Preparation of TiO 2-Coated Polyester Fiber Filter by Spray-Coating and Its Photocatalytic Degradation of Gaseous Formaldehyde

Heterogeneous photocatalytic oxidation (PCO) has shown great potential for indoor air purification of gaseous pollutants. Research attention has been drawn to the synthesis of new functional photocatalysts, as well as to kinetic studies of the influences of various reaction parameters, e.g., relative humidity. Nevertheless, when applied for practical use, the coating method and coating stability of are also important factors that will directly affect the removal efficiency. In the present study, a simple and economical spray coating method was developed to effectively immobilize TiO2 nanoparticles onto a polyester fiber filter at low temperature. Colloidal silica was added as a binder in the coating suspension. PCO efficiency evaluation of the coating was carried out using gaseous formaldehyde. The results indicate that the formaldehyde removal rate is associated with the amount of binder added. The highest removal rate was achieved for a coating with the TiO2 to binder (SiO2 equivalent) mass ratio of 1:1, with the results supported by XRF, SEM, FTIR and BET analyses. As compared to the conventional dip coating method, the spray-coated sample showed much higher PCO efficiency and stability, which may be mainly attributed to the more uniform dispersion of the catalyst and the stronger binding formed during high pressure spray coating. This coating method has great potential for large-scale applications of immobilized photocatalyst for indoor air purification.


