This application claims priority from US Patent Application Number 63107025, filed on October 29, 2020, entitled “A System for Estimating the presence of Microplastics in Liquids” the contents of which are hereby incorporated by reference in its entirety.

BACKGROUND OF THE INVENTION

The present application relates to a system and method for detecting the presence of microplastics in liquids.

Microplastics are an emerging contaminant and the need to detect their presence, particularly in water, is becoming increasingly more important.

The present system seeks to address this need.

SUMMARY OF THE INVENTION

According to a first example embodiment there is provided a system for detecting the presence of microplastics in liquids, the system comprising: a container for holding a liquid sample; a light source for illuminating the liquid and any particles contained in the liquid within the container with a selected illumination wavelength or warmth of light; a light receiver for receiving light reflected and/or emitted from the liquid and particles contained in the liquid at a receiving wavelength greater than or equal to the illumination wavelength or greater than or equal to the illumination warmth; a processor programmed to: access signal data received by the light receiver; and analyze the signal data to determine the presence of microplastics in the liquid sample.

The illumination wavelength may be between 370-400 nm.

The illumination wavelength may be between 435-460nm.

The illumination wavelength may be between 780-880 nm.

In one example, the illumination wavelength is supplemented by light greater than or equal to a Kelvin temperature of 3500.

The liquid sample and contents in the container are not altered by the addition of dyes or acids prior to illumination, neither are the liquid sample and contents in the container dried prior to illumination.

In one example, the light source sequentially illuminates the liquid and contents with a plurality of selected illumination wavelengths and/or warmth of light.

The sequential illumination wavelengths may include 370-400nm, 435- 460nm, and/or 780-880 nm.

The container, light source, light receiver and processor may be enclosed in a liquid resistant housing with a liquid inlet pipe through which liquid can pass into the container and a liquid outlet pipe through which liquid can pass out of the container and wherein only the liquid inlet pipe and the liquid outlet pipe are open to liquid outside the housing to allow the liquid to pass into and out of the housing.

The light source is preferably one or more Light Emitting Diodes (LEDs).

The light receiver is preferably a camera which takes an image of the illuminated liquid and particles contained in the liquid.

The data may be stored locally or exported to an external device.

The data may also be analyzed using artificial intelligence (Al), specifically machine learning (ML), as the method of characterization.

It will be appreciated that the innate autofluorescence of the particles is the primary indicator utilized by the system.

According to a first example embodiment there is provided a method for detecting the presence of microplastics in liquids, the method including: placing a liquid sample in a container; illuminating the liquid and any particles contained in the liquid within the container with a selected illumination wavelength or greater than or equal to the illumination warmth of light; receiving light reflected and/or emitted from the liquid and particles contained in the liquid at a receiving wavelength greater than or equal to the illumination wavelength or warmth; access signal data received by the light receiver; and analyze the signal data to determine the presence of microplastics in the liquid sample. BRIEF DESCRIPTION OF THE DRAWINGS

Figure 1 is a block diagram illustrating the various components of a system for detecting the presence of microplastics in liquids;

Figure 2 shows a first example embodiment of a system incorporating the components of Figure 1 ;

Figure 3 shows a second example embodiment of a system incorporating the components of Figure 1 ;

Figure 4 shows a third example embodiment of a system incorporating the components of Figure 1 ; and

Figure 5 shows the third example embodiment of Figure 4 enclosed in a housing.

DESCRIPTION OF EMBODIMENTS

Referring to the accompanying drawings, a system for detecting the presence of microplastics in liquids is illustrated.

Microplastics are defined as plastic particles less than 5mm along their longest dimension.

The system 10 for detecting the presence of microplastics in liquids includes a container 12 for holding a liquid sample.

A light source 14 is used for illuminating the liquid and any particles contained in the liquid within the container 12 with a selected illumination wavelength or a selected warmth of light. ln one example, the light source 14 is one or more Light Emitting Diodes (LEDs).

A light receiver 16 is used for receiving light reflected and/or emitted from the liquid and particles contained in the liquid at a receiving wavelength greater than or equal to the illumination wavelength or greater than or equal to the illumination warmth.

The light receiver 16 in one example is a camera which takes an image of the illuminated liquid and particles contained in the liquid.

