Silicon (Si), glass, and quartz are inorganic materials. Silicon is one of the earliest materials used to fabricate microfluidic chips because of its semiconductor properties, easy modification surface, and the availability of the technology (Nielsen et al., 2020). Inorganic materials possess advantages such as high stability, thermal and electrical conductivity, and solvent compatibility. Nonetheless, inorganic materials have several drawbacks. For example, Si has weak optical properties. Some studies showed that Si-based chips present a deficiency in transmittance in the visible light region but not in the infrared (IR) region. However, glass and quartz are reusable, biocompatible with biological samples, and have good optical properties. As such, the problem of transparent Si materials can be solved by making a hybrid material between glass and Si (Aralekallu et al., 2023).

A diversity of organic polymers are used to design microfluidic chips. Polymers are categorized into thermoplastic polymers, including PMMA, COC, COP, and PS; cured polymers, including PDMS; and volatile solvent polymers, including fluoroplastics. Thermoplastic polymers are rigid due to their structural arrangement, which makes them more resistant to changes in temperature and pressure (Damiati et al., 2022a). Topas and Zeonor demonstrate high efficiencies in their optical characteristics and are compatible with certain organic solvents. PDMS is a hybrid material between silicon and polymer that shows good transparent properties for optical detection methods (Aralekallu et al., 2023). PDMS and PMMA are the main polymers used for environmental monitoring research. PS is used for biological research, such as cell culture. In general, the fabrication processing of polymers is low-cost and more environmentally friendly without the need for a hazardous reagent, and they have good optical transmittance properties and elasticity. The main drawback of polymers is their incompatibility with many organic solvents, which affects device integrity and the reaction or analysis that operates inside a chip. For example, PDMS is highly compatible with commonly used solvents (e.g., chloroform, xylene, and ether) but is susceptible to swelling or chemical attack upon exposure to some acids and bases or nonpolar solvents (e.g., hydrocarbons, toluene, and dichloromethane) (Lee et al., 2003; Vinothkumar et al., 2011). PMMA, PC, and PS are highly compatible with alcohols but not with some organic solvents such as ketones and hydrocarbons. Furthermore, COC is a good choice for microfluidics due to its high resistance to acids and bases (e.g., hydrogen chloride, sulfuric acid, nitric acid, sodium hydroxide, and ammonia) and most organic polar solvents (e.g., acetone, isopropyl alcohol, and methanol). Overall, microfluidic materials are essential parts of the fabrication of devices because they can affect the precision of fluid behavior in microchannels.

Paper-based microfluidic materials represent a good alternative to organic and inorganic materials, especially the latter, which are expensive due to the technology processing techniques. Paper is a sheet containing a compressed hydrophobic/hydrophilic porous membrane that can be made from cellulose fiber or nitrocellulose to control the movement flow of a fluid on the paper via the capillary effect (Jin et al., 2022). This type of material is inspired by the traditional detection technique of paper chromatography (Wu et al., 2022) and has several advantages, such as being easy to modify and functionalize to enhance mechanical strength and conductivity; easily being able to place chemicals on paper; allowing the movement of fluids without needing pump pressure or an external pump; the easy visualization of an analyte’s color development when using colorimetry methods; and low-cost processing because there is no need for harsh fabrication conditions or a closed, clean room as it required for inorganic and organic materials (Ghaseminasab et al., 2023). The main drawback is the low specificity, which is unsuitable for volatile samples, surfactants, and organic solvents. This type of material has garnered more recent interest in the research area of water quality monitoring (Nishat et al., 2021).