INTRODUCTION
Indoor air pollution has become one of the top environmental risks to human health due to its significant contribution to human exposure to hazardous contaminants (Bruce et al., 2000;Bardana, 2001;Franklin, 2007;Lai et al., 2010;Hwang et al., 2011).The reasons for this include: people spend the majority of their time indoors; there is a wide and varied range of emission sources indoors; and for certain kinds of pollutant, e.g., formaldehyde and environmental tobacco smoke (ETS), the concentration levels are always higher in indoors compared with outdoors.Various of control technologies, such as filtration (Klimov et al., 2009), adsorption (Jee et al., 2005), ionization (Nishikawa and Nojima, 2001), ozone oxidation (Huang et al., 2012), etc., have been used for indoor air purification of different pollutants.
The application of heterogeneous photocatalytic oxidation (PCO) for air purification has been investigated intensively in the past two decades, particularly in converting indoor air volatile organic compounds (VOCs) into less toxic minerals, carbon dioxide and water (Zhao and Yang, 2003;Hodgson et al., 2007).This technique carries certain advantages such as: utilizing UV and/or visible light as initialization energy; non-selectivity of organic pollutants; potential to complete mineralization into CO 2 and H 2 O; and persistent effectiveness throughout the lifetime of the photocatalyst under ambient temperature.
As one of the best known carcinogenic VOCs, formaldehyde has been commonly identified in the typical indoor environment for its wide application in building materials and household products.In particular, the compressed wood products which use urea-formaldehyde (UF) resin as adhesive are one of the major emission sources indoors (Panagopoulos et al., 2011).Formaldehyde is also a by-product from cigarette smoking and fuel-burning (Daisey et al., 1994;Crump et al., 1997;Hodgson et al., 2000;Hippelein, 2004;Bari et al., 2011).Despite, high concentration levels of formaldehyde are also found in outdoor urban environment (Wang et al., 2010).Using heterogeneous PCO for the degradation of formaldehyde has been considered as a promising technique.Many previous studies focused on the degradation mechanism and limiting factors, such as relative humidity, initial pollutant concentration, light intensity, etc. (Peral and Ollis, 1992;Obee and Brown, 1995;Obee, 1996;Noguchi et al., 1998).In the past decade, more concerns have been raised on the effectiveness of photocatalyst immobilization for its potential on large scale application (Paschoalino et al., 2006;Žabová and Dvořák, 2009;Paz, 2010).
Catalyst immobilization is the key for successful implementation of heterogeneous photocatalytic oxidation (PCO) of indoor air pollutants.Substrate selection and effective coating method are two crucial factors in catalyst immobilization.Suitable substrates should not only have stable physical and chemical properties but also can provide large surface area for the catalyst to attach on.A desirable coating method should meet the following requirements: provides stable and durable coating; provides effective and sufficient contact between the coating and the pollutants; non-selective of different substrates; easy and energyefficient; economical and suitable for large-scale application.
Some commonly used substrates such as glass, aluminum, and stainless steel etc. are suitable for high temperature annealing treatment after coating.However, the stiffness and weight of the materials impose limitations for large-scale implementation.Textile-structured polymer materials, such as polyester (Mejía et al., 2010), cellulose acetate (Coronado et al., 2008), cotton textiles (Yuranova et al., 2006), and wool-polyamide (Bozzi et al., 2005), have also been studied in recent years for their low cost, flexibility and large surface area.Polyester, commonly referring to polyethylene terephthalate (PET), is one of the most stable and widely used artificial fabrics in the industry.In recent years, TiO 2 coated polyester materials have been studied for the application on indoor air purification (Dong et al., 2007;Park and Na, 2008), disinfection and anti-bacteria (Paschoalino and Jardim, 2008;Mihailovic et al., 2010), as well as self-cleaning surface (Sawada et al., 2003).Most of the previous works were focusing on the degradation/elimination efficiency rather than on the stability and durability of the coatings.
Sol-gel dip coating is a widely used immobilization method for fix-bed photocatalysis (McMurray et al., 2004).Although the performance of sol-gel dip coating is satisfactory in terms of coating stability, the high temperature (typically higher than 400°C) heat treatment step for annealing after coating is not applicable to the textile-structured substrates without losing their physical and chemical integrity (Lim et al., 2009).Other coating methods, such as spin coating, electrochemical deposition, electrophoretic coating, and sputtering etc., usually have complicated steps to achieve the preferable film thickness and are also costly and difficult for practical application (Byrne et al., 1998;Kasanen et al., 2011).
Large-scale application of PCO for air purification can be achieved through direct coating of commercial photocatalysts such as Degussa P25 TiO 2 (Paz, 2010).When using direct coating, the primary issue of interest is the stability of the coating.Poor adhesions have been reported on glass substrate by direct coating method such as dip coating or spin coating (Mills et al., 2002;Keshmiri et al., 2004), where even slight mechanical abrasion may remove the photocatalyst from the substrate surface.Thus, a binding agent is usually necessary for direct coating in order to form solid adhesion between catalyst and substrate.Either organic polymer or inorganic binding materials have been used for photocatalyst immobilization (Du et al., 2008;Lim et al., 2009;Henderson, 2011).However, the effect of binder amount on the PCO activity and stability of the coating is seldom discussed.The present study attempts to develop a simple and economical spray coating method to effectively immobilize P25 TiO 2 (Degussa) nanoparticle onto polyester fiber substrate.Colloidal silica (Ludox AS-40), which is a typical industrial binder but seldom be reported for being used as binder in direct coating of photocatalyst, was added into the coating suspension as a binding agent to increase the adherence between TiO 2 and the substrate.Without high temperature heat treatment, the coatings were finally cured at 60°C for stabilization.PCO of gaseous formaldehyde was carried out on the spray-coated samples.Different binder dosages were compared in order to find out the most stable and effective coating combination.One sample prepared by dip coating was also evaluated for comparison purpose.Mechanical stability test was taken to evaluate the effects of different methods (spray coating vs. dip coating) on the stability and durability of the coating.

