Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Effect of UV-LED fluence rate and reflection on inactivation of microorganisms Hajimalayeri, Adel 2016

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2016_november_ hajimalayeri_adel.pdf [ 3.99MB ]
Metadata
JSON: 24-1.0314125.json
JSON-LD: 24-1.0314125-ld.json
RDF/XML (Pretty): 24-1.0314125-rdf.xml
RDF/JSON: 24-1.0314125-rdf.json
Turtle: 24-1.0314125-turtle.txt
N-Triples: 24-1.0314125-rdf-ntriples.txt
Original Record: 24-1.0314125-source.json
Full Text
24-1.0314125-fulltext.txt
Citation
24-1.0314125.ris

Full Text

EFFECT OF UV-LED FLUENCE RATE AND REFLECTION ON INACTIVATION OF MICROORGANISMS   by  Adel Hajimalayeri  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  MASTER OF APPLIED SCIENCE in The Faculty of Graduate and Postdoctoral Studies (Chemical and Biological Engineering)   THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)   August 2016   © Adel Hajimalayeri, 2016 ii  Abstract  Ultraviolet light emitting diodes (UV-LEDs) are emerging as viable alternative to traditional UV lamps for water disinfection. UV-LEDs possess many advantages, some of which include robust configuration, long lifetime, and small size. With UV-LED being an incipient technology, there are also a number of knowledge gaps and challenges towards their widespread applications to water disinfection. The research presented here focuses on addressing two such knowledge gaps; first, the impact of reflection from the reactor wall on the fluence rate distribution inside a UV-LED reactor, and second, the effect of UV fluence rate on the inactivation of microorganisms.  Three UV-LEDs: 265nm, 275nm, and 285nm were utilized as the UV sources. The test microorganisms were E.coli ATCC 11229 and Fecal Coliform. Teflon (OD98), stainless steel and aluminum petri dishes were built and used as reflective materials. A new experimental method for evaluating the absorbed energy by a water sample during the UV-LED’s irradiation was proposed. This method was used to derive the E.coli kinetic models at wavelengths of interest. These kinetic models were useful for calculating the inactivation improvement due to reflection inside different reactors. At the same experimental condition, the E.coli log inactivation inside different reactors (reflective and non-reflective as control) was calculated and the kinetic models were utilized to back calculate the correspondent value for the UV fluence in the system. The ratio of the absorbed energy by the solution in a reflective material over that of the control experiment was reported as the improvement arising from reflection within the reactor. The maximum improvement in inactivation of E.coli was iii  around 100% obtained in a reactor made of Teflon followed by nearly 60% and 30% increase in aluminum and stainless steel, respectively. The results also showed that the E.coli inactivation kinetics followed the time-dose reciprocity rule meaning that the same level of inactivation was observed for both high and low fluence rates; whereas the low fluence rate resulted in more Fecal Coliform inactivation than the high fluence rate at the same UV fluence. These observations were consistent for all three LEDs.       iv  Table of contents  Abstract .......................................................................................................................................... ii Table of contents .......................................................................................................................... iv List of tables................................................................................................................................. vii List of figures .............................................................................................................................. viii List of abbreviations ......................................................................................................................x Acknowledgements ...................................................................................................................... xi Chapter 1: introduction .................................................................................................................1 Chapter 2: background .................................................................................................................4 2.1 UV light characteristics................................................................................................... 4 2.1.1 Ultraviolet light-emitting diodes (UV-LEDs) ............................................................. 6 2.2 Fundamental terms and concepts in photochemistry: ..................................................... 9 2.3 Literature review ........................................................................................................... 11 2.3.1 UV-LED water disinfection studies .......................................................................... 11 2.3.2 Time-dose reciprocity law evaluation ....................................................................... 17 2.3.3 Effect of reflection on UV fluence rate distribution ................................................. 21 2.3.4 Thesis objectives ....................................................................................................... 23 Chapter 3: Experiments ..............................................................................................................24 3.1 Experimental setup........................................................................................................ 24 3.1.1 UV-LED setup .......................................................................................................... 24 3.1.2 UV-LEDs .................................................................................................................. 27 3.1.3 Reflective petri dishes ............................................................................................... 27 v  3.1.4 Fluence rate measurement......................................................................................... 28 3.1.4.1 Chemical actinometry ....................................................................................... 29 3.2 Microbial test ................................................................................................................ 30 3.2.1 Test microorganisms ................................................................................................. 30 3.2.2 Preparation of inoculum ............................................................................................ 31 3.2.3 Growth curves ........................................................................................................... 31 3.2.4 Irradiation procedure ................................................................................................. 32 3.2.5 Enumeration of microorganisms’ colonies ............................................................... 32 Chapter 4: Results and discussion ..............................................................................................35 4.1 Spectrum analysis of UV-LEDs.................................................................................... 35 4.2 Fluence rate measurements ........................................................................................... 36 4.3 The UV-LEDs’ radiation outputs at their start-up conditions ...................................... 38 4.4 Growth curves ............................................................................................................... 41 4.5 Specular reflection measurement at 45° ....................................................................... 43 4.6 UV inactivation behavior of E.coli and Fecal Coliform ............................................... 44 4.7 Effect of reflection on E.coli inactivation ..................................................................... 46 4.7.1 Evaluation of the absorbed energy by E.coli sample during UV irradiation ............ 49 4.7.1.1 Calculation of the absorbed by the E.coli water sample during UV irradiation 50 4.7.1.2 Development of the E.coli inactivation kinetics based on the absorbed energy by the solution ................................................................................................................... 54 4.8 The effect of UV-LED’s fluence rate on E.coli inactivation ........................................ 58 4.9 The effect of UV-LED’s fluence rate on Fecal Coliform inactivation ......................... 59 Chapter 5: Conclusions and recommendations ........................................................................65 vi  5.1 Conclusions ................................................................................................................... 65 5.2 Recommendation for future work ................................................................................. 66 Bibliography .................................................................................................................................69 Appendix A: KI/KIO3 Actinometry (theory and experimental protocol) ..............................72 Appendix B: MATLAB code for calculation of irradiance, fluence rate, and reflection affected fluence rate on a surface of a square and a circle based on radiometer readings: .76 Appendix C: MATLAB code for calculation of the average reflection factors ......................78 Appendix D: UV-LEDs data sheet .............................................................................................79  vii  List of tables  Table   2.1 Comparison between UV-lamp and UV-LED ............................................................... 8 Table  4.1 The UV-LEDs’ characteristics at different  experimental trials ................................... 37 Table  4.2 The UV-LEDs’ fluence rates in reflective materials .................................................... 37 Table  4.3 Comparison between 45° specular reflectivity of different petri dishes ....................... 44 Table  4.4 Comparison between the absorbed energy by the sample at different wavelengths .... 52 Table  4.5 comparison between water factor and attenuation factor ............................................. 53 Table  4.6 The E.coli log inactivation results at different calculated energy per volume. ............ 54 Table  4.7 Comparison between the reflective materials’ improvement in log reduction of E.coli....................................................................................................................................................... 56  viii  List of figures  Figure  2.1 UV light in the electromagnetic spectrum (USEPA 2006) ........................................... 4 Figure  2.2 Applications of each band of UV light (set catalogue n.d.) .......................................... 5 Figure  2.3 Simplified band diagram of the 210nm LED, illustrating how an LED with AlN active region operates (adopted from Taniyasu et al. (2006)). .................................................................. 7 Figure  2.4 Reflection and refraction between a medium of refractive index n1 and another medium of refractive index n2 (Bolton 2000) .............................................................................. 11 Figure  3.1 Schematic diagram of the experimental setup taking into account the LED radiation pattern (Θ) (A) side view of the test apparatus (B) top view of the test apparatus ....................... 25 Figure  3.2 Experimental setup including all essential equipment: A) power supply; B) THORLABS single-axis translation stage; C) &D) THORLABS metric translating X/Y filter mount; E) HI200M stainless steel mini-stirrer; F) black box; G) heat sink; H) UVLED; and I) petri dish with magnet. .................................................................................................................. 26 Figure  3.3 Experimental step-up for measuring reflection ........................................................... 28 Figure  3.4 Setup for UV-LED irradiance measurement on the surface: (a) USB 2000+ UV/Vis spectroradiometer (b) X positioner (c) Y positioner (d) laboratory jack (e) heat sink (f) LED PCB (g) UV-LED (h) black box. ........................................................................................................... 29 Figure  3.5 Serial dilution .............................................................................................................. 33 Figure  3.6 Spread-plate method .................................................................................................... 33 Figure  4.1 Normalized spectral analysis for 265nm, 275nm, and 285nm UV-LEDs. .................. 35 ix  Figure  4.2 The UV-LEDs' performance within their start-up conditions: a) 265nm F.C=350mA b) 265nm F.C=35mA c) 275nm F.C =100mA d) 275nm F.C =10mA e) 285nm F.C =500mA f) 285nm F.C =50mA ....................................................................................................................... 41 Figure  4.3 The growth curves for the E.coli and the Fecal Coliform (cultured colonies (  )) ... 42 Figure  4.4 E.coli and Fecal Coliform UV susceptibility: a) 265nm b) 275nm c) 285nm. ........... 46 Figure  4.5 E.coli inactivation inside reflective materials: a) 265nm b) 275nm c) 285nm. .......... 49 Figure  4.6 Schematic diagram of experimental set-up for absorbed energy calculation .............. 50 Figure  4.7 Essential dimensions for calculating energy inside E.coli sample .............................. 51 Figure  4.8 The E.coli log inactivation versus modified fluence at different wavelengths ........... 55 Figure  4.9 The UV inactivation of E.coli under different UV-LEDs’ fluence rates. Data are expressed as the mean of logarithmic reduction in duplicate; standard error is indicated as bars at extreme points. .............................................................................................................................. 58 Figure  4.10 The UV inactivation of the Fecal Coliform under different UV-LEDs’ fluence rates. Data are expressed as the mean of logarithmic reduction in duplicate; standard error is indicated as bars at extreme points. .............................................................................................................. 60 Figure  4.11 The UV absorption of unexposed E.coli and Fecal Coliform sample at 265nm, 275nm, and 285nm........................................................................................................................ 61 Figure  4.12 The measured fluence rates in 9mm petri dish when organism sample was irradiated to: a) 0.4mW/cm2 at 265nm, 0.4mW/cm2 at 275nm, and 2mW/cm2 at 285nm b) 0.04mW/cm2 at 265nm, 0.04 mW/cm2 at 275nm, and 0.3mW/cm2 at 285nm. ...................................................... 