m nit r a t e . A MAS sp in r a t e of 33333 H z wasused. Spectra were collected with a 0.5-s acquisit ion delay.FIDs were processed using 13-Hz line broadening.

Solid s t a t e 31P MAS N MR sp e c t r a we r e ob t aine d a t 81. 0MHz on a Bruker Avance 200 spectrometer using a VTN probed e sig ne d for u se wit h a 7-mm r ot or . Samp le s we r e s t u d ie dusing a 3.0- µs pulse based on 90° pulse width calibrations withNH 4‚H 2P O4. Spectra were collected with an 8.0-s relaxationdelay and a MAS spin rate of 5 kHz. Proton-coupled spectra

were collected using a 0.3-s FID acquisition time. High-powerproton-decoupled spectra were collected using only a 35-msFID acquisition time. The samples were externally referencedt o N H 4‚H 2P O4 at 1.0 ppm.11 Solution 31P N MR sp e c t r a we r eobtained on a Bruker Avance 400 spectrometer at 161.9 MHz.Spectra were collected using a 6- µs pulse based on 90° pulsewidth calibrations w ith 85%phosphoric acid an d a 2-s relax-at ion d e lay .

Hydration Reactions. H y d r at ion r e ac t ions w e r e p e r -formed using freshly boiled and cooled deionized water insidea nitrogen-filled glovebag in the presence of ground LiOH toavoid CO2 contamination. All hydration reactions were carried

out at room t empera ture, typically 22-25 °C . N o at t e mp t wasma de to compensat e for th e exothermic na ture of the rea ctions.

Tricalcium s ilicate (C3S) (1.0 g, 4.4 mmol) a nd w a ter (0.3-

0.45 mL) were added to a plast ic vial a nd st irred wit h a m etalspatula for ca. 20 s. For 29Si MAS NMR investigations, 0.45mL of wat e r wa s u se d and hy d r at ion wa s a l lowe d t o cont inuef or u p t o 2 w e ek s . F o r S E M s t u di es , C 3 S s a m p le s w e r ehy d r at e d w it h 0. 3 mL of wat e r for 24 h and 2 we e ks.

Tricalcium aluminate (C3A) (1.0 g, 3.7 mmol) and water (0.3m L ) w e r e a d d e d t o a p l a s t i c v i a l a n d s t i r r e d w i t h a m e t a lspat ula for ca. 20 s. Sam ples for 27Al MAS NMR investiga tions

were hydra ted inside the N2 bag a nd then immediately packedinto ZrO2 rotors to allow for in situ observation of aluminatehy d r at ion. Samp le s for SE M we r e hy d r at e d for 4 h p r ior t oanalysis. The high-vacuum gold sputtering conditions weresufficient to halt hydration.

St udies on C3A hydra tion in t he presence of excess gypsumwere performed by mixing C3A (0.6 g, 2.2 mmol) an d C aS O4‚

2H 2O (0. 4 g , 2 .3 mmol) and t he n hy d r a t ing w it h w at e r (0.3mL ) and s t ir r ing for c a . 20 s . Samp le s for 27Al MAS NMRinve st ig at ions we r e hy d r at e d insid e t he N 2 b a g a n d t h e nimme diat e ly p ac ke d int o Zr O 2 r ot or s t o a l low f or i n s it uob se r vat ion of e t t r ing it e for mat ion. Samp le s for SE M we r ehydr a ted for 6 h before gold sputterin g. C3A/gypsum sa mplesfor XRD analysis were hydrated for 1, 3, and 24 h.

Ettringite Synthesis. Samp le s of p u r e e t t r ing it e we r eprepared by m odificat ion of th e previously described sy nthe-

sis. 1 Ca(OH)2 (1.59 g, 21.4 mmol) was dissolved in water (1.2L) and m ixed with a solution of CaS O4‚2H 2O (1.374 g, 8 m mol)in water (530 mL) for 1 h, at which point a solution of Al 2-(S O4)3‚18H 2O (1.78 g, 2.7 mmol) in wa ter (770 mL) wa s a dded.The result ing aqueous solution was st irred for 2 h and thenallowed to settle for 1 week. The resulting crystalline precipi-tate was collected by filtration. The formation of ettringite wasconfirm ed by XRD [J CP DS # 41-1451].

Hydration in the Presence of H6ntmp. The as-receivedsolution of H 6ntmp (10 g of a 50 wt %solution) was diluted to100-mL t otal solution volume to approximat e a 5 wt %st ocksolution. Separate solid samples of CaO, Ca(OH)2, and boeh-mite, [Al(O)(OH)], were hydrated with H 6ntmp solutions (0.5m L o f t h e 5 w t % H 6ntmp solution per 1.0 g of solid) for 31PM A S N M R a n a l y s i s . G y p s u m ( C a S O4‚2H 2O ) w a s h y d r a t e dwit h H 6ntmp (2 mL of the 5 wt %H 6nt mp solution per 1.0 g of

solid) for 31P MAS N MR a naly sis . H6ntm p (0.6 g of th e 50 wt% Aldrich solution, 0.1 mmol of H 6n t m p) w a s a d d ed t o asolution of a luminum n itra te (0.67 g, 0.1 mmol) and in wa ter(100 mL ), r e su lt ing in a p H of ab ou t 3. A p r e cipit a t e im-me d iat e ly forme d and w as ana ly zed b y 31P MAS N MR.

All C3A, C3S, a nd cement sa mples for NMR a na lyses werehydra ted by th e method described above using the 5 wt %stocksolution of H 6nt mp (0. 3 mL p e r 1. 0 g of sol id ) in p lac e ofdeionized water to create mineral pastes containing 1.5 wt %H 6nt mp u nle ss ot he r wise not e d . C e me nt samp le s we r e hy -drat ed by th ree different methods for 31P MAS NMR analysis.C e ment w as h y d r at e d wit h a solu t ion of H 6ntmp (0.5 mL per1.0 g of cement of a 2 wt %H 6ntmp solution) to yield a past econta ining 1 wt %H 6ntm p relative to the weight of dry cement.A second sa mple of cement w as hydra ted w ith w at er (0.3 mLper 1.0 g of cement) for 2 min, at which point a solution of

H 6ntmp (0.2 mL of a 5 wt %H 6ntmp solution) wa s mixed intothe past e and a llowed to react for 10 min before NMR ana lysis.The t hir d c e me nt samp le was hy d r at e d wit h a mor e d i lu t esolution of H 6ntmp (0.5 mL per 1.0 g of cement of a 0.6 wt %H 6ntmp solution) to yield a cement pa ste conta ining 0.3 wt %H 6ntmp relative to the weight of dry cement.

C3A sam ples for XPS a na lysis were hydrated w ith the 5 wt% H 6ntmp solution (0.5 mL per 1.0 g of solid) in a N 2 bag for4.5 h, wa shed with a cetone, and filtered. C3S sa mples for XPSana ly sis we r e hy d r at e d wit h t he 5 wt % H 6ntmp solution (0.5mL p e r 1. 0 g of sol id ) in a N 2 b a g f o r e it h e r 7 h a n d t h enwashed with acetone and filtered or hydrated for 1 week andanalyzed without rigorous drying. Cement clinker samples forXP S we r e s imilar ly hy d r at e d for 7 h , washe d wit h ac e t one ,and filtered. Samples for SEM were hydrated using the stock5 wt %H 6ntm p solution (0.3 mL per 1.0 g of solid) and a llowed

(10) Bish op, M. Thesis, Rice U niversity, 2001.(11) Annu al Reports on N M R Spectroscopy ; Academic P ress: New

York, 1986; Vol. 28.

3076 Ch em . M at er ., V ol . 15, N o. 16, 2003 B i sh op et al .



to hydrat e for the following t imes: C3A, 4 h; C3A an d gypsumsamples (0.6 and 0.4 g, respectively), 6 h; C3S, 24 h and 1week; C4AF, 6.5 h; cement, 7 h. Sa mples for liquid 31P N MRwere prepared by hydrating minerals with dilute solutions ofH 6nt mp .

