Published in Chemical Engineering Journal, Volume 293, 1 June 2016, Pages 102–111
Carbon nanofibers doped with nitrogen for the continuous catalytic ozonation of organic pollutants
J. Restivoa, E. Garcia-Bordejéb, J.J.M. Órfãoa, M.F.R. Pereiraa
aLaboratório de Catálise e Materiais (LCM), Laboratório Associado LSRE-LCM, Departamento de Engenharia Química, Faculdade de Engenharia, Universidade do Porto, Rua Dr. Roberto Frias, 4200-465
bInstituto de Carboquimica (ICB-C.S.I.C.), Miguel Luesma Castán 4, 50018 Zaragoza, Spain
Corresponding author: M. F. R. Pereira,
Abstract
Catalytic ozonation using carbon materials, in particular nanocarbons, has been appointed as an interesting alternative for the abatement of recalcitrant emerging organic pollutants. Efforts to achieve more efficient catalysts have been carried out, including carbon doping with heteroatoms. In this study, the effect of nitrogendoping of carbon nanofibers in their catalytic activity for the ozonation of organic pollutants was assessed. For this end, pristine and N-doped carbon nanofibers were prepared, both in powder and in structured forms. The former were tested in semi-batch ozonation experiments, while the latter were used in continuous ozonation experiments.It was observed that the presence of N-containing functionalities on the surface of the carbon nanofibers enhances their capability as catalysts for the studied reaction.
Keywords:ozonation, nitrogen-doped nanocarbons, structured catalysts
1.Introduction
Recent findings of the resistance of harmful products to conventional water treatments, as well as of the toxicity associated with the degradation products of these compounds, have led to a growing interest in the development of novel water treatment technologies that may efficiently remove recalcitrant emerging organic pollutants [1]. In particular, advanced oxidation processes such as catalytic ozonation have been shown to be an interesting method for the abatement of such pollutants [2-7].
Different materials have been used as catalysts for the ozonation process [8]. In particular, carbon materials such as activated carbons [2,3, 9-11], multi-walled carbon nanotubes (MWCNT)[12-16], carbon xerogels[17] and carbon nanofibers (CNF)[18-21] have been shown to have the potential to be effective solutions.
The modification of the surface properties of carbon materials have been known to influence their catalytic properties [22, 23]. In particular, the presence of surface heteroatoms, such as oxygen, sulfur and nitrogen, has been shown to significantly affect the catalytic activity of carbon materialsin catalytic ozonation[2, 16, 24-29].
The application of structured catalysts, in particular as honeycomb monoliths, has been appointed as an interesting alternative for the application of carbon materials as catalysts during the ozonation process [18-21].
In this work, a study of the influence of the presence of N-containing functionalities on the surface of CNF was carried out. For this end, catalytic ozonation experiments were performed in semi-batch and continuous operation, using CNF in powder form and in structured form,respectively. Oxalic acid and phenol were selected as model compounds, and further ahead atrazine, metolachlor and nonylphenol were used as model organic micropollutants.
2.Methods and materials
2.1 Preparation of catalysts
2.1.1 Powder carbon nanofibers
Carbon nanofibers in powder form were prepared using a 20% Ni on Al2O3catalyst, which was previously reduced under H2at 550 ºC. Carbon growth was carried out using a C2H6:H2(50:50) mixture at 600 ºC. Nitrogen functionalities were introduced on the surface of the carbon nanofibers replacing H2by NH3 during the growth phase. Both samples (without and with nitrogen) CNF and N-CNF were purified from the growth catalyst first under NaOH reflux at 80ºC for 4 h and later under HCl reflux at 100 ºC for 4 h. After purification, less than 1 wt% residual catalyst remains on the carbon material.
2.1.2 Structured carbon nanofibers
A well attached layer of entangled CNFs was grown on the walls of cordierite monoliths as reported elsewhere [30]. In brief, cordierite monoliths (from Corning, 22 mm diameter, 60 mm length, 400 cpsi) were washcoated with alumina by a dip-coating method similar to the sol-gel coating described by Nijhuis et al.[31]. Nickel was deposited by adsorption from a pH-neutral nickel solution as described elsewhere [32]. To growth the CNFs, the Ni/alumina coated monolith was fitted in a quartz reactor by wrapping it in quartz band. The reduction of the calcined catalyst was carried out in a hydrogen atmosphere at 550 ºC for 120 min (heating rate of 5 ºC/min). The monolith was then heated (5 ºC/min) to 600 ºC. When thistemperature was reached, 100 cm3(STP)/min of a C2H6:H2 (50:50) gas mixture was fed. For N-doping of grown CNF, H2 was replaced by NH3in the gas feed [33]. The CNF growth was allowed to proceed for 2 hours following up by cooling down under inert atmosphere.
