Skip to main content

Monolithic reactors structured at the nano and micro levels for catalytic water purification

Final Report Summary - MONACAT (Monolithic reactors structured at the nano and micro levels for catalytic water purification)

Executive summary:

The objective of the present work is the development of an emerging technology for the catalytic detoxification of water. In the preparation of the structured catalytic reactor, we aimed at achieving control of the structure at different length scales, from macroscopic scale (reactor level) to nanoscopic scale (catalyst nanoparticles). The global objective is to develop and apply novel nano-engineered materials based on carbon nanofibres with hierarchical structure to two catalytic processes for purifying different types of waters, improving the results obtained with conventional techniques without the production of undesirable by-products, looking for a potential long-term impact and decreasing the cost and the energy-use of other techniques. The target processes are two:

1. a reduction process such as the catalytic reduction of nitrates, bromates and perchlorates and
2. an advanced oxidation process such as catalytic ozonation of emerging contaminants from household and industry.

Some partners have been working on the optimisation of the nanocarbon coating on structured reactors (cordierite and metallic monoliths, sintered metal fibres, carbon cloth and activated carbon fibres). The structured catalyst has fluidodynamic and mechanical properties that render them suitable for a commercial application. There is negligible pressure drop, therefore it can be implemented in exiting processes without increasing significantly the energy costs. Furthermore, there is no release of nanomaterials in adhesion tests under real flow conditions. Therefore, the composites are suitable for using in catalytic water treatment from the point of view of durability and health safety.

Other partners have designed and implemented the plants to carry out the reactions in continuous. In parallel, other partner has developed a mathematical model to describe the water treatment processes. Detailed kinetic model of catalytic nitrates reduction and ozonation in water have been developed and validated. Some important results of the project are the development of a modelling methodology for multiphase reactions and a methodology for validation of model parameters. The process conditions for the operation in continuous have been optimised with the help of the experimental results and the input from the modelling.

The catalysts were tested in the ozonation of five micropollutants and in the reduction of nitrates, bromates and perchlorates. The five micropollutants selected are naturally present in some rivers and industrial waters and they are an antibiotic (erythromycin), a pharmaceutical (bezafibrate), two pesticides (atrazine, metolachlor) and an endocrine disruptor (nonylphenol). The experiments comparing single ozonation and catalytic ozonation under conditions relevant industrially showed that both methods led to comparable removal of the initial pollutant. However, catalytic ozonation was more effective in the removal of degradation intermediates bringing about less toxic by-products, according to microtox test. Although the process shows a high potential, still some modifications should be made in the catalyst towards the ultimate goal of complete mineralisation.

Catalyst based on skaergaardite (Pd-Cu) and palladium-tin (Pd-Sn) were supported on the carbon nanofibres (CNF) structured supports and tested in reduction of nitrates to nitrogen (N2). The challenge of this reaction is to obtain a catalyst completely selective to N2 avoiding ammonia (NH3) formation, which is also a pollutant of water. The catalyst PdSn/5%CNF/SMF demonstrated the best published activity and N2 selectivity (approximately 75 %) in an open flow reactor in a continuous mode using natural water, as far as we know. The improvement of the selectivity to N2 is attributed to the mesoporosity of the support. Nevertheless, the ultimate goal of zero NH3 production has not been reached yet. This precludes the commercialisation for treatment of drinking water where complete NH3 removal is necessary. However, this catalyst demonstrated effective and durable for the treatment of industrial waters with a high concentration of nitrates and high conductivity. In this particular application, this technology outperform other technologies, which are non selective to the pollutant and which concentrate the pollutant instead of destroying it.

Catalysts of Pd and Pd-Sn supported on CNF/structured reactors were also tested in the reduction of bromates to bromides. Bromates are a subproduct of ozonation which needs removal due to the strict regulation. The catalyst was very effective and stable for the reduction of bromates to innocuous bromide, which have significant advantages over conventional technologies such as adsorption, which accumulate the pollutant and need frequent replacement of the sorbent.

The structured catalyst prepared here show a high potential in several reactions tested. However, for commercialisation further research is needed for the validation of the technology in continuous tests on different type of waters at the pilot level (efficiency, reliability, technical and economical evaluation) and the integration in existing treatment plant via the use of the first industrial prototype.

Project context and objectives:

The objective of the present work, funded by the European Seventh Framework Programme (FP7) environment (ENV) nanosciences, nanotechnologies, materials and new production technologies (NPM), funding scheme FP7-ENV-NPM-2008-2, is the development of an emerging technology for the catalytic detoxification of water. In the preparation of the structured catalytic reactor we aimed at achieving control of the structure at different length scales, from macroscopic scale (reactor level) to nanoscopic scale (catalyst nanoparticles). The global objective is to develop and apply novel nano-engineered materials based on carbon nanofibres with hierarchical structure to two catalytic processes for purifying different types of waters, improving the results obtained with conventional techniques without the production of undesirable by-products, looking for a potential long-term impact and decreasing the cost and the energy-use of other techniques. The target processes are two:

1. a reduction process such as the catalytic reduction of nitrates, bromates and perchlorates and
2. an advanced oxidation process such as catalytic ozonation of emerging contaminants from household and industry.