62  x  List of abbreviations  ATCC American Type Culture Collection CFU Colony Forming Unit DNA Deoxyribonucleic Acid E. coli Escherichia coli LB agar Luria-Bertani Agar LED Light-emitting Diode MS2 Male-specific-2 OD Outer diameter PBS Phosphate Buffered Saline RNA Ribonucleic Acid US EPA United States Environment Protection Agency UV Ultraviolet                 xi  Acknowledgements  I would like to thank my supervisor Dr. Fariborz Taghipour in Chemical and Biological Engineering (CHBE) department for providing me with all necessary resources to complete my work as well as his support and guidance throughout this project. I would also like to bestow thanks upon Dr. Majid Mohseni, my second supervisor in the department of Chemical and Biological Engineering for his constant encouragement, advice, and for letting me work and learn from him during the past two years. He provided me with many helpful challenging questions to solve and always inspired me to do my best for everything that I do. This work was funded by a grant from Natural Science and Engineering Research Council of Canada (NSERC). I am also grateful to Ata Kheyrandish, my office mate, lab mate, and close friend who helped me to build my experimental set-up and taught me how to use UV-LEDs. I owe a great deal of thanks to Kai Song for his help getting started in the lab, acquiring materials, and mentoring the microbial aspects of this research. I also want to thank Majid Keshavarzfathi, my lab mate and close friend who provided me with essential information to improve different part of my research. In addition, of course, I offer my enduring gratitude to my parents, my brother, and my sister-in-law for their continued prayer and support throughout this project.  1  Chapter 1: introduction  Humans started to learn more about the microbial world and the necessity of water disinfection after the invention of the microscope by Hans and Zacharias Janssen in 1590. For the first time, instead of observing the microbial effects on the human race, researchers were able to observe bacteria. By developing an understanding of the microbial world, purification of water changed from simply increasing its aesthetic appeal towards disinfection (Morris 2012).  Modern disinfection systems include chlorine-based chemicals, ozone, and UV radiation disinfection. Chlorine gas as well as several chlorine containing compounds such as sodium hypochlorite, chlorine dioxide, and chloramine are the most common chemical disinfectants. Pure chlorine causes high levels of oxidative damage in microbes, which makes it an effective disinfectant; however, formation of disinfection by-products, such as trihalomethanes (THMs), is the main disadvantage of water chlorination. Indeed, the potential health risks associated with the consumption of disinfection by-products have compelled researchers to study alternative methods such as ozonation or UV radiation for the purpose of disinfection (V. M. Bakhir. 2003).  Ozone (O3) as a water disinfectant has been used for several decades. Given its relatively high oxidation potential (2.07V), ozone is a powerful oxidizing agent and also disinfectant. However, upon reacting with natural organic matter (NOM) in water, it leads to the formation of smaller and more biodegradable organic compounds which act as nutrient for microorganisms, and reduction in water biostability. Additionally, ozone requires high initial expenses for equipment as well as significant operational costs (Camel & Bermond 1998).  Ultraviolet (UV) irradiation inactivates microorganisms by disrupting their DNA structure and has many advantages, compared to conventional methods, which make it very 2  valuable for microbial disinfection. The main distinctive feature of UV is the capability of inactivation of Cryptosporidium and Giardia, two chlorine resistant pathogens. Additionally, the implementation of UV radiation does not produce any disinfection by-products, does not leave any residual taste/odor, and does not compromise public health by over- or under-dosing (Chatterley 2009). The conventional sources of UV are mercury-based lamps which have been around for over a century, with the first mercury based UV lamps used for drinking water disinfection in France in the early 1900’s.However, they have issues such as low lifetime (8000 to 10,000 hours) and big size which is why they are mostly placed at the center of the UV reactor (USEPA 2006).  Moreover, mercury is a hazardous chemical which, if the lamp breaks during installment or because of collision with foreign object, can expose the plant personnel and also compromise public health by entering the drinking water supply (USEPA 2006). Additionally, based on the United Nations Environment Program (UNEP) agreement, the mining and usage of mercury should be limited by the year 2020 (UNEP 2013).  Ultraviolet Light Emitting Diodes (UV-LEDs) are one emerging and promising alternative source of UV. UV-LEDs have many advantages compared to conventional UV lamps. They do not contain mercury, have a small size that makes them flexible for design purposes, do not need a warm up time, and their potential lifetime could be up to 100,000 hours. Moreover, they can be manufactured at specific wavelengths from 247nm to 365nm (Chatterley 2009). Currently, the main weaknesses of this new technology are its low efficiency (1%) and high cost; however, it is predicted that by 2020 the wall plug efficiency of UV-LEDs will be increased to 20% (Ibrahim et al. 2014) and it will become more economically feasible as the time passes (Chatterley 2009).  These features have attracted significant attention from the research 3  community as well as water treatment industry towards investigating different aspects of UV-LED applications. The study presented in this thesis focused on two important factors that must be considered in designing and finally building a UV-LED water disinfection reactor. First, this work focused on investigating the effect of reflection of the body of the reactor on microorganisms’ inactivation. To achieve this, Teflon, stainless Steel, and aluminum were chosen as most common reflective materials and E.coli ATCC 11229 was used as the most suitable model microorganism. Second, it aimed to evaluate the effect of fluence rate on microorganism inactivation. Two levels of fluence rates (with an order of magnitude difference) were applied on E.coli and Fecal Coliform at different wavelengths. In both parts, three different UV-LEDs of 265nm, 275nm, and 285nm in terms of their wavelengths and specifications were used. A batch reactor was constructed and used at different operating conditions based on each LED’s features. Actinometry as well as radiometry measurements were utilized for the calculation of fluence rate.  The following four chapters will provide more detailed information about what was done in this project. Chapter 2 will include the background information and a detailed review of literature. Chapter 3 will provide the experimental procedures applied in this project. Chapter 4 will have the results as well as the discussions about them. The last chapter, chapter 5 will include the overall conclusion and the recommendations for future work.  4  Chapter 2: background  2.1  UV light characteristics UV wavelengths lie between x-rays and visible light, ranging between 100nm and 400nm including major bands labeled as vacuum UV (100 to 200nm), UV-C (200 to 280nm), UV-B (280 to 315nm), and UV-A (315 to 400nm). (Fig 2.1)  Figure ‎2.1 UV light in the electromagnetic spectrum (USEPA 2006) UV-A or long wave UV is the specified range of wavelengths from 315nm to 400nm which is sometimes referred to as black light (USEPA 2006). It has a number of applications including light therapy in medicine, UV curing of printer inks and polymers, bug zappers as flies and other insects can be attracted to light emitting around 365nm, and medical imaging of cells. UV-A composes 95% of all sun’s UV light reaching the Earth’s surface. UV-B or medium wave UV is the specified range of wavelengths from 280nm to 315nm (USEPA 2006). DNA sequencing, protein analysis, medical imaging of cells and drug discovery are some UV-B applications. Solar UV-B radiation composes 5% of UV radiation at the surface of the earth and 5  is considered the main cause of DNA damage leading to skin cancer. UV-C or short wave UV is the name of the range of wavelengths from 200nm to 280nm (USEPA 2006). UV-C range has a number of applications from forensic analysis to optical sensors. Additionally, since the maximum absorption of microorganisms happens at wavelengths of 260nm to 270nm, UV-C range has been considered the most efficient range for water disinfection. Vacuum UV or V-UV is the fourth category of UV light, emerging from the UV-C band, contains wavelengths between 100nm and 200nm (USEPA 2006). It is called V-UV since at these wavelengths, emissions are absorbed in the majority of media like air or water, and vacuum is the only environment where they will not be absorbed immediately (Fig 2.2). Although the categories and specifications of UV light are subject to interpretation, these four groups are the most widely accepted.   Figure ‎2.2 Applications of each band of UV light (set catalogue n.d.)    6  2.1.1 Ultraviolet light-emitting diodes (UV-LEDs) H. J. Round of Marconi Labs first discovered that some inorganic materials could emit light if an electric voltage is impressed on them. This was considered the initial spark in the development of LEDs (wikipedia). Publications related to the development of UV-LEDs began to increase in the late 1980’s (Morris 2012). In 1988, Mishima made one of the first UV-LEDs. It was built from a cubic BN pn junction and had an emission spectrum focused around 340nm to 360nm and covered spectrum from 215nm to the visible red  (Mishima et al., 1988). Through the 1990’s, the experimentations resulted in mixing multiple semiconductors into ternary and quaternary compounds which improved the quality of the LEDs and also increased the ability to fine tune the devices. AlGaN and AlInGaN are the most typical ternary and quaternary compounds used in UV-LEDs which are normally deposited on an undoped AlN sample in order to produce wavelengths changing as a function of fine-tuning between the bandgap of AlN and the ternary or quaternary superlattices surrounding it. The theoretical minimum bandgap for each AlN based LED is just below 6.1 eV or just above 205nm since the band gap of the AlN component is 6.026eV (Guo & Yoshida, 1994) to 6.2eV (Levinshtein et al., 2001) at 300 K. The bandgap for other components such as GaN is 3.2eV to 3.39eV at 300K (Levinshtein et al. 2001) and 1.9eV to 2.05eV for InN at 300K (Guo & Yoshida, 1994; Levinshtein et al., 2001).  In 2006, a 210nm AlN LED was produced with AlN/AlGaN superlattices (Taniyasu et al., 2006). Fig 2.3 illustrates the simplified diagram adopted from the structure of this AlN/AlGaN UV-LED which was originally published by Taniyasu et al (2006). The main role of regions marked as p-type and n-type AlN/AlGaN superlattice is to decrease operating voltage (Taniyasu et al., 2006). As Fig 2.3 shows, when a voltage is applied on the whole semiconductor, the electrons start passing through the active region. when a hole with the same level of energy is 7  matched to that electron, recombination of electron with the hole takes place and during this process, a photon with the same energy as the recombination energy will be emitted.    Figure ‎2.3 Simplified band diagram of the 210nm LED, illustrating how an LED with AlN active region operates (adopted from Taniyasu et al. (2006)). In more modern cases, multi-quantum wells are implemented in the active layer of the LED to encourage recombination at a specific wavelength (Shur & Gaska 2010). When electrons are trapped at an energy state within the quantum well, they will be forced to recombine from the conduction band to the hole in the valence band and emit at specified wavelength of the band gap of quantum well. The use of these quantum wells not only increases the chance of right recombination but also improves the efficacy of the device (Shatalov et al., 2002). Table 2.1 includes the comparison between conventional sources of UV radiation, which are low pressure or medium pressure lamps, and the new UV technology, which is the UV-LED. Some of the data shown in this table such as UV-LED’s output power is based on the availability of the source in the market and it can be changed in future. 8  Table  ‎2.1 Comparison between UV-lamp and UV-LED Source Specification UV lamp UV-LED Wavelength Low-pressure mercury lamps have a peak wavelength at 253.7nm while medium pressure mercury lamps are polychromatic with peaks at different wavelengths.  247-365nm (Shur & Gaska 2010) Lifetime (Chatterley 2009) 8000-10000 h 100000 h (for visible LED)* Size (Chatterley 2009) ~30-80cm (Arc length) ~ 5-9 mm Energy Consumption Very high Low Output power 5.7-14.3 W (at 254nm) 5-13 mW (at UV-C range) Environment Disposal of mercury inside UV-lamps has environmental issues. Although UV-LEDs do not have mercury in their structure, they have the problems associated with the disposal of semiconductor’ materials.  * The current lifetime for UVC-LED is currently less than 100000.     