Solar energy is readily available in most climates and can be used for water purification. However, solar disinfection of drinking water mostly relies on ultraviolet light, which represents only 4% of the total solar energy, and this leads to a slow treatment speed. Therefore, the development of new materials that can harvest visible light for water disinfection, and so speed up solar water purification, is highly desirable. Here we show that few-layered vertically aligned MoS2 (FLV-MoS2) films can be used to harvest the whole spectrum of visible light (∼50% of solar energy) and achieve highly efficient water disinfection. The bandgap of MoS2 was increased from 1.3 to 1.55 eV by decreasing the domain size, which allowed the FLV-MoS2 to generate reactive oxygen species (ROS) for bacterial inactivation in the water. The FLV-MoS2 showed a ∼15 times better log inactivation efficiency of the indicator bacteria compared with that of bulk MoS2, and a much faster inactivation of bacteria under both visible light and sunlight illumination compared with the widely used TiO2. Moreover, by using a 5 nm copper film on top of the FLV-MoS2 as a catalyst to facilitate electron–hole pair separation and promote the generation of ROS, the disinfection rate was increased a further sixfold. With our approach, we achieved water disinfection of >99.999% inactivation of bacteria in 20 min with a small amount of material (1.6 mg l–1) under simulated visible light.
Rapid and energy-efficient water disinfection methods are urgently required to address global challenges related to energy and water scarcity1,2,3,4,5. Solar energy is an attractive renewable-energy resource and can be used for water disinfection via the solar disinfection of drinking water (SODIS), an approach employed throughout the world (particularly in developing countries that lack distribution systems for potable water) to disinfect water for consumption6. Sunlight can inactivate microorganisms via direct or indirect mechanisms. Direct mechanisms involve photons (usually ultraviolet B (UVB) or ultraviolet A (UVA)) that interact directly with nucleic acids or other essential macromolecules to cause inactivation. Indirect mechanisms involve photons (ultraviolet or visible) that interact with chromophores either within the organism or external to the organism to create reactive oxygen species (ROS) that subsequently cause inactivation. To date, most studies suggest that disinfection by ultraviolet photons is one of the most important means of inactivating waterborne pathogens7,8,9. However, energy in UVA and UVB collectively only accounts for 4% of the energy in the solar spectrum, so SODIS is time consuming (requires 6–48 h of exposure time)6. It is desirable to harness more effectively the energy in photons from the visible range to speed up the photoinactivation.
One way is to use a semiconductor-based photocatalyst10,11,12. When a photocatalyst absorbs light, it generates electron–hole pairs so that electrons and holes can react with water and dissolved oxygen separately to generate ROS. ROS, such as the hydroxyl radical, singlet oxygen and superoxide, are strong oxidants and can disinfect pathogens by damaging essential macromolecules13,14. Technically, the potentials (versus normal hydrogen electrode) for ROS production (pH 7) at the electron side for oxygen reductions are −0.33 V (O2/O2•−) and 0.28 V (O2/H2O2), and at the hole side for water oxidation are 1.1–1.9 V (H2O/OH•) and 0.82 V (H2O/O2)15,16,17,18 (Fig. 1a), and so it is possible for an efficient photocatalyst to use the whole visible-light range of the solar spectrum. However, the bandgaps of the most-studied metal oxide photocatalysts are typically too large for most of the visible part of the solar spectrum to be utilized. For example, TiO2 has a bandgap of 3.0–3.2 eV and only harvests ultraviolet light19. Lowering the bandgap of TiO2 and enabling its absorption of visible light usually involves complex synthesis, such as doping with nitrogen, sulfur or iron19,20,21,22,23,24, making hybrid composites of TiO2 with lower bandgap materials25,26 and advanced structural design27,28, which limit its practical utility for photocatalytic applications29,30. Besides lowering the bandgap of metal oxides, research has investigated the potential for new semiconductor materials with small bandgaps, such as C3N4 (2.7 eV, cutoff wavelength 459 nm)31,32 and red phosphorus (1.42 eV, cutoff wavelength 873 nm)33 as photocatalysts that utilize visible light. The disinfection efficacy, specifically the disinfection rate, of these materials is still far from practical applications25,31,34,35,36,37. To seek novel photocatalysts that can harvest visible light for fast water disinfection is extremely attractive.
