Shape analysis device

Abstract

According to the present invention, a first particle sensor, a second particle sensor and a device for characterizing one or more particles in a fluid sample are provided, the device comprising a first particle sensor and / or at least one second particle sensor. A method of characterizing one or more particles in a fluid sample is also disclosed.

Description

TECHNICAL FIELD

[0001] The present invention relates to an apparatus for characterizing one or more particles in a fluid sample. BACKGROUND

[0002] The discovery and deployment of new materials require their detailed characterization. For emerging materials used in the environment (energy, agriculture), healthcare (nanopharmaceuticals, therapeutics), or food industry (food, packaging), details such as size, shape, charge, and concentration are critical. Increasing evidence suggests that the properties of materials such as nanomaterials and their applications depend not only on the chemical nature of the material but also on the physical and mechanical properties of the material. Rapid, in situ, or online analysis within the manufacturing industry is limited. The current state-of-the-art uses microscopy (electronic or optical) which is costly, low throughput, and provides limited information on the material in its natural environment. Solution-based techniques include light scattering which can become complicated by flow. In addition, it can be difficult to handle multi-modal sample sets and turbid solutions. There is also a lack of techniques to quantify nanoscale object shape properties in a high-throughput manner.

[0003] Resistance pulse sensors (RPS) are an emerging technology over the past two decades, which has seen more prevalence in the application of biological samples such as DNA. Most of the research and commercial activities in RPS have been focused on DNA sequencing / analysis. The theory and application of RPS for nanomaterials are emerging and rapidly developing, and are ready to be developed and applied to the manufacturing of emerging nanomaterials.

[0004] The traditional approach to analyzing materials is to perform reactions / synthesis, extraction, and then analyze the product. These workflows are often referred to as batch reactors. The translation of batch synthesis to continuous flow platforms represents an increasing area of research over the past decade. They reduce production costs and reduce the variability of the product from batch to batch. However, the infrequent sampling / monitoring of reactions within the jet flow reactor limits the benefits of the flow process. Therefore, the improvement of microreactor systems lies in the fabrication of complete laboratories on a chip. These “lab-on-a-chip” devices ensure uniform mixing and jet behavior by integrating continuous online monitoring of the chemical products for high-throughput processing.

[0005] Sample mixtures require individual particle classification and analysis in order to determine the size, charge, and concentration of various particles. Current technologies have a fixed range, so they can only measure one type of material. There is a need for high-throughput online sensors for nanomaterials with single particle resolution, capable of handling particles from nanometers to microns.

[0006] An emerging manufacturing process for microfluidic systems is additive manufacturing (AM) or 3D printing. This is an advantageous alternative, mainly because it is able to build 3D designs from STL (Standard Tessellation Language) files without intermediate steps, thus minimizing labor, time and cost. AM allows for custom flow channels to be designed for and interfaced with the analysis platform. BRIEF DESCRIPTION OF DRAWINGS

[0007] The present application will now be described, by way of example only, with reference to the accompanying drawings in which:

[0008] Figure 1 is a perspective view of the apparatus according to an embodiment of the application;

[0009] Figure 2 shows the apparatus according to an embodiment of the application from above;

[0010] Figure 3 shows the apparatus according to an embodiment of the application from below;

[0011] Figure 4 shows the apparatus according to an embodiment of the application without lid;

[0012] Figure 5 shows the electrodes, O-ring groove and inlet / outlet when the lid is removed;

[0013] Figure 6 shows a perspective view of the lid of an embodiment of the application;

[0014] Figure 7 shows a plan view of the lid of an embodiment of the application;

[0015] Figure 8 shows a second configuration of the lid on the apparatus according to an embodiment of the application;

[0016] Figure 9 shows a second particle sensor according to an embodiment of the application;

[0017] Figure 10 illustrates the mechanism of interaction between the apparatus and the particles;

[0018] Figure 11 a) shows a schematic of the RPS setup according to an embodiment of the application; Figure 11 b) shows the signal from a translocation event;

[0019] Figure 12 shows a plan view of the electrode arrangement in an embodiment of the first particle sensor;

[0020] Figure 13shows a CAD drawing of a lid according to an embodiment of the application, the lid having a primary ridge;

[0021] Figure 14 shows a CAD drawing of a lid according to the application, the lid having a secondary ridge extending from the primary ridge;

[0022] Figure 15 shows a plot of current recorded at 1 volt in 0.25 mM KCI buffer compared to ridge height measurements;

