Performance of ultraviolet photocatalytic oxidation for indoor air applications: Systematic experimental evaluation

Journal of Hazardous Materials 261 (2013) 130–138

Contents lists available at ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Performance of ultraviolet photocatalytic oxidation for indoor air applications: Systematic experimental evaluation Lexuan Zhong a , Fariborz Haghighat a,∗ , Chang-Seo Lee a , Ness Lakdawala b a b

Department of Building, Civil and Environmental Engineering, Concordia University, Montreal, Quebec H3G 1M8, Canada DECTRON International Inc., Montreal, Quebec, Canada

h i g h l i g h t s

g r a p h i c a l

a b s t r a c t

• An innovative experimental set-up • • • •

was designed and constructed for testing UV-PCO. Test methodologies were developed to examine UV-PCO air cleaners for VOCs removal. VOCs type, inlet concentration, flow rate, irradiance, and RH have influence on PCO. Gas-phase ozonation with a variety of compounds was examined in a duct system. Formation of by-products generated from incomplete conversion was investigated.

a r t i c l e

i n f o

Article history: Received 23 February 2013 Received in revised form 6 July 2013 Accepted 9 July 2013 Available online xxx Keywords: Ultra-violet photocatalytic oxidation (UV-PCO) Air filters Photolysis Ozonation By-products

a b s t r a c t Photocatalytic oxidation (PCO) is a promising technology that has potential to be applied in mechanically ventilated buildings to improve indoor air quality (IAQ). However, the major research studies were done in bench-top scale reactors under ideal reaction conditions. In addition, no study has been carried out on the investigation of the ozonation and photolysis effect using a pilot duct system. The objective of this study is the development of methodologies to evaluate the performance of PCO systems. A systematic parametric evaluation of the effects of various kinetic parameters, such as compound’s type, inlet concentration, airflow rate, light intensity, and relative humidity, was conducted, and new interpretations were provided from a fundamental analysis. In addition, the photolysis effect under vacuum ultraviolet (VUV) irradiation for a variety of volatile organic compounds (VOCs) was examined for the first time in a pilot duct system. The performance comparison of ultraviolet C (UVC)-PCO and VUV-PCO was also discussed due to the presence of ozone. Moreover, the formation of by-products generated with or without ozone generation was fully compared to evaluate the PCO technology. © 2013 Elsevier B.V. All rights reserved.

1. Introduction Indoor air quality (IAQ) has received enormous attention for its impact on occupants’ health, comfort, and work performance.

∗ Corresponding author. Tel.: +1 514 848 2424x3192; fax: +1 514 848 7965. E-mail addresses: [email protected], [email protected] (F. Haghighat). 0304-3894/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.jhazmat.2013.07.014

Traditional dilution ventilation has limitation on protecting building occupants against chemical and/or biological agents and reducing energy consumption. The technology of adsorption filtration, such as granular activated carbons and zeolites, has been widely studied due to their promising removal performance. However, high pressure drop and adsorbent regeneration are the main obstacles in the applications of such adsorption air cleaners. Heterogeneous photocatalytic oxidation (PCO), as a promising advanced oxidation technology, has been suggested as an

L. Zhong et al. / Journal of Hazardous Materials 261 (2013) 130–138

alternative and energy efficient method to improve IAQ through the photocatalytic degradation of volatile organic compounds (VOCs) [1]. In literature, numerous researches have been carried out to examine the PCO of gaseous contaminants [2–7]. However, majority of available PCO data are based on laboratory bench-top equipped with a small PCO reactor where experiments were carried out under ideal reaction conditions, i.e. low volumetric airflow rates, tested with one or a few VOCs, etc. Therefore, the test results based on the ideal experimental conditions could be problematic and may not be scaled up to predict the performance in full-scale systems. Although a few researches have explored the feasibility of the PCO technology applied in HVAC systems [8–10], these applications are mainly aimed at designing a portable PCO air cleaner employed in a closed room or a chamber. Few studies [11,12] aimed to explore the PCO performance as a single-pass way employed in an HVAC system, and only several key parameters, such as residence time, irradiance type, and filter type, have been examined. The application of ozone (O3 )-producing lamps in ultraviolet (UV)-PCO air cleaners inevitably introduces O3 into a duct system. O3 is a very powerful and strong oxidant. In the past half century, kinetics and mechanism of the gas-phase reactions of O3 with VOCs under conditions relevant to the atmosphere were well examined in the atmospheric science field [13]. It is found that the action of O3 is extremely selective, and O3 usually plays a positive role to remove only alkenes and other VOCs containing unsaturated carbons. It is of interest to examine the ozonation effect in this project using a dynamic system with a relatively high O3 concentration (ppm), rather than using a traditional static chamber system with a low O3 concentration (ppb). Also, the photolysis impact on the removal of VOCs under VUV irradiation is an important photochemical phenomenon to be explored. Recently, a few studies investigated the removal performance of UV-PCO for toluene and benzene using a bench-top photocatalytic flow reactor with ozone-producing UV lamps [3,14,15]. To the best of our knowledge, no study has been carried out on the investigation of the ozonation and photolysis effect for a wide range of VOCs using a pilot duct system, which is one of the contributions of this study. The principal objective of this research is to develop methodologies to evaluate the performance of UV-PCO systems for IAQ applications. Therefore, this paper demonstrates a systematic evaluation of in-duct UV-PCO air cleaners equitably and thoroughly under the conditions relevant to the actual applications for a wide range of VOCs. In addition, the ozonation effect on the performance of in-duct PCO air cleaners has been fully examined for the first time. Moreover, a parametric evaluation of the effects of various kinetic parameters, such as VOCs’ type, inlet pollutant concentration, airflow rate, light intensity, and RH, on the PCO efficiency has been conducted for extending the existing knowledge on the PCO technology in indoor air applications. The formation of by-products generated from incomplete conversions has also been examined, which may have a profound impact on PCO technological developments.