MoS2, a prototypical layered transition-metal dichalcogenide, is an emerging semiconductor material with physical and chemical properties conducive to applications in transistors and electrocatalysts for hydrogen-evolution reactions38,39,40,41,42. Single- or few-layered sheets of MoS2 exhibit intriguing properties distinct from those of bulk MoS2. By decreasing the thickness of MoS2 to a few layers or a single layer, its bandgap changes from an indirect bandgap of 1.3 eV to a direct bandgap of 1.9 eV (ref. 43). Also, nanostructured MoS2 would benefit the separation of electron–hole pairs by decreasing the distances for electrons and holes to diffuse to the surface of the materials and also increasing the reaction sites44. Here we report the successful use of few-layered vertically aligned MoS2 (FLV-MoS2) as a photocatalyst for water disinfection under visible-light illumination (Fig. 1b). The FLV-MoS2 has great potential for photocatalytic water disinfection owing to its low cost and straightforward synthesis. FLV-MoS2 showed a much faster disinfection rate than TiO2 under both visible-light and sunlight illumination. As semiconductor materials are nonspecific in catalysing ROS production among other competing reactions (such as hydrogen evolution, oxygen reduction and oxidation), additional catalysts (Cu or Au) were deposited onto the MoS2 films to promote the production of ROS. With 5 nm of Cu or Au as both the reaction catalyst and to facilitate the separation of electron–hole pairs, FLV-MoS2 achieved a rapid water disinfection with 5 log (>99.999%) inactivation of E. coli within 20 or 60 min, respectively, compared with that of pristine FLV-MoS2 of 120 min.
Growth and characterization of FLV-MoS2
The FLV-MoS2 was grown by first sputtering Mo film on a glassy carbon (GC) substrate, which was then sulfurized at 500 °C for 10 min (ref. 41). The sulfurization was conducted in a sulfur-rich environment and the MoS2 film thickness was determined by the initial thickness of the Mo film sputtered. Transmission electron microscopy (TEM) images of the 40 nm MoS2 film are shown in Fig. 2a. It is clear that the as-grown MoS2 film is composed of vertically aligned layers of MoS2 with domain sizes of around 3–10 layers, as also shown in our previous studies41,42. Reports have demonstrated that MoS2 has a higher in-plane electrical conductivity40,41 and also a higher catalytic activity on the edge sites, which means this vertically aligned structure can have a better electron–hole transport from MoS2 to the electrolytes. For photoelectrochemical reactions, the material band structure is the key factor that directly determines the mechanism, products and efficiency. Therefore, the properties of the vertically aligned MoS2 were characterized. First, the bandgap was characterized by measuring the absorption spectrum of MoS2, as shown in Fig. 2b for 40 nm MoS2. The bandgap value was extracted from the relation between the absorption coefficient and photon energy for the indirect bandgap semiconductor. The plot of (αhν)1/2 versus hν is shown in Supplementary Fig. 1. The bandgap extraction gives 1.55 eV for 40 nm FLV-MoS2. This allows FLV-MoS2 to absorb light up to 800 nm wavelength and the solar spectrum utilization is also increased to ∼50% in energy compared with the 4% of TiO2. The Fermi level of FLV-MoS2 was characterized by scanning Kelvin probe force microscopy and the results are shown in Fig. 2c–e. The Au line was patterned on the MoS2 film as a reference because it has a stable work function of 5.1 eV. Figure 2d shows the line-scan data on the MoS2 film that crosses the Au pattern. The Fermi level of MoS2 on the line scan ranges from 5.23 to 5.26 eV with an average of 5.24 eV. The Fermi-level mapping on a 1.5 × 12 mm2 area (Fig. 2e) shows the uniformity of the as-grown FLV-MoS2 film and the averaging Fermi level across the entire area is ∼5.24 eV. The valence band (VB) position of FLV-MoS2 was confirmed as 5.55 eV using ultraviolet photoelectron spectroscopy and the results are shown in Supplementary Fig. 2. The band position of FLV-MoS2 with respect to the ROS reaction potential and the band position of bulk MoS2 are shown in Fig. 2f. The VB position for bulk MoS2 is 5.54 eV (Supplementary Fig. 2). It is clear that FLV-MoS2 is suitable for ROS generation for photocatalytic water disinfection.
