Tingjie Chena,
Min Niua,
Xiaodong Wangb,
Wei Weia,
Jinghong Liu*a and
Yongqun Xie*a
aCollege of Material Engineering, Fujian Agriculture and Forestry University, 350002, Fuzhou, Fujian, China. E-mail: fjxieyq@hotmail.com; 314515363@qq.com; Fax: +86 591 83789135; Tel: +86 591 83789307
bDivision of Wood Technology and Engineering, Luleå University of Technology, 93187, Forskargatan 1, Skellefteå, Sweden
First published on 15th October 2015
Poly-aluminum silicate sulphate (PASS) was synthesized in a mixed aqueous solution of sodium silicate and aluminum silicate via a sol–gel method for use in ultra-low density fiberboard (ULDF). The preparation conditions were optimized by using response surface methodology. The effects and interactions of the Si/Al molar ratio (X1), pH value (X2) and temperature (X3) on the internal bond strength of ULDF were investigated. Research showed that the optimum internal bond strength (10.23 ± 0.64 kPa) was obtained under a Si/Al molar ratio of 2:1, pH value of 8, and a temperature of 50 °C. Analyses of the Fourier transform infra-red spectroscopy spectra confirmed that Al–O–Si bonds were formed between polysilicate and Al or its hydrolysate. The particle size analysis showed that the average size of PASS was 7.52 μm. Some of the PASS entered the cell wall and made a contribution to the improvement of the mechanical properties of ULDF.
Silicon materials such as water glass and aluminum compounds can improve the heat resistance of wood-based materials. When aluminum salts are heated, they absorb a lot of heat from the dehydration reaction.11 In addition, silicon materials can also ameliorate the mechanical properties of wood composites.7,8,12 Poly-aluminum silicate sulphate (PASS) is one of the inorganic polymer coagulants which can be prepared through polymerization of poly-silicic acid and hydroxylated aluminium salts.13 The polysilicic acid can neutralize the positive charge in polyaluminum and also combine with Al and its hydrolysis products through Al–O–Si bonds to form hydroxyl-aluminosilicate due to its negatively charged polymer.14,15 PASS has many distinct abilities including charge neutralization, adsorption bridging and sweep coagulation. When it is stirred with fibers, it covers the surface of the fibers or even interacts with plant fibers which improves the properties of the fibers or fiber composites.16,17
It’s worth noting that the properties of inorganic polymer coagulants are affected by many factors such as aging temperature, silicon dose, pH value, and Si/Al ratio. For example, the coagulation efficiency of an aluminum–silicate polymer composite (PASiC) increases with rising basicity, however, the PASiC products tend to become cloudy or partly gelatinous in this case.18 Li et al. showed that the coagulation performance of a poly-silicic-cation coagulant was affected by the pH value. The Si/Al ratio also played a crucial role in the properties of inorganic polymer coagulants. Different positive charges and molecular weights of PASiC coagulants were obtained when they were prepared in different Si/Al ratios. For example, an increase in the Si/Al ratio would increase the amount of polymeric coordination.18–21
Although the coagulation performances of various inorganic polymer coagulants in water and waste water treatment have been already studied by many researchers, there is still no research on the effects of PASS with different characteristics on the mechanical properties of ULDF. To systemically study the influence of PASS on the mechanical properties of ULDF, PASS materials with different Si/Al molar ratios, pH values, and temperatures were prepared and added into the preparation of ULDF. Additionally, to obtain the optimal preparation conditions of PASS, a standard response surface methodology (RSM) called central composite design (CCD) was used. The chemical structures and particle sizes of the PASS materials were examined and analyzed.
