Interaction of Aluminum Metaphosphates in the Setting of Potassium Silicate Solutions in Terms of the Crystalline Phase Composition.

Aluminum phosphates are known as inorganic hardening agents for the setting of alkali silicate solutions, but only few studies have been published on the setting mechanism of potassium water glass. The solution behavior of two aluminum metaphosphates in alkaline environments were investigated photometrically determining the dissolved aluminum content. The crystalline phase composition of the hardened potassium silicate systems was determined by X-ray diffraction. New insights into the setting mechanism were obtained concerning the structure of the aluminum metaphosphate and the SiO2/K2O ratio of three different potassium silicate solutions. With increasing pH value aluminum tetrametaphosphate reacts rapidly and forms crystalline potassium tetrametaphosphate dihydrate by an ion-exchange-reaction. In parallel, a depolymerization of the cyclic metaphosphate structure occurs leading to potassium dihydrogen phosphate as final fragmentation product. With aluminum hexametaphosphate no ion-exchange reaction product was observed. Only potassium dihydrogen phosphate could be found in higher quantities compared to the reaction with aluminum tetrametaphosphate.

Introduction

Potassium silicate solutions, also known as potassium water glasses, are colloidal potassium silicates sols with a typical molar SiO2/K2O ratio ranging from 2.5 to 4.5.1 They are used in industrial applications as acid‐resistant mortars and cements. They are main components for industrial floors and concretes in the field of construction chemistry,2 as well as in silicate paints for facades.3 The setting behavior of potassium silicate solutions with inorganic hardening agents like aluminum metaphosphates has, so far, only been insufficiently studied. Since now, only a few studies based on the reaction of sodium silicate solutions with aluminum tetrametaphosphate have been published,4 while the interaction in the potassium silicate system and the impact of different aluminum metaphosphate modifications have not been in focus of research so far.

During chemical hardening an irreversible solidification reaction of the silicate binder system takes place, which leads to a high chemical resistance towards acid attack.4a, 5 A special feature of the potassium silicate binder is the formation of non‐hydrated potassium sulphates, when getting into contact with sulphuric acid. This is the main benefit towards sodium silicates, since the sodium sulphates contain crystal water which provoke component damage due to efflorescence, as a result of the incorporation of crystal water. The same phenomenon applies to the corresponding carbonates when reacting with CO2.2, 3 Aluminum metaphosphates, also referred as cyclophosphates, are aluminum salts of the ring‐shaped metaphosphoric acids with the general empirical formula (HPO3)n. They consist of corner connected PO4‐tetrahedra, which are present in ring sizes of n=4, 6, and 9.6 The only commercially pure product is aluminum tetrametaphosphate, Al4(P4O12)3. It has been reported that the synthesis of metaphosphates is generally difficult because, in addition to weight and temperature profile, water vapor partial pressure plays an important role in the representation of the respective modifications, which is considerably more difficult to adjust.7 For the other modifications, no synthesis routes are described leading to pure reaction products, which can subsequently be reproduced on an industrial scale. Most of the reported structure data are based on X‐ray diffraction studies were single crystals were manually extracted from the product mixture.8 Beside hardening agents, aluminum metaphosphates and their mixtures are used as high‐temperature binders in refractory products (green strength, thermal stability and abrasion resistance),9 in sealings of oxidic ceramics, in technical glasses, glass ceramics and as low‐temperature seals (low thermogravimetry, low softening temperatures, high thermal expansion coefficient, high UV transparency).9a

The aim of this work is the determination of a better insight into the setting mechanism of potassium silicate solutions in respect to the potassium water glass’ molar SiO2/K2O ratio. Furthermore, the influence of the crystalline phase composition in respect to the aluminum metaphosphate structure was investigated.