EXPERIMENTAL METHODS
The polyester substrates used in this study are nonwoven dry polyester fabrics, with a typical filter thickness of 10 mm and overall packing density of 0.055 g/cm 3 .The polyester fiber has diameter of about 25-40 μm and fiber density of about 0.24-0.38dtex.They are usually used as air filter media in the HVAC system for particle filtration and are provided by a local supplier (BioFit, Singapore).Titanium dioxide TiO 2 (Degussa P25) was purchased from Evonik Degussa (Germany).Ludox AS-40 colloidal silica was purchased from Sigma-Aldrich (USA).All other chemicals used for the preparation of spray/dip-coated samples as well as for photocatalytic activity test were chromatography grade and used without further purification as received (Sigma-Aldrich, USA).Milli-Q water was used as solvent at all time.
The polyester substrates were cut into the size of roughly 100  50  3 mm.All the substrates were washed with 0.5% nonionic washing agent Felosan RG-N (Bezema, Switzerland) at liquid-to-fabric ratio of 45:1 in an ultrasonic washer (Kudos, China) for 15 min and rinsed with Milli-Q water followed by sonication for 15 min for three times.Pure acetone was used in the final washing stage, and the substrates were dried in pure nitrogen gas flow at room temperature.
A Navite F-75G spray gun (Navite, China) was used for spray coating, with nitrogen as the carrier gas under a pressure of 65 psi.The spray distance from the nozzle exit to the substrate surface was 150 mm.To achieve uniform coating on the bulk body of the substrate, the substrate was spray-coated on both sides.Each side was sprayed evenly for 10 rounds.An interval of 15 min was taken every two rounds using a blow drier to evaporate water.The spraycoated substrates were finally cured at 60°C for 3 hr to stabilize the coatings.The substrate was weighed by a digital analytical balance (readability of 0.0001 g, Mettler-Toledo, Switzerland) before and after coating.
One dip-coated polyester sample (identified hereafter as sample D-2) with TiO 2 :binder (SiO 2 equivalent) mass ratio of 1:1 was also prepared.The substrate pre-treatment and preparation of coating suspension are identical to those used in preparing the spray coating samples.The substrate was dipped into the coating solution for 30 min with stirring and then dried by a blow drier for 15 min.Several rounds of dipping-drying process were taken until the amount of coating reached the same value as that of the spray-coated sample S-2.The coating was then also stabilized at 60°C for 3 hr.
Surface characterization of the spray/dip-coated samples was carried out by a JEOL JSM-5410LV (JEOL, Japan) scanning electron microscopy (SEM) with energy dispersive spectrometer (EDS) analysis.The amounts of TiO 2 and SiO 2 contents on the substrates were derived by gravimetric measurement together with X-ray florescence (XRF) analysis on a Bruker AXS SRS 3400 (Bruker, Germany) Xray florescence spectrometer with Rh source.Infrared spectroscopy was performed on a Perkin Elmer Spectrum 100 Optica (Perkin Elmer, USA) Fourier Transform Infrared Spectrometer (FTIR).The spectrum was taken in the range of 500-4000 cm −1 .BET surface areas of the samples were determined by nitrogen physisorption on an Autosorb-1 system (Quantachrome, USA).Table 1 shows the specific experimental purposes of the above mentioned instruments in the present study.
The photocatalytic oxidation (PCO) of gaseous formaldehyde was performed on a fix-bed photo-reactor which was designed according to the ISO standard 22197-1:200722197-1: (ISO, 2007) ) with special modifications based on the application in this study (see Fig. 1).The reactor body is made of polished stainless steel 316 L (Exel-Mitsui, Singapore).An optical window made of GE214 fused quartz glass   2. The detection limit of the analytical procedure (GC-FID) was about 358 pg/injection (1 μL).This was the amount of the analyte which gave a peak whose height was about 3 times the height of the peak given by the residual formaldehyde derivative in a typical blank front section of the sampling tube.The detection limit of the overall procedure (derivatization + desorption + analysis) then was determined as 358 ng per sample, which was equivalent to approximately 0.2 ppm in a 1.5 L gas sample at 24°C.This was the smallest quantitated amounts of analyte spiked on the front section of the sampling tube, with an average recovery of 95.8% ± 2.4% (n = 6).Before switching on the UV light lamps, the testing gas flow would continuously pass through the photo-reactors to reach the steady state, which was indicated by reaching the same values in inlet (C In ) and outlet (C Out ) concentrations.According to the preliminary test result, this process usually took about 1.5 to 2 hr in our current system.Therefore, in this present study, UV light lamps were turned on after 2 hr of system stabilization.The PCO removal rate of formaldehyde, η (%), was calculated as: (%) 100 The PCO experimental conditions are shown in Table 3.Four repeatable PCO experimental cycles were taken for each tested sample.The initial concentrations of formaldehyde averaged at 24.6 ± 2.8 ppm for all PCO tests, although the typical formaldehyde concentrations in a problem building range from 0.5 to 2 ppm.Since the baseline concentration of formaldehyde in the testing environment is typically 0.2-0.3ppm, a concentration that is about 100 times higher than the environmental baseline was chosen to avoid the noise impact.In addition, higher concentration also help to shorten the sampling time needed in our analytical process so that a higher temporal resolution can be obtained.
The mechanical stability test then was taken for each sample.An OS-20 Orbital Shaker (Boeco, Germany) was used for testing the mechanical stability of the coatings on