Synthesis of [Ca(H4ntmp)(H2O)]∞‚3.5(H2O)]. To an a que-

ous suspension of ca lcium hy droxide (0.74 g, 10 mmol, in 1 L)w a s a d d e d H 6ntmp (6.0 g of a 50% H 6nt mp mixt u r e, c a . 10mmol) until th e solution clear ed. The solution w as boiled down(t o c a . 100 mL ) and a l lowe d t o c ool t o r oom t e mp e r at u r e ,producing a crystalline precipitate of colorless blocks suitable

for s ing le -c ry st a l X-r ay d iff r ac t ion. This e xp er ime nt wa srepeat ed using 2 equiv of calcium hyd roxide (0.74 g, 10 mmol,in 1 L) to H 6ntmp (3 g of the 50%H 6ntm p mixture, ca. 5 mm ol).The solution never fully cleared. The insoluble material formedw a s a n a l y z e d b y 31P MAS N MR. The solu t ion f i lt r a t e w asboiled down (to ca. 50 mL) and cooled t o room temperature.Unit cell measurements confirmed the sample to be [Ca(H 4-ntmp)(H 2O)]∞‚3.5(H 2O)]. mp: 118 °C (dec). DR IFT IR (cm-1):1642 (s br), 1433 (s), 1314 (w), 1164 (s br), 1073 (s br), 1023(s), 937 (s), 805 (w), 550 (m). 13C C P MAS N MR: δ 55.4 (C H 2),53.7 (C H 2), 51.0 (C H 2). 31P C P M A S N M R : δ 11.7 (1P), 4.41(2P ).

Crystallographic Studies. Data for compounds [Ca(H 4-ntmp)(H 2O)]∞‚3.5(H 2O)] was collected on a Bruker CCD SMARTsystem, equipped with graphite monochromated Mo K R r ad ia-tion ( λ ) 0.71073 Å) and corrected for Lorentz and polarization

effects. The structure was solved using the direct methodsp r og r am XS 12 and d if fe r e nc e F ou r ie r map s and r e f ine d b yusing full-matrix least-squares methods. All non-hydrogenat oms we r e r e f ined wit h anisot r op ic t he r mal p ar ame t e r s .Hydrogen atoms were introduced in calculated posit ions andallowed to ride on the a t ta ched carbon atoms [d (C -H ) ) 0.95Å]. Refinement of posit ional and anisotropic thermal param-eters led to convergence. [Ca (H 4ntmp)(H 2O)]∞‚3.5(H 2O)]: C 3H 19-C a N O13.5P 3, M ) 417.75, monoclinic, a ) 11.252(2) Å, b )

8.475(2) Å, c ) 15.633(3) Å, β ) 90.25(3)° , U ) 1490.8(5) Å3,space group P 21/n (No. 14), Z ) 4, 6715 reflections mea sured,2166 u niq u e of whic h 1452 [F > 4σ (F ) ] w e r e u s e d i n a l lcalculations. The final R was 0.0464 and R w 0.1086, weight ingscheme (SHELXTL parameters) 00401, 0.

Results and Discussion

Reaction of H6ntmp with Calcium Salts. P oly-phosphates can enhance the solubil i ty of calcium hy-droxide in water, both because they reduce the pH ofthe solution and ar e capable of chelat ing Ca 2+.5 I n t hecase of H 6ntmp (I) the enhanced solubility is temporary.Addition of H 6nt m p t o a n aqueous suspension of C a-(OH)2 ini t ial ly results in a clear solution; however, awhite solid precipitates over time.

The soluble [Ca(H n ntmp)](4-n )- species wa s observedwith 31P NMR spectroscopy using dilut e solutions of Ca-(OH)2 and an equi m ol ar or s l i ght exc ess of H 6ntmp.With excess H 6ntmp the solution 31P NMR spectrumshows a tr iplet a t δ ) 9.0 ppm [J (P -H ) ) 12 Hz] while

under equimolar condit ions a broad singlet (δ ) 9.8ppm) is observed. The t ra nsition betw een th ese spectrais as a consequence of the slight increased acidity (pH6 versus pH 7) for solutions with an excess of H 6nt m pan d t he deprotonat ion of the [Ca(Hn ntmp)](4-n )- species.

The ra te of precipita t ion varies w ith the concentra -tions of both Ca 2+ a n d H 6ntmp, a nd th e solution pH. I fdi lut e sol ut ions are em pl oyed, a sm al l qua nt i t y ofcolorless crysta ls is initially formed. In contr a st, use ofmore concentrated solutions leads to rapid precipitationof a pol ycryst a l l ine pow der t hat i s c hemi cal l y and

stru ctura lly identica l to the single crysta ls (see below).From single-crysta l X-ra y d iffraction mea surements t hestructure of this crystal l ine material was determinedto be [Ca(H 4ntmp)(H 2O)]∞‚3.5(H 2O) (Figures 2-4); se-lected bond lengths and angles are given in Table 1.

The calcium atoms in [Ca(H 4ntmp)(H2O)]∞ have oc-tahedral coordination through five phosphonate interac-tions and a coordinat ed w at er [O(1)]. All t he C a -Odistances ar e within the r an ge previously reported forphosphona te compounds [2.305(2)-2.511(1) Å].13-16 Thereare t w o t ypes of phosphonat e m oi et y : one t erm i nal[P (1)] (II ) and two bridging [P(2) and P(3)] (II I).

The P -O distances in uncomplexed [H 4ntmp]2- were

reported t o be divided into t hree groups: the shortest(av erage 1.477 Å) associ at ed w i t h PdO b o n d s , t h elongest (avera ge 1.534 Å) a ssocia ted w ith P -OH bonds,and the intermediate (average 1.501 Å) described atP -O- b onds i n w hi c h t he oxygen c arr i es a negat i v echarge.17 Similar ly, in [Ca (H 4ntmp)(H2O)]∞, the P -OHbonds are the longest [1.560(4)-1.569(4) Å]. However,

(12) Sheldrick, G. M. S H E L X T L ; Bruker AXS, Inc.: Madison, WI,1997.

(13) Lan gley, K. J . ; Squat trito, P . J . ; Adan i, F.; Montoneri, E. Inorg.

C h i m . Ac t a 1996, 25 3 , 77.(14) R udolf, P. R.; C larke, E . T.; Mar tell , A. E.; C learfield, A. Acta

Crystallogr. 1988, C4 4 , 796.(15) Smith, P . H. ; Ray mond, K. N. Inorg. Chem. 1988, 27 , 1056.(16) Ca o, G. ; Lyn ch, V. M.; Swinnea, J . S . ; Mal louk, T. E . Inorg.

Chem . 1990, 29 , 2112.(17) D aly, J . J . ; Wheat ly, P . J . J . Chem. Soc. A 1967, 212.

Figure 2. View of th e sheet stru cture of [Ca (H4ntmp)(H 2O)]∞.Thermal ellipsoids a re shown at the 50%level. All hydrogenat oms and w at er of crystallization are omitted for clarity. P (2)and P(3) bridge two calcium atoms, and P(1) is associated withonly a single calcium atom.




th e uncomplexed PdO bond len gt h [1.488(4) Å] a ppea rsin the range of bond lengths found for the P -O/PdObonds associated with calcium [1.474(4)-1.503(3) Å]. Ont he b asi s of t he c harge b al anc e and t he rel a t i ve P -Odistances, one oxygen per phosphonate group is proto-nat ed along with t he a mine’s nitrogen.15,18 The Ca -O-Pangles a re grouped into t wo ra nges: 135.1(2)-142.9-(2)° and 169.6(2)-170.1(2)°. Although there is no clearcorrelation between the Ca -O-P angl e and ei t her t he

Ca -O or P -O bond dist a nces, there is a good correla tionbetween the Ca -O and P -O bond dista nces; see Figure5.

The structure of [Ca(H 4ntmp)(H2O)]∞ m a y b e d e -scribed as adopting hierarchical level of crysta l a rchi-tecture consisting of two t ypes of dimer (p r i m a r y level,Figure 2) intera ctions, creating a series of plan a r sheets(second ar y l ev el , Fi gure 3) w hi c h are coplana r andconta in intersheet conta cts via h ydrogen-bonded w at ersof crystal l izat ion (t e r t i a r y level, F igure 4). I t is t hisi nfi nit e sheet li ke s t ruct ure t hat i s i m port ant w i t h

(18) (a) Moteka itis, R.; Ma rtell , A. J. Coord. Chem. 1985, 1 4 , 139.(b) Sprankle, P.; Meggitt , W.; Penner, D. Weed Sci. 1975, 76 , 1304.

Figure 3. A planar sheet of [Ca(H 4ntmp)(H 2O)]∞. Atom color scheme: calcium (green), oxygen (red), phosphorus (purple), a nd

nitr ogen (blue).

Figure 4. Crystal packing of [Ca(H 4ntmp)(H 2O)]∞

viewed along the b -axis. All hydrogen atoms and water of crystallization areomitt ed for clarit y. Atom color scheme: calcium (green), oxygen (red), phosphorus (purple), a nd n itrogen (blue).




rega rd t o the function of H6ntmp as a cement inhibitor;see below.

