Colour in wastewater is normally caused by dissolved matters after the removal of turbidity [9]. The majority of industrial wastes are high in colour intensity, especially those of the textile industries. Dyes contained in wastewater that are not completely removed by treatment and are later discharged into drains are potentially carcinogenic due to their benzidine content and high compositional variability [21]. Off-line monitoring test (sample handling needed) is not suitable for use in this kind of monitoring unit operation at the wastewater treatment plant. The concentration of dyestuffs found in the environment can cause environmental risks due to its inherent eco-toxicology. Thus, reliable analytical methods are needed to detect critical concentrations of dye compounds in the environmental compartment [22].

Chemical Oxygen Demand (COD) is used as a measure of oxygen required to oxidize organic matter of a sample by a strong chemical oxidant. Potassium dichromate, for instance has been widely used for oxidizing a variety of organic substances completely. When the reaction is complete, colourimetric quantification of the residual oxidant is performed [9]. The residue becomes an important parameter and it is used extensively to determine the quality of wastewater quickly (around 2 h), compared with 5 days for a Biological Oxygen Demand test. A number of improvements have been carried out on the digestion process as shown in Table 1. However, most industrial wastewater sample contains chloride ions, aromatic hydrocarbons and pyridine, which are not oxidized unless a catalyst like silver sulphate (Ag₂SO₄) of used [23]. Consequently, secondary hazardous waste is produced and need special disposal.

Biochemical Oxygen Demand (BOD) is one of the widely used parameters to determine the relative oxygen requirements to oxidize the organic matter which present in industrial effluent when seeded with a microbial system [9]. A quantitative relationship for oxygen demand of microorganism to oxidized wastewater sample to carbon dioxide, water and ammonia is shown as an equation [23] below: (1) C n H a O b N c + ( n + a 4 ‐ b 2 ‐ 3 4 c ) O 2 → nCO 2 + ( a 2 ‐ 3 2 c ) H 2 O + cNH 3

Variations of oxygen demand requirement exist, which include using shorter and longer incubation duration, when using respirometric method for measurement of oxygen uptake (units mg O₂/L) [9]. There are a few methods for the determination of BOD, including microbial sensor demonstrated by Riedel et al. [17]. The authorized method for BOD adopted by the American Public Health Association Standard Methods Committee [9] requires time and the use of careful laboratory quality control practices to complete the test, in which cumbersome procedures are involved. As such, it is inadequate for on-line monitoring wastewater treatment unit operations.

Dissolved organic compound in wastewater contains aromatic structures, which have absorption peaks within the UV region [24,25]. Due to these characteristics, some linear relationships are established within them, which include water total organic carbon (TOC), nitrate, suspended solids (SS), COD, BOD and dissolved organic carbon (DOC). The extent of water pollution can be judged directly via the absorbance value after a water sample is irradiated with 254 nm UV light [26]. According to research [27,28], the UV light absorbance at 254 nm has been utilized to directly estimate the aggregate organic content of a water sample. Rapid and versatile monitoring of colour, BOD and COD using optical techniques such as UV-visible spectroscopy and fluorescence measurements was suggested in the research of Bourgeois et al. [29]. Optical properties such as absorption, scattering, re-emission and reflection can be used to rapidly identify the characteristic of wastewater, even in large batches of samples [25].

Essentially, fibre optic sensors (FOS) work like other electrical sensors except that the FOS uses a glass fibre instead of copper wire and light instead of electricity. A FOS modulates the properties of light including intensity, phase, polarization, or wavelength shift [30]. To date, much of the instrumentation improvements in this particular application have been recorded. Different sensor applications require different types of transducers (optical fibres) to manipulate the incident radiation. Thus, instrumentation for optical fibre sensor requires an overview.

A common notation evanescent wave (EW)-based method replaces a portion of the original fibre cladding with a modified cladding material (sensitive material) called taper fibre [52]. This taper fibre creates the evanescent field, resulting a light scattering phenomenon on the optical fibre structure. When radiation losses increase, the transmitted light decreases along the fibre [53]. One popular intensity modulation technique involves bending the fibre to induce radiation losses [36]. Thus, with interaction within fibre and target analyte, the optical absorption result, refractive index change or scattering information can be obtained.