The objectives of the project are:

1. growing a well attached, uniform and thick layer of nanocarbon materials (NCM) over the structured reactor prepared previously. The good adherence of the NCM layer has been verified by the weight loss in ultrasonic maltreatment. The uniformity of coverage has been checked in scanning electron microscopy (SEM) images by the absence of cracks and uncovered areas.
2. testing the catalyst in conventional slurry reactor with the objective of achieving higher efficiency and/or selectivity than commercial catalyst on conventional supports
3. implement plants for testing the catalysts in three phase systems
4. achieving high conversion in the catalytic reduction in continuo of the anions (nitrates, bromates and perchlorates) present in natural, concentrated and industrial waters. These anions should be reduced to negligible levels to comply with the strict regulation limits. In the case of nitrate to less than 50 mg/l, which is the maximum contaminant level (MCL) allowed for drinking water in European Union (EU). In this later reaction, special attention is paid to achieve high selectivities towards nitrogen, avoiding undesirable formation of NH3. Develop a catalyst effective and stable in the reduction of bromates (10 µg/L allowed level since 2009) and perchlorates.
5. achieving high rates of mineralisation of organic emerging pollutants by catalytic ozonation in continuous both in natural and concentrate water. They are an antibiotic (erythromycin), a pharmaceutical (bezafibrate), two pesticides (atrazine, metolachlor) and an endocrine disruptor (nonylphenol).
6. special attention has been paid to study by-products formed in simple ozonation and catalytic ozonation and its toxicity. To enhance the performance of nanocarbon materials as catalyst, low cost and non-leaching metal oxides (CeO2) and addition of hydrogen peroxide (H2O2) as co-reactant has been considered.
7. the catalyst should have a high durability under real conditions using either natural or industrial waters
8. characterising the catalyst using advanced characterisation techniques, such as in situ spectroscopy, high-resolution transmission electron microscopy (HRTEM), extended x-ray absorption fine structure (EXAFS) and relate it to the catalytic performance. This gives us feedback to modify the synthesis method in order to obtain improved catalyst for the target reactions.
8. life cycle assessment of the technology and validation of scaling of catalyst (new experimental data and new calculations)
9. to develop a multiphase flow simulation of a structured monolith reactor to be able to design an optimal reactor configuration and to develop a predictive hierarchical simulation model of the reactor for both ozonation and reduction reactions
10. modelling of the mechanism of catalytic ozonation and validation against new experimental data
11. molecular dynamics and fluidodynamic study of the CNFs based on advanced characterisation and mathematical model of topology of CNF layer.

Project results:

Preparation and activation of ceramic and metallic monoliths and catalyst for NCM growth, work package two (WP2) TLTL The objective of WP2 is the preparation of monolith metallic structures with a precise control over different channel sises. The monolithic structure (metallic and ceramic) has been coated with the catalyst for CNF growth (supported on hydrotalcite, zirconia, alumina or SiO2) by thermal spraying of powder precursor.

Norta prepared stainless steel foils coated by nickel (Ni) oxide using a spray pyrolysis method. Tests of the Spanish Science Research Council (CSIC) showed the influence of ultrasonic treatment on integrity and adhesion of catalytic support to metal substrate. The adhesion of sprayed layer to metal substrate was insufficient and, in some cases, partial depletion of layer was seen. Possible reason of this effect may be, as showed investigations, uncontrollable growth of NCM on the border between sprayed layer and steel substrate. Summarising the received result of first NCM growing tests five honeycomb structure 22 mm diameter and 45 mm length blocks were produced and sent to partners for growth tests in the monoliths.

Another substrates ceramic honeycomb structures were supplied and tested by CSIC. The work carried out at CSIC included the coating of cordierite monoliths with the growth catalyst. Cordierite monoliths were coated with Ni on alumina and zirconia by dipcoating in a sol of the precursor. The NCM grown on cordierite monoliths exhibited less than 2.5 wt% weight loss in ultrasound tests. Therefore, the adhesion is very good according to the standard test and then it is acceptable to be used in water treatment.

It is known that the required content of alumina to cover ceramic monoliths (honeycomb) can vary according to the final application of this. We have deposited a 9 % in weight of alumina with a smaller deviation to 5 % in all the studied monoliths. The percentage of alumina coating has been calculated by the difference of weight of the monolith before and after impregnation, whereas the wt% of Ni deposited on monoliths by electrostatic adsorption, was calculated by the analysis by inductively coupled plasma optical emission spectrometry (ICP-OES) of the residual Ni content of the aqueous solution after deposition, obtaining a 10 wt% with respect to alumina washcoating. A visual inspection of monoliths after impregnating with Ni and after CNF growth shows a homogeneous and uniform green and black colour, respectively, which indicates that Ni is deposited uniformly along the three dimensions of the monolith.

SEM characterisation of the monolith before and after covering with alumina showed the formation of a very thin layer alumina of approximately 1-2 µc of thickness.

In summary, the CNF adhesion was not good on metallic monolith but it was good on ceramic monoliths.

Preparation and activation of structured fibrous supports and catalysts for NCM growth (WP3)

The main achievements in WP3 according to the WP3 tasks:

1. In order to produce sintered metal fibres (SMF) and activated carbon fibres (ACF) supports with optimal chemical properties for catalytic applications, different methods of material surface treatment were developed. The SMF (Inconel, stainless steel) pre-treatment procedures (cleaning, thermal activation and acid treatment) were chosen with regard to further CNF growing. ACF were chemically modified and functionalised, such as cleaning, oxidative H2O2 and nitric acid (HNO3) treatment, to adjust surface chemistry for deposition of a catalytically active phase. The optimum treatment conditions were found in all cases.
2. The structured monoliths fabrication using cutting, corrugation, assembling and fastening procedures adapted to SMFInconel and SMFSS was successfully carried out by Norta.
3. The SMFInconel composition (around 60 % of Ni) allows after a special pre-treatment of the material a direct growing CNF resulting in CNF/SMF composites. The method of the SMFInconel pre-treatment was developed. Alternatively, SMFSS should be first coated by an oxide support layer. The multi-steps sol-gel technique of oxide layer deposition (precursor solution preparation, deposition, drying, thermal treatment) was developed and optimised. Various SMFSS-based composites (10 compositions) containing different metal oxides, such as aluminium oxide (Al2O3), magnesium oxide (MgO), zinc oxide (ZnO), nickel oxide (NiO) and mixed oxides were prepared. After characterisation and CNF growth test the most promising compositions: 5wt.%(Al2O3+MgO(1:1))/SMFSS, 5%(Al2O3+ZnO(1:1))/SMFSS were selected for further application. The most important advantage of prepared materials is the quality of the oxide coatings. The developed synthetic approach allows to create thin (avoiding internal mass-transfer limitations), homogeneous, crack-free, stable oxide layers on a metal surface.
4. Two effective methods of CNF-growth catalyst (Ni) deposition were developed. The first (in situ) method includes the introduction of a Ni-precursor into the impregnation solution used for the oxide layer deposition by a sol-gel technique. The composition 5wt.%(Al2O3+MgO(1:1)+NiO(20%))/SMFSS was selected for further testing and delivered to the partners (CSIC, NTNU). The second method was the direct deposition of metallic Ni on SMFSS surface. This was done by electroless Ni-deposition (plating) technique. The process technology (precursor solution preparation, deposition, drying, thermal treatment) was developed and optimised. The Ni loading was in the range of 0.5 to 5 wt %. The compositions 0.5wt%Ni/SMFSS and 5wt%Ni/ SMFSS were selected for CNF-growing and delivered to CSIC.
5. The prepared materials were characterised by different methods (ultrasonic bath test, corrosion resistance, active metal dispersion, SEM, etc.). They demonstrated homogeneity and stability of the coatings, their strong anchoring to the SMF surface.