9  2.2 Fundamental terms and concepts in photochemistry:  The followings are the definitions of the common terms used in this study and in the field of UV disinfection in general (Braslavsky et al., 1996; Bolton, J.R., 2010; Bolton, 2000). They are provided here to help readers better understand the discussions presented in this thesis. Radiant Energy [J]:  The total radiant energy emitted by a source over a given period. Radiant Power [W]:  The total power emitted in all directions from a source. Irradiance [mW/cm2]:  The total power from all upward directions incident on an infinitesimal surface element with area dS containing the point under consideration divided by dS. Fluence Rate [mW/cm2]:   The total radiant power incident from all directions onto an infinitesimally small sphere with cross-sectional area dA, divided by dA. Fluence or UV Dose [mJ/cm2]:  The total radiant energy from all directions passing through an infinitesimally small sphere of cross-sectional area dA, divided by dA, it equals to the multiplication of Fluence Rate (mW/cm2) and exposure time (second). Refraction and Snell’s Law: Snell’s Law governs the refraction properties of radiant energy transmitted through a surface (Fig 2.7) 𝑛1 sin 𝜃1 = 𝑛2 sin 𝜃2 10  where 𝑛1 and 𝑛2 are the refractive indices of the two media, 𝜃1and 𝜃2are the angle of incidence and the angle of refraction, respectively. Note that if 𝑛2>𝑛1 the angle of refraction (𝜃2) is less than the angle of incidence (𝜃1) Reflection and the Fresnel Law: Whenever radiant energy passes through an interface between two media of different refractive indices, a certain fraction of the radiant energy is reflected at the angle 𝜃𝑟 = 𝜃1; the rest passes through the interface into the second medium and undergoes refraction. The optics involving the description of this process is complicated because the amount reflected depends on the polarization of radiant energy. If 𝑟⊥ is the amplitude of the energy perpendicular to the plane of incidence and 𝑟ǁ is the amplitude of radiant energy parallel to the plane of incidence, then the Fresnel Law (Meyer-Arendt 1995) define these two amplitudes as  𝑟⊥ =𝑛1 cos 𝜃1 − 𝑛2 cos 𝜃2𝑛1 cos 𝜃1 + 𝑛2 cos 𝜃2 𝑟ǁ =𝑛2 cos 𝜃1 − 𝑛1 cos 𝜃2𝑛1 cos 𝜃2 + 𝑛2 cos 𝜃1   11   Figure ‎2.4 Reflection and refraction between a medium of refractive index 𝒏𝟏 and another medium of refractive index 𝒏𝟐 (Bolton 2000) The Reflectance for unpolarized radiant energy is given by 𝑅 =12[𝑟⊥2 + 𝑟ǁ2] The transmittance T refracted into the second medium is given by 𝑇 = 1 − 𝑅 Thus, the incident beam is split into two parts (Fig 2.5): a reflected part of fraction R and a refracted and transmitted part of fraction T.     2.3  Literature review 2.3.1 UV-LED water disinfection studies  This section focuses on a detailed analysis of the research conducted to-date in the field of the UV-LED disinfection. The comments are mainly related to the weaknesses in the methods that have been used in the past and how these methods can be modified. The knowledge gaps 𝜃1 𝜃2 𝜃𝑟 12  leading to the objectives of this project will be explained in the last two part of the literature review. Vilhunen et al. (2009) studied the effectiveness of either 269nm or 276nm wavelengths on inactivation of E.coli K12. The irradiation experiments were done for each individual wavelength from zero to 20 min exposure times, 5-minute intervals, in three different test media: Ultrapure water; water and nutrients; and water, nutrients and humic acid. For the same experimental condition, the radiant flux of the 269nm UV-LEDs (3.4mW) was almost half of the output power of 276nm UV-LEDs (6.1mW), but the 276nm inactivation was slightly higher than that of 269nm.The better performance of 269nm LEDs was attributed to the higher DNA absorption of bacteria at that wavelength. Additionally, decreasing the UV transmittance of the sample decreased the amount of inactivation in the batch system especially under long exposure times (15-20 min).The authors measured radiant flux of LEDs by using an integrated sphere and reported those values for the amount of energy received by the microorganisms inside their proposed reactor. The integrated sphere measurement is helpful for measuring the whole amount of LEDs’ output energy but since the exposed surface area of sample receives a portion of this energy, the results of the integrating sphere will overestimate the actual amount delivered to the sample. Using chemical actinometry or spectroradiometer measurement will lead to a better estimation of fluence rate within the containing microorganisms’ sample. For comparing the performance of different wavelengths (269nm and 279nm), Vilhunen et al. (2009) plotted the graph showing the number of microorganisms versus the exposure time. The authors concluded that the more efficient wavelength between 269nm and 279nm is the one that resulted in the fewer number of microorganisms under the same exposure time. This method is not a good way for comparing the effectiveness of different UV-LEDs because at different exposure times, both 13  fluence rate and wavelength will affect the inactivation. Fluence rate is important because it shows how much energy is delivered to the microorganism, while the importance of wavelength is because of the different microorganism’s DNA absorption at different wavelengths. A better way of comparing the performance of different LEDs based on their wavelengths is using the fluence-based graphs, where calculated fluence will be on x-axis instead of time. This method can compare the impact of different LEDs on inactivation at the same level of energy and only shows the effect of their wavelengths on inactivation.  Chatterley & Linden (2010) constructed a UV-LED collimated beam apparatus using three 265nm LEDs as part of their study. They compared the results of E.coli K12 inactivation in the UV-LED collimated beam apparatus with those of a conventional mercury lamp. It was found that at doses less than 20 mJ/cm2, there was no significant difference between the UV-LED and the mercury lamp collimated beam results. The authors used the 254nm collimated beam protocol (Bolton et al., 2003) for the UV fluence calculation in their proposed UV-LED collimated beam apparatus. There are two factors in the 254nm collimated beam protocol, water factor and reflection factor, which must be recalculate for the wavelengths other than 254nm if the protocol is being used for other wavelengths. The authors calculated water factor for the 265nm UV-LED, but they used the same reflection factor as 254nm UV lamp for the 265nm UV-LED. Additionally, a high level of the petri factor (more than 0.9) is the main criterion for showing collimation in the 254nm UV collimated beam apparatus; however, there is no guarantee that the same collimation exists in the UV-LED collimated beam apparatus. It is because of the UV-LED’s radiation pattern, a unique feature of the LED technology compared to the UV lamps. The radiation pattern of a UV-LED shows the pattern of its intensity at different angles. Because of this characteristic, it is possible to have high petri factor meaning that the 14  same irradiances over the surface of the sample compared to the central irradiance but the light is not still collimated. So, in sum, the collimated protocol and concept, which is commonly used for the UV lamps, cannot be implemented for the UV-LEDs. Bowker et al. (2010) compared the UV-dose response of non-pathogenic microorganisms (two coliphages, MS2 and T7, and one bacterium, E.coli ATCC 11229) in their proposed UV-LED collimated beam apparatus for 255nm and 275nm UV-LEDs with the inactivation kinetics resulted from conventional collimated beam system with 254 nm mercury lamp. They designed a collimated beam apparatus with UV-LEDs by developing a CFD model in COMSOL software. In this model, the parameters were LED’s configuration on a heatsink, objective wavelengths (255nm or 275nm; operated separately) and the distance between the sample surface and the light source. These parameters were optimized in order to have the maximum uniformity of irradiance over the sample surface and the minimum production cost.  After constructing the UV-LED’s collimated beam based on the results of the optimization, validation of numerical results was performed by determining the Petri factor (Bolton et al., 2003), and irradiance on the surface of the sample through experiments. The same trend of E.coli inactivation (log linearity at high fluences and shoulder effect at low fluences) was observed for all three UV sources but the effectivity of the 254nm lamp in terms of microbial inactivation was significantly higher than that of the 275nm LEDs and the 255nm LEDs. The higher inactivation for the 275nm LEDs compared to the 255nm LEDs was attributed to the germicidal wavelength of bacteria, which is close to 270nm. Although greater MS2 log inactivation was found with 254nm lamp in comparison to both LED wavelengths, the difference was not significant. Coliphage T7 showed slightly higher log inactivation with the 275nm LEDs in comparison to the 255nm LEDs, which was justified with the maximum DNA absorbance of T7 at 270nm. In their model, Bowker et al. 15  (2010) did not consider transmittance and absorbance of the sample; however, they applied the viewing angle of the UV-LEDs. Finally, they proposed the following formula for calculating the incident irradiance on the sample surface.  𝐼 =𝑃2𝜋𝑟2 (1 − cos(𝛼)) where I is irradiance; P is radiant power from the light source, r is distance and α is the viewing angle. Substituting the viewing angle of 60 in this formula leads to an incident irradiance, which is twice higher than the total amount of incident irradiance on a UV-LED’s surrounding hemisphere area, making this formula not reliable for accurate calculation. The reported initial concentration of E.coli was around 108cfu/ml in their study. This range of concentration has a high UV absorbance and it affects the result of microorganisms’ tests by reducing the amount of inactivation significantly. Their 255nm UV-LED results showed significantly less inactivation compared to the low-pressure mercury lamp. They concluded that the main reason for this discrepancy is the low irradiance rate (around 0.05mW/cm2) of their 255nm UV-LED which caused lower inactivation. The authors supported their idea by comparing their results with those of Sommer et al. (1998). A more plausible explanation, however, for the discrepancies in the Bowker et al. (2010) results can be overestimation of fluence based on their proposed mathematical model. Chevremont et al. (2012) examined the impact of four different factors: bacterial density, pH, exposure time, and wavelength (254nm, 280nm, 365nm, 405nm) towards inactivation of three strains of E.coli and two strains of E. faecalis. Either one UV-LED or a coupled configuration of one UV-C and one UV-A LED was mounted on an apparatus in a batch reactor 16  to expose a microbial sample for each individual experiment. The pH of the medium only affected the reduction of E. faecalis. About bacterial density, their results showed that it influenced the microbial reduction of E.coli CIP 6224 and E. faecalis ATCC 19433. However, the author considered bacterial density as well as pH of the medium as insignificant factors in their further analysis. Changing the wavelength influenced the inactivation of all strains, and exposure time only affected the inactivation of E.coli strains. Additionally, the study found that coupled wavelengths such as 280/365nm, 280/405nm and 254/365nm, had a maximum effect on inactivation of some strains of E.coli like ATCC 11303, while for the same experimental condition, the maximum inactivation for one of the E. faecalis strains (ATCC 33186) was achieved with a single wavelengths of 280 nm. At the end, the author concluded that coupled configurations were better than single wavelengths. They supported their claim by hypothesizing a cumulative effect of the power emitted by coupled LEDs instead of the effect linked to the selected wavelength. The design criteria for their experimental set-up were not declared clearly, which makes generalization of their results very hard. Although using exactly the same UV-LEDs with the same set-up configuration can lead to the same results, if the same wavelengths are provided with different UV-LEDs, there will be no guarantee that the same results will happen. This is because of the UV-LED’s radiation pattern; If two UV-LEDs have the same output power and the same viewing angles but their intensities follow different radiation patterns, the amount of energy received by a certain surface will be different. In order to solve this problem, there are reliable methods in terms of providing detailed information about the UV-LEDs. The first one might be reporting the output power of the UV-LED as well as its radiation pattern. The other might be reporting the average fluence rate of the UV-LED by implementing chemical actinometry or measuring the UV-LED’s irradiance with a radiometer. 17  Oguma et al. (2013) studied the application of UV-LEDs in both batch and flow-through water disinfection reactors. They used 265nm, 280nm and 310nm wavelengths separately or in combination to evaluate their disinfection efficiencies in term of E.coli K12 inactivation. In both batch and flow-through reactors, the authors analyzed the results based on time and fluence inactivation efficiencies. They found that 280nm and 265nm UV-LEDs had the highest time-based and fluence-based inactivation efficiencies, respectively. In their microbial tests, they used E.coli at its stationary phase, which can cause error in the results because at this phase, the rate of microorganism's growth and death will be the same. Therefore, UV radiation will not be the only factor for inactivation of bacteria. The right phase for microbial tests is the growth phase since all of the observed inactivation can be attributed to the disinfectant. The reported fluences in their study were calculated by modifying the results of the 254 nm UV lamp collimated beam method to other wavelengths. Some of the parameters in this calculation, like reflection factor, must have been changed for different wavelengths; however, they considered these factors the same as those of the 254 nm UV lamps in their fluence calculation, which can cause error in their reported fluences.  2.3.2 Time-dose reciprocity law evaluation The above literature review focused on analyzing major studies which have been done on LED biodosimetry experiments and inconsistencies mainly related to their methods of  measuring fluence distribution, microorganism preparation and experimental set-up. A few researchers focused on studying the effect of UV fluence rate on inactivation of different microorganisms. Knowledge of the microorganisms’ reaction to different levels of UV fluence rates is essential in designing a UV disinfection flow through reactor especially with UV-LEDs. It is because of the low output power of the UV-LEDs compared to the conventional UV lamps, 18  which makes their fluence rates decrease significantly with distance from the radiation source. The following review will shed light on the available literature that focused on validating the time-dose reciprocity rule in 254nm UV lamps. Harm (1968) examined the effect of UV dose fractionation on the survival rate of E.coli. In his study, four E.coli strains: E.coli B/r; E.coli B; E.coli Bs-1; and E.coli Csyn- as well as two types of phages: T1 wildtype, and T4 wildtype (Doermann) were used as tested organisms. The experimental set-up consisted of a 254 nm low pressure mercury lamp, placed in a metal housing with a shutter for making short exposures. The effective UV dose rate of 7.2ergs/mm2.sec (~7.2×10-4mW/cm2) was applied to a 2.5mL sample on a watch glass, covered with a quartz plate. At the same given UV doses, the tested microorganisms were irradiated either in a single or fractionated exposure condition. The bacterial cells’ results showed a higher survival rate in separated exposures compared to that of continuous irradiation. However, no difference was seen in phages’ reduction under both conditions. It was hypothesized that the deviation from the time dose reciprocity law, seen only in bacterial cells’ inactivation, was due to repair machineries taking place during the dark periods of experiment. In addition, the author hypothesized that the rate of cell repair mechanism depends on the level of sub lethal factors during the experiment, meaning that as the sub lethal factors increase, the repair efficiency would decrease. This idea was used by the author as an explanation for the shoulder effect in bacteria kinetic curves at low UV doses.   