A basic structure of an intrinsic fibre optic sensing part is shown in Figure 5. The total internal reflection (TIR) effect is observed in the resulting attenuation when the condition for critical angle criterion is achieved [54]. Interference between the incident and reflected signals at each point of TIR then generates a wave which extends beyond the optical fibre core. Light propagating along a tapered fibre is not confined to the core region but penetrates into the surrounding cladding region, the so-called penetration depth. The penetration depth can also be derived by Equation (4) [55]: (4) d p = λ 2 π n core [ sin 2 θ − ( n clad n core ) 2 ] 1 / 2where:

d_p is the penetration depth.

Fibre gratings have widespread application in telecommunications. Its shares common advantages with many other FOS relative to electrical sensor technology. Fibre gratings also have their own unique characteristics which include compatibility with multiplexing and distributed sensing capabilities [36]. There are two commonly used techniques associated with the fabrication of fibre gratings, namely, Fibre Bragg Grating (FBG) and Long Period Grating (LPG). A basic Bragg grating based sensor system working principle is shown in Figure 6.

FBG is known to be very susceptible to minute changes in refractive index. The fundamental principle of operation of an FBG sensor system lies in the monitoring of the wavelength shift of the reflected back Bragg signal [58]. These characteristics of fibre grating-based sensor systems are applied in conjunction with chemical sensor applications due to their ability to operate over a broad spectral range (LPG) and even individual gratings [59]. A simple formula for as grating is represented by Equation (5) [60]: (5) λ B = 2 . n eff . Λwhere:

λ_B is the Bragg wavelength.

The length and index of refraction of an optical fibre will change when it is subjected to an environmental perturbation. Interferometric techniques may be used to sense these changes with a high level of measurement sensitivity that matches those achieved with conventional sensor technology [61]. Various types of fibre optic implementation of the two beam interferometer sensor such as the Mach– Zehnder, Michelson; multi beam interferometer sensors such as Sagnac and Fabry–Perot, can be used to perform the detection [36]. Generally, interferometric sensors can provide a sophisticated of sensitivity to the measurement parameters if the technique used is appropriate. There are three major multiplexing arrangements characteristically used, namely Wavelength Division (WDM), Frequency Division (FDM), and Time Division (TDM) multiplexing [32].

Natural and synthetic dyes are extensively used in the food processing, pharmaceutical, textile industries and so on [92]. The wastewater that they generate is a common environmental problem, especially textile waste, which contains high level of dyes. Typically, such wastewaters contain organic compounds with complex structures that are highly soluble and easily hydrolyzed [79]. Approximately 20% to 40% of these dyes remain in the effluent and some are known to be toxic and carcinogenic [93]. Indeed, colour can be characterized by using visible region light source for absorption spectrum analysis. The strong absorption band can be tailored by ring substitution. In fact, in azo groups, structural geometrical changes occur upon absorption of light [72,94]. Colour can be fully characterized by the analysis of the absorption or transmission spectrum by means of optical fibre spectroscopy [77,95].

Spectrophotometry methods with a broad wavelength range have been chosen due to the fact the colour absorption band is highly correlated with the dye concentration in the substrate. This allows the synchronized and expeditious determination of the dyes studied [77]. A single spectral value (635 nm) for monitoring the dyebath was discussed by Sahin and Ulgen [78], and was used for the continuous monitoring system of denim yarns. Thus, choosing the suitable optical source for the specified application is very important. FOS requires data processing of the output signal to extract the desired measurement from the overwhelming amount of data. First stage information based on absorption spectra shape can be used for further classification by using optical spectral analysis. For example, optical spectral analysis accomplished through mathematical methods, such as principal component analysis (PCA) using partial least squares (PLS) algorithm to analyse interferences, usually exhibit overlapped spectra in dye mixtures [77]; or optical spectral analysis using multiple linear regression model to evaluate transmission and absorption spectra of dyes, and automated correction, have been applied in this research to get highly precise values for dye samples [79].

Oxygen demand is an important index for assessing the concentration of organic matter. Concentration of organic compounds can be estimated based on the amount of oxygen used for its complete degradation. Prediction of COD due to actual COD is related to oxidizeable minerals, carbohydrates, biodegradable organic matters and humic substances [28]. Due to the characteristics of aromatic substances and double bonds that exist in their molecular structures, most organic matters have absorption peaks, particularly in the UV region and absorbance at 254–280 nm [96,97].