Levels | |||||
---|---|---|---|---|---|
Coded-variables (Xi) | −1.682 | −1 | 0 | 1 | 1.682 |
Si/Al molar ratio (X1) | 0.32:1 | 1:1 | 2:1 | 3:1 | 3.68:1 |
pH values (X2) | 4.64 | 6.0 | 8.0 | 10.0 | 11.36 |
Temperature (X3, °C) | 16.36 | 30 | 50 | 70 | 83.64 |
Y = 10.23 + 0.63X1 + 0.41X2 + 0.33X3 − 0.055X1X2 + 0.078X1X3 − 0.077X2X3 − 2.31X12 − 1.48X22 − 0.81X32 | (1) |
Run no. | Coded levels | Internal bond strength (kPa) | |||
---|---|---|---|---|---|
Si–Al molar ratio (X1) | pH value (X2) | Temperature (X3, °C) | Experimental | Predicted | |
1 | −1(1:1) | −1(6.0) | −1(30) | 3.61 | 4.20 |
2 | 1(3:1) | −1 | −1 | 5.15 | 5.43 |
3 | −1 | 1(10.0) | −1 | 4.83 | 5.28 |
4 | 1 | 1 | −1 | 5.94 | 6.29 |
5 | −1 | −1 | 1(70) | 4.48 | 4.87 |
6 | 1 | −1 | 1 | 6.12 | 6.41 |
7 | −1 | 1 | 1 | 5.18 | 5.64 |
8 | 1 | 1 | 1 | 6.81 | 6.96 |
9 | −1.68(0.32:1) | 0(8.0) | 0(50) | 3.40 | 2.63 |
10 | 1.68(3.68:1) | 0 | 0 | 5.03 | 4.76 |
11 | 0(2:1) | −1.68(4.64) | 0 | 5.94 | 5.38 |
12 | 0 | 1.68(11.36) | 0 | 7.22 | 6.74 |
13 | 0 | 0 | −1.68(16.36) | 8.01 | 7.38 |
14 | 0 | 0 | 1.68(83.64) | 8.91 | 8.50 |
15 | 0 | 0 | 0 | 10.10 | 10.23 |
16 | 0 | 0 | 0 | 10.31 | 10.23 |
17 | 0 | 0 | 0 | 9.97 | 10.23 |
18 | 0 | 0 | 0 | 10.77 | 10.23 |
19 | 0 | 0 | 0 | 10.48 | 10.23 |
20 | 0 | 0 | 0 | 9.59 | 10.23 |
In general, the exploration and optimization of a fitted response surface may produce poor or misleading results unless the model exhibits a good fit.24–27 As shown in Table 3, the model’s fit was good because its p-value was less than 0.0001 and the lack of fit value was 0.0956. The determination coefficient (R2) of this model was 0.9667, which implied that 96.67% of the variations could be explained by the fitted model. The value of R2 was in reasonable agreement with Radj2 which indicated a high degree of correlation between the observed and predicted value.28 So, the values of R2 (0.9667) and Radj2 (0.9368) in this model proved that the regression model could explain the true behavior of the system well.
Source | Sum of squares | Degrees of freedom | Mean square | F-Value | p-Value |
---|---|---|---|---|---|
a p < 0.01 highly significant; 0.01 < p < 0.05 significant; p > 0.05 insignificant. | |||||
Model | 112.28 | 9 | 12.48 | 32.29 | <0.0001 |
X1 | 5.49 | 1 | 5.49 | 14.22 | 0.0037 |
X2 | 2.26 | 1 | 2.26 | 5.84 | 0.0362 |
X3 | 1.53 | 1 | 1.53 | 3.96 | 0.0745 |
X1X2 | 0.024 | 1 | 0.024 | 0.063 | 0.8075 |
X1X3 | 0.048 | 1 | 0.048 | 0.12 | 0.7317 |
X2X3 | 0.048 | 1 | 0.048 | 0.12 | 0.7317 |
X12 | 77.04 | 1 | 77.04 | 199.38 | <0.0001 |
X22 | 31.39 | 1 | 31.39 | 81.25 | <0.0001 |
X32 | 9.49 | 1 | 9.49 | 24.55 | 0.0006 |
Residual | 3.86 | 10 | 0.39 | ||
Lack of fit | 3.01 | 5 | 0.60 | 3.54 | 0.0956 |
Pure error | 0.85 | 5 | 0.17 | ||
Correlation total | 116.15 | 19 |
The corresponding variables would be more significant with greater F-values and smaller p-values.29 As can be seen in Table 3, the F-value (32.29) and p-value (less than 0.0001) implied that this model was significant and only a 0.01% chance that it could occur due to noise. On the contrary, the F-value (3.54) and p-value (0.0956) of the lack of fit implied that it was not significant and a 9.56% chance that it could occur due to noise. The p-values in this model indicated that X1, X2, and the three quadratic terms (X12, X22, and X32) affected the IB of the ULDFs significantly, whereas X3, X1X2, X2X3, and X1X3 were all insignificant to the response. The results also showed that the independent variable X1 was the most significant factor on the experimental effect of the IB of ULDF.
The results showed that the variables X1 and X2 played an important role in the IB of ULDFs, whereas X3 did not. At a Si/Al molar ratio of 1.8:1–2.2:1 and a pH value of 7.8–8.1, a maximal IB (10.23 kPa) could be determined (Fig. 3a). But, at a given Si/Al molar ratio, the IB of ULDFs decreased with too low or too high pH values. This was because the morphostructure of the fibers and the foaming system of the ULDF might be affected by the low or high pH value. Additionally, the particle size of PASS was small at low pH values which might limit them leaving in ULDFs, whereas the particle size was large at high pH values which would not be helpful to the distribution of Si/Al compounds on the fiber surfaces. On the other hand, the decrease in IB at higher Si/Al molar ratios was possibly due to the distribution of additives and their charge neutralization which was the major and effective mechanism for the absorption of PASS on the fiber surface.22,30 The IB of ULDFs increased with temperature, especially within the range from 45 to 60 °C, but declined slightly at higher temperatures (Fig. 3b). This was because the particle sizes of PASS were larger at higher temperatures which are always prone to agglomerate on the fiber surface. Fig. 3c shows the combined effect of pH value and temperature on the IB of ULDFs at a constant Si/Al molar ratio (2:1). The result was elliptical, indicating significant interactive effects of the two independent variables on the IB of ULDFs.31
As seen in Fig. 3, the interaction between the Si–Al molar ratio (X1) and the other two variables (X2, X3) was significant. This was because more and more hydroxyl ions were added as the pH value increased which made the formation of sediment between aluminium and silicate ions easier. The hydrolyses of aluminium and silicate were significantly affected by the temperature of the solution. Meanwhile, there was almost no interaction between the pH value (X2) and temperature (X3). The pH value of the solution might be affected by the temperature but not significantly.
