Sage Journals: Discover world-class research

Abstract

Boron chloride in the chlorosilane is hard to remove and directly impacts the quality of polysilicon produced by improved Siemens method. Silicon dioxide functionalized with mannitol and citric acid successfully removed boron chloride in organic solution. The effects of immersion concentration and drying temperature were studied to attain the best adsorption performance. The sorption phenomenon was described well by pseudo-second-order kinetic model and Langmuir adsorption isotherm; particle diffusion model proved that the adsorption on the active adsorption site was the rate determining step, with the formation of boronic esters. Thermal stability and degradation kinetic of the adsorbents were investigated by thermogravimetric analysis. The characterization results of Fourier transform infrared spectroscopy, thermogravimetric analysis, X-ray energy spectrometry, and BET showed the silicon dioxide was successfully functionalized by oxygen-containing functional groups, leading to be a better and safer substitute to traditional boron-selective chelating resin.

1. Introduction

In the face of an energy crisis and environmental degradation and growing demand for low carbon emission reduction, the unlimited, nonpolluting, and extremely abundant solar power is seen as an effective decarbonization scenario [1], and the market of photovoltaic technology is rapidly evolving with a compound annual growth rate (CAGR) equal to 34% between 2010 and 2020 [2]. Silicon-based solar cells account for more than 94% of the PV market due to the low cost, high photoelectric conversion efficiency, and stability [3]. The quality of polysilicon products greatly affects the photoelectric conversion efficiency and stability.

At present, improved Siemens method [4] is the main production process of polysilicon. Trichlorosilane is prepared by hydrogenation of silicon tetrachloride, and then trichlorosilane is reduced by hydrogen and deposited on the silicon core heater to produce the polysilicon. As the typical acceptor impurity, the trace boron impurities in the raw material of trichlorosilane directly affect the quality and conduction type of polysilicon products. The tolerance limit of boron impurity in solar-grade polycrystalline silicon required by China [5] is below $0.1 \times 10^{- 9}$ , and boron impurity in electronic-grade polycrystalline silicon [6] is required to be less than $5 \times 10^{- 11}$ . The trace boron compounds are complex in the trichlorosilane. Conventional wisdom holds that the main boron compound impurities are BCl₃, B₂Cl₆, or other complex forms. The boiling point of these impurities is too close to chlorosilane to be separated by multistage distillation [7]. The purification methods [8] of boron impurity from chlorosilane include extraction, complexation, adsorption, and partial hydrolysis [9, 10]. Some of these techniques are effective but have some limitation. The partial hydrolysis tends to lead to strong silicification of the device and generate large amounts of corrosive hydrogen chloride [11, 12]. The extraction method [13, 14] is highly efficient, but overall, a lot of work is still needed to find a suitable complexing agent, and there is a lack of in-depth research on the complexing process and mechanism in the actual industrial production.

Compared with other methods, the adsorption method is currently the most widely used in the industry, and the commonly used adsorption material includes activated carbon [15], ion-exchange resin [16, 17], or boron-selective chelating resin (e.g., Amberlite RA-743, Dowex BSA-1, and D564) [18, 19], which has a macroporous polystyrene backbone to which the functional group (glucosamine) is attached. Glucosamine can bond covalently to boron impurities, resulting an excellent boron removal performance. However, the chlorosilane is flammable and explosive. The large amount of flammable chelating resin will pose a huge hidden hazard in case of leakage accidents, resulting in the need for the safe and cheap boron adsorption agent.

The boron impurities like BCl₃ in chlorosilane exhibit similar electron deficiency to the H₃BO₃ in aqueous solution, which has been well studied, and some mentioned commercial boron-selective chelating resin has been developed and widely used in seawater desalination, boron halogenation, and other fields [19]. Further researches focus on material with higher selectivity and easier separation, and Oladipo and Gazi [20, 21] developed hydroxyl-enhanced magnetic material that performed better and can be easily separated from the suspensions by an external magnet, leading to be a techno-economic substitute to the existing boron adsorbents [22]. It is concluded that most reported boron adsorbents for H₃BO₃ in aqueous solution have electron-rich group that plays a key role in the boron removal, leading to the electron-rich oxygen-containing inorganic adsorbent for BCl₃ in this study.

As the inorganic and nonflammable material, the surface of silicon dioxide is covered with the large number of electron-rich hydroxyl groups [23], so it was applied in the removal of boron impurities [24]. Besides, it possesses the advantages of being easily modified, the stable physical and chemical properties, good mechanical strength, high specific surface area, and pore capacity [25].

The study is aimed at investigating boron chloride adsorption on silicon dioxide impregnated with different functional groups and attaining better removal performance, which was evaluated using equilibrium, kinetic, and thermodynamic studies.

2. Materials and Methods

2.1. Reagent and Instrument

All chemicals, silicon dioxide (CP, purchased from Tianjin Guangfu Technology Development Co., Ltd.), boron chloride in hexane (CP, provided by Aladding), methylamine-H, L-ascorbic acid, ammonium acetate, acetic acid, disodium EDTA dihydrate, tartaric acid, (CP, provided by Macklin), meglumine, mannitol, citric acid, salicylic acid (CP, bought from Heowns), and boron-selective chelating resin (D564) (provided by Shanghai Yuanye Bio-Technology Co., Ltd.), were directly used and without further purification.

Ultrasonic cleaner (SK7200B), Shanghai KD Ultrasonic Instrument Co., Ltd; UV-Visible spectropho-tometer (754 N), Shanghai Yidian Analytical Instrument Co., Ltd.