Of t he t w o t ypes of di mer i nt erac t ions show n i nFigure 2, the first may be considered t o consist of tw ocal ci um at om s b ridged b y t w o phosphonat e groups

forming an eight-membered ring analogous to thoseobserved for group 13 phosphona tes19 and phosphi-nat es ,20 as w ell as carb oxyl at es , a m i des, phosphora-mides, and tri f lates.21 Of the remaining phosphonategroups on each [H 4ntmp]2- l igand, one chelat es acalcium in the dimer [via O(11)] and the other bridgestw o a djacent dimeric units [via O(21) an d O(22)]. Thesecond dimeric subunit consists of a H 4nt m p l i gandchela ting one calcium a tom [via O(11) and O(31)] andbridges two other calciums [via O(23) and O(22)]. The[H4ntmp]2- l igand chelates in a wrap-around fashionusing t w o phosphonat e t erm ini , not t hrough a four-membered chelate ring associated with one phosphonatemoiety. This is in contr a st t o the stru cture reported for

[Ca(O3P C H 2NH 2C H 2C O2)]∞(H 2O)2.15,18

As noted a bove, the solid-sta te st ructure wa s deter-mined from a single crystal obtained under slow pre-cipitation conditions. To confirm that the structure of[Ca(H 4ntmp)(H2O)]∞ is r epresentat ive of the calcium-containing product obtained from the reaction under avariety of other concentra t ion condit ions, the X-ra y

powder pattern for a series of samples was comparedw i t h t he pow der pat t ern c al c ul at ed f rom t he s i ngl e-crystal dat a solution. The X-ra y powder di ffra ct ionpattern (Figure 6a) of a polycrystalline powder obtained

from t he react i on of C a( OH)2 w i t h H 6ntmp clearlymat ches (d -spacing a nd r elative intensity) the th eoreti-cal powder pattern generated using XPOW12 from th esingle-crysta l st ructure solution (Figure 6b). Thus, a llthe crysta lline products ar e of the same ma terial, a lbeitwith possible variat ions in wa ter of crystal l izat ion.

Sa mples of [Ca(H4ntmp)(H2O)]∞ yield a 31P MAS NMRspec t rum w i t h t w o peaks, a t δ ) 11.7 and 4.4 ppm(Figure 7a) due to the t erminal (II) and bridging (II I)phosphona te m oieties, respectively. A cryst a lline sa mpleof [Ca(H 4ntmp)(H2O)]∞ was exposed to concentratedN a O H a n d C a ( O H )2 t o det erm ine t he effect on t hestru cture of excess ba se a nd calcium, r espectively. The31P MAS NMR spectrum of [Ca(H 4ntmp)(H2O)]∞ exposed

to a concentra ted Na OH solution for 12 h reveals tha tthe proportion of termina l phosphona te groups (δ ) 11.7ppm, II ) has i nc reased w i t h respec t t o t he b ri dgi ngphosphonate groups (δ ) 4.4 ppm, II I); see Figure 7b.Therefore, the OH - ions compete favorably for calciumi n b asi c sol ut ions and di srupt som e of t he b ridgingphosphonat e intera ctions. In contr a st, a fter exposure ofthe crystals to Ca (OH)2 for 12 h, more bridging moieties(δ ) 4.4 ppm, I II ) are observed relative to the numberof terminal phosphonates (δ ) 11.7 ppm, II ) a n d a na ddit ional pea k nea r 20 ppm is present (Figur e 7c). Thepeak a t δ ) 20 ppm is due to th e [Ca 3(ntmp)] complex,based on a comparison to previous solution 31P N MRda ta which revealed [Ca (ntmp)]4- a t δ ) 18 ppm. In the

(19) Mason, M. R. J. Cl uster Sci. 1998, 9 , 1 and references therein.(20) (a) Ha hn, F. E.; Schneider, B. Z. Naturforsch. B 1990, 4 5 , 134.

(b) Landry, C . C. ; Hynes, A.; Bar ron, A. R. ; H aiduc, I . ; Si lvestru, C.Polyhedron 1996, 15 , 391.

(21) (a) Keys, A. ; Bot t , S . G. ; B arron, A. R. Polyhedron 1998, 17 ,3121. (b) Keys, A.; Barbarich, T.; Bott, S. G.; Barron, A. R. J . C h em .

Soc., Dalton Trans . 2000, 577.

Table 1. Selected Bond Lengths (Å) and Angles (deg) for[Ca(H4ntmp)(H2O)]∞a

Ca(1)-O(11) 2.296(3) C a (1)-O(21A) 2.266(4)Ca(1)-O(31) 2.344(4) C a (1)-O(22B ) 2.324(3)Ca(1)-O (32C ) 2.288(3) C a (1)-O(1) 2.386(4)P(1)-O(11) 1.483(3) P (1)-O(12) 1.488(4)P(1)-O(13) 1.569(4) P (2)-O(21) 1.474(4)P(2)-O(22) 1.487(3) P (2)-O(23) 1.560(4)P(3)-O(31) 1.503(3) P (3)-O(32) 1.483(4)P(3)-O(33) 1.565(4)O(11)-Ca(1)-O(22C) 176.3(1) O(21A)-Ca(1)-C(32B) 175.6(1)

O(31)-Ca(2)-O(1) 174.3(1) Ca (1)-O(11)-P (1) 142. 9(2)Ca(1A)-O(21)-P(2) 169.6(2) Ca(1D)-O(22)-P( 2) 13 5. 1(2 )Ca(1)-O(31)-P( 3) 1 42. 4(2 ) C a ( 1B )-O(32)-P( 3) 17 0. 1(2 )

a S ym m et r y t r a ns for m a t i ons us ed t o genera t e equiv a l ent a t -oms: A (1 - x , 1 - y , 1 - z ), B (1 - x , -y , 1 - z ), C (x + 1/2, 1/2 -

y , z - 1/2), D (x - 1/2, 1/2 - y , z - 1 /2).

Figure 5. P lot of P -O bond (Å) versus C a -O bond length(Å) in [Ca(H 4ntmp)(H 2O)]

∞.

Figure 6. X-ray powder diffraction of (a) polycrysta llineprecipitate from t he reaction of Ca(OH)2 a n d H 6ntmp and (b)the computer-simulated powder diffraction (XPOW) from thesingle-crystal structure solution of [Ca(H 4ntmp)(H 2O)]∞‚3.5-(H 2O).




sol id s t a t e , t he full y deprot onat ed l i gand m ust b eassociated with more calcium ions to balance the nega-t i ve ch a r g e. S a w a d a e t a l . r e por t ed t h a t t h e f u ll ymonoprotonat ed ligand (with t he proton on th e nitrogen)bound to calcium in a fac -chelate fashion in which allof the phosphonate-calcium interactions ar e termina l.22

In oi l f ield applicat ions, cement is hydrated with al imited amount of wa ter. The a mount of wa ter used isreported as weight of wa ter per weight of solid (w/s),with values of 0.3-0.7 w /s being u sed, depen din g on th eapplications. To provide a comparison of the speciation

close to actual “in-field” conditions, it is necessary tochara cterize the rea ction of H6ntm p with calcium oxideand hydroxide under similar conditions. The 31P MASNMR of [Ca(H 4ntmp)(H2O)]∞ wa s used to interpret thespectra of H 6nt m p reac t ed w i t h C aO, C a( OH)2, a n dC a S O4‚2(H 2O) (gypsum) with a limited a mount of wa terpresent . Samples were prepared by wett ing calciumoxide and calcium hydroxide with H 6nt mp (5 wt %) tocreat e pastes w ith t he consistency of freshly hydra tedcement. The 31P MAS NMR spectra for both samplesrevealed peaks a t δ ) 19 and 3 ppm. These peaks ar eal l very broad, suggesting tha t the phosphonates existin a multitude of different environments possibly dueto poor crystallinity. The peak at δ ) 19 ppm is due t o

a ful ly deprotonated phosphonate coordinated to cal-cium, while the peak at ca. 3 ppm is most likely due toa bridging moiety (i .e. , II I). Hydrat i on of gypsum ,C a S O4.2(H 2O), in t he presen ce of H 6ntmp results in theformat ion of [Ca (H 4ntmp)] (δ ) 11.6 and 4.4 ppm).

