Average values of EE,O for different wavelength irradiations in E. coli inactivation in water.
Ultraviolet light-emitting diode (UV-LED) is a newly emerging UV light source with a potential of replacing the conventional chemical methods, mercury UV lamps and xenon lamps in water disinfection applications. In this chapter, we will first give a general description on the status of E. coli disinfection in water by UV-LEDs. Then the main text will concentrate on our experimental studies. We will discuss the effects of single and combined UV-LED irradiation on E. coli in water, including the inactivation efficiency, the recover percentage after the UV-LED irradiation, the optimal wavelength for low energy consumption, differences in pulsed and continuous operations of UV-LEDs, effect of UVA-LED followed by UVC-LED irradiation and vice versa, and finally the effect of TiO2-assisted photocatalytic disinfection.
- E. coli
Millions of people including children die every year from infectious diseases caused by various waterborne pathogens . Among the pathogens, a group of bacteria called Escherichia coli (E. coli) is one of the known carrier of the diseases such as diarrhea, urinary tract infections, respiratory illness and pneumonia . Since E. coli are typically found in the environment, foods and intestines of humans and animals, they have been widely used as fecal indicator bacteria in water quality analysis . Numerous countries and world organizations put a limit count of zero per 100 ml E. coli for drinking water. Passing this limit, it is an indication of the presence of faecally related pathogens in water, and hence a potential risk of high level of microbial waterborne disease outbreak . Therefore, different water disinfection methods have been employed using E. coli as an inactivation target either in laboratory tests or in water disinfection plants. Among the different methods, the conventional use of chemicals such as chlorine can lead to introduction of disinfectant-resistance to bacteria , change of water taste and production of odor  and harmful disinfection by-products (DBPs) such as trihalomethane (THM) compounds, and haloacetic acids (HAAs) that are carcinogenic, mutagenic and reproductive toxicants . Ozone is reported as an effective alternative disinfectant to chlorine due to its ability of reducing microbiological challenge to downstream disinfection. However, the ozone is also known in forming DBPs, particularly bromate , that can cause irreversible effects on humans such as renal failure and deafness . The latest water disinfection method employs the use of ultraviolet (UV) light irradiation whose wavelength ranges from 100 to 400 nm. The UV light irradiation is currently attracting extensive attention in water and wastewater disinfection because of it is DBPs-free, and no need of chemicals that can cause ecological problems .
UV light is usually divided into four regions: vacuum (V) UV (100–200 nm), UV-C (200–280 nm), UV-B (280–315 nm) and UV-A (315–400 nm) . Note that, water and air absorb all wavelengths below 190 nm. Therefore, only the wavelengths between 190 and 380 nm can cause biological effects . Absorption of UV light by deoxyribonucleic acid (DNA)/ and ribonucleic acid (RNA) of a microorganism or virus inhibits its normal replication leading to cell death . However, the UV damaged DNA of some microorganisms particularly E. coli is known to undergo repair by mechanisms such as photo-repair that requires light in the wavelength range of 300-500 nm to activate a photolyase enzyme and dark repair that is light independent [13, 14]. This can greatly decrease the UV light disinfection efficiency hence posing a great chance of health risks of infection. The common UV light sources include: the sun, mercury pressure lamps, xenon lamps and newly emerging UV-light emitting diodes (UV-LEDs). Although the sun gives a cheap and green natural source of light, it is mostly unreliable and only UVA, and approximately 10% of UVB light reaches the earth’s surface . Mercury pressure lamps which exists in two types: low pressure (LP) and medium pressure (MP) mercury lamps emitting a monochromatic light at a wavelength of 254 nm and polychromatic light at a broad range of 185–600 nm respectively , are the commonly used UV light sources in the current water disinfection systems . However, these lamps are usually characterized with fixed wavelengths and limitations like short bulb lifetime, low energy efficiency, high operating temperatures and environmental pollution due to mercury . On the other hand, xenon lamps are characterized by a broad range of wavelength (200–1100 nm), with 40% being UV consisting of UVC, UVB and UVA in ratio about 20%, 8% and 12%, respectively . Therefore, the xenon lamp can exhibit both photochemical effect due to the effect of the UV light, photophysical and photothermal effects due to its high intense pulses . The three multi-target effects can lead to complete destruction of the cell wall and the nucleic acid structure of a microorganism . In addition, the xenon lamps have high penetration, high energy conversion, no pre-heating is needed, faster start-up and no ozone generation . Although the xenon lamp exhibits the above mentioned advantages over the sun and mercury pressure lamps, they have a high energy demand which is un-preferable especially in developing countries. The lamps are also limited in adjusting the duty rates and pulse frequency due to overheating that will affect disinfection efficiency . The newly emerging UV-LEDs are characterized with diversity in wavelengths within the UV range and have advantages such as environmental friendly (no mercury), compact and durable, faster start-up, potential to minimize energy consumption, longer lifetime, and a high frequency switching [23, 24, 25].
