**Study of Catalytic Combustion of Chlorobenzene and Temperature Programmed Reactions over CrCeOx/AlFe Pillared Clay Catalysts**

#### **Yingnan Qiu 1, Na Ye 1, Danna Situ 1, Shufeng Zuo 1,\* and Xianqin Wang 2,\***


Received: 15 January 2019; Accepted: 25 February 2019; Published: 2 March 2019

**Abstract:** In this study, both AlFe composite pillaring agents and AlFe pillared clays (AlFe-PILC) were synthesized via a facile process developed by our group, after which mixed Cr and Ce precursors were impregnated on AlFe-PILC. Catalytic combustion of organic pollutant chlorobenzene (CB) on CrCe/AlFe-PILC catalysts were systematically studied. AlFe-PILC displayed very high thermal stability and large BET surface area (*S*BET). After 4 h of calcination at 550 ◦C, the basal spacing (*d*001) and *S*BET of AlFe-PILC was still maintained at 1.91 nm and 318 m2/g, respectively. Large *S*BET and *d*001-value, along with the strong interaction between the carrier and active components, improved the adsorption/desorption of CB and O2. When the desorption temperatures of CB and O2 got closer to the CB combustion temperature, the CB conversion could be increased to a higher level. CB combustion on CrCe/AlFe-PILC catalyst was determined using a Langmuir–Hinshelwood mechanism. Adsorption/desorption/oxidation properties were critical to design highly efficient catalysts for CB degradation. Besides, CrCe/AlFe-PILC also displayed good durability for CB combustion, whether in a humid environment or in the presence of volatile organic compound (VOC), making the catalyst an excellent material for eliminating chlorinated VOCs.

**Keywords:** AlFe-pillared clay; CrCeOx; chlorobenzene; catalytic combustion; temperatureprogrammed reaction

#### **1. Introduction**

Chlorinated volatile organic compounds (CVOCs) are considered to be very harmful to the environment, not only a direct harm on human health but also destroy the ozone layer [1,2]. Today, the major industrial processes for CVOCs elimination involve direct combustion at very high temperatures (above 850 ◦C). This is a fairly expensive process and produces highly toxic byproducts or intermediates by incomplete combustion, such as dioxins, Cl2, and CO [2,3]. The low operating temperatures (<500 ◦C) and high selectivity into harmless product, make catalytic combustion an attractive option [4–6].

Due to the high toxicity of dioxins and the need for laboratory safety, model reagents, such as chlorobenzene (CB), are used to predict destruction behavior of dioxins on different catalysts [7,8]. Vanadia-based catalysts [9,10], precious metals (Pt, Pd, Ru) supported on zeolites [11,12], and various oxides [13,14] are employed for the catalytic combustion process. However, these catalysts often have some disadvantages of relative low catalytic performance, rapid deactivation caused by coking or chlorine poisoning, high price, and the formation of polychlorinated benzene [14]. Transition metal oxide catalysts including cobalt, copper, manganese, and chromium oxides can not only resist

deactivation caused by chlorine poisoning but also enhance catalytic activities by the modification with rare earth elements [15,16]. Doping ceria into transition metals oxides could improve redox property of metal oxides, increase the mobility of oxygen and the rate of chlorine removal or transfer, thereby enhancing their catalytic properties and reaction stability during CVOCs combustion [17]. The catalytic behavior is associated with the interaction between the catalyst and the reactant, thus, understanding the relationship between adsorption/desorption and combustion of CVOCs over the catalysts becomes very important. The information can not only explain their own catalytic characteristics, but also provide insights into catalytic macroscopic behaviors [18–20].

Catalytic combustion is a typical gas–solid, two-phase reaction, which mostly occurs on the surface of catalysts. The advantage of heterogeneous catalysis is that the catalyst does not need to be separated, so the process can be operated in uninterrupted flow. Nevertheless, if there is a need to separate the catalyst, it can be done in a much simpler way than a homogeneous catalyst. Heterogeneous catalysts have been well applied in many fields such as heterogeneous Suzuki cross-coupling reaction catalyzed by magnetically recyclable nanocatalyst [21], modulated large-pore mesoporous silica as an efficient base catalyst for a Henry reaction [22], and magnetically separable and sustainable nanostructured catalysts for heterogeneous reduction of nitroaromatics [23–25]. The disadvantage of heterogeneous catalysis is that the catalyst can only use the catalytic active points on its surface, which is slightly inefficient, but can be improved by increasing the specific surface area (*S*BET). Therefore, the active components are usually dispersed on the carrier with large *S*BET. In a catalytic combustion reaction, the carrier can not only support the dispersed active components, but also increase the stability, selectivity, and activity of catalysts. The support can also reduce the use of high-priced active components, thereby reducing the cost of the catalyst. Common metal oxide catalyst carriers include γ-Al2O3, TiO2, SiO2, ZrO2, or their complexes. Molecular sieve supported catalysts also show good catalytic combustion activity. The molecular sieve carriers studied include ZSM-5, β-molecular sieve, SBA-15, MCM-41, and so on.

Natural clay is a hydrated aluminosilicate, which can be designed by crosslinking or pillaring, which leads to the formation of a class of material known as pillared clays (PILCs). Moreover, because of its wide distribution, abundant reserves, and low price, it has its own advantages in replacing existing catalyst carriers. PILCs have large *S*BET, uniform pore size distribution, and high thermal stability. They are good catalytic materials and carriers. A popular research topic in recent years is synthesizing PILCs supported catalysts for VOCs catalytic combustion reactions [26–28], whereas the application for CVOCs combustion is rare and the structure-activity relationship was not clear. It has been reported in the literature that *S*BET, pore volume (*V*p), and thermal stability of PILC can be improved by the synthesis composite pillaring agents, and these mixed pillaring agents has been widely used in the past 30 years [29]. At present, many kinds of mixed pillaring agents have been synthesized, with aluminum pillaring agents being one of the components. However, defects in the preparation process still exist. Therefore, there is an urgent need to optimize involved unit operations and simplify procedures, especially to reduce the amounts of NaOH and AlCl3 solutions.

The present work intends to simplify the synthesis steps, using a high-temperature hydrothermal one-step method, to prepare AlFe composite pillaring agents, and then synthesize AlFe-PILC [30]. Compared with Na-Mt, AlFe-PILC had good structural characteristics, such as larger *S*BET and *V*p, so it had the potential for use as a catalytic support. How to improve the efficiency of low temperature catalytic combustion of CB and improve the stability of the catalysts are the key problems to be solved in the current catalytic combustion process. This paper intends to use AlFe-PILC as a carrier to prepare CrCeOx catalysts, and to study their application in the catalytic combustion reaction of CB, aiming at forming achievements in advanced catalytic materials and their applications. The adsorption/desorption properties and catalytic performance of CB were systematically studied in order to get a clear map about the structure–activity relationship for the catalytic reaction and explore its potential for further industrial application.

#### **2. Experimental Section**

#### *2.1. AlFe-PILC and Catalyst Preparation*

The starting material was the sodium form of montmorillonite (Na-Mt) (>100 mesh, Hengsheng Trading Co., Ltd., Baotou, China). AlFe pillaring agents were prepared with Locron L (Clariant, Switzerland, Al2(OH)5Cl·2-3H2O) and ferric nitrate solution. The following preparation of AlFe pillaring (the molar ratio of Al/Fe is 5) and AlFe-PILC using the similar method detailed in our previous research [30].

Cr/Na-Mt, CrCe(5:1)/Na-Mt, Ce/AlFe-PILC, Cr/AlFe-PILC, and CrCe/AlFe-PILC were synthesized by impregnating a Cr and Ce nitrate solution onto the equal volumes of supports (Na-Mt and AlFe-PILC) overnight, followed sequentially by drying and calcination at 500 ◦C for 2 h. Cr, Ce, or CrCe loading for each catalyst was 10 wt.%, and Cr/Ce molar ratios were adjusted to 2.5:1, 5:1, 7.5:1, and 10:1, respectively. All the reagents were analytically pure, and obtained by Shanghai Chemical Reagent Factory (Shanghai, China).

#### *2.2. Characterization*

The samples were characterized using X-ray diffraction (XRD) (PANalytical, Almelo, Netherlands) for *d*<sup>001</sup> value and phase composition. The *S*BET, mesopore area (*A*mes), *V*p, micropore volume (*V*mic), and pore size distribution of the samples were determined via N2 adsorption isotherms using a TristarII 3020 apparatus (Micromeritics Company, Atlanta, GA, USA). High-resolution transmission electron microscopy (HRTEM) on a JEM-2100F (JEOL, Valley, Japan) was employed to get the catalyst morphology and particle size. The chemical compositions of the catalysts were determined with energy dispersive X-ray spectroscopy (EDS) using an Oxford INCA instrument (Oxford Instruments, Warrington, UK). All the characterization methods for the samples have been reported and detailed in our previous research [30–32].

#### *2.3. Catalytic Performance Tests and Temperature-Programmed Reactions*

The activity of the catalysts was evaluated in a WFS-3010 microreactor (Xianquan, Tianjin, China). The degradation products were detected by mass spectrometry (MS, QGA, Hiden, U.K.). No byproduct other than H2O, CO2, and HCl was detected. Thus, the conversion was calculated based on CB consumption [31]. To further study the "mixture effect" of the feed gases, 1% (*v*/*v*) water vapor and 100 ppm toluene were also introduced. Besides, the durability of CrCe (5:1)/AlFe-PILC for the catalytic combustion of CB was investigated at a CB concentration of 500 ppm and gas hourly space velocity (GHSV) of 25,000 h<sup>−</sup>1.

H2 temperature-programmed reduction (H2-TPR) was conducted on a CHEMBET-3000 instrument (Quantachrome, Boynton Beach, FL, USA) to evaluate the reducibility of the catalysts. The sample (50 mg) was pre-treated in air at 300 ◦C for 0.5 h, and then the temperature was reduced to 100 ◦C. The flow rate of the reductive gas (5 vol.% H2/Ar, purified by deoxidizer and silica gel) was 40 mL/min, and the reaction temperature was elevated by 7.5 ◦C/min. The H2 uptake was determined using a thermal conductivity detector (TCD) detector (Shimadzu, GC-14C, Kyoto, Japan), and the H2O produced was absorbed using 5 Å zeolite [32].

Temperature-programmed desorption (TPD) and temperature-programmed surface reaction (TPSR) measurement were carried out in the same equipment as the catalytic performance tests to determine the adsorption capacity and the relationship between desorption performance and catalytic combustion properties [30,32]. Prior to the measurement, 350 mg catalyst was pre-treated in Ar (99.99%) at 300 ◦C for 30 min, then the temperature was decreased to 50 ◦C. Adsorption gas (40 mL/min) was a mixture of Ar (99.99%) and CB (about 500 ppm). The quantitative amounts were estimated by integrating the desorption curve. After the adsorption reached an equilibrium (CB concentration in the effluent gas was monitored using Gas Chromatography-Mass Spectrometry (GC-MS), the catalysts were purged by Ar (99.99%) for a period of time at 50 ◦C until CB concentration to constant. Then, the desorption and catalytic properties of the catalysts were measured from 50 to 500 ◦C with a heating rate of 7.5 ◦C/ min in 20 vol.%O2/80 vol.%Ar (without CB). The reactants and products (such as CB (m/z =112), CO2 (44), H2O (18), Cl2 (71), and HCl (36.5) were analyzed with an on-line MS apparatus.

O2 temperature-programmed desorption (O2-TPD) was also performed using the same apparatus. The catalyst (350 mg) was firstly treated in 10 vol.%O2/90 vol.%Ar at 300 ◦C for 0.5 h. After the temperature was slowly cooled down to room temperature, followed by an Ar purge (40 mL/min) for 30 min, the sample was heated from 50 to 900 ◦C with a heating rate of 7.5 ◦C/min in Ar flow. The signal of desorbed oxygen was monitored by the MS.

#### **3. Results and Discussion**

#### *3.1. Material Textural Properties*

#### 3.1.1. XRD Analysis

Figure 1 presented the XRD patterns (2*θ*: 10–80◦) of the Cr/Ce catalysts supported by Na-Mt and AlFe-PILC. The diffraction peaks belonging to cristobalite and quartz appear at 19.8◦ and 26.7◦ (2*θ*), respectively [33]. Cristobalite and quartz are two of the main components of montmorillonite. They have the characteristics of high temperature resistance, which can ensure the stability of the catalysts in a high temperature gas–solid continuous reaction. Therefore, they play an important role in catalyst components. The diffraction peaks of Fe2O3 appeared in all the AlFe-PILC based catalysts because the amount of Fe2O3 increased after AlFe pillaring process. The diffraction peaks of CeO2 appeared in Ce/AlFe-PILC catalyst. Compared with Cr/Na-Mt, the diffraction peak intensity of Cr2O3 in Cr/AlFe-PILC clearly decreased, and the result showed that the dispersion of Cr2O3 particles on AlFe-PILC was greatly improved. After adding Ce, the diffraction intensity of Cr2O3 for CrCe(5:1)/AlFe-PILC further decreased. On the one hand, the addition of Ce was beneficial to the dispersion of Cr2O3, and on the other hand, it may have been due to the reduction of Cr2O3 content. Notably, the CeO2 diffraction peaks were not found in CrCe(5:1)/AlFe-PILC, possibly because the small amount of CeO2 was highly dispersed on AlFe-PILC.

**Figure 1.** XRD patterns of the catalysts: (**a**) Cr/Na-Mt, (**b**) Ce/AlFe-PILC, (**c**) Cr/AlFe-PILC, and (**d**) CrCe(5:1)/AlFe-PILC.