131

Table 1 Physical properties of TiO2 /FGFs and TiO2 /CCFs air filters. Property

Units

TiO2 /CCFs

TiO2 /FGFs

Fiber diameter TiO2 loading BET surface area Average pore diameter Total pore volume

␮m wt% m2 /g nm cm3 /g

90 14.32 887.7 3.1 0.69

150 4.63 105.7 3.5 0.09

Low-pressure mercury lamps of each 18.4 W (Ster-L-Ray, Atlantic Ultraviolet Inc.) were employed: a G18T5L/U germicidal (UVC) lamp with a peak wavelength of 254 nm, and a G18T5VH/U ozone producing (VUV) lamp with a maximum emission at 254 nm and a minor emission at 185 nm. All lamps were powered by ballasts for ionization of the mercury vapor. Eight reagent grade chemicals were selected as representative of indoor air contaminants [16], which included toluene (99.9%), pxylene (99.9%), 1-butanol (99.9%), n-hexane (96%), octane (95%), MEK (99.9%), and acetone (99.5%) from Fisher Scientific Inc. (Canada), and ethanol (99%) from SAQ (Société des alcools du Québec – Québec Alcohol Board). 2.2. Experimental set-up To develop a methodology for evaluating the performance of UV-PCO, an innovative UV-PCO system was built up, and the schematic diagram of the test apparatus is shown in Fig. 1. The test rig was made of four parallel aluminum ducts with 0.3 m × 0.3 m (1 foot × 1 foot) inner cross section area. The system was able to provide up to 255 m3 /h (150 cfm) airflow rates and was equipped with a radial fan with speed control mounted at the end of each duct. This was an open-loop mode system, and the laboratory air was introduced directly to the system after passing through a pleated fabric pre-filter. The air containing evaporated VOCs was introduced into the PCO system through a stainless steel tube and mixed with laboratory air at the gas mixer chamber. The conditions of inlet mixer gases were monitored for humidity and temperature by a sensor (HMT 100, Vaisala) mounted at the center of the mixer chamber. The well-mixed gases were evenly fed into four ducts. The upstream of each of the ducts was fitted with a perforated stainless steel cross tube to collect air samples and an electronic low-flow probe at the center to monitor the airflow rate. The UVPCO reactor was designed to be versatile, so that different in-duct UV-PCO filters with various geometries could be installed. In this study, three ducts were equipped with three PCO filters irradiated with four UV lamps arranged in two banks (Fig. 1). The vertical distance between the surfaces of the UV lamps and the PCO filters was approximate 5 cm, and the distance between two lamps was

2. Experimental 2.1. Materials Two commercially available PCO air filters, titanium dioxide (TiO2 ) coated on fiberglass fibers (TiO2 /FGFs) and TiO2 coated on carbon cloth fibers (TiO2 /CCFs), were examined in this study. The physical properties of the two systems were characterized by scanning electron microscopy (SEM) for morphology and N2 adsorption isotherm for BET surface area and pore structure, which are given in Table 1.

Fig. 1. Schematic diagram of the UV-PCO system.

132

L. Zhong et al. / Journal of Hazardous Materials 261 (2013) 130–138

13.3 cm. Each lamp was 8.6 cm away from the reactor wall. In one of ducts, only two VUV lamps were installed without the PCO filters in order to examine the ozonation effect. Two pressure taps were mounted before and after each of the PCO reactors. After the UV-PCO reactor, there was a probe installed at the center of each duct to monitor RH and temperature of air stream at downstream, respectively. Downstream was also fitted with two cross sampling tubes and one bulkhead union to provide ports for the collection of VOCs and ozone. Detailed prequalification tests of the test rig are available in previous publication [17]. The effluent stream was then introduced into the adsorption module containing carbon and chemical absorbents to trap the residual VOCs and the generated by-products. For the application of O3 producing lamps, metal honeycomb coated with MnO2 post-filters were installed at the end of the duct system for residual O3 decomposition. The number of layers of MnO2 post-filters employed in experiments were determined by the O3 concentration generated downstream, and, for all experiments, the outlet O3 concentrations after the duct system were controlled in less than 50 ppb. 2.3. Contaminant generation system The selected VOCs were injected using an automatic syringe pump (KD Scientific). The laboratory compressed air was used as the carrier gas and its flow rate was controlled by a mass flow controller (Omega FMA 5400/5500). A chemically inert polytetrafluoroethylene (PTFE) tube was used as a contaminant vapor line, through which vaporized chemical was passed into the injection port. The port was placed on the top of the test rig in order to avoid condensation of the VOCs on the interior duct surface. A perforated cross stainless steel tube with a diameter of 4.8 mm, which was connected with the tube transporting gaseous pollutants, was installed at the center of the duct system to uniformly distribute VOCs in the four-duct system. 2.4. Analytical methods The inlet and outlet concentrations of VOCs and gaseous byproducts were qualitatively and quantitatively monitored by an online calibrated photo-acoustic multi-gas monitor (B&K 1302) equipped with an auto sampler (CBISS MK3) and an offline calibrated high performance liquid chromatography (HPLC, Perkin Elmer). The concentration of ozone in each effluent stream was measured by a calibrated six-channel ozone analyzer (Model 465L) which was programmed to alternatively and continuously take a sample from the downstream of each duct with an accuracy of ±1% of reading. For the HPLC analysis, potential carbonyl by-products were trapped on a high purity silica adsorbent coated with 2, 4-dinitrophenylhydrazine (2,4-DNPH) (Supelco LpDNPH S10L). Sample eluate was separated and analyzed by the HPLC with UV detection (360 nm) equipped with a C18 Brownlee validated micro-bore column (150 mm × 4.6 mm ID, 5 ␮m film thickness). Acetonitrile and distilled water were used as mobile phase with a flow rate of 1.0 mL/min. A gradient analysis method was developed: the ratio of 70% acetonitrile/30% water was held for 6 min, then the ratio increased to 100% acetonitrile/0% water and maintained for 3 min, and finally the ratio returned back to 70% acetonitrile/30% water for 4 min. The irradiance of 254 nm and 185 nm on the surface of TiO2 filter was monitored by two calibrated UV radiometers (Steril-Aire and ILT900-R). The locations of nine irradiance measurement ports were equally arranged at the center of the duct. The average of nine irradiance values represented the average irradiance on the surface of the PCO filter. The average irradiances of 254 nm and 185 nm