2.2 Characterization of catalysts
The textural characterization of the prepared carbon nanofibers was carried out by N2physisorptionat -196 ºCusing aMicromeritics ASAP 2020 apparatus, after outgassing for 4 h at 150 ºC. The pore volume was calculated from the adsorbed amount at a relative pressure of 0.99. The specific surface area was calculated by the BET (Brunauer, Emmet and Teller) method in the relative pressure range 0.01–0.10 following the ASTM-4365 standard.
The nature of the N-containing functionalities introduced on the surface of the carbon nanofibers was characterized by XPS using a ESCAPlusOmnicrom equipped with a Mg Kα radiation source to excite the sample. Calibration of the instrument was performed with Ag 3d5/2 line at 368.27eV. All measurements were performed under ultra-high vacuum better than 10-10torr. Internal referencing of spectrometer energies was made using the C 1s signal at 284.6 eV. The curve fitting of the spectra was performed using CASA XPS software after applying a Shirley baseline. For peak deconvolution, the full width at half maximum (FWHM) was fixed equal for all the peaks and with a maximum value of 2.5 eV.
2.3 Evaluation of catalysts
The prepared catalysts were evaluated in the catalytic ozonation of organic pollutants using two systems: asemi-batch system where the catalysts were used in powder form and acontinuous ozonation system where the catalysts were used in their structured form.The powder samples are identified as CNF and N-CNF (without and with nitrogen, respectively), and the honeycomb monolith structured samples are identified as HM-CNF and HM-N-CNF (without and with nitrogen, respectively).
The selected organic pollutants included oxalic acid, phenol, atrazine (ATZ), metolachlor (MTLC) and nonylphenol (NLP).
The semi-batch ozonation experiments were carried out using a conventional stirred tank reactor. Ozone was produced from pure oxygen using a BMT 802N ozone generator, at 50 g m-3(STP),and introduced into the reactor using a glass disperser at 150 cm3(STP)min-1. A volume of 700 mL of solution containing the pollutants at the desired concentration was used, and the amount of powder catalyst used, when applicable, was 100 mg. The solution was kept homogeneously stirred using a magnetic stirrer at 200 rpm.
The continuous ozonation experiments were carried out using a bubble column containing an internal loop. Ozone was produced from pure oxygen using a BMT 802N ozone generator, at 50 g m-3(STP),and introduced into the reactor using a glass disperser at the bottom of the bubble column at 20 cm3(STP) min-1. The solution containing the selected pollutant was fed into the reaction system using a peristaltic pump at 12 mL min-1, and the internal loop was kept flowing at all times at 60 mL min-1. The structured catalysts were placed inside the bubble column, thus operating in a multiphase flow where the gas and liquid simultaneously contact with the solid phase. The reaction conditions were optimized to achieve homogeneous axial dispersion of the phases throughout the channels of the monoliths, and to allow the formation of Taylor flow, which is known to enhance the performance of such systems due to decreased mass transfer resistance[34].
The concentration of oxalic acid and other organic acids was followed using an Elite LaChrom HPLC coupled with a UV-Vis detector. Separation was achieved using an Alltech OA-1000 chromatography column with a 5mM H2SO4mobile phase and detection was carried out at 200 nm. The concentrations of atrazine, metolachlor,nonylphenol and phenolwere followed using an Elite LaChrom HPLC coupled with a DAD detector. Separation was achieved using a Lichrocart C18-RP Puroshper Star chromatography column. For atrazine and phenol a MeOH:H2Omobile phase was used, and detection was carried out at 222 nm. For metolachlor an ACN:H2Omobile phase was used and detection was carried out at 196 nm. For nonylphenol an ACN:H2O mobile phase was used and detection was carried out at 190 nm.
Acute toxicity analyses were performed using an Azure Environment Microtox apparatus and procedure ISO/DIN 11348-3. The microorganisms used were the luminescent bacteria Vibrio fischeri from Hach Lange, which is used as representative of aquatic environments. The bacteria were exposed to samples after activation and 15 min incubation at 15 ºC, and the decrease in activity as function of the luminescence was measured after 30 min.
Total organic carbon (TOC), measured with a Shimadzu TOC-5000A apparatus, was used to assess the mineralization degree.
3.Results and discussion
3.1 Characterization of fresh catalysts
3.1.1 Powder carbon nanofibers
Textural characterization of the pristine and N-doped powder CNF is presented in Table 1.
Table 1 – Textural characterization of the pristine and N-doped CNF.
CNF / BET Area(m2 g-1) / Pore
Volume
(cm3g-1) / Average pore
Size
(nm)
CNF / 151 / 0.39 / 9.1
N-CNF / 318 / 0.62 / 6.2
Some differences are observable in the textural properties of pristine and N-doped CNF. Particularly, the specific surface area calculated by the BET method showed a significant increase when the N-CNF sample is considered. An increase in the pore volume was also observed, while the average pore size decreased.