Growth of NCM layer on the monoliths via catalytic chemical vapour deposition (CCVD) (WP4)

The objective of this WP is to develop a reliable and reproducible method to grow a well attached, uniform and thick layer of NCM over the structured reactor, such as metallic foils, sintered metal fibres, ceramic monoliths or carbon felt (CF). Different parameters such as good adherence, uniformity and desired thickness have been asked for fitting with the criteria of the reactor.

The main achievement of this WP is the growth of a well-adhered layer of entangled CNFs on structured reactors such ceramic monoliths, CF and sintered metal fibres).

CNF on ceramic monoliths

CSIC carried out the growth of CNF on ceramic monoliths coated with Ni/Al2O3. The CNF loading was about 15 wt%. This CNF coating was uniform in terms of thickness and coverage. The CNFs were of a narrow diameter distribution and formed a network of entangled CNF with a surface area of about 200 m2/g in the range of mesoporosity, mean diameter 17 nm. The adhesion of the CNFs to the monoliths was excellent.

NCM grown on graphite felt

As a matter of fact, this support presents a relative opening structure where the NCM can easily grow on a well dispersed catalyst particles. The catalyst deposition and NCM synthesis have been optimised in order to get a homogeneous and good anchored NCM layer. The thickness of this layer is difficult to estimate because of the tortuosity of the structure and the presence of NCM everywhere inside the felt. The weight increase in percentage is in that case a more appropriated parameter. In our case, percentages up to 100% can be easily reached without damaging or closing the structure. This structure is relatively easy to handle and can be used directly as a catalyst support for further reactions.

Carbon nanofibres of two types, i.e. fishbone (CNF) and platelet (CNP), were grown on graphite felt. Previously to the growth with ethane, Ni or nickel copper (NiCu) nanoparticles were deposited for CNF or CNP, respectively.

Test adhesion

The anchorage of CNFs is really good, with most of samples losing less than 2 wt% after ultrasonic treatment. No weight loss after ultrasonic treatment were noticed on CNP CNFs-CF materials.

BET measurements

After carbon nanofiber synthesis, the CF had a specific Brunauer-Emmett-Teller (BET) surface area of 80 m2/g.

SEM

Some differences between CNP and CNF can be noticed with SEM pictures. As a matter of fact, the CNP fibres seem much shorter but more homogeneous in diameter than CNF fibres. The adhesion is as well much better.

Raman spectroscopy

The two bands obtained on the as-received CF are typical for graphite although their intensities are relatively low compared with the signal from the carbon nanofibres. As expected, no second order on CNP CNFs has been detected according to the nature itself of the CNP fibres. Raman is a trustable tool to identify and distinguish CNF from CNP.

NCM grown on SMFInconel

Another support used is SMF provided by EPFL. Partners found this support easy to use and with good results in terms of NCM growth. Two different SMF were tested: Inconel and Stainless Steel. The first one seems to get the most promising result concerning the weight increase and specific surface area and the adhesion test was also excellent. The corrosion resistance was better than the initial SMFs, therefore they are feasible for water treatment. They observed also that the NCM on inconel are not well graphitised which is not that important to use as catalyst support because they have very high surface area.

CNF loading as a function of the growth time

The kinetics of CNF growth depends strongly on the gas composition. When the growth was carried out with a gas composition without inert, i.e. 50:50 (C2H6:H2) the carbon deposition occurs very fast at the beginning for 0.5Ni%/SMFSS, yielding more than 5 wt% carbon in the first 15 min of reaction. To have more control over the carbon yield, we carried the CNF growth with a diluted gas feed consisting of Ar:C2H6:H2 (50:25:25). For SMFinconel, in order to moderate the kinetics of CNF growth further, we used a gas feed more diluted with argon i.e. a composition of Ar:C2H6:H2 (80:10:10). Generally speaking, the carbon yield increases almost linearly in the first 30 minutes and later the growth rate slowed down or stabilised. This agrees with the mechanism usually reported in the literature in which the growth of CNFs consist of three stages, viz. nucleation, growth and deactivation. The stabilisation of carbon yield after certain time indicates that the catalyst is almost deactivated. However, pyrolitic carbon can still be deposited leading to thickening of CNFs.

SEM characterisation

In this second period we characterised samples of stainless steel sintered metal fibres (SMFSS) coated with a sol-gel of Al2O3+MgO+NiO. i.e. 5wt.%(Al2O3+MgO+NiO)/SMFSS.

The amount of Ni in the gel is crucial for the CNF yield. In samples containing 10% of Ni the weight increase after CNF growth is negligible even for gas feeds with high hydrocarbon content. For sample containing a 20% Ni the weight increase reaches a maximum growth in about 3.43 wt% independently of the gas composition. The higher CNF loading are grown on the sample with 50% Ni loading. The weight after CNF growth increases to about 9%, even for short growth times (15 min) and gas compositions with low hydrocarbon content (80:10:10). The different CNF yield gives rise to different morphologies after CNF growth.

Textural characterisation by N2 physisorption of different types of SMF after CNF growth

Both the pore volume and surface area follow this decreasing order for the different types of samples: SMFInconel higher than 5%Ni/SMFSS higher than NiO(53%)+ Al2O3+ MgO(1:1)/SMF higher than 0.5%Ni/SMFss.