Sommer et al. (1996) studied the effect of protraction of UV irradiation on different microorganisms in a laboratory UV irradiation devise. The reactor consisted of 10 low-pressure mercury lamps (253.7nm) mounted horizontally over the irradiation vessel. The lamps were regulated in order to emit different UV intensities and the exposure time was set by using a time-19  controlled diaphragm (100 mm by 25 mm) attached right under the UV lamps. The test organisms were a vegetative bacterial strain (E.coli ATCC 25922), a bacterial virus (Staphylococcus aureus phage A994) and bacterial spores (Bacillus subtilis ATCC 6633) as well as three haploid laboratory strains (YNN281, YNN282, and RC43a), two diploid strains (laboratory strain YNN281*YNN282 and commercial bakery yeast) of yeast (Saccharomyces cerevisiae), and spores of the latter diploid strain. For a given level of UV dose, each microorganism was exposed to three levels of UV intensities (0.02, 0.2, and 2W/m2). Newman-Keuls method and analyses of variance were applied to analyze the significance of difference of reduction rates by changing the intensities within certain selected UV doses. The study found that the bacterial spore (B. Subtilis) was the most UV-resistant test organism, followed by the bacteriophage (S. aureus phage), whereas the UV-susceptibility of vegetative bacteria (E.coli) was distinctly higher. The susceptibilities of S. cerevisiae diploid wild-type strains as well as haploid strains (YNN281, YNN282, and RC43a) and spores fell between those of E.coli and the bacteriophage. The status of UV susceptibility of the yeast spores was considered as a confirmation of their inactive state during the irradiation experiments. In addition, it was found that E.coli, the bacteriophage, the bacterial spores and the yeast spores followed Bunsen-Roscoe reciprocity law in their inactivation by UV light within the range of 0.02, 0.2, and 2 W/m2 of UV intensities. Conversely, all yeast strains showed higher inactivation at lower fluence rates compared to that of higher fluence rates at a given UV dose. The authors hypothesized that the inactive state of viruses and bacterial spores during UV exposure, which makes them unable to repair themselves through repair enzyme mechanisms, was the main reason that they follow the Bunsen-Roscoe law. The deviation of all vegetative yeast cells’ inactivation behavior from the time-dose reciprocity law was attributed mainly to different aspects of repair mechanisms. It was 20  because of the inactivation behavior of yeast spores (an inert state of yeast cells) which followed the reciprocity law. Since the spores are inactive during UV exposure as opposed to the active status of the cells, and repair mechanisms require active organisms, finding the reason through repair mechanisms were more reasonable. It was hypothesized that UV protraction influenced the efficiency of repair mechanisms and also it increased the possibility of lethal to sub lethal damage. Another hypothesis was the sensitivity of regulators of the repair mechanisms, which were only increased when low fluence rates were applied. At the end, the authors suggested further investigation of repair mechanisms for clarification of yeast cell behavior. In another study, Sommer et al. (1998) studied the UV (253.7nm) inactivation behavior of water related organisms and also the validity of the time dose reciprocity rule for kinetics of those organisms.  The experimental set-up was the same as the one used in their previous research in 1996. They selected three E.coli strains (ATCC 25922, ATCC 11229 and isolate from sewage), three bacterial viruses (MS2, ΦX174 and B40-8) and spores of Bacillus subtilis as test organisms. The irradiation tests were done at three levels of UV fluence rates (2, 0.2 and 0.02 W/m2) under the same UV doses. It was found that E.coli strains and phage ΦX174 are the most UV susceptible organisms, followed by B40-8, MS2 and finally bacterial spores. The study found that the rate of E.coli strains inactivation was around one log higher in the high dose rates compared to that of the low dose rates at only doses of 80-100J/m2. The authors hypothesized that it might be because of the repair enzymes of the cell, which were more influenced by high UV intensities. On the contrary, the kinetics of eukaryotic yeast cells’ inactivation was more susceptible to exposure time instead of UV intensity as it was reported by Sommer et al. (1996). Other organisms’ inactivation kinetics did not deviate from Bunsen-Roscoe law, which was 21  accredited to the inactive state of bacterial spores and phages during UV irradiation. In both cases, the results were reported without any error bars and were not analyzed statistically. 2.3.3 Effect of reflection on UV fluence rate distribution Evaluating the effect of UV reflection from the body of the reactor is important for modeling the fluence rate distribution more accurately in order to increase the efficiency of the UV disinfection reactor. Although, in the past, researchers developed and applied various numerical models to determine the distribution of UV fluence rate inside UV reactors (e.g., Jacob & Dranoff 1970; Blatchley 1997; Irazoqui et al. 1973; Bolton 2000), no study in the open literature is reported on testing and determining the impact of reflection on microorganism inactivation. Since determination of the UV-LED’s fluence rate distribution is hard because of the complicated reflection behavior of different wall-materials as well as other important factors such as reflection and refraction of sample at different angles, providing an experimental method for measuring the improved fluence rate as a result of mentioned factors is desirable.  Li et al. (2012) studied the effect of inner-wall reflection on fluence rate distribution by performing measurements with a micro-fluorescent silica detector (MFSD). They built a cylindrical quartz UV reactor. Black cloth, stainless steel tube, and aluminum foil were wrapped individually around the outside of the reactor in order to provide different reflection within the operating environment. The source of UV radiation was a 254nm mercury lamp placed at the center of the reactor and surrounded by a quartz sleeve. A micro-fluorescent detector was mounted inside the reactor, which could be moved in perpendicular and parallel directions to the lamp axis. Another MSFD was inserted within the gap between the quartz sleeve and the lamp to monitor its fluctuation. Both MSFD readings were displayed separately on a multimeter outside the reactor. The UV transmittance of the solution was controlled by implementing different 22  concentrations of potassium acid phthalate. For each reflective material, they used a spectrophotometer for measuring the spectral reflectance coefficient within the range of 210nm to 280nm and a vacuum UV analytical spectrophotometer for measuring the diffuse reflection property. Additionally, the results of this study were compared to numerical data calculated with UVCalc model, developed by Bolton (2000). The study found that the reflection coefficient of aluminum, stainless steel and black cloth at 254nm, all covered by quartz sheet, were around 80.5%, 26.1% and 11.1%, respectively. The amount of black surface reflection was accredited mainly to the quartz UV reflection. (Bolton, J.R. 2010 found that quartz can be 8% reflective when it is exposed to normal UV beams and this amount gets higher at larger angles). The authors reported that the high reflective material (quartz/aluminum foil) remarkably increased the average fluence rates and this effect was seen more significantly when a solution with high UVT was used as the experimental medium. Moreover, the comparison of experimental results and numerical data showed that the error of neglecting the effect of inner-wall reflectance was 35% for the stainless steel reactor. In terms of diffuse reflection property of the inner wall, it was concluded that higher diffuse reflectivity led to more uniformity of fluence rate distribution inside the reactor. In their research, the fluence rate measurement error associated with folded aluminum foil was mentioned. This error can be solved by using a rough aluminum surface. In addition, the authors suggested the similar study with the rough reflective surfaces to be done in order to clarify the significance of diffuse reflection in FR enhancement.  23  2.3.4 Thesis objectives  Based on the discussions provided in the last two sections of the literature review, the following topics were found as important knowledge gaps needed to be studied for the application of the UV-LEDs in water disinfection area: 1. The effect of the UV-LED’s fluence rate on inactivation of microorganisms 2. The effect of reflection from the UV-LED reactor wall on the fluence rate enhancement inside the reactor as well as finding a new method for reporting this improvement As it was discussed earlier, the methods that were previously used for UV-LED fluence calculation, as well as reporting the inactivation data are not reliable. In this study, it was attempted to use the most accurate methods for UV fluence calculation and consider important details in reporting results in order to prevent possible errors. Since, unlike conventional UV lamps, there is no well-defined collimated beam protocol for measuring UV-LED fluence, iodide/iodate actinometry and spectroradiometer measurement were used for calculating the fluence rate during irradiation. For the first objective of this research, 265nm, 275nm, and 285nm UV-LEDs were selected and E.coli and Fecal Coliform were used as test microorganisms. For the second objective, aluminum, stainless steel, and Teflon were chosen as the reflective materials and E.coli was the only tested microorganism. The details of experimental procedure as well as the results of this study are provided in the following chapters.  24  Chapter 3: Experiments  3.1 Experimental setup  3.1.1 UV-LED setup In developing the experimental setup with a UV-LED, safety is an important factor which must be considered. The apparatus had to be a closed system, which could protect others from UV radiation. The status of the water sample during the disinfection process could be flowing and static. Given the objectives of this project, a static well-mixed batch system was chosen. A magnetic stir bar was incorporated with the system in order to ensure that the sample was mixed uniformly during irradiation. Because of the low output power of the implemented UV-LEDs (Appendix D), only 10mL of the sample, containing test organisms, was placed in a petri dish to be disinfected for each experimental run. An additional consideration was eliminating incident emissions of ambient light in order to inhibit the possibility of photo-reactivation. Before starting the experiments, the ambient light in the lab was measured and there was no trace of light between 320nm and 370nm wavelengths. Hence, the assumption of no photo-reactivation is valid since the whole setup was contained in a black box.    25   Figure ‎3.1 Schematic‎diagram‎of‎the‎experimental‎setup‎taking‎into‎account‎the‎LED‎radiation‎pattern‎(Θ)‎(A) side view of the test apparatus (B) top view of the test apparatus  Fig 3.1 shows the experimental setup and the testing apparatus. Although UV-LEDs are considered to be point sources, they follow certain radiation patterns shown in the LEDs’ data sheets in Appendix D. Therefore, the intensity of UV radiation at the center of the petri dish was not expected to be the same as that near the edge. An important aspect of taking into account the UV-LED radiation pattern was calculating the distance between the LED and the surface of the sample in the petri dish, in order to make consistent experimental conditions. Based on different UV-LEDs’ radiation patterns, the distance was calculated and adjusted in a way that the UV-LED viewing angle energy covered the whole surface of the water sample. In addition, since a collimated beam apparatus was not used in this project, this method could be helpful to ensure that any other researcher in the field can repeat the experimental procedure of this project. Spectral analysis was performed using an Ocean Optics USB 2000+ UV/Vis spectrometer. A TTi EX355R Bench Power Supply generated the required operating power for UV LEDs. In order to make a vortex in the sample, HI200M Stainless Steel Cover Magnetic Mini-Stirrer was used. To fix the LED precisely downward on the center of the sample and to Stirrer Θ A 12cm 12cm Petri dish 5cm Stirrer B UV-LED Microorganism solution Petri dish 26  adjust the calculated distance between LED and the surface of the sample, THORLABS Metric Translating X/Y Filter Mount and THORLABS Single-Axis Translation Stage with standard Micrometer were used, respectively. Fig 3.2 shows the final experimental setup which includes all the equipment applied for the purpose of this project.  Figure ‎3.2 Experimental setup including all essential equipment: A) power supply; B) THORLABS single-axis translation stage; C) &D) THORLABS metric translating X/Y filter mount; E) HI200M stainless steel mini-stirrer; F) black box; G) heat sink; H) UVLED; and I) petri dish with magnet.  B C A E D F H G I 27  3.1.2 UV-LEDs The UV-LEDs were purchased from LG and Nikkiso companies (Appendix D). Given the goal of analyzing the impact of different fluence rates and different material reflections on inactivation of microorganisms, the decision was made to investigate germicidal wavelengths of UV spectrum and study the efficacy of each wavelength in the project. The three wavelengths used were the 265nm (Nikkiso), 275nm (LG), and 285nm (Nikkiso). 3.1.3 Reflective petri dishes To assess the influence of surface reflection on fluence enhancement inside the UV-LED water disinfection reactor, aluminum, Teflon (odm-98), and stainless steel were selected as reflective materials. In order to have comparable results with quartz petri dish, similar petri dishes (1cm height, 3mm thickness, and 5cm inner diameter) were built. The inner surface of each petri dish was polished to make the diffuse portion of its reflection as uniform as possible. For measuring the reflection coefficients of the chosen materials at 265nm, 275nm, and 285nm, an experimental set-up was designed and built. In this set-up, the light source was tuned to emit the UV light at wavelengths of interest and the reflected rays from the surface were collected with a UV Vis 2000+ Spectroradiometer (Fig 3.3). The absolute value of UV irradiance from the source was measured by placing the Spectroradiometer directly in front of the light source. Finally, the ratio of the irradiance of reflected beams to the irradiance of emitted beams from the source was considered as the reflectivity for each material.     