Results

Solid State Synthesis of Al2P6O18

Aluminum hexametaphosphate was synthesized in a new solid‐state reaction route in a purity of >96 wt% and impurities of aluminum tetrametaphosphate. In contrast to reported synthesis routes a nearly pure product could be determined.8b, 8c

Dissolved Aluminum Content

The concentration of the dissolved aluminum content of the aluminum metaphosphates in alkaline sodium hydroxide solutions was determined by UV/Vis photometry. The results of the aging tests of aluminum tetrametaphosphate at the four different pH values is displayed in Figure 1. A strong dependence of the dissolution rate on the pH value of the alkaline solution can be observed. According to the initial weight of the aluminum tetrametaphosphate (1 g/l), a complete dissolution of the aluminum ions corresponds to a maximum concentration of 102 mg/l. The ageing at pH value of 11.0 shows a slight increase in the dissolved aluminum ion concentration to 8 mg/l within the 30 days measuring period, which corresponds to a total solution of 8 %. With increasing pH value, the concentration of dissolved aluminum ions from the aluminum

Figure 1

Concentration curve of the aluminum ions of the aluminum tetrametaphosphate in solution during ageing at different pH values determined by UV/Vis photometry.

tetrametaphosphate increases significantly, which is in good agreement to reported studies.4a At pH 12.5, half of the aluminum is dissolved after 15 days, and after 30 days almost the complete aluminum content is dissolved.

Figure 2 displays the results of the aging tests of aluminum hexametaphosphate (1 g/l) at pH values ranging from 11.0 to 12.5. Again, the maximum possible aluminum ion concentration is 102 mg/l. In comparison to aluminum tetrametaphosphate, aluminum hexametaphosphate shows a completely reverse dissolution behavior as a function of the pH value. Although the proportion of dissolved aluminum rises steadily with increasing ageing time in all measurements, the concentration of dissolved aluminum ions decreases with increasing pH value. At pH 11.0, the entire aluminum content is dissolved after 34 days. With increasing pH, the amount of dissolved aluminum ions decreases, and reaches 43 % at a pH value of 11.5, approx. 30 % at pH=12.0 and 15 % of the maximum concentration at a pH value of 12.5.

Figure 2

Concentration curve of the aluminum ions of the aluminum hexametaphosphate in solution during ageing at different pH values determined by UV/Vis photometry.

Since the dissolution behavior of the aluminum content of the aluminum metaphosphates is in contrary to each other, it is assumed that the reactivity of the hardening agents and the setting mechanism differ from each other. This result confirms reported observations on the growth in stability of metaphosphate structures with increasing ring size towards cleavage in alkaline environment.10

X‐Ray Powder Diffraction and Rietveld Refinement

In Figure 3 the X‐ray powder diffraction patterns of the mixtures of potassium silicate binders K1, K2 and K3 with aluminum tetrametaphosphate, respectively K1AXX, K2AXX, and K3AXX, are illustrated.

Figure 3

X‐ray powder diffractograms of the test series K1AXX, K2AXX and K3AXX. ▪ Al4(P4O12)3, • KH2PO4, ▴ K4P4O12 ⋅ 2 H2O, ▾ KZnPO4 ⋅ 0.8 ♦ H2O, ZnO

The broad background signal with its center at about 28° 2 theta presents the amount of X‐ray amorphous phase content.4a With increasing hardener content, the X‐ray amorphous content decreases. The qualitative phase assignment of the diffraction pattern results that beside unreacted hardener, a cubic potassium‐containing metaphosphate phase, potassium tetrametaphosphate dihydrate, is formed in small quantities. In addition, a monoclinic potassium dihydrogen phosphate phase is generated. A zinc‐containing phosphate phase, potassium zinc phosphate hydrate, is formed by the homogenization process of the powder samples with zincite as internal standard, which is obtained in several diffraction patterns. The KH2PO4 structures can be described as the final reaction products of the depolymerization of the metaphosphate structure, which is formed regardless of the aluminum metaphosphate modification of the system. With increasing pH value of the potassium silicate solution, the formed KH2PO4 content decreases and is nearly missing in the samples with potassium silicate solution K3. The relatively lower formation of crystalline hydrogen phosphate, Na2HPO4, in sodium silicate systems can be contributed to existence of the additional crystalline sodium diphosphate structures, which could not be found in any potassium silicate sample.4a, 11 Since the diphosphate structures can be described as intermediate reaction products in the depolymerization of the cyclic metaphosphate structure, either corresponding potassium diphosphate structures are only present in the X‐ray amorphous phase content, or they are decomposed to the potassium dihydrogen phosphate. The last option can describe the stronger formation of the potassium hydrogen phosphate as a result of higher depolymerization. In contrast to reported reaction products of the setting of sodium silicate solutions with aluminum tetrametaphosphate, no alkali carbonate, alkali hydrogen carbonate structures or diphosphate structures could be determined.4a, 12 This can be described by a different reaction behavior of the potassium silicates compared to sodium silicate binders. Another explanation can be attributed to the different sample pretreatment routine having a significant impact on the formation of the phase composition.4a, 11 For the quantitative crystalline phase analysis and the quantification of the X‐ray amorphous phase content, all diffractograms were evaluated using the Rietveld method.

