IntroductionMagnesium potassium phosphate (MKP) cements are normally composed of dead-burnt magnesia and monopotassium phosphate (KH₂PO₄), which react in water and yield the main hydrate of K-struvite (MgKPO₄·6H₂O). MKP cements have been frequently used as rapid repair materials in the field of civil engineering and as solidification/stabilization agents in the field of waste managements, due to the unique properties, such as fast setting, high early strength, low drying shrinkage, strong bonding to old Portland cement concrete substrates, low pH value, and so on. Boron compounds are the de facto standard retarders of MKP cements. Normally the increase of boron compound dosage could prolong MKP cement setting, but result in slower strength development, especially at early ages. Therefore, in order to obtain sufficient setting time and early strength at the same time, research attempts on seeking new retarders have been carrying out in recent years. Aluminum salts, Al(NO₃)₃ and Al₂(SO₄)₃, were among those reported chemicals, showing promising retardation effect, especially at low M/P molar ratios. However, influence of aluminum salts on properties and hydration of MKP cements are yet to be explored, as several important aspects remain unclear, such as their impact differences on MKP cement early and late strengths, roles of Al³⁺ and of different anions on governing the hydration. Therefore, in this work influence of three common aluminum salts, AlCl₃, Al(NO₃)₃ and Al₂(SO₄)₃, on properties and hydration of MKP cements at magnesium-to-phosphate (M/P) molar ratio of 4 and water-to-cement (w/c) ratios of 0.27 and 5.00 was studied.MethodsThe raw materials for the preparation of MKP cements were industrial-grade dead-burnt magnesia and KH₂PO₄. The chemicals in analytical grade were borax (Na₂B₄O₇·10H₂O), Al₂(SO₄)₃·16H₂O, Al(NO₃)₃·9H₂O, and AlCl₃·6H₂O. The MKP cement pastes and suspensions were prepared at same M/P molar ratio of 4, and at w/c ratios of 0.27 and 5.00. The aluminum salts were dissolved in the portioned mixing water before use, and the Al concentrations for the pastes and the suspensions were 600 mmol/L and 32.4 mmol/L, respectively. For comparison purpose, borax was used as reference, of which the dosages were 5.15% and 6.08%, by weight of cement, equal to the dosages of Al₂(SO₄)₃·16H₂O and Al(NO₃)₃·9H₂O, respectively. Vicat needle test was used to determine the paste final setting time. Flexural and compressive strengths of the pastes with the dimensions of 25 mm × 25 mm × 125 mm were measured after 1, 3 h, 1, 7, 28, 90 d and 400 d. Additional strength measurements were carried out after 6 h and 12 h for the paste with Al(NO₃)₃, due to the low strengths within the first 3 h. Hydration heat of the pastes up to 24 h was measured using an isothermal calorimeter (TAM) at 20 ℃, using internal mixing method. Moreover, hydrate assemblages of the pastes after 28 d and 400 d were analyzed using XRD, TGA and FTIR methods. To track the aqueous evolution, pH and electrical conductivity of the MKP cement suspensions were continuously monitored up to 24 h at 25 ℃. Also the aqueous compositions were determined using ion chromatography and inductively coupled plasma after 15, 33 min, 3 h, 1 d and 7 d.Results and discussionCompared to the reference paste, the additions of Al₂(SO₄)₃, Al(NO₃)₃ and AlCl₃ could extend the setting time by around 2.5, 3.1 times and 5.1 times, respectively, corresponding to around 25%, 45% and 80% increases as compared to borax at 6.08%. It indicates better retardation effect of the aluminum salts and the anions taking effect in the order of Cl^-, NO₃^-, and SO₄^2-. In contrast to Al(NO₃)₃ and Al₂(SO₄)₃, AlCl₃ leads to clear sample expansion and cracks, suggesting its unsuitable use in MKP cement-based materials. Moreover, compared to the reference paste, both Al(NO₃)₃ and Al₂(SO₄)₃ lower the early strengths. The strength weakening effect of Al₂(SO₄)₃ mainly occur within the first 3 h, and of Al(NO₃)₃ is more pronounced and lasts longer up to around 12 h. However, thereafter, the strengths of the pastes with Al(NO₃)₃ and Al₂(SO₄)₃ gain fast, leading consequently to similar or even slightly higher strengths at late ages compared to the reference paste.Al(NO₃)₃ shows better effects on slowing down hydration heat release and on decreasing cumulative heat up to 24 h, compared to Al₂(SO₄)₃, in agreement with the setting time results. Both Al(NO₃)₃ and Al₂(SO₄)₃ do not change the main hydrate, which remains K-struvite, but slightly decrease the formation contents. Compared to Al₂(SO₄)₃, Al(NO₃)₃ lowers more the system pH value and extends hydration stages to later reaction times, consistent with the results of setting time and calorimetry, revealing stronger effect of NO₃^- than SO₄^2- under same Al³⁺ concentration. The aqueous composition results suggest well that Al³⁺ is precipitated rapidly at the beginning of reaction, leading to certain amorphous Al-containing phases. At low w/c of 0.27 the anions of NO₃^- and SO₄^2- are precipitated as crystalline KNO₃ and K₂SO₄, filling micro-pores in hardened paste matrices, which indicates attention required on matrix volume stability as high dosages of these aluminum salts are used.ConclusionsThe main findings of this work are concluded as follows. All the investigated aluminum salts (AlCl₃, Al(NO₃)₃ and Al₂(SO₄)₃) could retard MKP cement setting. Furthermore, they show better retardation effect than borax at the same dosage. In contrast to Al(NO₃)₃ and Al₂(SO₄)₃, AlCl₃ could lead to sample expansion and cracks, indicating its unsuitable use in MKP cement-based materials. Moreover, both Al₂(SO₄)₃ and Al(NO₃)₃ could lower the cement early strengths, which occurs within the first 3 h and 12 h, respectively, consistent with their retardation effects. But they show no adverse impacts on the paste strengths at late ages, compared to the reference paste. Compared to Al₂(SO₄)₃, Al(NO₃)₃ is more effective on reducing system pH, extending hydration stage, slowing down hydration heat release, and on inhibiting slightly K-struvite precipitation, which contribute to better retardation effect. In addition to K-struvite, small contents of amorphous Al-containing hydrates and of crystalline KNO₃ and K₂SO₄ are precipitated in the hydrated matrices, affecting the paste properties.