#### 3.1.2. N2 Adsorption/Desorption

Table 1 summarizes the textural properties of samples. *SBET* and *A*mes of Na-Mt were only 51 and 41 m2/g, and the values of *V*<sup>p</sup> and *Vmic* were 0.076 and 0.0043 cm3/g, respectively. AlFe-PILC's *SBET* and *V*<sup>p</sup> reached 318 m2/g and 0.195 cm3/g, respectively, indicating that the formed composite AlFe polycation was relatively large and thus the clay layers were further stripped to form more porous structures. The *Vmic* of AlFe-PILC was 0.077 cm3/g, and it was about 39.5% in *V*p. Compared with the Na-Mt and AlFe-PILC support, the supported Cr or CrCe catalysts exhibited lower *SBET* and *V*<sup>p</sup> values, indicating that some of the Cr and Ce ions migrated into the pores and clay layers, and thus blocked some of the pores. It was worth noting that a large number of micro-mesoporosity in AlFe-PILC support was favorable for good dispersion of active species and rapid diffusion of reactants, thus significantly enhanced their catalytic activity of various reactions.

**Table 1.** Characteristics of the samples: values of surface area and pore volume.


<sup>a</sup> Calculated from BJH method. <sup>b</sup> Total pore volume estimated at P/P0 (relative pressure) = 0.99. <sup>c</sup> Calculated from the *t*-plot method.

In Figure 2a, N2 adsorption/desorption isotherms for all the materials were type IV, while its type H3 adsorption–desorption hysteresis appeared at P/P0 above 0.45, indicating that the material had a mesoporous structure and the pores in the material were slit pores formed by layer-like structures. The adsorption amount of Na-Mt was low; however, AlFe-PILC had a pronounced increase in adsorption because more pores were formed by AlFe polyoxycations. The addition of Cr2O3 and CeO2 to Na-Mt and AlFe-PILC decreased N2 adsorption capacity and thus pore volume, indicating the doped cations entered and/or blocked the pores of Na-Mt and AlFe-PILC. In Figure 2b, the average mesoporous diameters of AlFe-PILC materials were distributed in a narrow range of 3.96 nm and were wider than the pore-diameter distribution range of Na-Mt (3.10 nm), confirming the pore size was increased after pillaring. The average pore size of CrCe(5:1)/AlFe-PILC was in a narrow region of approximately 3.65 nm. The stability of AlFe-PILC support was good and the mesoporous structure was not destroyed.

**Figure 2.** *Cont.*

**Figure 2.** Characteristics of the samples: (**a**) N2 adsorption/desorption isotherms, and (**b**) pore-size distributions.

#### 3.1.3. HRTEM Analysis

Figure 3 shows HRTEM picture and the EDS spectra of CrCe(5:1)/AlFe-PILC. It can be seen that CrCe(5:1)/AlFe-PILC had a layered structure and the active particles (5–10 nm in size) were uniformly distributed throughout the support, and the layered structure of AlFe-PILC was not damaged after loading active ingredients. Al, Fe, Cr, Ce, O, and other elements were identified in the EDS spectra, which confirmed that the active species (Cr and Ce) were successfully loaded on the surface of AlFe-PILC. The results indicated that AlFe-PILC was a good support for highly dispersed active species. All these properties were conducive to improving the catalytic degradation of CB.

**Figure 3.** *Cont.*


**Figure 3.** HRTEM picture and the EDS spectra of CrCe(5:1)/AlFe-PILC.

#### *3.2. Catalytic Performance Test*

#### 3.2.1. CB Combustion and Durability Test

Figure 4 presents the conversions of CB combustion on various catalysts. In Figure 4a, Cr/Na-Mt exhibited poor performance and did not fully convert CB until 460 ◦C. Cr/AlFe-PILC caused complete degradation of CB at 320 ◦C, about 140 ◦C lower than the degradation temperature of CB required for Cr/Na-Mt. The conversion of Ce/AlFe-PILC was negligibly low, and it was 87%, even at a reaction temperature of 500 ◦C. Ceria doping significantly improved the catalytic activities of Cr/Na-Mt and Cr/AlFe-PILC. In addition, the molar ratios of Cr/Ce (2.5, 5, 7.5, and 10) had an effect on the catalytic performance of CrCe/AlFe-PILC (Figure 4b). The catalysts exhibited a lower performance when the Cr/Ce ratio was less than 5, possibly indicating Cr2O3 was the active species and CeO2 acted as an assistant. The catalyst performance decreased when the Cr/Ce molar ratio was larger than 5, possibly because less oxygen vacancies existed with a relatively lower amount of CeO2. Therefore, the content of CeO2 was one of the key factors to improving the performance of CrCe/AlFe-PILC. In particular, CrCe(5:1)/AlFe-PILC had the highest catalytic performance and could completely degrade CB at about 290 ◦C. No Cl2 or other byproducts were detected, showing that the catalyst had good selectivity for HCl without producing secondary pollution.

Figure 5 shows the curves of CB over CrCe(5:1)/AlFe-PILC in the continuous reaction process. There was no significant drop for catalytic activities within 1000 h tests, suggesting that the CrCe(5:1)/AlFe-PILC catalyst was durable. Moreover, this catalyst also displayed good catalytic performances in the presence of 100 ppm toluene or 1% water vapor, further indicating its high potential for industrial application.

**Figure 4.** (**a**) CB conversions vs. temperature over Cr/Na-Mt, CrCe(5:1)/Na-Mt, Ce/AlFe-PILC, Cr/AlFe-PILC, and CrCe(5:1)/AlFe-PILC. (**b**) The effect of Cr/Ce molar ratios on CB catalytic combustion over CrCe/AlFe-PILC.

**Figure 5.** Lifetime test performed for CrCe(5:1)/AlFe-PILC at 280 ◦C. CB concentration: 500 ppm; gas hourly space velocity (GHSV): 25,000 h<sup>−</sup>1; catalyst amount: 350 mg.

#### 3.2.2. Effect of CB Concentration and Gas Hourly Space Velocity

Figure 6 presents the effect of CB inlet concentrations on the catalytic performance of CrCe(5:1)/AlFe-PILC. The change of its inlet concentration had a great influence on CB conversion from 500 to 2500 ppm. Furthermore, when the inlet concentration was in the range of 500 to 1500 ppm, CB conversion increased appreciably. This was primarily because low concentration CB only provided a small amount for chemisorbed CB on catalyst active sites and could act as the controlling factor of the reaction. However, as the concentration of CB continued to increase, CB conversion decreased until it was completely prohibited, which may be related to chemisorbed oxygen on the catalyst active sites becoming the reaction controlling factor [34]. The result indicated that CB degradation combustion proceeds via a Langmuir–Hinshelwood (L-H) mechanism, and this catalyst could be used for removing CB waste gases with a wide range of concentrations.

**Figure 6.** The effect of inlet concentration on CB catalytic combustion over CrCe(5:1)/AlFe-PILC. CB concentration: 500–2500 ppm; GHSV: 25,000 h<sup>−</sup>1; catalyst amount: 350 mg.

Figure 7 shows the effect of GHSV on the CB catalytic combustion activities over CrCe(5:1)/AlFe-PILC. GHSV is the gas hourly space velocity. To calculate this parameter, the flow rate of feed gas (involved inert and main components) can be adjusted. Then, GHSV is the ratio of gas flow rate in standard conditions to the volume of the catalyst. Increasing GHSV slightly decreased the catalytic performance, indicating that this catalyst was highly effective for CB destruction in different reaction conditions. The catalyst active sites were already fully occupied, even with the lowest GHSV used in this work, and more reactant molecules provided by high GHSV could not be chemisorbed and reacted. Thus, high temperature was required to obtain the same conversion with high GHSV.

**Figure 7.** CB conversions vs. temperature over CrCe(5:1)/AlFe-PILC under the conditions of CB concentration at 500 ppm and GHSV at 25,000–40,000 h<sup>−</sup>1.

#### *3.3. Temperature-Programmed Reaction Studies*

#### 3.3.1. H2-TPR Analysis

H2-TPR profiles of the catalysts are shown in Figure 8. The reduction of Fe2O3 species was obvious in all the catalysts (γ peak), which was from the relatively high contents of Fe2O3 (4.45% in the original clay) [35,36]. Compared with the γ peak area from Cr/Na-Mt, the areas from Cr/AlFe-PILC and CrCe(5:1)/AlFe-PILC increased, revealing that more iron oxide species were formed as Fe2O3 pillars in AlFe pillaring. In the case of Ce/Na-Mt, it was beneficial for the reduction of surface and bulk CeO2 to have two reduction peaks at 541 and 745 ◦C. There were two reduction peaks below 650 ◦C in the Na-Mt and AlFe-PILC-supported Cr catalysts, which indicated that peaks α<sup>1</sup> and α<sup>2</sup> were the reduction peaks of the surface and inside Cr2O3, respectively. For CrCe(5:1)/AlFe-PILC, peaks α<sup>1</sup> and α<sup>2</sup> were divided into two or three peaks, which suggested the better-dispersed Cr2O3 on the AlFe-PILC support. Compared with Cr/Na-Mt, the reduction peaks of CrCe(5:1)/AlFe-PILC systematically shifted to lower temperatures, indicating CeO2 improved the reducibility of Cr2O3 by increasing Cr2O3 dispersion and lattice oxygen mobility. The peak β<sup>2</sup> of CrCe(5:1)/AlFe-PILC at 588 ◦C was the reduction peak of bulk CeO2, and the peak of surface CeO2 overlapped with peak α2. The CeO2 reduction peak was shifted toward lower temperatures compared with that from Ce/Na-Mt. This shift occurred because Cr2O3 underwent a stronger oxidation process and could be more easily reduced, allowing it to interact with CeO2 to produce a reduction peak at a lower temperature. It suggested that the interaction between Cr2O3 and CeO2 species weakened the Ce-O bond and promoted the reduction of CeO2. The α peak temperatures followed: Cr/Na-Mt > Cr/AlFe-PILC > CrCe(5:1)/AlFe-PILC. The results indicated that the interaction between Cr2O3 and CeO2 species could improve the mobility of oxygen species in the catalysts, thus improving the reduction of both Cr2O3 and CeO2 species.

**Figure 8.** H2-TPR spectra: (**a**) Ce/Na-Mt, (**b**) Cr/Na-Mt, (**c**) Cr/AlFe-PILC, and (**d**) CrCe(5:1)/AlFe-PILC.

#### 3.3.2. TPD and TPSR Analysis

The adsorption/desorption of CB, catalytic combustion behavior, and the evolution of the main products (CO2, H2O, and HCl) over the catalysts were investigated using CB-TPD/TPSR techniques (Figure 9). As it was mentioned previously, the CB combustion on CrCe(5:1)/AlFe-PILC catalyst proceeded via an L-H mechanism, where the adsorption of reactants on the catalyst active sites was a critical step. In Figure 9a, CrCe(5:1)/Na-Mt and CrCe(5:1)/AlFe-PILC showed different CB adsorption capacities. The CB absorption capacities of CrCe(5:1)/AlFe-PILC (44.8 μmol/g) was obviously stronger than CrCe(5:1)/Na-Mt (7.9 μmol/g) by integrating over the absorption spectra. The above results fully proved that clay materials with larger *S*BET, *V*p, and *d*001-value favor CB adsorption. In Figure 9b, the temperature of CB desorption peaked for CrCe(5:1)/Na-Mt and CrCe(5:1)/AlFe-PILC were 145 ◦C and 198 ◦C, respectively, indicating that the interaction of CB and CrCe(5:1)/AlFe-PILC was stronger than with CrCe(5:1)/Na-Mt. Therefore, CB could remain inside the pores or outside the surface of CrCe(5:1)/AlFe-PILC for a longer time, being conducive to the adsorption and catalytic degradation of CB. The results indicated that improved structure and the strong interaction between CrCe mixed oxides with AlFe-PILC enhanced the adsorption of CB.

**Figure 9.** *Cont.*

**Figure 9.** Performance of CB adsorption/desorption and catalytic combustion over CrCe(5:1)/Na-Mt and CrCe(5:1)/AlFe-PILC. (**a**) CB adsorption of CrCe(5:1)/Na-Mt and CrCe(5:1)/AlFe-PILC; (**b**) CB desorption of CrCe(5:1)/Na-Mt and CrCe(5:1)/AlFe-PILC; (**c**) CB desorption/catalytic combustion of CrCe(5:1)/Na-Mt; (**d**) CB desorption/catalytic combustion of CrCe(5:1)/AlFe-PILC.

As shown in Figure 9c,d, CB desorption was accompanied with CB combustion under O2/Ar, and the adsorbed CB species reacted with lattice O from CrCeOx to form CO2, H2O, and HCl. CB was completely reacted over CrCe(5:1)/AlFe-PILC at about 300 ◦C, while it needed about 440 ◦C on CrCe(5:1)/Na-Mt. CO2, H2O, and HCl were detected, but CO and Cl2 were not detected, indicating that the catalysts in the study had high selectivity to HCl and CO2 formation. It was notable that the peak temperature of the products for CrCe(5:1)/Na-Mt was at 413 ◦C, which was much higher than that of CB desorption peak temperature (145 ◦C). However, the peak temperature of product for CrCe(5:1)/AlFe-PILC was at 275 ◦C, which was close to that of CB desorption (198 ◦C). This phenomenon can explain why CrCe(5:1)/AlFe-PILC had the highest CB degradation activities compared to other catalysts in this work. The larger overlapped region between CB desorption and catalytic combustion, the better the catalytic performance. Therefore, tuning the CB adsorption and catalytic properties was a key to designing an efficient catalyst for CB catalytic combustion.

In order to find out the relationship between the oxygen species absorbed on the catalyst surface and the catalytic properties, O2-TPD were investigated from 50 to 900 ◦C. The O2-TPD plots for Cr and CrCe metal oxide catalysts consisted of oxygen desorption regions shown in Figure 10. There were three types of desorption peaks, the α desorption peak, the β desorption peak, and the γ desorption peak. Furthermore, these three peaks could be assigned to superoxide ion O2 −, peroxide ion O2 <sup>2</sup>−/O−, and lattice oxygen ion O2−, respectively [37,38]. Increasing the temperature is beneficial to increase the rate of desorption and transformation of superoxide species into O2 <sup>2</sup>−, O−, and Olattice<sup>2</sup><sup>−</sup> [39]. It can be seen that the α and β desorption peaks follow: CrCe(5:1)/AlFe-PILC < Cr/AlFe-PILC < CrCe(5:1)/Na-Mt < Cr/Na-Mt, which was in good agreement with the aforementioned catalytic performance of CB combustion. It was worth mentioning that the total amount of surface-active oxygen species, in terms of the sum of α and β desorption areas, follows the same sequence of the peaks. It can be observed from the γ desorption peak that adding CeO2 increased the desorption area of lattice oxygen ion O2<sup>−</sup> compared with the non-doped catalyst. Thus CrCe(5:1)/AlFe-PILC exhibited the highest oxidation performance since electrophilic Oads (O2 −, O2 <sup>2</sup>−, O−) played a critical role in the complete oxidation of organic compounds [40].