[0023] Figure 16 shows a pulse measured from an 80 micron particle in a 150 micron microfluidic channel;

[0024] Figure 17 shows an exploded view of an apparatus according to an embodiment of the application;

[0025] Figure 18 shows a CAD drawing of a solid state nanopore scaffold;

[0026] Figure 19 schematically shows the operation of an apparatus comprising a first particle sensor and a second particle sensor and corresponding pulses;

[0027] Figure 20a is a schematic of an apparatus comprising a plurality of second particle sensors;

[0028] Figure 20b is a schematic of an embodiment of a second particle sensor comprising a scaffold;

[0029] Figure 20c is a schematic showing fluid flow in relation to a second particle sensor;

[0030] Figure 20e -g shows different flow set-ups for a second particle sensor;

[0031] Figure 21 shows a workflow describing the use of an apparatus;

[0032] Figure 22 (a) shows a schematic of an RPS set-up and signal. A nanoparticle passes through the pore and produces a time-current signal. The signal can be characterised by its amplitude Δip, the width across the pulse (e.g. full width at half maximum (FWHM)) or its shape.

[0033] Figure 22 (bi) is a TEM image of carboxyl-coated iron oxide nanorods. Figure 22 (bii) is a SEM image of CPC200 polystyrene particles. Scale bar is 1 μm;

[0034] Figure 23a) shows predicted normalized pulse shapes for nanorods and nanospheres. Figure 23 b) shows recorded average normalized resistance pulses for nanospheres (231) and nanorods (230). Figure 23 c) shows recorded average normalized pulses for nanorods with aspect ratios of 5 (232), 2.8 (233), 1.8 (234) and nanospheres (235);

[0035] Figure 24 Recorded pulse data is shown. Each pulse containing 301 data points is isolated. Pulse analysis is done by one of five models. A) Pulse amplitude, B) blockade width at fractions of 0.75, 0.5, 0.25 of the pulse height. C) Non-normalized spline fit and D-E) Spline fit on normalized pulse amplitude;

[0036] Figure 25 Visualization of key spline segments that identify particle shape as determined by a multiple t-test with Bonferroni correction is shown;

[0037] Figure 26 (a) shows a plot of the ratio of rods to spheres predicted using data model D versus known ratios;

[0038] Figure 27 a) Schematic of the sensing area when using an acetate film, flat lid and ridge lid, b) Schematic of the expected resistance pulse when a particle passes through the sensing area;

[0039] Figure 28 Schematic of a first particle sensor with different lid designs is shown, demonstrating how the size of the microfluidic channel can be adjusted via providing a polymer layer on the lid surface and via applying increased pressure to the lid;

[0040] Figure 29 (a) is a plot of measured current versus flow rate for a flat PDMS lid, voltage = 5.64 V, 0.25 mM KCl. b) Current-voltage plot without flow, c) Current-voltage plot with 100 mbar pressure applied to the liquid, d) Baseline current of the device with 100 mbar flow pressure at voltage = 0.5 V (100 mM), 0.6 V (50 mM), 8.5 V (0.25 mM), 100, 50 and 0.25 mM KCl (from left to right);

[0041] Figure 30 (a) is an example of current traces from a ridge lid design, particle diameter 20 x 10 -6 m, 8.5 V. b) Particle diameter 20 x 10 -6m, particle count vs pressure at 8.5V. c) particle diameter = 20 x 10 -6 m, particle count vs concentration at 8.5V. d) particle diameter = 30 x 10 -6 m, pulse amplitude distribution at 5.64V. e) particle diameter = 20 x 10 -6 m, pulse amplitude distribution at 7.5V. f) average pulse shape for each of the gaskets in e, averaged over 60 particles. All examples run on 0.25mM KCl;

[0042] Figure 31 (a) shows the average pulse shape for a) spherical algae (300) to rod-shaped algae (305). b) average pulse shape for rod-shaped algae at 50 mbar (310), 100 mbar (312), 125 mbar (312) and 150 mbar (316) pressure; -6 m, blockade distribution at 5.64V, 0.25mM KCl, using two different (low-left, high-right) screw tensions. b) as a control, the device was sealed using an acetate tape (Gen-1). Particle diameter = 30 x 10 -6 m, blockade distribution at 5.64V, 0.25mM KCl. Increasing the tension on the screws allows the RPS to measure smaller particles. c) particle diameter = 20 x 10 -6 m, blockade distribution at 8.5V, 0.25mM KCl, d) particle diameter = 10 x 10 -6 m, blockade distribution at 0.6V, 100mM KCl. e) particle diameter = 2 x 10 -6 m, blockade distribution at 1V, 100mM KCl;

