Effects of organic additives on calcium hydroxide crystallisation during lime slaking

Abstract Organic compounds, often used in cement systems as admixtures, may affect the crystallisation and carbonation kinetics of Ca(OH)2, an important phase of hydrated cement. Here, we investigated changes in Ca(OH)2 morphology in the presence of 3 organic compounds, commonly encountered in cement and lime-based materials: sucrose, pectin and calcium lignosulfonate. The additives were introduced either before or after lime slaking to determine the influence of temperature. Ca(OH)2 crystals and supernatant solutions were characterised at time of slaking and after 6 months of ageing using scanning electron microscopy, X-ray diffraction and optical emission spectroscopy. Our results indicate that the morphology of Ca(OH)2 crystals is modified by the characteristics of the organic molecules which promote formation of Ca(OH)2 with habits that can result in faster carbonation, an effect that is detrimental to cement used in reinforced concrete. These effects are enhanced when the additives are introduced before slaking, likely as a result of thermal degradation.


Introduction
The aim of this work is to investigate changes to calcium hydroxide (portlandite, or Ca(OH) 2 ) crystals morphology in presence of various organic compounds commonly used in cement and lime pastes. Gaining a deeper understanding of these changes allows tailoring some properties of the cement-and lime-based materials to specific applications.
Portlandite is the second most abundant phase in hydrated cement. It forms by hydration of lime (CaO) and as a by-product of the hydration of alite and belite. It is crucial for the durability of reinforced concrete in providing an alkaline reservoir, which protects rebars from corrosion [1][2][3]. One of the most common deterioration mechanisms of concrete is carbonation of portlandite, which results in pH reduction and depassivation of reinforcing steel elements [2].
The carbonation rate of portlandite, as well as many other fresh and hardened properties of lime and cement systems, are highly affected by the morphology of portlandite crystals. In hydrated cement pastes, dispersed microcrystalline portlandite is regarded as beneficial to mechanical properties as opposed to large crystals or agglomerates [4]. In lime pastes, the microstructural evolution of portlandite upon prolonged storage in water (e.g. during the so-called traditional practice of "ageing" of lime putties) from large crystals with prismatic habit to smaller crystals with platelet habit is known to improve the rheological properties as well as the carbonation rate of lime mortars [5,6].
Besides the "ageing", the presence of other phases can also cause changes in portlandite morphology. Sulfates and tricalcium aluminate are known to induce a transformation from large clusters to dispersed hexagonal platelets of portlandite [7][8][9]; nitrates and chlorides favour a rod-like prism habit [7,8]; silicates favour large agglomerates and irregular shapes [8].
However, while inorganic molecules have been extensively studied, little is known about organic compounds. In cement and concrete systems, these compounds are often embedded as admixtures with various purposes: triethanolamine (TEA) and tri-isopropanolamine (TIPA) are used as set accelerators; polycarboxylate-polyether copolymers or lignosulfonates are used as superplasticizers; fatty acid salts are used as water repellents and air-entraining agents [10,11]. Berger and McGregor [7] showed that organic compounds strongly influence the formation, growth and morphology of Ca(OH) 2 crystals embedded in cement pastes. Young [12] observed portlandite growth poisoning and an increase in portlandite supersaturation as a direct result of the adsorption of organic molecules onto Ca(OH) 2 nuclei. Barker et al. [13] https://doi.org/10.1016/j.cemconres.2020.106254 Received 21 July 2020; Received in revised form 2 October 2020; Accepted 2 October 2020 rationalised the influence of organic additives on the growth of Ca (OH) 2 crystals by proposing four criteria: (i) growth poisoning of specific Ca(OH) 2 crystal faces, (ii) modification of Ca ions solubility, (iii) adsorption on Ca(OH) 2 crystal surfaces (indicated by changes in the zeta potential), and (iv) substitution for Ca or OH ions within the crystal structure with organic nano-inclusions.
Further insights on the effect of organic compounds on Ca(OH) 2 crystals modifications come from research on air lime. Historically, lime mortars were often produced by mixing not only lime, water and aggregate, but also a wide variety of organic additives to modify the properties of the mix. Organic additives used in lime (e.g. oils, fatty acids, plant extracts, animal glues, egg, blood, beer, casein, fruit juices, sticky rice, etc.) are known to affect the characteristics of mortars such as the pore structure and setting process. Polysaccharides (e.g. found in mucilage and other plant extracts) have an effect on the pore size distribution of the mortars and on the strength as well as on the carbonation rate [14][15][16][17][18], proteinaceous materials (e.g. animal glue and blood) improve mechanical strength and carbonation rate [17,19], and oils affect the microstructure of lime pastes and mortars by altering the carbonation pathway of Ca(OH) 2 , fostering the formation of amorphous carbonate phases [17,20]. However, because of the variety of compounds used as admixtures and the little scientific research carried out to date, there is little information on how organic additives used in lime-based materials can benefit the modern construction industry.
This paper presents the result of an investigation on the effect of three organic additives -namely sucrose, pectin and lignosulfonate -on the microstructure and mineralogical characteristics of the Ca(OH) 2 produced by lime slaking in excess of water. The research aims at improving properties of the materials currently available. Analyses are carried out on the freshly prepared putty and after 6 months from slaking in order to gather information on how additives influence the lime "ageing" process. The ageing time range has been selected conveniently for various reasons. The outcome of this investigation can be beneficial to the cement industry, as it contributes to understand how the phase portlandite evolves in presence of additives commonly used in cements up to 6 months from setting as early hydration is crucial in determining the final concrete characteristics. Of course, in a cement system, portlandite is embedded in a more complex matrix, where interactions with other phases may affect its evolution. However, our study although limited to the lime-water-additive system, provides a valuable starting point to disentangle the complexity of the interactions present in a cement sample.
Moreover, the acquired knowledge will be beneficial to the lime industry. Manufacturing costs can be cut if additive-modified lime putties could be used after a reduced time from production with respect to traditional lime products, which instead need several months or years of maturation before application. An extensive literature on the ageing of pure lime systems [5,6,[21][22][23] has investigated the process in a time range that varies from a few months to years of storage under water. In this study, we have investigated the lime-additive systems at time "zero" and after 6 months, representative of a minimum ageing time after which the ageing effects are observable in the material. The goal is to verify whether the same effects can be detected in a reduced ageing time in presence of organic molecules in comparison to the pure lime putty.
In the following section, a brief review of some of the literature on the organic compounds used in this research is provided.