under the airflow of 170 m3 /h were 25–36 W/m2 and 1–3 W/m2 , respectively. 2.5. Experimental conditions and procedure The target concentration of the selected challenge gases was a sub-ppm level (0.25–2 ppm) which closely represented indoor air pollution conditions. All PCO experiments were carried out at different setup values for the parameters, except humidity was unregulated in this system. Table 2 summarizes the detailed experimental conditions based on which to fully examine the impacts of various experimental parameters on the UV-PCO performance. The three-step injection procedure was developed and adopted for all UV-PCO tests, which included the following details. First, the preparation work contained calibration of the air sampling pumps and the sampling cartridges, and calibration of the online multigas monitor for the selected VOCs. Meanwhile, the sampling and measurement system was established by placing the sampling lines and sampling pumps in position and connecting potassium iodide (KI) ozone scrubber (selective removal of O3 and no VOCs lost by adsorption) in the sampling lines, setting up the real-time test system to monitor the airflow rate, temperature, and RH in each duct, setting up the online measurement system to monitor VOCs and O3 at upstream and downstream of each duct, and setting up an appropriate contaminant generation system. Second, the proper PCO filters and UV lamps were installed in the designated position in each PCO reactor. The fans were turned on and were set at an appropriate airflow rate; the multi-gas analyzer and the O3 monitor were turned on to measure the background for 30 min; and then UV lamps were switched on to get a stable UV output. When the experimental conditions became stable, the PCO reaction could be initiated by first injection of a challenge VOC with an appropriate injection rate, and the real-time concentration was recorded by the online measurement system. Once the steady-state condition was reached, DNPH samples were taken at sampling rate of 1.3 L/min for 1.5 h to explore the generation of by-products. For all VOCs at inlet concentration of 500 ppb, DNPH samples were taken twice in four ducts to check the repeatability of by-products. Upon DNPH sampling completion, the UV-PCO test was ended by stopping the injection while the measurement was continued. Then the UV-PCO test was repeated with second injection and third injection. In order to avoid of catalyst deactivation resulting from the high concentration, the order of injection rate was in accordance with the expected concentration from low to high. The whole duration of a UV-PCO test for a VOC with three concentration levels lasted approximate 10 h. Before starting of a UV-PCO test for another VOC, each set of PCO filters was irradiated under UV lamps with fans running for a period of around 10 h to regenerate catalyst and to remove the residue VOCs in the PCO system. 2.6. Quantification method The experimental data collected from the upstream and the downstream measurement ports is employed to calculate the effectiveness of a UV-PCO air cleaner. Single-pass removal efficiency, t (%), is determined by the amount of the removed pollutant from the air stream after it goes through the air cleaner, and it is defined as follows: t =

Q (Cup,t − Cdown,t ) Cup,t − Cdown,t × 100 = × 100 QCup,t Cup,t

(1)

where t (%) is the single-pass efficiency of a pollutant; Cup,t (mg/m3 ) is the contaminant concentration at upstream as a function of time; Cdown,t (mg/m3 ) is the contaminant concentration at

L. Zhong et al. / Journal of Hazardous Materials 261 (2013) 130–138

133

Table 2 Experimental conditions. Parameter

Experimental conditions

Measurement accuracy (by manufacturer)

Max. test standard deviation

Single VOC

Toluene, p-xylene, ethanol, 1-butanol, acetone, MEK, hexane, octane 0.25–2 41–255 (25–150 cfm) 10–60 16–43 20–25

–

–

±5% of the reading ±1.3% of the reading ±0.17% of the reading ±5% of the reading ±0.17 ◦ C

±0.27 ±12.2 (7.2 cfm) ±2.4 ±1.0 ±0.8

Inlet concentration (ppm) Volumetric flow rate (m3 /h) RH (%) Light intensity (W/m2 ) Temperature (◦ C)

downstream as a function of time and Q (m3 /s) is the airflow rate through an air cleaner. 3. Results and discussion 3.1. Influence of the targeted molecules 3.1.1. Effect of VOCs type Fig. 2 presents the conversion of each VOC at an inlet concentration of 500 ppb under three experimental scenarios (TiO2 /FGFs + UVC, TiO2 /FGFs + VUV, and TiO2 /CCFs + VUV). For two types of air filters, the order of single-pass removal efficiency of the selected chemical classes follows the sequence of alcohols > ketones > aromatics > alkanes. These observations agree with the photocatalytic oxidation rates reported by Hodgson et al. [11] and Obee and Hay [18]. This sequence is also in the same trend found in the adsorption capacity of TiO2 coated air filters for the selected chemical classes [14]. This implies that adsorption process, to be more specific, the intermolecular force is one of the key factors influencing the photocatalytic activity. For non-polar alkanes adsorbed in the solid phase dispersion forces are the main intermolecular force, which is weaker than van der Waals interactions for aromatics. Due to the high dipole moment of the carbonyl group, dipole–dipole interactions for ketones are stronger than van der Waals attractions for aromatics. In addition to van der Waals interactions, hydrogen bonding plays a greater role for the attraction between alcohols and hydrated catalyst surface. It can be also observed from Fig. 2 that single-pass UV-PCO removal efficiency of TiO2 /CCFs filter is distinctly higher than that of TiO2 /FGFs filter for all VOCs. Photocatalytic activity depends not only on the properties of challenge VOCs but also on the features of substrates supporting TiO2 . Generally, the larger specific surface area helps to increase photocatalytic activity since more active sites are provided through coating of TiO2 nano-particles on the larger BET surface of TiO2 /CCFs [19]. According to Table 1, the BET surface area of TiO2 /CCFs was 887.7 m2 /g, which was roughly 8 times higher than that of TiO2 /FGFs leading to a higher photocatalytic activity for TiO2 /CCFs. It is worth mentioning that RH was unregulated (15–45%) for all tests, and the cross influence of RH on the UV-PCO performance for different VOCs was not considered here. Table 3 shows the values of RH for the tests of the targeted molecules.