28   Figure ‎3.3 Experimental step-up for measuring reflection  3.1.4 Fluence rate measurement The fluence rates of each UV-LED on the surface of the sample were determined by USB 2000+ UV/Vis spectroradiometer and confirmed by chemical actinometry. The details of each method are discussed below: 3.1.4.1 Spectroradimeter measurement For each measurement, the UV-LED of interest was mounted downward on a height-adjustable heat sink attached to the top of the setup (Fig. 3.4), while on the bottom, the upward detector attached to a XY positioner was capable of measuring the fluence rate distribution on a given area. After fixing the distance between the UV-LED and the detector, the irradiance, the normal element of the received UV light by a detector, was measured in 0.5 cm increments on a 5cm×5cm square (equal to the diameter of the petri dish). The raw data was modified using a MATLAB code (Appendix B) to calculate fluence rate on a circular surface. The effects of refraction and reflection (Snell’s Law and Fresnel law) from the surface of the sample were implemented in the code.    Reflective Material  Spectroradiometer  Light source 45° Emitted beam Reflected beam  Probe holder 29   Figure ‎3.4 Setup for UV-LED irradiance measurement on the surface: (a) USB 2000+ UV/Vis spectroradiometer (b) X positioner (c) Y positioner (d) laboratory jack (e) heat sink (f) LED PCB (g) UV-LED (h) black box.  3.1.4.1 Chemical actinometry Chemical actinometry is one the most common methods for UV fluence measurement. This method is based on the photochemical principles. Light sensitive chemicals such as KI/KIO3, uridine, and ferroxalate solutions are used to absorb energy from the UV source. For the purpose of this research, KI/KIO3 actinometer was used for measuring the fluence rates at different wavelengths. There are two advantages of using this actinometer. First, it is optically opaque to the light with a wavelength shorter than 290 nm. So, all the UV light within the germicidal range could be absorbed. Second, the solution is optically blind to the light with a wavelength longer than 330 nm meaning that it will not absorb the UV-A and visible background light.  30  Based on the method described by Rahn et al. (2003), the KI/KIO3 solution was prepared. The solution consisted of 0.6M KI, 0.1M KIO3, and 0.01M Na2B4O7 in 100mL water. At 300nm and 352nm, the absorbance of the solution was recorded. After exposing the solution to a specific UV-LED, the absorbance of the sample was measured again at 352nm. The exposure times were chosen such that the absorbance of irradiated samples at 352nm were less than 1 because at higher values than 1, the data becomes unreliable (Chatterley 2009). The theory behind this method as well as the details of calculation and preparation of the solution are provided in Appendix A. For UV-LEDs used in this study (265nm, 275nm, and 285nm), the quantum yields of 0.57, 0.43, and 0.29 were calculated based on linear relationship between wavelength and quantum yield in the range of 254nm (Φ=0.73) to 284nm (Φ=0.3) (Rahn et al. 2003). The average reflection factors (α) of 0.0349, 0.0318, and 0.0339 were calculated by utilizing both Snell’s and Fresnel’s equations for 265nm, 275nm, and 285nm LEDs, respectively. For this calculation, the sample refraction factors of: 1.37 at 265nm; 1.366 at 275nm; and 1.363 at 285nm were used and the average amount was found based on incident angles changing from 0 to half of each LEDs’ viewing angle with one degree increments.(the MATLAB code is provided in Appendix C) 3.2 Microbial test 3.2.1 Test microorganisms The approval of the United States Environmental Protection Agency (USEPA) for a water disinfection system is based on achievement of 3-log inactivation of Cryptosporidium and Giardia in that system (USEPA 2006). Due to the main focuses of this project, the effect of fluence rate and reflection on inactivation, E.coli (American Type Culture Collection #11229, Manassas, VA) and Fecal Coliform were used as test microorganisms. E.coli ATCC 11229 is a 31  typical surrogate representing the pathogenic E.coli O157:H7 strain. A Fecal Coliform sample was isolated from wastewater samples obtained from the UBC Waste Water Treatment Pilot Plant. 3.2.2 Preparation of inoculum The E.coli sample was received as a powder from ATCC and revived based on manufacturer’s guideline. In the propagation step, one colony of microorganism (E.coli or Fecal Coliform) was detached from a freshly incubated agar plate and added to 40ml of sterile LB (nutrient broth) in a sterile 50ml vial. The solution was incubated at certain temperature (37°C for E.coli and 42°C for Fecal Coliform) and mixed at 200rpm to ensure that constant oxygen levels were provided throughout the stock. Following overnight incubation, 1mL stationery-phase stock solution was mixed with 9mL fresh LB and incubated again for 90 minutes for E.coli and 150 min for Fecal Coliform to achieve their log-growth phase. Because of the high absorbance of the broth solution, it was washed three times in Phosphate buffered saline (PBS). Each time, the sample was placed in two sterile 10-mL centrifuge tubes and spun at 33000 rcf for 5 minutes. Then, the supernatant was extracted and disposed, while 9 mL of fresh PBS was added to re-suspend the colonies and reduce the opacity because of the growth media.    3.2.3 Growth curves Growth curves for both E.coli and Fecal Coliform was sketched based on absorbance of unwashed stock solution at 600nm (OD 600). After starting the incubation, the optical density at 600nm (OD600) was measured at different time intervals. The UV-Vis spectrophotometer was used to measure the absorbance of the samples at 600nm in a 3.5 mL quartz cuvette and the zero was set using sterile fresh LB. In order to clean the cuvette for each reading, it was rinsed three times with distilled water and dried out between measurements. Samples for each microorganism 32  were cultured at two points along their curves to compare their OD600 values to their concentrations. 3.2.4 Irradiation procedure For each experimental run in either first part or second part of this research, 10mL of washed sample (E.coli or Fecal Coliform) underwent photochemical testing. Following exposure, the serial dilutions and enumeration steps were applied, respectively. 3.2.5 Enumeration‎of‎microorganisms’‎colonies The enumeration of tested microorganisms followed a similar method to that of Cho et al. (2010). The spread plate technique was used to enumerate both E.coli and Fecal Coliform before and after UV exposure. The colonies were cultivated on the agar plate (LB). LB agars can promote the growth of Fecal Coliform or E.coli, while suppressing the growth of other microorganisms. E.coli or Fecal Coliform colonies appear as white spots on this kind of agar. Two major steps in the spread plate technique are serial dilution of the sample (Fig 3.5) and spreading the sample on plates (Fig 3.6). First, the microorganism’s sample is diluted with the dilution factor of 10-1 based on following procedure:  The bench top (working surface) was cleaned with ethanol (75%). Providing fire with a Bunsen burner made the whole environment sterile.    The centrifuge tubes were labeled based on their corresponding dilution factors.  1 mL of microorganism solution (exposed or unexposed) was pipetted to the centrifuge tube, containing 9mL fresh PBS, and was mixed with the vortex maker device.  step 3 was repeated for further dilutions  33   Figure ‎3.5 Serial dilution  Figure ‎3.6 Spread-plate method  Second, each diluted solution is spread on agar plates. The details can be described as follows:    The essential information such as microbe name and dilution factor was written on each plate  For each experimental run, the last solution (for example 10-6) was mixed again by the vortex maker device. 20µL of the solution was dropped slowly at the center of petri dish. 34   The solution on the surface of the agar was spread by a sterile stainless steel spreader while the plate was turning slowly on a turntable  The spreading step was done on three plates (triplicate).  The plates were inverted and incubated for almost 12 hours.  The number of colonies on each plate was counted to calculate the number of colony forming units per 1mL. For example, if the number of colonies of one dilution were 𝑎1, 𝑎2, and 𝑎3, the average number of colonies was calculated as follows: 𝑥 =𝑎1 + 𝑎2 + 𝑎33                                         Therefore, for the dilution factor of α, the concentration (CFU/mL) of E. coli/ fecal coliform in that dilution was determined by the following formula:  𝐶 =1000 ∗ 𝑥20 ∗ 𝛼                                            This method was just suitable for 20 - 200 colonies per petri dish. If the number of colonies exceeded this range, the solution was diluted (for more than 200 CFU/dish) or concentrated (for less than 20 CFU/dish).  35  Chapter 4: Results and discussion   4.1  Spectrum analysis of UV-LEDs The spectral measurement for LG and Nikkiso UV-LEDs was conducted with an Ocean Optics USB 2000+UV/Vis spectrometer. The UV intensity was measured as a function of wavelength. Fig 4.1 shows the normalized spectral results.  Figure ‎4.1 Normalized spectral analysis for 265nm, 275nm, and 285nm UV-LEDs.    The measured peak values for all the three UV-LEDs matched well with the manufacturers’ specifications. In addition, the values for the full width at half maximum (FWHM), measured across the spectra at 50% of the peak irradiance, for 265nm, 275nm and 285nm UV-LEDs were 9.37nm, 10.47nm and 11.97nm, respectively. From the lowest to the highest wavelength UV-LEDs, the measured FWHM were slightly lower than manufacturers’ 00.20.40.60.811.2250 255 260 265 270 275 280 285 290 295 300 305 310 315 320Nominal Irradiance wavelength(nm) 265nm275nm285nm36  specifications, which are 11nm, 12nm, and 13nm, respectively. The shorter the FWHM, the more reasonable to consider the UV-LED as the monochromatic source.   4.2 Fluence rate measurements In order to apply different fluence rates on the surface of the sample during irradiation, LEDs’ forward currents were changed rather than changing the distance between the LED and the sample. This method was considered better since the radiation pattern of UV-LEDs at both high and low fluence rates remained the same. The desired distance between each LED and the sample was calculated based on following formula:  Distance between the UV − LED and the sample′s surface =𝑟𝑎𝑑𝑖𝑢𝑠 𝑜𝑓 𝑡ℎ𝑒 𝑝𝑒𝑡𝑟𝑖 𝑑𝑖𝑠ℎtan(12∗ 𝑣𝑖𝑒𝑤𝑖𝑛𝑔 𝑎𝑛𝑔𝑙𝑒 𝑜𝑓 𝑡ℎ𝑒 𝐿𝐸𝐷)    For example: Distance between the LG 275nm UV-LED and the sample’s surface = 0.5∗5𝑐𝑚 tan(0.5∗124) = 1.33 cm  The operating conditions for each LED, exposing its high and low fluence rate, are shown in Table 4.1. The fluence rate data calculated based on Iodide/Iodate actinometry and radiometer measurements were correlated well at different experimental conditions. The maximum and minimum differences between actinometry and radiometer data were 13% and 4% for the 285nm and 265nm UV- LEDs, respectively.       37  Table ‎4.1 The UV-LEDs’ characteristics at different  experimental trials Wavelength Viewing angle (degree) Distance (cm) Forward Current (mA) Fluence Rate from Actinometry (mW/cm2) ± standard deviation Average Fluence rate from Radiometer  (mW/cm2) Difference (%) 265nm 130 1.17 350 0.412 ± 0.004 0.468 12.0% 35 0.037 ± 0.001 0.042 11.3% 275nm 124 1.33 100 0.381 ± 0.001 0.410 7.1% 10 0.040 ± 0.002 0.041 4.0% 285nm 130 1.17 500 2.010 ± 0.025 1.825 10.2% 50 0.282 ± 0.001 0.250 13.0%  Table 4.2 displays the fluence rate values in Teflon, aluminum, and stainless steel petri dishes as determined by chemical actinometry. No significant difference was observed, likely because the absorption of UV by actinometry solution happens in a very thin layer close to the surface of the solution, and would not be affected by the reflection of the inner surface.  Table ‎4.2 The UV-LEDs’ fluence rates in reflective materials Wavelength Distance (cm) Current (mA) Material Fluence rate (mW/cm2) ± standard deviation 265nm 1.17 350 Teflon 0.409±0.005 Stainless Steel 0.418±0.003 Aluminum 0.409±0.003 275nm 1.33 100 Teflon 0.378±0.004 Stainless Steel 0.387±0.001 Aluminum 0.378±0.001 285nm 1.17 500 Teflon 1.994±0.003 Stainless Steel 2.041±0.006 Aluminum 1.995±0.002 38  4.3 The UV-LEDs’‎radiation outputs at their start-up conditions   The stability of each UV-LED’s output power was tested by measuring its irradiance over a certain time under the conditions given in Table 4.1. Fig 4.2 displays the change of the output power in 265nm, 275nm, and 285nm UV-LEDs through time at their highest and lowest values measured during the experimental operating conditions. All the three LEDs showed stable output power when they were operated at their low forward currents. Over the first 5 seconds and 10 seconds, the irradiance of 265nm UV-LED and 285nm UV-LED at their maximum operating conditions decreased by about 5% and 9%, respectively. Chatterley & Linden (2010) observed the same change in their study; the irradiance of their proposed UV-LED batch reactor decreased by 7% in the first 10 min. In our study, the 275nm LED’s irradiance did not change at 100 mA forward current. It should be noted that, the reductions mentioned here were considered in fluence calculations in this project.   1.361.381.41.421.441.461.481.51.521.540 5 10 15 20 25 30 35 40irradiance (mW/cm2) Time(second) a 39       0.1100.1110.1120.1130.1140.1150.1160.1170.1180.1190.1200 5 10 15 20 25 30 35 40Irradiance (mW/cm2) Time (second) b 0.670.6750.680.6850.690.6950.70 5 10 15 20 25 30 35 40Irradiance (mW/cm2) Time(second) c 40      0.06000.06100.06200.06300.06400.06500.06600.06700.06800.06900.07000 5 10 15 20 25 30 35 40iradiance (mW/cm2) Time(second) d 4.254.34.354.44.454.54.554.60 5 10 15 20 25 30 35 40Irradiance (mW/cm2) Time(second) e 41   Figure ‎4.2 The UV-LEDs' performance within their start-up conditions: a) 265nm F.C=350mA b) 265nm F.C=35mA c) 275nm F.C =100mA d) 275nm F.C =10mA e) 285nm F.C =500mA f) 285nm F.C =50mA   4.4 Growth curves The growth curves of the E.coli ATCC 11229 and the Fecal Coliform are presented in Fig 4.3. As the graphs show, Fecal Coliform needs about two hours more time to reach growth phase compared to the E.coli strain at the same growing condition. The following equations can be helpful to estimate the tested organisms’ initial concentration (CFU/mL) by using the absorbance of microbial stock solution at 600nm (OD 600).   E.coli ATCC 11229: OD600 ×2.5×108 (0.1<OD600<0.8) Fecal Coliform: OD600×2.6×108 (0.15<OD600<0.85) 0.560.5650.570.5750.580 10 20 30 40 50Irradiance (mW/cm2) Time(second) f 42    Figure ‎4.3 The growth curves for the E.coli and the Fecal Coliform (cultured colonies (  ))  This information is necessary in order to ensure that the test organism is in its growth phase during the experiments; otherwise, the results of inactivation would not be accurate. At the 0.00E+005.00E+071.00E+081.50E+082.00E+082.50E+083.00E+083.50E+084.00E+0800.20.40.60.811.21.40 2 4 6 8 10 12 14 16 18 20 22 24 26Concentration (CFU/ml) OD 600 Time (hours) E.coli ATCC 11229 0.00E+005.00E+071.00E+081.50E+082.00E+082.50E+083.00E+083.50E+084.00E+084.50E+085.00E+0800.20.40.60.811.21.41.61.80 2 4 6 8 10 12 14 16 18 20 22 24 26Concentration (CFU/ml) OD 600 Time (hours) Fecal Coliform 43  stationary phase, the natural death of organisms plays an important role in reducing their number and that is an extra factor beside UV irradiation for inactivation. Morton & Haynes (1969) found that E.coli in the stationary or late log phase was about 1-2 log more sensitive to UV radiation compared to E.coli in early log phase. The cultured colonies used in this project had the OD600 of around 0.5 in their initial condition. 