A paper-based analytical device (µ-PAD) is a microfluidic device that is considered one of the lowest-cost and fastest analytical techniques. Xiong et al. (2022) reported on the development of a µ-PAD-based microfluidic device with smartphone app integration for the simultaneous detection of Cu(II), Ni(II), Fe(III), nitrite, and pH (Figure 3A). The device depends on the colorimetric detection method in which the color intensity is proportional to the analyte concentration. The working principle of this system is to measure the color intensity released from a chromogenic reaction that occurs between the reagents, bathocuproine, dimethylglyoxime (DMG), phenanthroline, Griess reagent, and bromothymol blue (BTB), which are added to certain zones in the µ-PAD and the target analytes, Cu(II), Ni(II), Fe(III), nitrite, and pH, respectively. Then, the integration app with the smartphone camera analyzes the captured picture to record the concentrations of the analytes based on the color intensity of each zone. The detection limits and linear ranges of the device were 0.4 ppm and 3.8–400 ppm, 1.9 ppm and 2.9–1,000 ppm, 2.9 ppm and 2.8–500 ppm, 1.1 ppm and 2.3–90 ppm and 5-9 for Cu(II), Ni(II), Fe(III), nitrite, and pH respectively. The recoveries and RSD % ranges were 97.9%–98.4% and 3.12%–4.35%, 98.2%–105.1% and 3.23%–4.34%, 104%–107.6% and 1.38%–2.65%, and 96.7%–106.4% and 3.05%–3.12% for Cu(II), Ni(II), Fe(III), and nitrite respectively. They compared their obtained results of spiked samples with those of inductively coupled plasm–mass spectrometry (ICP–MS) as a standard method for validation were relative standard deviation (RSD) was less than 5%. The smartphone-app-integrated μ-PAD detection system has excellent compatibility and high accuracy and reliability without a significant difference from the ICP–MS method.

In addition, the same type of µ-PAD and reagents were used for the microfluidic-based colorimetric detection of Cu(II), Ni(II), and Fe(III) in water in a study conducted by Aryal et al. (2023). They modified the detection process technique using capillary flow to drive the media in the microfluidic system, without redundantly pipetting the samples into the paper to enhance the color intensity and reduce the detection time to 8 s. Moreover, they used ImageJ software to measure the color intensity of the captured picture by reading the average grey scale. Based on their results, they obtained nearly the same detection limits as the previous study for Cu(II) 0.3 ppm and Ni(II) 2 ppm but a lower one for Fe(III) 1.1 ppm. The quantification limit were 1, 6.67, and 3.67 ppm for Cu(II), Ni(II), and Fe(III) respectively. They achieved good recovery between 80%–110% in more than 90% of real sample matrices from drinking water and rivers with acceptable range of precision and accuracy less than 15% RSD (Figure 3B).

Manbohi and Ahmadi (2022) also successfully executed the quantitative detection of inorganic nutrients, designing µ-PAD microfluidic device integrated with a smartphone app for the simultaneous detection of phosphate, nitrite, and silicate in coastal water. Nutrients are widely used in agriculture for plant growth. They can be carried to waterbodies via rainfall, so simultaneously monitoring the nutrients in waterbodies is crucial to avoid the overgrowth of aquatic plants and eutrophication. The researchers' device can be used for onsite monitoring and is based on the principle of colorimetric detection, which relates the intensities of RGB components in the captured picture to the concentrations of the analytes. The detection limits were 1.52, 0.61, and 3.74 μg/L for phosphate, nitrite, and silicate, respectively. The linear ranges were between 5 and 100 μg/L for phosphate (R ² = 0.9909), 5 and 100 μg/L for nitrite (R ² = 0.9819), and 10 and 600 μg/L for silicate (R ² = 0.9933). Their sensor exhibits good agreement compared with a spectrophotometric method using a real coastal water sample, where the recovery percentages were 92%–108%, 95%–105%, and 94%–103% for phosphate, nitrite, and silicate, respectively. Furthermore, it shows advantages in the total analysis time less than 6 min and generating real-time maps and sharing the results via a network.

Hydrazine (N₂H₄) is an inorganic compound widely used in industries and has a carcinogenic effect on human health. Humans may be exposed to it by drinking contaminated water. Hence (Ghaseminasab et al., 2023), developed a portable tool for detecting hydrazine that does not require pretreatment or reagents and is based on a microfluidic paper-based colorimetric device (µ-PCD). Furthermore, it requires the naked-eye colorimetric monitoring of N₂H₄. The designed µ-PCD is fabricated from paraffin as it is an inexpensive material and modified with silver nanomaterials (silver nanoprisms (AgNPrs)), silver nanowires (AgNWs), and silver citrate (AgCit)). In this portable device, the AgNWs and AgNPrs show outstanding colorimetric results for N₂H₄ detection among a linear range of 0.08–6 M and 0.02–5 M respectively, as well as lower limit of detection 800 μM and 200 µM were stable for 90 and 60 days, respectively (Figure 3C). Generally, the reported studies that depended on paper-based microfluidic chip materials showed the simplicity of the detection procedure and could identify the contaminant primarily with the naked eye. However, to know the actual concentration, they integrated with a mobile app that needed an efficient algorithm.