Both pore volume and surface area exhibit a dependence with the CNF yield. The pore volume decreases with the CNF growth, while Surface area exhibited a maximum for some types of SMFs.

Conclusions

The graphite nanofibres presents very good anchorage to the support either ceramic monoliths, CF or SMF. There are some differences between the CNF grown on the different supports. The highest CNF loading corresponds to CF support. These nanoshaped macro-supports are suitable for the use in the two catalytic reactions involved in the project.

Deposition of active phase for catalytic water treatment (WP5)

The first objective is to functionalise the surface of CNF with O or N heteroatoms without affecting the adhesion of the CNF. The second objective is to deposit the metal active phase (Pd, Pd-Cu, Pd-Sn) in a dispersed way and its characterisation by advanced instrumental techniques. This will enable to relate the catalytic behaviour to the structure of the catalyst.

Funtionalisation of the NCM

We generated oxygenated surface groups by treating the CNF/SMF composites with two oxidising agents in liquid phase, i.e. HNO3 and H2O2 and O2 in gas phase. We used two compositions of HNO3, i.e. concentrated and 1 Molar, different durations of the treatment and two different temperatures, i.e. room temperature and 90ºC. Regarding H2O2 treatment, we used different durations and two temperatures, i.e. room temperature and 90 ºC.

We created Nitrogen functionalities by growing the CNFs with NH3 in the feed gas and also by post-treatment of H2O2-oxidised CNFs with NH3 at 600 ºC- The as-prepared CNFs have basic character as determined by the point of zero charge (PZC).

The adhesion of CNF after the different treatments was checked by maltreatment in ultrasound during 15 min, which is a conventional method to check adhesion resistance. To check the corrosion we used the standard test ASTM D3987-85. The treatment with diluted 1M HNO3 and with H2O2 led to an excellent adhesion (no release of particles in ultrasound test). Therefore, these treatments are suitable for the application of the composite to water treatment. The functionalisation with nitrogen in situ during growth is also suitable because the adhesion was excellent.

The corrosion resistance of the CNF/structured reactor composites, as determined by ASTM D3987-85, was even better than the initial structured reactor. Therefore, the CNF coating not only does not increase the corrosion but acts also as a protective coating. Therefore, the corrosion resistance was excellent for all the CNF/structured reactors.

Deposition of the active phase.

The active phase of Pd, Cu has been deposited on the different CNF/structured reactors by different methods such as:

1. Simple impregnation: This method led to good results for CNF/SMF and CNF/ACF but was not suitable for monoliths. CSIC deposited the metals on CNF/monoliths and this led to large particles of Pd (about 15 nm) and large Cu aggregates. Furthermore, the distribution of the metals along the length of the monolith was not uniform.
2. Impregnation of the metals by ion exchange: This method led to a better distribution of the metals along the length of the monoliths.
3. Impregnation of second metal (Cu) by electroless deposition or reductive deposition. This method brought about the deposition of Cu in the close proximity of the first metal Pd.

Characterisation of active phase using TEM and EXAFS of catalyst supported on CNF/CF

Two different carbon structures grown onto CF have been used as supports for the bimetallic Pd/Cu catalysts, namely CNF and CNP.

The bimetallic Pd/Cu catalysts were prepared in two steps and a brief summary is given below:

1. Impregnation of PdCl2, followed by drying at 100°C and calcination at 250°C
2. Impregnation of Cu(NO3)2 followed by drying at 100°C and calcination at 250°C.

Two samples, PdCu CNP GF CNP and PdCu CNP GF CNF, have been examined by transmission electron microscopy (TEM). Small sections of each material were cut from the bulk and dispersed in isopropanol. Drops of the resulting suspension were placed onto an amorphous carbon support before examination. This allows the structure of the carbon fibre and metal particle components of the samples to be examined.

Images recorded in TEM and scanning (S)TEM mode show the structure of the support material and the distribution of the metal. Dark-field STEM images show strong contrast from the latter.

The basic structures of the components of the support are clearly seen. The metal component shows some separation of Pd and Cu, some aggregation of particles and particle sizes of usually a few nanometers, but also some larger particles especially containing Pd. The composition of apparently individual particles can be measured by energy-dispersive x-ray spectroscopy (EDS), although in the Pd-Cu/CNF CF sample, EDS analysis was complicated due to fluorescence from the Cu TEM support grid. For the Pd-Cu/CNP CF sample the measurements were performed using a Ti grid to avoid the fluorescence interference.

The metal particles are predominantly around 5 nm in sise. Some agglomeration of particles can be seen and it is obvious from EDS mapping that Cu and Pd particles co-exists in close vicinity. Usually the EDS spectra of particles show contribution from both Pd and Cu. However, for the large particles, larger than 20 nm, only Pd is detected.

The TEM images confirm the unusually high Cu dispersion in these catalysts which is also reflected in the EXAFs analysis below. The high Cu dispersion is most likely due to Pd being deposited first on the carbon support and hence improving and stabilising the anchoring of Cu particles.

The Pd nanoparticles appear to have spherical shape (Figure 8). The typical size is less than 5 nm. Cu particles are in the same sise-range, but tend to have more irregular shapes. The metal particles (dark spots) may be either Cu or Pd or both. The bright-field TEM contrast is not able to distinguish between the two metals.

X-ray absorption spectroscopy (XAS)

The XAS data were collected at the Swiss-Norwegian Beamline (SNBL), European Synchrotron Radiation Facility (ESRF) in France. The ESRF provide electron beam energies of 6 GeV and a maximum current of 200mA. Spectra were collected in transmission mode using a double Si(111) crystal monochromator for the Cu - (8979 eV) and Pd K edges (24350 eV).