Figure 4 displays the X‐ray powder diffraction patterns of the samples with potassium silicate solution K2 and aluminum hexametaphosphate as hardening agent (K2BXX). Since the reflections have low intensities due to the high Zincite content (50 to 80 wt%), an enlarged section of the diffractogram displays the relevant phases. In addition to the proportion of aluminum hexametaphosphate, which remains unreacted, only potassium dihydrogen phosphate was found as a crystalline phase. Figure 5 illustrates the graphical evaluation of the Rietveld refinement of sample K1A30.The multi‐color curve corresponds to the measured diffractogram, with the colors indicating the different phase reflexes. The diffractogram calculated by Rietveld refinement is shown in red. The grey curve corresponds to the difference between the experimentally determined diffractogram and the calculated diffractogram, which indicates the quality of the Rietveld fit.

Figure 4

X‐ray powder diffractograms of the sample series K2BXX with an enlarged section. ▪ Al2P6O18, • KH2PO4, ♦ ZnO.

Figure 5

Result of the Rietveld refinement of the diffraction pattern of sample K1A30.

The results of the quantitative phase analysis of all the sample series are summarized as a bar graph in Figure 6. In all samples, the X‐ray amorphous fraction decreases continuously with increasing hardener content. For samples with K1 as binder, the proportion decreases from 81 % for K1A05 to 60 % for K1 A30. For the samples with K2, the X‐ray amorphous proportion drops from 93 % to 46 %, and from 95 % to 68 % with K3AXX. The samples with K2 and aluminum hexametaphosphate show a decrease in the X‐ray amorphous fractions from 96 % to 37 %. As a result, the average X‐ray amorphous fraction is the lowest with the K2BXX sample series, and the highest with the K1AXX sample series. The samples of the K1AXX series reveal that the amount of the unreacted aluminum tetrametaphosphate rises with the increasing proportion of hardener. The potassium tetrametaphosphate dihydrate remains in the range of 2±1 % despite the increase in hardener. The monoclinic potassium dihydrogen phosphate shows no clear correlation with the metaphosphate amount and is in the range between 1 and 7 %. In the K2AXX samples, the proportion of unreacted aluminum tetrametaphosphate is initially lower than for K1A05 and K1A10, whereas it increases significantly to 28.5 % with K2A20. The amount of potassium tetrametaphosphate dihydrate grows continuously with increasing hardener content from approx. 2 % for K2A05 to 11 % for K2A30. The formation of crystalline potassium dihydrogen phosphate begins with 15 wt% of the hardening agent and increases from 5 % to 11 % for K2A30. The samples with potassium silicate binder K3 show that the amount of the unreacted hardening agent is considerably lower than the samples with K1 and K2. The amount of potassium tetrametaphosphate dihydrate grows from 3 % to 8 % with increasing metaphosphate content. Potassium dihydrogen phosphate could only be detected in sample K3A30, with around 1 %. In the K2BXX sample series, the amount of unreacted aluminum hexametaphosphate rises from 4 % to 52 %.

Figure 6

Result of the quantitative phase analysis by Rietveld refinement. K1–K3=potassium silicate solutions, A=aluminum tetrametaphosphate, B=aluminum hexametaphosphate, C=unreacted metaphosphate, D=potassium tetrametaphosphate dihydrate, E=potassium dihydrogen phosphate, F=potassium zinc phosphate hydrate, G=X‐ray amorphous content.