**Figure 10.** O2-TPD profiles of Cr/Na-Mt, CrCe(5:1)/Na-Mt, Cr/AlFe-PILC, and CrCe(5:1)/AlFe-PILC.

#### **4. Conclusions**

In this paper, AlFe-PILC supported CrCe mixed oxides are synthesized and used for adsorption/desorption and catalytic combustion of CB. A series of characterization methods were used to investigate the structure and redox properties of these materials, including HRTEM-EDS, H2-TPR, TPD/TPSR, and O2-TPD. Comparing the results of the *S*BET, *V*p, and *d*001-value, AlFe-PILC performed better than Na-Mt. Without doubt, AlFe-PILC synthesized in this study constituted a class of porous materials with excellent properties. A large number of micro-mesoporosity and Ce was added to the AlFe-PILC to optimize its structure and improve the dispersion of Cr2O3 particles on the AlFe-PILC. XRD analysis and HRTEM images clearly revealed the stable layered structure with the *d*<sup>001</sup> value ≈1.91 nm and well-dispersed active species in AlFe-PILC. The addition of Ce and optimized structure of support greatly improved the oxidative property of Cr2O3. CB-TPD experiments reveal that the optimized structure coupled with the strong interaction between CrCe metal oxides and AlFe-PILC enhanced CB adsorption capacity and adsorption strength. CB-TPSR results showed that the larger the overlapped region between CB desorption and the catalytic combustion, the better the catalytic performance. In particular, CrCe(5:1)/AlFe-PILC show an excellent catalytic property, and stability was due to the lower temperature of completely degraded CB (approximately 290 ◦C) and the conversion remained stable for 1000 h. CB catalytic combustion on CrCe/AlFe-PILC catalyst was via a Langmuir–Hinshelwood mechanism, and adjusting adsorption/desorption properties was one of the most important factors for designing efficient catalysts. CrCe/AlFe-PILC also exhibited good durability for CB destruction, both in the humid condition and in the presence of toluene; therefore, this catalyst deserves wide attention and it is a potential prospect for industrial application.

**Author Contributions:** S.Z. designed the experiments; Y.Q. and D.S. performed the experiments; S.Z. and X.W. wrote the paper; N.Y. provided some experimental equipment and guided the experiments.

**Funding:** Zhejiang Public Welfare Technology Research Project (LGG19B070003), the Foundation of Science and Technology of the Shaoxing City (2018C10019) and 2018 Zhejiang Province Innovation Training Program for College Students (2018R432031).

**Acknowledgments:** We would like to acknowledge the financial support from Innovation Team of Huzhou South Taihu Elite Program. Moreover, we would also grateful to Zhejiang Da-Feng Automobile Technology Co., Ltd for the related experiment and test.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## *Article* **Combined E**ff**ect of Pressure and Carbon Dioxide Activation on Porous Structure of Lignite Chars**

#### **Natalia Howaniec**

Department of Energy Saving and Air Protection, Central Mining Institute, Pl. Gwarkow 1, 40-166 Katowice, Poland; n.howaniec@gig.eu; Tel.: +48-32-259-2219

Received: 3 April 2019; Accepted: 22 April 2019; Published: 23 April 2019

**Abstract:** Lignite is an important natural resource with the application potential covering present and future energy systems, including conventional power plants and gasification systems. Lignite is also a valuable precursor for the production of porous materials of tailored properties for various environmental applications, including the removal of contaminants from gaseous or liquid media. Although the lignite-based activated carbons are commercially available, various approaches to produce carbon materials of desired properties are still being reported, covering temperature, partial oxidation and chemical activation effects on surface and structural properties of these materials. Limited data is, however, available on the effects of pressure as the activation parameter in shaping the porous structure of carbonaceous materials, in particularly lignite-derived. In the study presented the combined effect of carbon dioxide activation and pressure in the range of 1–3 MPa at the temperature of 800 ◦C on the development of porous structure of lignite chars was reported. The study was also focused on poor-quality resources valorization by using a relatively low calorific value, low volatiles and high ash content lignite as a carbon material precursor. The results showed that the application of pressure in carbon dioxide-activation process at 800 ◦C results in generation of chars of comparable or higher specific surface area than the carbon materials previously received with demineralization and carbon dioxide activation of lignite. They also proved that the combined pressure and carbon dioxide activation may be effectively applied in conversion of low quality lignite into valuable porous materials.

**Keywords:** lignite; porous structure; carbon dioxide; activation; pressure

#### **1. Introduction**

Lignite is a considerable element of world coal resources [1]. As such, it is not only an important energy resource applied currently in conventional power plants but also a fuel particularly suitable for gasification and co-gasification systems for its higher reactivity when compared to bituminous coals [2–4]. Lignite is also a valuable parent material for the production of porous materials of tailored properties for various industrial applications [5–8]. The majority of them concerns sorption processes in the removal of contaminants, e.g., phenol, mercury, sulphur oxides, copper, and organic compounds from gaseous or liquid media [9–14]. Although the lignite-based activated carbons are commercially available, various approaches to produce carbon materials of desired properties are still being reported. The vast literature is available on the application of the temperature and various oxidizing agents [5,7,11–17] as well as acidic or basic treatment [7,10,18–20] in shaping the surface and structural properties of lignite-derived carbon materials. Limited data is, however, available on the effects of pressure as the activation parameter in shaping the porous structure of carbonaceous materials [21–23], in particularly lignite [24–26]. A few studies considering the development of porous structure of bituminous coals under carbon dioxide atmosphere [27,28] or inert gas [24,29,30] and elevated pressure are available. Porous structure development of chars is also considered in the literature in terms of lignite suitability and chars reactivity in gasification process, in particularly with the incorporation of carbon dioxide in a valorization cycle as a gasification process reactant [20,31–34]. Previous studies showed that the values of surface area of bituminous coal chars developed under carbon dioxide atmosphere in general drop with process pressure [27,28]. Swelling properties have been reported to influence the porous structure of coal chars under inert gas atmosphere and pressurized conditions, and although this effect is enhanced with increased coal rank, it also depends on volatiles and specific petrographic components content [24,29] and vary with pressure values [5,30]. The amounts of particular mineral components, differing in rates of expansion, e.g., kaolinite, quartz, pyrite, and calcite affect the porous structure development of lignite chars at increased temperature. The mineral matter in bituminous coal-derived chars was also reported to be severely affected qualitatively and quantitatively when treated with carbon dioxide at 900 ◦C. The inorganic amorphous phase was decomposed and the reduced mineral forms were oxidized and reacted with aluminosilicates forming calcium and iron minerals [35].

The combined effect of carbon dioxide activation and pressure on the development of porous structure of lignite chars has not been reported so far. Therefore, the experimental study on the application of pressure in the range of 1–3 MPa in carbon dioxide-activation of lignite chars was performed and its results are presented in this paper. The lignite of relatively low calorific value and high ash content was selected as a precursor which makes the study valid also in the context of poor quality resources valorization.

#### **2. Materials and Methods**

The lignite of relatively low calorific value, high ash and sulfur content, provided by Polish opencast mine from Szczercow deposit, was selected as the chars precursor. Lignite of Szczercow deposit is characterized by high huminite maceral group content, of approx. 82% vol., including densinite share of 44% vol. and atrinite content of 23% vol. The liptinite and intertinite contents are of 7 and 4% vol. Highly porous macerals, like textinite, ulminite and atrinite amount in total to approximately 37% vol. [33]. The mineral matter content is approximately 8% vol. with the dominance of clay minerals. The proximate and ultimate analyses of lignite tested were performed in an accredited laboratory in compliance with the relevant standards and are given in Table 1.


<sup>1</sup> PN-G-04560:1998 with the use of automatic thermogravimetric analyzers LECO (St. Joseph, MI, USA): TGA 701 and MAC 500; <sup>2</sup> PN-G-04516:1998 calculated by difference; <sup>3</sup> PN-G-0484:2001 with the use of an automatic analyzer TruSpec S by LECO; <sup>4</sup> PN-G-04571:1998 with the use of an automatic analyzer TruSpec CHN by LECO; <sup>5</sup> PN-G-04513:1981 with the use of LECO calorimeters: AC-600 and AC-350.

Lignite sample of 1 g was heated in a high-pressure thermogravimetric analyzer (Rubotherm GmbH, Bochum, Germany) with the heating rate of 20 ◦C/min in an argon atmosphere to the final process temperature of 800 ◦C (see Figure 1) and pressurized to the final process pressure of 1, 2 or 3 MPa, respectively. When the final process temperature and pressure were reached carbon dioxide was introduced to the reactor with a flow rate of 100 mL/min for 120 min.

**Figure 1.** Schematic diagram of the system for carbon materials preparation at high temperature and pressure and with the use of carbon dioxide activation.

The resulting carbon materials were outgassed at 120 ◦C overnight and analyzed in terms of their porous structure parameters with the application of a gas sorption analyzer Autosorb iQ (Quantachrome Instruments, Boynton Beach, FL, USA). Based on the nitrogen sorption isotherm data acquired at −196 ◦C, the specific surface area and pore size distribution were determined with the application of the multi-point BET method [36] and the Density Functional Theory (DFT) [37], respectively. The total pore volume was quantified as the volume at the relative pressure of 0.99. The narrow micropore area and volume were further analyzed on the basis of the carbon dioxide isotherm at 0 ◦C and the Monte Carlo (MC) method [38]. The surface properties of the resulting carbon materials were also explored with the use of a scanning electron microscope SU-3500N (Hitachi High-technologies Corporation, Tokyo, Japan).

#### **3. Results and Discussion**

The porous structure of carbon materials produced from low quality lignite at the temperature of 800 ◦C and under the pressure of 1–3 MPa with a carbon dioxide activation step was complex and composed of micro- and mesopores. High uptake at low relative pressures, which may be seen in Figure 2, presenting the exemplary nitrogen isotherm for lignite chars tested, is indicative of micropores present in the porous structure of lignite chars. The occurrence of a hysteresis loop proves that the material is also rich in mesopores and its profile reveals the irregular, slit-like shape of pores [39].

**Figure 2.** Nitrogen isotherm (−196 ◦C) for lignite chars produced at 800 ◦C, under 2 MPa and with carbon dioxide activation.

The results of the porous structure parameters of carbon materials produced under the pressure of 1–3 MPa and at atmospheric pressure, for comparison purposes, are given in Table 2. It may be seen that the rise in pressure resulted in an increased specific surface area values. The highest increase, of 9%, was observed with the change in the process pressure from the atmospheric to 1 MPa. Further increase in pressure in the range 1–3 MPa gave the rise in the specific surface area of 6–7% per 1 MPa.

**Table 2.** Properties of porous structure of lignite chars determined with the use of nitrogen sorption isotherm at −196 ◦C and carbon dioxide sorption isotherm at 0 ◦C.


The average pore diameter showed a decrease with pressure in the range 0.1–2 MPa (see Table 2). The difference in values of the average pore size of carbon materials produced under 2 and 3 MPa was within the experimental error. It implies that the pressure-enhanced development of smaller pores resulting from the devolatilization, moisture release and partial oxidation of carbon with carbon dioxide under the pressure of up to 2 MPa was counteracted with merging of pores in larger structures under the pressure of 3 MPa. A similar trend was also observed previously for lignite [25,26] and bituminous coal [22,26], as well as biomass chars [21,26] with no carbon dioxide activation, although the limiting values of pressure varied between 2 and 3 MPa for various parent materials. The differences in the values of the total pore volume for chars activated with carbon dioxide under the pressure of 0.1–2 MPa were within the experimental error. However, under the highest pressure tested, of 3 MPa, an increase of approximately 8% in the total pore volume was observed (see Table 2).

The pore size distribution (PSD) data showed an increase in a pore volume with pressure applied mainly because of increasing volume of mesopores of a diameter over 5 nm in lignite chars with carbonization pressure from 0.1 to 3 MPa (see Figure 3a). The pore volume of small micropores (diameter below 1 nm) determined based on the nitrogen isotherm increased with a change in the pressure from atmospheric to elevated, and was comparable for chars generated under 1–3 MPa. The variation in process pressure seemed to have no measurable effect on the development of 1–2 nm micropore volume with the carbon dioxide activation at 800 ◦C, which may be related to closure of

these micropores or their merging in larger structures at higher pressures. The latter seems to be also demonstrated by the increasing share of mesopores of a diameter over 5 nm. These pores had a dominant role in the increase in the total DFT pore volume with pressure which amounted to 4, 11 and 17% with pressure rise from atmospheric to 1, 2, and 3 MPa, respectively. The dominant share of micropores in shaping the pore area is also visible (see Figure 3b) as well as the positive effect of pressure applied on the development of the smallest pores (of a diameter below 1 nm) and the total DFT area of pores (Figure 3b). Under 3 MPa the share of mesopores of a diameter over 2 nm in the pore area also slightly increased which again may be indicative of merging of pores in larger structures under the highest pressure tested. The total DFT area of pores increased with change of the process pressure from atmospheric to 1, 2 and 3 MPa of 15%, 18% and 24%, respectively.

**Figure 3.** Distribution of: (**a**) pore volume and (**b**) area based on DFT method and nitrogen isotherm (−196 ◦C) for lignite chars produced at 800 ◦C, under 0.1–3 MPa and with carbon dioxide activation.

The area and volume of narrow micropores (diameter range of 0.45–1.5 nm), determined based on the carbon dioxide sorption isotherm, showed a clear increase with a change in process conditions from atmospheric to pressurized (Table 2). This was caused mainly by an enhanced development of pores of a diameter 0.65–0.85 nm (see Figure 4). No meaningful difference was however observed with further changes in pressure from 1 to 3 MPa in terms of narrow microporosity, except for a slight increase in the volume of pores of a diameter in the range of 0.85–1.05 nm under 3 MPa, giving a slight rise to the total MC pore volume, which may be the effect of structural rearrangements under the maximum pressure tested.