The N-doped carbon nanofibers were characterized by XPS. The amount of each nitrogen species identified on the surface of the N-doped CNF are presented in Table 2.
Table 2 – Relative amounts of nitrogen groups on the surface of N-doped CNF determined by XPS analysis.
Sample / PyridinicGroups / Pyrrolic
Groups / Quaternary
Groups / N/C
BE
(ev) / RA
(%) / BE
(ev) / RA
(%) / BE
(ev) / RA
(%)
As-grown
N-CNF / 398.4 / 69.7 / 399.8 / 1.3 / 400.8 / 29.0 / 0.026
Purified N-CNF / 398.4 / 43.6 / 399.8 / 25.8 / 400.8 / 30.6 / 0.021
Where: BE – Binding energy; RA – Relative amount
Pyridinic and quaternary nitrogen functionalities were detected by XPS on the surface of the as-grown N-CNF, together with a small amount of pyrrolic groups. The analysis carried out after purification with NaOH-HCl showed changes in the relative amounts of these functionalities. Namely, a much larger relative amount of pyrrolic groups was identified. The purified N-CNF fibres were used for the catalytic experiments presented below.
3.1.2 Structured carbon nanofibers
The characterization of the structured CNF catalysts has been thoroughly described elsewhere, for both pristine [30, 35] and N-doped [33] carbon nanofibers.
For brevity, only the nature and amount of the N-containing functionalities found on the surface of the HM-N-CNF structured catalyst are reproduced here (Table 3).
Table 3 –Characterization by XPS of nitrogen groups on the surface of N-doped structured CNF (HM-N-CNF).
Sample / Pyridinicgroups / Pyrrolic
groups / Quaternary
groups / Oxidized N / N/C
BE
(ev) / RA
(%) / BE (ev) / RA (%) / BE (ev) / RA
(%) / BE (ev) / RA (%)
HM-N-CNF / 398.4 / 52.6 / 400.3 / 14.4 / 401.2 / 25.2 / 404.5 / 7.8 / 0.074
Where: BE – Binding energy; RA – Relative amount
When compared with the powder N-CNF, a wider relative amount of nitrogen was found in the structured catalyst. Furthermore, some oxidized nitrogen was also detected in this sample.
3.2 Ozonation experiments
3.2.1 Powder carbon nanofibers
The first approach to the evaluation of the influence of N-doping of CNF on their performance as catalysts in ozonation was carried out using oxalic acid as a model compound, in a semi-batch system. The dimensionless removal of oxalic acid during catalytic ozonation using pristine and N-doped CNF samples is presented in Figure 1. The result obtained by single ozonation is also included for comparison purposes.
It is clear from Figure 1 that the addition of CNF (either pristine or N-doped) to the ozonation system enhanced the removal of oxalic acid from solution. The catalytic activity of carbon nanomaterials in this reaction is well-known [12, 16, 18-21, 24, 36, 37].
On a more interesting note, similarly to what was reported formultiwalled carbon nanotubes [38], the N-doped CNF showed better performance in the removal of oxalic acid than the pristine CNF sample. It has been previously reported that the introduction of nitrogen-containing functionalities on the surface of carbon materials might enhance their catalytic activity.Studies performed using activated carbons have shown that the inclusion of basic nitrogen functionalities on the surface enhances their catalytic activity in the ozonation reaction; this effect has been attributed to either an increase in the rate of surface reactions due to larger availability of free electrons on the surface, or to the reaction of ozone with pyrrol surface functionalities leading to the formation of hydroxyl radicals that are released and are able to further react with the organic pollutants in the liquid bulk [25, 26, 28]. On the other hand, studies carried out using carbon xerogels[29] and carbon nanotubes [38] have suggested that the main factor for the improvement of the catalytic activity by introduction of nitrogen functionalities is the increase in the surface density of free-electrons.
The results obtained in the catalytic ozonation of oxalic acid support the idea that the increase in electronic density on the surface, promoted by the N-functionalities, favours the reduction of ozone, as recently proposed [25, 26, 28, 29, 39]. It is interesting to notice that the activity observed for the CNF samples is very close to that observed for MWCNT [38], for both pristine and N-doped samples.
The same samples were also used in the catalytic ozonation of phenol. The dimensionless concentration of phenol during these experiments is presented in Figure 2.
As it was observed when MWCNT were used [38], the presence of a catalyst in the ozonation of phenol does not significantly alter its conversion, likely due to its fast reaction with molecular ozone [40, 41].
The dimensionless concentration of TOC during ozonation experiments using CNF catalysts is presented in Figure 3, as a measure of the mineralization degree.While the mineralization degree was slightly improved with the addition of the catalysts, there was not much difference between the pristine and the N-doped CNF.