A processor 18 is programmed to control the system. The processor will have software executing thereon to access signal data, typically in the form of images, received by the light receiver and analyze the signal data to determine the presence of microplastics in the liquid sample. The processor may take the form of at least one microprocessor and/or microcontroller.

In the illustrated examples below, the processor 18 does the image processing but it will be appreciated that this could be done remotely whereby the processor described includes a processor located remotely from the remainder of the components. In this example, the light receiver data (typically camera images) is transmitted to the remote processor.

A memory 22 is used for storing the data required to be used by the system.

In one example the processor 18 is comprised of an Arduino Uno microprocessor together with a microcontroller to control the system. The Arduino Uno is an open-source microcontroller board developed by Arduino™.

Microprocessors can be used to control electronic components that include timing and running code in a loop in order to execute a list of tasks repetitively. Additionally, Arduinos and other microprocessors can be controlled by other microcontrollers with code (e.g. Python) embedded in the microcontroller.

The prototype system used a camera connected via USB as the light receiver 16, which camera could not be controlled by the Arduino microprocessor directly.

Rather, the camera and the Arduino are controlled through a microcontroller with the ability to run a Python script. Python is used to send signals to many devices all in one script, solving the problem of communication between the Arduino microprocessor and the camera used.

Additionally, microprocessors have memory components attached separately. Microcontrollers also have the ability to store data which is needed to retain testing images and a master script. The Python script on the microcontroller sends a signal to the Arduino microprocessor to execute a task. The Arduino microprocessor then controls the other electrical components of the device, for example to make the LEDs turn on and off and to make the valves (described below) open and close.

Once executed, the Arduino microprocessor sends a signal back to the microcontroller indicating it is ready for the next command.

Once the Python code receives positive feedback from the Arduino that the LEDs were excited, the next step would be to capture the image.

Rather than send a signal to the Arduino, the Python script sends a signal to the camera through the microcontroller to capture and save the image in the memory 22.

Once the image is captured and saved, the Python script continues communicating with the Arduino. It will be appreciated that there are many other microprocessors available which could be integrated into the system in place of the Arduino microprocessor and microcontroller described above.

The components of the system described above can be put together in a number of different configurations.

For purposes of illustration, the system will be described with reference to the liquid being water, which is one application of the system described, and the system is used to detect the presence of microplastics in the liquid, in this case water.

A first example embodiment is illustrated in Figure 2.

In this embodiment, the container 12 is a cuvette which can be removed from the housing 28 in order to be filled with water. A cap is placed on the cuvette and it is inserted into the housing 28.

It is envisaged that this embodiment could be used by citizen scientists as it features a simple manual sample collection via the cuvettes.

The cuvette 12 would then be inserted into the housing 28 which consists of a light-tight box where LEDs 14 are used to excite the water sample.

The water sample can then be rapidly analyzed, and the results are displayed to the user via the display 26.

A second example embodiment is illustrated in Figure 3 where the container 12 is a glass or see-through tube.

This device includes an inlet pipe 30 with an opening 32 via which water can enter into the device and pass through the pipe into the container 12. An outlet pipe 34 allows water to pass from the container 12 and out of an opening 36.

The goal of this embodiment is to passively detect microplastics. Chambers could be added to this embodiment to control the flow of water through the device allowing for sedimentation to occur.

It is envisaged that in this embodiment the device would be anchored in a body of water with continuous flow to drive water into the container 12.

Filters will be placed inside the inlet pipe 30 to filter out any coarse sediments. These filters will need to be cleaned periodically.

In this example embodiment, a light source in the form of LEDs 14 are placed next to the container 12 to illuminate the contents thereof.

Figure 4 illustrates a further embodiment in which the inlet pipe 30 bifurcates and directs water towards two containers 12a and 12b.

It will be appreciated that this doubles the number of samples that can be tested in deployment.

Light sources 14 are located alongside each of the containers 12 to illuminate the contents thereof.

The containers 12 used in this example are UV-B transparent glass tubes.

This embodiment also includes valves 38 which control the flow of water into the containers 12. The valves are typically solenoid valves.

The valves 38 are placed in the liquid inlet pipe and the liquid outlet pipe adjacent to the containers, and are opened and closed to allow liquid to pass into the containers and then closed to keep liquid in the containers. The opening and closing of the valves is controlled by the processor 18. In this way the two containers 12 are controlled as isolated units using the valves 38.

The valves 38 will isolate the container 12a or 12b from the remainder of the system, allowing samples to be taken from the flow of water through the system.

