Introduction

Our understanding of the nitrogen cycle has shifted from how to enhance nitrogen fixation to promote the production of food and energy to the realization that an elevated level of reactive nitrogen poses a growing influence on human health and environmental systems1,2. Human activities such as fossil fuel combustion, industrial nitrate production, the use of artificial nitrogen fertilizers and the release of nitrogen in waste have drastically altered the global nitrogen cycle1,3. The Haber–Bosch and Ostwald processes are the most well-known industrial nitrate-producing methods for making fertilizer from nitrogen (N2) in the presence of noble metal catalysts at high temperatures and pressures4,5. However, an excess level of nitrate emitted into drinking water systems is toxic to humans, for example, resulting in ‘blue baby syndrome’ in children6. It can also lead to eutrophication and algae blooms in freshwater and estuaries1,2,6.

Nano-sized titanium dioxide (TiO2 <100 nm) is used as an additive worldwide because of its high refractive index, strong UV-light-absorbing capability, and superior sterilizing, deodorizing and anti-fouling properties7,8. It accounts for 70% of the total production of pigments worldwide. Millions of tons of TiO2 are produced annually and its production is expected to increase continually9,10. TiO2 nanomaterials, widely used in white paints, foods, consumer products and household products, are disposed of down drains, discharged in faeces/urine, washed off of surfaces, discharged into sewage and eventually enter wastewater-treatment plants. Furthermore, the intake of nanoscale TiO2 through foods by human beings is estimated to be on the order of 1 mg Ti per kg of body weight per day, with children identified as having a higher exposure rate because of the greater content of nanoscale TiO2 in sweets10.

Thus far, TiO2 has been widely used as an environmentally friendly photocatalyst and additive10,11,12,13. However, there has been no report on direct nitrate production from atmospheric N2 and O2 through TiO2-mediated photocatalysis under ambient conditions and its potential negative influence on the environment, given its increasing use worldwide. In this study, we conduct both indoor and outdoor experiments to explore potential environmental pollution problems induced by such a TiO2-mediated photocatalysis, and to establish a cost-effective photocatalytic process for nitrate production. The first-principles density functional theory calculations are applied to elucidate the formation mechanisms of intermediate gas products on the (001) surface of nano-TiO2. The results suggest that nano-sized titanium dioxide may catalyse the formation of nitrate via the photochemical reaction of atmospheric nitrogen and oxygen.

Results

Photocatalytic nitrate production

Scanning electron microscopy was used to image the surface morphology and structure of the TiO2-carbon paper used in the photocatalytic process. The carbon paper covered with TiO2 displayed a cracked-mud structure, resulting in an increased actual surface area (Fig. 1a,b). The X-ray diffraction spectrum of the TiO2, in which the peaks correspond to anatase and rutile phases (Fig. 1a,b), exhibited the same structure as that of the commercial TiO2 nanoparticles. P25 is a 15/85 mixture of rutile/anatase, whereas E171 is anatase. This result indicates that the coating of TiO2 on the carbon paper surface did not change the structure of the TiO2.

Figure 1: Structural characterization of TiO2 and nitrate concentration profiles.
figure 1

(a) Scanning electron microscopy image and X-ray diffraction spectrum of the P25 on the TiO2-carbon paper, scale bar, 5 μm; (b) scanning electron microscopy image and X-ray diffraction spectrum of the E171 on the TiO2-carbon paper; A represents anatase and R represents rutile, scale bar, 10 μm; and (c) concentration profiles of nitrate in the open photocatalytic reactor filled with ultrapure water (P25 N2/O2: the N2/O2 (3:1/v:v) mixture was purged into the reactor with the P25-carbon paper inserted; the tests were conducted in triplicate; the error bar at each data point reflects the s.d. of the three repeated measurement of concentrations. P25 Air: air was purged into the reactor with the P25-carbon paper inserted; E171 Air: air was purged into the reactor with the E171-carbon paper inserted; P25 Cr(VI): air was purged into the reactor with the P25-carbon paper inserted and 0.01 mM Cr2O72− was dosed as an electron scavenger).