Microfluidic devices can be made from different materials to enhance the properties of detections. Tazawa et al. (2023) introduced a microfluidic device using a glass-molding process for the onsite measurement of residual chlorine in tap water. The chip was fabricated from a glass substrate and then coated with diamond-like carbon (DLC) to prevent the mineral’s adsorption effect (Figure 4A). The device can detect residual chlorine in the range of 0.1–1 ppm. In addition (Phuong et al., 2023), fabricated a sensitive device for cyanide monitoring in water based on the hybrid coating nanomaterial of Au_core–Ag_shell. The fiber optical sensor was clad with cyanide and formed a cyano–metal complex that led to a change in the refractive index of the localized surface plasmon resonance peak (Figure 4B). This method is more environmentally friendly due to the low toxicity of the nanomaterial compared with the reagent used in the traditional detection method. They reported the lowest detection method was 8 × 10⁻¹¹ M over the concentration range of CN⁻ between 0–150 µM. In general, microfluidic chips with different materials and integration methods show a significant advantage in detecting various chemical contaminants that can pose serious risks to human health and the environment.

Meanwhile, microfluidic devices have made a breakthrough in the field of in situ sensors due to the low consumption volume of samples and reagents, which can lead to analyzing a maximum number of samples. Some researchers have tried to build more advanced in situ sensors automated without human intervention. Sonnichsen et al. (2023) proposed an automated LOC device for analyzing total alkalinity in situ in seawater (Figure 4C). The developed microfluidic device enabled fluid mixing, chemical reactions, and optical detection on a single platform. The device demonstrated satisfactory results of the analysis in the 5–25°C range, and a 1 L acid/indicator bag with a 500 mL standard bag was enough to analyze 532 samples and 62 standard measurements due to the advantage of low reagent consumption in microfluidics. Moreover, this device provided instant and permanent results of the seawater’s total alkalinity. Total alkalinity is an efficiency indicator of ocean alkalinity enhancement (OAE) for the capture and removal of CO₂ from the air and dissolved in the ocean by being converted to bicarbonate, which can reduce the effect of CO₂ and decrease the crisis of increasing temperatures and climate change impacts. However, the main problem is if the ocean is acidified due to saturation with bicarbonate, which can affect the ability to capture CO_2, so monitoring the total alkalinity is very important to take quick action to reduce the acidification influence by enhancing the ocean’s alkalinity.

Another study by Morgan et al. (2022) reported on a fully automated phosphate analyzer based on an inlaid microfluidics absorbance cell. The microfluidic chip fabricated from polymethyl methacrylate (PMMA) consists of a syringe pump and solenoid valves to aid in automated fluid control. Their developed device achieved a 15.2 nM detection limit and a 50.8 nM quantification limit over a dynamic range of 0.2–10 µM. Also, the sensor shows high in situ measurements underwater with (<1.5%) RSD. However, these types of automated sensors need highly expert scientists to build a good architecture sensor. Both studies presented excellent models with high efficiencies.

The advantages of using microfluidic chips in the detection of inorganic contaminants in water can be applied to the detection of organic contaminants. Per- and poly-fluoroalkyl substances (PFASs) and microplastics are new analytes that the EPA considers toxic to human health. PFASs are a large group of organic substances, including perfluorooctanesulfonate (PFOS). A recent study showed that approximately 45% of PFASs detected in US drinking water and the substances humans are most exposed to are PFOA and PFOS (Smalling et al., 2023). Rapid detection sensors are necessary to protect human health. The microfluidics technique could keep up with new types of contaminants. Cheng et al. (2020) present microfluidic impedance sensors based on a metal–organic framework (MOF) for PFOS analysis (Figure 5A). The principle of the sensor is based on the design of a receptor probe made from the MOF along with a microelectrode in a microfluidic channel through which the fluid passes to capture PFOSs. They present the advantages of the flow technique to overcome the diffusion resistance of the solution and allow for rapid measurement with a detection limit 0.5 ng/L. Zhang et al. (2023) produced a microfluidic device for integrated sample processing and counting for microplastic analysis. The device shows simplicity in analysis and reduces costs compared with the traditional technique. The method is semi-automated to perform the digestion, filtration, and staining of microplastics in the river water sample inside the microfluidic chip, which is fabricated from a double layer of PMMA. The microplastics are counted and detected with a fluorescence microscope, and the results are then processed with a software video (Figure 5B). In addition (Gong et al., 2023), developed a model that combined a microfluidic chip and machine learning to identify tiny-sized (smaller than 50 µm) microplastics in seawater. The chip made from PDMS was used to trap a tiny pristine microplastic particle to overcome the frequent overlapping of peaks encountered in direct analysis using Raman spectroscopy (Figure 5C). Moreover, using a microfluidic chip can trap a single particle and increase the accuracy of detection and identification with real samples. This developed model is much more complicated when combined with different machine learning models compared with the previous study, but it can save a lot of time in identifying new types of microplastics.