Wastewater contains turbidity and suspended solid (SS) which can strongly affect the transmitted optical signals [98]. Measurement corrections have to be made using different wavelengths simultaneously. Research by Matsche and Stumwohrer in 1996 [27] showed that by using two different light source wavelengths, 254 nm and 350 nm, in wastewater, biological treatment plants can reduce the effect of light absorption from turbidity and suspended solids. This important finding in ultraviolet spectroscopy for rapid determination COD was then applied in research [80,82].

An alternate approach would be the use of optical techniques to monitor the quality of wastewater. According to investigation by Dornbush and Ryckman [99], 250 nm is a useful wavelength for absorbance measurements which can be applied in analysing organic contaminants.

In addition, according to [100], oxygen demand can be evaluated by using the 280 nm wavelength as absorbance value. Studies have illustrated the high correlation between UV absorption and organic matter in wastewater, Zhao et al. [81] have recently demonstrated a COD monitoring system that uses the UV-Vis spectrometry absorption method. With a wide optical range of 200 to 720 nm, a wider range of COD concentration can be covered, thus, COD concentration can be calculated from 30 to 1,000 mg/L via UV absorbance within an interval of 5 min. The main role of this research is to determine the capability of monitoring COD continuously and its potential use in on-line monitoring for surface waters.

Wu et al. [24] then proposed a similar sensor but with the addition of near infrared (IR) transmission for comparisons to UV absorbance as a method to quantify COD. Hydrocarbons in water can be determined by absorption spectroscopy in the infrared region [101]. In the ultraviolet and visible region, particular organic contaminants can be measured quantitatively, but no qualitative information can be provided. Near infrared (NIR) spectra are capable of providing quantitative as well as some qualitative information. Meanwhile infrared (IR) spectra can grant sufficient information for a distinct identification of the analyte [102]. The NIR method is appropriate for short paths because the light scattering coefficient is larger than that of the UV method in water. NIR transmission illustrates the technical advantages of rapid, highly selective, in situ monitoring [103].

A fluorescence approach for COD sensing was discussed by Hur et al. [3]. In their work, synchronous fluorescence spectra have been studied to estimate the COD concentration of sewer samples. In order to achieve a more sensitive fluorescence-based COD sensor, Hur et al. [82] used a fluorescence excitation-emission matrix with PARAFAC, combined with UV absorption indices, for real time BOD, COD and Total Nitrogen concentration evaluation of urban river waters using 220 nm and 254 nm optical sources. Optical techniques can be advanced by using fibre optics. This advanced technique had been introduced by Fang and Dai [83] for predicting the COD value of waste water. It only took 62 seconds of response time and the detection range from 0 to 350 mg/L was covered.

Contemporary study reviews are more focused on absorption-based and fluorescent-based techniques for BOD fibre sensor application [28]. The absorption of excited and emitted wavelength by the sample can result in a fluorescence intensity attenuation known as the inner filtering effect. Inner filter effects are caused by the existence of other chromophores in the solution, which absorb in the same wavelength as water samples [104]. The common solution to minimize the effect is to apply some form of mathematical correction during the calculations for the diluted samples [88], most often based on the measured absorbance of the sample [105] or fluorescence intensity multiplied by a correction factor [106].

Larsson et al. [107] and Ahmad et al. [84] suggested using Raman scatter peaks to compensate for the attenuation of the incident excitation radiation. Henderson et al. [104] utilized the relationship between concentration and fluorescent intensity to relate a non-linear effect with wavelength and left the data uncorrected. Special attention and correction should be paid towards the fluorescence spectra, even though altering the pH of the original sample is not recommended [108]. As observed in the research by Reynolds et al. [109], fluorescence intensities increase with pH until 10 for raw sewage samples. According to the research of Patel-Sorentino et al. [110], similar results were obtained for fluorescence intensity, which increased to 10 due to the structure of humic substances that changed with the modifications in their environment. Most problems arise at highly concentrations of contaminant samples, in such case, filtration, dilution and absorption tests out for inner filter effect are suggested.

Despite this usefulness as a monitoring tool, however, fluorescence studies have simply relied on “sing e peak i king” to quantify the desired fluorescent components. Synchronous fluorescence can be used by fixing the excitation and emission wavelengths uniformly to get the spectra from a scan of the entire section of the excitation wavelength. This is an advanced way to obtain a better resolution of fluorescence spectra. This is because it can capture all fluorophores contained in the samples [111].