Sucrose
Sucrose is one of the most common sugars. It is a disaccharide made of α-glucose and β-fructose groups linked together by a (1 → 2) glycosidic bond (Fig. 1a) [24]. The hydroxyl groups make the molecule highly water-soluble (2100 g/L) [25].
Sucrose has been commonly used as hydration retarder in cement mixtures since the '30s [26]. The underlying mechanism of retardation seems related to the adsorption of sucrose onto hydrating cement particles and hydration products forming a temporary barrier for further hydration and resulting in C-S-H nuclei-poisoning [10,27,28]. As regards to lime systems, the use of sugars has been documented in historic lime mortars of various countries where it has shown to improve mechanical properties and durability of the mixes [14][15][16]29,30]. Currently, sugar is used in the development of novel nano-structured limebased materials [31].
Similarly to what has been found on the interaction of this molecule with C-S-H, sucrose has been observed to adsorb onto Ca(OH) 2 nuclei, poisoning their growth [12,32,33].

Pectin
Pectin is a polysaccharide (Fig. 1b) mainly composed of linear chains of α(1 → 4)-linked galacturonic acid units, partially methylated at the carboxyl group [34]. Pectin-like polysaccharides are the main compounds found in plant extracts traditionally added to lime mortars, e.g. nopal juice (extracted from cladodes a plant common in South America) in historic Mayan and Aztec mortars [34,35]. Rodriguez-Navarro et al. [36] showed that the addition of nopal juice in water used for lime slaking induces Ca(OH) 2 crystals modification resulting in a lime putty of improved rheology with properties similar to those of aged putties. Pure pectin interacted with Ca(OH) 2 electrostatically and through H-bonds, acting in a 3-fold manner: (i) crystallisation inhibitor ("nuclei poisoning effect"), leading to an increase of nanosized portlandite crystals; (ii) "habit modifier", leading to the formation of thin plate-like particles; (iii) "colloidal stabiliser", leading to an increase in colloidal stability of lime putty dispersions. Recently, it has been shown that pectin affects the nonclassical Ca(OH) 2 crystallisation both in the pre-nucleation stage, by stabilising Ca(OH) 2 pre-nucleation clusters (PNCs) and liquid precursor and therefore delaying the formation of amorphous Ca(OH) 2 (precursor of crystalline portlandite), as well as in the post-nucleation stage, by stabilising amorphous Ca(OH) 2 and delaying the formation of metastable crystalline Ca(OH) 2 and in turn of stable crystalline portlandite [3].

Lignosulfonate
Lignosulfonates are a class of compounds that induce microstructural modifications of Ca(OH) 2 crystals. As by-product of the pulping industry, low price lignosulfonates currently find application in the production of concrete as dispersants and superplasticizers [10,37]. During wood pulp production, a process called sulfite pulping separates lignin through acidic cleavage of ether bonds. During this process, carbocations form and readily react with HSO 3 − to form sulfonate groups. The general structure of lignosulfonates is shown in Fig. 1c.
Lignosulfonate is a water-soluble polymer containing both hydrophilic groups, i.e. hydroxyl, hydroxyphenyil and sulfonate, and hydrophobic components, i.e. carbon chains [38]. As superplasticizers for cement, lignosulfonates have the ability to intercalate cement particles by overcoming the van der Waals attractive inter-particle forces that cause agglomeration [10]. Their dispersive effect is a result of both adsorption of lignosulfonate on cement particles and formation of a physical layer that prevents the agglomeration of cement particles [39][40][41][42]. Commercial lignosulfonates are also used as set retarders. However, the retardant effect in hydration seems related to the sugar impurities found in their formulation rather than to the lignosulfonate itself [43]. Few studies have investigated the effect of lignosulfonate and molecules with sulfonate groups on Ca(OH) 2 . Jawed et al. [44] observed that lignosulfonate inhibits growth of Ca(OH) 2 crystals. Barker et al. [13] have shown that additives containing sulfonate moieties modify Ca (OH) 2 crystals' habit and size, as a result of the adsorption of the organic molecules onto Ca(OH) 2 crystal surfaces, affecting nucleation and growth. The authors observed that the Ca(OH) 2 crystal lattice remained unaltered, suggesting that organic molecules are not embedded into the Ca(OH) 2 structure. Kirchgessner and Lorrain [45] and Lee et al. [42] found an inverse correlation between calcium lignosulfonate content and Ca(OH) 2 particle size as a result of the de-agglomeration caused by the hydrophobic portion of the calcium lignosulfonate molecule. The lowest particle size was obtained with 1.5% lignosulfonate content. Yilmaz and Glasser [46] showed that sulfonate-based superplasticizers (sulfonated melamine formaldehyde, commonly used in cement pastes) modifies the habit of Ca(OH) 2 crystals from prism to platelets. Rodriguez-Navarro et al. [3] showed that lignosulfonate acts on the preand postnucleation stages of Ca(OH) 2 crystallisation similarly to pectin, as previously described in Section 2.2.

Sample preparation
A total of 7 putties were prepared as listed in Table 1: a control putty and 6 putties produced by using the 3 organic molecules that are the focus of this research. For each molecule, two batches were prepared: one batch was prepared by slaking the CaO with an aqueous solution containing the dissolved additive, whereas another batch was prepared by slaking the CaO in pure water and, immediately after the slaking, the additive was added to the putty (i.e. as received). In the following paragraphs, modified putties produced via addition of additives in solution are referred to as samples "A", whereas those obtained via direct addition of the solid additives are referred to as samples "B" according to the nomenclature in Table 1.
The control putty was prepared by slaking 333 g of CaO in 1 L of deionised (DI) water, in order to obtain a 1:3 mass ratio [36,[48][49][50][51]. Batch 'A' of the sucrose-modified putties was produced by slaking 333 g of CaO with 1 L of a 5% wt sucrose solution [31]. For the batch 'B', 333 g of CaO were slaked in 1 L of distilled water as for the control. After the initial violent stage of the hydration reaction, as the mixture ceased boiling, it was hand-stirred for 10′ to promote slaking. At this point, 50 g of sucrose were added to the lime paste and mixed thoroughly. Batch 'A' of the pectin-modified putty, was produced by slaking 333 g of CaO with 1 L of 0.5% wt pectin solution [36]. The solution was prepared by slowly adding dry pectin to distilled water, heated on a magnetic hot plate at 80 °C and with a stirring speed of 1500 rpm, to favour the dissolution of pectin and prevent clumping. Batch 'B' was prepared similarly to batch 'B' of the sucrose-modified putty, however, in this case after the initial violent slaking stage, 5 g of pectin were gradually added to the mix, while the constant and vigorous mixing and the high temperature of the putty (about 80 °C right after slaking) promoted the dissolution of pectin. Similarly, batch 'A' of the CaLSmodified putty was prepared by slaking 333 g of CaO with 1 L of a 1.5% wt solution of CaLS [42]. For batch 'B' the same amount of dry CaLS (15 g) was added to the lime paste after the initial violent slaking stage. After production, all putties were stored into airtight containers under excess of DI water.