**Figure 4.** Narrow microporosity determined based on the carbon dioxide isotherm (0 ◦C) for lignite chars produced at 800 ◦C, under 0.1–3 MPa and with carbon dioxide activation: (**a**) Micropore volume, and (**b**) micropore area.

The activation with carbon dioxide under the increased pressure seems to be resulting in an increased average pore diameter and the total pore volume, as well as a slightly lower specific surface area when compared to values observed previously for lignite-derived materials with pressure, as the only activation agent in carbonization step at 1000 ◦C [25] or carbon dioxide as the only activation agent at 900 ◦C [8]. These effects are clearly also related to the temperature and composition of a parent material since in the studies with CO2-only activation at 750–900 ◦C [5] similar values, of 0.336–0.365 cm3/g to the reported in this study were observed, but for a lignite of a considerable higher volatiles content (48%) and significantly lower ash content (5%), which made it more suitable as a porous material precursor. This proves that the increased pressure and high temperature may be successfully applied instead of partial CO2-oxidation for production of lignite-derived chars of well-developed surface area and relatively low average pore diameter.

The application of pressure in carbon dioxide activation process at 800 ◦C enabled production of chars of only slightly lower specific surface area than the ones received with demineralization and carbon dioxide activation step of lignite of a considerably lower ash content (8–9%) than in the study presented here, which implies that pressure may be also considered as an alternative to chemical activation [20]. The total pore volume reported for lignite chars produced with pressure and carbon dioxide activation within the study presented here doubled the values observed for lignite chars of the volatiles content as high as 50% at 800 ◦C, with chemical demineralization and carbon dioxide activation [11].

These results show that the combined pressure and carbon dioxide activation may be effectively applied in utilization of poor quality parent materials (low volatiles, high ash content) for porous materials development. They also prove that the combination of pressure and carbon dioxide activation results in the development of porous structure of lignite-derived materials of a comparable specific surface area and of a similar or higher total pore volume than observed for complex demineralization, chemical activation and carbon dioxide treatment of the respective parent materials.

As it can be seen from the SEM (define) images presented in Figure 5, the surface of char particles generated under 0.1 and 1 MPa was visibly smoother (Figure 5a,b) than of chars produced under the elevated pressure of 2 and 3 MPa (Figure 5c,d). However, even for the chars generated under 1 MPa a clear difference may be noticed, consisting in a less dense texture and a more complex structure with some cracks and roughness, when compared to chars developed under atmospheric pressure, resulting from volatiles release and lignite carbonization under carbon dioxide atmosphere. This is in line with the variations in the average pore diameter of chars as described above, as well as increased specific surface area and micropore area with pressure applied (Table 2). The chars developed under 2 and 3 MPa (Figure 5c,d, left) have a more pumice-like structure than a plate-like structure characteristic for chars developed under atmospheric pressure and 1 MPa (Figure 5a,b, left). The chars produced under 2 and 3 MPa (Figure 5c,d, right) showed also visibly more expanded cavities and larger cracks than chars generated under lower pressures. They are likely to be composed of a considerable amount of micropores and smaller mesopores of irregular shape as demonstrated also by the nitrogen isotherms shape and DFT data (Figure 2). This means that the elevated pressure resulted in an enhanced porosity of chars under the experimental conditions applied, though there may also have occurred some rearrangements in the porous structure resulting from thermal annealing observed previously for carbon dioxide treatment of bituminous coal chars at the temperature of 800–900 ◦C [34].

**Figure 5.** SEM images of lignite chars generated under carbon dioxide atmosphere at 800 ◦C and under the pressure of: (**a**) 0.1 MPa, (**b**) 1 MPa, (**c**) 2 MPa and (**d**) 3 MPa.

#### **4. Conclusions**

On the basis of the experimental study performed and presented within the paper the following conclusions may be drawn:


**Funding:** This research was funded by the Ministry of Science and Higher Education, Poland, grant number 10171019.

**Conflicts of Interest:** The author declares no conflict of interest.

#### **References**


© 2019 by the author. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## *Article* **CO Adsorption Performance of CuCl**/**Activated Carbon by Simultaneous Reduction–Dispersion of Mixed Cu(II) Salts**

#### **Cailong Xue, Wenming Hao, Wenping Cheng, Jinghong Ma \* and Ruifeng Li**

College of Chemistry and Chemical Engineering, Taiyuan University of Technology, Taiyuan 030024, China; xuecailong35@163.com (C.X.); haowenming@tyut.edu.cn (W.H.); chengwenping@tyut.edu.cn (W.C.); rfli@tyut.edu.cn (R.L.)

**\*** Correspondence: majinghong@tyut.edu.cn; Tel.: +86-351-6111353

Received: 15 April 2019; Accepted: 14 May 2019; Published: 16 May 2019

**Abstract:** CO is a toxic gas discharged as a byproduct in tail gases from different industrial flue gases, which needs to be taken care of urgently. In this study, a CuCl/AC adsorbent was made by a facile route of physically mixing CuCl2 and Cu(HCOO)2 powder with activated carbon (AC), followed by heating at 533 K under vacuum. The samples were characterized by X-ray powder diffraction (XRD), inductively coupled plasma optical emission spectrometry (ICP-OES), N2 adsorption/desorption, and scanning electron microscopy (SEM). It was shown that Cu(II) can be completely reduced to Cu(I), and the monolayer dispersion threshold of CuCl on AC support is 4 mmol·g−<sup>1</sup> AC. The adsorption isotherms of CO, CO2, CH4, and N2 on CuCl/AC adsorbents were measured by the volumetric method, and the CO/CO2, CO/CH4, and CO/N2 selectivities of the adsorbents were predicted using ideal adsorbed solution theory (IAST). The obtained adsorbent displayed a high CO adsorption capacity, high CO/N2, CO/CH4, and CO/CO2 selectivities, excellent ad/desorption cycle performance, rapid adsorption rate, and appropriate isosteric heat of adsorption, which made it a promising adsorbent for CO separation and purification.

**Keywords:** CuCl/AC adsorbent; CO adsorption; monolayer dispersion; isosteric heat; adsorption isotherms

#### **1. Introduction**

With the rapid development of C−1 chemistry recently in the chemical industry, carbon monoxide (CO) as a significant resource has been widely applied to prepare a large variety of chemical products, such as formic acid, acetic acid, oxalic acid ester, carbonic acid two methyl ester, anhydride, etc. [1,2]. Most of these preparations need high-purity CO. The main methods of producing CO are the steam reforming of natural gas and coal gasification [3]. In addition, a significant amount of CO is discharged as a byproduct in tail gases from different industrial flue gases including carbon black tail gas, silicon carbide furnace gas, yellow phosphorus tail gas, coke oven gas, blast furnace gas, etc. [4–6]. From both processes, the obtained CO is mixed with N2, H2, CH4, CO2, and vapor. CO is toxic to humans because it combines with hemoglobin in the blood to form carboxy-hemoglobin hindering the transportation and release of oxygen in the blood, which leads to death [7]. Moreover, even a trace amount of CO can poison the noble catalysts, such as the proton-exchange membrane fuel cells, which restrict the CO content below 0.2 ppm to protect the platinum electrocatalyst [8,9]. Thus, the separation and purification of CO from different gas mixtures have significance both industrially and environmentally.

Among the proven technologies for CO separation and purification, adsorption processes, such as pressure swing adsorption (PSA) and temperature swing adsorption (TSA) have the advantages of convenient operation, low energy consumption, low operating cost, etc., and have been widely used in CO separation [10–14]. Adsorbent plays a crucial role in the adsorption based gas separation process, it has been found that many porous materials, such as activated carbons [15,16], zeolites [17,18], and metal-organic frameworks (MOFs) [19,20], have adsorption capacities to a certain extent. However, it is difficult to separate and purify CO from gas mixtures by using these materials directly since their adsorption capacity and selectivities are low. Cu(I) adsorbents for CO separation have received extensive attention for their high CO adsorption capacity and high selectivity, since CO molecules can form a π-complexation bond with Cu(I) ions on the adsorbent, which are stronger than the interaction caused by van der Waals forces [21–26]. More importantly, π-complexation bonds are still weak enough to be broken by normal engineering operations, such as increasing temperature or reducing pressure, and the adsorbed CO can be easily desorbed, which makes it a suitable adsorbent in PSA and TSA systems [23]. Two approaches are used for making Cu(I) adsorbents. In the first process, Cu(I) adsorbents are prepared by impregnating Cu(II) salts into a porous support including zeolites, activated carbons (ACs) and MOFs, etc. [20,25,27], and then reducing Cu(II) to Cu(I) using reducing gases, such as H2 or CO. However, it is difficult to control the reduction degree, and Cu(II) is easily over reduced to Cu. In the second process, Cu(I) adsorbents are prepared by direct dispersion and impregnation of CuCl. Hirai et al. [28,29] and Tamon et al. [30] obtained Cu(I)/AC adsorbents by using dispersing reagents, such as concentrated hydrochloric acid or organic solvents, to disperse CuCl onto the AC surfaces, and then drying at 403 K in N2. Xie et al. [31] prepared CuCl/zeolite adsorbents by dispersing CuCl powder spontaneously onto the surfaces of zeolites at 623 K in an inert atmosphere, which displayed high adsorption capacity and selectivity for CO. When using CuCl as a starting material, the adsorbent preparation has to be carefully performed in a dry inert atmosphere, to prohibit the oxidation and hydrolysis of Cu(I). In our previous work, we successfully obtained Cu(I) π-complexation adsorbents with an aqueous solution of equimolar CuCl2 and Cu(HCOO)2 as starting materials by the traditional impregnation method followed by activating at the temperature of 583 K [32].

Herein, the purpose of this work is to develop CO adsorbent using a solid-state auto reduction–dispersion method with CuCl2 and Cu(HCOO)2 as the initial material. Then, X-ray powder diffraction (XRD), inductively coupled plasma optical emission spectrometry (ICP-OES), N2 adsorption/desorption and scanning electron microscopy (SEM) were employed to characterize the samples. Pure component CO, CO2, N2, and CH4 adsorption isotherms on the adsorbents were measured in a volumetric method. The CO/CO2, CO/CH4, and CO/N2 selectivities of the adsorbents were predicted by using ideal adsorbed solution theory (IAST). The adsorption isotherms were fitted with the Langmuir–Freundlich model, and the corresponding heats of adsorption were calculated. The cyclic CO adsorption on adsorbent was performed to evaluate its repeated availability during the adsorption and desorption cycles. Furthermore, the CO adsorption rate on adsorbent was discussed and reported.

#### **2. Materials and Methods**

#### *2.1. Materials*

Copper formate tetrahydrate (Cu(HCOO)2·4H2O, 98%) and cupric chloride dihydrate (CuCl2·2H2O, 99%) were purchased from Alfa Aesar Chemical Co. Ltd.(Ward Hill, MA, USA). Activated carbon (AC) was purchased from Chengde Jingda Activated Carbon Manufacturing Co. Ltd. (Chengde, China).

#### *2.2. Preparation of CuCl*/*AC Adsorbents*

CuCl/AC adsorbents were synthesized following two steps. First, the AC was physically mixed with CuCl2 and Cu(HCOO)2 powder to obtain CuCl/AC adsorbent precursors. Then, the obtained precursors were dried at 373 K and activated in a tube furnace at 533 K for 4 h under vacuum. The obtained precursors and CuCl/AC adsorbents were marked as Cu(II)-x/AC and Cu(I)-x/AC

(x <sup>=</sup> 2, 3, 4, 5, 6), in which the loading of copper is 2, 3, 4, 5, and 6 mmol·g−<sup>1</sup> AC, respectively. The as-synthesized CuCl/AC adsorbents were stored in vacuum dry storage in a desiccator.

#### *2.3. Adsorbent Characterization*

Powder X-ray diffraction (XRD) patterns of the samples were recorded by a Shimadzu LabX XRD-6000 system (Kyoto, Japan) in the 2θ range of 5 to 35◦ using CuKα1 (λ=1.54056 Å) radiation operated at 40 kV and 30 mA. The pore volume and surface area of the samples were calculated from N2 adsorption/desorption isotherms measured on a surface area and pore size analyzer (QUADRASORB SI, Quantachrome Inc., Boynton Beach, FL, USA) after activating the samples at 393 K for 4 h under vacuum. The specific surface areas (SBET) were determined using the BET (Brunauer–Emmett–Teller) method under relative pressure in the range of 0.01 to 0.20. The adsorbed amount of N2 at p/p0 = 0.98 was employed to calculate the total pore volume (VTotal). Scanning electron microscope (SEM, Hitachi S4800, Hitachi Ltd., Tokyo Japan) was used to observe the samples' morphology. Cu contents were measured by inductively coupled plasma optical emission spectrometry (ICP-OES, Thermo-ICAP6300, Thermo Fisher Scientific Co., Ltd., Waltham, MA, USA).

#### *2.4. Adsorption Measurements*

Before the adsorption measurements, the samples were degassed under vacuum while heating up to 393 K for 4 h. The CO, CO2, CH4, and N2 adsorption isotherms were measured at the temperature required using a static volumetric apparatus (NOVE1000e, Quantachrome Inc., Boynton Beach, FL, USA). During the adsorption measurements, the temperature was maintained by circulating ethanediol-water from a bath with setting temperature. The adsorption capacity was determined from the adsorption isotherm measured at 298 K. Ultrahigh purity grade CO (99.99%), CO2 (99.999%), CH4 (99.99%), and N2 (99.999%) were used without any purification.

