Cement Energy and Environment

to the total signal were determined from the individual fitted distributions by reconstructing individual echo decays. Hence, the amplitude of the reconstructed first echo was compared to the measured first echo to confirm the validity of the technique. In order to reduce noise, a Gaussian low-pass filter, with a standard deviation of 80 min (4 experimental points), was applied to the series of first echo amplitudes as a function of the curing time. The applied filter was effective in removing high frequency fluctuations while revealing low frequency features of interest and facili tating useful derivative plots that required no additional smoothing (see Supplementary Material, Fig. S1 and animation). -Long Tz - ShonTi - - . Fhling - Data I I ~ : ~ I .-I II I .WI An eight-channel TAM Air calorimeter (TA Instruments) was used to monitor the isothermal heat production from hydrating cement paste at 23 ·c. A time lag existed between the contact of water and cement and the start of measurement due to time for mixing (2 min), sample weighing into the glass ampoule, and stabilization of the block temperature (typically 45 min). 3.0 Results and discussion 3.1 T2 distributions and the meaning of the T2L and T2S peaks The development of multimodal distributions of T2 values (Fig. 2) during early cement hydration is an established feature of 1 H NMR relaxometry and there is general consensus that the discrete peaks are 112.5 h~--c::±=::·::::""':::!' :...,~./£__ ~_..£::.- __ representative of water confined in different structural levels within the cement [17,20,27-29]. Various authors have developed schemes, with much commonality between them, for the assignment of T2 peaks to separate structural elements (e.g . 11 , 21]. T2 values in the millisecond range (T2L) are attributed to water in capillary pores, which initially is interstitial water between the cl inker grains but then 28.5 h I -----=~----- ~ lllllltlllll 39.2 h .... .. 1 o- 5 1 o- 4 1 o- 3 1 o- 2 10-1 Transv. Relaxation Time- T 2 (s) l fig. 1. Examples oflog-normal fits ofT 2 dlStributiOIISfollowlllg laplace inversions ofa:ho decays from hydrating ore with no retardo~nL The ventral dashed line shows the mimmum echo time of Lhe instrument Ill Transverse Relaxation li11e •T 2 (s) f.. lt\~auif;~,j:;:c;tp.u;.,f{f(..u»A !XIN4.iliiJSl'f \!lltlt(!J,~. \ll:';!t.:rO IDO!J: 1 ltU.oi1Hne111nl.ill46hdb:5W•t.t llii'0.1Wic.llour.rofLIItl$r.ID10!L evolves to become inter-hydrate water with pores of about 10 nm. In comparison, T2 values in the hundreds of microseconds range (T2S} are attributed to water in gel pores of about 3 nm, conceptualized as existing between the gel globules that comprise C-S- H [21] . The validity of this assignment scheme for our system will be considered later in Sections 3.3 and 3.7. The key hydrogen proton populations to which added water is proposed to convert during the early hydration of OPC are shown schematically in Fig. 3. The shapes of the T2L and T2S peaks were in part determined by the value of the regularization parameter a used in the Laplace inversions of the 56 •