Table 3 RH for the tests of the targeted molecules. VOC

RH (%)

Ethanol 1-Butanol Hexane Octane Acetone MEK Toluene p-Xylene

14.8 19.2 31.8 43.2 43.5 21.8 44.8 34.5

± ± ± ± ± ± ± ±

1.3 2.1 1.5 2.2 1.4 1.3 2.4 2.3

TiO2 /FGFs filter presents hydrophilic property, and VOCs with high polarity show higher affinity to the surface of TiO2 /FGFs filter, whereas TiO2 /CCFs filter belongs to a non-polar substrate which prefers to adsorb non-polar VOCs. In addition, the lighter VOC in the same chemical class always shows higher photocatalytic activity than the heavier VOC for the TiO2 /FGFs filter, which is opposing for the TiO2 /CCFs filter. This is attributed to the fact that for the TiO2 /FGFs filter with less adsorptive ability, intermediates of small molecular weight generated from UV-PCO of light VOCs are less competitive with light VOCs for adsorption and photocatalytic reaction at active sites, and they are also more easily further oxidized or desorbed under humid conditions resulting in more active sites available. However, for the TiO2 /CCFs filter with strong adsorption ability, van der Waals interaction is the dominant force, which increases with molecular weight. Hence, the heavier VOC of each group demonstrates more active in the UV-PCO. 3.1.2. Effect of inlet concentration The UV-PCO experiments were conducted using three different inlet concentration levels with all selected VOCs to examine its effect on the removal performance of UV-PCO. Usually the inlet concentration was 250 ppb, 500 ppb and 1000 ppb, except that maximum 800 ppb was selected for 1-butanol due to difficult evaporation and minimum 500 ppb was used for acetone because of detection limit. The effect of the inlet concentration on the singlepass removal efficiency for various VOCs is shown in Fig. 3. The trend of a lower inlet concentration resulting in higher removal efficiency was observed for all VOCs. The same behavior was also observed by Jeong et al. [3] when they studied the photodegradation of toluene in the range of inlet concentrations from 0.6 to 20 ppm under VUV irradiation. This can be interpreted by the limited adsorption capacity of the fixed active sites at the catalyst surface. The amount of molecules effectively participating in the UV-PCO reaction is not enhanced in the same ratio as an increase of the inlet concentration resulting in a decrease of removal efficiency. Moreover, the competitive effect between multiple by-products and the challenge VOC to some extent inhibits the adsorption of a VOC, especially when its inlet concentration is high. Hence, compared with the challenge concentrations, the number of the active sites resulting from low BET surface area of TiO2 /FGFs is a limiting factor in this study. This result agrees with the conclusion made by Sleiman et al. [5] that PCO is suitable for the photodegradation of gaseous effluents at low ppb concentration levels. It should be noted that although RH was not constant between tested VOCs, it was almost constant for each individual one for three concentration levels (shown Table 3). 3.2. Influencing parameters 3.2.1. Effect of airflow rate Fig. 4 shows the effects of airflow rate on the conversion of ethanol. The airflow rate varied from 69 m3 /h (45 cfm) to 255 m3 /h (150 cfm) for the TiO2 /FGFs filter and from 41 m3 /h (25 cfm) to 170 m3 /h (100 cfm) for the TiO2 /CCFs filter. It is evident from the

134

L. Zhong et al. / Journal of Hazardous Materials 261 (2013) 130–138

Fig. 2. Single-pass removal efficiencies of various VOCs under three different experimental scenarios (inlet concentration = 500 ppb, RH = 15–45%, airflow rate = 170 m3 /h, irradiance = 27–36 W/m2 ).

Table 4 PCO conditions at different airflow rates. PCO filter

Airflow rate (m3 /h)

TiO2 /FGFs

69 89 127 187 255

TiO2 /CCFs

41 81 127 170

Face velocity (m/s)

Residence time (ms)

8 13 26 47 78

0.21 0.26 0.38 0.56 0.76

45 37 25 17 12

17 45 99 188

0.12 0.24 0.38 0.51

79 40 25 19

Pressure drop at PCO (Pa)

results that the conversion decreased gradually with an increase in airflow rate. The same behavior was also observed for acetone, hexane and toluene when the airflow rates were increased. Also, these trends are in accordance with results reported previously [3,6,11]. In addition, the curve of ethanol result is somewhat consistent with the finding reported by Hodgson et al. [11] that the relationship between the reaction efficiency and the residence time is approximated reasonably well by an exponential function. The airflow rate of 41–255 m3 /h corresponding to a face velocity of 0.12–0.76 m/s (shown in Table 4) was used in this study. However, by changing the airflow rate, the residence time differed a lot.