Fourier transform infrared (FT-IR) spectrum was collected on a Thermo Fisher Nicolet iZ10 using KBr as standard reference in the range of 400–4000 cm^-1. The thermogravimetric analyses (TGA) were performed on a NETZSCH 209F3 derivatograph. The data recording occurred in 250 ml∙min^-1 outflow air, within the 30–950°C temperature range and at the 20 K∙min^-1 rates of heating. The specific surface area was tested and calculated by Micromeritics.

2.2. Silicon Dioxide Impregnation

The process of impregnation was described in Figure 1, and about 5 g of silicon dioxide was impregnated in the 100 cm³ impregnating solution (weight concentration range of 0.2–3%) for 12 h at 90°C. The impregnated silicon dioxide was collected by filtration, washed by hexane, and vacuum dried at 70, 90, 110, 130, 150, and 170°C to constant weight. The impregnating substance used was meglumine (ME), mannitol (MA), citric acid (CA), tartaric acid (TA), and salicylic acid (SA).

Figure 1

Schematic illustration of the process of silicon dioxide impregnation.

2.3. Static Adsorption

Due to the flammable and explosive property of chlorosilane, the boron chloride in chlorosilane was replaced by the boron chloride in hexane in the study. The tests were carried in the screw capped bottles with polyethylene gaskets filled with N₂, which prevents the boron chloride from water steam.

During static experiments, adsorbent (0.1, 0.2, or 0.5 g) was put into the 15 or 30 ml hexane solution of boron chloride at the concentration of 585 ppm at room temperature for 2 h.

In kinetic experiments, the boron concentration was measured by sampling at regular intervals until the concentration remained constant, which means the adsorption equilibrium reached.

In order to investigate the dependence on the initial boron concentration, the concentration of the solution varied between 234 ppm and 23400 ppm, while the 0.1 g of adsorbents was weighted into the 15 ml of the hexane solution of boron chloride.

The adsorption capacity ( $q$ ) of boron chloride is given as $\begin{matrix} (1) & q = \frac{(c_{0} - c) \times V}{m} \end{matrix}$ where $c_{0}$ and $c$ are the initial and residual boron chloride concentration in the solution, mg∙l^-1; $V$ is the volume of boron chloride solution, l; $m$ is the amount of adsorbents added in the solution, g.

Boron chloride in hexane was hydrolysed to form boric acid solution [26], then it was measured spectrophotometrically using azomethine-H, which will form yellow complex with boric acid after 6 h of waiting time, and then the absorbance value was tested at 415 nm, with a 1 cm cuvette against the blank test [27].

3. Results and Discussion

3.1. Static Adsorption Experiments of Adsorbent

As polyols and organic acids have typical oxygen–containing functional groups that definitely enhance the boron adsorption as reported [28, 29], the adsorption ability of boron chloride on impregnated silicon dioxide showed different increase with different impregnating agent in Figure 2. The result points to silicon dioxide impregnated with mannitol (SD + MA) and citric acid (SD + CA) as the best adsorbents of boron chloride in this study, which can be attributed to the structure and number of oxygen-containing functional groups. The structure of impregnating agent is shown in Figure 3, MA and ME are both polyols, and MA has more hydroxy than ME, resulting little better performance than ME. Compared with CA and TA, SA has less hydroxy and carboxy, leading to a lower adsorption capacity.

Figure 2

Comparison of boron adsorption capacity on SD, SD + ME, SD + MA, SD + CA, SD + TA, SD + SA, and D564.

Figure 3

Structure of impregnating agent (ME, MA, CA, TA, and SA).

3.2. The Effect of Impregnating Concentration

Figure 4 presents the adsorption capacity of boron chloride on impregnated silicon dioxide prepared with different impregnating solution concentration, and the adsorption capacity of boron chloride on impregnated silicon dioxide increased at early stage and then decreased with increasing impregnating solution concentration. The maximum adsorption capacity of SD + CA is 44.37 mg∙g^-1at 1.2% of impregnating solution, and the maximum adsorption capacity of SD + MA is 40.49 mg∙g^-1 at 1% of impregnating solution, which is attributed to the increasing loading of the oxygen-containing active sites on the surface of silicon dioxide with the increasing concentration of impregnating solution. However, with further increase in concentration, as shown in Table 1, some functional groups may agglomerate and block the limited loading sites of silicon dioxide, resulting in the decrease in the pore size, pore volume, and decrease in adsorption capacity of impregnated silicon dioxide. When the concentration of mannitol impregnation solution reached 3%, the pore channels may be blocked or even disappeared, the internal diffusion of boron chloride mainly control the adsorption, and boron chloride cannot enter the pore channels of impregnated silicon dioxide, which caused the adsorption capacity of SD + MA to drop sharply, even lower than that of blank silicon dioxide.

Figure 4

The effect of concentration of impregnating solution on boron adsorption capacity on SD + MA and SD + CA.

Table 1

The pore size of different adsorbents impregnated by solution of different concentration.

Adsorbents	Concentration of impregnating solution W%	Surface area (m²·g^-1)	Pore volume (cm³·g^-1)	Pore size (A)
SD	\	307	0.9994	130.2
SD + MA	1.0	234	0.9611	164.0
SD + MA	3.0	287	0.9209	128.2
SD + CA	1.0	287	0.9599	133.7
SD + CA	3.0	266	0.8376	126.1

3.3. The Effect of Drying Temperature

As can be shown in Figure 5, with the increase of the drying temperature, little change in adsorption capacity indicates that the drying temperature had little effect on the increase of active oxygen-containing group sites. But the increasing temperature may promote the bonding of mannitol to the hydroxyl groups on the surface of silicon dioxide.

Figure 5

The effect of drying temperature on boron adsorption capacity on SD + MA and SD + CA.