In the photoreactor under UV irradiation, the nitrate concentration increased almost linearly with the reaction time when purified air or N2/O2 (3:1/v:v) gas was purged into the reactor, and an average nitrate-producing rate of 3.51±0.06 mg m−2 h−1 was obtained in the two cases using the P25-carbon paper (Fig. 1c). The tests with N2/O2 gas confirmed that nitrate was directly formed from N2 and O2 through the TiO2-mediated photocatalysis. In addition, the similar levels of the nitrate-producing rate in both systems indicate that the interference of nitrogen oxides in the purified air was negligible for all the indoor experiments. For the E171-carbon paper, the rate decreased to 2.96±0.11 mg m−2 h−1. With the dose of Cr(VI) as an electron scavenger14, the rate decreased to 0.29±0.08 mg m−2 h−1 in the initial 96 h, but then increased to 3.42±0.63 mg m−2 h−1 after Cr(VI) became exhausted (Fig. 1c), implying that the photogenerated electrons (e) might have an important role in such a nitrate-producing process. No nitrate was detected in the reactor when purified air or ultrapure N2 gas was injected but without photocatalysis (that is, in the absence of UV irradiation, in the presence of UV and carbon paper but without TiO2, or without carbon paper and TiO2 at all). In addition, no nitrate was formed when ultrapure Ar gas was injected and UV irradiation carried out. However, a small amount of nitrate with an average producing rate of 0.18±0.06 mg m−2 h−1 was detected when ultrapure N2 gas was injected in the presence of UV irradiation. No detectable ammonia or nitrite was measured in all the experiments. These results indicate that the photocatalytic nitrate-producing reaction occurred with the catalysis by P25 or E171 under UV irradiation. Thus, purified air was directly used in the subsequent indoor experiments. The nitrate-producing rate decreased slightly over time, which was attributed to the acidification of the photocatalyst surface.

Photocatalytic nitrate production under ambient conditions

In the well-controlled climatic chamber at fixed illumination intensity, the nitrate-producing rate initially increased with the increasing relative humidity but decreased once the relative humidity reached a certain height (Fig. 2a). Nitrate production also occurred under solar illumination due to the presence of UV light in sunshine (Fig. 2b,c). Furthermore, the nitrate-producing rate was closely correlated with the sun illumination intensity at a constant relative humidity (100%), which was achieved through sprinkling ultrapure water on the TiO2-carbon paper surface. More nitrate was harvested at noontime (11:00–13:30) and afternoon (13:30–16:00) when a higher illumination intensity was available, compared with that in the morning (8:30–11:00) and at evenfall (16:00–18:30) (Fig. 2b).

Figure 2: Profiles of nitrate production rate under various conditions.
figure 2

(a) Effects of relative humidity on the nitrate production rate and illumination intensity in the climatic chamber; (b) profiles of nitrate production rate and illumination intensity of sunlight in the outdoor photocatalytic process under sunlight irradiation with ultrapure water dropped upon the TiO2 (P25)-carbon paper surface on 20 September 2010, in Hefei, China (31.51 °N, 117.17 °E) and (c) profiles of nitrate production rate and illumination intensity of sunlight in the outdoor photocatalytic process under sunlight irradiation without the dropping of ultrapure water on the TiO2-carbon paper surface on 26 July 2012, in Hefei, China. All the tests were conducted in triplicate, and the error bars indicate the s.d. of three repeated measurements.