Here, residence time is defined as the thickness (0.95 cm) of the air filter divided by the face velocity. Table 4 presents the residence time and the pressure drop for the tested airflow rates of two air filters. Decreasing the airflow rate helps to increase the residence time so that more VOCs can be adsorbed to the catalyst surface and adsorbed molecules have more chances to participate in the reactions with hydroxyl radicals, and then to be oxidized. As a consequence, a higher conversion rate can be achieved. It was observed that the pressure drop increased as the airflow rate enhanced for two filters, which was in accordance with the conclusion reported by Destaillats et al. [12]. The pressure drop of the TiO2 /CCFs filter increased more compared to that of TiO2 /FGFs due to the high resistance resulting from the high BET surface area of the TiO2 /CCFs filter. It should be noted that in the case of VUV irradiation, O3 concentrations were varied when airflow rate changed. Photolysis also plays a role for the removal of VOCs, which will be discussed in Section 3.3.1. Therefore, a lower airflow rate results in higher single-pass removal efficiency. The multi-pass method may be an alternative to relatively extend the residence time of VOCs for the PCO technology applied in a HVAC system.

3.2.2. Effect of light intensity The effect of light intensity on the performance of UV-PCO air cleaners was examined at 170 m3 /h airflow rate with the

Fig. 3. The effect of inlet concentration on conversion of various VOCs under UVC irradiation for the TiO2 /FGFs air filter (RH = 15–45%, airflow rate = 170 m3 /h, irradiance = 27–30 W/m2 ).

L. Zhong et al. / Journal of Hazardous Materials 261 (2013) 130–138

135

Fig. 4. The effect of airflow rate on ethanol, acetone, hexane, and toluene conversion (inlet concentration = 500 ppb, RH = 40–60%, irradiance = 24–39 W/m2 ).

3.2.3. Effect of RH Water vapor plays a dual role in the UV-PCO through the following ways: on the one hand it provides hydroxyl radicals by chemical decomposition of adsorbed water; on the other hand excessive

Single-pass removal efficiency

TiO2/FGFs+UVC Acetone

Ethanol

Toluene

Hexane

20% 0.0411x0.4223

y= R² = 0.9312

15%

10%

y = 0.0175x0.5850 R² = 0.9119

y = 0.0228x0.4834 R² = 0.9688

5% 0% 10.0

y = 0.0072x0.4713 R² = 0.9312

20.0

30.0 Irradiance (W/m2)

40.0

50.0

Fig. 5. The effect of irradiance on ethanol, acetone, hexane, and toluene conversion (inlet concentration = 500 ppb, flow rate = 170 m3 /h, RH = 55–62%).

water vapor competes with challenge VOCs for the same surface of TiO2 . The optimal humidity level is determined by the balance between conversion promotion through chemical processes and inhibition through physical interactions [1]. Due to the limitation of the test facility, RH inside the duct was uncontrolled. RH examining experiments were conducted when the laboratory RH conditions achieved to the expected levels. The effect of water vapor on the conversion of VOCs was examined by applying humidity levels from 10% (2300 ppm) to 60% (16,000 ppm) to 500 ppb ethanol. Fig. 6 shows the single-pass removal efficiency for ethanol as a function of RH at different experimental scenarios. A decrease in the conversion rate of ethanol from 26% to 11% was observed when RH of air increased from 10% to 57% for the TiO2 /FGFs filter under UVC illumination. In the case of VUV exposure, the conversion of ethanol for TiO2 /CCFs and TiO2 /FGFs decreased from 40% to 23% and from 30% to 13%, respectively. For the tested other VOCs, such as acetone, hexane, and toluene, water vapor also shows inhibition effect on their conversion rates. These

Ethanol Single-pass removal efficiency

lighter VOCs of each selected group (ethanol, hexane, acetone, and toluene). The configurations of one, two, three, and four UVC lamps standing in a row with two TiO2 /FGFs filters were used in each duct, and the vertical distance between air cleaners and UV lamps was kept constant. The range of irradiance employed was 16–43 W/m2 . The experimental results of single-pass removal efficiency for tested VOCs are shown in Fig. 5. The increase trend of conversion rate with an increase of the irradiance was observed for all VOCs. The trend was correctly described by a power function, and the power exponent was in the range of 0.4–0.6, which is in consistent with the reaction order of 0.5 for high absorbed light intensity (greater than 10–20 W/m2 ) reported by Obee and Brown [20]. It should be noted that the removal efficiency of ethanol was lower than that of acetone due to the possible reason of partial deactivation of the catalyst in this case.

45% 40%

TiO2/CCFs+VUV TiO2/FGFs+VUV

O3=1017.1ppb

TiO2/FGFs+UVC

1269.2

35%

1428.4

O3=2372.5

30%

1112.9 1119.4

25%

1077.1 1321.7

20%

1061.1

15%

903.9

10% 5% 0%

20%

40%

60%

RH Fig. 6. The effect of RH on ethanol at different experimental scenarios (inlet concentration = 500 ppb, flow rate = 170 m3 /h, irradiance = 24–36 W/m2 ).

136

L. Zhong et al. / Journal of Hazardous Materials 261 (2013) 130–138

Table 5 Rate constants k for the gas-phase reactions of O3 with various compounds. k (cm3 molecule−1 s−1 )

Compound Alkane (i.e. ethane, hexane) Acyclic monoalkene (i.e. ethene, hexene) Cycloalkenes, cyclodialkenes, cyclotrialkenes (i.e. d-limonene, pinene) Monocyclic aromatic (except styrene, i.e. toluene, xylene) Oxygen-containing compounds not containing double c–c bonds (i.e. alcohols, ketones)

T (K)

Techniqueb

Lifetimea

−23

≤10 10−16 –10−18 10−14 –10−16

298 ± 2 298 ± 2 295 ± 2

S-IR/CL S/F-IR/CA/CL/UV/FTIR/MS S/F-CL

≥30 years 1–10 day 0.03–3 h

≤10−20 ≤10−20

297 ± 2 297 ± 2

S-CL S/F-CL/IR/CA

≥30 years ≥3 years

The rate constants were calculated when an O3 concentration of 1 × 1012 molecule cm−3 (i.e. around 40 ppb at ground level) was used. S stands for static system; F stands for flow system; MS stands for mass spectrometry; IR stands for infrared absorption spectroscopy; FTIR stands for Fourier transform infrared absorption spectroscopy; UV stands for ultraviolet absorption; CL stands for chemiluminescence; and CA stands for chemical analysis. a