Therefore, due to their characteristics and advantages, the UV-LEDs have arisen as a very promising UV light sources in water disinfection applications as demonstrated in literature [23, 26, 27, 28]. Especially, UV-LED reactors can be utilized in small scale since they can be photovoltaic powered, which is convenient in remote areas since they can be photovoltaic powered [29, 30, 31]. Although the wall plug efficiency (WPE) of UV mercury lamps (15–35%) is higher than that of UV-LEDs (10%), the latter is expected to be improved significantly, being similar to the case seen in visible LEDs whose WPE is currently around 80% [32, 33]. In water disinfection, the UV-LED irradiation can be applied in two modes: (i) pulsed light (PL) and (ii) continuous wave (CW) mode. Whereas PL irradiation is a fast non-thermal technology for decontamination based on the application of short pulses of high intensity of light , CW application on the other hand is based on the application of low light intensity . Furthermore, the diverse nature of the UV-LED wavelengths allows for tailored irradiation in which the wavelengths can be irradiated at the same time (simultaneous) or one after the other (sequential). During the disinfection applications, the mechanism of the two irradiation modes can either be photolytic or photocatalytic. In photolytic disinfection, only UV light is involved such that the absorbed photons inactivate the pathogen . Meanwhile, photocatalytic disinfection involves combining UV light and a photocatalyst such as TiO2, that has the ability to absorb UV light of appropriate photon energy (Eq. (1)), and in an air-saturated or water environment, radicals such as OH• and •O2− that are highly destructive towards microorganisms are produced . Therefore, this chapter discusses effects of UV-LED irradiation on E. coli in water, including inactivation efficiency, recover percentage after the UV-LED irradiation, and energy consumption, in terms of single and combined wavelength, PL and CW operations, simultaneous and sequential modes, and finally the effect of TiO2-assisted photocatalysis.
where E is the photon energy, h is the plank’s constant =, c is the speed of light in a vacuum = and is the wavelength of the UV light (m).
2. Indices of inactivation and repair performance for UV-LED disinfection
2.1 Evaluation of inactivation
2.1.1 Inactivation efficiencies
The inactivation efficiency of E. coli was analyzed by calculating log inactivation using Eq. (2).
where N0 and N are the colony count (CFU/mL) before and immediately after inactivation, respectively.
2.1.2 Synergistic inactivation efficiencies
Synergistic effect of combined wavelengths on the E. coli inactivation is compared from the results of log inactivation by combined UV-LEDs and the results from the sum of log inactivation by individual UV-LEDs. Therefore, the synergy values were calculated using the relation:
Synergy (Log units) = Log inactivation by combined UV-LEDs – Sum of log inactivation by individual UV-LEDs.
2.2 Evaluation of repair
2.2.1 Repair efficiencies
where N0 is the cell number before UV irradiation (CFU/mL), N is the immediate cell number after UV irradiation (CFU/mL), Nt is the cell number after repair for a period of time, t (CFU/mL).
where S is the survival ratio at time t (%); N0 and Nt have the same meaning as above.
2.2.2 Repair kinetics
220.127.116.11 Modeling photo-repair
where Sm is the maximum limit of the microorganisms’ survival by repair and S0 is the survival ratio immediately after UV irradiation, k2 is the growth second-order repair rate constant.
Note that k2 is not a pure repair rate constant, it is rather a model parameter that is adjusted to predict the experimental data whose physical meaning is related to the time required to reach Sm and then the stabilization phase [38, 39]. Therefore, a pure repair rate constant, K (Eq. (6)) can be obtained from the derivatives of Eq. (5) and its maximum value, Kmax (Eq. (7)) is obtained when S reaches half of Sm .