A stainless steel in situ cell was designed for the experiment. The cell can be used both for gas- and liquid phase reactions and is equipped with Kapton windows and carbon gaskets. The cell was heated by cartridge heaters located at the upper- and lower part of the cell and the power was provided by a Delta power supply. The experiments were performed in two steps: First the catalysts were reduced by heating to 150°C at a heating rate of 2°C/min in 25 ml/min 5%H2/He. The temperature was held for 2 hours prior to cooling down to ambient temperature in He. In the second step the liquid was continuously saturated by flowing 110 ml/min CO2 and 110 ml/min 5%H2/He through the solution before the gases were send to the exhaust system. The nitrate-containing water was pumped through the cell at a rate of 5 ml/min. The liquid went through a closed loop during the reaction. The reaction was performed at ambient pressure and temperature.

EXAFS data were collected prior to and after the reduction/reaction step as well as in between. Continuous EXAFS scans of the Pd and Cu edges were performed during this process.

XANES results obtained during in situ reduction

A linear combination routine was used for determination of the oxidation states of Cu and Pd in the samples. The references used are Cu foil, Cu2O and CuO for the Cu K-edge and Pd foil and PdO for the Pd K-edge, respectively. The Cu is present as CuO in the as prepared sample.

The reduction of Cu starts immediately when the temperature is increased before it levels out at around 50°C and the degree of reduction stays fairly constant until 90°C. After 90°C the reduction starts to further increase. At the target temperature the Cu is present in three different states; approximately 83% Cu(0), around 12% Cu2O and approximately 5% CuO. The values are not changing during the two hours duration of the reduction.

The palladium is partially reduced in the as-prepared sample. The degree of reduction (35% Pd(0)) stays constant until the temperature reach approximately 70°C. The Pd is completely reduced at temperatures above 100°C.

EXAFS results obtained during in situ reduction

EXAFS analyses give information about the coordination number, distances and disorder. The coordination number can be used to obtain an average particle sise, depending that the value is lower than the bulk. Fourier transformation was performed on the Pd and Cu K-edge spectra in the 3-15Å region in k-space.

The results of the EXAFS analyses show that the Pd in the as-prepared catalysts is a mixture of Pd metal and PdO. The first and third peak corresponds to the Pd-O and Pd-Pd bonds in the PdO. The second peak with a distance of 2.76 Å is from Pd metal. The coordination number obtained for the PdO is lower than the bulk PdO, 4 and 12 respectively. This indicates that the PdO particles are smaller compared to the bulk phase. Also, the coordination number obtained for the metallic Pd is lower than the tabulated value of 12 for the first shell with a distance of 2.75 Å of the bulk phase of Pd(0). This corresponds to a Pd particle size obtained from XAS in the range 2 to 3 nm. The Pd is fully reduced after the reduction procedure and also after the reaction no changes in the Pd oxidation state can be observed. This indicates that the reaction conditions do not lead to oxidation of the Pd. Comparison of the coordination numbers of pre-reduced catalyst and catalyst after the reaction reveal that it is only minor changes in the Pd surroundings. A slight increase in the Pd-Pd coordination number is an indication of sintering, but the values are within the experimental error limit.

Pd is fully reduced after the reduction procedure and no changes in the Pd oxidation state can be observed after the reaction. Comparison of the coordination numbers of pre-reduced catalyst and catalyst after the reaction reveal that it is only minor changes in the Pd surroundings. A slight increase in the Pd-Pd coordination number is an indication of sintering.

The distances obtained for the first and second peak are in agreement with the CuO model component. Also, the coordination number of the first Cu-O shell is the same as the model within in the error limit of the technique. The value of the Cu-Cu shell is lower compared to the reference which is four. This indicates small particles, which is also confirmed by the TEM images.

In situ reaction study

The Cu(I) fraction increases at the expense of metallic Cu when the nitrate containing water was introduced to the catalyst. Only small changes can be observed in the CuO (Cu(II)) fraction. The ratio of Cu(0)/Cu(I) seem to become stable after approximately 100 min.

Characterisation of Pd-Cu supported on carbon with CNP structure (Pd-Cu/CNP-CF)

The Cu deposited on the CNP structure was partially reduced already at room temperature. Reduction treatment for 2 hours at 150°C only lead to a 10% higher degree of reduction of the Pd. Comparison of the reduction profiles for the two support shows that the support plays a role in the reduction of the metals.

Introduction of nitrate polluted water causes a complete oxidation of the catalyst within 5 minutes. The main Cu fraction in the sample is present as CuO.

The carbon structure influences the reduction behaviour of Cu. Cu is easier to reduce when it is deposited on the CNP carbon structure (CNP). The final degree of reduction is the same for both samples with different support materials; 80%Cu(0)/20%Cu(I) respectively. Cu is partially oxidised to Cu(I) in the most active catalyst (Pd-Cu/CNF-CF) whereas Pd is completely reduced and maintains in the reduced state during reaction. The average Pd particle size from XAS is 2 to 3 nm (compared to approximately 5 nm from TEM). The Cu particles seem to be of the same size as the Pd particles. No bimetallic interaction can be detected in the catalysts.

Partial conclusions of characterisation

The carbon structure influences the reduction behaviour of Cu. Cu is easier to reduce when it is deposited on the CNP carbon structure (CNP). The final degree of reduction is the same for both support materials, 80%Cu(0)/20%Cu(I) respectively.

1. Cu is partially oxidised to Cu(I) in the most active catalyst, Pd-Cu/CNF-CF
2. Pd is completely reduced and maintains in the reduced state during reaction
3. the average Pd particle size from EXAFS is 2-3 nm, with some larger particles
4. the Cu seems to have similar particle size to the small Pd particles. This is confirmed both by EXAFS coordination numbers and TEM images
6. no bimetallic interaction can be detected in the catalysts, albeit Cu and Pd are mostly detected in the EDS signal from the small particles. Large particles contain Pd only and no pure Cu.

Characterisation by STEM of PdCu catalyst on CNF/SMF and CNF/ACF and catalyst for nitrate/bromate reduction

Due to the STEM technique, bimetallic nanoparticles are much more contrasted against the carbon support. Moreover, sometimes this technique allows discriminating between different metals, because the contrast is a function of the atomic number.