In contrast to the formed potassium tetrametaphosphate dihydrate with aluminum tetrametaphosphate, no potassium hexametaphosphate phase could be detected. Beside unreacted metaphosphate, only potassium dihydrogen phosphate could be detected starting from K2B10 with around 13 %, which does not evidence a clear tendency with increasing hardener content. It could be demonstrated, that the chemical hardening agent used in all samples did not react completely. The sample series K3AXX contained the smallest amount of unreacted metaphosphate. This implies that the reactivity of the aluminum tetrametaphosphate augments with increasing alkalinity and thus, lower SiO2/K2O molar ratio. Comparing the quantities of unreacted hardener in respect to the metaphosphate structure between samples K2AXX and K2BXX, the remaining metaphosphate dominates for aluminum tetrametaphosphate until sample K2A20. On the contrary, its content for K2A30 is nearly half of its quantity in K2B30. Consequently, it can be concluded that the aluminum hexametaphosphate decomposes much stronger up to 20 % hardener content, which means that it has a weaker stability against the alkaline environment of the potassium silicate system. Sample K2B30 shows an opposite result, since hardly any hexametaphosphate decomposes and the stability increases considerably towards the aluminum tetrametaphosphate. This can be attributed to an inhomogeneous distribution of the hexametaphosphate units within the sample. Regardless of the amount of aluminum metaphosphate as hardening agent fractions of unreacted hardener remain in the samples. With higher hardener content the relative amount of unreacted metaphosphate increases, in good agreement to reported results.4a, 11 While crystalline silicate containing fractions are observed in the setting of sodium silicate solutions with aluminum metaphosphates,4a in potassium silicate solutions no evidence can be found in the presence of crystalline silicate structures, which can be explained as the silicate phases must be in the x‐ray amorphous content.

Discussion

The studies have shown that chemically hardened potassium silicates with aluminum tetrametaphosphate undergo an ion‐exchange reaction by forming potassium tetrametaphosphate dihydrate K4P4O12 ⋅ 2 H2O. The aluminum ions of the metaphosphate structure exchange with potassium ions from the potassium silicates. Both the reactant, aluminum tetrameta‐phosphate, and the newly formed potassium metaphosphate modification are based on a cubic crystal lattice with similar lattice parameters within the unit cell. During chemical initiated hardening of the potassium silicate K2 with aluminum hexametaphosphate, no analogous ion‐exchange reaction could be detected. The reason for the absence of an analogous reaction is to be found in the different dissolution behavior of the two aluminum metaphosphates in alkaline environment in respect to the pH value and the fact, that an ion‐exchange reaction in aluminum hexametaphosphate would require an energetically unfavorable symmetry change from a monoclinic to a cubic crystal system.13 It could be demonstrated, that with increasing pH value, the proportion of potassium tetrameta‐phosphate dihydrate formed by the ion‐exchange reaction between the aluminum tetrametaphosphate and the potassium silicate binder increases. However, this has not been observed on the basis of quantification, since the highest proportion of potassium tetrametaphosphate dihydrate is present in the samples with K2. However, when comparing the amounts of unreacted metaphosphate, it can be concluded, that the strongest decomposition of metaphosphate took place in the samples with K3. Due to the reaction of the metaphosphate with the alkaline potassium silicate solution, the pH value of the metaphosphate decreases continuously, which results in an accelerated setting of the binder system. Due to the faster hardening, the mobility of the ions decreases significantly, which could lead to comparatively smaller amounts of potassium tetrametaphosphate dihydrate being formed than in the K2AXX sample series, despite the higher pH value. When comparing the amounts of potassium dihydrogen phosphate, the largest amounts are to be expected in the K3AXX sample series, as a result of the higher alkalinity of the water glass. Potassium dihydrogen phosphate could not be detected in any sample of the K3AXX sample series. Consequently, an accelerated setting can also be assumed in this case, which means that these phases cannot be formed at all.

Finally, the results can be summarized as follows: the chemically initiated hardening of potassium silicate solutions with aluminum tetrametaphosphate results in a dissolution of the aluminum ions in the alkaline potassium silicate environment and leads to an ion‐exchange reaction through the partial replacement of the dissolved aluminum ions with potassium ions from the water glass. Subsequently, the cyclic phosphate structure is depolymerized to form potassium dihydrogen phosphate as the final crystalline decomposition product. Since there are no indications of crystalline silicon‐containing or aluminum‐containing phases, it can be assumed that these are present in the X‐ray amorphous phase fraction. It can be deduced that during the setting of the silicate network structure additional alumosilicate phases are formed by incorporating the dissolved aluminum into the silicate network.