18.104.22.168 Modeling dark repair
where M is the mortality, a zero-order decay rate constant, while the other parameters have the same meaning as in Eq. (5). Note that, S, Sm, S0, k2, M and t in Eqs. (5) and (8) have a clear physical significance.
2.3 UV-LED technical parameters
2.3.1 Emission spectrum and optical power
The action spectrum of a microorganism is directly related to the LED emission spectrum i.e., the wavelength and the full width at half maximum [41, 42, 43, 44]. Therefore, the determination of the LED emission spectrum before any experimental study is crucial. In this chapter, UV-LEDs with emissions at 265, 280, 310 and 365 nm, optical power of 1.8, 1.6, 1.3, 100 mW respectively at current of 20, 20, 20, 350 mA achieved at voltages of 6.0, 4.0, 6.0 and 4.0 V respectively (Great Bright Company, China) were used. The optical power was measured by an integrating sphere. Meanwhile the emission spectra measured with Spectro 320 Optical Scanning Spectrometer exhibited peak wavelengths at 267, 275, 310 and 370 nm with full widths at half-maximum of about 12, 10, 9 and 8 nm respectively (Figure 1).
2.3.2 Fluence measurement
The log inactivation of most pathogens is proportional to the applied UV light fluence as given in Eq. (9), where k is the inactivation rate constant that varies from one microorganism to another.
Therefore, determining of fluence is critical for UV-LED disinfection applications. The common UV fluence determination methods include: Radiometry and chemical actinometry (iodide-iodate (KI) and ferrioxalate (FeOx) actinometry). For UV-LEDs, fluence determination protocol employing the two methods for pathogen inactivation is well described in . Therefore, this chapter employed radiometry only in the UV-LED fluence determination.
Average incident irradiance, (mW/cm2) was first determined using IL-1700 radiometer with a SED 270 detector (International Light, USA), at the same distance as the water surface level from the UV-LEDs (L cm). Average fluence (mJ/cm2) inside the Petri dish was then calculated using Eq. (10) .
where PF the petri factor, DF the divergence factor calculated using Eq. (11), WF the water factor calculated using Eq. (12), RF the reflection factor taken to be 0.975 , t (s) the exposure time and CF is the collimation factor which was taken to be 1.
where L (cm) is the distance between microbial suspension surface and the UV-LED and D (cm) the microbial suspension depth (Figure 2).
where I (mW/cm2) and Iλ (mW/cm2/nm) are the total radiant power of the UV-LED and the radiant power at λ of the UV-LED, respectively, αλ (cm−1) is the decadic absorption coefficient of the microbial suspension at λ, and D (cm) is the microbial suspension depth. The decadic absorption coefficient is the absorbance for 1 cm path length.
2.3.3 Electrical energy determination
The electrical energy (EE,N) for a specific N-log inactivation of microorganisms can be determined using Eq. (13).
where EE,N is the electrical energy for a specific -log reduction of each sample, (in kWh/m3), d (cm) is the internal diameter of the Petri dish (Figure 2) and FN is the fluence required for N-log inactivation (mJ/cm2). The value of 3.6 103 is a unit conversion constant for W and kW, s and h, mL and m3. V is the volume of the sample (mL). C is the wall plug efficiency calculated using Eq. (14)  and WF is the water factor calculated using Eq. (12) .
where Poutput is optical power (mW) of the UV-LEDs, Pinput is the applied electrical power (mW), IA is the applied current (mA), VA is the applied voltage (V), and FA is the radiant flux (mW).
3. Disinfection performance of UV-LED
The UV-LED disinfection efficiencies were discussed in four parts: (i) inactivation; (ii) repair; (iii) synergistic effect; and (iv) electrical energy efficiency.