To collect samples, first the valve closest to outlet 36 adjacent one of the containers 12a and 12b is closed, and then after a delay to allow the container 12a or 12b to fill with water, the valve closest to inlet 32 is closed.

In this example, when testing is being conducted in one branch, the other testing chamber remains open allowing for continuous flow through the system.

This is done to prevent contaminants from accumulating outside the testing chamber when the valves to the testing chamber are closed.

The alternation of testing chambers also allows for testing periods intermixed with continuous flow periods thereby allowing for previous sample remains to be flushed from the testing chamber 12 and the LEDs 14 time to cool after use.

Additionally, using one chamber at a time draws less power, potentially reducing the risk of electrical failure.

Figure 5 illustrates the components from Figure 4 enclosed in a housing 40 which has a base 40a and a lid 40b connected together via hinges 42.

Located in the base of the housing are the mechanical components described above as well as the electronic components referred to in Figure 1. The housing 40 is a liquid resistant housing with only the liquid inlet pipe 30 and the liquid outlet pipe 34 open to liquid outside the housing to allow the liquid to pass into and out of the housing.

The hinged lid 40b could easily be lifted by maintenance personnel when necessary and then easily resealed using clamps to apply pressure between the device lid and a rubber liner below the lid, and the base.

It will be appreciated that inside the housing, the UV-B tubing 12 and the LEDs 14, are thereby encompassed by a light tight box to ensure the highest quality images and mitigate user and environmental safety hazards from exposure to UV light.

On the outside of the housing 40, typically beneath the base 40b is an attachment point for an anchoring mechanism. The anchor allows the housing to remain in the center of flow, preventing drift, and making the housing easy to retrieve.

The anchor can be detached from the housing for transport purposes. Before deploying the housing, the anchor would need to connected to the base of the housing. A rope drum can also be used for the anchor rope to be tied off. This removeable anchor system increases portability.

As mentioned above, the inlet pipe 30 typically has a filter, such as a mesh filter, located inside the pipe to prevent debris from entering the system.

The housing 40 exterior is designed to complement the interior. Preventing filters from clogging is an important design consideration as during deployment filters may become clogged due to debris larger than the mesh size accumulating on the surface. In order to mitigate this, the system 10 was designed such that water could flow though the housing 40 in both directions.

After a set time period, rudder flaps (not shown) controlled by processor 18, and modelled after kayak rudders, would be lowered from the base 40b by a solenoid to cause the housing 40 to rotate in the water flow about the fixed anchor point.

It will be appreciated that as the housing 40 was designed with rotation in mind, the housing 40 was engineered to be symmetrical about all axes.

Internally, the housing 40 is symmetrical as well to maintain a symmetrical mass distribution for rotation.

For extended periods of deployment, the housing is programmed to rotate clockwise and then counter-clockwise to prevent from over-rotating the anchor.

The operation of the illustrated in Figures 4 and 5 will now be described in more detail.

While this embodiment has a different physical construct to the embodiments described in Figures 2 and 3, it will be appreciated that the essential principles to detect microplastics in the water will be the same even where there are some differences due to the physical construct of the system.

Firstly, unlike prior art systems, the liquid sample and contents in the container are not altered by the addition of dyes or acids prior to illumination and certainly do not need to be dried prior to illumination.

This is very advantageous to allow the system to be easily be used even by a layperson at the site where the water sample has been collected.

Prototypes of the system described above used a plurality of LEDs as the light source 14.

The plurality of LEDs included one or more of white light LEDs (with a temperature greater than or equal to a Kelvin temperature of 3500), LEDs with an illumination wavelength of between 370-400 nm, LEDs with an illumination wavelength between 435-460nm and LEDs with an illumination wavelength between 780-880 nm.

All of the LEDs selected to be used in a particular application are controlled by the processor 18 to illuminate the sample liquid and particles contained therein.

In one example, the light sources sequentially illuminate the liquid and contents with a plurality of selected illumination wavelengths and/or colour intensities of light.

For example, the sequential illumination wavelengths include at least some of 370-400nm, 435-460nm, and 780-880 nm.

Using one of the embodiments described above, a water sample is entered into the testing chamber 12.

First, the sample is photobleached with UV light to mitigate biological fluorescence interference in the images [1]. During the wait time of photobleaching, sediments suspended in the water column would be able to settle.