CaO
Scanning electron microscopy (SEM), X-ray diffraction (XRD), and X-ray fluorescence (XRF) were used to characterise the CaO used in these tests. For the analysis, a sample of about 50 g of CaO granules were placed inside a glove box filled with nitrogen gas (N 2 ) to prevent reaction with atmospheric CO 2 and H 2 O. The granules were crushed with an agate mortar and pestle and the powder produced was sieved to obtain particles with diameter < 500 μm. During crushing and sieving, CO 2 concentration was monitored using a K30 10,000 ppm CO 2 Sensor from CO2Meter.com and constantly kept < 200 ppm by washing the glovebox with N 2 when needed. SEM analyses were performed using a Tescan Mira3 microscope. The CaO powder was sprinkled on carbon tape fixed on a stab and subsequently coated with a 5 nm-thick Pt layer. Analyses were carried out in high-vacuum mode at 10 kV voltage. Image processing and calculations were performed using ImageJ v1.53a software.
XRD analyses were performed with a Rigaku SmartLab instrument on CaO particles passing a 230 mesh sieve and pressed onto a glass sample holder. The measurements were performed in parallel beam geometry, X-ray source was a Cu tube producing K α radiation, scans were collected in the 10°-90° 2-θ range with steps of 0.5° and at 50 kV  and 40 mA. Quantitative phase analysis was performed using the Rietveld method implemented in the Rigaku SmartLab Studio II software. Quantitative XRF analysis was performed using a Spectro Xepos benchtop XRF analyser. The 'Geochemistry traces' method was used as internal calibration, and resolution was 131.5 eV at 5.89KeV. Pressed pellets of a mixture of CaO and an inert wax binder (Fluxana Cereox wax, C 38 H 76 O 2 N 2 ) were produced to improve consistency of the results.

Putties solid phase
Solid phase and the supernatant solution of all putties were analysed separately. For the solid phases, a sample of about 10 g of putty was collected from the core of each batch, rinsed with ethanol and dispersed in a quote of the same solvent into a sealed vial until analysis. It is assumed that the bulk of the putties, where the samples were collected from, is homogeneous.
Micromorphological characterisation of the putties was obtained using the same SEM as for CaO characterisation. For the sample preparation, 3 quotes of each dispersion were transferred onto metal stabs with a pipette and vacuum-dried for 1 h at 50 °C. The solid residue was then coated with a 5 nm-thick platinum layer using a sputter coater. Image analysis (crystal size measurements) has been carried out with the software ImageJ v. 1.53a [52].
The mineralogical characterisation was obtained using a Rigaku SmartLab X-ray diffractometer (XRD) in the same operational conditions as previously described in Section 3.3.1. A droplet of each dispersion was transferred onto a zero-background Si wafer with a pipette and vacuum-dried for 1 h at 50 °C, in order to produce an oriented aggregate, with Ca(OH) 2 crystals lying on their {00.1} faces [5]. For each oriented aggregate, 3 different scans were collected. All parameters calculated from XRD data (platelet abundance, crystallite size, microstrain, D hkl described in the following paragraphs) are the average of 3 scans and error is calculated by standard deviation.
To estimate the platelet abundance in the putty samples the method proposed by Rodriguez-Navarro and colleagues in 1998 [5] was used. The mean crystallite size was estimated from the XRD data using the Halder-Wagner method [53] implemented in the Rigaku SmartLab Studio II software package. To gain further insight on the development of crystallite facets in relation to the presence/absence of the additives and the effect of ageing, the mean crystallite size values of single portlandite (00.1), (10.1) and (10.0) reflections were estimated from XRD data using the Scherrer equation [54]: where D hkl is the mean crystallite size in the direction perpendicular to the hkl plane, k is the grain-shape dependent constant 0.92, λ is the wavelength of the incident beam (1.5408 Å for Cu-K α radiation), θ is the Bragg reflection angle, and B 2θ is the line broadening at half-height of the diffraction peak (in radians).
The particle size distribution of the putty was measured via laser diffraction using a Mastersizer 3000 (Malvern Panalytical; LD-PSD from now on). Agglomeration of particles was reduced by using ethanol as dispersant and through sonication. During the measurements, the laser obscuration was kept within 15-20%. For each sample, at least 3 consecutive runs were carried out in order to obtain RSD% within the thresholds recommended by ISO13320 [55] (< 3% for D50 percentile and < 5% for D10 and D90 percentiles of the particles population).

Putties supernatant solution
The pH of the supernatant solutions was measured using a Hannah HI-991300 pH meter (0-14 pH range) that was calibrated prior to analysis using the two-points calibration method with the 7.01 and 10.01 buffer solutions. Total Ca concentration in the supernatant solutions was measured through Inductively Coupled Plasma -Optical Emission Spectrometry (ICP-OES) using a Perkin Elmer Optima 8000 spectrometer. For this analysis a sample of 10 mL of solution from each batch was collected with a syringe and then filtered through a 0.1 μm PTFE syringe filter to remove any solid suspended particle. It is assumed that, upon filtering, the Ca detected during the ICP-OES analysis is only due to the presence of Ca ions in aqueous solution, in equilibrium with solid Ca(OH) 2 , and not from solid Ca(OH) 2 or CaCO 3 particles in suspension. The filtered solutions were then diluted (x100 or x1000 depending on the Ca concentration) in a 2% H 2 NO 3 solution to obtain a concentration of Ca within the linear range of the instrument.