3.4. Characterization

Functional groups identified by FT-IR spectra in Figure 6 confirm the successful functionalization of mannitol and citric acid on the surface of silicon dioxide, and the FT-IR spectra of impregnated adsorbents show the increase in the band at 3400 cm^-1 (ν-OH) [30] and the appearance of new peak at 2904 cm^-1 and 2844 cm^-1 (ν-CH) [31], suggesting the successful grafting of polyol groups on the silicon dioxide supports. The appearance of a new peak at 1760 cm^-1 (ν-C=O) and 1410 cm^-1 (δ-C-OH) [30] and the wide and scattered peak at the range of 3200 cm^-1–2500 cm^-1 [30] suggest the loading of carboxylic acid on the silicon dioxide.

Figure 6

FT-IR spectra of SD, SD + MA, and SD + CA.

Active silicon dioxide and impregnated silicon dioxide were analyzed by X-ray energy spectrometry (EDS). It was found in Table 2 that the proportion of elemental Si on the surface of the impregnated silicon dioxide decreased, and the content of elemental C increased significantly compared to the activated silicon dioxide, which was attributed to the loading of polyhydroxy organic compounds on the surface of silicon dioxide.

Table 2

The relative percentage of C, O, and Si in adsorbents determined by EDS.

Adsorbents	Concentration of impregnating solution W%	C%	O%	Si%
SD + MA	1.0	15.75	61.25	22.99
SD + CA	1.0	13.35	56.26	30.39

Figure 7 shows the TGA curves of the studied adsorbents from ambient temperature to 950°C at heating rates of 20°C∙min^-1. In these curves, the mass loss of all adsorbents started at about 95°C and ended around 500°C. The mass residue was mainly SiO₂, the product of inorganic silicon combustion, and the thermal decomposition of analyzed polymers occurred in two or three well-defined stages.

Figure 7

TGA and DTG curves for (a) SD, (b) SD + MA, and (c) SD + CA.

(b)

(c)

Thus, in the case of silicon dioxide, the first stage began at about 40°C and ended at about 97.5°C, characterized by weak DTG peak at 74.8°C, which can be mainly attributed to the adsorbed water. This stage involved maximum 0.98% weight loss and can be considered nondegradable. The second stage, which began at around 360°C and ended at about 630°C, can be attributed to the decomposition of hydroxyl groups on the surface of silicon dioxide.

The SD + MA exhibits two distinct stages of decomposition. Thus, the first stage began at 69.6°C and ended at about 122°C, with DTG peak at about 103.1°C, which can be mainly attributed to the residual solvent, and this step involved 4.23% mass loss. The second stage started at around 311.7°C and ended at around 494.5°C with DTG peaking at 397.4°C, mainly due to the thermal decomposition of mannitol, which involved 11.25% mass loss.

The steady mass loss in SD + CA started at about 92.8°C, ended at 950°C, and was characterized by a DTG peak of 249.6°C, which was due to the mass loss of the citric acid on the silicon dioxide, and the weight loss rate was 15.17%. The difference between silicon dioxide and impregnated adsorbents shows that mannitol and citric acid are indeed loaded onto the surface of the silicon dioxide.

3.5. Thermal Degradation Kinetic of Adsorbents

In order to explain the thermal degradation of synthesized adsorbents, it is necessary to know the whole kinetic triplet [32, 33]. At each stage of the beginning and end of decomposition, kinetic analysis was performed using Freeman-Carroll method [34] to calculate the activating energy, pre-exponential factor, and reaction order from a single TGA curve based on the following equations: $\begin{matrix} (2) & \frac{d α}{d t} = A e^{- E / RT} {(1 - α)}^{n} \end{matrix}$ or $\begin{matrix} (3) & \frac{Δ In (d α / d t)}{Δ In (1 - α)} = - \frac{E}{R} \frac{Δ (1 / T)}{Δ In (1 - α)} + n \end{matrix}$ where $d α / d t$ is the rate of reaction, s^-1, $α$ is the conversion of reaction, $n$ is the order of reaction, $R$ is the gas constant, 8.31 J∙mol^-1∙K^-1, $T$ is the degradation temperature, K, $A$ is the pre-exponential factor, and $E$ is the apparent activation energy in J∙mol^-1.

The results of the thermal degradation kinetic were shown in Figures 8 and 9.

Figure 8

Thermal degradation kinetic curves of (a) SD + MA in 69.6–103.1°C and (b) SD + MA in 311.7–397.4°C.

(b)

Figure 9

Thermal degradation kinetic curves of SD + CA in 92.8–249.6°C.

The kinetic parameters corresponding to the degradation of impregnated adsorbents are presented in Table 3.

Table 3

Kinetic properties for adsorbents using Freeman-Carroll method.

Adsorbents	SD + MA			SD + AC
Parameter	$E$ (KJ∙mol^-1)	$n$	$R^{2}$	$E$ (KJ∙mol^-1)	$n$	$R^{2}$
Stage I	21.14	1.543	0.9691	12.17	0.5974	0.9602
Stage II	15.74	0.6244	0.9694

As can be seen from the data in Table 3 that the activation energy and the order of reaction are not always a constant value, proving that the thermal degradation of mannitol-modified silica is a complex reaction.

3.6. Adsorption Kinetics Experiments of Adsorbents

Adsorption equilibrium curves described in Figure 10 show that the adsorption rate of boron chloride in solution is fast. The adsorption capacity increased rapidly with the increase of adsorption time in the first 10 minutes, after which, it slowed down as the adsorption equilibrium was reached. This is a typical development rule in adsorption process. As the adsorption progresses, the diffusion rate of solute to adsorbent slows due to the decrease of adsorption concentration [35]. The experimental results obtained were used to study the rate-limiting step.

Figure 10

Effect of reaction time on removal of boron chloride on SD, SD + MA, and SD + CA.