The photocatalytic reaction under ambient conditions was affected by the relative humidity and the sun illumination intensity. The greatest nitrate-producing rate was obtained in the morning (8:30–11:00) with the highest relative humidity and medium illumination intensity. The rate decreased with the reduced illumination intensity from noontime (11:00–13:30) to evenfall (16:00–18:30), which had almost the same relative humidity levels (Fig. 2c and Supplementary Fig. S2a). These results imply that such a nitrate-producing process could occur on the surface of ubiquitous areas coated with TiO2 under ambient conditions, and that the reaction rate is closely associated with the relative humidity and sun illumination intensity.

Detected intermediate products

A continuous nitrate-producing reaction was also observed in the photocatalytic experiments carried out in the airtight glass reactor with a quartz glass window (Fig. 3a). Compared with the chromatogram of the gas in the 5-l flask before and after the photocatalytic reaction, an intermediate gaseous product was detected with a retention time of 1.25 min, which was identified as NO by the Fourier transform IR (FTIR) analysis (Fig. 3b). The FTIR analysis of the gas post reaction revealed an increase in the NO concentration of 200 ppb, compared with the samples analysed before the reaction (see Supplementary Fig. S2b, c). No N2O was detected during this process. Thus, NO was the detected intermediate product in this photocatalytic process.

Figure 3: Concentration of nitrate and gas chromatogram.
figure 3

(a) Concentration profiles of nitrate in the airtight glass reactor; the tests were conducted in triplicate and (b) partial enlargement chromatogram of the gas in the airtight reactor before and after the reaction with its entire range shown in the inset. The error bars in a refers to the s.d. of three measurements of nitrate concentrations.

Theoretical analysis of photocatalytic mechanisms

The above observations suggest that NO was the initial product of this type of photocatalytic process on a TiO2 surface. The reaction in the presence of Cr(VI) indicates that photogenerated electrons (e) had an important role in this process, although the involvement of photogenerated holes (h+) could not be neglected. Thus, the following photocatalytic NO formation is proposed.

The intermediate formation pathways (1) and (2) illustrated in Supplementary Fig. S1 show the reactions occurring with O2 and e on the conduction band (CB), and with H2O and h+ on the valence band (VB) of the TiO2 (001) surface (Supplementary Fig. S3), respectively.

To explore the mechanism that has plausible energies and barriers, a number of mechanistic scenarios have been examined. We first describe the results concerning the preferred reaction mechanism on the CB (Supplementary Fig. S1). The geometry structures of the optimized key reaction intermediates are illustrated in Supplementary Fig. S4 (1~5) for each step of NO formation (detailed description is given in Supplementary Note 1). In the photocatalytic process, the ΔrGmθ values of the elementary step and overall process (Supplementary Table S1) indicate that the NO formation on the CB of the TiO2 (001) surface is thermodynamically favourable compared with the positive ΔrGmθ (173.38 kJ mol−1) without added photolytic energy. Moreover, ΔrGm becomes more negative with the increasing temperature (Supplementary Fig. S5), indicating that a high temperature is beneficial to this process. The lower energy barrier (Ea) of the photocatalytic NO formation reaction shown in the overall potential energy graph (Fig. 4a) reveals that the pathway of the CB mechanism is an energetically feasible reaction with a higher reaction rate.

Figure 4: Calculated potential energy profile for NO formation.
figure 4

Calculated potential energy profile of each step for NO formation on the conduction band (a) and the valence band (b) of TiO2 (001) surface, according to Supplementary Fig. S1.