35%

1600

30%

1400 1200

25%

1000 20%

Efficiency 800

15%

Ozone concentration

10%

600

Ozone concentration (ppb)

Single-pass removal efficiency by photons

b

400

5%

200

0%

0 250 500 1000 250 500 800 250 500 1000 250 500 1000 250 500 1000 250 500 1000 500 10002000 250 500 1000 ethanol (ppb) 1-butanol (ppb) hexane (ppb)

octane (ppb)

toluene (ppb) p-xylene (ppb) acetone (ppb)

MEK (ppb)

Fig. 7. Single-pass removal efficiencies of various VOCs at three concentration levels by photons (RH = 35–58%).

observations can be interpreted as the competition for adsorption between the VOC and water molecules. In addition, the presence of abundant water may enhance the possibility of electron–hole recombination, which is an unfavorable process for PCO of VOCs. According to ASHRAE Standard 55-2010 [21], a RH between 40% and 60% is recommended for a healthy and comfortable indoor environment. Compared with VOCs concentrations (typically ppb levels) in the context of indoor air applications, water vapor exists in large excess so that it is unlikely that hydroxyl radical concentrations become rate limiting. Competitive adsorption and electron–hole recombination can be deemed as the dominating interaction processes under the condition that the RH is achievable in buildings and HVAC systems. 3.3. Ozone influence 3.3.1. Ozonation and photolysis When only VUV lamps are presented in a duct, a target compound could be broken down by photolysis effect as well as ozonation effect. The gas-phase reactions of O3 with various classes of organics under the conditions relevant to the atmosphere were discussed by Atkinson and Carter [13], and the bimolecular rate constants are summarized in Table 5. From this table, it can be found that O3 usually plays a positive role in removing alkenes. For alkanes, aromatics and oxygen-containing organics, the reactions are very slow, at the room temperature with the bimolecular rate constant of ≤10−20 cm3 molecules−1 s−1 . Hence, these reactions are of negligible atmospheric importance. Therefore, the dominant process tends to be photolysis since photons with an energy at the ultraviolet wavelength can affect the chemical bonds. The photons output relates with the number of VUV lamps. Here, experiments to investigate the photolysis effect were conducted by placing different numbers of VUV lamps in each duct so as to

establish four photon energy levels. Eight types of single compound with three concentration levels were employed using the same methodology described in Section 2.5. Fig. 7 shows the single-pass removal efficiencies with similar concentrations for various VOCs at the 170 m3 /h (100 cfm) airflow rate in the absence of any PCO filter. It clearly shows acetone and MEK were scarcely removed. This low reactivity in the chemical reaction is due to the weaker electron withdrawing power of the carbonyl group compared to the hydrocarbons. Hence, ketones interact with photons much less readily than the other VOCs do. In a duct system, the photolysis effect on the elimination of ketones is negligible. The results also indicate the strong interaction of photons with the heavier VOCs in the same class resulting in higher removal efficiency, especially for aromatics. For example, when toluene and p-xylene were at the similar inlet concentrations, the single-pass removal efficiency of p-xylene by the photolysis effect was around twice of toluene. Moreover, these results further show that under the condition of the same amount of VUV lamps, an increased concentration of a challenge VOC significantly reduced the removal efficiency. 3.3.2. Ozone-involved UV-PCO The electron affinity (EA) of O3 is 2.103 eV, and is considerably larger than that of O2 (0.44 eV) or the oxygen atom (1.46 eV) [22]. Thus, excited electron resulting from absorbance of UV photon with TiO2 is captured more efficiently in the presence of O3 . The possible set of reaction steps considered in this scenario is described as follows: e− + O3 → O− + O2 CB

(2)

e− + O2 → O− 2 CB

(3)

− O− 2 + O3 → O2 + O3

(4)

L. Zhong et al. / Journal of Hazardous Materials 261 (2013) 130–138

137

Table 6 Comparison of gas-phase UV-PCO by-products generated with or without ozone. Compound (ppb)

UV lampa

Gas-phase UV-PCO by-products detected by HPLC (ppb)

Ethanol (478)

UVC VUV

Formaldehyde (27.0–29.7), acetaldehyde (168.9–174.0), acetone (1.6–3.3) Formaldehyde (13.6–33.9), acetaldehyde (88.1–101.7), 2-butenal (2.2–6.3), propanal (6.1–7.8)

1-Butanol (488)

UVC VUV

Formaldehyde (36.9–37.1), acetaldehyde (20.9–22.1), propanal (23.8–24.0), butanal (79.1–80.6) Formaldehyde (7.8–16.9), acetaldehyde (15.9–20.7), 2-butenal (7.3–9.2), propanal (24.5–26.5), butanal (22.0–49.9)

Hexane (498)

UVC VUV

Formaldehyde (26.1–26.3), acetaldehyde (13.0–14.3), acetone (1.1–2.9) Formaldehyde (14.9–23.3), acetaldehyde (15.3–21.4), 2-butenal (5.3), propanal (3.1–6.6), butanal (4.3), tolualdehyde (1.4–3.9), hexanal (0.6–1.1)

Octane (504)

UVC VUV

Formaldehyde (21.1–23.0), acetaldehyde (9.5–9.7), acetone (1.5–2.6), 2-butenal (2.8), propanal (0.3–1.7), butanal (0.6) Formaldehyde (20.7–40.1), acetaldehyde (18.9–20.3), acetone (18.2–25.2), propanal (4.4–4.8), butanal (0.2–4.8), pentanal (1.0–2.8)

