1. Introduction

Global population increase and rapid industrialization have put severe strains on clean water resources [1, 2]. It is estimated that around 4 billion people worldwide have limited access to clean water and millions of people die of waterborne diseases annually [3]. The most common methods to treat water involve using physical, chemical, or biological processes that nonetheless do not destroy many of the pollutants in the water, mainly highly complex organic compounds chief amongst them halogenated organics [4]. Amongst these organics produced by industrial and agricultural usage are dyes, surfactants, pesticides, and herbicides [1].

There are currently several technologies used to obtain high-purity water. Phase-transition treatments such as adsorption or coagulation merely transfer the pollutants from one phase to another creating the problem of ultimately disposing of the material [5]. Other methods like sedimentation, filtration and membrane technologies can create secondary pollutants and have high operating costs [6]. Air and stream stripping removes the contaminants into the gas phase creating gas pollution [7] while a proven method like adsorption by activated carbon produces a hazardous solid that must be removed [8]. Biological processes may also be ineffective because of the presence of compounds disabling microbe activity [9]. Chlorination, the most commonly used treatment process, generates carcinogenic byproducts [10, 11]. Generally, the most effective processes are based on the use of hydroxyl radical compounds like peroxides and ozone; however, these processes use high-energy UV light coupled with the strong oxidants generally of high toxicity [12]. Indeed, some of these processes are destructive in nature and suffer from very low efficiencies which lead to high treatment costs if complete mineralization is required.

The short-comings in these conventional technologies has led to the development of new approaches based around advanced oxidation (AO) processes [13]. AO processes create in situ transitory species which are capable of mineralizing the compounds of interest [14]. One particularly promising technology in this area is heterogeneous photocatalysis which use semiconductor materials like TiO₂ and low energy UV photons to achieve the degradation of organic compounds and the deactivation of harmful microbes [3]. The appeal of this technology besides the low costs is its ability to completely mineralize the pollutants into harmless compounds; organic compounds are degraded to carbon dioxide while others are reduced to common anions like nitrates, sulfates and chlorides [15]. Practically any pollutant including aliphatics, aromatics, dyes, surfactants, pesticides, and herbicides have been demonstrated in previous literature to be completely mineralized [16].

Although photocatalysis is a very promising technology, there are relatively few pilot-scale plants that are currently in operation [17]. While TiO₂ can be activated by sunlight, the intensity of peak solar illumination is too low to enable high capacity for water treatment for a lot of applications without using a lot of land area [18]. On the other hand, the use of external UV lights needs to justify the extra cost by enabling higher water treatment capacity to still be competitive [19]. Amongst existing reactors are slurry-type reactors where the TiO₂ catalysts are suspended in solution [20]. These reactors suffer from poor light illumination and require down-stream collection of the TiO₂ particles [17, 20]. On the other hand, fixed bed reactors, using immersion-type, external-type and distributive-type lamps, generally a variation of the annular type reactors suffer from low surface area to volume ratio [21]. Progress has been made to improve these reactor surfaces by using innovative micro and nano-structures to better distribute light and reagents locally such as with integrated ZnO wires [22] and TiO2/PDMS nanocomposites [23]. Other developments include the proposal of innovative-type reactor designs for large-scale applications. Multiple Tube Reactors (MTR) and Tube Light Reactors (TLR) reactors both use waveguides to deliver the UV photons deep into the reactor depth where they are gradually released [24, 25]. Apart from imparting more than 400 times improvement over the classical annular reactors, these designs also have the possibility of being scaled for high-capacity water-treatment applications [25].

One limit of such reactors however is the uneven distribution of light delivered throughout the reactor space [26]. Because these reactor models both rely on waveguides, there is a decay of the internally transmitting light resulting in non-uniform release of the light [26]. One possible solution is to employ waveguides with engineered scatterers that are spatially distributed precisely to scatter the light uniformly within the reactor space. We have utilized such waveguide designs in the past for use in algal photobioreactors [27]. Employing single-waveguide reactors, we were able to demonstrate uniform scattering illumination across the length of the waveguide when integrated over the entire range of scattering angles in the near-field regime [27]. Such uniform illumination schemes were shown to be superior to their counterparts chemically etched to have uniform surface roughness in delivering more than 40% improvements in growth rates to algal cultures [27].