Photocatalytic disinfection performance of FLV-MoS2
The photocatalytic disinfection performance of FLV-MoS2 was then examined. First, the photocatalytic effect of FLV-MoS2 was confirmed by comparing the inactivation efficiencies of bacteria with FLV-MoS2 under visible light (without UVA and UVB) to those of the controls—FLV-MoS2 incubated with bacteria in the dark, and visible-light illumination of the bacterial suspension without FLV-MoS2. A solar simulator equipped with an ultraviolet-blocking filter was used as the light source. The Gram negative bacterium E. coli was used as a process indicator in all the experiments unless otherwise specified. The bacterial concentration at each time point of the experiment was normalized to the starting concentration at time 0 and the results are shown in Fig. 3a. Visible light alone or FLV-MoS2 in the dark both showed <50% disinfection efficiency after 120 min, whereas FLV-MoS2 under visible light showed >99.999% E. coli inactivation in 120 min, after which the bacteria could not be detected and no recovery was observed (Supplementary Fig. 3). The disinfection efficiency of FLV-MoS2 was also compared with that of an equal mass of bulk MoS2 (suspended flakes of ∼2 µm) under the same experimental conditions (Fig. 3b). Bulk MoS2 showed a 54% efficiency over 120 min, which is much lower than that of FLV-MoS2 of >99.999%. This suggests that decreasing the size of MoS2 to only a few layers (∼2–6 nm thick) can increase the photocatalyst properties of MoS2. Decreasing the layer thickness of MoS2 increases the ROS generation because of both the bandgap widening and the diffusion-distance shortening of the electrons and holes to the material surface. Also, the disinfection performance was compared between an equal mass of FLV-MoS2 and horizontal MoS2 film (Fig. 3b). The horizontal MoS2 film was made of stacked MoS2 nanoflakes, with each flake 1–6 nm in thickness. As both samples have larger bandgaps than that of bulk MoS2, the disinfection rates were much higher. Between the FLV-MoS2 and horizontal MoS2, the vertical direction configuration gave a higher disinfection rate. This could be the result of a higher in-plane conductivity of the electron/hole in MoS2 and also to more active edge sites exposed on the FLV-MoS2 film.
To eliminate the thermal effect potentially caused by infrared light, the disinfection efficiency of E. coli in water with FLV-MoS2 under visible light combined with infrared light and under only infrared light was compared. The light spectrum is shown in Fig. 3c and the disinfection performance is shown in Fig. 3d. With infrared light alone, the disinfection efficiency (83%) was much less than that of visible light combined with infrared light together (>99.999%). Additionally, a temperature-control experiment also eliminated the thermal effect (Supplementary Fig. 5). Therefore, it is confirmed that visible-light photocatalysis instead of a thermal effect induces most of the disinfection. The disinfection effect of FLV-MoS2 on a model Gram positive bacterium, Enterococcus faecalis, and another strain of E. coli was also demonstrated. No live bacteria were detected after 120 min of illumination (Supplementary Figs 6 and 7). This suggests that FLV-MoS2 can be used to inactivate a variety of bacteria.