Gravimetric and X-Ray Fluorescence (XRF) Measurement
To compare the photocatalytic activities of the coated polyester samples, TiO 2 and SiO 2 (binder) loading amounts on the overall bulk of the substrates should be confirmed in the first place by gravimetric measurement and X-ray fluorescence spectroscopy (XRF).The overall substrate was first weighted by an analytical balance and then scanned in XRF to obtain the composition weight percentages of TiO 2 and SiO 2 component.The loading amount of each component then can be calculated by substrate weight × composition weight percentage.The results are shown in Table 4.For each tested sample, the values of the weight percentages (%) of TiO 2 and SiO 2 are the average of four repeatable XRF measurements taken on the whole substrate.
All five samples have comparable TiO 2 loading amount, with average of 0.24 ± 0.01 g.For the spray-coated samples, loading amount of SiO 2 increases from S-1 to S-4, which corresponds to the different binder dosages in the coating suspensions.The measured mass ratios of TiO 2 :SiO 2 for all five samples are approximately consistent with the theoretical values as 2:1, 1:1, 1:2 and 1:4 through S-1 to S-4.Consequently, the comparison of photocatalytic activities of all five freshly prepared (before stability test) samples was based on equivalent photocatalyst loading amount.

Scanning Electron Microscopy (SEM)
Figs. 3(a)-3(e) show the scanning electron microscopy of the four spray-coated samples (a-d) and the one dip-coated sample (e).Spray-coated sample S-2 in Fig. 3(b) shows the most uniform TiO 2 dispersion and less coagulation.The agglomerate size of TiO 2 attached on the polyester fiber increases with increasing binder loading amount.Spraycoated sample S-4 has the highest binder loading amount and also shows heavy coagulation in Fig. 3(d) which might limit the contact between TiO 2 and the reactant species and further affect the PCO efficiency.For the dip-coated sample D-2 in Fig. 3(e), a large agglomerate size can also be observed due to the coagulation formed in the dip-coating process, which affects the effective surface area of the coating and then further reduces its PCO activity.The following BET and PCO tests will verify the above observations.

Energy Dispersive Spectroscopy (EDS) Analysis
EDS test was carried out simultaneously with SEM analysis.Both Ti and Si signals were observed on the spectroscopy.The presence of TiO 2 and SiO 2 on the fiber surface was confirmed again.The average elemental percentages of Ti, Si are 3.97%, 6.50% on sample S-2, and  4.19%, 4.66% on sample D-2, respectively, which are correlated to the TiO 2 and SiO 2 loading amount in the whole bulk of the substrates measured by XRF.This surface elemental analysis may be used as an alternative when XRF is not available.