3.1 E. coli inactivation efficiency
Comparative experiments with or without TiO2 confirmed that, after 40 min of stirring in the dark, no inactivation occurred (data not shown). This indicates that UV light is the key requirement in both photolytic and photocatalyic inactivation. In both the photolytic and photocatalytic experiments, lower wavelengths were found to have a higher inactivation efficiency than longer wavelengths (267 > 275 > 310 > 370 nm) . Specifically, in photolytic inactivation, an average fluence of 5, 7, 800 and 900 mJ/cm2 was required by the 267, 275, 310 and 370 nm UV-LEDs, respectively per order of log inactivation. Note that, a 4-log inactivation is required especially in Austria and Germany  in the inactivation of most microorganisms. Therefore, the 267 and 275 nm UV-LEDs required an average fluence of 12 and 15 mJ/cm2, respectively for the 4-log to be achieved in E. coli inactivation. Meanwhile the other UV-LEDs required a relatively higher fluence for the same 4-log inactivation to be achieved [48, 49]. This finding indicated that UVC wavelengths have a higher germicidal effect in the inactivation of E. coli as also confirmed by their relatively higher average inactivation rate constant (k) of 0.4 and 0.3 for the 267 and 275 nm UV-LEDs, respectively compared to insignificant <0.03 for the 310 and 365 nm UV-LEDs. The finding was also consisted with the other studies in literature as reviewed in Ref. . The DNA of most microorganisms is believed to have an absorption maximum of light between 260 and 270 nm , hence confirming the findings.
In photocatalytic disinfection, addition of TiO2 (1.0 g/L) resulted an interesting finding. Whereas the inactivation efficiency was increased in both the 310 and 370 nm UV-LEDs by the addition of TiO2, that for the 267 and 275 nm UV-LEDs was drastically decreased . Note that, anatase phase of TiO2 that was used in our work has a bandgap of around 3.20 eV . Therefore, in an air saturated or water environment, UV photon energy, is required to induce the generation of the reactive OH• radicals from the TiO2 surface. The photon energy from the 267, 275, 310 and 370 UV-LEDs was calculated and found to be , , , and , respectively. This indicates that, UVA wavelength is the most appropriate in photocatalytic disinfection as was confirmed by a significant enhanced inactivation efficiency by the 370 nm UV-LED when anatase phase of TiO2 was added in the E. coli suspension . The enhanced inactivation efficiency by the 370 nm UV-LED with TiO2 is therefore attributable to the huddle effect of the UV photons and radicals. Other than their lower capability of radical production from the TiO2 surface due to UV photon energy not within the optimum, the inactivation efficiency by the 267 and 275 nm UV-LEDs decreased with addition of the TiO2 due to a screening effect by the TiO2 which protected the E. coli against the strong UV photon of the UV-LEDs .
In another experiment, PL and CW UV-LED irradiation showed similar inactivation efficiency at equivalent average fluence . Meanwhile, 267 nm UV-LED still had a slightly higher inactivation efficiency than the 275 nm UV-LED (Figure 3), which is in agreement with previous findings explained in the preceding paragraphs and also confirmed by reports in Ref. . Although different UV-LEDs were employed, similar findings were also reported in other studies reported in literature [55, 56, 57]. However, an enhanced inactivation efficiency by PL over CW UV-LED irradiation is reported [58, 59, 60, 61]. These discrepancies could be attributed mainly to unequal fluences between the PL and CW UV-LEDs, which is key in microbial inactivation. PL from xenon lamps is reported to cause enhanced inactivation efficiency than CW UV irradiation by mercury lamps . The finding is due to xenon lamps’ broad-spectrum UV content, short duration intense pulses and the high peak power which can lead to three multi-target mechanisms (photochemical, photophysical and photothermal) . It should be noted that, the PL irradiation produced by xenon lamps is much different from that of the UV-LEDs in terms of emission spectrum, intensity, frequency switching. Therefore, the inactivation mechanisms of the PL xenon lamp may not apply to the UV-LEDs whose wavelengths are just within 200–400 nm and if a single UV-LED is used, almost a monochromatic wavelength is obtained compared with the broad range (200–1100 nm) from the xenon lamp. In addition, the current peak power of the UV-LEDs is still low (mW) which requires more improvements , compared to that of xenon lamps which is relatively high (kW) . Unless the optical power is significantly improved, the E. coli inactivation efficiency by PL and CW UV-LED will still be equivalent. The only significant advantage of PL over CW UV-LED is its ability to suppress the heat generated during the UV-LED operation [54, 56]. This is due to the PL irradiation’s ability to generate heat only during the short pulse and a cooling period can occur between each pulse. This ability was clearly observed when the PL showed a lower solder temperature as compared to the CW UV-LED at similar driving currents and ambient temperature (Figure 4).