Conclusions

It could be proven, that the mechanism of setting of potassium water glasses is strongly dependent on the structure of the aluminum metaphosphate hardener. The reaction of potassium silicate solution and aluminum tetrametaphosphate is a two‐stage process. In the first stage the aluminum ions are dissolved from the cyclic phosphate matrix and are exchanged by the potassium ions of the water glass. The drop of the pH value leads to a polymerization of the amorphous silicate binder matrix. In the second stage a depolymerization of the cyclic phosphate structure takes place and leads to the final crystalline product, potassium dihydrogen phosphate. Intermediate structures of the ring depolymerization could not be reported by the X‐ray studies. It is assumed, that these structures are present in the amorphous phase content. Moreover, the formation of an X‐ray amorphous alumosilicate binder network is expected, since no silicon and no aluminum containing phases could be detected by PXRD. It has already been described in literature that, crystalline sodium carbonate and crystalline di‐ and oligo‐phosphates as well as hydrogen phosphates are formed during the chemical hardening of sodium water glasses with aluminum tetrametaphosphate. In this paper there is no evidence that analogous phases exist. The absence of potassium carbonate phases can be explained with the controlled setting conditions (closed sample containers and climatic chamber) as well as with the conscious forego of thermal pre‐treatment. The higher condensed phosphate structures expected during depolymerization of the cyclic phosphate structure could not be detected in any of the samples. It can be supposed, that these phases are formed in the X‐ray amorphous phase fraction, and can, therefore, not be detected by X‐ray diffraction techniques.

Experimental Section

Synthesis of Aluminium Hexametaphosphate, Al2P6O18 (Type B)

1.44 g (18.5 mmol) aluminium hydroxide Al(OH)3 (Sigma Aldrich, reagent grade) and 8.56 g (64.8 mmol) diammonium hydrogen phosphate (NH4)2HPO4 (Sigma‐Aldrich, ≥98 %) were used in a molar ratio of 1 : 3.5. The reactants were manually ground, homogenized and heated to T=425 °C in a corundum crucible in a muffle furnace at a heating rate of 5 K/min, and cooled down to ambient temperature after a holding time of 60 hours. The strongly acidic residue was cleaned in several washing processes with distilled water to remove the phosphoric acid formed from the hydration of the excess P2O5. After subsequent drying, the result was: 4.5 g of a fine, white powder of aluminium hexametaphosphate with minor impurities of aluminium tetrametaphosphate.

Solubility Tests of the Aluminium Metaphosphates Used in Alkaline Environments

A controlled alkaline environment was adjusted using sodium hydroxide as an aqueous solution at four different pH values of typical water glasses (pH=1.0, 11.5, 12.0, and 12.5). Aluminium tetrametaphosphate and aluminium hexametaphosphate were exposed to a concentration of 1 g/l in the alkaline environment for different periods of time (3 days, 5 days, 8 days, 15 days, 21 days and 34 days) and, subsequently, filtered off.

UV/Vis‐Photometric Determination of the Dissolved Aluminum Content

The photometric analysis of the dissolved aluminium content of the samples were carried out on a photoLab 7600 UV/Vis (WTW). The concentration range of the chosen Al test (Spectroquant® 114825, Merck) was between 0.02 and 1.20 mg/l a 10 mm cuvette setup. The samples were diluted with distilled water and measured in a semi‐quantitative analysis for the respective concentration range.

X‐Ray Powder Diffraction

Potassium water glasses of the Betol series (Woellner GmbH, Ludwigshafen, Germany) with the SiO2/K2O molar ratios 4.01 (K1), 3.52 (K2) and 3.09 (K3) were used. The respective chemical hardener was added to the water glass in proportions of 5, 10, 15, 20, and 30 wt% and homogenized in an overhead mixer at 90 rpm for a period of four days. The solidified samples were stored in a climatic chamber at 25 °C and a relative humidity of 70 % for six weeks until complete setting. After manual grinding, zincite was added as an internal standard (10 80 wt%) and homogenized. For the XRD analysis, the samples were prepared via back‐loading into the sample carriers to reduce the degree of preferential orientation. The samples were measured on a Bruker D8 Discover in a Bragg‐Brentano geometry with a Cu Kα tube. The phase assignment of the reflexes was performed by comparison with indexed reflexes from the PDF 2014 database (ICDD).

Quantification by Rietveld Refinement

For the Rietveld refinement, structural data were taken from the Find it database (ICSD) as well as from the primary literature; the lattice parameters (a, b c, α, β, γ), crystallite size, and micro stresses were refined. The emission profile for laboratory diffractometers proposed by Berger was used to describe the Cu Kα emission characteristics,14 partially extended with a Cu Kß line, when measurements were performed without a Ni filter.

Conflict of interest

The authors declare no conflict of interest.

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supplementary

Click here for additional data file.