4.5 Specular reflection measurement at 45°  Table 4.3 displays the specular reflectance of different petri dishes at 45°. Even though the values are not accurate because of considering measurement at only one angle, they were considered as helpful values for comparing the materials’ reflectivity, relatively. According to this data, the reflectance of aluminum is almost twice higher than that of stainless steel and it is half that of Teflon. The reflectance values for both stainless steel and aluminum are less than the ones reported by Li et al. (2012) (30% for stainless steel and 80% for aluminum which both were covered with quartz during the measurement). Their data are more accurate than the results provided here because of the apparatus that they used for measuring the specular reflection; however, they are not valid for this project and the difference between the surface finish may play a determining factor in calculating the reflectivity.        44  Table ‎4.3 Comparison between 45° specular reflectivity of different petri dishes    Material  Wavelength  Aluminum Stainless Steel Teflon 265nm 54 23 90 275nm 48 29 93 285nm 48 24 89  4.6 UV inactivation behavior of E.coli and Fecal Coliform The inactivation of the test organisms under standard conditions of 0.2mW/cm2 UV fluence rate suggested by Sommer et al. (1998) are shown in Fig 4.4. At 265nm, the inactivation kinetics of both E.coli and Fecal Coliform reveal high UV susceptibility of these organisms. A 4-log reduction is already reached at a fluence of around 8mJ/cm2. This value is slightly higher than the reported one by Sommer et al. (1998) for the same strain of E.coli at 254nm (4-log reduction at UV dose of 10mJ/cm2). Beside the fact that different apparatus were used in both studies, the difference might be explained by higher DNA absorbance of bacteria at around 260nm (USEPA 2006).  There is a linear relationship between the log inactivation and fluence values of 0-9 mJ/cm2 for E.coli and in the fluence range of 4-9 mJ/cm2 for Fecal Coliform. At fluences less than 4mJ/cm2, the inactivation of Fecal Coliform is showing a shoulder effect and hence, is less than that of E.coli. 45    01234560 2 4 6 8 10Log inactivation (Log N0/N) Fluence (mJ/cm2) a E.coliFecal Coliform00.511.522.533.544.50 2 4 6 8 10Log inactivation (Log N0/N) Fluence (mJ/cm2) b E.coliFecal Coliform46   Figure ‎4.4 E.coli and Fecal Coliform UV susceptibility: a) 265nm b) 275nm c) 285nm.  As the wavelength increases from 265nm to 285nm, the UV susceptibility of both E.coli and Fecal Coliform decreases. A 4-log reduction for E.coli and Fecal Coliform happens at fluences around 9mJ/cm2 and more than 10mJ/cm2 for 275nm and 285nm, respectively. The lower log inactivation at higher wavelengths is believed to be the result of lower DNA absorbance of microorganisms at those wavelengths. A shoulder effect is observed for E.coli inactivation at 275nm and fluences less than 4mJ/cm2, while the kinetics of Fecal Coliform inactivation follows a log linear reduction at fluences of 0-9 mJ/cm2 and the same wavelength. Overall, at given wavelengths, the UV susceptibility of both E.coli and Fecal Coliform is the same and it decreases as the wavelength increases. 4.7 Effect of reflection on E.coli inactivation Fig 4.5 illustrates the E.coli log reductions in different reflective petri dishes as well as the reduction in quartz petri dish (control) at two fluences for each wavelength. As expected, the log inactivation of the tested organism appeared to improve because of the reflection in all cases; 00.511.522.530 2 4 6 8 10 12Log inactivation (Log N0/N) Fluence (mJ/cm2) c E.coliFecal Coliform47  however, the significance of this improvement was different at different conditions. At 265nm, the stainless steel’s reflection slightly increased the E.coli reduction, while the inactivation of E.coli was improved markedly by around 1 and 2 log with aluminum and Teflon, respectively. At 275nm, the reduction of E.coli within stainless steel was almost 1 log more than that of quartz, aluminum resulted in about 50% more log inactivation than that of quartz and no colony was seen on the agar plates covered with irradiated samples in Teflon at fluences of 4 and 5 mJ/cm2. It was concluded that more than 2 log increase compared to inactivation in the quartz petri dish had happened in the Teflon at 275nm. At 285nm, the log inactivation in both stainless steel and aluminum was a bit higher than that in the quartz petri dish, and Teflon performed significantly better than the others with increasing the E.coli reduction by about 1.5 log. Although actinometry results did not show any difference in fluence rate values inside different petri dishes (the results shown in Fig 4.5), it is reasonable to assume that the reflection must have increased the fluence rate values inside these petri dishes. Therefore, another path was used to find the effective fluence rate values inside the UV reflective materials. The details and discussions around this approach are provided in the next section.         48     0.000.501.001.502.002.503.002 4Log inactivation Fluence(mJ/cm2) a QuartzStainless SteelAluminumTeflon0.000.501.001.502.002.503.003.502 4Log inactivation Fluence (mJ/cm2) b QuartzStainless SteelAluminum49   Figure ‎4.5 E.coli inactivation inside reflective materials: a) 265nm b) 275nm c) 285nm.  4.7.1 Evaluation of the absorbed energy by E.coli sample during UV irradiation As it was discussed before, applying the iodide/iodate actinometry resulted in the same amount of fluence rate for each individual wavelength inside different petri dishes. On the other hand, higher log inactivation of the E.coli inside the reflective petri dishes compared to that of the quartz petri dish means higher number of photons inside those reflective materials. Therefore, if the amount of the energies (aka photons) reaching the microorganisms inside different petri dishes could be calculated, the comparison would be used to calculate the effect of reflection on microorganisms’ inactivation. In order to do so, the E.coli inactivation kinetic inside the quartz petri dish was needed because based on the log inactivation in a reflective material; the improvement of energy within a reactor made of that material, because of its 0.000.501.001.502.002.503.003.504 8Log inactivation Fluence(mJ/cm2) c QuartzStainless steelAluminumTeflon50  reflection, could be calculated. The set-up shown in Fig 4.6 was assembled in order to find the absorbed energy by the E.coli solution during UV-LED’s irradiation. A quartz petri dish (90% transparent in UV range) with similar size to other petri dishes was used for holding the water sample containing E.coli and was placed on a 9 mm glass petri dish including chemical actinometry solution. The distance between sample surface and the UV source was adjusted based on the viewing angle of each LED (Table 4.2). A black cover was used in order to ensure that the UV photons emitted from the angles outside the range of the LED’s viewing angle would not be absorbed by the actinometry solution.    Figure ‎4.6 Schematic diagram of experimental set-up for absorbed energy calculation  4.7.1.1 Calculation of the absorbed by the E.coli water sample during UV irradiation Based on the results of chemical actinometry in the 5mm and 9mm petri dishes, the average UV fluence rate inside the E.coli sample during UV irradiation could be calculated. The following illustrates a sample calculation for case of the 265nm UV-LED.  Wavelength: 𝜆 = 265𝑛𝑚 Heat sink Stirrer  Sample Dark cover Actinometry solution LED 51  Viewing angle: 𝛩 = 1300 Area of the 5mm petri dish: 𝐴1 =𝜋4 × (5)2 = 19.63𝑐𝑚2 Exposed area in the 9mm petri dish: 𝐴2 =𝜋4× (5.6 + 2 × (0.8 × tan(65)))2 = 64.06𝑐𝑚2 Average fluence rate in the 5mm petri dish: 𝐸′1 = 0.4 𝑚𝑊/𝑐𝑚2 Average fluence rate in the 9mm petri dish: 𝐸′2 = 0.1 𝑚𝑊/𝑐𝑚2 Exposure time: 𝑡 = 50 𝑠𝑒𝑐𝑜𝑛𝑑𝑠  Figure ‎4.7 Essential dimensions for calculating energy inside E.coli sample   𝑇ℎ𝑒 𝑎𝑏𝑠𝑜𝑟𝑏𝑡𝑖𝑜𝑛 𝑓𝑎𝑐𝑡𝑜𝑟 𝑐𝑎𝑙𝑐𝑢𝑙𝑎𝑡𝑒𝑑 𝑏𝑎𝑠𝑒𝑑 𝑜𝑛 𝑈𝑉 − 𝐿𝐸𝐷′𝑠 𝑒𝑚𝑖𝑠𝑠𝑖𝑜𝑛 (𝐴𝐹)= −log (𝐸𝑛𝑒𝑟𝑔𝑦𝑖𝑛/𝐸𝑛𝑒𝑟𝑔𝑦𝑜𝑢𝑡) 𝐴𝐹 = − log (0.1 ∗ 50 ∗ 64.060.4 ∗ 50 ∗ 19.63) = 0.0884  𝐴𝑡𝑡𝑒𝑛𝑢𝑎𝑡𝑖𝑜𝑛 𝐹𝑎𝑐𝑡𝑜𝑟 (𝐴𝑡𝐹) = (1 − 10−𝐴𝐹)/(𝐴𝐹𝑙𝑛(10)) 𝐴𝑡𝐹 =1 − 10−0.08840.0884 ∗ ln(10)= 0.9048 𝑀𝑜𝑑𝑖𝑓𝑖𝑒𝑑 𝑈𝑉 − 𝐿𝐸𝐷 𝑓𝑙𝑢𝑒𝑛𝑐𝑒 𝑟𝑎𝑡𝑒  𝑏𝑦 𝑐𝑜𝑛𝑠𝑖𝑑𝑒𝑟𝑖𝑛𝑔 𝑡ℎ𝑒 𝑎𝑏𝑠𝑜𝑟𝑝𝑡𝑖𝑜𝑛 𝑓𝑎𝑐𝑡𝑜𝑟(𝑀𝐹𝑅)= 𝐸1′ ∗ 𝐴𝑡𝐹 𝑀𝐹𝑅 = 0.4 ∗ 0.9048 = 0.3619 (𝑚𝑊/𝑐𝑚2)  Actinometry Solution Sample LED Θ/2 5 mm 5.6 cm 5 Θ 3 mm 52  The same method was used for calculation of the modified UV-LED’s fluence rate for the 275nm and 285nm UV-LEDs. The results are reported in Table 4.4.  Table ‎4.4 Comparison between the absorbed energy by the sample at different wavelengths 𝜆 (𝑛𝑚) 𝛩 (𝑑𝑒𝑔𝑟𝑒𝑒) 𝐴1  (𝑐𝑚2) 𝐴2 (𝑐𝑚2) 𝐸′1 (𝑚𝑊𝑐𝑚2) 𝐸′2 (𝑚𝑊𝑐𝑚2) 𝑡𝑖𝑚𝑒 (𝑠) 𝐴𝐹 𝐴𝑡𝐹 𝑀𝐹𝑅 (𝑚𝑊𝑐𝑚2) 265 130 19.63 64.06 0.4 0.1 50 0.0884 0.9048 0.3619 275 124 19.63 58.21 0.4 0.11 70 0.0886 0.9046 0.3618 285 130 19.63 64.06 2 0.6 20 0.0092 0.9895 1.979  The absorption factor represents the absorption of UV energy emitted by LEDs using Beer-Lambert concept, taking into account the variation in intensity and incident angle. Based on Beer-Lambert Law, at specific wavelength and path length, the absorbance can be calculated using the following formula: 𝐴 = log (𝐼0/𝐼) where 𝐼0 is the intensity of light incident on the sample (mW/cm2), and 𝐼 is the intensity of light transmitted through the sample (mW/cm2). However, there are two main differences between absorption factor calculated in this work and the one from Beer-lambert Law. First, in Beer-Lambert Law, it is assumed that the light is emitting perpendicularly on the sample, while, for the calculated absorption factor based on UV-LED’s emissions, this assumption is not valid. Second, the sample is emitted by light source in Beer-Lambert Law calculation receives one uniform beam, while, the irradiated sample by the UV-LED receives more than one beam and with different intensity at different angles. Non-uniformity of the UV-LED’s emission at different angles is because of the UV-LED’s radiation pattern.  53  The attenuation factor introduced here represent the absorption of UV energy by water using the water factor concept, taking into account the path length of different UV rays traveling in water at different incident angles. Water factor takes into consideration the absorption of the solution and is used in UV collimated beam protocol for dose calculation. The formula for water factor calculation is as follows. 𝑊𝑎𝑡𝑒𝑟 𝑓𝑎𝑐𝑡𝑜𝑟 =1 − 10−𝑎𝑙𝑎𝑙𝐿𝑛(10) where 𝑎 is sample absorbance and 𝑙 is the sample depth. Table 4.5 compares the water factor and the attenuation factor for the system irradiated with UV-LEDs with different wavelengths.  Table ‎4.5 comparison between water factor and attenuation factor 𝑈𝑉 − 𝐿𝐸𝐷 𝑎 𝑎𝑙 𝐴𝐹 𝑊𝑎𝑡𝑒𝑟 𝑓𝑎𝑐𝑡𝑜𝑟  𝐴𝑡𝐹 265 0.2884 0.1442 0.0884 0.8509 0.9048 275 0.2969 0.1485 0.0886 0.8470 0.9046 285 0.2970 0.1485 0.0092 0.8470 0.9895  Comparing the last two columns of Table 4.5, the attenuation factors are higher than the water factors for all the wavelengths. The reason can be the path length in water factor calculation. The sample thickness (path length) in water factor calculation is assumed 0.5 cm; however, because the UV-LED’s rays are not collimated, they go through different path lengths inside the sample. Most of these paths are longer than 0.5 cm, which might be the reason that the attenuation factors are higher than the water factors. 54  Although the proposed way of calculating the UV-LED’s fluence rate in this thesis has some errors because what was discussed above, it was found to be the best method for generating E.coli kinetic data for different UV-LEDs experimentally.  4.7.1.2 Development of the E.coli inactivation kinetics based on the absorbed energy by the solution The E.coli containing water samples were exposed to different incident UV fluences (𝐸′1 × 𝑡) in the experimental set-up described in Fig 3.1. The microbial preparation and the enumeration part were done by the procedures described in the methodology section of this thesis. Based on the attenuation factors (AtF) provided in the Table 4.5, the measured incident fluences on the surface of the sample were changed to the fluence absorbed by the sample. Table 4.6 includes E.coli inactivation results of the three LEDs at different fluences as well as their corresponding fluences. Table ‎4.6 The E.coli log inactivation results at different calculated energy per volume. 𝑊𝑎𝑣𝑒𝑙𝑒𝑛𝑔𝑡ℎ 𝐸′1 × 𝑡𝑖𝑚𝑒 𝑀𝐹𝑅 × 𝑡𝑖𝑚𝑒 𝐿𝑜𝑔 𝑖𝑛𝑎𝑐𝑡𝑖𝑣𝑎𝑡𝑖𝑜𝑛 ± 𝑆𝑡𝑎𝑛𝑑𝑎𝑟𝑑 𝑑𝑒𝑣𝑖𝑎𝑡𝑖𝑜𝑛 265 2 1.81 0.77 ± 0.17 4 3.62 1.30 ± 0.10 5 4.52 1.80 ± 0.20 8 7.24 3.42 ± 0.31 275 4 3.62 0.92 ± 0.11 5 4.52 1.16 ± 0.27 8 7.24 2.55 ± 0.15 9 8.14 3.05 ± 0.23 285 4 3.96 0.40 ± 0.02 8 7.92 1.14 ± 0.20  Fig 4.8 shows the inactivation kinetics of E.coli based on the modified UV fluence at different wavelengths. 55   Figure ‎4.8 The E.coli log inactivation versus modified fluence at different wavelengths  The E.coli kinetic of inactivation derived from the modified fluence was used in order to calculate the inactivation improvement because of implementing each reflective material at provided wavelengths. At the given fluences on the surface, the UV-LED fluence inside the reflective materials was generated based on the log reduction of tested organism by using the kinetic equation of the E.coli inactivation at each provided wavelength. The ratio of effective energy absorbed by microorganism’s solution inside reflective material over the absorbed energy inside quartz petri dish was taken as a main criterion for reporting inactivation improvement because of that specific reflective material.      y = 0.4994x - 0.3239 R² = 0.9733 y = 0.4818x - 0.9132 R² = 0.9934 y = 0.187x - 0.34 R² = 1 00.511.522.533.540 1 2 3 4 5 6 7 8 9Log inactivation Modified Fluence (mJ/cm2) 265nm275nm285nm56  Table ‎4.7 Comparison between the reflective materials’‎improvement‎in‎log‎reduction‎of‎E.coli  Material Wavelength Log inactivation improvement ± Standard deviation Aluminum 265 1.44 ± 0.36 275 1.78 ± 0.20 285 1.27 ± 0.10 Stainless Steel 265 1.10 ± 0.26 275 1.28 ± 0.08 285 1.14 ± 0.15 Teflon 265 2.08 ± 0.80 275 𝑁𝑎𝑁 285 2.55 ± 0.98   One of the important factors affecting the energy distribution inside a container made of reflective material is the diffuse reflectance of the surface. There are two types of reflection: specular and diffuse. Specular reflection is that obtained with a mirror in which the reflected image matches the illuminated object. On the other hand, in diffuse reflection, every element of the surface becomes a new UV source radiating in all directions outward from the surface (Li et al., 2012). The diffuse reflection highly depends on the surface properties mainly the surface material and the surface roughness. Therefore, the reflection behavior of a material at different wavelengths and the roughness of its surface are the main surface related factors causing the variation in the numbers in the last column of Table 4.6. The other reason is the transmittance of the microorganism sample. Li et al. (2012) showed that fluence rate distribution in the high reflective materials is very sensitive to UV transmittance of the sample. In our study, the average UV transmittance of the sample was about 75%; however, the fluctuation around this number could cause high standard deviations especially for results obtained on Teflon.   