As described in the preferred VB mechanism (Supplementary Fig. S1), the structures of the key reaction intermediates are optimized and illustrated in Supplementary Fig. S4, numbered from 6–12 sequentially. The overall potential energy graphs for all of the steps with the transition states are illustrated in Fig. 4b. The thermodynamic parameters of the elementary step (Supplementary Table S2) indicate that the VB mechanisms can proceed with |e|U value of 2.56 eV in Steps V1~V3. For instance, Step V2 proceeds through the splitting of H2O on a Ti-5c site (7+h+→8+H+) to produce an adsorbed OH that simultaneously leads to Ti-O rupture (State 8, Supplementary Figs S1 and S4). In the photocatalytic process, the ΔG value of the elementary step involving proton and hole should deduct the energy (|e|U) of electron-hole pair generation. The U is the electropotential of a hole (at the maximum of the VB) with respect to standard hydrogen electrode (U=0, ΔGH=0), and thus, |e|U represents the energy required to generate the electron-hole pair (a detailed description is given in Supplementary Note 2). However, Step V4 is the rate-controlling step in the VB mechanism. The |e|U value of 2.56 eV is not sufficiently large to overcome the high Ea to drive the elementary reaction. In Step V4, the second H2O molecule attacks the O-O bond to form the structure of Ti-O-OH O-Ti (State 10), and proton removal with the photocatalytic evolution leaves oxygen radicals. Considering an overpotential of 1.22 V applied on the (001) surface, the ΔG of Step V4 can become negative because the |e|U equals 3.78 eV (2.56+1.22 eV). Furthermore, considering that the minimum CB of TiO2 is higher in energy relative to standard hydrogen electrode by about 0.2 eV, the UV light above 3.98 eV (0.2+3.78 eV, ~310 nm) is sufficiently large to drive the photocatalytic NO formation reaction with H2O and h+ on the VB of the TiO2 (001) surface.

The above mentioned protocols demonstrate that such a nitrate-producing process can be driven by the TiO2-mediated photocatalysis. Both thermodynamic and kinetic analyses reveal that the NO-forming process can spontaneously occur through the proposed CB mechanism, with a low Ea in every step. However, the VB mechanism with a high Ea needs overpotential to drive the photocatalytic reaction. As a consequence, the CB mechanism with a low Ea can be the dominant pathway for the NO formation from quantum chemical calculations, and the VB mechanism can also possibly occur with h+ at a low reaction rate. This is in agreement with the experimental results when both Cr(VI) dosing and ultrapure N2 injection were applied.

Discussion

As the most important and widely used white paint component (~110 μg Ti mg−1) in the coating industry, the application of TiO2 on the exterior walls of buildings and its consequent exposure to the atmosphere under UV light (for example, sunshine) irradiation is inevitable10, and in some cases extensive. Millions of tons of TiO2 are produced per year and nearly 70% is used as a pigment in paints. Thus, the continual increasing production of nitrate on the surfaces of a large number of TiO2-coated buildings worldwide may be expected. Given the calculated nitrate-producing rate under ambient conditions (Fig. 2c), the increasing nitrate amount is estimated to be thousands of tons per day with the rough assumption that half of the pigment TiO2 is exposed to the atmosphere under sunshine irradiation and 1 kg of TiO2 can cover 5–10 m2 of a building’s surface15. Considering the large amount of pigment TiO2 used worldwide, the global production of nitrate per year from such processes, which is approximately equal to the NOx-N emission of biogenic substances from soils into the atmospheric environment in the United States of America in 2002 (ref. 16), could be impressive. Although these estimates are necessarily rough, this value suggests an undiscovered effect on the global nitrogen cycle as an abiotic process17,18. Owing to the universal personal use of TiO2 to provide whiteness and opacity to products such as coatings, plastics, papers, inks, food and pills, an increasing number of toilets with TiO2 coating inside, and buildings with TiO2 coating outside in the form of glass coatings, cement, tiles and ceramics, the nitrate formation is expected to constantly increase10. The excess presence of nitrate in the environment is toxic to humans and aquatic lives. If the nitrate produced in TiO2-coated buildings is washed into water bodies, it can also contribute to water eutrophication and algae blooms, which are recognized as environmental disasters18.

It should be noted that nano-sized TiO2 particles (for example, <100 nm) are also widely used as a UV absorber in cosmetic and personal-care products, with a TiO2 content ranging from 0.7 to 5.6 μg mg−1 (ref. 10). Our results imply that although TiO2 particles are able to protect the skin from UV light, they may also produce nitrate simultaneously. At the present stage, it is unclear whether the nitrate formed on skin surfaces can penetrate the body, but it warrants further investigations.