Quicklime characterisation
The mineralogical and microstructural characteristics of the CaO are shown in Fig. 2. The XRD pattern (Fig. 2a) shows only CaO peaks whereas the microstructure of CaO particles (Fig. 2b) shows a porous fabric with pores of uniform size (approximately 0.1 μm diameter).
Chemical composition of the CaO as measured via XRF analysis is  shown in Table 2, and confirms the high purity of the lime with approximately 98.9% Ca content and minor inclusions of Mg (0.5%) and Si (0.4%).

SEM analysis
Selected SEM images of all putties are reported in Fig. 3 and in Supplementary Material. The figure shows that Ca(OH) 2 crystals in all putties are varied in size and shapes. In the fresh control putty (Figs. 3a, S1 and S2), crystals are commonly interlocked with 4 shapes visible: (i) short rod-like prisms of size of 0.5 ± 0.3 μm along the < 00.1 > direction; (ii) hexagonal platelets of size 1 ± 0.4 μm along the < 10.0 > direction with overdeveloped basal {00.1} faces; (iii) crystals with bipyramidal morphology and well-developed lateral {10.0} and edge {10.1} faces; (iv) nanogranular crystals. In the aged control putty, such granular nm-sized crystals appear more abundant whereas the μmsize crystals are less numerous (Fig. S3). The latter display a hexagonal platelets morphology and have similar size as in the fresh putty, with stepping appearing on the overdeveloped {00.1} faces. The {10.0} faces of the prisms show marked cleavage signs along the < 10.0 > direction.
Micromorphology of the putties produced in the presence of organic additives is shown in Fig. 3b-d. In sample M1A (putty modified using sucrose solution), crystals display a granular nano-sized habit (typically 100-200 nm) without a discernible geometry. Crystal facets are not visible, neither in the fresh nor in the aged sample (Figs. 3b, S4 and S5).
In the freshly-slaked M1B putty (modified using solid sucrose), crystals are facetted and display a short rod-like morphology similar to that of the control putty, but with bigger average size (1 ± 0.3 μm, Fig. S6). In the related aged sample (M1B), the rod-like habit is less expressed in favour of hexagonal platelet morphologies. Stepping of basal {00.1} faces and stacked-layers lamellar structure are visible as well as nanogranular crystals (Fig. S7).
In the fresh M2A sample (pectin solution; Fig. 3c), as well as in the fresh and aged M2B samples (pectin solid), the overall morphology is similar to the aged control putty: the majority of the crystals display a nanogranular habit (mean size 50-100 nm) with few crystals μm-size and hexagonal platelet habit. The hexagonal crystals have well-developed and dominant basal {00.1} faces and small {10.0} faces, with the latter showing marked cleavage signs in the < 10.0 > direction (Fig.  S8). In the aged M2A sample (pectin solution, Fig. S9), less μm-sized hexagonal platelets and more nanogranular crystals are visible, resembling the micromorphology of M1A sample (sucrose solution).
In sample M3A (CaLS solution; Fig. 3d), portlandite crystals exhibit both facetted and regular hexagonal plate-like habit, with largely overdeveloped basal {00.1} and small prism {10.0} faces. The crystals size distribution range between 0.1 and 1 μm. Smaller crystals also exhibit a facetted regular platelet-like shape and very few nanogranular crystals are visible. Lateral {10.0} faces do not show (or show only minimal) cleavage signs, and some overdeveloped {00.1} faces show a nanogranular surface structure (Fig. S12). The aged M3A putty shows a similar microstructure, with extensive drying-induced agglomeration and S14). In the aged M3B putty, the nanogranular crystals seem more abundant (Fig. S15). Table 3 shows the results of quantitative phase analysis for all samples. Results show that portlandite is the main phase identified in all samples, with Bragg peak positions of all samples matching well with those of a portlandite reference pattern (COD Card No. 7020138). Traces (< 5%) of calcite and monohydrocalcite are detected in fresh M1A, M2A and M3B putties. Formation of these carbonates is most likely related to a longer contact with air of the samples during preparation compared to other samples however, we cannot exclude that the presence of additives may have played a role in promoting the carbonation reaction since in the control samples no traces of carbonates were found.

XRD analysis
The platelets abundance A 00.1 and mean Halder-Wagner crystallite size of the control and modified putties at 0 and 6 months are shown in Fig. 4. With the exception of M2A (pectin solution), the platelets abundance A 00.1 of all other samples increases with ageing (Fig. 4a). A 00.1 for M1A (sucrose solution), aged M2A (pectin solution) and M3 samples (CaLS) is comparable to the control, while A 00.1 is slightly higher in M1B (sucrose solid), fresh M2A and M2B (pectin solid) than the control. Fig. 4b shows the mean crystallite size for all samples, both fresh and aged. Results show that the crystallite size is lower in the modified putties than in the control. The values in the control sample range between 35 and 36 nm, while in most of the modified putties between 34 and 19 nm, except for the aged M2B (pectin solid) where the size is 37 nm. A remarkably low mean crystallite size is measured in M1A (sucrose solution) samples, with values of 13-14 nm. Similar to the platelet abundance A 00.1 , the mean crystallite size increases with ageing. The microstrain ε, calculated using the Halder-Wagner method is negligible with values range 0.00-0.13% regardless of the presence of the additives and ageing time (Fig. 4b). Fig. 5 shows D 00.1 , D 10.1 and D 10.0 against the ageing time for each putty.
In the control putty (Fig. 5a), the crystallite size undergoes a slight increase both along and perpendicular to the c axis, as well as a slight increase in the edge (10.1) faces.
In M1A (sucrose solution, Fig. 5b), data show a decrease in the (10.1) facet size while the crystallite dimension in the directions parallel and perpendicular to the c axis are stable overtime and similar to each other, suggesting an equiaxed crystallite morphology. Notably, the crystallite size in all directions is included in the range 40-300 Å, which is remarkably smaller than the control (400-500 Å). Sample M1B (sucrose solid, Fig. 5c) also shows a decrease of (10.1) facet dimension, while the size along and normal to the c axis slightly increase after 6 months. Data showing that D 00.1 > D 10.0 suggest that crystallites are slightly elongated in the < 00.1 > direction.
Samples M2 (pectin, Fig. 5d, e) show a similar trend to M1B. In sample M3A (CaLS solution, Fig. 5f), a slight decrease in the directions parallel and perpendicular to the c axis is shown. The crystallite morphology also seems slightly elongated in the < 00.