Diversity of UV-LED wavelengths allows UV-LED for tailored irradiation like involving 2 or more wavelengths either in simultaneous or sequential manner. It is well known that, irradiation by UVC and UVB can induce lesion formation in the genomic DNA of a microorganism [66, 67, 68]. Meanwhile, irradiation by UVA causes formation of active substances such as reactive oxygen species that have lethal effects to a microorganism . Due to their different inactivation mechanisms, this part of the chapter therefore concentrated only on simultaneous and sequential irradiation involving a combination of UVC(or UVB) and UVA wavelengths. Note that, “UVC(B)” used here and henceforth in this chapter stands for UVC or UVB. Compared to sum of corresponding single wavelength, simultaneous irradiation of 267, 275 or 310 with 370 nm UV-LED led to lower log inactivation values of 1.27, 1.23 and 0.64, respectively. Similarly, lower log inactivation of 0.92, 0.90 and 0.63 was also obtained in sequential irradiation of 267, 275 and 310 nm followed by the 370 nm UV-LED, respectively (Figure 5). These results indicate that the 370 nm UV-LED irradiation could have functioned in repairing the already UV damaged DNA, rather than further damaging it [70, 71]. This assumption could be possible since the 370 nm is within the range of photo-repair light, 300–480 nm [13, 14]. On the other hand, higher log inactivation of 2.15 and 2.13 were achieved in sequential irradiation of 370 nm followed by 267 or 275 nm UV-LEDs, respectively. This log inactivation was also higher than that from the sum of corresponding single wavelength UV-LED irradiations, except for sequential irradiation of 370 nm followed by 310 nm UV-LEDs which achieved 0.98 log inactivation (Figure 5). Although the 370 nm (UVA) radiation can repair an already UV damaged DNA, the radiation on the other hand has an adverse effect when irradiated on un UV damaged DNA . This phenomenon is known as concomitant photo-repair phenomenon in which inactivating light itself has the potential to photo-repair the UV-injured DNA . Note also that, the 310 nm (UVB) is within the photo-repair light (300–480 nm), the 310 nm could have a concomitant photo-repair phenomenon similar to the 370 nm wavelength. While, 310 nm is still found to produce lesions in DNA that damage microorganisms. These findings are consistent with the other studies in literature [71, 73].
3.2 E. coli repair efficiency
As mentioned earlier in the introduction section, E. coli has the ability to undergo repair after damage from UV light irradiation. In all the experiments conducted, photo-repair was more dominant with an average of above 5% of photo-repair and negligible or no dark repair occurred [48, 49], demonstrating that photo-effect is the dominant mechanism of E. coli repair. The dominance of photo-effect in E. coli repair was also reported in other studies in Refs. [40, 74, 75]. Considering the 267, 275 nm UV-LED, the same observation was confirmed by the highest rate of photo-repair constant, > 4% compared to that of dark repair, < 0.02% . By analyzing the photo-repair after photolytic inactivation, 275 and 370 nm wavelengths were found to be appropriate in suppressing the photo-repair. In addition, when the same wavelengths were applied, 275 nm followed after the 370 nm UV-LED irradiation has a much lower percentage of photo-repair compared to the simultaneous irradiation of 275 nm and 370 nm. This observation is attributed mostly to the damage of E. coli’s membrane at 370 nm , and as well as both DNA and proteins at 275 nm . Note that, no significant difference was observed in the percentage of photo-repair for PL and CW UV-LED irradiation . However, the addition of TiO2 led to an insignificant % of E. coli photo-repair (<1%) and for dark repair, mortality was registered . The observation is attributed to the concomitant effect of the photons from the UV-LEDs and the radicals generated from the surface of UV irradiated TiO2 that led to more damage to the E. coli. In addition, the mortality in the dark repair is attributed to a residual disinfecting effect of the .