In this paper, we demonstrate a photo-catalytic water-splitting device with multiple stacks of waveguides. The waveguides are fabricated with both uniform and engineered scatterer designs to test the validity of the scatterer design. We use a TiO₂ sol-gel on top of each layer of waveguides to degrade methylene blue to test organic degradation for water cleaning. At 10 mL/min capacity, we demonstrate close to 67% degradation of the organic dye for the reactors with the engineered scatterer waveguides along with close to 4.5-5 times improvement over the uniform scatterer design. We also offer a path for upscaling the reactor for high-capacity use.

2. Methods

2.1 Fabricating the scattering pattern

SU-8 pillars were fabricated by spinning a 2.8 μm thick layer of SU-8 2002 (Microchem, USA) on 1 mm borosilicate glass slides and patterned through the use of hard contact exposure on the Suss MA6-BA6 Contact Aligner. As shown in Fig. 2, each SU-8 pillar was 5 µm by 5 µm. The scattering scheme involving the uniform density of SU-8 pillars had a coverage density of 25%, i.e. each pillar was separated from the other on a square lattice by 5 µm. Gradient densities of SU-8 pillars were achieved by varying the distance from each SU-8 pillar along the length of the waveguide. The chemically etched waveguides were fabricated by applying Glass Etching Cream (Armour Etch) to glass slides for seven hours followed by rigorous rinsing in water. This etching cream isotropically etched the glass to create a uniform characteristic surface roughness.

2.2 TiO₂ sol-gel

To create the TiO2 sol-gel, we added 0.23 mL of acetylacetone to 70 mL of water and vigorously mixed the solution. 7 g of anatase TiO₂ nanopowder (50 nm, Sigma Aldrich, USA) was mixed in slowly to disperse the powder over the solution. 0.2 mL of Triton X-100 was then added to facilitate the spreading of the colloid on the coverslips after which 1.4 g of polyethylene glycol was added into the mixture. The solution was then left to mix for several days before it was ready to use.

2.3 Fabricating the waveguides

50 µL of the TiO₂ sol-gel were applied to 80 µm coverslips. As shown in Fig. 1(b), the sol-gel was distributed evenly over the cover slides to create a uniform thin film. The coverslips were then dried for 2 hours at 80 °C after which they were calcinated at 550 °C for 6 hours in a controlled atmosphere furnace in atmospheric air to treat the catalyst film. The microcover slips with the TiO₂ coating were then affixed to the 1mm glass slides with the scatterers facing towards the coverslip as shown in Fig. 2. For the chemically etched scattering scheme, instead of SU-8 pillars as shown in Fig. 2, the glass slide had a homogenous roughness that scattered the light. The coverslip was sealed in place with PDMS to create air cladding at the interface to better scatter the light.

 

Fig. 1 (a) Picture of the stacked waveguide reactor; (b) SEM of the thin-film of TiO₂ catalyst on a cover-slip slide.

Download Full Size | PDF

 

Fig. 2 Schematic of the waveguide with the gradient scatterers enabling uniform illumination.

Download Full Size | PDF

2.4 Reactor Fabrication and Operation

The reactor frames were printed using a 3D printer with a photocurable resin (VeroClear, Objet Geometries Inc., Israel). The dimension on the frame was 7.5 cm x 2.5 cm x 3 cm (length x width x height). As shown in Fig. 1(a), the reactor frames consisted of 10 parallel slots reactor separated by 2 mm along the height of the reactor for each waveguide. Each layer slot had an influent port for input media and an effluent port for output media on the other side. After the waveguide scatterers were inserted in each layer, polydimethylsiloxane (PDMS) (Dow Chemicals, Midland, Michigan, USA) was used to affix the waveguides to each layer of the reactor. In all of our experimental runs, we used a 100 W Hg lamp as the UV source placed an inch away from the reactors and focalized by a lens. We calculated the UV light intensity, at 365nm, incident on the side of the reactor to be close to 0.2W/cm^2 assuming that this accounted for roughly 10% of total optical power output of the Hg lamp. Our reactors utilized Methylene Blue as the test organic dye. Degradation experiments were performed using 3x10⁻⁶ M Methylene Blue in 1x PBS buffer to keep the media around neutral pH. The percentage degradation of the dye was monitored by analyzing its absorbance at 670 nm. The activity was calculated by multiplying the flow rates by the percentage of the dye that gets degraded.