Some important organic pollutants need rapid treatment action due to their effects. COD is an important water index that indicates the degree of organic species in water that can compete with organisms for oxygen and the efficiency of plant treatment. Dichromate or potassium permanganate digestion is the traditional detection method used in the laboratory. It harms the environment due to the reagent’s toxicity and the volume used for each sample.

Li et al. (2022) present a COD detection technology using a t-structure microfluidic chip based on the ozone chemiluminescence detection method principle, offering sensitivity, rapidity, and no formation of byproduct pollutants (Figure 5D). This method shows COD detection results consistent with those obtained using the traditional potassium dichromate method with an average deviation of less than 5%. In addition (Jiang et al., 2020), developed a multiplexed chip for simultaneously detecting five parameters for monitoring water quality, including COD. The method is based on the color reduction of permanganate from purple to green, followed by analyzing the color intensity of the captured picture with a smartphone using MATLAB (Figure 5E). This device represents a quantification range for COD-low concentration (1–50) mg/L and COD-high concentration (50–250) mg/L. The percentage error of the analysis using the method device compared with a standard method was 3% for the sample from river and 1% for the sample from industrial waste. Compared with the previous study, this chip demonstrates the simplicity of fabricating the device, and the small amount of reagent required reduces the amount of chemical waste produced via the reactions.

Pesticides are also considered a type of organic contamination. They are introduced into waterbodies via agricultural and non-agricultural practices, and some types pose cariogenic risks to human health. Fernández-Ramos et al. (2020) developed a microfluidic paper based on the optical detection of organophosphate pesticides and carbamate. The determination process was based on inhibiting the catalytic activity of acetylcholine esterase (AChE) in the presence of organophosphorus and carbamate, which can be hydrolyzed with acetylcholine chloride (AChCl), leading to color changes (Figure 6A). Then, ImageJ software was used to calculate the intensity of the captured picture with a camera. The microfluidic chip used was a double µ-PAD, which helped separate the channel reagent path added for the detection process to avoid the direct contact of the substrate with the enzyme before adding the samples. This device showed a good recovery percentage with real samples between 97.7% and 102.3% with a detection limit of 0.24 μg/L for carbaryl and 2.00 for chlorpyrifos. Also, with reproducibility between 4.2%–5.5%. In addition, Moulahoum (2023) reported a method for detecting atrazine, a pesticide used for crops, based on fabricating a biosensor from a laser-printed µ-PAD chip in a Y shape using gold and silver nanoparticles as reagents for colorimetric detection (Figure 6B). The µ-PAD chip was fabricated useing a laser-printed to reduce the devices cost compared to fabrication with wax or inkjet printer. The color sensing was based on the aggregation of metallic nanoparticles due to the ligand exchange with the pesticide. The captured picture of the color signal was analyzed using ImageJ software. The detection limit was 3.5 µM using AgNPs and 10.9 µM using AuNPs. The recovery percentage of the spiked water samples with atrazine was in good agreement with the spiked concentration between 100.2% and 103.4% for AgNPs-based sensor and 117%–122% for AuNPs-based sensor. Both studies show the simplicity of the devices and detection. Nevertheless, the pesticides consist of different species types that can interfere. Table 1 shows different microfluidic platforms to detect chemical contaminants in water using various detection methods.