The disinfection performance of FLV-MoS2 was compared with that of the well-studied photocatalyst TiO2. TiO2 film was grown through the atomic-layer-deposition method followed by an annealing process. The thickness of TiO2 was kept the same as that of FLV-MoS2. The Raman spectra of both FLV-MoS2 and TiO2 are shown in Fig. 3e and TiO2 was confirmed to be in the anatase phase. A comparison of the disinfection performances of FLV-MoS2 and TiO2 is shown in Fig. 3f. Owing to the large 3.2 eV bandgap of TiO2 (which corresponds to a light wavelength of 387 nm), illumination under the solar spectrum with an ultraviolet-light filter (UVF) did not induce any photocatalytic ROS generation. Hence, the E. coli concentration decreased just 44% after 120 min of illumination, which is similar to that of the light control. This confirms that TiO2 was unable to harvest visible light for photocatalytic water disinfection. Using real sunlight as the light source (13:00 local time, 30 March 2013, Stanford, California—the spectrum is given in Supplementary Fig. 8), the disinfection performances of FLV-MoS2 and TiO2 were also compared (Fig. 3f). FLV-MoS2 showed a 5 log inactivation of bacteria concentration (>99.999% inactivation efficiency) and no live bacteria were detected within 60 min of real-sunlight illumination. This gives a first-order disinfection rate of 0.18 min−1 (R2 = 0.99) for FLV-MoS2 under real sunlight. TiO2 showed a log inactivation efficiency of 3 log (99.9%) in 60 min, which corresponds to a disinfection rate of 0.12 min−1 (R2 = 0.96). Thus, the photocatalytic effect of FLV-MoS2 was much better than that of TiO2 for E. coli disinfection under both visible light and real sunlight.
Disinfection performance enhanced by additional catalysts
Normally, the performance of a photocatalyst is governed by two critical properties, namely (1) the utilization of the solar spectrum by the semiconductor photocatalyst and (2) the efficacy of the generated electron–hole pairs for ROS production. For the second property, there is competition between the generated electrons and holes to produce ROS and other physical processes, such as electron–hole recombination, trapping in the material during diffusion and participating in other reactions (such as hydrogen evolution, oxygen reduction and oxidation). Semiconductor materials are usually nonspecific in catalysing ROS production. Hence, to promote the production of ROS, the addition of catalysts is crucial. Nano-size noble metals are used to decorate the semiconductor material surface to serve as ROS catalysts and to form metal–semiconductor junctions to enhance electron–hole separation. It has been shown in the literature that, with the addition of nanosize noble metals, ROS production can be increased dramatically45,46,47.
To accelerate the photocatalytic disinfection rate, 5 nm of Cu or Au was deposited onto the FLV-MoS2 to catalyse the ROS generation and also to improve the electron–hole pair separation. Cu and Au are good catalysts for ROS generation and they preferentially catalyse the two-electron oxygen reduction reaction for H2O2 generation over a four-electron transfer reaction (Supplementary Fig. 9). Cu and Au are deposited via a thermal evaporation process. The morphologies of as-grown Cu–MoS2 and Au–MoS2 were characterized by TEM and scanning electron microscopy (SEM), and the images are shown in Fig. 4a and Supplementary Fig. 10. The thermal evaporation process did not change the morphology of the MoS2 film. No obvious intercalations of Cu or Au were observed because the layer spacing was ∼0.63 nm and ∼0.62 nm for Cu–MoS2 and Au–MoS2, respectively, which is similar to that of pristine FLV-MoS2, ∼0.60 nm (Supplementary Fig. 11). The existence of the Cu and Au films was confirmed by X-ray photoelectron spectroscopy (XPS) (Fig. 4b). The XPS data showed signature peaks for Cu 2p3/2 at 932.6 eV and for Au 4f7/2 at 83.9 eV. Besides the catalytic effect, Cu and Au can also increase the separation of electron–hole pairs. The work functions of Cu and Au are 4.7 and 5.1 eV, respectively. When in contact with p-type FLV-MoS2, the Schottky junction facilitates the electron migration to the Cu or Au surface and the electrons participate in the oxygen-reduction reactions to produce ROS, as explained in the schematic in Fig. 4c. This should further promote the oxygen-reduction reaction for ROS generation.