The method explained here for calculating the inactivation improvement because of reflection can be used for analyzing other possible reflective materials for building a UV reactor. 57  This method does not depend on the material; however, it depends on the tested microorganism. Since the received energy by microorganism at different conditions is back calculated from the kinetic model, different kinetic models result in different energies. MS2 and Bacillus subtilis are the indicators for the purpose of reactor validation in US and Europe, respectively. MS2 dose-response curve follows a log linear model but the Bacillus subtilis kinetic has three sections; shoulder, log linear and tailing. Therefore, it is expected that the results of back calculating the UV fluence from these two indicators will not be the same. Therefore, to use this method, first the target microorganism should be selected. Then, the kinetic model can be derived based on the method presented in this study. After that, in any material, the essential UV fluence for different level of inactivation can be back calculated based the kinetic model, and is specific to the tested organism.              58  4.8 The effect of UV-LED’s fluence rate on E.coli inactivation The E.coli sample was exposed to two different levels of fluence rates for each of the three wavelengths under the experimental conditions (0.4mW/cm2, 0.04mW/cm2 for 265nm; 0.4mW/cm2, 0.04mW/cm2 for 275nm; and 2mW/cm2, 0.3mW/cm2 for 285nm). The results are shown in Fig 4.9.   Figure ‎4.9 The UV inactivation of E.coli under different UV-LEDs’ fluence rates. Data are expressed as the mean of logarithmic reduction in duplicate; standard error is indicated as bars at extreme points.  As it was discussed previously, the E.coli inactivation rate was higher at 265nm followed by that of the 275nm and 285nm. The results of E.coli inactivation at 275nm is consistent with the study completed by Bowker et al. (2010). Fig 4.9 clearly shows that the inactivation of E.coli 01234560 1 2 3 4 5 6 7 8 9 10 11Log inactivation (Log N0/N) Fluence mJ/cm2 265nm (0.4 mW/cm2)265nm (0.04 mW/cm2)275 nm (0.4 mW/cm2)275 nm (0.04 mW/cm2)285nm (2 mW/cm2)285nm (0.3 mW/cm2)59  ATCC 11229 with UV-LED depends only on the UV fluence and not on the UV fluence rates within the range of 0.04 to 0.4mW/cm2 for 265nm and 275nm, respectively, and 0.3mW/cm2 to 2mW/cm2 for 285nm. In other words, the kinetics of E.coli inactivation is following Bunsen-Rosque law at different UV-C wavelengths provided by the UV-LEDs. This results are in contrast to the study conducted by Sommer et al. (1998) who found around 1 log higher inactivation at UV fluence rate of 2W/cm2 compared to that of 0.02 W/cm2 for the same strain of E.coli at 254nm UV lamp reactor in the range of 8 to 10mJ/cm2. The deviation can be because of the different wavelengths, experimental set-ups and conditions used in both studies.  4.9 The effect of UV-LED’s fluence rate on Fecal Coliform inactivation The Fecal Coliform sample was tested in the same experimental conditions as those of the E.coli solution. As shown in Fig 4.10, protraction of UV irradiation results in higher inactivation of Fecal Coliform and this observation was consistent in all three wavelengths. The difference between log inactivation at low fluence rate and that at high fluence rate becomes greater as the UV fluence increases in all three cases. For example, at 275nm the same log inactivation can be seen for both high and low fluence rates up to the fluence of 4mJ/cm2, while the difference of around 2 log reduction is observed between inactivation at highest and lowest fluence rates at the fluence of 8mJ/cm2. Based on the trends shown in Fig 4.10, it is expected that at higher fluence values than those tested experimentally in this study, the difference between log inactivation of the high and low fluence rates becomes even more significant.  60   Figure ‎4.10 The UV inactivation of the Fecal Coliform under different UV-LEDs’ fluence rates. Data are expressed as the mean of logarithmic reduction in duplicate; standard error is indicated as bars at extreme points. The reason for greater inactivation of Fecal Coliform at low fluence rate compared to that at high fluence rate under the given fluences might be the high UV absorption of the Fecal Coliform sample at low fluence rates. The set-up shown in Fig 4.6 was used in order to test this hypothesis. E.coli or Fecal Coliform sample with the same initial concentration, was exposed to the UV-LEDs under their high and low fluence rates (0.4mW/cm2, 0.04mW/cm2 for 265nm; 0.4mW/cm2, 0.04 mW/cm2 for 275nm; and 2mW/cm2, 0.3mW/cm2 for 285nm) and the fluence rates, delivered to the actinometry solution within the 9mm petri dish, was calculated. In order to ensure that the initial concentrations of both organisms were the same, the UV absorbance of their stock solution was measured by UV-Vis spectrophotometer at 265nm, 275nm, and 285nm. Fig 4.11 displays the UV absorption of E.coli and Fecal Coliform samples at the three tested wavelengths. 0.000.501.001.502.002.503.003.504.000 1 2 3 4 5 6 7 8 9 10 11Log inactivation Fluence (mJ/cm2) 265nm (0.04 mW/cm2)265nm (0.4 mW/cm2)275 nm (0.04 mW/cm2)275 nm (0.4 mW/cm2)285nm (0.3 mW/cm2)285nm (2 mW/cm2)61   Figure ‎4.11 The UV absorption of unexposed E.coli and Fecal Coliform sample at 265nm, 275nm, and 285nm.  Fig 4.12 displays the fluence rates measured in the 9mm glass petri dish at both high and low UV fluence rates.  00.050.10.150.20.250.30.35260 265 270 275 280 285 290UV absorption Wavelength (nm) E.coliFecal Coliform00.10.20.30.40.50.60.7260 265 270 275 280 285 290fluence rate (mW/cm2) wavelength (nm) a E.coliFecal Coliform62   Figure ‎4.12 The measured fluence rates in 9mm petri dish when organism sample was irradiated to: a) 0.4mW/cm2 at 265nm, 0.4mW/cm2 at 275nm, and 2mW/cm2 at 285nm b) 0.04mW/cm2 at 265nm, 0.04 mW/cm2 at 275nm, and 0.3mW/cm2 at 285nm.  As shown in Fig 4.12, the mentioned hypothesis can be rejected because the amount of fluence rate received by the actinometry solution, placed under the quartz petri dish, is nearly equal for both E.coli and Fecal Coliform samples and it is consistent for all three wavelengths tested. It can also be concluded that both tested microorganisms absorbed the same amount of energy during the irradiation under the high and low UV fluence rates. Therefore, the deviation from time-dose reciprocity rule for the Fecal Coliform kinetics should be explored in other ways. Another hypothesis can be the insufficient mixing at short exposure times at high fluence rates compared to longer exposure times at low fluence rates. The long exposure time or mixing time decreases the chance of having agglomeration in a microbial sample which can cause blocking effects for UV exposures; however, as the fluence increases, the difference between 00.020.040.060.080.10.12260 265 270 275 280 285 290fluence rate (mW/cm2) Wavelength (nm) b E.coliFecal Coliform63  high and low fluence rate inactivation curves also increases in these results, making this hypothesis at least invalid for this study. The deviation in inactivation behavior of Fecal Coliform at different UV intensities was observed previously for different types of microorganisms. Sommer et al. (1996) found that more inactivation of the yeast cells was achieved for a given dose at 254 nm when the samples were exposed to low UV fluence rate (0.02 W/m2) compared to high UV fluence rate (2W/m2). They also showed that the difference was more pronounced after a certain UV dose (75 J/m2). Since the fecal coliform sample used in this project includes more than a single type of bacteria, it would be plausible that each of these bacteria has different behavior in terms of UV inactivation at various UV fluence rates. Therefore, Fig 4.8 shows their overall reaction to the UV irradiation at different UV fluence rates. Another reason might be different time-dose effects on the efficiency of repair systems. UV radiation results in DNA changes that are almost exclusively damaging to all organisms. However, UV irradiation affects almost all organisms in their natural environment (mostly UV-A and some levels of UV-B) because of the UV light passed to the Earth’s surface through the atmosphere. As a result, they are equipped with repair mechanisms. One such mechanism is dark repair or nucleotide excision repair (Rastogi et al., 2010). In this process, when DNA damage is recognized, the damaged part and the surrounding DNA are excised while the section of single stranded DNA remains on the normal double stranded macromolecule. Then, the normal cellular replication mechanism fills the gap by re-synthesizing the excised section of DNA using the other strand as a template. In simpler microorganisms, their DNA is not protected in a membrane-enclosed nucleus, another repairing mechanism called photo-reactivation or light repair is used for coping with UV radiation. Rather than the removal of DNA, this mechanism utilizes 320-370nm photons to activate the enzyme photolyase, 64  catalyzing the reverse reaction of UV-induced dimerization. As a result, the DNA damage is reversed, thereby inhibiting mutation and permitting the replication to proceed. In this study, dark repair was the only possible mechanism for reactivation of bacteria cells since the room light was blocked by a black box from the samples during UV exposures. About fecal coliform, the reactions and the components involved in repairing of the exposed sample at dark condition might be faster and more active at high UV fluence rates, respectively. Therefore, the higher UV inactivation of fecal coliform happened at lower UV fluence rates. This idea was primarily proposed by Sommer et al. (1998)  as a way to justify the deviation from time-dose reciprocity law seen in yeast cells and it has not been rejected so far by other researchers.         65  Chapter 5: Conclusions and recommendations  5.1 Conclusions This research focused on investigating the impact of reflection as well as fluence rate on the inactivation of E.coli ATCC11229 and Fecal Coliform by UV-LEDs. Three different UV-LEDs, 265nm, 275nm and 285nm, as well as four different reactor materials were examined in this study. Teflon, aluminum, stainless steel and quartz (as control) were used as the reactor materials. The followings highlight some of the main conclusions from this research:  At the surface-averaged UV fluences of 2 and 4 mJ/cm2 for both 265nm and 275nm UV-LEDs and 4 and 8 mJ/cm2 for 285nm UV-LED, the log inactivation of E.coli in Teflon, aluminum and stainless steel container were approximately 100%, 60% and 30% higher than that of the quartz, respectively.   No significant difference was observed between E.coli ATCC11229 inactivation data at the surface-averaged fluence rates of 0.4mW/cm2 and 0.04mW/cm2 at 265nm, 0.4mW/cm2 and 0.04mW/cm2 at 275nm and 2mW/cm2 and 0.3 mW/cm2 at 285nm.  For similar experimental conditions, Fecal Coliform showed different behavior compered to E.coli; higher inactivation was obtained at low fluence rates. This observation was consistent for all the three UV-LEDs.  The average fluence rates on the sample’s surface calculated by iodide/iodate actinometry and spectroradiometer matched well for the three tested UV-LEDs at their high and low input currents. The maximum difference between two methods was 13% for 285 nm UV-LED, working at its low input current. 66   No significant difference was observed among the average fluence rates on the surface of the sample in Teflon, stainless steel, aluminum and quartz petri dishes.    Based on the new experimental method proposed for evaluating the absorption of the sample under UV-LED’s irradiation, the attenuation factor was defined. For all the three tested UV-LEDs, the calculated attenuation factors were lower than the water factors.   The calculated attenuation factors were used to modify the calculated average fluence rate on the sample’s surface to the one absorbed by the microorganism sample during irradiation. The inactivation data of E.coli based on the absorbed UV fluences were proposed as the E.coli kinetics at the three tested wavelengths.   The relative reflectivity at 45° of Teflon, aluminum and stainless steel was about 90%, 50%, and 25%, respectively.   At the three tested wavelengths, there was no significance difference between the Fecal Coliform and the E.coli inactivation behavior at the surface-averaged fluence rate of 0.2mW/cm2.       There was no significant difference between the absorption of the E.coli sample and the Fecal Coliform sample under UV-LEDs’ exposures at their high and low fluence rates. 5.2 Recommendation for future work  Based on the results of this study, the following topics and areas are recommended for future studies:  The apparatus used in this study for measuring the reflectivity of aluminum, stainless steel and Teflon surfaces resulted in very good approximate numbers for 67  the specular reflectance at 45°; however, developing an apparatus for measuring the reflection at different angles and at different points on the surface would result in more accurate estimation of the total reflection.  It would be beneficial to study the effect of utilizing reflective materials as a thin layer around quartz reactors under the same experimental conditions. The advantages of this idea would be addressing the concerns such as diffusion of wall materials into the water and cleaning the wall surface over a long operating time..  The deviation observed in Fecal Coliform behavior from time-dose reciprocity law could be attributed to different hypotheses such as improvement of the repair systems under low fluence rates or different UV sensitivity of components involved in repair machineries; however, it would be helpful to confirm these hypotheses through further investigations. Future work could focus on evaluating the repair efficiency of different microorganisms under different UV fluence rates.  