Based on eqn (1) which was derived from the computational program, the optimal PASS conditions for ULDFs were a Si/Al molar ratio of 2.14:1, pH value of 8.26, and a temperature of 54.13 °C. Taking the practical operating conditions into consideration, some conditions were modified as follows: a Si/Al molar ratio of 2:1, pH value of 8, and a temperature of 50 °C. Under these conditions, an average of 10.23 ± 0.64 kPa was obtained, which is close to the model predicted value of 10.34 kPa. Compared to the control specimen (2.5 kPa), the IB of ULDF under the optimal conditions was increased by 313.6%. These results confirmed that the model adequately reflected the expected optimization and eqn (1) was satisfactory and accurate.
The peaks at around 3400 and 1666 cm−1 were attributed to the –OH stretching vibrations and –OH bending vibrations, respectively. The peaks for aluminum sulfate at 1104, 931, and 606 cm−1 were attributed to the SO42−, Al–O–Al, and Al–OH vibrations, respectively. The peaks for polysilicic acid at around 1380, 1114, 798, 619 and 441 cm−1 were ascribed to the contribution of the silicon–oxygen tetrahedral, Si–O–Si or SO42−, Si–O–Si, SO42−, and O–Si–O vibrations, respectively. Comparing the infrared spectra of PASS to polysilicic acid and aluminum sulfate, the peak at 1100 cm−1 might be attributed to the Si–O–Si, SO42−, or Si–O–Al vibration. This is due to the broad band in the range of 1200–1000 cm−1 which usually corresponds to the mixed overlap of Si–O–Si and Al–O–Si bonds.33 In addition, the peak for PASS at 931 cm−1 was weakened. This is due to the polysilicic acid which could combine with Al and its hydrolysis product through Al–O–Si bonds to form hydroxyl-aluminosilicate, which led to the decrease in Al–O–Al bonds. Combining the results of Tang et al. and this study, the structure of the Si sol and PASS could be deduced as shown in Fig. 5.32
As can be seen, the distribution of PASS' particle size in Fig. 6, they was mainly distributed from 2 μm to 15 μm. The average and largest particle sizes were 7.52 μm and 18.98 μm, respectively. Combined with the cumulative curve, it could be found that there was 26.02% of PASS whose particle size was less than or equal to 5.25 μm. Due to the porosity of the fibers which have many pits ranging from 0.1 μm to 5.0 μm on their cell wall, some of the PASS with a smaller particle size could enter and leave the cell wall or cell cavity.22 Therefore, the PASS could not only deposit on the fiber surface through charge neutralization or adsorption bridging, but enter the cell cavity through the fiber pits and then become a sediment to enhance the mechanical properties of ULDF.4
On the contrary, the efficiency of coagulants in waste water treatment might be influenced by increasing the Si/Al ratio and pH value. The results were different from the PASS for ULDFs. The study by Yang et al. showed that for a given Si/Al ratio, an increase in pH value increased the DOC removal efficiency.21 This was because that higher pH value produced larger molecular sized products which would enhance aggregating efficiency. Additionally, for a given pH value, the DOC removal efficiency increased then decreased when the Si/Al ratio increased. This was because of the interaction between hydrolyzed Al species and polysilicic acid which would decrease the positive charge and increased the molecular weight of the coagulant.11
In this study, for a given Si/Al ratio and temperature, an increase in pH value would increase the particle size of PASS. Different particle sizes of PASS were obtained at different pH values so the IB increased and later decreased in Table 4. The smaller particle size of PASS which was obtained under the conditions of lower pH values could enter into fibers; they might easy to run off when they were added in the preparation of ULDF.30 On the other hand, an increase in pH value would promote agglomeration and probably increased the molecular weight in PASS. The mechanical properties of ULDF would be affected by the large particle size of PASS which could influence the distribution of PASS on the fiber surface.
PASS | Si–Al molar ratio | Temperature (°C) | pH value | Particle size (μm) | IB (kPa) | DOC removal efficiency |
---|---|---|---|---|---|---|
This study | 2 | 30–40 | 4 | 0.40 | 6.24 | — |
6 | 0.90 | 7.41 | — | |||
8 | 5.13 | 9.95 | — | |||
12 | 12.59 | 8.73 | — | |||
Yang et al.21 | 0.05 | Ambient temperature | 4 | — | — | 9.1% |
6 | — | — | 27.1% | |||
8 | — | — | 20.8% | |||
0.02 | Ambient temperature | 4 | — | — | 7.7% | |
0.05 | — | — | 9.1% | |||
0.10 | — | — | 3.4% |
Taking these factors into consideration, the suitable particle size of PASS which played an important role in the mechanical properties of ULDF should be prepared. Combined with the results of FTIR and particle size, the optimal preparation conditions of PASS were valid for preparing ULDFs.
This journal is © The Royal Society of Chemistry 2015 |