Infrared Spectroscopy of the Spray/Dip-Coated Samples
The IR spectra of the spray-coated sample S-2 and the dip-coated sample D-2 are shown in Fig. 4. For sample S-2, TiO 2 bands are observed in the range of 500-800 cm −1 with strong adsorption band appearing at 600 cm −1 representing of TiO 2 matrixes in six-fold coordination (Mohamed et al., 2002).The bands at 1060 cm −1 and 1080 cm −1 can be assigned to the asymmetric stretching and bending mode of the Si-O-Si banding (Shafei et al., 1995;Mohamed et al., 2002).A band at 950 cm −1 can be assigned to the stretching of Si-O species of the Ti-O-Si linkage which is formed by the inclusion of Ti 4+ into the SiO 2 matrixes (Yamashita et al., 1998).Combining all the information provided, it implies that TiO 2 species were embedded into SiO 2 matrixes through the high pressure spray coating approach.The bands at 1150-1240 are associated to the ester group vibrations in the polyester structure.For sample D-2, TiO 2 bands between 500 and 800 cm −1 , as well as the asymmetric stretching band of Si-O-Si at 1060 cm −1 are also observed (Shafei et al., 1995;Mohamed et al., 2002).However, there is no strong adsorption band which can represent the formation of Ti-O-Si linkage observed on the IR spectrum of sample D-2, indicating that the immobilization of TiO 2 on the substrate through SiO 2 binder by the dip coating method is not as strong as those using the spray coating method in this study.

Stability of the Spray/Dip-Coated Samples
The coating amounts on all five polyester samples were measured and compared before and after the stability test.
The result is shown in Table 4.The immobilized TiO 2 by spray coating has very good adhesion on the polyester fiber surface.After continuous shaking for 72 hr, the losses of TiO 2 content on the spray-coated sample S-2, S-3 and S-4 are all less than 3.6%.Sample S-1 has a relatively higher loss at 16.7%, which might mainly due to the insufficient binder as it has the lowest binder amount.The dip-coated sample D-2 has lost its TiO 2 significantly at 33.4% after the stability test, which indicates the poor adhesion of the coating on the polyester fiber surface using dip coating.This result again verifies the observation in the IR spectra that strong bond between photocatalyst and the binder was formed by spray coating, but not by dip coating.

Surface Area Measurement by BET
BET measurements of the spray/dip-coated samples are shown in Table 5. Spray-coated sample S-2 has the highest BET area of 95.10 m 2 /g.BET areas of spray-coated sample S-3 and S-4 dropped significantly to 64.56 m 2 /g and 47.30 m 2 /g, respectively.This result also confirms the observation on the SEM images that the excessive amount of binder causes larger agglomerate size which in turn leads to smaller effective surface area.Spray-coated sample S-1, which has the smallest binder loading, has a smaller BET area compared to sample S-2, indicating that certain amount of binder is preferred to generate uniform coating.
All the spray-coated samples have larger BET area than that of the dip-coated sample D-2 (49.97 m 2 /g), except for sample S-4 (47.30 m 2 /g) which has the largest binder loading.After the stability test, dip-coated sample D-2 has big decrease (about 14.9%) in surface area due to the significant loss of TiO 2 content; while spray-coated sample S-2 has only a small decrease in surface area (about 2.5%).