The adsorption kinetics curves were analyzed using the pseudo-first-order kinetic equation as follows [30]: $\begin{matrix} (4) & \ln (q_{e} - q_{t}) = \ln q_{e} - k_{1} t \end{matrix}$

The pseudo-second-order adsorption kinetic equation can be expressed as follows [36]: $\begin{matrix} (5) & \frac{t}{q_{t}} = \frac{1}{k_{2} q_{e}^{2}} + \frac{t}{q_{e}} \end{matrix}$ where $q_{e}$ and $q_{t}$ are the adsorption capacity, mg∙g^-1, at equilibrium and at time $t$ , respectively, $t$ is the adsorption time, min, $k_{1}$ is the pseudo-first-order adsorption rate constant, min^-1, and $k_{2}$ is the pseudosecondary adsorption rate constant, g∙mg^-1∙min^-1.

The correlation coefficients and the rate constants for the pseudo-first-order kinetic model and pseudo-second-order kinetic model are shown in Figure 11, Figure 12, and Table 4, respectively.

Figure 11

The pseudo-first-order kinetic model fitting of adsorption kinetics curves of SD, SD + MA, and SD + CA.

Figure 12

The pseudo-second-order kinetic model fitting of adsorption kinetics curves of SD, SD + MA, and SD + CA.

Table 4

Kinetic parameters of adsorption of boron chloride on SD, SD + MA, and SD + CA.

Adsorbents	$t_{e}$ (min)	$q_{e}$ (exp) (mg·g^-1)	Pseudo-first-order kinetic model				Pseudo-second-order kinetic model
Adsorbents	$t_{e}$ (min)	$q_{e}$ (exp) (mg·g^-1)	$q_{e}$ (fit) (mg·g^-1)	$k_{1}$ (min^-1)	$χ$	$R^{2}$	$q_{e}$ (fit) (mg·g^-1)	$k_{2}$ (g.mg^-1 min^-1)	$χ$	$R^{2}$
SD	40	33.42	22.59	0.1512	0.4794	0.8279	33.81	$1.725 \times 10^{- 2}$	$1.153 \times 10^{- 2}$	0.9998
SD + MA	10	34.02	33.52	0.2939	0.01493	0.9417	34.00	$1.817 \times 10^{- 2}$	$5.882 \times 10^{- 4}$	0.9987
SD + CA	10	34.38	32.28	0.3892	0.06506	0.9848	34.26	$4.691 \times 10^{- 2}$	$3.503 \times 10^{- 3}$	0.9996

Note: $q_{e} (\exp)$ and $q_{e} (fit)$ are the equilibrium experiment results and calculated value according the model, respectively.

$R^{2}$ and $χ$ can be used to assess how well the kinetic model coincides with the experiment data, $χ = |q_{e} (\exp) - q_{e} (fit)| / q_{e} (fit)$ ; if the experimental results are similar to the data from the model, $χ$ will be small else it will be large [29]. Larger $R^{2}$ and smaller $χ$ obtained from the pseudo-second-order kinetic model indicate that the adsorption of boron chloride on silicon dioxide can be described by the second-order kinetic reaction model better. It can be concluded that the adsorption rate is mainly controlled by chemisorption.

The control steps of the speed of adsorption process are usually investigated using the diffusion model in the particle [37], and the formula is as follows: $\begin{matrix} (6) & q_{t} = k_{n} t^{1 / 2} + C \end{matrix}$ where $q_{t}$ is the adsorption capacity of adsorbents on boron chloride at a certain time, mg∙g^-1, $t$ is the adsorption time, min, $k_{n}$ is the diffusion rate constant in the particle, and $C$ is a constant.

Typically, the particle diffusion model is three-stage linear, including boundary layer diffusion, internal diffusion, and adsorption site adsorption. Because the first stage reaction speed is very fast, it will not be the control step of adsorption reaction, so the adsorption data were fitted in two stages and shown in Figure 13.

Figure 13

Multisegment fitting curve of the intraparticle diffusion model of adsorption of boron chloride on SD, SD + MA, and SD + CA.

As can be seen from the Table 5, the $R^{2}$ of two stages is greater than 0.9, and the fit lines do not pass through the point of origin, indicating that the adsorption process is a complex process mixed with physical adsorption and chemical adsorption. The diffusion coefficient $k_{n}$ of section II is smaller than that of section I, which proves that the adsorption of boron chloride is controlled by the adsorption on the active adsorption site.

Table 5

Diffusion parameters in particles of adsorption of boron chloride on SD, SD + MA, and SD + CA.

Adsorbents	Section I			Section II
Adsorbents	$k_{n}$	$C$	$R^{2}$	$k_{n}$	$C$	$R^{2}$
SD	9.082	1.485	0.9560	1.307	25.17	0.9687
SD + MA	10.55	0.7507	0.9545	0.4509	29.77	0.9714
SD + CA	11.01	2.484	0.9479	0.9822	28.65	0.9549

3.7. Determination of Adsorption Isotherm

The distribution of boron between the liquid phase and the solid phase is an index of equilibrium position during the adsorption.

Adsorption isotherm was shown in Figure 14, silicon oxide adsorption of boron chloride can be roughly divided into several stages at concentrations of boron chloride ranging from 234 ppm to 2340 ppm, the adsorption capacity increased with the increase of the concentration of boron chloride solution, the increasing trend slowed down within 2340 ppm to 9360 ppm, and the adsorption capacity basically remained unchanged within 9360 ppm to 23400 ppm.

Figure 14

Adsorption isotherms of removal of boron chloride on SD, SD + MA, and SD + CA.

The adsorption isotherms were analyzed by two-parameter models: Langmuir, Freundlich, Temkin, and Dubinin-Radushkevich and three-parameter models: Redlich-Peterson, Sips, Toth, etc. [38].