Reactions 1 and 2 suggest that NO can be produced in this process and released into the air. The NO formed from the TiO2-mediated photocatalysis would immediately react with O2 and H2O to form NO2 and nitrate19. NO is a prominent air pollutant that can lead to the formation of acid rains, photochemical smog or ozone and threaten human health20,21,22,23. In addition, when NO is exposed to oxygen, it can be converted into NO2, a brown toxic gas and air pollutant24. The unstable nitrous acid (HONO)25,26, which can be formed by NO2 disproportionation on a range of surfaces including TiO2, fresh soot aerosol surfaces, glass, quartz and steel, can be rapidly decomposed into NO2, NO and H2O27,28, or be converted into NO and HO through photolytic decomposition29. Thus, HONO may also participate in the TiO2-mediated photocatalytic nitrate-producing process. It is worth noting that the nitrate-producing rate under outdoor conditions was higher than that in the indoor experiments (Figs 1c, 2). This might be partially attributed to the presence of nitrogen oxides outdoors. The greatest nitrate-producing rate under ambient conditions was obtained in the morning (8:30–11:00) with the highest relative humidity and medium illumination intensity (Fig. 2c). This might be partially associated with the peak NO/NOx concentration in the morning rush hour in urban areas26,30. Such a rapid nitrate reduction can help decreasing the NO/NOx concentration in the air. Therefore, attention should be given to the worldwide use of TiO2 and its potential environmental impacts.

Such a direct TiO2-mediated nitrate-producing process under UV or sunlight irradiation may have potential industrial applications because of its gentle reaction conditions and the abundant and free raw materials, that is, atmospheric N2 and O2. In the current well-known industrial nitrate-producing processes, for example, the Ostwald process4,5, NH3 is used as the raw material and noble metal catalysts and high reaction temperatures are required to complete the nitrate-producing reaction. However, this TiO2-mediated photocatalytic process could be completed under sunshine at ambient temperature, and thus has application potential. Further investigation is warranted to increase both its nitrate-producing rate and yield for practical use.

In summary, we have demonstrated that nitrate could be formed from abundant atmospheric N2 and O2 on TiO2 surfaces under either UV or sunlight irradiation. NO is suggested as an intermediate product in this process and its formation mechanisms are elucidated by both experimental measurements and theoretical calculations. This might be the first report on nitrate production from air in relation to the globally increased use of TiO2 (for example, P25 and E171) under ambient conditions. Such a catalytic reaction has a double-edged consequence: it has the potential to become an increasing pollution source, but it could also be used as an alternative cost-effective alternative to the Ostwald process for nitrate production.

Methods

Setup and operation of photocatalytic reactors

A quartz glass cylinder (6.5 cm in height and 6 cm in diameter) filled with 150-ml ultrapure water, except when 0.01 mM Cr2O72− was added as an electron scavenger14, was used as the photoreactor for photocatalytic reaction. The commercial TiO2 nanoparticles, P25 (Degussa Co., Germany), and the widely used food-grade TiO2, E171 (<250 nm, Jianghutaibai Co., China), were, respectively, painted on a chemically stable carbon paper (GEFC Co., China)31 and were inserted into the photoreactor for tests. A 30-W low-pressure mercury lamp was used as the source of UV light. The lamp was fixed outside the reactor against the TiO2-carbon paper surface at a 10 cm distance, with an intensity of 4.81±0.51 μW cm−2 at 253.7 nm. Air was purged into the reactor constantly and slowly through a flask, half of which was filled with ultrapure water for cooling and purification. In the control experiment, a mixture of ultrapure N2 (99.999%) and ultrapure O2 (99.995%) (3:1/v:v) was purged into the reactor 1 h before UV irradiation and lasted 72 more hours under UV irradiation. Blank experiments were also carried out with the injection of either purified air or ultrapure N2 gas but in the absence of either UV irradiation or TiO2 photocatalyst (with or without the carbon paper inserted into the photoreactor), or with the injection of ultrapure Ar gas or ultrapure N2 gas and in the presence of UV irradiation. The solutions in the reactor were stirred constantly. The morphology of the prepared TiO2-carbon paper was imaged using SEM (FEI Co., the Netherlands) and its crystal structure was characterized using X-ray diffraction (X’ Pert PRO, Philips Co., the Netherlands). The effect of relative humidity on this photocatalytic process was evaluated in a climatic chamber under a constant temperature (25±0.1 °C) but with different set humidities.