Particle size distribution analysis
The results of LD-PSD analysis of the modified putties in comparison to the control sample are shown in Fig. 6.
The control sample shows a trimodal particle size distribution (Fig. 6a). The sample is characterised by: i) small particles with size between 0.3 and 1.0 μm and a maximum at 0.5 μm; ii) particles 1.0-15 μm size with a maximum at ~4 μm (this is the predominant group, encompassing ~46 vol%); iii) larger particles with size between 15 and 80 μm and a maximum at ~30 μm. In the aged control, the PSD shows again a trimodal distribution with a predominant group of medium-sized particles (Fig. 6b), and the maximum of the central peak is slightly higher compared to the fresh putty (~5 μm instead of ~4 μm) and the group of large particles spread over a wider range reaching a size of 110 μm.
The PSD of fresh M1A (sucrose solution, Fig. 6c) is similar to the control, although the maximum of the mid-sized group of particles is lower (~3 μm) and the distribution is slightly narrower (maximum size is 51 μm). The PSD of the aged M1A putty (Fig. 6d) is narrower than the fresh sample, quasi-bimodal and shifted towards smaller values Table 3 Rietveld quantitative phase analysis from XRD data.   Fig. 6d, e) is polydisperse and has a long tail of large particles (max. size 144 μm). The percentage in volume of large particles (> 10 μm) is higher in the aged putty than the fresh.
The PSD of M2A (pectin solution) shows again a trimodal distribution (Fig. 6f, g). The maximum of the large particles group is at 98 and 86 μm in the fresh and aged putties, respectively, and the volume of these groups (> 90%) is significantly higher than in the control (57%).
The PSD of the M2B (pectin solid) shows a sample with more polydispersed and wider particle size than the control sample (Fig. 6h,  i). Similar to M2A, the large particles group's volume (> 90%) and the max size (976 and 240 μm in the fresh and aged sample, respectively) are significantly higher than the control.
The (trimodal) PSD width of M3A (CaLS solution, Fig. 6j, k) is comparable to that of the control, but the vol% of particles > 10 μm is significantly higher (84 and 72% in the fresh and aged putties, respectively).
The PSD of M3B (CaLS solid, Fig. 6l, m) is also trimodal and shows, similarly to the control, an increase in vol% of particles with size > 10 μm.

Supernatant solution analysis
The pH of the supernatant solutions on the control and modified putties for fresh and 6 month-old samples are reported in Table 4. The pH values are in the range 13.2-13.8 regardless of the presence or absence of the organic additive or of the ageing time. These values are above the theoretical pH of a saturated aqueous solution of Ca(OH) 2 at 20-25 °C [56], however, it is known that glass electrodes are affected by the so called 'alkaline error' at pH > 9 that can result in higher readings [57]. The total Ca concentration in the supernatant solutions measured using ICP-OES are reported in Fig. 7. Results show that all Ca concentrations are in the range 790-1010 mg/L regardless of the presence or absence of additive and ageing time, with the exception of the supernatant in samples M1 (sucrose), with 2.8-5.1 g/L, i.e. ~3-5 times higher than the other samples.

Control putty
Microstructure of the fresh control putty (Fig. 3a) showed textural features that are typical of portlandite crystals obtained through lime slaking with excess of water [48], characterised by crystals of irregular shape and wide size distribution with a rod-like habit (Fig. 3a). With ageing, the size of the basal {00.1} faces seems to increase at the expense of prism {10.1} faces overtime, acquiring a plate-like habit. With regard to the crystal shape, although direct correlations between platelet abundance A 001 and age of the putty cannot be established (because of the intrinsic variability across various types of lime), our results (Fig. 4a) suggest that platelets abundance increases with the ageing, in good agreement with observations by Rodriguez-Navarro et al. [5] and Cazalla et al. [58]. The signs of cleavage normal to the < 00.1 > direction observed at the SEM (Fig. 3) have been reported by various other authors [5,6,59] and are likely due to the transformation of the non-equilibrium prismatic habit to the equilibrium platelet habit, which entails the preferential dissolution of prism {10.0} and edge {10.1} faces compared to the basal {00.1} face with, eventually, separation in the direction perpendicular to the c axis. Such preferential dissolution is due to the different coordination number of the Ca on the surface of these faces [4]. In fact, Ca ions on the {10.1} and {10.0} faces are 4-fold and 5-fold coordinated whereas they are 6fold coordinated on the {00.1} face, like in the bulk. Water molecules and hydroxyl groups have a strong affinity for the under-coordinated sites and most likely scavenge them [60].
As regards the reduction in crystallite size upon ageing reported by other authors [5,6,22,23,58], our XRD data do not support this scenario, showing a slight increase in mean crystallite size after 6 months (Figs. 4b and 5a). Such increase can be ascribed to the secondary precipitation of portlandite occurring during ageing, as previously  observed by Rodriguez-Navarro et al. [5]. Secondary crystals are likely to have higher crystallinity than primary crystals as the nonclassical Ca (OH) 2 crystallisation takes place through a cascade of dissolution/reprecipitation steps with formation of phases (PNCs -dense liquid precursor -amorphous Ca(OH) 2 -Ca(OH) 2 mesocrystals -Ca(OH) 2 crystals) of increasing structural order [3]. Hence, secondary precipitation of Ca(OH) 2 is likely to bring about higher crystallinity, resulting in an increase in crystallite size.

Modified putties
In the putties with addition of organic molecules, various effects on the crystal morphology could be observed. In the following sections, the effect of each additive is discussed in detail, considering, first, the results of sample "A" (putty modified with additive in solution), followed by the results of sample "B" (putty modified with solid additive). Further considerations emerging by the comparison of the two batches (i.e. 'A' and 'B') are reported at the end of each section.