The concentration of the main primary intermediates formed (benzoquinone and hydroquinone) was followed (not shown), and it was observed that the presence of a catalyst did not significantly affect their evolution. Moreover, both compounds were quickly removed from solution, even in the case of single ozonation. A much more significant difference was observed in the case of oxalic acid, which was observed to accumulate in solutionthroughout the ozonationprocessuntil 3 h (duration of the experiments). The concentration of oxalic acid during these experiments is presented in Figure 4a, while the final accumulated TOC values at the end of the reactions, where the contribution of oxalic acid and of other unidentified organic compounds is discriminated, is presented in Figure 4b.
It is clear from Figure 4a that the addition of a catalyst to the ozonation system reduces the amount of oxalic acid accumulated during the ozonation of phenol. This effect might be attributed either to the enhanced removal of oxalic acid from solution, and/or hindering of the formation of oxalic acid due to changes in the reaction pathway.
Observation of Figure 4b indicates that the main contribution for the remaining TOCafter 3h during experiments with CNF catalysts was the increase in the released amount of organic intermediates other than oxalic acid, when compared with the single ozonation experiment. While a slight increase in the formation of these other organic intermediates was observed in the case of N-CNF, a conclusive statement on the role of the N-containing functionalities in this case cannot be formulated. The observed differences in the formation of oxalic acid and of other organic intermediates for the CNF and N-CNF catalysts are not sufficiently significant for such a conclusion. The presence of nitrogen species on the surface of CNF are, however, expected to play a role in the ozonation of phenol, due to changes in the electronic density on the surface of the fibres [42]. Moreover, changes in the degradation mechanism of organic pollutants during catalytic ozonation with N-containing carbon materials have been previously reported [25, 26, 28].
3.2.2 Structured carbon nanofibers
After assessment of the influence of N-doping of CNF in the catalytic performance for the ozonation process, continuous experiments were carried out using the structured CNF catalysts.
First, the pristine and the N-doped CNF structured catalysts(samples HM-CNF and HM-N-CNF, respectively) were tested in the continuous ozonation of oxalic acid. The removals measured at steady state are presented in Figure 5.
While the removal of oxalic acid was less extensive when the HM-N-CNF monolith was used, the normalization by the amount of carbon on the structured catalysts inverts this trend. In fact, the HM-N-CNF structured catalyst allows the removal at steady state of approximately more 0.05 mM of oxalic acid per gram of CNF than the other sample. This observation agrees with what was observed in semi-batch experiments, where the amount of catalyst used was the same for the experiments with the different samples. Thus, the doping of the CNF with nitrogen improves the activity of the CNF in the catalytic ozonation of oxalic acid.
Further experiments were carried out using selected organic micropollutants: atrazine, metolachlor and nonylphenol. When comparing the removals of the parent pollutants in the single and catalytic ozonation experiments (not shown), it is clear that the inclusion of a catalyst does not significantly alter the removal achieved, due to quick reaction of these pollutants with ozone [9, 19, 21, 43-49]. Furthermore, no significant difference between the removals achieved was observed when N-doped or pristine CNF structured catalysts were used. In fact, there is no evidence to support the idea that surface reactions might be improving the oxidation of these pollutants [12]. In the case of oxalic acid, it is known that surface reactions are contributing to the oxidation of the pollutant during ozonation, at similar conditions to those used here [2], and thus it was possible to conclude about the role of the nitrogen functionalities in the improvement of the removal rate obtained. The mineralization degree, as measured by TOC removal, is a better measure of the efficiency of the catalysts, and is much more closely related to surface reactions, or radical reactions in the liquid bulk, than the oxidation of the parent pollutant [9, 12, 19, 21]. The mineralization degree of the three organic pollutants, obtained during continuous ozonation experiments using HM-CNF and HM-N-CNF structured catalysts, is presented in Figure 6.The mineralization degrees achieved at steady statein the presence of the CNF and N-CNF structured catalysts were higher than those observed in the case of the non-catalytic experiments (not shown). This is attributed to the oxidation of intermediates resistant to direct ozonation, by either reactions occurring on the surface of the CNF or by reaction with hydroxyl radicals in the liquid bulk [12, 18, 19, 21]. On the other hand, only slight differenceswere observed between the two catalysts; the CNF sample showing a somewhat improved performance when compared with the N-doped sample. However, similarly to what was concluded in the experiments with oxalic acid, the TOC removal normalized by the amount of CNF is higher when the N-CNF structure is considered. Thus, it is possible to conclude that the presence of nitrogen on the CNF improves the rate of surface reactions capable of oxidizing compounds resistant to direct ozonation, or increases the production of hydroxyl radicals that may react with compounds resistant to direct ozonation in the solution bulk.