Reaction of H6ntmp with Tricalcium Silicate.The t ri- a nd d ica lcium silica tes (C3S a nd C 2S, respec-tively) comprise over 80 wt %of most P ortla nd C ements.I t i s know n t hat C 3S i s t he m ost i m port ant phase i ncement for strengt h development during t he first month,wh ile C2S rea cts much more slowly, a nd contr ibutes to

the long-term st rengt h of the cement. 1 B oth the si l icatephases react w ith w at er as shown below to form calciumhydroxide a nd a cal ci um -silicon-hydrat e r i gi d gel ,C -S -H (eqs 1 and 2). 23,24

The 29Si MAS N MR spect rum of anhydrous C 3Sshows five peaks in the range normally associated withS iO 4 species (Figure 8a). Hydration of C3S and subse-quent forma tion of C -S -H is readily observed by 29S iMAS NMR .25 During hydra tion, the a nhydrous si licat espec i es gi v e w ay t o t he hydrat ed di m ers and hi gherpolysilicate chain species found in C -S -H (Figu re 8b,c).The evolution of the dimeric silicate chains is markedby the growth of the broad peak at δ ) -78 ppm.

The C -S -H gel also has a unique morphology thatcan be ea si ly observed by S EM. The a nhydrous gra insappear smooth and are typically a few micrometers widein size (Figure 9a). After hydration, tapered rods of

(22) Sa wa da, K. ; Ka nada , T.; Naganuma , Y. ; Suzuki, T. J . C h em .

Soc., Dalton Trans . 1993, 2557.

(23) The detailed structure of C -S -H is not completely known;however, i t i s general ly agreed upon that i t consists of condensedsi l icate tetrahedra sharing oxygen atoms wi th central , calc ium hy-droxide-like Ca O2 lay ers (consisting of hexagonal layers of octa hedrallycoordinat ed calc ium atoms and tetrahedra l ly coordinat ed oxygenatoms). It i s proposed that the structure is most similar to ei therTobermorite or J ennite, both of which share a skeletal silicate chainstructure, Taylor, H. F. W. J. Am. Ceram. Soc. 1986, 69 , 464.

(24) The hyd ra tion of the calcium silicat es proceeds via four dist inctphases. The f i rst 15-20 min, termed the pre-induct ion period, i sm a r k e d b y r a p i d h e a t e vol u t io n. D u r i n g t h i s p er i od c a lci u m a n dhydroxyl ions are released into the solut ion. The next phase is theinduction period, during w hich time ca lcium oxide continu es to dissolve,producing a pH near 12.5. It has been suggested that , in pure C3S,the induction period may be the length of time it takes for C -S -H t obegin nucleation. Alternatively, the induction period may be causedby t he development of a smal l a mount of an impermeable calcium-

silicon -hydrate (C -S -H) gel at the surface of the part ic les, whichslows down t he migrat ion of wa ter to t he inorganic oxides. As th e initialC -S -H gel is t ra nsformed into the more permeable layer, hy drat ioncontinues and the induct ion period gives way to the third phase ofhydra tion, the accelerat ory period. Ha rdening (setting) occurs near t heend of the t hird period. The fourth st age is t he decelerat ory period inwhich hydration slowly continues until the reaction is complete. Therate of hydrat ion in the deceleratory period is determined ei ther bythe slow migration of water through C -S -H to the inner, unhydrat edregions of the pa rticles, or by t he migra tion of H+ through the C -S -Hto the anhydrous CaO an d SiO2, and the migrat ion of Ca 2+ a n d S i4+ t othe OH - ions left in solution. (a) Ramanchandran, V. S.; Feldman, R.F.; B eaudoin, J . J . Concr ete Science ; Heyden and Son Ltd. : Phi ladel-phia, P A, 1981. (b) Stein, H . N.; S tevels, J . J . Ap p l . C h em . 1964, 14 ,338. (c) Grutzeck, M.; Kwan, S.; Thompson, J . ; Benesi, A. J . M a t e r .

Sci. Lett. 1999, 18 , 217.(25) B ar nes, J . R.; Clagu e, A. D. H.; C layden, N. J . ; Dobson, C. M.;

Hayes, C. J . ; Groves, G. W.; Rodger, S. A. J. M ater. Sci. Lett. 1985, 4 ,1293.

Figure 7. 31P MAS NMR spectra of (a) [Ca(H 4ntmp)(H 2O)]∞‚

3.5(H 2O) showing a 2:1 ra tio of bridging to termina l phospho-nates, (b) [Ca(H 4ntmp)(H 2O)]∞‚3.5(H 2O) exposed to Na OH, a nd

(c) [Ca (H4ntmp)(H 2O)]∞‚

3.5(H 2O) exposed to Ca (OH)2. Spinningsideban ds a re ma rked (*).

Figure8. 29Si MAS NMR of (a) anhydrous C3S, (b) C3S after22 h of hydration, and (c) C3S after 42 h of hydration.

2(CaO)3(SiO2) + 7H 2O f (CaO)3(SiO2)2‚4(H 2O) +

3Ca(OH)2 (1)

2(CaO)2(SiO2) + 5H 2O f (CaO)3(SiO2)2‚4(H 2O) +

Ca (OH)2 (2)




C -S -H grow from the C3S grains. 26 Over time, these

needles become more int erlocked a nd t a ke on a foil-likemorphology and eventually C -S -H connects adjacentparticles (Figure 9b).

The hydr at ion of C3S in the presence of H 6ntmp (1.5w t %) w a s f ol low e d b y S E M , N M R , a n d XP S . S E Mimages of C3S hydrated for 24 h in the presence of H 6-ntm p (Figure 9c) do not a ppea r a ppreciat ively differentfrom an nonhydrat ed sa mple (Figure 9a). The la ck ofhydration is confirmed by 29Si NMR spectroscopy, w hichreveal s t ha t i n t he presence of H 6nt m p no di m eri c

silicate species (i.e. , no peak a t -78 ppm) are formed,even after 2 w eeks of h ydra tion. This should be com-pared to the spectrum of C3S hydra ted without reta rd-ers , i n w hi c h a s i gni f i c ant am ount of C -S -H formsafter just 1 da y of hydra tion (Figure 8b).

XPS analysis was performed on C3S samples, withand without H 6ntm p (1.5 wt %), hydr a ted for 7 h a nd 1week for both samples. The Ca:Si ratio for a sample ofC3S cha nges from 2.6 to 2.4 during th a t t ime, consistent

with the formation of C -S -H (eq 1). The Ca:Si ratio int h e s a m p l e h y d r a t e d i n t h e p r e s e n c e o f H 6n t m p i sinitially low (1.8); however, it reached a value of 3.0 after1 week of hydra tion. The low ini t ial Ca :Si ra t io couldbe consistent with either a calcium complexation ornucleation poisoning mechan ism; however, t he highfinal C a :Si ra tio does not fit either model. The increasein the Ca :Si ra tio is consistent with the precipitat ion ofa calcium-rich phase such as the phosphonate complex.The 31P MA S N MR spec t ra of C 3S hydrat ed w i t h alimited a mount of wa ter (0.3 w/s) and H 6ntmp (0.75 wt%) revea ls a lar ge peak a t 19 ppm, due t o [Ca 3(ntmp)],and several smaller peaks spanning the region from 10to 1 ppm associat ed with both t erminal (II ) and bridging

(II I) phosphonat es.Reaction of H6ntmp with Tricalcium Aluminate.

Although the aluminate and ferrite phases comprise lesstha n 20%of the bulk of cement, th eir reactions ar e veryimportant in cement and dramatically affect the hydra-tion of the ca lcium silica te pha ses. Relative to C3S, th ehydration of C3A [(CaO)3(Al2O3) ] i s v ery fas t . I n t heabsence of any a dditives, C3A reacts w ith wa ter to formtw o intermediat e hexagonal pha ses, C2AH8 a nd C 4AH13(eq 3).27 C 2AH8 and C 4AH13 a re m et ast ab l e phasesthat spontaneously transform into the ful ly hydrated,thermodynamically stable cubic phase, C3AH6 (eq 4).

The al um inum coordi nat i on i n C 3A i s di st ort edtetrahedral, whereas the aluminum in C3AH6 exists inhighly symmetrical, octahedral [Al(OH)6]3- uni t s a l -lowing for hydration to be conveniently followed by 27AlNMR spectroscopy. Although a luminum is spin 5/2 a ndin general exhibi ts a large q uadr upolar coupling con-sta nt , severa l dist inct species can be observed duringhydration.28 The chemical shifts of the hexa gonal pha ses,

C2AH8 a nd C 4AH13, lie within 10 ppm of C3AH6 a ndthe l ines due to a l l of these phases a re broader tha n 10ppm at both 52.1 and 130.3 MHz; however, the transi-t ion from t he hexagonal pha ses to the cubic phase canbe observed as an upfield shift even on a 200-MHz (27Al

(26) The C -S -H takes on a more spiked, needlel ike appeara nceas it becomes partially dehydrated under the high-vacuum conditionsnecessary for SE M imaging.