#### **3. Results and Discussion**

#### *3.1. Characterization of Samples*

The XRD patterns of the copper loaded AC samples before and after activation are presented in Figure 1. Before activation, the diffraction peaks of Cu(HCOO)2 and CuCl2 [33,34] can be observed in the Cu(II)-x/AC samples, and the reflection intensities increased with the increase of copper loading. After activation, the diffraction peaks of CuCl2 and Cu(HCOO)2 disappeared, and the Cu(I)-5/AC sample displayed only a relatively weak peak (2θ = 28.5◦) of CuCl [35], suggesting that Cu(HCOO)2 and CuCl2 were transformed into CuCl after activation. Meanwhile, the absence of CuCl diffraction peak on Cu(I)-2/AC, Cu(I)-3/AC, and Cu(I)-4/AC samples might be due to the well dispersion of CuCl on the AC surface beyond the detection limit of XRD [36]. By further increasing the copper loading to 5 and 6 mmol·g−1, the appearance of the characteristic peak of CuCl implies that the crystal size of CuCl on the AC surface increased with the increase of CuCl loading, which was able to be detected by XRD with the CuCl loading higher than 5 mmol·g<sup>−</sup>1.

Figure 2a,b show the representative SEM images of the copper loaded AC samples before and after activation. It can be clearly observed that the particles of copper species present on the AC surface for the Cu(II)-4/AC. However, the particles on the AC surface disappeared after activation, which implies that the activation process contributes to the CuCl dispersion on the AC surface. Figure 2c,d show selected-area and the element mapping analyses of Cu(I)-4/AC. It revealed that the copper particles are uniformly dispersed on the AC surface. The good dispersion of CuCl on Cu(I)-4/AC observed by SEM agreed well with the XRD results in Figure 1.

**Figure 1.** X-ray powder diffraction (XRD) patterns of activated carbon (AC), Cu (II)-x/AC (**a**) and Cu(I)-x/AC (**b**).

**Figure 2.** Scanning electron microscopy (SEM) images of Cu(II)-4/AC (**a**), Cu(I)-4/AC (**b**), and selected-area element mapping analyses of Cu(I)-4/AC (**c**,**d**).

Figure 3 shows the N2 adsorption/desorption isotherms of the Cu(I)-x/AC samples at 77 K. The N2 adsorption gradually decreased with increasing CuCl loading. Table 1 lists the textural parameters of AC and Cu(I)-x/AC. It can be observed that the total pore volume (VTotal) and BET surface area (SBET) gradually decreased with the increase of CuCl loading, indicating that CuCl had been loaded into the pores of the parent AC. As the CuCl loading increased, more and more surface within the pores were occupied by CuCl, which may result in a further decrease of VTotal and SBET. As CuCl was well dispersed on the surface of AC when the CuCl loading was below 4 mmol/g, the average pore sizes decreased with the increase of CuCl loading. Smaller pores were filled with the further increase of the CuCl loading. Therefore, the increased average pore size resulted from the percentage increase of the available larger pores in AC, which is similar to the observation by Ramli et al. [37].

**Figure 3.** N2 adsorption/desorption isotherms of Cu(I)-x/AC.

**Table 1.** The parameters of pore structure, loading and utilization coefficient of CuCl for Cu(I)-x/activated carbon (AC).


<sup>a</sup> average pore diameter of adsorbent; <sup>b</sup> calculated from starting material; <sup>c</sup> obtained by ICP-OES.

It can be observed from Figure 4 that the CO adsorption capacity of CuCl/AC increased with CuCl loading in the range of 0 to 4 mmol·g<sup>−</sup>1. The maximum value of adsorption capacity was 45.4 cm3·g<sup>−</sup>1. With the continuous increase of the CuCl loading, the CO adsorption capacity of CuCl/AC with the copper loading of 5 mmol·g−<sup>1</sup> was almost the same as that with 4 mmol·g−1. The decrease of CO adsorption occurred with further increasing the copper loading to 6 mmol·g−1. This phenomenon can be ascribed to the following reason. The more copper loaded, the more active sites of CuCl/AC adsorbents present, which would enhance CO adsorption. However, the increase of copper loading also resulted in a decrease of surface area for CuCl/AC adsorbents, as shown in Table 1. As a result, the increase of the adsorbed amount from the increased active sites and the decrease of adsorbed amount from the decrease of surface area were in a dynamic balance in the copper loading range of 4 to 5 mmol·g<sup>−</sup>1. When the copper loading reached 6 mmol·g<sup>−</sup>1, on the one hand, the amount of adsorbed CO decreased because the decrease of surface area; on the other hand, the Cu(I) started to agglomerate on AC surface with considerable copper loading, resulting in the low utilization of active sites.

**Figure 4.** CO adsorption isotherms on AC and Cu(I)-x/AC at 298 K.

In addition, Table 1 lists the measured Cu contents with ICP-OES, which are closely approximate to the values in the raw material. The utilization coefficient of surface CuCl is described from the equation:

$$\eta = \frac{q\_{CO}}{n\_{\text{CuCl}}} \times 100\%$$

where η is the utilization coefficient of CuCl, *qCO* is the actual CO adsorption capacity at 298 K and 100 kPa, *nCu*Cl is the mole of CuCl per gram CuCl/AC adsorbent. According to this equation, the utilization coefficients were calculated and presented in Table 1. It can be seen that the utilization coefficient decreased with the increasing of CuCl loading, which means that the high CuCl loading on AC could not guarantee high utilization of CuCl, since not all Cu(I) species can be utilized.

#### *3.2. Adsorption Selectivities of CO to CO2, CH4, and N2*

Figure 5a gives the adsorption isotherms of pure CO, CO2, CH4, and N2 on Cu(I)-4/AC in the pressure range of 0 to 100 kPa. The adsorption of CO2, CH4, and N2 on Cu(I)-4/AC increased almost linearly with pressure, while the adsorption isotherm of CO on Cu(I)-4/AC presented a type-I isotherm [38], that is the CO adsorption increased sharply with pressure at a low pressure range, implying the adsorption of relatively strong CO-Cu(I) π-complexation, which is propitious to separate CO from CO/CO2/CH4/N2 mixed gas.

**Figure 5.** Adsorption isotherms of CO, CO2, CH4, and N2 on Cu(I)-4/AC and Langmuir–Freundlich (L-F) fitting lines (**a**), and ideal adsorbed solution theory (IAST)-predicted adsorption selectivities (**b**).

g The Langmuir–Freundlich (L-F) model and IAST were employed together to calculate the CO/CO2, CO/CH4, and CO/N2 selectivities with the equimolar CO/CO2, CO/CH4, and CO/N2 mixture. The L-F model can be expressed as

$$q = q\_m \frac{bp^{1/n}}{1 + bp^{1/n}}$$

where *q* is the adsorbed amount, *p* is the pressure and *qm* is the saturation adsorbed amount, *b* is the adsorption affinity and *n* is the corresponding deviation from the Langmuir isotherm.

First, the adsorption isotherm of pure CO, CO2, CH4, and N2 were fitted by the L-F model [38]. After that, the CO/CO2, CO/CH4, and CO/N2 selectivities were predicted by IAST theory [39,40]. Finally, the relevant selectivities curves along with the increase of pressure were obtained. Figure 5b shows that the CO/CO2, CO/CH4, and CO/N2 selectivities decrease gradually with increasing pressure. Nevertheless, the CO/CO2, CO/CH4, and CO/N2 selectivities on Cu(I)-4/AC were still up to 2.6, 8.0, and 34.3 at 100 kPa, respectively, which suggests that it has the potential for the effective separation of CO from the gas mixtures. Table 2 lists the benchmark materials for CO adsorption. Cu(I) adsorbents have higher adsorption capacity than the conventional porous adsorbent. The Cu(I)-4/AC adsorbent prepared in this study has relatively high CO/CO2 selectivity among the selected adsorbents.


**Table 2.** Comparisons with adsorbents in the literature.

<sup>a</sup> the temperature of adsorption measurements; <sup>b</sup> adsorbed amount of gases at 100 kPa.

#### *3.3. Isosteric Heat of Adsorption*

Isosteric heat of adsorption is a significant thermodynamic parameter to characterize the interaction between the adsorbate and the adsorbent and to design a gas adsorption separation process, which can be calculated by Clausius–Clapeyron equation [41] as

$$\left[\frac{\partial \ln P}{\partial (1/T)}\right]\_{\overline{q}} = -\frac{\Delta H\_s}{RT}$$

where *P* is the pressure, *R* is the ideal gas constant, *T* is the experimental temperature, *q* is the adsorption amount, and Δ*Hs* is the isosteric heat of adsorption. In this work, the experimental isotherms and the L-F model predicted isotherms of CO at different temperatures of 273 K, 293 K, and 298 K (as shown in Figure 6) were used to calculate Δ*Hs* of CO adsorption on AC and Cu(I)-4/AC. The conventional Langmuir–Freundlich (L-F) adsorption model correlated the experimental results and the fitting parameters, which are listed in Table 3. The experimental data fit well with L-F model, as can be seen by the high values of R<sup>2</sup> (the coefficient of the experimental data and the fitting data). Δ*Hs* can be derived from the slopes of the plots of *lnP* versus 1/*T* at given adsorption amounts, as shown in Figure 7. It was shown that the isosteric heats of CO adsorption on Cu(I)-4/AC are remarkably much higher than those on AC. The result indicates that the π-complexation interaction between CO and Cu(I) is stronger than the van der Waals interaction of CO with the parent adsorbent. Usually, <sup>Δ</sup>*Hs* is <sup>&</sup>lt;20 kJ·mol−<sup>1</sup> for common physical adsorption and >80 kJ·mol−<sup>1</sup> for chemical adsorption [42]. The values of <sup>Δ</sup>*Hs* on Cu(I)-4/AC maintained around 50 kJ·mol−<sup>1</sup> in the whole pressure range, suggesting that the strength of complex adsorption is between physisorption and chemisorption. Such isosteric heat is not only propitious to adsorb CO, but also liable to desorb CO with a normal engineering operations (evidence as shown in Figure 9).

**Figure 6.** CO adsorption isotherms of AC and Cu(I)-4/AC at 273 K, 288 K, and 298 K.

ΔΗ


**Table 3.** Langmuir–Freundlich (L-F) fitting parameters of CO isotherms on AC and Cu(I)-4/AC.

**Figure 7.** Isosteric heats of AC and Cu(I)-4/AC as the function of the adsorbed amount of CO.

#### *3.4. Adsorption Kinetics of CO*

In addition, it is also crucial that the CO adsorption rate need to be quite rapid for potential applications of CuCl/AC adsorbent in the adsorption-driven separation of CO from gas mixtures containing CO, CO2, CH4, and N2. Typically, the requirement of the adsorption process in industrial applications is shorter than 1 min [43]. Here, we studied the time-dependent adsorption of CO on Cu(I)-4/AC adsorbent by releasing a small amount of CO and studying the adsorbed amount as a function of time as shown in Figure 8. Cu(I)-4/AC showed a relatively rapid adsorption rate, which reached 96% of the CO capacity within 25 s. The rapid CO adsorption rate suggests that Cu(I)-4/AC can meet the requirements for industrial application of adsorbent to separate CO from gas mixtures containing CO, CO2, CH4, and N2 in a PSA process.

**Figure 8.** Uptake kinetics for Cu(I)-4/AC at 298 K.

#### *3.5. Cycle Adsorption of CO on Cu(I)*/*AC*

In the actual processes of gas separation, an ideal adsorbent not only needs to have high adsorption capacity and high selectivity but also needs to exhibit a stable cyclic adsorption performance in long-term adsorption/desorption cyclical operation. The pure CO cyclical adsorption/desorption isotherm at 298 K was evaluated for six times (The degassing between each cycle was carried out at 353 K under vacuum). As shown in Figure 9, the maximum amount of CO adsorption was reduced by about 1.3% after six cycles of adsorption and desorption, suggesting that the CO adsorption process using CuCl/AC adsorbent is stable under the investigated conditions. Its stable adsorption behavior indicates that the CuCl/AC has broad application prospects in selective adsorption of CO. It must be noted that the parent gases for the separation of CO must be pretreated to remove moisture in industrial processes, since the Cu(I) in the CuCl/AC can be oxidized to Cu(II) in the form of copper chloride once in contact with water vapor and O2 (especially under light condition [44,45]), which then cannot form complexation with CO.

**Figure 9.** Adsorption and desorption cycles for CO at 298 K on Cu(I)-4/AC (the degassing between each cycle was carried out at 353 K under vacuum).

#### **4. Conclusions**

CuCl/AC adsorbents for the separation of CO have been successfully obtained using CuCl2 and Cu(HCOO)2 as the initial material by a solid-state auto dispersion method. CuCl2 and Cu(HCOO)2 can be transformed into highly dispersed CuCl with activation at 533 K under vacuum atmosphere. The CO adsorption capacity increased with transformed CuCl loading until 4 mmol·g−<sup>1</sup> and then decreased afterward. The CO adsorption capacity of Cu(I)-4/AC achieved 45.4 cm3·g<sup>−</sup>1, and the CO/CO2, CO/CH4, and CO/N2 selectivities were up to 2.6, 8.0, and 34.3 at 100 kPa, respectively. In addition, the isosteric heat of adsorption on Cu(I)-4/AC was about 50 kJ·mol<sup>−</sup>1. The CO adsorption capacity almost remains constant during six times cyclical adsorption and rapid adsorption kinetics at the adsorption process. Those excellent properties of Cu(I)-4/AC adsorbent would make it a promising adsorbent for CO separation and purification.

**Author Contributions:** Conceptualization, J.M.; methodology, C.X. and J.M.; formal analysis, C.X., W.H. and W.C.; resources, R.L.; data curation, W.C.; writing—original draft preparation, C.X.; writing—review and editing, W.H.; supervision, J.M.; funding acquisition, J.M. and R.L.

**Funding:** This research was funded by Key Scientific and Technological Project of coal fund of Shanxi province (No.FT201402-03) and Shanxi Provincial Key Innovative Research Team in Science and Technology (No.2014131006).

**Acknowledgments:** This work was supported by Key Scientific and Technological Project of coal fund of Shanxi province (No.FT201402-03) and Shanxi Provincial Key Innovative Research Team in Science and Technology (No.2014131006).