The disinfection performances of Cu–MoS2 and Au–MoS2 are shown in Fig. 4d. A solar simulator with a UVF was used as the light source, as in the disinfection experiments. For FLV-MoS2 without any catalyst, after 120 min of illumination no live E. coli were detected. With 5 nm of Cu, the disinfection rate was enhanced and within only 20 min no live bacteria were detected, which is equivalent to a disinfection efficiency of >99.999%. For Au–MoS2, the illumination time to achieve >99.999% disinfection efficiency was also shortened relative to FLV-MoS2 to 60 min. The disinfection rates for Cu–MoS2 and Au–MoS2 were 0.57 min−1 (R2 = 0.99) and 0.19 min−1 (R2 = 0.96), respectively, under a solar simulator with a UVF. These are much faster compared with that of FLV-MoS2, 0.087 min−1 (R2 = 0.99). The disinfection enhancement confirmed the efficacy of Cu and Au to separate electron–hole pairs more efficiently and catalyse the oxygen reduction reactions for ROS generation. With Cu and Au as catalysts the disinfection rate is enhanced sixfold and twofold, respectively. After the photocatalytic disinfection experiments, the concentration of Cu or Au was measured. No detectable Au was present in the water sample and the Cu concentration was 3.25 ppb (µg l–1), which is <1% of the Cu catalyst mass loading.
Photocatalytic disinfection mechanism
Furthermore, to prove directly the disinfection mechanism by ROS in the FLV-MoS2 and Cu–MoS2 systems, and also to prove the catalytic effect of Cu, ROS concentrations in both systems were measured and a series of ROS-scavenging experiments were conducted. The results are shown in Supplementary Figs 12–14 and Supplementary Discussion. It was shown that four ROS (O2•−, 1O2, H2O2 and OH•) were present in both the FLV-MoS2 and Cu–MoS2 systems with concentrations of H2O2 > O2•− > 1O2 > OH•. Moreover, for each ROS, the concentration in the Cu–MoS2 system was higher than that in the FLV-MoS2 system. This is consistent with the fact that Cu–MoS2 demonstrated a higher disinfection rate and also proves the effectiveness of Cu in catalysing ROS generation. Six scavenger-quenching experiments were conducted to study the disinfection contribution of each ROS as well as of the electrons and holes. The results suggest that H2O2 has the strongest effect in inactivating bacteria, and that oxygen-reduction-related ROS dominates the disinfection effect. All these results are consistent with the band structure of MoS2, which favours oxygen reduction to produce ROS. Also, FLV-MoS2 and Cu–MoS2 proved to be stable after multicycle tests (Supplementary Fig. 15).
The bacterial disinfection performance of Cu–MoS2, with a disinfection rate of 0.57 min−1 (R2 = 0.99) under a solar simulator with a UVF is, so far, among the best based on a review of the literature in terms of speed and log reduction in bacterial concentration (Fig. 4e and Supplementary Table 1). Compared with those that achieve total disinfection (no live bacteria detected) within 30 min, such as ZnO/Cu (ref. 48), graphene oxide (GO)–CdS (ref. 34) and sulfonated graphene oxide (SGO)–ZnO–Ag (ref. 35), the material mass loading used in the case of Cu–MoS2 is ∼60 times less than that of the others. However, for fixed photocatalysts grown on substrates with a similar material mass loading, such as TiO2–CdS (ref. 25), the disinfection rate of Cu–MoS2 (>99.999%) is much faster than that of TiO2–CdS (∼90%) in 20 min. Hence Cu–MoS2 shows great advantage as a new effective visible-light photocatalyst for water disinfection.