(27) The structure of C2AH8 is not precisely known, but structuresconta ining [Ca 2Al(OH)6][Al(OH)4]‚n H 2O or [Ca 2Al(OH)6][Al(OH)3(H2O)3]-OH have been proposed. C4AH13 has a layered structure based onthe calcium hydroxide structure in which one out of every three Ca 2+

i s replaced by an Al 3+ or Fe3+ w i t h a n O H - anion in the interlayerspace to balance the charge. All of the aluminum centers in C4AH13are octahedral .

(28) (a) Skibsted, J . ; Hendricson, E.; J akobsen, H. J . I norg. Chem.

1993, 3 2 , 1013. (b) Fa ucon, P .; Cha rpentier, T.; Bert ra ndie, D.; Nonat,A.; Virlet, J .; P etit , J . C. Inorg. Chem. 1998, 37 , 3726.

Figure 9. SE M ima ges of (a) an hydrous C3S, (b) C3S a fter 2weeks of hydration, and (c) C3S a fter 24 h of hydrat ion in t hepresence of H 6nt mp (1.5 w t %).

2(CaO)3(Al2O3) + 21H 2O f

(CaO)4(Al2O3)‚13(H 2O) + (CaO)2(Al2O3)‚8(H 2O) (3)

(CaO)4(Al2O3)‚13(H 2O) + (CaO)2(Al2O3)‚8(H 2O) f

2(CaO)3(Al2O3)‚6(H 2O) + 9(H 2O) (4)




observed frequency 52.1476 MHz) spectrometer. Thehi ghl y di s t ort ed anhydrous t et rahedral a l um i num i sdifficult to observe on a 200-MHz instrument, but isreadi ly ob serv ed a t 500 MHz (ob serv ed frequency

130.3187 MHz). At 52 MHz, t he t etrah edral phase isnot observed; however, the growth of the intermediatehexagonal phases C2AH8 and C4AH13 shown by thesignal near δ ) 8 ppm and the subsequent downfieldshi f t o f t h i s peak t o δ ) 11 ppm to indicate C3AH6forma tion is readily apparent . In contra st , a t 130 MHz(Figure 10) the tra nsi t ion from a tetra hedral pha se (δ) 40-80 ppm) to an octahedral phase (δ ) 8 ppm) a ndthen a downfield shift to the highly symm etrical C 3AH6species is clearly observed.

The hy dra tion of C3A in t he presence of H 6ntmp (1.5wt %) wa s monitored using 27Al MAS spectroscopy. Ona 200-MHz spectrometer, no tetra hedral aluminum isobserved. The octahedral species appears at 6 ppm,

sl ightly upfield of the normal shift expected for thepart ial ly hydra ted, hexagonal calcium a luminat e inter-mediates. This peak continues to grow in for the firsthour and then shifts sl ightly to δ ) 7.8 ppm after 2 h,suggesting the formation of C2AH8 and C4AH13 (cf. ,Figure 10b). After 3 days of hydration the spectrumshow s a s li ght i ncrease i n s ignal i nt ensi t y , b ut noC3AH6 is observed. This is in contrast to the spectraobserved for C3A hydr at ion in t he a bsence of addit ives,wh ich shows significa nt C 3AH6 format ion a fter just 30min.

C3A hydration with H 6ntm p (1.5 wt %) wa s invest i-gated by 31P MAS NMR spectroscopy. Pea ks a t δ ) 18.4and 3.4 ppm were observed, a spectrum most similar

to that observed for calcium hydroxide hydrated with1.5%H 6ntm p. No peaks due to aluminum phosphonat ecomplexes wer e observed,10 indicating tha t under theseconditions, reaction with calcium is favored over alu-minum.

The hy dra tion of C3A in t he presence of H 6ntmp (1.5w t %) w a s a l so f ollow ed b y S E M . Af ter 4 h , t h edevelopment of a foil-type layer on the C3A particles isobserved, similar to that observed for the hydration ofpure C3A but with m uch smaller feat ures. XP S a na lysisof C3A hydr a ted for 4 h in the presence of H 6ntmp (2.5w t %) reveal s a P : C a ra t i o i s 1 . 6 (i .e ., c a . 2 C a permolecule of H 6ntmp). In a ddit ion, there is a change inthe Ca :Al ra t io from 0.9:1 in t he a nhydrous mineral to

11.6:1. This should be compared to the Ca:Al ratio of

1.7:1 observed for the uninhibited paste. This resultindicates that aluminum is buried under a Ca -P-richlayer.

Reaction of H6ntmp with Tricalcium Aluminatein the Presence of Gypsum. I f t h e v e ry r a p id a n dexothermic hyd rat ion of C3A is al lowed to proceedunhindered i n cem ent , t hen t he set t i ng occurs t ooqui ckly and t he cem ent does not develop s t rengt h.Therefore, gypsum [calcium sulfate dihydrate, CaSO4‚

2(H 2O) ] i s added t o s l ow t he C 3A hydrat i on. I n t hepresence of gypsum, tricalcium aluminate forms theneedlelike mineral ettringite, [(CaO)3(Al2O3)(CaSO4)3‚

32(H 2O)] (eq 5). The st ructure of ettr ingite consists ofcolumns of calcium, a luminum, a nd oxygen surr ounded

b y w at er and sul fa t e i ons , as show n i n Fi gure 11. I finsufficient sulfat e is present , th en t he et tringi te w illeventua lly react w ith excess C3A to form the monosul-fat e, C3A‚C a S O4‚12(H 2O) (eq 6). However, th e rea ctions

tha t a ffect hy dra tion inhibition occur dur ing the induc-

tion period, and the monosulfate forms after the end ofthe induction period. Thus, we have assumed that thisl as t s t ep i n t he hydrat i on of C 3A in t he presence ofgypsum is not important to the investigat ion into themechanisms of hydration inhibition.

The 27Al MAS of C3A hydrated in the presence ofexcess gypsum indicat es tha t the et t ringi te forma tionproceeds via partially hydrated calcium aluminate spe-cies, possibly C2AH8 and C4AH13. The formation ofettringite may also be seen from th e growth of crystal-line rods from the surface of C3A (Figure 12). The rodsgenera lly ran ge from 1 to 2 µm in length a nd a re about10-20-nm w ide when gr own in situ fr om pastes of C3Aand gypsum. Phase conformation of et tringi te is ob-

Figure 10. 27Al MAS NMR at 130.3187 MHz of C3A (a) and(b ) a f t e r 30 min hy d r at ion. Sp inning s id eb and s ar e mar ke d(*).

Figure 11. Ettringite columns (a) consisting of octahedralaluminum, tetrahedral oxygen, and 8-coordinate calcium. Thecoordina tion sphere of each calcium is completed by wa ter a ndsulfate ions. The pa cking of the columns (b) represented bylar ge circles; the sma ller circles represent chan nels conta iningwat e r and su lfa t e ions.

(CaO)3(Al2O3) + 3 C a S O4‚2(H 2O) + 26H 2O f

[(CaO)3(Al2O3)(CaSO4)3‚32(H 2O)] (5)

[(CaO)3(Al2O3)(CaSO4)3‚32(H 2O)] +

2(CaO)3(Al2O3) + 4H 2O f

3(CaO)3(Al2O3)(CaSO4)‚12(H 2O) (6)




tained from its low-angle powder X-ray tag (d ) 9.8 Å)tha t al lows i t to be easi ly dist inguished from residualcalcium sulfate (d ) 7.7 Å).

The effect of H 6ntmp on the h ydra tion of C3A in thepresence of gypsum w a s investigat ed using 27Al and 31P

MAS NMR and XRD. The 31P MAS NMR spectrum ofC3A hydrated in the presence of excess gypsum with adilute H 6ntmp solution is markedly different from thecompar able spectru m wit hout gypsum. In t he presenceof excess gypsum, there is a small peak at δ ) 16 ppmand a larger peak at δ ) 6 ppm with a second maximumnear δ ) 3 ppm and a broad shoulder th at extends outt o -10 ppm, the lat ter feat ure suggesting tha t complex-at ion t o aluminum should not be ruled out .29

27Al MAS NMR of C3A hydrated in the presence ofgyspum and 1.5 wt %H 6ntmp reveals initia l accelerated

forma tion of ettr ingite, followed by a long period of slow reactivi ty.30 Figure 13 compar es the hydra tion of C3Aw i t h e x ce ss g y ps u m w i t h a n d w i t h ou t t h e H 6nt m p

solution (1.5 w t %) dur ing t he first 30 min of hyd ra tion.The 10-min spectr um of C3A wit h gyps um a nd H 6nt m pshow s et t r ingit e (cent ered a t δ ) 13 ppm ) a n d apart i a l ly hydrat ed cal ci um al um inat e (C 2AH8 andC4AH13, δ ) 8 ppm) present in a bout a 1:1 ra tio (Figure13a). This spectrum remains unchanged after 30 min.In contra st , the spectra in the a bsence of the reta rder(Fi gure 13b ) show t he t ransi t i on from t he hydrat edcalcium aluminate intermediate (δ ) 8 ppm) to ettring-i te (δ ) 13 ppm). After 30 min, the spectrum taken inthe absence of H 6ntmp shows almost complete conver-sion to ettringit e, while the spectru m wit h H 6ntmp stillshows a significant amount of part ial ly hydrated cal-cium aluminat es.