Colloidal particulates in the water sample may not settle out but are likely too small to be detected by the camera 14 and are not of concern for interference with microplastic detection.

Next, the white light LEDs, then 385 nm LEDs, and lastly 448 nm LEDs are activated sequentially. It will be appreciated that the exact sequence of the lights could be varied.

An image is captured and saved during each excitation. Once all images are captured, depending on the embodiment in use, the solenoid valves 38 open, releasing the sample and the testing process would then begin on the opposite branch or a new sample can be inserted. As described above, in one example the processor 18 is implemented using a combination of a microprocessor connected to a microcontroller executing a Python script, for example.

The Python script would place a heavy importance on timing and would vary with each embodiment. The functions of the main loop of code would proceed as outlined for the third embodiment.

Wait times are dependent on deployment location and would need to be programmed into the system prior to use.

For time the device is in operation functions could include:

1 . Main Loop a. Close test chamber exit valve b. Pause to fill c. Close test chamber entrance valve d. Pause for settling and photobleaching e. Light white LED f. Take white light picture g. Save white light picture h. Turn off white light LED i. Turn on 385 nm LED j. Take 385 nm photo k. Save 385 nm photo l. Turn off 385 nm LED m. Light 448 nm LED n. Take 448 nm picture o. Save 448 nm picture p. Turn off 448 nm LED q. Open both test chamber valves to empty sample r. Pause to flush test chamber

Repeat process in the opposite test chamber 2. After x number of cycles of the main loop described above, the Python script carries out the following steps: a. Activate solenoid to drop rudder flap into flow and initiate device rotation. b. Wait for rotation initiation. c. Activate opposite solenoid to drop opposing rudder flap slow device rotation. d. Wait. e. Raise rudder flaps. f. Resume main loop

Once the images are collected, visually distinguishing microplastics from other substances can be improved and executed more rapidly from image analysis. With the ability to create many training images of known samples, machine learning (ML) is used for this.

The ML process starts with data collection. In order to teach an algorithm how to identify microplastics, or train it, the algorithm is shown images that contain microplastic samples, images that do not contain microplastics, and images that contain objects which may appear in microplastic samples that are not microplastics (like fertilizer, paper, soil).

Images used to create a data set for training the ML algorithm are produced experimentally, using the same process outlined above. Images used in the training process are categorized as training images, validation images, or test images.

Training images are used throughout the algorithm improvement cycle to modify the algorithm while validation images are only to be used intermittently to evaluate the training progress. Test images are only to be used on the final algorithm, so that model changes are made from these images to improve the algorithm’s predictive capabilities.

Image processing may involve image scaling, colour alteration, or other tools. Most importantly, the training data set must be labelled. Image analysis software, like the open source tool Imaged, used by the University of Warwick in Coventry, U.K. to identify microplastics in ocean water, is used to aid with this process [2],

During image processing, images are labelled to identify which pixels contain microplastics. Labels allow the algorithm to learn how to distinguish microplastics from other objects in samples. Once labelled, to create a sufficient number of images to train, validate, and test the algorithm, the image set is expanded. Data augmentation is the most common method used to create more test images to improve model performance and reduce the amount of time spent manually collecting images.

After each iteration of model training with the training data set, referred to as an epoch, model outputs are evaluated. Based on labels in the data, models are able to self-evaluate and auto update algorithm parameters to improve performance.

Once changes are implemented, a new epoch begins, and the training data will be resubmitted through the algorithm. The cycle repeats until the model meets the applicant’s specifications.

Algorithms are written in many coding languages, including Python. As the electronics portion of the system will be controlled with a Python script, the ML algorithm could be integrated into the Python script, allowing for in-situ image processing.

It is envisaged that the ML will occur off the system and the data then planted back in the memory for the processor 18 to use. Prototype testing of the system described above was conducted by the inventors and will be described below in more detail.

Firstly, initial testing was conducted to determine the validity of fluorescence as a microplastic detection method in water.

The first phase employed a spectrofluorometer, which is a machine used to measure the fluorescence signature of a sample at a specified excitation or emission wavelength range. The second phase required testing to occur outside of the tightly controlled environment of a spectrofluorometer to determine if experimental results could be recreated with a preliminary sensor prototype.