The (5%Pd+2.5%Cu)/ACF catalyst metal nanoparticles have a quite broad size distribution with an average diameter of around 7.7 nm. EDX analysis revealed that the larger particles are composed of both Pd and Cu. The smaller particles seem to be only Pd. It can be explained by the fact that palladium deposited on the first step of the material synthesis by adsorption of the precursor followed by its reduction is homogeneously distributed in the form of less than 2 nm nanoparticles through the ACF microporous network. On the second step of the synthesis, because of the diffusion limitations, copper is catalytically deposited rather on easily accessible Pd nanoparticles placed on the outer activated carbon fiber surface than on Pd nanoparticles seated within the ACF micropores.

Bimetallic nanoparticles in the (2%Pd+1%Sn)/ACF are smaller (d around 2.3 nm) than in the previous Pd-Cu sample. They have a quite narrow size distribution. EDX analysis revealed that both Pd and Sn are homogeneously distributed over the ACF support. All particles are composed of both Pd and Sn. It is worth to remind that tin was deposited on the second step of the synthesis through the precursor impregnation followed by H2 reduction and hence it was able to penetrate into the ACF microporous network and to reach all Pd nanoparticles.

Monodispersed metal nanoparticles (d around 2.0 nm) are uniformly distributed over the CNF surface. EDX analysis confirmed that nanoparticles are composed of both Pd and Cu. The uniform contrast inside each nanoparticle seems to exclude that the metals are segregated. They form a bimetallic mixture with homogeneous distribution of the two metals.

The (0.3%Pd+0.15%Sn)/5%CNF/SMF catalyst is also very homogeneous with metal nanoparticles densely dispersed on the CNF surface. In this case nanoparticles have bimodal size distribution with peaks at 2.0 and 7.0 nm. EDX analysis revealed that the larger particles are composed of both Pd and Sn. At the same time, it is clearly seen from the high magnification image that many nanoparticles have a core-shell structure. EDX analysis of core and shell indicated that the core consists of Pd whereas the shell (around 2 nm thickness) consists of Sn. The smaller particles have the same contrast than the shell regions and could be ascribed to pure Sn.

In the catalytic reduction of nitrates, the catalyst consisting of Pd-Sn showed higher selectivity to N2 than the Pd-Cu which is related to the specific interaction between Pd and Sn which leads to a core-shell structure.

Catalytic batch test in conventional reactors (WP6)

The objectives proposed for WP6 included the testing of the monolithic catalysts prepared by the project partners in conventional slurry tank or trickled bed reactors. As to obtain the material in appropriate powder form, the supporting monoliths were to be crushed or scraped and the resulting powder was to be tested in the catalytic reduction or advanced oxidation processes.

Multiphase structured reactor hydrodynamics and reaction simulation (WP7)

Process modelling of catalytic nitrate hydrogenation was performed by combing the computational fluid dynamics (CFD) hydrodynamics model and mechanistic reaction kinetics and mass transfer.

Different scenarios regarding flow patterns corresponding to different mass transfer efficiencies and structures of reactor channels were simulated by using this combined model.

Different process working methods, i.e. single-pass and recycling-loop, were suggested based on simulation results. A single pass reactor may be feasible for low concentrations of nitrates, however reaction selectivity with the current catalysts is not high enough. For the higher concentrations of the nitrates, corresponding to the industrial water effluent, e.g. from MEL Chemicals, a loop reactor is suggested based on the simulation results.

These results will be validated with the new data of reduction of nitrates in highly concentrated solutions and with new catalysts based on carbon nanofibre supports and presented in the final report.

The combined model can be further tuned based on the properties of different catalysts (corresponding to different mass transfer coefficients and different reaction parameters) for novel process design.

The three phase contact (gas-liquid-solid catalyst) established using the so called Taylor flow enhances the mass transfer with respect to bifasic contact established in conventional packed beds.

Using monoliths of a cell density of 400 cpsi the gas distribution through the channel is not always optimal and sometimes big bubbles are formed due to the coalescence of small bubbles. The gas distribution is better using monoliths of a la lower cell density such as of 64 cpsi.

Conclusion

The CNF/structured reactors outperform conventional reactors as long as good mass transfer is guaranteed. For the CNF/SMF and CNF/CF structured catalyst this occurs for almost all experimental conditions all the ranges due to the good mixing of the gas/liquid catalyst. For the CNF/monoliths, the enhanced performance is found when taylor flow (or segmented flow) is stablished with a good distribution of gas and liquid through the monoliths channels. This distribution is better when using monoliths of cell density around 64 cpsi.

Catalytic test of structured reactor in reduction in continuo (WP9)

The objective of this work package is to test the monolithic catalyst in real conditions and the optimisation of process parameters in the reduction with H2 of nitrates, bromates and perchlorates in water. The objective is to study the behaviour of the catalyst in the set ups implemented in deliverable 8.1 for reduction reactions in continuous during long term operation using waters of different origin such as natural water, concentrate water and industrial water.

The main results are the following:

1. The catalysts tested for the nitrates reduction are stable and active at least during 55 hours of reaction. The conversion obtained with the different catalysts is constant in the first hours of reaction and only a slow catalyst deactivation is observed during 24 hours with the catalyst with the lowest Pd content or in the reaction with a lower catalyst mass. Nevertheless, it seems that after this partial deactivation the catalyst activity remains stable until the end of the reaction. The selectivity of the different catalysts tested for the nitrate reduction is constant along the reaction, obtaining a better selectivity with the catalysts containing tin and palladium than with the catalysts containing copper and palladium.
2. The Pd/Sn catalysts studied are efficient and stable in reducing the nitrates present in an industrial wastewater with a high conductivity and with a high nitrate concentration in batch and in continuous reaction.
3. The only tested catalyst that showed some activity for the perchlorates reduction is the catalyst containing 5 % Pd-5 %Re supported on a CNF/SMF support, that present a 25 % of perchlorate conversion.
4. Pd-catalysts supported on active CF or monoliths are active for bromate reduction in batch reactions. Nevertheless, the best results are obtained with the catalyst 0.3%Pd/5%CNF/SMFinc.
5. This catalyst can remove the bromates present in polluted water but the hydrogen partial pressure and the contact time influence significantly the catalyst activity and should be controlled for an adequate performance.
6. Pd-Sn catalyst present some activity for the reduction of the nitrates and bromates present both in the same polluted water but nitrates only are reduced after bromates reduction.