Fluorescence-based techniques with biosensors can be dividing into two general types: sandwich assays and direct methods. For the sandwich assay method, a primary immobilized cell is bonded to the surface of the fibre. The presence of target bio-molecule that bind (fluoresce) with the immobilized cell can then be excited by incident light and emission fluorescent [62]. Relations between fluorescence properties and natural waters were reported as early as 1910 by Dienart [112]. Preliminary results on the fluorescence and scattering properties of sewage have been established as well in 1993 [113]. In addition, the total integrated fluorescence intensity contributed by suspended material, the dissolved biodegradable or non-biodegradable organic matter can be correlated to the changes in BOD of water samples [84]. This technique used in BOD detection was demonstrated by Ahmad et al. [90] and Hur et al. [82]. A scheme of BOD biosensor is shown in Figure 8.

A BOD optical biosensor is generally designed with a respirometric and microbial type probe. Karube et al. [114] used Pseudomonas putida, and Riedel et al. [17] used Trichosporon cutaneum cells as the isolated bacterium for BOD detection. However, this probe has its limitations; oxygen consumption during the detection process will degrade the electrode. Electrochemical sensors based on the Clark-type oxygen electrode have been widely developed as the transducer of BOD sensors. Conceivably, the electrode used for oxygen measurement can be replaced by an optical (fluorescent) sensor. The major benefit of optical fibres over electrodes in the case of BOD measurements is that there is no oxygen consumption during measurement. Thus, no depletion of oxygen can occur [85] and it also does not require the frequent replacement of electrodes.

In addition, the direct method uses a fluorescent dye directly to measure the fluorescence quenching, where the reactive dye can react with specific compounds [62]. When used with an external fibre optic probe, the method takes advantage of ruthenium complexes. Ruthenium complexes are an attractive class of compounds for oxygen-stimulated fluorescence quenching in sensing applications. This is due to their high photochemical stability, high extinction coefficients, and absorption spectra the matches with the emission spectrum, for example, blue LED [115]. As a result, this fluorophore was immobilized in a photopolymerized hydrogel made from poly(ethylene glycol) diacrylate (PEGDA), [116] and was investigated for luminescence based tests suing fibre optic oxygen sensors [87].

A microbial system should offer low selectivity, high assimilability and high bio-oxidation activity for a broad range of organic compounds [114]. Due to the fact each microbial species has its metabolic deficiencies, the universality of BOD sensor is restricted. As a result, it is essential to select an appropriate microorganism for a biosensor [117]. The first fibre optic microbial sensor for determination of BOD was demonstrated by Wolfbels et al. [85] with a fluorescence-based technique. A simple description had been shown as Figure 9 below.

Trichosporon cutaneum cells immobilized in poly(vinyl alcohol) offer a biocompatible microenvironment for microorganisms and allow prompt response (diffusion) of organic compounds. Besides, tris(4,7-diphenyl-l,l-O-phenanthroline) ruthenium(II) perchlorate is used as the oxygen indicator. Excitation light (from the bottom) passes through the polyester support and excites fluorescence in the oxygen sensitive layer. Then, the emitted light is collected and guided by the fibre bundle to the photo detector. This research showed that estimation of BOD can be made rapidly, in about 5 to 10 min. It is possible to use this probe in situ for sewage plant effluents and municipal sewage monitoring, however, the short lifespan of the microbes and the thickness of cells are limitations that have to be overcome.

The strains were also studied by Karube et al. [86] using a conventional microorganism, Pseudomonas putida. This research demonstrates that there is a requirement for a maintenance-free microbial BOD sensor that has long term stability and is not affected by heavy metals. Nevertheless, the reproducibility of BOD sensors based on particular cells or activated sludge is often poor. Consequently, mixed strains on microorganisms immobilized within a single sensor were developed [117]. Kwok et al. [88] modified this sensor by immobilizing activated sludge cum Bacillus subtilis on oxygen sensing films and placing it underneath the sample vials. In an extensive study, Chen et al. [89] applied multiple microorganisms—B.licheniformis, D. maris and M. marinus—into a single microbial BOD biosensor.