Cellulose-based catalyst design for efficient chlorate reduction

1 Introduction

During water disinfection processes or chloralkali electrolysis, the formation of toxic species such as chlorate ( ${\mathit{ClO}}_3^{-}$ ) occurs. The reduction of such undesirable species is important from both an environmental and an industrial point of view (Brinkmann et al., 2014; Jakab-Nácsa et al., 2020; Ren et al., 2020; WHO, 2005). Due to a side reaction during chloralkali electrolysis, oxygen is formed at the anode and chlorate at the anolyte. Thus, brine is produced which is contaminated with chlorate, but recycled in the process. However, the chlorate contamination will reduce the quality of caustic soda and could damage the equipment which will further decrease the efficiency of the procedure (Brinkmann et al., 2014). It is important to implement a prevention-, reduction-, and/or post-treatment into the process which could effectively reduce the amount of the above mentioned by-product (Brinkmann et al., 2014; Maslamani et al., 2020). In chloralkali plants, the commonly used method is thermal hydrochloric acid treatment, which is energy- and chemical-intensive (about 90 °C, pH 0, conversion 60–90%) (Ren et al., 2020). In contrast to this, by applying a suitable catalyst, less chemicals and energy is necessary and higher conversion could be achieved during a heterogeneous catalytic hydrogenation and the hydrogen which is formed during the electrolysis can also be used (Brinkmann et al., 2014; Ren et al., 2020).

Recently carbon supported platinum group metal (Pd, Pt, Ru, Ir, Rh) catalysts were tested in the catalytic hydrogenation of chlorate with promising results (Chen et al., 2017; Kuznetsova et al., 2012; Rutger et al., 2001). However, such powder-based catalysts have a disadvantage, as their separation from the reaction medium is difficult. It is preferable for industrial applications, especially in flow catalytic processes to use macroscopic structured materials (Akhtar et al., 2014) such as zeolites (Yilmaz and Müller, 2009) and monoliths (Hosseini et al., 2020), because they are easy to handle.

Due to environmental issues, there has been an increasing focus on renewable and degradable raw carbon based materials (Nasrollahzadeh et al., 2020). Cellulose is a good example of such materials which is widely available, inexpensive, and nontoxic and thus, its potential application in the field of heterogeneous catalysis is intensively researched (Xie et al., 2019). Cellulose-supported catalysts have been tested with great success in case of various chemical transformations including the Suzuki reaction (Dong et al., 2020), cycloaddition (Bahsis et al., 2018), degradation of methyl orange (Tavker et al., 2020), nitrobenzene (Li et al., 2020), p-nitrophenol (Lin et al., 2011; Wu et al., 2013) and methylene blue reduction (Wu et al., 2013).

Several efficient cellulose supported noble metal catalysts have been developed (Dong et al., 2020; Li et al., 2013). Gold nanoparticle-cysteamine/carboxymethyl cellulose (Tan et al., 2010) and cellulose nanocrystal-supported gold nanoparticles (Wu et al., 2014) have also been created before. Furthermore, controlled shape Pt nanoparticles with high catalytic activity have been prepared by using wood nanomaterials (Lin et al., 2011). Palladium nanoparticles on modified cellulose support have been utilized as catalyst in the hydrodeoxygenation of vanillin (Li et al., 2018). Flow-through catalytic experiments have also been performed, by using Au containing (Xie et al., 2020, 2019) or hybrid CNF/TiO₂ (Lucchini et al., 2018) cellulose monoliths. These systems are not only efficient catalysts, but their production can be carried out in a green and environmentally friendly way (Wu et al., 2014; Li et al., 2013; Quignard and Choplin, 2001).

In this research, cellulose beads (CB) have been used as support material in the design of palladium and platinum catalysts for catalytic hydrogenation of chlorate. The catalysts have been tested in both batch and continuous flow reactors. The effect of Fe₂O₃ as catalyst promoter has also been studied, because previously it was shown that the (per)chlorate content of brine can be reduced by using iron combining it with UV light (Gurol and Kim, 2003).

2 Experimental

2.4

2.4 Hydrogenation tests

2.4.1

2.4.1 Hydrogenation in batch reactor

The measurements were performed in a side-inlet gas washing bottle with fritted disc which was placed in a Julabo circulator. The hydrogenation of aqueous solution of potassium chlorate (Eq. (1)) (100 ml, 200 mg/dm³) was carried out at 80 °C, with 40 sccm nitrogen and 100 sccm hydrogen. In each case 1 g catalyst was used. The experiments lasted for 3 h and sampling was carried out after 0, 5, 15, 30, 45, 60, 90, 120, 150, and 180 min. For each sample (1 ml) 0.1 g potassium iodide and 1 ml hydrochloric acid were added and then, it was diluted to 50 ml with distilled water to apply redox reaction (Eq. (2)).