To summarise, a catalyst based on Pd-Sn on CNF/SMF is the most active and selective to N2 (80 % selectivity). However, the production of NH3 renders the process non feasible commercially for drinking water treatment. It is more feasible the application to industrial concentrated waters such as industrial wastewaters (MEL chemicals) with a mix of pollutants because the structured catalyst showed a high activity and no deactivation during long term experiments. This makes the comparison with other technologies favourable for catalytic reduction such as adsorption.

There is a need to reduce Bromate after ozonation process. The best approach is to reduce its formation. When this is not possible, another technology for the reduction of bromate must be applied.

The comparison of catalytic reduction with other existing technologies for the reduction of bromates indicates that the catalytic reduction using structured catalyst is a promising technology for the reduction of bromates (also in combination with nitrates). Pd structured catalysts supported on active CF or monoliths are active for bromate reduction, obtaining the best results with the catalyst 0.3%Pd/5%CNF/SMFinc. This catalyst can remove the bromates present in natural and industrial water but the water matrix and the hydrogen partial pressure influence significantly the catalyst activity and should be controlled for an adequate performance.

Catalytic test of structured reactor in ozonation in continuo (WP9)

The objective is testing the structured catalysts in the ozonation of the five selected organic micro pollutants during long term operation in continuous and to study the ozonation byproducts and its toxicity. The five selected micro-pollutants were atrazine, metolachlor, erythromycin, bezafibrate and nonylphenol. These pollutants were spiked in different concentrations simulating natural water and concentrated water. Among the catalytic materials tested by the partners, a monolith coated with CNF was identified as a promising catalyst for the catalytic ozonation experiment.

The experiments were carried out in parallel at VERI and Porto University in continuos using the CNF/monolithic catalyst. Porto University put the emphasis in studying ozonation byproducts and its toxicity. After testing the monolithic reactor in ozonation using oxalic acid as a model compound in previous period, it is important to tests with the target pollutants identified within the project. Preliminary experiments, using water close to the expected actual conditions, were run at UP, with water prepared and later analysed by Veolia. Later, Veolia conducted additional experiments on a pilot plant specially designed for the project, using conditions similar to those found in real conditions. At UP the focus was put on the study of the potential to fully mineralise the pollutants, using catalytic ozonation under semi-batch operation and on the optimisation of a system to replicate these tests in continuous operation.

In this study, VERI selected the target pollutants representatives of water pollution, studied the elimination of these target pollutants with concentrations representative of real water effluents and designed and built a small-scale pilot allowing single ozonation, catalytic ozonation and single adsorption experiments to be performed simultaneously.

Removal of initial pollutant in real conditions in continuous

In this study, catalytic ozonation is compared to single ozonation in order to highlight any eventual benefit of using a catalyst. VERI tested the monolithic catalyst in closed-industrial experimental conditions using two different water matrixes: natural and concentrate. Different configurations were tested allowing us to adjust the experimental conditions that need to be used for industrial applications in order to achieve highest removal of the pollutants. It consisted in adjusting multiple parameters such as flow rate, contact time, volumes, ozone demand.

Initially, the results obtained indicated that the monoliths does not provide any significant added value in terms of the initial pollutant removal in the catalytic ozonation of the selected pollutants compared to single ozonation experiments, regardless of the water matrix used. These results are related to the experimental conditions used here and using the selected molecules. It was recognised that it is not only interesting to follow the initial pollutant but also to study the byproduct formation and its toxicity, because previous results pointed out that catalytic ozonation is more effective in the removal of intermediates of initial pollutant decomposition than simple ozonation and some of those intermediates may be more toxic than the initial pollutant.

Study of byproducts formation and its toxicity in the ozonation of micropollutants in continuous.

To assess the effect of catalytic ozonation on the removal of ozonation byproducts, i.e. degree of mineralisation of initial pollutant and evaluate the toxicity of byproducts and compare with simple ozonation, Porto university designed suitable experiments.

The application of a catalyst during the ozonation of the selected organic compounds generally improves their mineralisation. Although adsorption can play a significant role on the removal of the pollutant from solution, the same does not happen with the by-products formed. Also, during adsorption, mineralisation, which is the goal of the process, is not achieved.

In general, the inclusion of the catalyst does not improve the removal of the selected compounds. This was already known from previous experiments in semi-batch with the catalysts in powder form, as well from experiments made using real effluents by Veolia. However, it is noticeable that the catalytic ozonation experiments contributed to a relevant further mineralisation of the pollutants. A long term adsorption experiment was made using MTLC. After less than 1 hour the concentration of MTLC in the outlet was equal to the concentration in the feed solution, meaning that the adsorption on the catalyst in steady state is negligible. This performance can be considered analogous to the other pollutants.

Regarding the toxicity of the effluent, in general, the ozonation treatment process increases the inhibition of activity of the lumniscient bacteria used for the Microtox tests. However, it is important to underline that the toxicity degree was lower in continuous operation than that obtained in the semi-batch experiments, probably due to the different extent in the degradation chain obtained. The inclusion of a catalyst decreases the effluent toxicity in all the cases. Particularly, in the case of erythromycine and metolachlor, the catalytic ozonation effluent presents a value of toxicity slightly lower than the initial.

Thus, the application of a monolithic catalyst enhances the mineralisation of effluents containing the selected pollutants and reduces the toxicity when compared with the single ozonation process.

Performance of modified catalysts and processes

The addition of highly dispersed ceria to the surface of the multiwalled carbon nanotube (MWCNT) generally improves its catalytic activity. The addition of hydrogen peroxide (H2O2) to the solution also seems to improve the ability to achieve larger mineralisation values. The main role of H2O2 is on the decomposition of ozone into radicals which will react with the recalcitrant by-products formed during reaction. This means the removal of the parent compound is slower when H2O2 is present since it competes with the pollutants to react with the dissolved ozone.