In conclusion, we have demonstrated the efficient harvesting of visible light for photocatalytic water disinfection with a novel material, FLV-MoS2. By decreasing the domain size, the bandgap of MoS2 was increased from 1.3 eV (bulk material) to 1.55 eV (FLV-MoS2). This enabled the FLV-MoS2 to generate ROS successfully for bacteria inactivation in water. The FLV-MoS2 showed a faster disinfection than the most-studied photocatalyst, TiO2. With the additional deposition of Cu or Au to assist electron–hole pair separation and also to catalyse the ROS production reactions, FLV-MoS2 showed a rapid inactivation of >99.999% bacteria in only 20 or 60 min, respectively. The promising performance of FLV-MoS2 and Cu–MoS2 on bacteria shows great potential as photocatalysts for the visible-light inactivation of pathogens (bacteria, viruses and protozoa) in water.
Edge-terminated MoS2 films were grown inside a single-zone, 12 inch horizontal tube furnace (Lindberg/Blue M) equipped with a 1 inch diameter quartz tube. For the 40 nm FLV-MoS2 film, the substrates (1 × 2 cm) were sputtered with a 10 nm thick Mo film as a precursor and then sulfurization at a base pressure of 100 mTorr Ar environment. The tube furnace was quickly raised to the reaction temperature of 500 °C in 20 min and kept at 500 °C for 10 min of reaction41.
Bacteria (E. coli (JM109, Promega and ATCC K-12) and Enterococcus faecalis (ATCC 19433)), were cultured to log phase, harvested by centrifugation at 900g, washed twice with deionized (DI) water and suspended in DI water to ∼106 c.f.u. ml–1. Photocatalytic disinfection was performed using a solar simulator (Newport) calibrated at air mass 1.5G (100 mW cm–2) as the light source and a UVF (Clarex) and red-pass filter (Clarex NIR-70) to tune the light spectrum. The FLV-MoS2 sample size was 1 cm × 2 cm × 40 nm and the water volume was 25 ml. Bacterial concentrations were measured at different times of illumination using standard spread-plating techniques. Each sample was serially diluted and each dilution was plated in triplicate onto trypticase soy agar and incubated at 37 °C for 18 h. The experiments for the solar-disinfection performance were conducted in duplicate. The light spectra (solar simulator with a UVF or solar simulator with a red-pass filter) were measured at the same place at which the sample was set during the disinfection experiment. The solar spectrum during the real-sunlight experiment was estimated using a simple model of the atmospheric radiative transfer of sunshine. The disinfection rates (k) were estimated using Chick's law: ln(C/C0) = −kt where k is in units of per time, C is the concentration of bacteria and C0 is the concentration at t = 0 of the experiments. The slope of the line of best fit to ln(C/C0) versus t for each experiment was used to estimate k.
Characterizations were carried out using TEM (aberration-corrected FEI 80-300 environmental Titan (S) TEM microscope at 300 keV), Raman spectroscopy (WITEC Raman spectrometer), XPS (SSI SProbe XPS spectrometer with an Al Kα source), SEM (FEI Nova NanoSEM 450) and a scanning Kelvin probe (KP Technology Model 5050). The absorption measurement used a xenon lamp (69911, Newport) as the light source coupled with a monochromator (74125, Newport).
We acknowledge the Stanford facilities, Stanford Nanocharacterization Laboratory and Soft & Hybrid Materials, for characterization. Y.C. acknowledges support from the US Department of Energy, Basic Energy Sciences, Materials Sciences and Engineering Division, under contract DE-AC02-76SF00515. We thank G. M. Stewart for his help with the schematic drawing.
About this article
Facile synthesis of cubic Ag/Ag2O composites and its shape-dependent photo-catalytic activity examination
Journal of Materials Science: Materials in Electronics (2019)