Outdoor experiments under solar light irradiation were also carried out. Ultrapure water was dropped on the TiO2-carbon paper surface at a rate of 3 ml h−1 to eliminate the interference of relative humidity and was collected. Then, ultrapure water was used to wash the TiO2-carbon paper and the beaker several times, and the water samples were collected into a 25-ml volumetric flask for measurement every 2.5 h. A luminometer (ZDS-10, Xuelian Co., China) was used to measure the illuminance intensity on the TiO2-carbon paper surface every 0.5 h. Experiments using the TiO2 (P25 or E171) on carbon paper under solar light irradiation but in the absence of ultrapure water were also conducted for comparison.

Analysis of intermediates and final products

To identify the possible intermediate products formed in this process, a separate photocatalytic experiment was carried out in a 250-ml airtight glass reactor, in which 150-ml ultrapure water and 0.1-g TiO2 (P25) powders were added. The reactor was equipped with a quartz glass window (2 × 4 cm2) and a sample outlet capped using a rubber stopper. The liquid samples were taken from the sample outlet of the reactor by using an injector and were used for analysis. There was a gas circuit between this reactor and a 5-l flask filled with air. A peristaltic pump was used to ensure sufficient airflow through the reactor during the 3-day reaction process. The gas samples in the airtight reactor before and after the 3-day photocatalytic reaction were analysed with a gas chromatograph (6890, Agilent Co., USA) equipped with an electroncapture detector, and a FTIR (TENSOR 27, Bruker Co., Germany) equipped with a mercury-cadmium-teluride detector for qualitative and quantitative analyses of the potential NOx intermediates.

Both nitrate and nitrite were analysed using an ion chromatograph (ICS-1000, Dionex Co., USA) and ammonia was determined using a spectrophotometer (UV2540, Shimdzu HIMADZU, Japan) at a wavelength of 420 nm after the Nessler reaction. All of the organic and inorganic reagents used were of analytical grade, unless otherwise stated. All of the solutions were prepared with ultrapure water (Millipore Co., USA). All of the tests were conducted in triplicate.

Quantum chemical calculations

The formation mechanisms of intermediate gaseous products on the (001) surface of the nanoscale TiO2 was explored with first-principles density functional theory calculations (details are provided in Supplementary Methods). The (001) facet, which is the minor surface in the equilibrium shape of the anatase nanocrystals in TiO2 P25 (Degussa), is found to be the dominant active site and contributes to the unique properties of TiO2 nanoparticles in the form of anatase. Its existence in anatase TiO2 has been demonstrated theoretically and experimentally32,33,34. The structures of the reactants, intermediates and products in the reaction process at the (001) surface of nanoscale TiO2 were optimized in water solvation. The potential energy profiles and the Gibbs free energy changes of each step for the intermediate formation were calculated. Geometry parameters of O-O and N-N in the intermediate states of pathways were calculated. Finally, the reaction mechanisms for the intermediate product formation were elucidated.

Additional information

How to cite this article: Yuan, S.-J. et al. Nitrate formation from atmospheric nitrogen and oxygen photocatalysed by nano-sized titanium dioxide. Nat. Commun. 4:2249 doi: 10.1038/ncomms3249 (2013).