Langmuir equation assumes that the maximum adsorption occurs when the surface is covered by a monolayer of adsorbate, which is effective for monolayer sorption onto a surface with a finite number of identical sites [36], as shown below: $\begin{matrix} (7) & q_{e} = \frac{k_{l} q_{l} c_{e}}{1 + k_{l} c_{e}}, \end{matrix}$ where $q_{e}$ is the adsorption capacity at adsorption equilibrium, mg∙g^-1, $c_{e}$ is the concentration of boron chloride at adsorption equilibrium, mg∙l^-1, $q_{l}$ is the maximum theoretical monolayer adsorption capacity, mg∙g^-1, and $k_{l}$ is the adsorption equilibrium constant, which is related to the enthalpy change of adsorption.

Freundlich equation is a purely empirical model, which is indicative of the surface heterogeneity of the sorbent, and is given by the following equation: $\begin{matrix} (8) & q_{e} = k_{f} c_{e}^{1 / n_{f}} \end{matrix}$ where $k_{f}$ is the Freundlich constant, $1 / n_{f}$ is the heterogeneous coefficient, it is generally considered that substances with $1 / n_{f}$ between 0.1 and 0.5 are easy to be adsorbed, and when $1 / n_{f} > 2$ , they are difficult to be adsorbed.

Temkin equation: $\begin{matrix} (9) & q_{e} = B \ln c_{e} + A \end{matrix}$ where $B$ is Temkin’s isothermal energy constant, and $A$ is Temkin’s isothermal energy constant.

Dubinin-Radushkevich isotherm model is used to describe sorption on both homogeneous and heterogeneous surfaces [39]. It can be expressed as follows: $\begin{matrix} (10) & q_{e} = q_{D} \exp \{- A_{D} {[In (1 + \frac{1}{c_{e}})]}^{2}\} \end{matrix}$ where $q_{D}$ is the adsorption capacity of adsorbent, mg∙g^-1, and $A_{D}$ is related to the free energy of adsorption.

The linear correlation coefficient shown in Table 6 indicates that the adsorption process is most consistent with the Langmuir equation, indicating that monolayer adsorption occurs on the surface of adsorbents. The $1 / n_{f}$ in the Freundlich equation is less than 0.5, suggesting the favorable adsorption process [40].

Table 6

Two parameters adsorption isotherm parameters of boron chloride on SD, SD + MA, and SD + CA.

Adsorbents	$q_{e} (\exp)$ (mg·g^-1)	Isotherms and parameters
Adsorbents	$q_{e} (\exp)$ (mg·g^-1)	Langmuir	Freundlich	Temkin	Dubinin-Radushkevich
SD	249.6	$q_{l} = 250.0$	$K_{f} = 82.42$	$B = 30.78$	$q_{D} = 226.0$
		$k_{l} = 1.471 \times 10^{- 3}$	$1 / n_{f} = 0.1152$	$A = - 43.00$	$A_{D} = 2.843 \times 10^{4}$
		$Δ q_{t} = 0.4000$	$Δ q_{t} = 10.46$	$Δ q_{t} = 14.95$	$Δ q_{t} = 23.54$
		$χ = 1.600 \times 10^{- 3}$	$χ = 4.021 \times 10^{- 2}$	$χ = 5.652 \times 10^{- 2}$	$χ = 0.1041$
		$R^{2} = 0.9898$	$R^{2} = 0.8573$	$R^{2} = 0.8429$	$R^{2} = 0.9088$

SD + MA	275.8	$q_{l} = 277.8$	$K_{f} = 103.3$	$B = 30.42$	$q_{D} = 250.8$
		$k_{l} = 2.091 \times 10^{- 3}$	$1 / n_{f} = 0.1024$	$A = - 12.29$	$A_{D} = 1.012 \times 10^{4}$
		$Δ q_{t} = 2.000$	$Δ q_{t} = 10.00$	$Δ q_{t} = 15.47$	$Δ q_{t} = 25.00$
		$χ = 7.199 \times 10^{- 3}$	$χ = 3.503 \times 10^{- 2}$	$χ = 5.313 \times 10^{- 2}$	$χ = 9.968 \times 10^{- 2}$
		$R^{2} = 0.9977$	$R^{2} = 0.8619$	$R^{2} = 0.8419$	$R^{2} = 0.8863$

SD + CA	332.6	$q_{l} = 312.5$	$K_{f} = 43.65$	$B = 45.32$	$q_{D} = 276.7$
		$k_{l} = 4.555 \times 10^{- 3}$	$1 / n_{f} = 0.2176$	$A = - 100.5$	$A_{D} = 6.181 \times 10^{3}$
		$Δ q_{t} = 20.05$	$Δ q_{t} = 50.96$	$Δ q_{t} = 18.73$	$Δ q_{t} = 55.91$
		$χ = 6.416 \times 10^{- 2}$	$χ = 0.1329$	$χ = 5.327 \times 10^{- 2}$	$χ = 0.2021$
		$R^{2} = 0.9981$	$R^{2} = 0.8492$	$R^{2} = 0.9262$	$R^{2} = 0.8761$

The adsorption isotherms were analyzed by three-parameter models as follows:

Redlich-Peterson equation: $\begin{matrix} (11) & q_{e} = \frac{A_{RP} C_{e}}{1 + B_{RP} c_{e}^{g}} \end{matrix}$ where $A_{RP} and B_{RP}$ are the Redlich-Peterson constants;

Sips model: $\begin{matrix} (12) & q_{e} = \frac{q_{s} {(K_{s} C_{e})}^{1 / n_{s}}}{1 + {(K_{s} C_{e})}^{1 / n_{s}}} \end{matrix}$ where $K_{s}$ is the adsorption equilibrium constant, and $q_{s}$ is the adsorption capacity of adsorbent, mg∙g^-1;

Toth equation: $\begin{matrix} (13) & q_{e} = \frac{q_{T} K_{t} c_{e}}{{[1 + {(K_{t} c_{e})}^{t}]}^{1 / t}} \end{matrix}$ where $K_{t}$ is the adsorption equilibrium constant, $q_{T}$ is the adsorption capacity at equilibrium, mg∙g^-1, $t = 1$ , which proves that the adsorption process occurs on the monolayer adsorption surface.