(1)

$${\mathit{ClO}}_3^{-}+{3H}_2\to {\mathit{Cl}}^{-}+3H_{2}O$$

(2)

$${\mathit{KClO}}_3+6KI+6HCl\to {3H}_{2}O+{3I}_2+7KCl$$

Thus, iodine formed in proportion to the chlorate concentration and appears with yellow colour in the solution. This colour change was followed with UV-6300PC spectrophotometer at 351 nm wavelength. The spectrophotometric method was calibrated by using potassium chlorate solutions with different concentrations (0, 50, 100, 150 and 200 mg/dm³).

2.4.2

2.4.2 Hydrogenation in a continuous flow system

The aqueous solution of potassium chlorate (200 mg/dm³) was flowed through the reactor (Fig. 1.) controlled by a peristaltic pump (0.5 ml/min). The reaction was performed at 80 °C, and the hydrogen flow rate was 100 ml/min, while 6 g catalyst was used. The hydrogenated solution is transferred by the pump to a branch where aqueous solution of potassium iodide (28 g/dm³) and sulfuric acid (4.5 mol/dm³) were added. As in the batch hydrogenation, iodine was formed in proportion to the chlorate concentration (Eq. (2)). The iodine containing solution was transmitted to the spectrophotometric measuring cell and the colour change was used to follow the concentration of chlorate.

Close

Fig. 1

(A) Schematic representation of the continuous chlorate hydrogenation set-up: 1) chlorate solution; 2) solution outlet; 3) H₂ flow inlet; 4) gas outlet; 5) catalyst; 6) KI solution; 7) H₂SO₄ solution; 8) thermostatic mixing loop; 9) spectrophotometric measuring cell; 10) waste collecting dish; 11) heating flow outlet; 12) heating flow inlet; (B) spectrophotometric measuring system.

Export to PPT

In order to monitor the chlorate concentration continuously, a special spectrophotometric cell was constructed (Fig. 1) which consisting of a flowthrough cell, a light source and a light sensor.

The cell is a square cross-section cuvette with input and output openings. Although the absorption maximum of I₃⁻ ion in aqueous media is about 353 nm, the molar absorption coefficient is high enough at the emission maximum of a UV LED (∼380 nm) to be able to measure and follow the changes of the chlorate concentration without applying a dedicated monochromator unit. Therefore, a UV LED and a phototransistor were used as a light source and a detector, respectively. The light source and the detector were placed on the opposite side of the cell at the same height. In this way a rather compact device has been developed with a small sample volume (∼2 cm³) which led to a short dead time necessary to detect any change in the concentration. The applied LED was operated with a current generator to achieve stable emission intensity. The collector electrode of the phototransistor was connected to a power supply (+9V), while the base electrode was not connected to any voltage, so no bias base-voltage was applied. However, the light sensitivity decreased, and therefore no electric signal could be measured at daylight illumination at the connecting point, the loading resistor, and the emitter electrode of the transistor. Despite the reduced sensitivity, the emission of the LED produced a sufficient output signal to carry out measurements in the 1–250 mg/dm³ concentration range.

The output signal was led to an AD converter (ADC Protoboard, manufactured by Mikroelectronica Ltd.), which was connected to a Raspberry Pi B card (credit-card-size ARM-based computer, manufactured by Raspberry Pi Foundation). As an operating system, a special Linux distribution, Raspbian was used. A special measuring and monitoring program was written to carry out the necessary tasks, such as the calibration and recalibration of the system, continuous monitoring of the chlorate concentration and recording the data.

3 Results and discussion

Four different palladium and platinum containing, cellulose bead (CB) supported catalysts, Pd-Pt/CB-Fe₂O₃, Pt-Pd/CB-Fe₂O₃, Pd-Pt/CB, and Pt-Pd/CB have been successfully prepared. Iron oxide have been used as a promoter material in two cases, and the specific surface area of these samples is about two times larger (Pd-Pt/CB-Fe₂O₃: 0.4177 m²/g and Pt-Pd/CB-Fe₂O₃: 0.5212 m²/g) than in the case of their counterparts (Pd-Pt/CB: 0.2124 m²/g and Pt-Pd/CB: 0.1937 m²/g).