3. Results and Discussion

The light was illuminated into the waveguide through the side where it was totally internally transmitted until it was released either by the presence of a rough surface or by striking a SU-8 pillar with a higher index of refraction (NA = 1.59) than the glass itself. According to Fresnel equations, when light strikes an interface from a medium of lower to higher refractive index, part of the light is transmitted and reflected. As long as the coupling coefficient between the pillars and the waveguide is small, we can expect that both the intensity of the internally reflected wave and scattered wave will decrease exponentially for the case of a constant density of SU-8 pillars along the transmitting direction of the waveguide.

As demonstrated in a previous work [27], if the extinction coefficient is known for a uniform density of scatterers, one can then precisely distribute the scatterers along a gradient distribution to ensure uniform intensity of scattering along the length of the waveguide as:

Of course, the total illumination across the surface of the waveguide will be composed of different scattering profiles over all the scattering angles. To do this, we used shallow channel dyes as reported in previous works (not shown here) [27]. The scattered light was used to excite fluorescent dyes in a channel depth of 300 µm within the surface of the waveguides. Images taken over the length of the waveguide allowed for the determination of the relevant extinction coefficient for Eq. (1). This modeled the 2 µm photocatalyst layer suspended ~120 µm above the waveguide reasonably well.

Borosilicate glass is highly absorbing of higher energy UV photons with wavelengths shorter than 350 nm while SU-8 is highly absorbing for wavelengths shorter than 360 nm. Because the anatase phase of TiO₂ is only activated by UV photons smaller than 380 nm, this meant that the wavelength of activation corresponded to 365 nm, the only peak in between 360 nm and 380 nm on the Hg lamp spectrum.

At 1 mL/min going through each layer (10 mL/min for entire reactor), the photocatalytic water-cleaning reactor based on the chemical etch scatterers, constant SU-8 scatterers and gradient SU-8 scatterers yielded 15%, 12% and 67% degradation percentages respectively. As shown in Fig. 3, this yielded between 4.5 times and 5 times improvement in degradation activity for the reactor with the gradient scatterers over the uniform scatterers. We suggest this improvement of the gradient scatterers is because of its more uniform illumination. The uniform scatterer schemes including both the chemical etch and constant SU-8 scatterer scheme exponentially decay in illumination intensity; the front of the waveguides have much more than optimal intensity leading to photo-saturation, while the back of the waveguides are illuminated with less than optimal intensities leading to sub-optimal degradation rates.

 

Fig. 3 Degradation Percentages of reactors with different scattering schemes for light delivery through the waveguides at capacity of 10mL/min or 1mL/min/layer.

Download Full Size | PDF

Another salient advantage of our system is also the augmentation of mass transport. Similar optofluidic planar reactors for water cleaning [22, 28] in the past have showed tremendous improvements albeit they were much thinner than the slot channels in our reactor (100 µm vs 1 mm). The augmentation of mass transport is reported in previous experiments [29] as well and is explained through the following Eq.:

To test the reaction rates, we performed degradation reactions at different flow rates from 0.5mL/min to 4mL/min over one layer. To calculate the degradation reaction, we used to following Eq.:

 

Fig. 4 Degradation Rate Constant with respect to the different flow rates.

Download Full Size | PDF

To achieve higher capacity photocatalytic water cleaning reactors will require upscaling the dimensions of the reactor to make the treatment costs more competitive. The gradient distribution for the SU-8 scatterers can be generalized and employed for longer sizes of waveguides. To keep the improvements in mass transport, we suggest that the distance between the waveguides is kept as small as possible. In addition, while the SU-8 scatterer fabrication that we used on our waveguides is very expensive, there are cheaper fabrication methods possible such as hot embossing, casting and molding. Finally, the use of UV LEDs could provide the cheap source of low-energy UV photons required to activate the reactor.

4. Conclusion

In this paper, we demonstrate a novel photocatalytic reactor for water cleaning with augmented mass transfer and optical transfer characteristics for high-capacity applications. We use a TiO₂ sol-gel to create a porous thin film of catalyst and stack the layers closely to improve mass transport. In addition, we use SU-8 pillars as scatterers distributed across the waveguide in a gradient to enable uniform illumination across the length of the waveguide. We achieve 67% degradation of the Methylene Blue dye at 10mL/min capacity and show a 4.5-5 times improvement over typical waveguide light delivery configurations. The degradation rate constant is characterized in flow rate experiments and shown to be between 0.37 – 0.54 min⁻¹.