The spectrofluorometer used in testing provided information regarding the intensity of fluorescence as a function of excitation, which was used to pinpoint optimal excitation wavelengths to use in the sensor. Based on background microplastic research, testing was performed on PE, PS, PP, and PET, the most prevalent microplastics in the environment.

The results from testing lead were that three of the four plastics exhibited emission behaviour between 385-420 nm when excited by light between 385- 390 nm. Given the overlap in this range, it was determined that this would allow for the detection of the most types of plastics for the initial prototype.

The primary takeaway from this testing was that fluorescence is a feasible option for the sensing of microplastics in water. It was able to provide a level of recognition of the plastics, as well as differentiate between dirt and other materials.

During further testing, it was important to understand if common contaminants and materials in rivers were known to fluoresce, as this could interfere with the sensors results. The literature showed that it was very common for biological material to fluoresce and this fluorescence occurs at many different wavelengths for different organic materials.

An early model of the sensor using 448 nm as the excitation wavelength detected biologicals which were observed in bright green in the test images.

It is important to be able to distinguish between sediment and microplastics and this was a key success for the sensor.

Another common contaminant in rivers is fertilizer as agricultural runoff is known to contaminate surface water. It is unknown from the literature if fertilizer fluoresces even though there are some studies that show ammonia, which is commonly found in fertilizers, can fluoresce. However, an early version of the prototype was able to excite a fertilizer bead but additional testing on a different brand of fertilizer did not show fluorescence results.

From literature, it is known that paper fluoresces. Paper fluorescence is visible when excited between 280-450 nm [3].

During testing of tap water in a prototype device, the water was placed in a tube and inserted into the test setup. First, a white light LED was used to illuminate the sample. Two images were captured by the camera during white light illumination. This process was repeated for 385 nm LEDs and 448 nm LED illumination. All samples were tested in duplicate, meaning that the outlined procedure, from sample mixing to LED excitation and image capturing was repeated twice for each sample, resulting in four images at each light excitation.

The goal of experimental testing was better understanding of how microplastics fluoresce in different sample environments. Each contaminant was also tested by itself in water to see if and how the contaminant fluoresced. The following contaminants were selected for experimental testing from the literature investigation: biologicals, paper pulp, fertilizer, and sediments.

Experimental testing showed potential for interference from contaminants which would need to be mitigated. The plurality of excitation wavelengths and images serve as a mitigation measure.

One qualitative observation when testing was the speed at which non-plastic samples settled. After inducing turbulence, the sample was loaded and pictures were taken, with a maximum of 10 seconds elapsing before the first image was captured to avoid having the material settle. As it was observed that environmental materials settled, in the future the testing rig would be set to the higher mounting locations, so that light could be directed to the top portion of the testing tube to only target microplastics. In doing so, the testing cell could isolate the response of plastics by only allowing plastic to be hit by the excitation light.

During experimental testing it was observed that air bubbles in the samples tended to refract and distort light.

It was noted that the response of bubbles in 448 nm light was different to that of plastic in 448 nm light. As a result, it is possible to combine the results of the 385 nm-excited images with the 448 nm-excited images to help differentiate between plastic and air bubbles.

Although performed in unideal conditions, experimental testing was considered successful. Based on the testing, it appears that materials commonly found in environmental samples may not impede microplastic fluorescence testing, as many do not fluoresce.

Thus it will be appreciated that the system provides the rapid quantification of microplastics in liquid using fluorescence as a microplastic detection method in water, machine learning and image analysis in a portable solution that can be used in-situ. It will also be appreciated that the innate autofluorescence of the particles is the primary indicator utilized by the system.

It is envisaged that the system can be used by researchers in the field and by water and wastewater facilities. Additionally, the bottling industry interested in water quality will also be able to use the system described.

References

[1] C.-R. K. Ishan Barman, G. P. Singh, and R. R. Dasari, “Effect of photobleaching on calibration model development in biological Raman spectroscopy,” Journal of Biomedical Optics. [Online].

[2] Alexander S. Tagg, Melanie Sapp, Jesse P. Harrison, and Jesus J. Ojeda Analytical Chemistry 2015 87 (12), 6032-6040 DOI: 10.1021/acs.analchem.5b00495.

[3] A. C. Croce and G. Bottiroli, “Autofluorescence spectroscopy and imaging: a tool for biomedical research and diagnosis,” European journal of histochemistry : EJH, 12-Dec2014. [Online].