Sucrose-modified putties
Results of the SEM analysis show that portlandite crystals formed by slaking the CaO in sucrose solution (Fig. 3) are characterised by an irregular shape with jagged edges and no facets developed. This suggests that the sucrose inhibits the growth of all faces. Similar irregularities were reported also by Martinez-Ramirez et al. [31] in portlandite precipitated in sucrose solution. Inhibition of the crystal growth is also supported by the XRD data (Figs. 4b and 5), suggesting that the crystallites in lime slaked with sucrose solution are more than 2 times smaller than the control putty, and from 0.3 to 2.3 times smaller than all other modified putties. The Ca concentration in the supernatant solution is ~0.5-2.5 times higher than the solubility of Ca(OH) 2 at ambient T (i.e. 1.7 g/L [61]). This suggests that some of the Ca atoms present in the supernatant must not be in its free Ca 2+ form (otherwise the solution would be highly supersaturated with respect to portlandite and the latter would eventually precipitate), but it is complexed to sucrose or its degradation by-products [31,62]. A high Ca concentration is also found after 6 months, suggesting that such Ca-sucrose complexes are stable.
LD-PSD of the fresh sample is significantly narrower than the control and becomes quasi-monodispersed after 6 months of ageing (Fig. 6). The smaller particle size in the fresh putty is likely the result of the nucleation inhibition effect of sucrose also observed in other studies during cement hydration [12,[26][27][28]32]. In the presence of sucrose, Ca (OH) 2 nucleates at high supersaturation, leading to a high nucleation density and to the formation of smaller, more numerous Ca(OH) 2 crystals. With ageing, larger particles (> 10 μm, likely agglomerates of smaller crystals) are not detected in the system, suggesting that sucrose effectively prevents the formation of agglomerates acting as a dispersant, by either surface adsorption, growth poisoning or through calcium chelation [63].
Substantial differences are observed between the putty slaked in sucrose solution and the putty slaked in pure water with the subsequent addition of solid sucrose. In the latter case, portlandite crystals display a prismatic habit with well-developed {10.0} and {10.1} faces and their overall microstructure observed by SEM is comparable to the control putty (Fig. 3). The average crystallite size is only slightly lower than that of the control, as opposed to the marked reduction measured in the putty slaked in sucrose solution (Fig. 4b). The LD-PSD results (Fig. 6) seem comparable to the other modified putties, while the PSD of the putty slaked in sucrose solution is remarkably narrower.
Despite these differences, the Ca concentration in the supernatant solution of the putty modified with dry sucrose is similar to that of the putty slaked in sucrose solution. This suggests that the complexing activity of sucrose towards Ca atoms (responsible for the high Ca concentration measured in solution) is not the only factor that accounts for the distinct characteristics observed between portlandite crystals in the putty slaked in sucrose solution and in the putty with addition of solid sucrose. To explain these differences, in the following sections (Sections 5.3.1 and 5.3.2) we consider the possible effects of alkaline and thermal degradation on sucrose molecule.

Sucrose alkaline degradation
While the glycosidic linkage of sucrose is unstable under even slightly acidic conditions and is readily hydrolysed, the sucrose molecule is known to be stable in alkaline conditions (pK a = 12.6) [24,64]. However, under highly alkaline conditions (pH > 12), the molecule can be subject to various degradation processes, such as hydrolysis, with formation of reducing sugars and other organic acids [65,66]. The degradation products of reducing sugars (i.e. easily-oxidisable sugars defined by the presence of a specific moiety which gives also lower stability to alkaline conditions) and organic acids contain the HO-C-C=O group (similar to carboxyl but with hydroxyl and carbonyl linked to two distinct, adjacent carbon atoms), responsible for the adsorption of sugars on the surface of Ca(OH) 2 particles according to Bruere [65]. A further mechanism proposed in the literature [67] for sucrose alkaline degradation leads to the formation of another disaccharide (a glucosyl fructose derivative) which readily degrades mainly into lactic acid, both of which contain the HO-C-C=O group. However, since both sucrose-modified putties have pH > 12 (Table 4) these degradation processes, which are only related to the pH, have likely occurred in both batches.

Sucrose thermal degradation
When lime is slaked with an aqueous solution of sucrose, the heat generated by the reaction of CaO with water can produce a thermal degradation of sucrose. In fact, the temperature in the core of the mix can reach 200 °C [30], above the melting point of sucrose which is between 160 and 186 °C [68,69].
Several mechanisms have been proposed in literature to describe the thermal degradation of sucrose and its transformation upon heating. Oxidation takes place after melting at 185-190 °C and ends completely at 440 °C [70]. In the melting temperature range (160 to 189 °C), partial decomposition of sucrose into glucose, fructozone and fructose through hydrolysis can occur [69]. At 185 °C, the primary reactions of thermal degradation are reported to be the splitting of the glycosidic bond and formation of sucrose derivatives, i.e. stereoisomers with different configuration on the pyranose ring and anhydrous forms [71].
Our experimental results suggest that the temperature plays a critical role on the effect that sucrose can have on the growth of portlandite crystals. This suggests that in the solution surrounding the putty slaked with solid sucrose, the main species are sucrose and the products of its alkaline degradation. These compounds, have only a minor effect on the growth of portlandite crystals as microstructure (Fig. 3) and crystallite size (Fig. 5) in our sample are rather similar to that the control putty. Differently, it is likely that in the putty slaked with the sucrose solution, the heat generated by the slaking process resulted in the formation of compounds such as glucose, fructose and other reducing sugars derivatives, which contributed to a significant reduction in crystallite size (Fig. 5) and distinct microstructure made of small, jagged-edged crystals (Fig. 3). The mechanisms by which these compounds interact with Ca(OH) 2 , other than complexation, is not yet clear and should be further investigated. It is possible, however, that these new molecules affect the stability of PNCs, delaying the nucleation of Ca(OH) 2 , as recently proposed in various studies on the effect of other organic additives on the crystallisation of Ca(OH) 2 and CaCO 3 phases [3,36,72].