The $g$ of the Redlich-Peterson, $n_{s}$ of Sips, and $t$ of Toth models obtained in Table 7 are not equal to 1, indicating that the adsorption process occurs on the heterogeneous surface of multilayer adsorption, which is inconsistent with the analysis results of Langmuir isothermal adsorption model. The linear correlation coefficients are all less than 0.9, indicating that the three-parameter adsorption isothermal model is less suitable than the two-parameter model for the adsorption process of boron chloride on silicon oxide.

Table 7

Three parameters adsorption isotherm parameters of boron chloride on SD, SD + MA, and SD + CA.

Adsorbents		Redlich-Peterson	Sips	Toth
SD	$q_{e} (\exp) = 249.5$	$A_{RP} = 0.6920$	$q_{s} = 223.4$	$q_{T} = 232.6$
		$B_{RP} = 2.261 \times 10^{- 3}$	$K_{s} = 5.149 \times 10^{- 3}$	$K_{t} = 2.272 \times 10^{- 3}$
		$g = 1.024$	$n_{s} = 0.2572$	$t = 1.707$
		$Δ q_{t} = 12.42$	$Δ q_{t} = 26.11$	$Δ q_{t} = 16.99$
		$χ = 0.05218$	$χ = 0.1169$	$χ = 0.07308$
		$R^{2} = 0.8703$	$R^{2} = 0.9116$	$R^{2} = 0.8777$

SD + MA	$q_{e} (\exp) = 275.8$	$A_{RP} = 1.417$	$q_{s} = 249.2$	$q_{T} = 247.5$
		$B_{RP} = 5.573 \times 10^{- 3}$	$K_{s} = 8.713 \times 10^{- 3}$	$K_{t} = 3.593 \times 10^{- 3}$
		$g = 0.9944$	$n_{s} = 0.2165$	$t = 134.0$
		$Δ q_{t} = 8.062$	$Δ q_{t} = 26.62$	$Δ q_{t} = 28.32$
		$χ = 0.03011$	$χ = 0.1068$	$χ = 0.1144$
		$R^{2} = 0.8555$	$R^{2} = 0.8969$	$R^{2} = 0.8632$

SD + CA	$q_{e} (\exp) =$ 332.6	$A_{R P} = 1.879$	$q_{s} = 321.0$	$q_{T} = 326.4$
		$B_{RP} = 8.993 \times 10^{- 2} 3$	$K_{s} = 4.211 \times 10^{- 3}$	$K_{t} = 7.041 \times 10^{- 3}$
		$g = 0.9565$	$n_{s} = 1.197$	$t = 0.7387$
		$Δ q_{t} = 14.24$	$Δ q_{t} = 11.60$	$Δ q_{t} = 6.183$
		$χ = 0.04514$	$χ = 0.03613$	$χ = 0.01892$
		$R^{2} = 0.9611$	$R^{2} = 0.9575$	$R^{2} = 0.9596$

3.8. Chemisorption Mechanism

After completing adsorption, FT-IR spectra of adsorbents in Figure 15 show the appearance of new peak at 1400 cm^-1 (as-B-O) [41], suggesting the chemisorption of boron chloride on the adsorbents.

Figure 15

FT-IR spectra of SD, SD + MA after adsorption, and SD + CA after adsorption.

Due to the significant electron deficiency of the boron, it can form typical covalent coordination bonds with electron donor, and the following mechanism of nucleophilic addition reactions is proposed in Figure 16 for the chemisorption of boron chloride on modified adsorbents.

Figure 16

The mechanism of chemisorption of boron chloride on adsorbents.

The lone pair of electrons of polyhydric alcohols was coordinated to the electron deficient boron, and after accepting electrons, the Cl^- left with a pair of electrons, due to the strong polarity of the B-Cl bond, followed by the leaving of proton on the methanol. The total reaction is $3 ROH + BC l_{3} = = B {(OR)}_{3} + 3 HCl$ , similar to the hydrolysis of BCl₃ [42].

4. Conclusions

In this study, the silicon dioxide modified by oxygen–containing functional groups (–OH and –COO–) shows a better performance of the boron chloride removal in polysilicon production than that of chelating resign under the test conditions. The adsorption performance and adsorption mechanism of boron chloride are firstly discussed and described in detail, which are the relative lack in the previous industrial methods.

This study has given encouraging results, and organic-inorganic hybrid materials could be applied to reduce boron chloride. We wish to carry out the adsorption tests in the condition of flammable and explosive chlorosilane, which is closer to the industrial production, and this part has been left for future research.

Footnotes

Data Availability

The data used to support the findings of this study are available from the corresponding author upon request.

Conflicts of Interest

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Acknowledgments

Our research was supported by Tianjin University and Zhejiang Shaoxing Institute of Tianjin University.