Pd and Pt aggregates have been located on the surface of the cellulose beads (Fig. 2.). In the case of the iron oxide containing samples, the surface is richly covered with the promoter as well (Fig. 2, A and B).

Close

Fig. 2

SEM images of the designed cellulose-based catalysts: (A) Pd-Pt/CB-Fe₂O₃, (B) Pt-Pd/CB-Fe₂O₃, (C) Pd-Pt/CB, (D) Pt-Pd/CB.

Export to PPT

On the SEM images, nanoparticles are visible with a diameter of a few tens of nanometres (SI Fig. 1). However, due to the lower resolution of the SEM, particles with <10 nm diameter can not be located, thus, to get a better resolution, the nanoparticles have been detached from surface of the cellulose beads and these were examined by HRTEM. On the HRTEM images of the catalysts (SI Fig. 2) palladium and platinum nanoparticles have been identified. There is no significant difference between the size of the nanoparticles in the different samples. In each case, the particles are <10 nm and their average size is ∼4.9 nm. It was also confirmed by XRD measurements, that the size of the Pd and Pt nanoparticles is about 10 nm, and the anatase particles are 45 nm, while the hematite particles are bigger than 100 nm.

The XRD analysis (Fig. 3.) confirmed that elemental palladium nanoparticles present in the sample, as the peaks found at 40.0° and 45.9° 2ϴ angles can be identified as Pd (1 1 1) and Pd (2 0 0) reflexions (ICDD card number 046-1043). At 39.6° 2ϴ degree another peak has been located which corresponds to the Pt (1 1 1) reflexion (ICDD card number 00-004-082). In case of the iron oxide containing catalysts, peaks at 24.1°, 33.2°, 35.6° 40.9°, 49.5°, 54.1°, 62.5°, and 64.0° 2ϴ degrees have been found, which can be associated with the (0 1 2), (1 0 4), (1 1 0) (1 1 3), (0 2 4), (1 1 6), (2 1 4), and (2 0 8) reflexions of α-Fe₂O₃ (ICDD card number 33-0664). The wide peaks at 12.2°, 19.2° and 21.6° belongs to the (1 0 0), (1 1 0) and (0 2 0) planes of cellulose (French, 2014). Anatase were also identified in the samples, as peaks at 25.3°, 37.8°, and 48.0° 2ϴ degrees have been appeared (ICDD card number 21-1272) which corresponds to (1 0 1), (0 0 4), and (2 0 0) reflexions of TiO₂. The presence of anatase can be explained by the production method of the cellulose beads, because TiO₂ is the most widely used solid matrix (principal pillar supporting) during the adsorption of the epichlorohydrin in the initial step of cellulose production (Lei et al., 2003).

Close

Fig. 3

XRD patterns of the designed catalysts and the applied cellulose beads.

Export to PPT

FTIR measurements were also performed to identify the surface functional groups of the catalysts. These are important because appropriate groups could promote the adsorption of noble metals on the surface through ion exchange adsorption, electrostatic interactions, and complex formation. There is no significant difference between the coverage of the surface of the catalysts (Fig. 4.). The wide band located between 3100 cm⁻¹ and 3500 cm⁻¹ indicates the presence of —OH stretching (Maaloul et al., 2020; Yang et al., 2007). The peak at 2888 cm⁻¹ can be associated with C—H stretching (Jiang et al., 2017; Yang et al., 2007; Zhang et al., 2018). The band at 1643 cm⁻¹ indicates C⚌C vibration occur, but only in case of the pure cellulose beads (CB) (Yang et al., 2007). At 1371 cm⁻¹ a peak indicating O—H bending vibration in each sample (Mazlan et al., 2019). Strong peaks about 1000 and 1200 cm⁻¹ corresponds to C—O and C—O—C groups (Maaloul et al., 2020; Zhang et al., 2018). The peak at 895 cm⁻¹ belongs to C—H bending vibration (Szymanska-Chargot and Zdunek, 2013), while the Fe₂O₃ containing samples show bands about 541 cm⁻¹, which are related to Fe—O stretching (Alves et al., 2019).

Close

Fig. 4

FTIR spectra of the designed catalysts and the applied cellulose beads.

Export to PPT

The palladium, platinum and iron contents of the catalysts were determined by using ICP (Table 1). The Pt-Pd/CB catalyst contain ∼1.4 times more precious metal than its iron oxide promoted counterpart, Pt-Pd/CB-Fe₂O₃. There is no significant difference in the minor precious metal content of the catalysts with or without promoter.