Reaction of H6ntmp with Tricalcium Alumino-ferrite. Tetracalcium aluminoferrite (C4AF) reactsmuch like C3A, i .e., w i t h t h e f o r m a t i o n o f a n i r o n -substituted ettringite in the presence of gypsum. How-ever, hydra tion of the ferrite phase is much slower t ha nt hat of C 3A.31 I n a d d it i on , i f i n a ce me nt t h er e i si nsuf fi ci ent gypsum t o conv ert a l l of t he C 4AF t oettringite, then an iron-rich gel forms at the surface of

the si licat e pa rt icles, wh ich is proposed t o slow theirhydration.32

SEM i m ages of part i a l l y hydrat ed C 4A F show analuminat e foil forms a t the surfa ce of C4AF part icles,which closely resembles the early hydration of C3A.SE M a lso clear ly shows the development of et tringi tefrom C4AF pastes hydrated with gypsum. C4AF hydra-tion in the presence of H 6ntmp was fol lowed by SEM.C 4AF hydrat ed for 6. 5 h i n t he presence of H 6nt m pconta ins smooth part icles tha t a re not connected. C4AF

hydration, in the presence of gypsum, results in theformat ion of ettringit e is confirmed by SE M (Figure 14a)and XRD. In contrast , in the presence of H 6nt m p t heC 4AF p a r t icl es a p pe a r t o h a v e a t h in , a m or ph ou scoa t i n g a n d con t a i n p la t e li ke a n g u la r s h a pe s t h a ta p pe a r t o h a v e t h e s a m e coa t i n g (F i gu r e 14b ). I nsummary, t he addit ion of H 6ntmp to C4AF completelyinhibits hydration, even in the presence of gypsum.

Cement Hydration in the Presence of H6ntmp.The hydration of cement is obviously far more complextha n th e sum of the hydra tion reactions of the individualm i neral s . Fi gure 1 depi ct s a t ypical cem ent grai n ,showing the larger silicate particles surrounded by themuch smaller C3A and C4AF particles. The setting of

cement ca n be broken down into severa l distinct periods.The more reactive aluminate a nd ferri te pha ses reactfirst , a nd t hese reactions dra mat ical ly a ffect the hydra -tion of the silica te phase. Scrivener a nd P ra tt used TEM

(29) (a) Zakowsky, N.; Wheat ley, P. S . ; Bul l , I . ; At t f ield, M. P. ;Morris, R. E . D a l t o n T r a n s . 2001, 2899. (b) Edgar, M.; Carter, V. J . ;G r e w a l , P . ; S a w e r s, L . -J . ; S a s t r e , E . ; Tu n s t a l l, D . P . ; C o x, P . A. ;Lightfoot, P.; Wright, P. A. Chem. M ater . 2002, 14 , 3432.

(30) The ra pid formation of et tringi te wi th the dudra t ion of C3Aand gypsum in th e presence of H 6ntmp is confirmed by XRD.

(31) This di f ference in hydra t ion rate ma y be due to th e fac t th ati ron is not a s free to migrate in the past es as a luminum, which maycause the formation of a less permeable iron-rich layer at the surfaceof the C4AF particles and isolated regions of iron hydroxide. (a) Mehta,S.; J ones, R.; Caveny, B.; Chatterji , J . ; McPherson, G. En v i r on m e n t al

Sca n n i n g El e ct r o n M i c r osc op e ( ESEM ) ex a m i n a t i o n o f i n d i v i d u a l l y

h y d r a t ed Por t l a n d C em en t p h a ses ; H a l l ib u r t on R e se a r ch C e n t er :Duncan, OK. (b) Taylor, H. F. W.; Newbury, D. E . Cem. Concr. Res .1984, 14 , 565.

( 3 2 ) R a m a c h a n d r a n , V . S . C o n cr et e Ad m i x t u r es H a n d b o ok , 2 nd

Ed i t i o n ; Noyes Publicat ions: Pa rk Ridge, NJ , 1995.

Figure 12. SEM image of C3A hydrated in the presence ofgypsum showing the formation of ettringite.

Figure 13. 27Al MAS NMR spectra of C3A (0.6 g) and gypsum(0.4 g) hydra ted for 10, 15, a nd 30 min, (a) w ith H 6ntmp (1.5

w t %) a n d (b ) w i t h ou t H 6n t m p, s h ow i n g t h e r a p id b utincomplete forma tion of ettringite (δ ) 13 ppm) in t he presenceof the “retarder” H 6nt mp .




to develop the w idely a ccepted model depicted in F igure15. 33

In the first few minutes of hydration (Figure 15b), the

aluminum an d iron pha ses react with gypsum to forman amorphous gel at the surface of the cement grainsa nd short r ods of ettring ite grow. After t his initial periodof reac t i v i t y , c em ent hydrat i on s l ow s dow n and t heinduction period begins. After about 3 h of hydration,the induction period ends and the acceleratory periodbegins. During t he period from 3 to 24 h, a bout 30%ofcement reacts to form calcium hydroxide and C -S -H .The dev elopment of C-S -H in t his period occurs in t wophases. After ca. 10 h of hydration (Figure 15c), C3Shas produced “outer C -S -H,” which grows out from theettringite rods rat her tha n directly out from the surfa ceof the C3S particles. Therefore, in the initial phase ofthe rea ction, the silicate ions must migra te through th e

aluminum-and iron-rich phase to form the C -S -H. Inthe lat ter pa rt of th e accelera tory period, a fter 18 h ofhydration, C3A continues to react with gypsum, forminglonger et tringi te rods (Figure 15d). This network ofet t r i ngi t e and C -S -H appears t o form a “hydrat i ngshell” about 1 µm from the surfa ce of a nhydrous C3S.A sm al l am ount of “ i nner C -S -H” forms inside thisshell. After 1-3 days of hydrat ion, reactions slow an dthe deceleratory period begins (Figure 15e). C3A reactswith ettr ingite to form some monosulfat e. “Inner C -S -

H” continues to grow nea r t he C3S surfa ce, na rrowing

the 1- µm gap between the “hydrating shell” and anhy-drous C3S. The ra te of hydra tion is likely to depend ont he di f fusi on rat e of w at er or i ons t o t he anhydroussurfa ce. After 2 weeks of hydra tion (Figur e 15f), the ga pb et w een t he “hydrat i ng shel l ” and t he grai n i s c om -pletely filled with C -S -H. The original, “outer C -S -

H” becomes more fibrous.

Cement hydration can be monitored by 29Si MAS,albeit w ith lower resolution t han the pure C3S past es

due to the presence of additional silicate phases (C2S)an d pa ra ma gnetic iron-conta ining minerals. A compari-son of t he spect ra i n Fi gure 16 rev eal s ext ensiv eC -S -H format ion (δ ) -78 ppm) during t he first d ayof hydration. This should be compared to Figure 8,which shows the hydration of the pure C3S paste. Incontr ast , cement hydra ted for 12 da ys with H 6nt mp (1.5wt %) reveals only a sma ll amount of C-S -H format ion,as indicated by th e shoulder a t -78 ppm (Figu re 16c).It is the early reactions in the hydration of cement thatare crucial to control l ing the length of the inductionperiod. B y the t ime a significa nt a mount of C-S -H h a sbeen formed a nd can be seen by th e emergence of a pea ka t -78 ppm in the 29Si MAS NMR (i.e. , Figure 16b),

the induction period is over. Thus, w ith r egard to th esi licat e pha ses, the a ddit ion of H 6nt m p result s i n t hesignif icant ret ardat i on of t he hydrat i on process , i npar ticular t hrough t he inhibition of the induction period.