References

Berberi

Thodhorjani

Hoxha

Muda

Photovoltaics: between a bright outlook and uncertainty

Energy Science & Engineering 2013 1

7581943

2 72 80

10.1002/ese3.10

2-s2.0-85020151724

Allouhi

Rehman

Buker

M. S.

Said

Up-to-date literature review on solar PV systems: technology progress market status and R&D

Journal of Cleaner Production 2022 362, article 132339

7581943

10.1016/j.jclepro.2022.132339

Chen

Morita

Chen

Wang

Boron removal for solar-grade silicon production by metallurgical route: a review

Solar Energy Materials & Solar Cells 2019 203, article 110169

7581943

10.1016/j.solmat.2019.110169

2-s2.0-85071984699

Yasuda

Morita

Okabe

T. H.

Processes for production of solar-grade silicon using hydrogen reduction and/or thermal decomposition

Energy Technol (Weinh) 2014 2

7581943

2 141 154

10.1002/ente.201300131

2-s2.0-84904809389

China National Standards

Solar-grade polycrystalline silicon

2017

China National Standardization Management Committee

GB/T 25074-2017, 2017-11-01

China National Standards

Electronic-grade polycrystalline silicon

2014

China National Standardization Management Committee

GB/T 12963-2014, 2014-12-31

Qian

Huang

Progress in the technology of high efficient removal of trace boron and phosphorus from trichlorosilane

Chemical Industry and Engineering 2018 35

7581943

2 42 48

10.13353/j.issn.1004.9533.20161071

Zhu

Research development in the technology of boron removal from trichlorosilane

Materials Review 2013 27

7581943

5A 29 32

10.3969/j.issn.1005-023X.2013.09.007

Wang

Processing crude trichlorosilane, by performing gas stripping treatment of crude trichlorosilane, and subjecting dichloro-dihydro-containing silicon and mixture of trichlorosilane and silicon tetrachloride to first purification treatment

2015 2015-139710, CN104261412-A; CN104261412-B

10.

Chen

Lin

Zhang

Method of removing impurities in trichlorosilane mixed gas, involves spraying trichlorosilane mixed gas in gas-steam mixture wetting tower using silicon tetrachloride liquid and adsorbing silicon dust by silicon tetrachloride liquid

2010 2011-B02394, CN101913609-A

11.

Cao

Mao

Jiang

Sun

Jin

Method of removing phosphorus-boron impurities from trichlorosilane, involves contacting trichlorosilane gas containing phosphorus-boron impurities with steam and inert gas, performing partial hydrolysis, distilling and purifying

2010 2011-B08030, CN101920961-A; CN101920961-B

12.

Guo

Tang

Zhao

Removing boron and phosphorus during polysilicon production by modified Siemens process, comprises e.g. reacting silicon powder and hydrogen chloride to prepare synthetic trichlorosilane gas and reacting it with wet gas

2010 2011-B08030, CN101759186-A; CN101759186-B

13.

Mueh

E. H. R.

Installation and method for reducing the content in elements, such as boron, of halosilanes

US Patent 2011052474(A1), 2011

14.

Winter

W. E.

Diaz

Purification of halogenated silicon compounds

US Patent 3069239(A), 1962

15.

Chen

Zhang

Method of absorbing parts per billion level boron and phosphorus impurities in trichlorosilane, comprises taking liquid, rectifying and purifying trichlorosilane, entering into each vaporizer, heating, passing into fixed bed and adsorbing

2016 2016-14255Y, CN105329902-A; CN105329902-B

16.

Miyao

Ishida

Nezu

Purification of chlorosilane for silicon crystal growth involves circulating trichlorosilane which contains phosphorus trichloride and trichlorinated boron as impurity to anion exchange resin and disproportionating trichlorinated boron

2015 2015-70904R, JP2015202991-A; JP6184898-B2

17.

Wang

Yao

Zhang

Bai

Peng

Purifying trichlorosilane involves cooling trichlorosilane, and then adsorbing cooled trichlorosilane in adsorption tower for adsorption of boron and phosphorus to get purified trichlorosilane

2015 2015-591862, CN104828827-A; CN104828827-B

18.

Yang

Luo

Shi

Min

Boron-removing device in manufacturing trichlorosilane comprises a first and a second boron-removing tower, which are connected in series and communicated via rectifying tower and low styrene content resin filler layers in two towers

2011 2011-J90255, CN201890773-U

19.

Wang

Zhang

Synthesis of a novel boron-specific chelating resin and its adsorption to boron from salt lake brine

The Chinese Journal of Process Engineering 2004 4

7581943

6 502 507

10.3321/j.issn:1009-606X.2004.06.004

20.

Oladipo

A. A.

Gazi

High boron removal by functionalized magnesium ferrite nanopowders

Environmental Chemistry Letters 2016 14

7581943

3 373 379

10.1007/s10311-016-0554-6

2-s2.0-84962648930

21.

Oladipo

A. A.

Gazi

Hydroxyl-enhanced magnetic chitosan microbeads for boron adsorption: parameter optimization and selectivity in saline water

Reactive and Functional Polymers 2016 109

7581943

23 32

10.1016/j.reactfunctpolym.2016.09.005

2-s2.0-84989332740

22.

Oladipo

A. A.

Gazi

Efficient boron abstraction using honeycomb-like porous magnetic hybrids: assessment of techno-economic recovery of boric acid

Journal of Environmental Management 2016 183

7581943

917 924

10.1016/j.jenvman.2016.09.059

2-s2.0-84992123351

23.

Zhuravlev

L. T.

Concentration of hydroxyl groups on the surface of amorphous silicas

Langmuir 1987 3

7581943

3 316 318

10.1021/la00075a004

2-s2.0-0022722235

24.

Sugimura

Matsuoto

Kobayashi

Purification of chlorosilane e.g. dichlorosilane used as raw material for preparing polycrystalline silicon, involves contacting crude chlorosilane containing boron compound with ion-exchange resin and silica adsorbent

y 2011-C09644, WO2011024276-A1

25.

Sears

G. W.

Determination of specific surface area of colloidal silica by titration with sodium hydroxide

Analytical Chemistry 1956 28

7581943

12 1981 1983

10.1021/ac60120a048

2-s2.0-27644560025

26.