Acetone (479)

UVC VUV

Formaldehyde (16.9–17.6), acetaldehyde (8.0–8.5) Formaldehyde (11.4–18.3), acetaldehyde (7.0–7.3)

MEK (479)

UVC VUV

Formaldehyde (22.3–23.0), acetaldehyde (11.3–28.6), acetone (1.0–2.5), hexanal (0.3–0.4) Formaldehyde (24.8–35.3), acetaldehyde (28.7–39.5), acetone (1.2–1.6), hexanal (0.4–1.2)

Toluene (485)

UVC VUV

Formaldehyde (10.3–12.2), acetaldehyde (4.6–5.8), acetone (3.6), 2-butenal (2.2), butanal (0.4–1.3), benzaldehyde (0.3–0.4) Formaldehyde (4.4–17.4), acetaldehyde (5.3–7.8), acetone (2.9–4.2), 2-butenal (4.6–12.2), butanal (0.2–1.8), benzaldehyde (1.5–3.7)

p-Xylene (484)

UVC VUV

Formaldehyde (16.6–17.2), acetaldehyde (5.9–6.4), acetone (2.1–2.7), butanal (0.1–0.3), tolualdehyde (1.7) Formaldehyde (17.7–33.4), acetaldehyde (8.9–12.1), acetone (1.4–2.9), 2-butenal (8.2–11.0), butanal (0.1–0.2), tolualdehyde (1.4–6.0)

a

UVC (with TiO2 /FGFs). VUV (with TiO2 /FGFs or TiO2 /CCFs) (the HPLC detection limit of each analyte is 3 ng).

− O− 3 + H2 O → OH · +OH + O2

(5)

The enhancement of electron capture rate due to the participation of O3 in PCO process not only reduces the possibility of recombination of electron–hole pairs, but also more effectively produces hydroxyl radicals through complicated chain reactions. Consequently, the single-pass removal efficiency of TiO2 /VUV was usually higher than that of TiO2 /UVC for the tested VOCs (see Fig. 3) due to the higher generation rate of hydroxyl radicals in the presence of O3 . Photolysis effect during TiO2 /VUV process also played a significant role. Jeong et al. (2005) and Yang et al. [3,23] also came to the same conclusion that employment of VUV in PCO technology may be more effective and more economical than the TiO2 /UVC process for the treatment of gas streams. 3.3.3. Gas-phase by-products generated with or without ozone The formation of by-products in the photocatalytic oxidation of the selected VOCs was investigated. Table 6 lists all the value ranges of major gas-phase UV-PCO by-products detected by the HPLC when the challenge gas was around 500 ppb. Formaldehyde and acetaldehyde were produced as the UV-PCO reaction products for all experiments in the absence or in the presence of ozone. Usually, yields of formaldehyde and acetaldehyde are proportional to the inlet concentrations of a challenge VOC. It is evident that the formation of by-products generated from incomplete oxidation by the UV-PCO is closely related to the nature of a challenge VOC. For example, the amount of acetaldehyde produced from the PCO of ethanol was 10 times higher than its amount produced from the PCO of other VOCs. Hence, the by-products derived from each VOC were somewhat different, which were propanal and butanal for 1butanol, benzaldehyde for toluene, tolualdehyde for p-xylene, and etc. In the case that VUV lamps were employed in the absence of an air filter, the detected by-products generated by the photolysis were not significantly different from those produced from UV-PCO, but they were generated at lower concentrations. In addition, compared with UVC-PCO, the involvement of O3 induced by the VUV lamps enhanced the mineralization of acetaldehyde and butanal as a by-product of partially oxidized ethanol and 1-butanol,

respectively. This is attributed to more hydroxyl radicals generated from photolysis of O3 . Moreover, more products were produced when employing the VUV lamps than those using the UVC lamps due to the photolysis. Detailed discussion about the impacts of different experimental conditions on the by-products generation could be found in Farhanian et al. [24].

4. Conclusion An innovative UV-PCO duct system experimental set-up was designed and constructed, and in-duct test methodologies were developed to investigate the performance of UV-PCO air cleaners for various VOCs removal. The single-pass removal efficiency of two types of air filters (TiO2 /FGFs and TiO2 /CCFs) under UVC or UVV illumination in removing VOCs with various physical properties ranks as follows: alcohols > ketones > aromatics > alkanes. It was observed that the PCO removal efficiency increased by decreasing of the inlet concentration, reducing the airflow rate, increasing of the irradiance, or decreasing of RH, respectively. Competition of adsorption sites, bond energy, residence time, and the number of effective hydroxyl radicals are new interpretations of these observations based on a fundamental analysis of reaction mechanisms. For the first time, it was investigated that the photolysis under VUV irradiation played a critical role in the removal of alkanes, alcohols, and aromatics at three concentration levels in a duct system. Additionally, the removal efficiency reduced with an increase in the inlet concentrations of a challenge VOC. Furthermore, the conversion rates of the VOCs by the VUV-PCO were higher than that by the UVC-PCO due to the presence of ozone. Formation of by-products had a close relationship with the PCO reaction mechanisms of different VOCs. The appearance and the concentrations of the by-products depended on the experimental conditions, the nature of a challenge VOC, as well as the characteristics of the catalysts. Therefore, by knowing the intrinsic kinetics, the design of a PCO air purifier working under optimal conditions can reduce the generation of by-products, which is the future research direction to enhance the practicability of the PCO technology.