27Al M AS N MR ca n b e u s ed t o m on it or ce me nthydration. Unhydrated cement shows only 1 peak at 78ppm, due to the tetr ah edral a luminum in C4AF (Figure17a), since the tetrahedral aluminum centers in C3Aare too distorted for observation on a 200-MHz spec-trometer. During hydration, the observed peak trans-f or m s t o t h a t of t h e oct a h e d ra l ca l ci um a l u mi na t ephases (including C3AH6; δ ) 11 ppm); see F igure 17b.This transition from tetrahedral aluminum (ca. 80 ppm)to octahedral aluminum near (ca. 11 ppm) can be usedto est imate the extent of cement hydra tion. In cementpastes hydra ted without inhibitors, the tra nsforma tionto the octa hedra l calcium alumina te pha ses is completewell before the end of 2 weeks. Cement hydration withH 6ntm p is severely inhibited, an d only a fra ction of thetotal aluminum content has been hydrated even after2 weeks (Figure 17c). The addition of H 6ntm p does nott ot a l ly s h ut d ow n t h e h y d r a t ion of t h e a l u mi nu mconta ining minera ls, as occurs for C3S, but the hydr a-tion rate is clearly much reduced.

31P MAS NMR of cement hydrated with H 6ntmp (1wt %) reveals t wo broad pea ks nea r δ ) 18 and 4 ppmdue to the terminal (II ) an d bridging (II I) phosphonat e

moieties, respectively (Figure 18a). Under these condi-t ion s t h e r a t io of t er m in a l (II ) a n d b rid gin g (II I)phosphonate is approximately 2:1. This is the inverseof that found in crystal l ine [Ca(H 4ntmp)(H 2O)]∞. Thera tio of these peaks is dependent on th e exact hydra tionconditions.

A sam pl e of c em ent w as f i rs t hydrat ed for 2 m i nwithout H 6ntmp to al low ini t ial hydration of the alu-m i nat e phases , and t hen H 6ntmp (1 wt %) wa s a ddedalong with addit ional water. This delayed addit ion ofthe inhibi tor is actual ly a common practice in the oi lwell cementing industry, as many retarders are foundto retard cement for longer with delayed addition. Theresult ing spectrum reveals that more of the bridging

(33) S crivener, K. L.; Pra tt , P . L. M ater. Res. Soc. Symp. Proc . 1984,31 , 351.

Figure 14. S EM ima ges of C4AF (a) hydra ted in the presenceof excess gypsum an d (b) hydrat ed in the presence of excessg y p su m and H 6nt mp .




phosphonate species formed (Figure 18b). The phospho-nic acid, H 6ntmp, is known to be a very a ct ive retardera t very low weight percenta ges, ca . 0.1-0.3%. Therefor e,

a t hird cement sa mple was hydr at ed with less H 6nt m p(0.3 wt %), a nd the spectrum reveals tha t the r at io of

Figure 15. Schematic representation of anhydrous cement (a) and the effect of hydration after (b) 10 min, (c) 10 h, (d) 18 h, (e)1-3 da ys, a nd (f) 2 weeks (Adapted from K. L. S crivener, P h.D. Thesis, Un iversity of London, 1984).

Figure 16. 29S i M AS N MR of (a ) d r y C l a ss H P o rt l a n dCement, (b) after hydration for 24 h, and (c) after hydrationfor 12 days in the presence of H 6nt mp (1.5 w t %). Figure 17. 27Al MAS N MR (52. 1 MH z) of (a) a nhy d r ou s

cement, (b) cement hydrated for 2 weeks, and (c) a samplehydrated for 2 weeks in the presence of H 6nt mp (0.5 w t %).




t erm inal t o b ridging phosphonat es (Fi gure 18c) i ssimilar to tha t found in crysta lline [Ca (H 4ntmp)(H 2O)]∞(Figu re 7).

B ecause t he low concentra t ion and delayed addit ionconditions are those under which the phosphonic acidis known to be more act ive, Figure 18c should moreaccurately reflect the speciat ion in the field. On thebasis of 31P N MR dat a , w e propose t hat a phosponat ecomplex (or series of complexes) with a structure notunlike that of [Ca(H 4ntmp)(H2O)]∞ is most l ikely re-sponsible for the retardation of cement sett ing. Fur-thermore, since the delayed a ddition of the phosphonica c id a l low s f or s om e C 3A a n d C 4AF t o r ea c t w i t hgypsum to form ettr ingite and t he amorphous gel at the

surface of the cement gra ins, this mat erial may directthe heterogeneous precipitation of the active calciumphosphonate creating a coating, retarding further hy-dration.

SE M ma y be used to observe the sur face morphologyduring hydrat i on. Anhydrous cem ent grai ns appearsmooth and relatively featureless, with particles rangingi n s i z e f rom c a . 1-50 µm . A ft er hydrat i on, C -S -Happears as a coat i ng on t he grai ns w i t h som e c harac -teristic C -S -H rodlike regions a nd some sma ll crysta lsof calcium hydr oxide present.

X P S anal ysi s of c em ent hydrat ed w i t h H 6nt m p re-veals almost no a luminum a t t he surface of the gra ins.

The Ca:Al ratio increases from 15:1 for the anhydroussample to 78:1 for t he sa mple hydra ted w ith H 6ntmp.This should be compared to the Ca:Al rat io of 25:1observed for t he cement sa mple hydra ted w ithout H 6-ntm p. In ad dition, the Ca :Si ra tio is 3.8:1, exactly tw icethe rat io of 1.9:1 found for cement hydrated withoutinhibitors. The a mount of H 6ntmp used (2.5%relativeto the w eight of cement) results in a Ca :P ra t io of 2:1.Thus, in the presence of H 6ntmp the surfa ce of hydra tedcement gra ins is rich in calcium a nd phosphorus, a nddeficient in silicon and aluminum, consistent with theformation of a calcium phosphonate coating.

The extra calcium found at the surface supports adissolution/reprecipita t ion mechanism (see below),

w hereb y t he H 6ntmp promotes the dissolution of cal-cium from the minera l phases a nd t hen reprecipita tesa C a p hos ph on a t e s pe ci es on t o t h e s u rf a ce of t h ehydrat i ng gra i ns.

Conclusions

The reaction of H 6ntmp w ith calcium sal ts r esults inthe initial formation of soluble [Ca(H n ntmp)](4-n )-, wh ich

precipita te over time a s [Ca(H4ntmp)(H2O)]∞. The ra tioof bridging to termina l phosphonat e groups is dependenton the pH (terminal phosphonate groups increase withincreased pH) and the calcium content ([Ca 3(ntmp)] isformed at high Ca:H 6ntmp ratios).

F or t h e h y d ra t i on of C 3S i n t h e p re se nce of t h ephosphonic a cid, 29Si N MR i ndicat es t hat no C -S -Hforms, while XPS reveals a Ca:Si rat io begins Si-richan d then chan ges to Ca-rich, an d 31P MAS NMR revealstha t the pa ste conta ins var ious calcium phosphonates.For the hydration of C3A in the presence of H 6ntmp,27Al MAS NMR reveals that the hydration of C3A isseverely inhibited; 31P MA S N MR dat a i ndi c at es t hatt h e H 6n t m p i s r ea c t in g p ri ma r i ly w i t h ca l ci um a s

opposed to aluminum; and XPS dat a r eveal tha t a Ca -

P -rich layer buries the a luminum a t t he surfa ce of C3A.H 6ntmp enhances calcium solubili ty, promoting thedissolution of calcium from anhydrous mineral phasesa nd forming a meta sta ble soluble calcium phosphona tespecies. However, once formed, th e Ca -P complexes arenot very soluble an d precipitat e almost immedia tely inbasic solutions. P recipitat ion of t hese complexes formsa layer on t op of th e hydra ting C3A part icles, buryingthe a luminum-rich la yer, creat ing a higher Ca :Al ra t ioat the sur face. For the hydr at ion of C3A in the presenceof H 6ntmp and excess gypsum, (a) 27Al MAS NMR showsfast i n it i a l form at i on of e t t r ingit e ac com panied b yanother hydra ted calcium a lumina te (most probably

C2AH8 an d C4AH13), followed by a period of very slow rea ctivity in w hich more ettringite slowly forms; (b) 31PMAS NMR reveals t ha t less of the species a t 19 ppm isformed relative to pastes hydra ted without gypsum, [Ca -(H 4ntmp)] may be present , a nd some H 6nt m p m ay b ecoordinated t o aluminum. It ha s been proposed tha t t hephosphonic acids inhibited t he tr an sforma tion of gelat i-nous et tringi te to crystal l ine et tringi te by adsorbingont o t he surfac e of t he grai ns .9 O n t h e b a s i s o f t h eforegoing, such a mechanism is not likely since H 6nt m pappears to cata lyze, rat her tha n inhibit , the formationof ettringite.