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## *Article* **E**ff**ect of High Pressure on the Reducibility and Dispersion of the Active Phase of Fischer–Tropsch Catalysts**

#### **Simón Yunes 1, Miguel Ángel Vicente 2, Sophia A. Korili <sup>3</sup> and Antonio Gil 3,\***


Received: 28 May 2019; Accepted: 10 June 2019; Published: 13 June 2019

**Abstract:** The effect of high pressure on the reducibility and dispersion of oxides of Co and Fe supported on γ-Al2O3, SiO2, and TiO2 has been studied. The catalysts, having a nominal metal content of 10 wt.%, were prepared by incipient wetness impregnation of previously calcined supports. After drying at 60 ◦C for 6 h and calcination at 500 ◦C for 4 h, the catalysts were reduced by hydrogen at two pressures, 1 and 25 bar. The metal reduction was studied by temperature-programmed reduction up to 750 ◦C at the two pressures, and the metal dispersion was measured by CO chemisorption at 25 ◦C, obtaining values between 1% and 8%. The physicochemical characterization of these materials was completed by means of chemical analysis, X-ray diffraction, N2 adsorption-desorption at −196 ◦C and scanning electron microscopy. The high pressure lowered the reduction temperature of the metal oxides, improving their reducibility and dispersion. The metal reducibility increased from 42%, in the case of Fe/Al2O3 (1 bar), to 100%, in the case of Fe/TiO2 (25 bar).

**Keywords:** Fischer–Tropsch; supported iron oxide; supported cobalt oxide; reducibility; dispersion

#### **1. Introduction**

Supported cobalt and iron catalysts have been extensively studied in the Fischer–Tropsch reaction for the conversion of synthesis gas obtained from natural gas due to their high activity, high selectivity to long-chain paraffins, and low activity in the formation of water [1]. Despite this huge amount of research work, there is still no agreement in the scientific community about the active phases in the reaction between CO and H2. Thus, for example, in the case of Fe catalysts there is a high degree of consensus about the fact that its carbides, and not metallic Fe, are the active phase. Likewise, the two factors that can control the activity of the catalysts are the degree of reduction of the metallic precursor as well as the shape and size of the metal particles formed, characteristics that are related through the dispersion, the distribution of the particles on the support [2].

In general, the type and structure of the support affect the dispersion, particle size, and reducibility, and as a consequence the activity of the supported metal catalysts [2,3]. The acidity of the supports, as well as the presence of dopants, are other factors that affect the reducibility and dispersion of the metallic phase [2,4]. Other influencing preparation variables are the metal precursor, the solvent, the metal content, the method of preparation, and the pretreatments before the catalytic tests. For example, in the case of cobalt, CoO can react with the supports both during the synthesis and during the reduction treatment, resulting in various mixed compounds like CoAl2O4, Co2SiO4

or CoTiO3 [5–10]. These mixed compounds require far too high reduction temperatures to reduce the metal. Under this way, cobalt metal can be lost by sublimation and increase the particle sizes by sintering, resulting in catalysts with lower performance. The reduction temperatures can be decreased by increasing the pressure, also allowing less cobalt to be lost, and decreasing sintering. Other possible action is the promotion with noble metals (Pd, Pt, Re, and Ru have been used) to enhance the reducibility of the oxides, to improve the metal dispersion, to reduce the catalyst deactivation, etc. [11].

The reaction conditions, such as temperatures up to 350 ◦C, pressures up to 55 bar, and the presence of fluids under supercritical conditions, also control the Fischer–Tropsch synthesis process [12]. It is highly recommended to characterize the catalysts under these conditions to have a more realistic view of the effect of the different variables on the properties of the solids.

The aim of this work is to evaluate the effect of pressure on the reducibility and dispersion of Co and Fe catalysts prepared from various supports. Some examples of this effect have been published previously but only for cobalt catalysts supported on carbon nanofibers [13,14]. In this work, a comparative study also considering Fe as active phase and other catalytic supports is presented.

#### **2. Materials and Methods**

#### *2.1. Preparation of the Catalysts*

The Co and Fe catalysts, with a nominal metal content of 10 wt.%, were prepared by incipient wetness impregnation of the supports. The salts used were Co(NO3)2•6H2O (Panreac, Castellar del Vallés, Barcelona, Spain) and Fe(NO3)3•9H2O (Riedel-de Haën-Honeywell, Madrid, Spain), respectively. The commercial supports used were γ-Al2O3 (Spheralite 505, Procatalyse, Rueil Malmaison, France), SiO2 (Aerolyst 350, Degussa, Frankfurt, Germany), and TiO2 (Aeroxide TiO2 P25, Degussa). Prior to their use, all of them were calcined in air at 500 ◦C for 4h. After impregnation, all the catalysts were dried at 60 ◦C for 6 h and calcined again at 500 ◦C for 4 h.

#### *2.2. Characterization Techniques*

The metal content was determined by inductively coupled plasma-atomic emission spectroscopy (ICP-AES) using Varian Vista-MPX equipment with radial vision (Varian, Palo Alto, CA, USA). The crystalline structure of the catalysts was analyzed by X-ray diffraction (XRD) in a Siemens D5000 diffractometer (Siemens, Munich, Germany). The scanning electron microscopy analyses were carried out at CLPU (Salamanca, Spain) using a Carl Zeiss SEM EVO HD25 (Zeiss Microscopy, Jena, Germany). The textural characterization of the supports and the catalysts synthesized was carried out by adsorption-desorption of N2 (Air Liquide, 99.999%) at −196 ◦C using a static volumetric method in Micromeritics model ASAP 2010 equipment (Norcross, GA, USA).

The temperature-programmed reduction (TPR) and the pulse chemisorption experiments were carried out in an automated and controlled Effi Microactivity-reference PID Eng & Tech, denoted as Micro Catalyst Characterization and Testing Center, MCCTC, Madrid, Spain (see Figure 1). This equipment is mainly used to carry out catalytic reactions of any kind at a pressure between atmospheric and 200 bar. The configuration of the system allows to characterize the catalyst in situ, especially when carrying out catalytic tests in which the deactivation of the catalyst takes place. Under these conditions, it is necessary to characterize the catalyst without removing it from the reactor, avoiding its contact with the atmosphere.

The equipment consists of a hot box where all the pipes, valves, furnace, and reactor are arranged, normally kept warm at a temperature up to 200 ◦C to prevent the condensation of vapors. The equipment is connected in line to an MKS Instruments mass spectrometer, CirrusTM single-quadrupole model, which constitutes the system's gas analysis system (Andover, MA, USA).

The system has a tubular steel reactor of 32 mm in length and 9 mm in internal diameter, with the catalytic bed placed inside it on a porous plate. The catalytic bed is fixed between two layers of quartz wool. The reactor is located inside a longitudinal opening cylindrical furnace that allows the thermal regulation of the system at a maximum working temperature of 1100 ◦C. A type K thermocouple placed in contact with the catalytic bed, which is part of the temperature measurement and control system, is placed inside the reactor. The flow of gases enters the reactor in a descending way and the reaction products exit from the bottom, being then introduced into the analysis system. The flow rate of all the gases entering the reaction system is measured through mass flow controllers. The gases are preheated in the hot box of the system and directed towards a six-way valve that pneumatically directs the path of the current directly to the reactor tube or to the gas outlet. In this valve, a loop (0.521 cm3 NTP) was adapted, which allows to dose an active gas such as CO in order to determine the dispersion of the active phase of any type of catalyst.

**Figure 1.** Flow-diagram of the Micro Catalyst Characterization and Testing Center (MCCTC) connected to a mass spectrometer.

For the TPR experiments, 0.3 g of catalyst was pretreated by heating from room temperature to 150 ◦C at a heating rate of 10 ◦C/min, and maintained at 150 ◦C for 2 h, all the process under a flow of N2 (99.999%) of 50 cm3/min. The samples were then reduced using a 2.5% H2/Ar (99.999%) mixture with a flow of 400 cm3/min, from room temperature to 750 ◦C, with a heating rate of 10 ◦C/min, and a pressure of 1 and 25 bar. This temperature was maintained until the baseline returned to zero, indicating a complete reduction. Two commercial pure oxides, Co3O4 (Sigma-Aldrich España, Madrid, Spain, 99%) and Fe2O3 (Sigma-Aldrich, 99.99%), were used as references for calibrating the reduction profiles and quantifying the amount of H2 consumed by the catalysts studied.

The dispersion of the active phase was determined by repeating the reduction treatment, both at 1 and 25 bar pressure, but finishing it at 500 ◦C and maintaining this temperature for 4 h to ensure that the baseline was recovered. Next, the reducing mixture was replaced by an inert gas at the same temperature to ensure the elimination of all traces of H2. In the next step, the temperature was lowered to room temperature, and the dosing was carried out using a calibrated loop of 0.521 cm3 of CO (99.999%) until the saturation of the signal was observed, almost three identical peaks. With the volume of chemisorbed CO, the metallic content and the stoichiometric ratio CO/metal, 1:1 in this case, the percent dispersion of the metals was determined using the following Equation (1),

$$D(\%) = \frac{V\_{ads} \cdot f\_{\text{a}} \cdot \mathcal{W}\_{\text{a}}}{V\_{mol} \cdot \mathcal{M}(\%)}\tag{1}$$

where *D*(%) is the metal dispersion, *Vads* is the total volume of CO chemisorbed, *fa* is the stoichiometry factor, *Wa* is the atomic mass of the active metal, *Vmol* is the molar volume of the CO and *M*(%) is the mass percent of metal present in the catalysts.

In the case of the chemisorption temperature, the value was selected to avoid the contribution to the physical adsorption of CO on the supports. For the stoichiometric relationship, it was assumed that each molecule of CO interacted with one Co or Fe atom on the surface [15,16].

#### **3. Results and Discussion**

The XRD results indicated (see Figure 2) the formation of the cobalt spinel Co3O4 in the case of the cobalt catalysts and the hematite phase Fe2O3 in the case of iron catalysts. No mixed phases involving the supports and the active phases were detected, although in the case of alumina, Al(III) cations from the support should isomorphically incorporate to both active phases, and in fact reducibility studies strongly suggested this possibility (vide infra).

**Figure 2.** XRD patterns of the supported cobalt (**left**) and iron (**right**) oxide catalysts.

Selected micrographs of the cobalt catalysts are included in Figure 3. The images confirmed that the structures of the particles of the supports were not significantly modified by the impregnation with the metallic solutions and the subsequent calcination.

**Figure 3.** SEM images of the supported cobalt (**left**) and iron (**right**) oxide catalysts. Top: Alumina-supported catalysts; middle: Silica-supported catalysts; bottom: Titania-supported catalysts.

The adsorption isotherms were of type II according to the IUPAC classification [17] for γ-Al2O3 and TiO2 supports and type IV in the case of SiO2 support. The presence of metallic oxides did not change the shape of the adsorption isotherms or the shape of the hysteresis cycles that originated in the desorption process. This result may indicate that the metal oxides were well dispersed on the surface of the supports. The textural properties, specific surface area, and pore volume of the supports and of the catalysts are included in Table 1, in which the metal content of the catalysts is also included.


**Table 1.** Textural properties derived from N2 adsorption at −196 ◦C and metal content by inductively coupled plasma-atomic emission spectroscopy ICP-AES.

\* Total pore volume, calculated from N2 adsorption at p/p<sup>o</sup> = 0.98.

The TPR profiles corresponding to the series of Co catalysts are included in Figure 4. In all the cases, a reduction peak can be observed at 350–465 ◦C, which coincided with the maximum peak of reduction of pure oxide Co3O4 to Co0. The reduction shoulder observed at lower temperature can be related to the reduction of Co3O4 to metallic Co in two steps (Co3<sup>+</sup> <sup>→</sup> Co2<sup>+</sup> <sup>→</sup> Co0), as has been discussed by Arnoldy and Moulijn [18]. These results confirmed the presence of Co3O4. In the case of the catalyst with Al2O3 as support, the higher reduction temperatures may be due to a greater oxide-support interaction. Al(III) can isomorphically incorporate to the spinel phase. The peak of reduction centered at higher temperature (600–1000 ◦C), which was observed in the Co/Al2O3 catalyst, suggested the reduction of Co2<sup>+</sup> species present in the form of cobalt spinels.

**Figure 4.** Reduction profiles corresponding to the supported cobalt (**left**) and iron (**right**) oxide catalysts.

In two previous works, Jacobs et al. [19] and Borg et al. [20] reported that the metal-support interactions affected the reduction of cobalt species and the strength of such interactions for various supports decreased in the order Al2O3 > TiO2 > SiO2. In the case of Al2O3 and as a result of the metal-support interactions that arose from the diffusion of cobalt ions into alumina lattice sites of octahedral or tetrahedral geometry, the formation of CoAl2O4 was proposed, and as a consequence the reducibility of the cobalt was hindered. TiO2 is known to exhibit the strong metal-support interaction (SMSI) effect, where during the reduction of the cobalt oxide, the partial reduction of the support also takes place, encapsulating or decorating the cobalt particles [21]. The nature of hydroxyl groups and their concentration and distribution on the silica surface play the main role in the dispersion of cobalt particles. In this type of support, the formation of Co2SiO4 has also been reported.

The TPR profiles corresponding to the series of Fe catalysts are also included in Figure 4. The reduction of Fe2O3 was related to a first peak at 400–460 ◦C (3Fe2O3 → 2Fe3O4) and a second broader peak located at high temperatures that reflected the reduction to metallic iron (2Fe3O4→6Fe). The shoulder at 630 ◦C indicated an intermediate reduction state (2Fe3O4 → 6FeO → 6Fe).

The volumes of H2 consumed by weight of metal oxide, the degree of reducibility of the catalysts, and the dispersion of the metal phases for the catalysts studied are included in Table 2. The results are given for the two working pressures, 1 and 25 bar.


**Table 2.** Results of H2 consumption, metal reducibility and dispersion.

The low values of metal dispersion can be related to the high metal contents of the synthesized catalysts. It was also indicative of the large size of the metal particles. For the two metals considered, the catalysts synthesized using TiO2 as support had greater degrees of oxide reduction. This result can be attributed to the weak interaction existing between the metallic oxide and the support and characteristic of the SMSI effect. The situation was very different for the catalysts supported on γ-Al2O3, which showed lower degrees of oxide reduction, behavior that can be explained through strong oxide-support interactions, even with the formation of spinels involving cations from the support, that is, Fe(II) or Co(II) as divalent cations, and Fe(III) or Co(III) from the precursors and Al(III) from the support as trivalent cations, this effect seeming to be more significant in the case of Co-samples [5,18]. These effects decreased with pressure, as the reducibility of the metal oxides and the metal dispersion increased.