Photocatalytic Oxidation (PCO) of Formaldehyde
Fig. 5 shows the variation of formaldehyde PCO removal rate within 105 min reaction (irradiation) time by sample S-2.The purpose of this test was to figure out when the system would provide a stable degradation after light turned on.Four repeatable experimental cycles were taken under the same conditions.Within the first 15 min after light turned on, as the substrate sample had already adsorbed HCHO and reached saturation during dark condition, the difference between inlet and outlet concentrations was not obvious when HCHO continuously flowing through the substrate sample.From 15 to 60 min, the removal rate increased progressively when more active sites were available and slowed down when the reaction process approaching the stable state.After 90 min, the removal rate did not change further.Therefore, in this continuous flow system, removal efficiencies of different samples were compared after light turned on for 90 min.No evident degradation was observed in either the absence of TiO 2 coating (non-loaded substrate) or UV irradiation.
The PCO degradation efficiencies of formaldehyde by different spray/dip-coated samples before and after the stability test are shown in Fig. 6.The spray-coated sample S-2 has the highest formaldehyde removal rate both before (92.86% ± 2.58%) and after (91.96% ± 2.42%) the stability test.For all spray-coated samples, there is no obvious loss of photocatalytic activity after the stability test, except for sample S-1, which is mainly due to the loss of TiO 2 loading on sample S-1 after stability test (see Table 4), indicating that small binder dosage can not provide satisfactory adherence of TiO 2 on the polyester substrate.
For the dip-coated sample D-2, significant reduction of removal rate, from 75.28% ± 3.13% to 40.93% ± 3.64%, is observed after stability test due to the significant loss of TiO 2 loading (shown in Table 4).It is worth to note that although there is a slightly higher TiO 2 loading on sample D-2 than that on sample S-2 before the stability test, the large agglomerate size formed during the dip coating process restricts its photocatalytic activity, resulting in the lower PCO efficiency than that of sample S-2.One of the important findings in this study is the effect of binder dosage on the PCO activities of the spray-coated samples, which is clearly illustrated in Fig. 6.For the freshly prepared spray-coated samples with equivalent TiO 2 loading amount, formaldehyde removal rate increases in the beginning and then decreases accordingly when the binder dosage rises from sample S-1 to S-4.Formaldehyde removal rate of sample S-2 is higher than that of sample S-1 due to the more uniform dispersion of TiO 2 coating and smaller agglomerate size.Further increasing the binder dosage in sample S-3 and S-4 leads to large agglomerate size thus decreases the effective surface area as seen in SEM images and BET test results.In addition, large amounts of binder might also wrap TiO 2 and limit the contact between TiO 2 and formaldehyde molecules.Small surface area combined with less contact lead to the decreasing photocatalytic activities of sample S-3 and S-4.With respect to the coating methods, the spray coating method developed in the present  study shows great advantage over dip coating method in providing more stable and durable immobilized photocatalysts on the polyester fiber surface.

CONCLUSION
Effective immobilization of TiO 2 photocatalyst onto polyester fiber filter substrate was successfully conducted via spray coating approach.A strong dependence between the binder (SiO 2 ) dosage and the coating structure was observed.The optimal coating was achieved for TiO 2 :SiO 2 mass ratio of 1:1, which shows the most uniform TiO 2 dispersion, the smallest coating agglomerate size and the highest PCO removal rate of gaseous formaldehyde.Certain amount of binder provides strong adherence of TiO 2 on the polyester fiber surface and also improves the formation and dispersion of small coating agglomerate.Higher binder dosage increases the coating agglomerate size and further limits the contact between TiO 2 and formaldehyde molecules.Another sample prepared by conventional dip coating method with TiO 2 :SiO 2 mass ratio of 1:1 shows bigger agglomerate size, smaller surface area and lower photocatalytic activity in comparing with the corresponding spray-coated sample.
The immobilization of TiO 2 on polyester substrates by spray coating is stable and durable, with only little loss of TiO 2 loading and no obvious decline on PCO efficiency after a 72 hr mechanical stability test.The stability of the dip-coated sample is not satisfactory with significant loss of TiO 2 loading and according decrease of photocatalytic.The spray coating method developed in this study is costeffective and can be easily scaled up for the practical application of heterogeneous photocatalytic oxidation in indoor air purification.

Fig. 5 .
Fig. 5. Variation of PCO removal rate of formaldehyde within 105 min reaction time by sample S-2.

Fig. 6 .
Fig. 6.PCO removal rate of formaldehyde on different spray/dip-coated samples before and after stability test.

Table 1 .
Specific experimental purposes of the instruments used for sample characterization.

Table 2 .
GC-FID test conditions for formaldehyde derivative.

Table 4 .
Loading amounts of TiO 2 and SiO 2 on different samples before/after stability test.
a spray-coated.b dip-coated.c measured by analytical balance.d B/ST -Before stability test.e A/ST -After stability test.f measured by XRF in oxide testing mode for 4 repeatable measurements on the whole filter substrate.g calculated by total filter weight  weight percentage.h ratio of the average loading amount for each component.

Table 5 .
BET area of different spray/dip-coated samples.