For c em ent hydrat i on w i t h H 6ntmp, (a) 27Al MAS

NMR shows that only hal f of the aluminum has beenconv ert ed t o t he hydrat ed, oct ahedral form af t er 2weeks; (b) 29S i M A S N M R a n d S E M r e v e a l t h a t n oC -S -H is formed, although some ettringite is formed;(c) 31P MAS NMR shows tha t H 6ntm p prefers t o complexcalcium, and i t is l ikely that [Ca(H 4ntm p)] is formed;(d ) XP S r ev ea l s t h a t a C a - a n d P -r ich l a y er f or m s ,burying the a luminum.

Dissolution/Reprecipitation: An AlternativeMechanism for Cement Hydration Inhibition.Therear e four primary models for cement hydra tion inhibi-tion: calcium complexat ion, nucleation poisoning, sur-face a dsorption, a nd protective coating/osmotic burst -ing.1

Figure 18. 31P MAS NMR spectra of cement hydrated with(a ) 1.0 wt %H 6ntm p, (b) delay ed add ition of 1.0 w t %H 6nt mp ,an d (c) 0.3 wt %H 6nt mp .




Ca lcium complexat ion involves either removing cal-cium from solution by forming insoluble salts or chelat-ing calcium in solution to prevent C -S -H formation.34

However, if the retarder were acting solely by calciumcomplexation, then one molecule of retarder would berequir ed per calcium ion in solution, an d good inhibitorsare used in much smaller quanti t ies, on the order of0.1 -2 wt %of cement. In addit ion, there is no simplecorrelat ion betw een either ca lcium binding str ength or

calcium salt solubility and reta rding a bility. 35 Yet it ha sbeen shown that , in pure systems, that is , of C3S andC2S, the l ime concentrat ion in solutions is the mostimportant factor in determining t he precipitat ion ofC -S -H .36 Therefore, although calcium complexationmust play s ome role in inhibition, other mechanisms ofinhibi t ion must be at work a s well .

Inhibi t ion by nucleation poisoning is w here th e re-ta rder blocks the growt h of C-S -H or Ca(OH)2 crystalsthrough inhibiting agglomerates of calcium ions fromforming the necessary nucleites. Nucleation inhibitorsact on th e surface of sma ll clusters; therefore, less tha none molecule of retar der per ca lcium ion is required toproduce dram at ic results.37 In a relat ed manner, surface

adsorption of inhibi tors directly onto the surface ofeit her t he anhydrous or (m ore l ikel y) t he part i a l lyhydrated mineral surfaces blocks future reactions withw a t e r .

The fina l possible mecha nism for hyd ra tion inhibitionwas original ly posi ted to explain the existence of theinduction period in C3S and cement hydration in the

ab senc e of ret arders ; how ev er, i t m ay b e appl i ed t oinhibition in general. In this mechan ism, a semiperme-able layer at the surface of the cement gra in forms a ndslow s dow n t he m i grat i on of w at er and l engt hens t heinduction period. Osmosis will drive wa ter t hrough th esemipermeable membrane toward the unhydrated min-eral , and eventual ly the flow of water creates higherpressure i nside t he prot ect i v e c oat i ng and t he l ayerb ur s t s. H y d r a t i on i s t h e n a l low e d t o c on t in u e a t a

normal rate.38 This final m echa nism is undoubtedly th emechan ism by ma terials such a s l igosulfat e operate.

The a bility of phosphonic acid reta rders to a ct simul-taneously in several different capacities would explaintheir increased efficiency over more conventional re-t arders . On t he b asi s of t he foregoi ng result s , a new model of hydr a tion inhibition is shown in Figure 19. Inthis model, cement is modeled as a lar ge calcium silica tepart icle in a C3A/C4AF mat rix. U pon hydra tion, H 6-ntmp first promotes th e dissolution of calcium, creat inga m et ast ab l e solub le C a phosphonat e com plex a ndal lowing water and gypsum to at tack the aluminum atthe surfa ce of the cement grains a nd form ettringite.

The phosphonic acid does not a ct by simple ca lciumcomplexat ion. I f t his were the case, then a far greateramount of H 6ntmp would be necessary to produce thesame dra mat ic results tha t a re observed when H 6nt m pis used to inhibit cement hydr a tion. On the bas is of ourinvestigat ion of the reaction of H 6nt m p w i t h v ari oussoluble ca lcium sa lts, we propose tha t t he Ca phospho-na te complex then undergoes a r earra ngement (perhapsa s s h ow n i n F i gu r e 20) a n d p re ci pi t a t es on t o t h ep a r t ia l ly h y d ra t e d s u rf a ce s. I t i s p os s ib le t h a t t h ecalcium phosphonates prefer to precipitate directly ontothe hydr at ing mineral pha ses, thereby blocking future

(3 4) R a m a c h a n d r a n , V. S . ; F e ld m a n , R . ; B e a u d oi n , J . Concrete

Science: Tr eati se on Cur rent R esearch ; Heyden and Son. Ltd. : Phi la-delphia, 1981.

(35) Young, J . F. Cem. Concr. Res. 1961, 2 , 415.(36) Ga rraul t-Ga uff inet , S . ; Nonat , A. J . C r y st . G r o w t h 1999, 2 00 ,

565.(37) Thomas, N.; B irchall , J . Cem. Concr. Res. 1983, 13 , 830. (38) D ouble, D . Philos. Trans. R. Soc. London 1983, A 30 , 53.

Figure 19. Schematic representation of the H 6ntmp inhibit ion of cement showing (a) t he phosphonic a cid promoting calciumdissolution, allowing water and gypsum to react with the aluminum phases at the surface of the cement grain, (b) the formationof a meta sta ble calcium phosphonat e, which precipitates onto the hydra ting Al surfaces (c), forming a ba rrier to wa ter a nd sulfat ediffusion.




reactions of these phases with water. Heterogeneous

precipita tion of [Ca(H4ntmp)(H 2O)]∞ onto the surfacesof hydrat i ng m i neral s , as opposed t o hom ogeneousc ryst a l grow t h, w oul d b e enhanc ed b y t he ab i l i t y ofterminal phosphonate moieties on the [Ca(H 4ntmp)-(H 2O)]∞ ol ig om er t o b in d t o t h e p a r t ia l ly h y d r a t edsurfaces. Terminal phosphonate moieties, and hencestrong attraction to aluminum surfaces, should exist inmany calcium phosphonate structures, and hence ingeneral promote heterogeneous crystallization.

If present on the surfa ce of the hydrat ing grains, thelay ered sheetlike coa ting of [Ca (H4ntmp)(H2O)]∞ (Figure4) would inhibit both further dissolution of calcium anddi ffusi on of w a t er t hrough t o t he m i neral surfac e.Furt herm ore, t he O‚‚‚O di st anc es b et w een adjac entuncoordinated phosphonate oxygens on the face of the[Ca(H 4ntmp)(H2O)]∞ sheets (8.47 and 9.60 Å) are closeto the value proposed by Coveney and co-workers 10 t obe the optimum binding distance for t he inhibi t ion ofettringite growth .

We propose that the mechanism by which H 6nt m pinhibits cement hydr at ion consists of two steps: First ,di ssol ut ion, w hereb y cal ci um i s ext rac t ed from t he

surface of the cement grains, exposing the aluminum-

rich surface to enhanced hydration; second, precipita-tion, whereby the soluble calcium phosphonate oligo-merizes either in solution or on the hydrate surface toform an insoluble polymeric Ca phosphonate. The Caphosphonate binds to the surface of the cement grains,i nhib it i ng furt her hydrat i on b y ac t i ng a s a di ffusi onbarrier to water as well as a nucleation inhibi tor.

Acknowledgment. Financial support for this workw a s prov ided b y Ha l li burt on Energy Serv ices. TheBruker CCD Smart System Diffractometer of the TexasC en t er f or C r y st a l log r a ph y a t R ice U n i ve rs it y w a spurchased with funds from the Robert A. Welch Foun-dation. The Bruker Avance 200 and 500 NMR spec-

trometers were purcha sed with funds from ONR G ra ntN00014-96-1-1146 and NSF Grant CHE-9708978, re-spectively.

Supporting Information Available: X-ray crystallo-g r ap hic d at a for [C a(H 4ntmp)(H 2O)]∞‚3.5(H 2O)] and selectedNMR and SEM images (PDF). This material is available freeof cha rge via the I ntern et a t h tt p://pubs.a cs.org.

CM0302431

Figure 20. Simu lat e d fa c -chelate structure (left) may form in aqueous solutions by the complexation of calcium by H 6n t m p a n dthen rearrange to give the bis-chelate calcium dimer (right) that is the structural subunit in [Ca(H 4ntmp)(H 2O)]∞.


A New Mechanisms for Cement Hydration Inhibition

Documents

Transcript of A New Mechanisms for Cement Hydration Inhibition