All four catalysts were tested in chlorate hydrogenation using a batch set-up. Better results have been obtained by using the iron oxide promoted samples during the hydrogenation (Fig. 5, A). The highest activity was achieved with the Pd-Pt/CB-Fe₂O₃ catalyst, and it reduced the initial chlorate concentration from 200 mg/L to ∼15 mg/L (Fig. 5, A). The other three catalysts, Pt-Pd/CB-Fe₂O₃, Pd-Pt/CB and Pt-Pd/CB were also efficient, but the remaining chlorate concentrations after 3 h hydrogenation were much higher, about 30, 53, and 48 mg/L, respectively. Catalytic tests were also performed with cellulose beads containing iron oxide and with unsupported Fe₂O₃ (SI Fig. 3), but no significant activity was observed. To the best of our knowledge, the role of Fe₂O₃ in the case of chlorate hydrogenation has not been studied before. However, it was found that iron oxide promote the fast and effective adsorption/desorption of reactants and products on the surface of catalysts (Tang et al., 2021; Xu et al., 2021). It has also been observed for Pt or Pd-containing catalysts that Fe₂O₃ promotes oxygen adsorption and water formation which is advantageous in dehydrogenation processes such as chlorate reduction (Hensley et al., 2014; Li et al., 2021).

Close

Fig. 5

(A) Comparison of the catalytic activity of the cellulose based catalysts with different Pd and Pt contents, and with or without Fe₂O₃ promoter in batch chlorate reduction experiments. (B) Reuse tests were also performed with the most active catalyst (Pd-Pt/CB-Fe₂O₃) for five cycles. (C) The lnc vs t diagram with the corresponding calculated k values of chlorate hydrogenation carried out in the presence of the different catalysts.

Export to PPT

In our previous study (Sikora et al., 2020), commercially available palladium-containing catalysts have been examined in the hydrogenation of chlorate ions. The best Pd/C catalyst achieved 93% conversion under the same conditions. Although in that case less (200 mg) catalyst was used for the hydrogenation, but the noble metal content (10 wt%) of the tested commercially available catalysts was higher. The synthesized Pd-Pt/CB-Fe₂O₃ catalyst could achieve the same conversion with a lower metal content, furthermore, the Pd/C catalysts could not be tested and applied in a flow-through system due to their powder-based form.

By applying the cellulose beads supported Pd-Pt bimetallic catalyst, the hydrogenation reaction occurs according to a pseudo-first-order kinetics, which was verified by linear regression on the initial measurement points of the ln c_NB vs. reaction time plot (Fig. 5, C). The reaction rate constants (k) were determined by applying linear regression using Eq. (3) as follows:

Reuse tests were also performed with the most active catalyst (Pd-Pt/CB-Fe₂O₃) for 4 more cycles (Fig. 5/B. There is no significant difference between the first two hydrogenations, however with further uses the amount of reduced chlorate starts to decrease slightly. After the 5th hydrogenation cycle about 3 times more chlorate (50 mg/L) remained than in the first cycle.

The decrease in catalytic activity after the cycles, is caused by metal leaching (ICP results, Table 1.). Based on the ICP results, the metal loss is high, although the catalytic activity is still reasonable compared to the metal loss. In order to prevent metal leaching, the surface of the cellulose can be modified to increase its adsorption capacity and thus, the binding of the precious metal particles to create more stable catalysts (Hokkanen et al., 2016).

The most promising catalyst, Pd-Pt/CB-Fe₂O₃ was also tested in a continuous flow system (Fig. 6). The hydrogenation was monitored for more than 3 h, and the chlorate concentration decreased continuously over time and then reduced to 0 in about 160 min.

Close

Fig. 6

Catalytic activity of the Pd-Pt/CB-Fe₂O₃ catalyst in a chlorate hydrogenation experiment performed in a continuous flow set-up.

Export to PPT

4 Conclusion

Cellulose bead supported palladium and platinum catalysts have been successfully designed and prepared. The size of the noble metal particles is <10 nm. The catalysts have low specific surface area (<1 m²/g) and low precious metal content (<0.6 wt%), however they showed excellent catalytic activity in the hydrogenation of chlorate. The effect of iron oxide promoter material is also tested. Although the Pd-Pt/CB catalyst achieved only 72% conversion, the addition of iron oxide to the system increased this by more than 20% in a batch set-up. The Pd-Pt/CB-Fe₂O₃ catalyst was successfully applied in a continuous hydrogenation process, and within 160 min it was able to reduce the chlorate concentration to zero.