Chen

J. S.

Lin

H. M.

Yang

M. H.

Spectrophotometric determination of trace and ultratrace levels of boron in silicon and chlorosilane samples

Fresenius' Journal of Analytical Chemistry 1991 340

7581943

6 357 362

10.1007/BF00321582

2-s2.0-0000929511

27.

Joanna

Jerzy

Jolanta

Maria

Marian

Piotr

Removal of boron dissolved in water

Environmental Progress & Sustainable Energy 2007 26

7581943

1 71 77

10.1002/ep.10180

2-s2.0-34248210672

28.

Çelik

Z. C.

Can

B. Z.

Kocakerim

M. M.

Boron removal from aqueous solutions by activated carbon impregnated with salicylic acid

Journal of Hazardous Materials 2008 152

7581943

1 415 422

10.1016/j.jhazmat.2007.06.120

2-s2.0-39749196188

17692458

29.

Oladipo

A. A.

Ahaka

E. O.

Gazi

High adsorptive potential of calcined magnetic biochar derived from banana peels for Cu2+, Hg2+, and Zn2+ ions removal in single and ternary systems

Environmental Science and Pollution Research International 2019 26

7581943

31 31887 31899

10.1007/s11356-019-06321-5

30.

Liu

Kim

Qing

Guo

Lee

Boron adsorption using a new boron-selective hybrid gel and the commercial resin D564

Colloids and Surfaces A: Physicochemical and Engineering Aspects 2009 341

7581943

1-3 118 126

10.1016/j.colsurfa.2009.03.045

2-s2.0-67349256931

31.

Wang

Zhang

Novel organic-inorganic hybrid mesoporous materials for boron adsorption

Colloids and Surfaces A: Physicochemical and Engineering Aspects 2006 275

7581943

1-3 73 78

10.1016/j.colsurfa.2005.06.075

2-s2.0-31544466496

32.

Sava

Burescu

Lisa

Study of thermal behavior of polyimides containing pendent-substituted azobenzene units

Polymer Bulletin (Berlin) 2014 71

7581943

6 1359 1373

10.1007/s00289-014-1128-z

2-s2.0-84901243265

33.

Liu

X. Y.

W. D.

Evaluating the thermal stability of high performance fibers by TGA

Journal of Applied Polymer Science 2006 99

7581943

3 937 944

10.1002/app.22305

2-s2.0-33645136942

34.

Freeman

E. S.

Carroll

The application of thermoanalytical techniques to reaction kinetics: the thermogravimetric evaluation of the kinetics of the decomposition of calcium oxalate monohydrate

The Journal of Physical Chemistry 1958 62

7581943

4 394 397

10.1021/j150562a003

2-s2.0-0008047480

35.

Bouguerra

Mnif

Hamrouni

Dhahbi

Boron removal by adsorption onto activated alumina and by reverse osmosis

Desalination 2008 223

7581943

1-3 31 37

10.1016/j.desal.2007.01.193

2-s2.0-38749137770

36.

Bouguerra

Marzouk

Hamrouni

Equilibrium and kinetic studies of adsorption of boron on activated alumina

Water environment research : a research publication of the Water Environment Federation 2009 81

7581943

12 2455 2459

10.1002/j.1554-7531.2009.tb00254.x

20099630

37.

Russo

Rossano

Salucci

Tesser

Salmi

Di Serio

Intraparticle diffusion model to determine the intrinsic kinetics of ethyl levulinate synthesis promoted by Amberlyst-15

Chemical Engineering Science 2020 228

7581943

115974

10.1016/j.ces.2020.115974

38.

Wei

Shuqin

Haorong

Gaowa

Investigation of phosphate removal from aqueous solution by both coal gangues

Water Science and Technology: a Journal of the International Association on Water Pollution Research 2017 76

7581943

4 785 792

10.2166/wst.2017.241

2-s2.0-85027724911

28799925

39.

Kaftan

Ö.

Açıkel

Eroğlu

A. E.

Shahwan

Artok

Synthesis, characterization and application of a novel sorbent, glucamine- modified MCM-41, for the removal/preconcentration of boron from waters

Analytica Chimica Acta 2005 547

7581943

1 31 41

10.1016/j.aca.2005.03.043

2-s2.0-23044515840

40.

Demey

Barron-Zambrano

Mhadhbi

Miloudi

Yang

Ruiz

Sastre

A. M.

Boron removal from aqueous solutions by using a novel alginate-based sorbent: comparison with Al2O3 particles

Polymers (Basel) 2019 11

7581943

9 1509

10.3390/polym11091509

2-s2.0-85072519072

31527526

41.

Zhang

Chen

Zheng

Zhao

Synthesis of poly ethylene glycol boronic ester

Fine Chemicals 2009 26

7581943

7 652 655

42.

Delue

N. R.

Crano

J. C.

Boron tri:chloride prodn. by chlorinating borate ester - in presence of free radical initiator, reducing phosgene formation (NL 30.9.80)

1980 1980-71869C, DE3011246-A

Silicon Dioxide Functionalized with Oxygen–Containing Groups for Enhanced BCl 3 Adsorption

Abstract

1. Introduction

2. Materials and Methods

2.1. Reagent and Instrument

2.2. Silicon Dioxide Impregnation

2.3. Static Adsorption

3. Results and Discussion

3.1. Static Adsorption Experiments of Adsorbent

3.2. The Effect of Impregnating Concentration

3.3. The Effect of Drying Temperature

3.4. Characterization

3.5. Thermal Degradation Kinetic of Adsorbents

3.6. Adsorption Kinetics Experiments of Adsorbents

3.7. Determination of Adsorption Isotherm

3.8. Chemisorption Mechanism

4. Conclusions

Footnotes

Data Availability

Conflicts of Interest

Acknowledgments

References