138

L. Zhong et al. / Journal of Hazardous Materials 261 (2013) 130–138

Acknowledgements The authors would like to express their gratitude to the Natural Sciences and Engineering Research Council of Canada for supporting this research project through a CRD grant, and Circul-Aire, Inc. for the financial support and providing the support for the design and construction of the experimental set-up. References [1] L. Zhong, F. Haghighat, P. Blondeau, J. Kozinski, Modeling and physical interpretation of photocatalytic oxidation efficiency in indoor air applications, Build Environ. 45 (2010) 2689–2697. [2] W.K. Jo, K.H. Park, Heterogeneous photocatalysis of aromatic and chlorinated volatile organic compounds (VOCs) for non-occupational indoor air application, Chemosphere 57 (2004) 555–565. [3] J. Jeong, K. Sekiguchi, W. Lee, K. Sakamoto, Photodegration of gaseous volatile organic compounds (VOCs) using TiO2 photoirradiated by an ozone-producing UV lamp: decomposition characteristics, identification of by-products and water-soluble organic intermediates, J. Photochem. Photobiol. A Chem. 169 (2005) 279–287. [4] A.K. Boulamanti, C.A. Korologos, C.J. Philippopoulos, The rate of photocatalytic oxidation of aromatic volatile organic compounds in the gas-phase, Atmos. Environ. 42 (2008) 7844–7850. [5] M. Sleiman, P. Conchon, C. Ferronato, J.M. Chovelon, Photocatalytic oxidation of toluene at indoor air levels (ppbv): towards a better assessment of conversion, reaction intermediates and mineralization, Appl. Catal. B: Environ. 86 (2009) 159–165. [6] Q.L. Yu, H.J.H. Brouwers, Indoor air purification using heterogeneous photocatalytic oxidation. Part I. Experimental study, Appl. Catal. B: Environ. 92 (2009) 454–461. [7] A.L. Loo Zazueta, H. Destaillats, G. Li Puma, Radiation field modeling and optimization of a compact and modular multi-plate photocatalytic reactor (MPPR) for air/water purification by Monte Carlo method, Chem. Eng. J. 217 (2013) 475–485. [8] W. Chen, J.S. Zhang, Z. Zhang, Performance of air cleaners for removing multiple volatile organic compounds in indoor air, ASHRAE Trans. (2005) 1101–1114. [9] J. Disdier, P. Pichat, D. Mas, Measuring the effect of photocatalytic purifiers on indoor air hydrocarbons and carbonyl pollutants, J. Air Waste Manage. 55 (2005) 88–96. [10] O. Debono, F. Thevenet, P. Gravejat, V. Hequet, C. Raillard, L. Lecoq, N. Locoge, Toluene photocatalytic oxidation at ppbv levels: kinetic investigation and carbon balance determination, Appl. Catal. B: Environ. 106 (2011) 600–608.

[11] A.T. Hodgson, H. Destaillats, D.P. Sullivan, W.J. Fisk, Performance of ultraviolet photocatalytic oxidation for indoor air applications, Indoor Air 17 (2007) 305–316. [12] H. Destaillats, M. Sleiman, D.P. Sullivan, C. Jacquiod, J. Sablayrolles, L. Molins, Key parameters influencing the performance of photocatalytic oxidation (PCO) air purification under realistic indoor conditions, Appl. Catal. B: Environ. 128 (2012) 159–170. [13] R. Atkinson, W.P.L. Carter, Kinetics and mechanisms of the gas-phase reactions of ozone with organic compounds under atmospheric conditions, Chem. Rev. 84 (1984) 437–470. [14] D. Kibanova, J. Cervini-Silva, H. Destaillats, Efficiency of clay-TiO2 nanocomposites on the photocatalytic elimination of a model hydrophobic air pollutant, Environ. Sci. Technol. 43 (2009) 1500–1506. [15] N. Quici, M.L. Vera, H. Choi, G. Li Puma, D.D. Dionysiou, M.I. Litter, H. Destaillats, Effect of key parameters on the photocatalytic oxidation of toluene at low concentrations in air under 254 + 185 nm UV irradiation, Appl. Catal. B: Environ. 95 (2010) 312–319. [16] D.W. VanOsdell, Evaluation of test methods for determining the effectiveness and capability of gas-phase air filtration equipment for indoor air applicationsphase I: literature review and test recommendations, ASHRAE Trans. 100 (1994) 511–523. [17] C.S. Lee, L. Zhong, D. Farhanian, C. Flaherty, F. Haghighat, Development of a parallel test system for the evaluation of UV-PCO systems, in: 7th International Cold Climate Conference, Calgary, Alberta, Canada, November 2012, 2012. [18] T.N. Obee, S.O. Hay, The estimation of photocatalytic rate constants based on molecular structure: extending to multi-component systems, J. Adv Oxid. Technol. 4 (1999) 1–6. [19] L. Zhong, C.S. Lee, F. Haghighat, Adsorption performance of titanium dioxide (TiO2 ) coated air filters, J. Hazard. Mater. 243 (2012) 340–349. [20] T.N. Obee, R.T. Brown, The estimation of photocatalytic rate constants based on molecular structure: extending to multi-component systems, Environ. Sci. Technol. 29 (1995) 1223–1231. [21] ANSI/ASHRAE Standard 55, Thermal Environmental Conditions for Human Occupancy, American Society of Heating Refrigeration and Air Conditioning Engineers, Inc., Atlanta, 2010. [22] S. Matejcik, P. Cicman, A. Kiendler, J.D. Skalny, E. Illenberger, A. Stamatovic, T.D. Mark, Low-energy electron attachment to mixed ozone/oxygen clusters, Chem. Phys. Lett. 261 (1996) 437–442. [23] L. Yang, Z. Liu, J. Shi, Y. Zhang, H. Hu, W. Shangguan, Degradation of indoor gaseous formaldehyde by hybrid VUV and TiO2 /UV processes, Separat. Purif. Technol. 54 (2) (2007) 204–211. [24] D. Farhanian, F. Haghighat, C.-S. Lee, N. Lakdawala, Impact of design parameters on the performance of ultraviolet photocatalytic oxidation air cleaner, Build. Environ. 66 (2013) 148–157.