3.3 Synergistic effect
During the E. coli inactivation, different wavelengths were combined and their synergistic effect was evaluated. The irradiations were performed in both simultaneous and sequential manner. From the results obtained, simultaneous irradiation involving 267/275, 267/310 and 275/310 wavelength combinations from the UV-LEDs did not yield synergy in E. coli inactivation . Note that, the 267 and 275 nm belong to the UVC, meanwhile the 310 nm belong to the UVB. The UVC and UVB have similar inactivation mechanism [66, 67, 68], which explains the absence of synergy in this case. Although UVC(B) and UVA wavelengths are reported to have different disinfection mechanisms as highlighted in the introduction section, interesting findings were found both in simultaneous and sequential irradiation on E. coli inactivation. Simultaneous irradiation of 267, 275, 310 nm and their combination with 370 nm UV-LED led to lower log inactivation compared to the sum of log inactivation of the corresponding single wavelengths. Similarly, lower log inactivation was achieved for 267, 275 and 310 nm followed by 370 nm UV-LED irradiation (Figure 6). These findings highlighted the concomitant photo-repair phenomenon of the 370 nm UV-LED. It should be noted that, the 370 nm is within the range of photo-repair light (300–500 nm). Therefore, other than damaging the E. coli bacteria, the 370 nm light could have performed the role of photo-repair as also discussed in previous studies [71, 72]. No synergy was found for 370 nm followed by 310 nm UV-LED irradiation. However, synergistic effect was found for 370 nm followed by 267 or 275 nm UV-LED (Figure 6). Because the 370 nm light can cause cell membrane damage, when irradiated with UVA first then followed by the UVC wavelengths, more damage was realized, leading to the synergy ultimately. However, irradiating 310 nm UV-LED after the 370 nm could have resulted to the repair of the E. coli since the 310 nm UV-LE is within the photo-repair light, hence absence of synergy in that case.
3.4 Electrical energy efficiency
To make a viable decision in choosing an appropriate UV-LED to be applied in disinfection applications, it is necessary to determine the electrical energy efficiency (EE,N) of the UV-LEDs for microorganism inactivation in water. For the combined wavelengths, the sequential irradiation of UVA followed by UVC-LED showed higher inactivation and repair repression efficiencies of E. coli compared to the other combinations. Therefore, the electrical energy efficiency per order of magnitude (EE,0) was determined only for single wavelength irradiation (in both photolytic and photocatalytic) and UVA followed by UVC(B)-LED irradiation on the E. coli in water (Table 1). In both photolytic and photocatalytic disinfection, the 275 nm UV-LED required lower EE,O. Although the addition of TiO2 to the E. coli suspension led to an increase in the EE,O for the 267 and 275 nm UV-LEDs, that for the 310 and 370 nm UV-LEDs decreased. Meanwhile, for the 370 nm followed by 275 nm UV-LED irradiation, it required lower EE,O than the other combination manners. The lower EE,O for the 275 nm UV-LED, and 370 nm followed by 275 nm UV-LED irradiation is mainly attributed to the higher wall plug efficiencies of these two kinds of UV-LEDs [48, 49]. A similar finding has also been reported in Ref. . Note that, the decrease in EE,O for mostly the 370 nm UV-LED in photocatalytic disinfection is attributed two things: (i) its higher wall plug efficiency; and (ii) its photon energy being within the required to induce radicals on TiO2 surface.
In this chapter, recent achievements about E. coli disinfection in water by UV-LEDs has been highlighted, as well as a general description on UV-LEDs. The main text concentrated more on our experimental studies in which the effects of single and combined UV-LED irradiation on E. coli in water, including the inactivation efficiency, the recover percentage after the UV-LED irradiation, the best wavelength for low energy consumption, differences in PL and CW operations of UV-LEDs, combination with UVA-LED followed by UVC-LED irradiation and vice versa, and finally the effect of TiO2 photo-catalyst, were discussed. Whereas the 267 nm UV-LED showed higher inactivation efficiency, the 275 nm UV-LED was more competitive with comprehensive consideration of higher repressive ability on E. coli repair and higher electrical energy efficiency. For photocatalytic disinfection, the 370 nm UV-LED was the most appropriate. Although PL UV-LED was found to be effective in suppressing temperature rising than CW operation, the two modes showed insignificant difference in E. coli inactivation and repair efficiency. For the combined wavelengths, UVA (370 nm) followed by UVC (275 nm) irradiation was effective in all aspects of inactivation, repair and electrical energy efficiencies.
We acknowledge the support from the National Key R&D Program of China (no. 2016YFB0400803), and the China Government Scholarship Council, no. 2017GXZ023553.
Conflict of interest