#### **4. Conclusions**

The effects of the high pressure of on the reducibility of catalysts based on cobalt and iron oxides on three commercial supports, γ-Al2O3, SiO2, and TiO2, have been presented. Two effects have been observed, the high pressure lowered the reduction temperature, decreasing the sintering of the metal oxide particles, while the pressure improved the reducibility of the metal oxides to an almost total reduction value. These two effects gave rise to a greater dispersion of the active metal phase, which may result in an increase in the activity of the catalysts.

**Author Contributions:** All the authors conceived, designed, and performed the experiments, analyzed the data, and drafted the manuscript.

**Funding:** This research was jointly funded by the Spanish Ministry of Economy and Competitiveness (AEI/MINECO) and the European Regional Development Fund (ERDF), grant number MAT2016-78863-C2-R. A.G. thanks Santander Bank for funding through the Research Intensification program.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## *Article* **Biosorption of Methylene Blue Dye Using Natural Biosorbents Made from Weeds**

**Francisco Silva 1,\*, Lorena Nascimento 2, Matheus Brito 3, Kleber da Silva 4, Waldomiro Paschoal Jr. 2,\* and Roberto Fujiyama 1,\***


Received: 20 May 2019; Accepted: 11 July 2019; Published: 5 August 2019

**Abstract:** The purpose of this work is to make use of vegetables that, although widely found in nature, there are few applications. The weeds used here, *Cyanthilium cinereum* (L.) *H. Rob* (CCLHR) and *Paspalum maritimum* (PMT) found in the Amazon region of Belém state of Pará-Brazil, contribute to the problem of water contamination by the removal of the methylene blue dye through the biosorption process, taking advantage of other materials for economic viability and processing. The influences of parameters such as, biosorbent dose, contact time, and initial concentration of dye were examined. The characterizations were realized using SEM to verify the morphology of the material and spectroscopy in the FTIR region. As for the adsorption mechanism, the physical adsorption mechanism prevailed. The time required for the system to reach equilibrium for both biosorbents was from 50 min, following a kinetics described by the pseudo-second order model. The adsorption isotherm data for PMT were better adjusted to the Langmuir model and the biosorption capacity (*qmax*) value was (56.1798 mg/g). CCLHR was better adjusted to the Freundlich model and its maximum biosorption capacity was 76.3359 mg/g. Thus, these weed species are promising for the biosorption of methylene blue dye in effluents.

**Keywords:** biosorption; weed; methylene blue dye; natural biosorbents; adsorption isotherms; adsorption kinetics

#### **1. Introduction**

The occurrence of weeds in the Amazon Region is considered the most serious biological problem faced by cattle ranchers, as well as their control, one of the highest components of the cost of farms production [1]. It is noteworthy that these plants are undesirable and most of the time they are extracted from nature and discarded or eliminated by chemical processes. Another problem in several countries of the world are the industrial processes that generate significant amounts of effluents containing heavy metals and dyes that affect the quality of water one of the resources the most used by living beings. Water is fundamental to the existence and maintenance of life and for this, it must be present in the environment in appropriate quantity and quality [2].

When colorants are present in aquatic environments, color is generally the first impact to be recognized in an effluent because very small amounts of synthetic dyes in water (<1 ppm) are highly visible [3]. This substance causes serious problems of aesthetic nature in receiving water bodies, even when present in small quantities [4]. Dyes, besides affecting the aesthetic value of water bodies, interfere with the penetration of sunlight into the aquatic environment and thus retard photosynthesis, inhibit the growth of aquatic biotics and interfere with the solubility of gases in bodies of water [5]. In the case of effluents from the textile industries, the dye concentration generally ranges from 10 to 200 mg/L, thus being quite visible [6].

The techniques most commonly used in wastewater treatment are reverse osmosis, ion exchange, adsorption, precipitation [7], membrane filtration [8,9], photocatalysis [10–12] and flocculation [13]. Among these methods, adsorption is one of the most effective methods [7,14] and most feasible due to its cost-effective and easy handling [15,16]. Among the adsorbents most applied stand out the zeolites, polymer-based porous materials [17] and, mainly, the activated carbon, most used due to its high surface area, however its use for dye removal is still very expensive, a fact that limits its wide application in the treatment of textile effluents [18]. This has led many researchers to look for more economic and effective adsorbents as potential substitutes for activated carbon [19], resulting in the interest of adsorbents from biomass to be used as sustainable biosorbents.

It is noteworthy that these plants are undesirable and most of the time they are extracted from nature and discarded or eliminated by chemical processes, which is observed two problematic: one is to give a useful end to vegetal species that in the majority of the times causes disorder to different human activities like agricultural, forestry, animal husbandry, ornamental, nautical, energy production between others [20]. The other is chemical contamination of water which is a worldwide concern, in the case here specified by dyes. Compared with other methods, the removal of dyes from aqueous solutions by the adsorption process proved to be an excellent alternative for effluent treatment, as well as an economical technique [21]. The authors report that the use of biological materials for the removal of dyes from aqueous solutions is commonly referred to as biosorption and has now attracted a great deal of interest in scientific knowledge and in the community as sustainable and ecological materials for the production of alternative sorbents. These materials are called biossorvents [22,23].

Aware of the above problems and the search for solutions for the chemical contamination of water, the objective of this research was to evaluate the biosorption potential of biosorbents produced using weeds as *Cyanthilium cinereum* (L.) *H. Rob* (CCLHR) and *Paspalum maritimum Trin* (PMT), collected in the state of Pará amazon region, aiming at the removal of the methylylene blue dye (MB) from aqueous solutions.

The CCLHR and the PMT were characterized by Scanning Electron Microscopy to investigate their morphologies and by Fourier transform infrared spectroscopy (FTIR) to detect the presence of functional groups present in the material that corroborate for their use in the removal of methylene blue (MB) of aqueous solution. PMT is a native species of tropical America, occurring in Central America and the Caribbean, northern Brazil and the coastal zone, from Northeast to South. In Brazil the highest concentrations occur from Pará to Bahia [24]. The CCLHR also known as Vernonia cinerea belongs to the Asteraceae family. The species is native to tropical Africa, tropical Asia, India, Indochina, tropical South America, West India and the US state of Florida [25].

Recent studies indicate that approximately 12% of the synthetic dyes are lost during manufacturing and processing operations and that about 20% end up entering the environment through effluents from industrial wastewater treatment plants [26]. Dyes have a complex chemical structure that is stable to light, heat, oxidizing agents and are also resistant to aerobic digestion [27,28]. Methylene blue is a cationic dye widely used in the textile industry for the dyeing of cotton and wool fabrics. When untreated, uncontrolled discharge into rivers and lakes affect not only the transparency of the waters, but also limits the passage of solar radiation, reducing the natural photosynthetic activity and causing changes in the aquatic biota and causing acute and chronic toxicity of these ecosystems [29,30].

As has already been synthetically mentioned, an appropriate alternative method, which has proved to be quite effective for removal of dyes from aquatic environments is biosorption, a subcategory of adsorption, which uses as biological raw material where the lingnocellulosics are included. In this class of materials, agricultural byproducts have been shown to be efficient because, in addition to being abundant, they are inexpensive and have a relatively low impact on the environment. In comparison to

other effluent treatment methods, biosorption substantially reduces the costs associated with financial investment in the process as a whole [31,32].

Several works that deal with the removal of methylene blue are presented in the literature using biosorbents obtained from lignocellulosic materials in the in natura form such as coconut fiber [33], banana peel [34], mint tailings [35], pineapple peel [36], cashew nutshell [37] pine leaves [38], tea residues [39], corn straw, pupunha palm [40], trunk of the papaya tree [41]. In this paper, the biosorbents presented for the removal of methylene blue dye (MB) were produced from weeds that can be defined as any plants that grow spontaneously in a place of human activity and cause damage to this activity, be it agricultural, forestry, livestock, ornamental, nautical, energy production etc. [42].

This work has objective produce biosorbents through weeds (PMT and CCLHR) and realize biosorptions test to verify removal efficiency MB from aqueous solutions. In the biosorption assays, the influence of parameters such as dosage of biosorbents between (0.05–0.5 g), initial concentration of dye in the range of (10.0 and 50.0 mg/L), and contact time (10 and 80 min) were evaluated. The experimental data of biosorption isotherms were evaluated by the Freundlich and Langmuir models and biosorption kinetics by the pseudo-first-order and pseudo-second-order models. The maximum biosorption capacity (*qmax*) values were 56.1798 mg/g for the PMT, whose experimental biosorption isotherm data were better adjusted with the Langmuir model and 76.3359 mg/g for the CCLHR, whose experimental data of biosorption isotherms were better fitted to the Freundlich model. Finally, the research result shows the biosorbent's potential for removal MB from waste water. It is expected that this research will contribute as an alternative in the problematic of the water chemical contamination, using a simple reproduction process and a wide availability of raw material for biosorbents production.

#### **2. Materials and Methods**

#### *2.1. Preparation of Biosorbent*

The used biosorbent were produced from weeds (*Cyanthilium cinereum* (L.) *H. Rob e Paspalum maritimum Trin*). The weeds found in the Amazon region, Belém-PA, Brazil. They were collected manually within the Universidade Federal do Pará (UFPA). After harvesting, the stems were extracted and cut into lengths of approximately 5 mm. The trimmed stems were washed in distilled water and introduced in an oven at 400 ◦C for 30 days, to reduce humidity and to avoid attack of microorganisms. After 30 days, the dried samples were again crushed and washed with distilled water until we did not observe more of the coloration in the solution. The wet samples were placed in an oven at 800 ◦C for 24 h. The prepared biosorbents from weeds were stored in airtight plastic containers in order to avoid humidity, and these were utilized in the biosorption assays.

#### *2.2. Solutions and Reagents*

In the present study of biosorption, the used adsorbate was the methylene blue dye (MB), classified as basic or cationic. A stock solution of 100 mg/L was prepared separately, which was diluted to between 10 and 50 mg/L. The molecular structure and chemical formula of the Cationic dye of methylene blue are shown in Table 1.


**Table 1.** Dye characteristics.

#### *2.3. Used Equipments at the Characterization*

In the Fourier Transform Infrared Spectroscopy (FT-IR) of the biosorbents was used Perkin Elmer Spectrum Two, in order to verify functional groups present in the samples. The biosorbents morphological was investigated by using Tescan scanning electron microscope (SEM) (Vega3 SB).

#### *2.4. Biosorption Experimental Procedure*

The biosorption is biomass ability to adsorb surface pollutants by carboxylic and phenolic functional groups, in which neutral pH make deprotonated and negative charge removes cations from solution by means of process as complexation, ionic exchange and adsorption [43].

The biosorption studies for the assessment of weeds (*Cyanthilium cinereum* (L.) *H. Rob* and *Paspalum maritimum Trin*) for the removal of MB dye from aqueous solutions was conducted by means of the batch biosorption procedure using 50 mL–pH 7 of solution, submitted to constant agitation speed of 150 rpm by magnetic stirrer (QUIMIS–Q221 MAG model), without temperature control. We analyzed the influence of parameters such as biosorbent dosage between 0.05–0.5 g, initial dye concentration in the range of 10.0 and 50.0 mg/L and contact time between 10–80 min. During each procedure at predetermined time intervals, solution samples were taken for residual analysis of MB concentrations using spectrophotometer. Equations (1) and (2) were used to calculate the percentage of removal and the biosorption capacity, respectively:

$$R\% = \frac{(c\_0 - c)}{c\_0} \times 100\% \tag{1}$$

$$q(t) = \frac{(c\_0 - c)}{m} \times \mathbf{V} \tag{2}$$

where:

*R*% = Percentage of removal *c*<sup>0</sup> = Initial concentration (mg/L) *c* = Concentration at time t *V* = Volume (L) *q*(*t*) = Biosorption capacity at time t *m* = Biosorbent mass.

#### **3. Results**

#### *3.1. Characterization of PMT and CCLHR Biosorbents*

Figure 1 shows the FTIR results for the identification of the present functional groups in the PMT and CCLHR species. In the range of 3000–3720 cm<sup>−</sup>1, a broad and low intensity band was observed, peaks at 3330 cm−<sup>1</sup> (Figure 1a) and 3320 cm−<sup>1</sup> (Figure 1b), characterizing the presence of O–H (alcohols and phenols) [44]. Peaks were also observed at 1630 cm−<sup>1</sup> (Figure 1a) and 1613 cm−<sup>1</sup> (Figure 1b), which indicated the C = O stretching in organic groups of carboxyl and bending vibration of the functional group –OH [45]. The intense band at 1035 cm−<sup>1</sup> (Figure 1a) and 1033 cm−<sup>1</sup> (Figure 1b) peaks confirm the functional groups O–C–O of the cellulose and lignin structure.

The biosorption of MB on the adsorbent may be due to the electrostatic attraction between these groups and the cationic dye molecule. At pH above 4, the carboxylic groups are deprotonated and negatively charged carboxylate ligands (–COO–) bind to the positively charged MB molecules. This confirms that the biosorption of MB by adsorbent was an ion exchange mechanism between the negatively charged groups present in adsorbent and the cationic dye molecule [46].

**Figure 1.** Infrared spectrum of *Paspalum maritimum* (PMT) (**a**) and CCLHR (**b**) samples, in the range of 400–4000 cm<sup>−</sup>1.

The Scanning Electron Microscopy images of PTM and CCLHR are shown in Figure 2 with magnification of 2190× and 1720×, respectively The SEM analyzes showed a large amount of pores implying a wide surface area, which facilitates the MB biosorption process [40]. This indicates a necessary requirement of these lignocellulosic materials as potential biosorbents. In general, the PMT and CCLHR presented different morphologies along their surface demonstrating different pores sizes and heterogeneous surfaces.

**Figure 2.** Scanning electron microscope (SEM) image of (**a**) PMT with magnification of 2190× and (**b**) CCLHR with magnification of 1720×.