Pectin-modified putties
In the putties modified with pectin, the effect of the additive on the crystallisation of portlandite are not as clear as for the sucrose, as both crystallite size and microstructure (Fig. 3d, e) are broadly comparable to the control putty. A higher amount of granular nm-sized is observed in the putty slaked with an aqueous solution of pectin, as confirmed by the XRD data (Fig. 4b). This enhanced formation of small size Ca(OH) 2 crystals induced by the addition of pectin is likely to be due to growth poisoning, as a result of the chemisorption of pectin mainly on the basal {00.1} faces [3,32,36]. In fact, the layered structure of portlandite crystals is characterised by sheets of Ca atoms coordinated with 6 OH groups, with the layers stacked along the < 00.1 > axis and held together by H bonds [73]. According to computer simulations [36], the {00.1} surface is positively charged with CaOH 2 + species, similarly to the MgOH 2 + species found on the {00.1} face of the isostructural brucite [74]. The interaction between the deprotonated carboxyl groups of pectin and the positively-charged {00.1} portlandite surfaces could occur via chemisorption through the following reaction [36]: (2)

Pectin alkaline degradation
In high alkaline environment (pH > 12), pectin tends to (i) depolymerise through β-elimination with subsequent reduction of molecular weight, degradation of pectin molecule and formation of sugar residues [36,75,76], and (ii) de-esterification of the methoxy groups [76,77]. Only limited complexation between carboxyl groups and divalent cations, such as Ca, to form egg-box structures (i.e. gelling behaviour) is observed when pectin is in an alkaline, Ca-rich solution [36,75]. As a result of depolymerisation, a variety of reducing sugar end units including xylose and galactose are produced [36]. As explained in Section 5.3.1, reducing sugars can adsorb onto Ca(OH) 2 particles and their presence in solution could account for the granular crystals formation observed in the pectin-modified putties.

Pectin thermal degradation
Thermal degradation of pectin is a rather complex process as various factors including composition, degree of polymerisation and degree of esterification affect its thermal behaviour [78]. The literature suggests that pectin thermal degradation occurs in the range of 180-270 °C [78][79][80] which is only partially within the temperature of the lime putty during slaking. Hence, the heat produced during the slaking could partially degrade the pectin, in good agreement with our experimental results which show altogether a less marked effect on portlandite morphology compared to sucrose.
The putty slaked in pectin solution had more abundant nm-crystals and μm-crystals, displaying a more tabular shape, with respect to the putty where the pectin was added after slaking, where μm-sized rod-like crystals were visible. In addition to the chemisorption of pectin on the {00.1} basal faces of portlandite, the thermal treatment during the slaking may have promoted depolymerisation via β-elimination which leads to formation of reducing sugars [81,82]. The ionised hydroxyl and carboxyl groups of these species are able to adsorb onto Ca(OH) 2 nuclei, hindering their growth. This is supported by our XRD data which showed that the crystallite size of Ca(OH) 2 slaked in pectin solution was smaller than the control. However, because β-elimination is favoured in pectin with high degree of methylation, the de-esterification caused by alkaline conditions should reduce heat-induced depolymerisation [82]. Thus, the formation of reducing sugars (consequence of the depolymerisation) is limited by the alkaline-induced de-esterification and their quantity should not be as high as in the putty slaked in sucrose solution, where the effect of crystallite size reduction is much clearer.

CaLS-modified putties
XRD results showed a crystal size reduction in the CaLS-modified putties compared to the control sample, which is in agreement with results of other researchers [42,45]. The microstructure of CaLS-modified putties displays a variety of crystal shapes and sizes. However various μm-sized crystals with a clear tabular shape and with significantly overdeveloped {00.1} basal faces are observed (Fig. 3), which is in agreement with previous studies [42,46] indicating that LS molecules can adsorb onto Ca(OH) 2 surfaces. Although some papers suggest a complexing ability of LS towards Ca ions in Portland cement mixtures [39,41,83,84], the stability of CaLS complexes is much lower than Ca-sucrose complexes [85], as also clearly shown by our analyses of the supernatant solutions which show that the Ca concentration in the CaLS-modified samples is comparable to the control and pectinmodified samples (the pectin has also a reduced tendency to bind Ca [36,75]).
Thus, these results suggest that the microstructural effects observed in SEM images (i.e. the tabular habit of Ca(OH) 2 crystals) is the result of the preferential absorption of LS onto {00.1} basal faces through a reaction analogous to Eq. (2), in this case between the sulfonate moieties of LS (see structure in Fig. 1) and the CaOH 2 + species present on the basal Ca(OH) 2 faces [36]: (3)

CaLS alkaline and thermal degradation
In alkaline conditions, lignosulfonates are known to depolymerise and to form of catechol-like structured monomers (mostly vanillin) with an overall increase in phenols, i.e. hydroxyl moieties linked to aromatic rings, and decrease in methoxy groups [86][87][88]. All of these processes are linearly dependent only within the T range of 70-90 °C [88]. It is, therefore, possible to deduce that the heat developed during the slaking process could have partially contributed to a structural modification of the added CaLS. We should note that vanillin, the compound mainly formed upon thermal treatment, has a pK a of 7.4 at 25 °C [89] hence, at the high pH of the solution (pH > 12, Table 4), this compound exists in the negatively-charged deprotonated form, which can potentially absorb on the positively-charged {00.1} basal Ca(OH) 2 faces (following Eq. (3)) and interact with Ca 2+ ions (complex formation). The occurring of these reactions and their extent should be further investigated. The microstructural and mineralogical evidence on CaLS-modified putties (Figs. 3,4,5) showed that the observed overall results (formation of platelet-like crystals and crystallite size reduction) were more marked in the putty slaked in CaLS solution than in the putty produced with addition of solid CaLS. This suggests that the structural modifications CaLS underwent upon thermal treatment through the slaking affected the Ca(OH) 2 crystallisation process by contributing to the growth of the {00.1} face in some Ca(OH) 2 crystals and the reduction of the mean crystallite size, through mechanisms that are also likely to be ascribable to the destabilisation of Ca(OH) 2 PNCs [3].