The ozonation of the micropollutants seems to create compounds which are more toxic than the parent compound. However, when a catalyst is present, the toxicity is smaller. This should be due to the formation of less toxic compound, e.g. organic acids, which are formed at the end of the degradation chain of the parent compounds.

Although mineralisation is improved with catalytic ozonation, total mineralisation is no yet achieved. The generally reduced toxicity using catalytic ozonation in comparison to ozonation alone is also a sign of the potential of this process.

Contrary to what was initially expected, the nitrogen (N) content on the CNF does not necessarily influence positively the activity of the catalysts. Additionally, the increase on the loading of carbon on the catalysts? surface does not correspond to a proportional increase on the activity, probably due to difficult of access to the lower layers of the CNF layer, due to an increase on the thickness of this same layer.

The increase of the contact time with the catalyst further improved the removal of total organic carbon (TOC) from the solution, as well as the toxicity of the system's outlet. In addition, it has been shown that the application of the honeycomb catalysts in a triphasic system under Taylor flow results in improved mineralisation degrees when compared to a biphasic system. Taylor flow is known to improve the mass transfer from the gas phase to the solid phase, thus enhancing the catalytic decomposition of ozone into highly reactive radicals.

The system to test the ozonation of organic pollutants in triphasic continuous operation was optimised, with a significant increase in activity when using oxalic acid as a model compound and compared to the biphasic system.

Comparative study between NCM/structured reactors and with conventional supports and reactors (WP10)

The objective is comparing the structured catalytic reactors implemented in this project with conventional catalyst supports and reactors. To this end, a exhaustive literature revision has been made since the beginning of the project to compare our catalyst with catalyst used in the literature. In addition, the consortium has used conventional catalyst supports such as activated carbon fibres (ACF), activated carbon, alumina, etc. as a bench mark for the different reactions. Conventional batch reactors and slurry reactors have been used as a benchmark to compare with the structured reactors implemented in this project.

Structured fibrous catalysts are very suitable for natural water denitrification due to the rational organisation of catalytic bed resulting in a low pressure drop, highly homogeneous distribution of the fluid flow and reduced mass transport limitations. The mesoporous CNF/SMF support allows avoiding internal diffusion limitations and provides a full accessibility to the active phase.

Pressure drop measurements were performed in order to see whether the monolith create a pressure drop or not. Different configurations of monoliths were tested (i.e. one or two monoliths in series) and different flow rate values were tested. To this aim, two different types of differential pressure apparatus were used allowing us to obtain different values to be compared. A pressure drop value of 0.3 mbar was measured indicating that the monoliths channels are not problematic for the hydrodynamic flow to operate.

This pressure drop was compared with a conventional reactor (paked bed of particles) and the packed bed exerts an order of magnitude higher backpressure (3 mbar) than the structured packings. The CNF/structured reactors outperform conventional reactors as long as good mass transfer is guaranteed. For the CNF/SMF and CNF/CF structured catalyst this occurs for almost all experimental conditions all the ranges due to the good mixing of the gas/liquid catalyst. For the CNF/monoliths, the enhanced performance is found when taylor flow (or segmented flow) is established with a good distribution of gas and liquid through the monoliths channels. This distribution is better when using monoliths of cell density around 64 cpsi.

Comparison with conventional supports

Pd-Sn catalyst supported on a CNF/SMF catalyst shows lower selectivity to undesired ammonia (NH3) and nitrite than a conventional support (ACF). In conclusion, structured fibrous catalysts are very suitable for natural water denitrification due to the rational organisation of catalytic bed resulting in a low pressure drop, highly homogeneous distribution of the fluid flow and reduced mass transport limitations. The mesoporous CNF/SMF support allows avoiding internal diffusion limitations and provides a full accessibility to the active phase.

Comparison with conventional reactors

Comparison of structured reactor with conventional reactor (batch) in reduction reactions

From the experiments in structured reactors in continuous and those in batch mode both for nitrate and bromate reduction, it can be concluded that structured catalysts containing CNFs are more active than conventional catalysts for the bromate and for the nitrate reduction reactions. These catalysts can be used in continuous structured reactors as CSTR or PFR, with no decrease of activity with respect to the batch reactor. This indicates that there is no diffusional limitation in continuous experiments and the process benefits of been in continuous and with low pressure drop. These results validate the use of the continuous structured reactors for these reduction reactions.

Comparison of structured reactor with conventional reactor (batch) used for ozonation reactions and with non-catalytic ozonation

The honeycomb monolith catalysts were selected to be tested in the continuous ozonation of oxalic acid. The dimensionless removal obtained with these catalysts is compared to that obtained using activated carbon in a packed bed under the same experimental conditions.

It is very clear that catalytic ozonation, when compared to the conventional ozonation treatment, significantly improves the mineralisation of the pollutants here considered. In addition, it can be seen that the application of the system in triphasic improves the performance due to the hydraulic regime, Taylor flow, which enhances the mass transfer between the phases (see Table 5). Furthermore, the increase of the contact time with the catalyst, by adding more monoliths in series, improved the mineralisation of the pollutants.

Potential impact:

The project has resulted in the direct employment of 15.5 full time equivalent (person working fulltime for a year). Among the persons involved in the project 14 are female and 17 are male. Therefore, there are no large differences in terms of gender.

The project has given rise to significant number of educational activities such as one summer school, open days, master student projects and three PhD thesis.

The technology developed has implications for the cleaning of pollution in waters to comply with European legislation. The technology will save energy and material resources in the cleaning process of companies. Therefore, it has the potential to contribute to make industry more competitive and less contaminant.

The project has given rise to significant number of educational activities such as four PhD Thesis, about eight master projects, summer school and open days at Universities once a year.

The dissemination activities so far include 15 publications in peer reviewed journals of high impact index and 46 oral and poster communications in conferences and three PhD thesis defences. The details of these dissemination activities are given below.

Furthermore it has been coverage in international press (The Guardian), an interview for Commission brochure and website.

List of websites:

http://www.monacat.eu

Webmaster:

Enrique Garcia Bordejé, c/ Miguel Luesma Castán 4, 50018 Zaragoza, Spain