The characterization results by SEM and FTIR show the predominance the physical adsorption mechanism due to the porosity of the material and the electrostatic interaction between the biosorbent and the methylene blue.

#### *3.2. The E*ff*ect of Dosage*

The biosorption capacity represent the biosorbate mass can be retained by the biosorbent mass while percentage of removal is related to speed which the biosorbate flows from solution to biosorbent surface. The MB biosorption by the CCLHR and PMT plant species was examined by dosage variation of 0.05 to 0.5 g in the concentration 15 mg/L, solution volume of 50 mL and agitation speed 150 rpm. For both species, we observed that the increase in the mass of produced biosorbents led to an increase at the percentage of removal, with CCLHR from 80.96% to 98.15% and PMT from 92.33% to 95.88%. However, increasing the dosage from 0.05 to 0.5 g caused a decrease in biosorption capacity, where for CCLHR form 12.11 to 1.47 mg/g, whereas for PMT from 13.85 for 1.44 mg/g. The increase in the percentage of removal with increasing dosage is because the larger amount of mass provided a larger number of active sites available for the biosorption, which causes the increase of the percentage of removal as already reported in the literatures [47,48]. The decrease of the biosorption capacity with the increase of the biosorbent dosage can be explained by the unsaturation of a certain number of active sites, since the volume and concentration remained fixed to a higher mass value, in which should be distributed the same amount of dye. Also, there is the particle aggregation due to the increase of the biosorbent dosage, which causes a decrease in the surface area and the increase of the diffusion path to be travelled by the adsorbate inside the biosorbent [48–54]. Figure 3 shows the curves of biosorbent dose vs. biosorption capacity and biosorbent dose vs. percentage of removal of PTM and CCLHR.

**Figure 3.** Biosorbent dose vs. biosorption capacity vs. percentage of removal: (**a**) PMT and (**b**) CCLHR.

#### *3.3. E*ff*ect of Initial Dye Concentration*

The initial dye concentration in the range of 10–50 mg/L was studied for the assessment of MB biosorption using a biosorbent dosage of 0.05 g. The biosorption capacity increased with increasing concentration from 9.03 to 41.99 mg/g (MB) for CCLHR biosorbent and from 9.31 to 41.67 mg/g (MB) for PMT biosorbent. The percentage of removal decreased from 90.36% to 83.98% for CCLHR biosorbent and from 93.18% to 83.35% for PTM biosorbent. At lower initial concentrations of MB there are relatively few dye molecules and a large number of available adsorption sites, which are present in the biosorbent masses, thus, it leads to a better interaction of the adsorbate with the biosorption sites, hence resultant to the higher percentage of MB removal. With the increase in the initial concentrations of MB occurs the gradual decrease in the percentage of removal due to the saturation of active biosorbents sites, since with the elevation of concentration the number of the dye molecules increases significantly [55]. The increase of initial dye concentration from 10 mg/L to 50 mg/L provided an increase in the biosorption capacity, since higher concentrations contribute to a decrease in the adsorbate mass transfer resistance of solution to the adsorbent surface, thus filling possible active sites still unoccupied in low concentrations [56,57] (Figure 4).

**Figure 4.** Initial concentration vs. percentage of removal vs. biosorption capacity: (**a**) PMT and (**b**) CCLHR.

#### *3.4. Adsorption Isotherms*

Isotherms are diagrams showing the variation of equilibrium concentration of adsorbent with the liquid phase concentration at a temperature. These models are used to illustrate the biosorbent interaction with the biosorbate and provide the relationship between the biosorption capacity and the liquid phase concentration of biosorbate under equilibrium condition at constant temperature [58].

#### 3.4.1. Langmuir Isotherm

The Langmuir model is used in the investigation of dye biosorption from liquid solution [41]. The model based on the assumption that exists a defined number of active sites, the biosorption process occurs on a homogeneous surface through monolayer formation without any interaction with the biosorbed molecules and that all the sites has equivalent energy [56,59,60] The Langmuir isotherm [61] is represented by Equation (3).

$$q\_{\varepsilon} = \frac{q\_{\text{max}} k\_L c\_{\varepsilon}}{1 + k\_L c\_{\varepsilon}} \tag{3}$$

Where:

*qe* = amount of solute adsorbed per gram of adsorbent at equilibrium (mg/g);

*qmax*: maximum biosorption capacity (mg/g);

*kL* interaction constant of adsorbate/adsorbent (L/mg);

*ce*: equilibrium concentration of adsorbate (mg/L).

From Equation (3), we can obtain the linearized form:

$$\frac{c\_{\varepsilon}}{q\_{c}} = \frac{1}{q\_{\text{max}}}c\_{\varepsilon} + \frac{1}{k\_{L}q\_{\text{max}}}\tag{4}$$

And so, plot a chart *ce*/*qe* in function of *ce* which allows you to calculate the which allows you to calculate the *qmax* e *kL* being that 1/*qmax* is the angular coefficient of the line and the 1/*kLqmax* is the intercession with the ordinate axis.

In Langmuir's model a widely used indicator in terms of analysis is called the separation factor *RL* which is calculated on the basis of *c*<sup>0</sup> e *kL* according to Equation (5).

$$R\_L = \frac{1}{1 + k\_L c\_0} \tag{5}$$

*c*<sup>0</sup> = initial concentration (mg/L).

The value of the constant *qmax* is related to the adsorbed species concentrations on the surface. When the biosorption capacity reaches this value it means that all available sites (sites that the adsorbate binds to the adsorbent) have been filled. The constant *kL* is related to the free energy of adsorption, which corresponds to the affinity between the surface of the adsorbent and the adsorbate [58]. *RL* indicates the mode of adsorbate biosorbent interaction and allows to classify adsorption isotherms in unfavorable (*RL* > 1), linear (*RL* = 1), favorable (0 < *RL* < 1), or irreversible (*RL* = 0) [62].

#### 3.4.2. Freundlich Isotherm

According to Freundlich model, the multilayer adsorption occurs on heterogeneous adsorbent surfaces and the higher energy sites on the surface are first occupied and the binding force decreases with the increase in the degree of occupation of the active sites, which reduces the adsorption with time [56,59].

The Freundlich model is represented by [63]:

$$q\_{\mathfrak{e}} = k\_{\mathbb{F}} \mathfrak{c}\_{\mathfrak{e}}^{\frac{1}{n}} \tag{6}$$

According to Cai L. et al. [44], the linearized form of Equation (6) is described by:

$$
\log q\_{\mathfrak{c}} = \log k\_{\mathfrak{F}} + \frac{1}{n} \log c\_{\mathfrak{c}} \tag{7}
$$

*kF* and *n* are the Freundlich constants related to biosorption capacity and biosorption intensity, respectively.

The adsorption isotherms obtained experimentally for PMT and CCLHR (Figure 5) were adjusted to the Langmuir and Freundlich models, for initial concentrations in the range between 10 and 50 mg/L. The values of the constants were calculated by the linearized forms of the respective models. Figures 6 and 7 show Freundlich and Langmuir models for both biosorbents (PMT and CCLHR), respectively. The results for the parameters obtained in both biosorbents are shown in Table 2. Analyzing the values of the correlation coefficient in Table 2 and Figure 8, we observed that the experimental data of the adsorption for the PMT is better adjusted to the Langmuir model, while the CCLHR to the Freundlich model. It is justified by the values of the respective linear correlation coefficients (R<sup>2</sup> closer to unity) in each case.

**Figure 5.** Adsorption isotherm of MB: (**a**) PMT, (**b**) CCLHR.

**Figure 6.** Freundlich isothermal adsorption equation fitting of methylene blue: (**a**) PMT, (**b**) CCLHR.

**Figure 7.** Langmuir isothermal adsorption equation fitting of methylene blue: (**a**) PMT, (**b**) CCLHR.


**Table 2.** Adsorption isotherm parameters for methylylene blue (MB).

In relation to the constants *kL*, *qmax* and 1/*n* (highest value of *n*) for the PMT and CCLHR Table 2 we noticed that the highest value of *kL* and lower value of 1/*n* were found for the PMT, although the higher value of *qmax* was for CCLHR. Both models indicated higher MB affinity for PMT surface. The higher value of *qmax* for the CCLHR was possibly due to the fact of the greater diffusion in the pores and intraporos of the CCLHR than PMT. Therefore, the CCLHR presented better efficiency in relation to PMT at the conditions investigated in this work.

Various biosorbents have been applied in removal MB from the aqueous solution, as reported in the previous literature, for comparison purposes. We can compare the results with other authors in term the maximum capacity adsorbed in the conditions optimized by each authors: Carica papaya wood (*qmax* = 32.25 mg·g<sup>−</sup>1) [41], Cornbread (*qmax* = 106.383 mg·g−1) and pupunha palm (*qmax* = 78.989 mg·g−1) [40], Potato shell (*qmax* = 48.7 mg·g−1) [64], Scenedesmus (*qmax* = 61.69 mg·g−1) [65]. Through the comparative study with the Table 2, we can conclude that PMT and CCLHR are between the most efficient adsorbents prepared for industrial wastewater treatment.

The values of *RL* between 0 and 1 Langmuir model, whose variation with the initial concentration is shown in Figure 9, for both biosorbents and of *n* between 1 and 10 Freundlich model are in Table 2, confirm that the biosorption was favorable for both the biosorbents (PMT and CCLHR), ie, the adsorbate prefers the solid phase than liquid [58].

**Figure 8.** Adsorption isotherms with the Langmuir and Freundlich models: (**a**) PMT, (**b**) CCLHR.

**Figure 9.** Curve of RL by the initial concentration for PMT and CCLHR.

#### *3.5. Adsorption Kinetics*

The biosorption with the interaction time between the biosorbents and MB was evaluated in the range of 10 to 80 min, a concentration of 15 ppm and a mass of 0.05 g was used. According to Figure 10, we observed that in the first 10 min occurred a rapid increase in the percentage of removal and percentage of biosorption by both biosorbents. After this period, they became slower and remained practically constant from 50 min. The results are due to the fact that in the initial phase of the biosorption, the dyes particles to be biosorbed were almost entirely present in the solution with high probability of accessing the biosorbents surface and the active sites are unoccupied at the beginning of the process. With the increased of the time occurred a decrease in the concentration due to the migration of MB to unoccupied sites, which hindered the biosorption process by increasing the competition of residual dye particles for the remaining available sites. Results in terms of this behavior have already been reported in the literature [59,66–68].

**Figure 10.** Contact time vs. percentage of removal vs. biosorption capacity: (**a**) PMT and (**b**) CCLHR.

In the literature, there is an expressive amount of linear kinetic models that are used to evaluate the controlling mechanism of the biosorption process such as the chemical reaction, diffusion and mass transfer [69]. Frequently, the most used models are pseudo-first-order and pseudo-second-order, which were also used in this work. The equations and their linearized forms are shown in Table 3.

**Table 3.** Kinetic equations of biosorption.


In order to confirm the experimental data, we utilized the pseudo-first-order and pseudo-second-order models, where kinetic biosorption parameters of MB for initial concentration of 15 ppm are shown in Table 4. In view of these results, it was observed that for both biosorbents the model that best represented the experimental data was the pseudo second order, since R<sup>2</sup> is closer to unity. Figures 11 and 12 show the behavior of the linearized form of said models through which the parameters of Table 4 were calculated.

**Table 4.** Comparison of the pseudo-first-order and pseudo-second-order models for the biosorption of MB on PMT and CCLHR.


**Figure 11.** Plots of pseudo-first-order kinetic model for the biosorption: (**a**) PMT and (**b**) CCLHR.

**Figure 12.** Plots of pseudo-second-order kinetic model for the biosorption: (**a**) PMT and (**b**) CCLHR.

#### **4. Conclusions**

The characterizations by FTIR and SEM showed, respectively, functional groups (hydroxyl, carbonyl and carboxyl) and high porosity surface, these factors confirm the produced biosorbents by PMT and CCLHR weeds present appropriate physical-chemical properties to adsorption process. The adsorption kinetics showed high remove percent and experimental data were most be adjusted by pseudo-second-order model. The maximum biosorption capacity (*qmax* = 56.1798 mg·g−<sup>1</sup> and *qmax* = 76.3359 mg·g−<sup>1</sup> for PMT and CCLHR, respectively) showed equivalent biosorption value to literature used by MB removal from industries wastewater. The experimental data of the adsorption for the PMT is better adjusted to the Langmuir model, while the CCLHR to the Freundlich model. The range *RL* and *n* results indicated favorable biosorption. Overall, the raw material showed potential for applications with low cost biosorbent, can be a viable alternative and had and with ecological appeal to remove Methylene Blue dyes by several industries segments.

#### **5. Patents**

In this work, the used plant raw materials are filed with the National Institute of Industrial Property (INPI)-Brazil, with the code BR 10 2019 008806-0. Additionally, eight weed species divided among the Cyperaceae, Poaceae, Amaranthaceae, Asteraceae families have already been tested and showed biosorption potential.

**Author Contributions:** Conceptualization, F.S. and R.F.; Methodology, F.S., L.N. and M.B.; Formal analysis, R.F., W.P.J. and K.d.S.; Investigation, F.S., L.N. and M.B.; Resources, F.S. and R.F.; Writing—Original draft, F.S.; Writing—review and editing, R.F., W.P.J. and K.d.S.; Project administration, R.F. and F.S.; Supervision, R.F.

**Funding:** This research received no external funding.

**Acknowledgments:** All authors acknowledge financial support from PROPESP/UFPA, CAPES, CNPq. The authors K.S. and W.P.J. gratefully acknowledge support from Universidade Federal do Pará. The author LN gratefully acknowledge support from PIBIC/UFPA for the financial support. In addition, the authors acknowledge the use of the facilities at LABNANO-AMAZON/UFPA.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

MDPI St. Alban-Anlage 66 4052 Basel Switzerland Tel. +41 61 683 77 34 Fax +41 61 302 89 18 www.mdpi.com

*Materials* Editorial Office E-mail: materials@mdpi.com www.mdpi.com/journal/materials

MDPI St. Alban-Anlage 66 4052 Basel Switzerland

Tel: +41 61 683 77 34 Fax: +41 61 302 89 18