It would be useful to study the effect of the UV fluence rate on the inactivation of viruses and spores under UV-C wavelengths. Based on US surface water regulation, the indicator for validating a UV disinfection reactor is MS2, while in Europe, Bacillus subtilis spore has been reported for this purpose. For both cases, it is expected that their inactivation behaviors do not deviate from time-dose reciprocity law because viruses and spores are inactive during UV irradiation and no repair mechanisms happen in them during the UV exposure. If experimental work confirms that, the inactivation behavior of MS2 and Bacillus subtilis follow the time-dose reciprocity law, it would be more reasonable to attribute the deviation from Bunsen–Rosque law 68  observed in Fecal Coliform inactivation behavior to their repair mechanisms efficiencies under different UV fluence rate conditions.69  Bibliography  Anon, Web search (25/7/2015). Available at: https://en.wikipedia.org/wiki/H._J._Round. Blatchley, E.R., 1997. Numerical modelling of UV intensity: Application to collimated-beam reactors and continuous-flow systems. Water Research, 31(9), pp.2205–2218. Bolton, J.R., 2010. Ultraviolet Applications Handbook, third., Edmonton, AB, Canada: ICC Lifelong Learn Inc. Bolton, J.R., 2000. Calculation of ultraviolet fluence rate distributions in an annular reactor: Significance of refraction and reflection. Water Research, 34(13), pp.3315–3324. Bolton, J R; and Linden, K G. 2003. Standardization of methods for fluence (UV dose) determination in bench-scale UV experiments, J. Environ. Eng., 129(3): 209–215. Bowker, C. et al., 2010. Microbial UV fluence-response assessment using a novel UV-LED collimated beam system. Water Research, 45(5), pp.2011–2019.  Camel, V. & Bermond, A., 1998. The use of ozone and associated oxidation processes in drinking water treatment. Water Research, 32(11), pp.3208–3222. Chatterley, C., 2009. UV-LED irradiation technology for point-of-use water disinfection in developing communities. Chatterley, C. & Linden, K., 2010. Demonstration and evaluation of germicidal UV-LEDs for point-of-use water disinfection. Journal of Water and Health, 8(3), pp.479–486. Chevremont, A. et al., 2012. Multivariate optimization of fecal bioindicator inactivation by coupling UV-A and UV-C LEDs. Desalination, 285, pp.219–225.  Cho, M. et al., 2010. Mechanisms of Escherichia coli inactivation by several disinfectants. Water Research, 44(11), pp.3410–3418.  70  Guo, Q. & Yoshida, A., 1994. Temperature Dependence of Band Gap Change in InN and AlN. Japanese Journal of Applied Physics, 33(5R), pp.2453–2456.  Harm, W., 1968. Effects of Dose Fractionation on Ultraviolet Survival of Escherichia Coli. Photochemistry and Photobiology, 7(1), pp.73–86.  Ibrahim, M. a S. et al., 2014. Evaluating the impact of LED bulb development on the economic viability of ultraviolet technology for disinfection. Environmental technology, 35(1-4), pp.400–406.  Irazoqui, H.A., Cerdá, J. & Cassano, A.E., 1973. Radiation profiles in an empty annular photoreactor with a source of finite spatial dimensions. AIChE Journal, 19(3), pp.460–467.  Jacob, S. & Dranoff, J., 1970. Light intensity profiles in a perfectly mixed photoreactor. AIChE Journal, 16(3), pp.359–363.  Levinshtein, M.E., Rumyantsev, S.L. and Shur, M.S., 2001. Properties of Advanced Semiconductor Materials: GaN, AIN, InN, BN, SiC, SiGe. John Wiley & Sons. Li, M. et al., 2012. Impact of reflection on the fluence rate distribution in a UV reactor with various inner walls as measured using a micro-fluorescent silica detector. Water Research, 46(11), pp.3595–3602.  V. M. Bakhir., 2003. Disinfection of drinking water: problems and solutions.  Meyer-Arendt, J.R., 1995. Introduction to Classical and Modern Optics, Prentice Hall.  Mishima, O. et al., 1988. Ultraviolet light-emitting diode of a cubic boron nitride pn junction made at high pressure. Applied Physics Letters, 53(11), pp.962–964. Morris, J.P., 2012. Disinfection of Bacillus Subtilis Spores Using Ultraviolet Light Emitting Diodes. Morton, R. a & Haynes, R.H., 1969. Changes in the Ultraviolet Sensitivity of Escherichia coli 71  During Growth in Batch Cultures. Journal Of Bacteriology, 97(3), pp.1379–1385. Oguma, K. et al., 2013. Application of UV light emitting diodes to batch and fl ow-through water disinfection systems. Desalination, 328, pp.24–30.  Rastogi, R.P. et al., 2010. Molecular Mechanisms of Ultraviolet Radiation-Induced DNA Damage and Repair. Journal of Nucleic Acids, pp.1–32.  set catalogue (25/7/2015), Application. Available at: http://s-et.com/applications.html. Shatalov, M. et al., 2002. Deep ultraviolet light-emitting diodes using quaternary AlInGaN multiple quantum wells. IEEE Journal of Selected Topics in Quantum Electronics, 8(2), pp.302–309.  Shur, M.S. & Gaska, R., 2010. Deep-Ultraviolet Light-Emitting Diodes. IEEE Transactions on Electron Devices, 57(1), pp.12–25.  Sommer, R. et al., 1996. Increased inactivation of Saccharomyces cerevisiae by protraction of UV irradiation. Applied and environmental microbiology, 62(6), pp.1977–83.  Sommer, R. et al., 1998. Time dose reciprocity in UV disinfection of water.pdf. Water Science and Technology, 38(12), pp.145–150. Taniyasu, Y., Kasu, M. & Makimoto, T., 2006. An aluminium nitride light-emitting diode with a wavelength of 210 nanometres. Nature, 441(7091), pp.325–328. United Nations Environment Programme (UNEP), 2013. Minamata convention on mercury. USEPA, 2006. Ultraviolet disinfection guidance manual for the final long term 2 enhanced surface water treatment rule. Environmental Protection, 2(11), pp.1–436. Vilhunen, S., Särkkä, H. & Sillanpää, M., 2009. Ultraviolet light-emitting diodes in water disinfection. Environmental Science and Pollution Research, pp.439–442.  72  Appendix A: KI/KIO3 Actinometry (theory and experimental protocol) Originally Prepared by Jim Bolton and Mihaela Stefan Bolton Photosciences Inc (2011) The composition of KI/KIO3 actinometer is KI, KIO3 and Na2B4O7. The role of Na2B4O7 in this solution is to ensure that the solution has a constant pH of 9.2; the solution will not turn to acidic condition, which can lead to the oxidation of 𝐼−. The principle behind this method is the following photochemical reaction:  8𝐼− + 𝐼𝑂3− + 3𝐻2𝑂 + ℎ𝑣  →  3𝐼3− + 6𝑂𝐻−                        In this chemical reaction, the number of 𝐼3− formed (N) has a linear relationship with the number of photons absorbed by actinometer solution (P) expressed by following formula:  𝑁 = 𝑃 ∗ 𝛷                                                        Φ [moles 𝐼3−/mole photon] is the quantum yield. The number of 𝐼3− formed in this reaction could be inferred from the increase of its concentration (ΔC [mole/L]).  𝑁 = ∆𝐶 ∗ 𝑉                                                    V [L] is the volume of actinometer solution. The concentration of 𝐼3− can be determined spectrophotometrically, and it is proportional to increase of absorbance of actinometer solution at 352 nm, so:  ∆𝐶 =∆𝐴𝐵𝑆𝜀                                                   ε [L ∙ mole-1.cm-1] is the adsorption coefficient and Δ ABS= ABS (before UV exposure) – ABS (after UV exposure). Energy contained in UV light (E) is the product of moles of photon and the photon energy (U), so:  𝐸 = 𝑃(𝑚𝑜𝑙𝑒𝑠) ∗ 𝑈 (𝐽𝑚𝑜𝑙𝑒)                                        73  Assuming that the exposure area is A and the exposure time is t, so the formula for the UV fluence calculation could be expressed as following:  𝑈𝑉 𝐹𝑙𝑢𝑒𝑛𝑐𝑒 (𝑚𝐽𝑐𝑚2) =∆𝐴𝐵𝐶 ∗ 𝑉𝜀 ∗ 𝜑 ∗ 𝐴∗ 𝑈                       And the irradiance (I) could be expressed as: 𝐼 (𝑚𝑊𝑐𝑚2) =∆𝐴𝐵𝐶 ∗ 𝑉𝜀 ∗ 𝜑 ∗ 𝐴 ∗ 𝑡∗ 𝑈                        When using this to calculate the irradiance of a UV source, the reflection of UV light by water should also be taken into consideration. Assuming the reflection factor is α, so the irradiance of UV source (IL) is:  𝐼𝐿 (𝑚𝑊𝑐𝑚2) =∆𝐴𝐵𝐶 ∗ 𝑉𝜀 ∗ 𝜑 ∗ 𝐴 ∗ 𝑡 ∗ 𝛼∗ 𝑈                        Example: for 5.0 mL actinometer solution in a 10 mL beaker (cross-sectional area 3.80 cm2), the absorbance at 352 nm (in a 1cm × 1cm quartz cuvette) before irradiation is found to be 0.021, call this A352(blank). After irradiation for 3.0 min, the absorbance at 352 nm is 0.526, call this A352(sample). The following calculations illustrate how the photon irradiance and the irradiance are calculated: [I-3] = [A352(sample) - A352(blank)]/27,636 = (0.526 – 0.021)/27,636 = 1.827 × 10-5 M moles I-3= [I-3] × V(L) = 1.827 × 10-5 × 0.005 = 9.137 × 10-8 moles  einsteins (moles of photons) = moles I-3 / Φ = 9.137 × 10-8 / 0.60 = 1.523 × 10-7 einsteins photon irradiance (E’P) = einsteins/(area × time) = 1.523 × 10-7 / (3.80 cm2 × 180 s)  = 2.226 × 10-10 einstein s-1 cm-2 74  irradiance (I) = E’P × photon energy at 253.7 nm (U253.7) The irradiance must be corrected for the 2.5% that is reflected from the water surface, so the incident irradiance on the water surface is:  IL = I/0.975 = (2.226 × 10-10 × 471,576)/0.975 J einstein-1 = 1.077 × 10-4 W cm-2 = 0.1077 mW cm-2  The following procedure should be used for the actinometry test.  a. 100mL of the KI/KIO3 actinometry stock solution is prepared by weighing out 9.96g of KI, 2.14 g of KIO3 and 0.381g of sodium tetraborate (Na2B4O7•10H2O) (This generates a solution that is 0.60 M in KI, 0.10 M in KIO3 and 0.01 M in Na2B4O7•10H2O). Dissolve in about 60mL of distilled water and add to a 100mL volumetric flask and make up to 100mL with distilled water. The solution should be made up fresh each time and should not be used after standing for more than 4h. b. Using a caliper (if possible) to measure the internal diameter of a 10 mL beaker and hence calculate the cross-sectional area (Area). c. Measure the absorbance of the actinometry stock solution in a 10 mm pathlength quartz cell at 300nm and 352nm. These values should be approximately 0.58 and 0.02, respectively. Call the latter value A352(blank). d. Measure the irradiance at the center of the beam with the radiometer. The irradiance should be approximately 0.1 – 0.3 mW/cm2 – call this E(before). e. Add 5.0mL of the actinometry stock solution and the 3mm × 12mm Teflon- coated stir bar to a 10mL beaker, place the beaker in the center of the beam at the same position as the radiometer detector head, and raise the platform so that the top of the solution to be irradiated will be at the same level as the reference marker on the radiometer detector head. 75  f. Irradiate for an exposure time of 2.5min (this is for an irradiance of 0.1mW/cm2; adjust this time according your irradiance level) and measure the absorbance at 352nm – call this A352(sample). g. Repeat (f) for exposure times of 2 and 3 times the time exposed in f (e.g. 5.0 min and 7.5 min for the example given). h. Replace the beaker with the radiometer detector; lower the platform to the same level as in (d) and record the meter reading – call this E(after).  i. Calculate the irradiance using the above formula.       76  Appendix B: MATLAB code for calculation of irradiance, fluence rate, and reflection affected fluence rate on a surface of a square and a circle based on radiometer readings: % plotting and Calculating the average fluence rate on a surface of a % square for 275nm UV-LED clear all clc  % radiometr data a=[1.1847   1.81524 2.91788 4.22193 5.6382  6.23363 5.80203 4.85287 3.38785 2.08179 1.3008 2.05138 2.99593 4.91263 7.65673 10.9441 12.8538 11.6195 8.82545 5.56781 3.13487 2.05309 3.40672 5.34025 8.34453 13.8439 21.481  26.3078 23.3899 15.7006 8.90185 5.39523 3.4053 5.17283 8.91322 15.2814 24.5184 40.2177 53.2651 44.6408 26.5156 15.555  9.64333 5.54286 6.87849 12.9703 24.9566 44.4502 65.9837 83.4646 69.5337 47.8126 27.0005 14.53   7.48166 7.84171 15.4175 31.0017 60.0352 89.3852 81.8791 92.5227 67.1083 35.2928 17.5691 8.91575 7.28039 13.9296 27.1317 49.5205 74.7695 98.5369 82.838  58.2453 32.4509 16.5821 8.50318 5.91016 10.1853 17.6485 30.6383 55.9977 72.7847 59.1704 35.6684 21.7464 12.2599 6.77099 3.98363 6.39571 10.8602 20.1261 32.5839 39.3978 34.181  21.6286 12.5654 7.65215 4.69333 2.37668 4.00245 7.4195  12.4077 17.7614 20.0443 17.8054 12.5112 7.58691 4.54232 2.8914 1.58056 2.65289 4.67752 6.99263 9.33494 10.4249 9.31014 6.97725 4.52844 2.62694 1.61793]; dx = 0.5;dy = 0.5;  %cm - Step sizes Z = 1.33; %cm - Distance between LED and plate y_size = size(a,2);x_size = size(a,1); X = -2.5:dx:2.5; Y = -2.5:dy:2.5; [xm,ym]=meshgrid(X,Y); Angle = acot(Z./sqrt(xm.^2+ym.^2)); Fluence = a ./ cos(Angle); figure; surf(xm,ym,Fluence) AveIrr = mean(mean(a)) AvFlu = mean(mean(Fluence))  %% Effect of Reflection n1 = 1; %reflection Index of Air n2 = 1.364; % Reflection Index of Water (Actinometry solution) Thi = Angle; 77  Rs = abs((n1.*cos(Thi)-n2.*sqrt(1-((n1./n2.*sin(Thi)).^2)))./(n1.*cos(Thi)+n2.*sqrt(1-((n1./n2.*sin(Thi)).^2)))).^2; Rp = abs((n1.*sqrt(1-((n1./n2.*sin(Thi)).^2))-n2.*cos(Thi))./(n1.*sqrt(1-((n1./n2.*sin(Thi)).^2))+n2.*cos(Thi))).^2; R = (Rs + Rp)./2; % Reflection Factor Fluence_with_Reflection = Fluence.*(1-R); AvFlu_ref = mean(mean(Fluence_with_Reflection))  %% Inter polation  % Plotting and calculating on a circle surface xxx = -2.5:.01:2.5; yyy = -2.5:.01:2.5; [xxx_mesh,yyy_mesh] = meshgrid(xxx,yyy); temp_Irr = interp2(xm,ym,a,xxx_mesh,yyy_mesh,'spline'); temp_Flu = interp2(xm,ym,Fluence,xxx_mesh,yyy_mesh,'spline'); temp_Flu_Ref = interp2(xm,ym,Fluence_with_Reflection,xxx_mesh,yyy_mesh,'spline'); % figure % surf(xxx_mesh,yyy_mesh,temp_Irr) r = 2.5;  %cm - plate radius r_mat = 0:.01:2.5; Th_mat = 0:1*pi/180:360*pi/180;  [Th_mesh,r_mesh] = meshgrid(Th_mat,r_mat); [Xq,Yq] = pol2cart(Th_mesh,r_mesh); a_Circle = interp2(xxx_mesh,yyy_mesh,temp_Irr,Xq,Yq,'spline'); Area_mesh = pi./180.*1.*(r_mesh.*.01+.01.^2); AverageIrr_Circle = sum(sum(Area_mesh.*a_Circle))/(pi*r^2) r = 2.5;  %cm - plate radius r_mat = 0:.01:2.5; Th_mat = 0:1*pi/180:360*pi/180;  [Th_mesh,r_mesh] = meshgrid(Th_mat,r_mat); [Xq,Yq] = pol2cart(Th_mesh,r_mesh); a_Circle = interp2(xxx_mesh,yyy_mesh,temp_Flu,Xq,Yq,'spline');   figure('Color','white','NumberTitle','off','Name','PolarPlot3d v4.3'); polarplot3d(a_Circle,'PlotType','surfn','PolarGrid',{20 100},'TickSpacing',8,...                   'AngularRange',[0 360]*pi/180,'RadialRange',[0 2.5],...                    'RadLabels',3,'RadLabelLocation',{180 'max'},'RadLabelColor','red'); Area_mesh = pi./180.*1.*(r_mesh.*.01+.01.^2); AverageFlu_Circle = sum(sum(Area_mesh.*a_Circle))/(pi*r^2) r = 2.5;  %cm - plate radius r_mat = 0:.01:2.5; Th_mat = 0:1*pi/180:360*pi/180;  [Th_mesh,r_mesh] = meshgrid(Th_mat,r_mat); [Xq,Yq] = pol2cart(Th_mesh,r_mesh); a_Circle = interp2(xxx_mesh,yyy_mesh,temp_Flu_Ref,Xq,Yq,'spline'); Area_mesh = pi./180.*1.*(r_mesh.*.01+.01.^2); AverageFlu_Ref_Circle = sum(sum(Area_mesh.*a_Circle))/(pi*r^2)  78  Appendix C: MATLAB code for calculation of the average reflection factors  %% The following code is for calculation of average reflection factors for 275nm LED clc clear n1 = 1; %Refraction Index of Air n2 = 1.366; % Refraction Index of Water  Thi = (0:1:62).*(pi/180); Rs = abs((n1.*cos(Thi)-n2.*sqrt(1-((n1./n2.*sin(Thi)).^2)))./(n1.*cos(Thi)+n2.*sqrt(1-((n1./n2.*sin(Thi)).^2)))).^2; Rp = abs((n1.*sqrt(1-((n1./n2.*sin(Thi)).^2))-n2.*cos(Thi))./(n1.*sqrt(1-((n1./n2.*sin(Thi)).^2))+n2.*cos(Thi))).^2; R = (Rs + Rp)./2; % Reflection Factor Av_ref = mean(R)                                       79  Appendix D: UV-LEDs data sheet 1. Nikkiso 265nm  80   81   82   83   84   85   86   87   88   89   90   91  2. LG 275nm  92    93        94    95    96    97    98    99    100    101    102    103    104  3. Nikkiso 285nm  105   106   107   108   109   110   111   112   113   114   115   

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0314125/manifest

Comment

Related Items