Ageing effect in modified putties
The changes induced by ageing in the modified putties are comparable to those induced in the control putty, in terms of crystal shape and size and system chemistry evolution. Remarkably, the effects induced by the additives in the fresh putties (e.g. modifications in crystal habit, calcium concentration in solution, etc.) are kept also after ageing. In the CaLS modified putties, the hexagonal-platelet shape observed in the fresh samples has been observed after ageing, suggesting that the series of dissolution/reprecipitation steps occurring during the maturation process are not able to alter the crystal habit imprinted by the organic molecule. Such a preservation of the portlandite crystal habit over time can be related to the nonclassical crystallisation of Ca(OH) 2 , which entails the aggregation of meso-or nanocrystals. This crystallisation pathway can allow the preservation of crystals' morphological features with respect to the classical route. As recently shown by Rodriguez-Navarro et al. [3], such a preservation of morphological features is enhanced in presence of organic additives.
As regards the effects of ageing to the overall crystal morphology, the abundance of nanogranular crystals seem to increase with ageing in all modified putties. As described in Section 5.1, the nanogranular habit is assumed by secondary crystals, which are known to form upon ageing of pure lime putty [5]. A similar trend was in fact observed in the control sample in this study. Secondary precipitation is not hindered in additive-inclusive samples and, on the contrary, seems to be promoted as extensive areas of the samples display nanogranular crystals and could be consistently observed at the SEM. Arguably, the presence of the additives contributes to an increase in nucleation density linked with the stabilisation of Ca(OH) 2 PNCs, brought about by various types of organic molecules including lignosulfonates and carbohydrates [3]. The crystallite dimensions seem stable after 6 months of ageing, according to the crystallite size calculated in the directions normal to the (00.1), (10.0) and (10.1) planes. Differences across the used additives (mainly regarding the crystallite size of all directions rather than the crystallite shape) were detected while the putties were freshlyslaked, and maintained throughout the ageing period. A general trend of increase in crystallite size could be noted in the modified putties over the 6 months of ageing and, as described in Section 5.1, it can be related to the secondary precipitation taking place during ageing and is likely enhanced by the presence of the organic additives.

Implications for cement applications
As recently proposed by Rodriguez-Navarro et al. [3], the precipitation of Ca(OH) 2 might play a crucial role in marking the end of the induction period of cement and the beginning of hydration. This is because the precipitation of Ca(OH) 2 makes the solution undersaturated with respect to alite and belite, which start to dissolve, hydrate and precipitate as C-S-H. The organic additives included in this study, and compounds of similar molecular structure, are commonly introduced in cement pastes as admixtures [10,26,37] but along with their desired effect, other side effects may be triggered by their presence. For instance, these compounds are known to affect the crystallisation path of Ca(OH) 2 , retarding its precipitation as a result of nucleation inhibition effect the mechanism thereof is not well established yet and is likely related to the stability of PNCs [3,12,[26][27][28]32]. This would also prolong the induction period and the beginning of hydration reaction, which is likely to be the mechanism by which these organic compounds work as set retarders when added in small percentage to unhydrated cement.
Our results show that the presence of the tested additives can modify the habit of Ca(OH) 2 especially when added before hydration of CaO and that such habit is preserved after ageing, meaning that the organic molecules and their degradation by-products keep playing an active role in affecting Ca(OH) 2 crystal growth. This effect has important implications as the morphology of portlandite is known to affect the mechanical properties of cement [9,4].
Our results also show that the tested additives lead to an overall crystallite size reduction of Ca(OH) 2 measured in the additives-modified putties with respect to plain Ca(OH) 2 , which has been related to lower mechanical strength of portlandite crystals and to a higher proneness to failure [90] as well as to a higher reactivity towards CO 2 resulting in faster carbonation [36]. Thus, the presence of those additives which led to a significant crystallite size reduction in Ca(OH) 2 (sucrose and CaLS) could negatively affect the performance and durability of concrete by losing the alkaline reservoir provided by Ca(OH) 2 . The crystallite size reduction might be the result of several mechanisms. It can be argued that the intercalation of the additive between Ca(OH) 2 particles is the underlying mechanism. However, this process would result in high microstrain which can be detected by XRD [13,36]. The results presented in this work show no or low microstrain in all samples, suggesting that the additives were not embedded into the portlandite lattice, in agreement with previous studies [36,65]. Additives molecules and agglomerates may act as nucleation centres for portlandite, a process that is well-known in biochemistry, e.g. in crystallisation of biogenetic CaCO 3 [91].
Lastly, the large Ca(OH) 2 crystals systematically observed in the CaLS-modified putties were also observed by Rodriguez-Navarro et al. [3]. The authors argued that such large crystals (their crystals, formed by chemical precipitation starting from CaCl 2 and NaOH solutions, were in the order of hundreds of μm whereas ours, obtained by CaO slaking, in the order of few μm) are more prone to mechanical failure, in particular along the < 00.1 > direction, which might lead to a cement paste of reduced strength.

Conclusions
This study investigated the effect of three organic additives commonly used in lime and cement mixtures on the mineralogical, crystallographic and morphological characteristics of portlandite crystals. Results indicate that: • lime slaking in sucrose solution produces changes in the morphology of Ca(OH) 2 crystals, which display a peculiar irregular shape and a significantly smaller crystallite size than those usually found in pure Ca(OH) 2 . This effect was not as marked in the putty slaked in pure water with later addition of solid sucrose. We infer that this difference is consequence of the thermal treatment that sucrose undergoes during the slaking, which leads to a series of degradation processes that foster the formation of species, such as reducing sugars and lactic acid, that adsorb onto Ca(OH) 2 particles and affect their growth. The Ca concentration in the supernatant solution indicates that sucrose has high affinity for Ca ions, and forms stable complexes over the time since Ca concentration remains similar at time 0 and after 6 months of storage.
• the effect of pectin on Ca(OH) 2 morphology is not very marked, both in terms of crystal shape and crystallite size, which are similar to crystals in a pure Ca(OH) 2 system. The high molecular weight of this compound make pectin more resistant than sucrose to thermal degradation, which accounts for the little effect observed on morphology.
• the addition of CaLS to the lime pastes produced remarkable changes in the morphology of portlandite crystals, which systematically displayed a regular plate-like shape with some crystals of size (up to few μm) higher than those seen in crystals of pure lime putty (up to 1 μm). This morphological change is related to a preferential adsorption of CaLS on the basal faces of portlandite.
Overall, the addition of organic additives reduces the crystallite size of Ca(OH) 2 . Upon ageing, while in the control putty portlandite crystals tend to overdevelop the basal faces transforming into plate-like crystals, when additives are present Ca(OH) 2 crystals tend to overdevelop less stable faces. These effects altogether point to the formation of a Ca (OH) 2 phase more susceptible to mechanical failure and more prone to carbonation, which is detrimental when the reaction occurs in reinforced concrete systems.