[0043] Figure 32 shows the signal obtained from a) microplastic particles running at 50mM KCl, 0.4V voltage, b) microplastic particles running at 0.25mM KCl, 7.5V voltage, c) algal particles running at 50mM KCl, 0.7V and d) 0.25mM KCl, 7V. In all images the scale bar is x=10 seconds, y=0.5nA;

[0044] Figure 33 shows a) average pulse shape for spherical algae (300) to rod-shaped algae (305). b) average pulse shape for rod-shaped algae at 50 mbar (310), 100 mbar (312), 125 mbar (312) and 150 mbar (316) pressure;

[0045] Figure 34 shows a method according to an embodiment of the application;

[0046] Figure 35 shows a method according to an embodiment of the application;

[0047] Figure 36 A method of using a classifier according to embodiments of the application is shown; and

[0048] Figure 37 A system according to embodiments of the application is shown. DETAILED DESCRIPTION

[0049] According to the present application, there is provided a first particle sensor comprising:

[0050] a base comprising a microfluidic channel, wherein the microfluidic channel comprises a first electrode and a second electrode positioned along the microfluidic channel, wherein the first particle sensor is configured such that when a fluid sample comprising at least one particle passes along the microfluidic channel and flows past the first electrode and the second electrode, each of the at least one particle is recorded as a pulse (e.g. a resistance pulse or a conductance pulse).

[0051] According to another aspect of the present application, there is provided a second particle sensor comprising:

[0052] a holder housing a membrane and an electrode, wherein the membrane comprises at least one aperture, wherein the second particle sensor is configured such that when a fluid sample comprising at least one particle passes through the at least one aperture of the membrane, the electrode detects the at least one particle in the fluid sample and records a pulse (e.g. a pulse resistance or a conductance pulse).

[0053] According to another aspect of the present application, there is provided an apparatus for characterising one or more particles in a fluid sample, comprising:

[0054] (i) a first particle sensor as described herein; and / or

[0055] (ii) at least one second particle sensor as described herein,

[0056] wherein the apparatus further comprises a fluid inlet and a fluid outlet connected by the first particle sensor, wherein, in use, a fluid sample passes in through the inlet, along the microfluidic channel of the first particle sensor, past the first electrode and the second electrode, and out through the fluid outlet, and wherein each of the at least one particle is recorded as a pulse (e.g. a resistance pulse or a conductance pulse).

[0057] According to another aspect of the present application, there is provided an apparatus for characterising one or more particles in a fluid sample, the apparatus comprising:

[0058] a first particle sensor comprising a base comprising a microfluidic channel, wherein the microfluidic channel comprises a first electrode and a second electrode positioned along the microfluidic channel, and a detection region between the first electrode and the second electrode;

[0059] an inlet; and

[0060] an outlet,

[0061] wherein, in use, a fluid sample comprising at least one particle flows in through the inlet, along the microfluidic channel of the first particle sensor, past the first and second electrodes, and out through the fluid outlet, wherein the passage of the at least one particle through the detection region is recorded as a pulse (e.g. a resistance pulse).

[0062] According to a further aspect of the application there is provided a device for characterising one or more particles in a fluid sample, the device comprising:

[0063] an inlet;

[0064] an outlet;

[0065] a microfluidic channel extending between the inlet and the outlet and providing an axial flow path for fluid flowing therealong;

[0066] a first particle sensor for detecting the passage of a particle moving along the axial flow path;

[0067] a first electrode and a second electrode positioned along the microfluidic channel;

[0068] wherein, in use, one or more particles flowing along the microfluidic channel are detected by the first particle sensor and the passage of the one or more particles is recorded as a pulse (e.g. a resistance pulse).

[0069] The first particle sensor can be aligned with the flow path, or can extend at an obtuse, acute or normal angle to the axial flow path.

[0070] According to a further aspect of the application there is provided a device for characterising one or more particles in a fluid sample, the device comprising:

[0071] an inlet;

[0072] an outlet;

[0073] a microfluidic channel extending between the inlet and the outlet and providing an axial flow path for fluid flowing therealong;

[0074] a particle sensor for detecting the passage of a particle moving along the axial flow path, the particle sensor extending at an angle to the microfluidic channel;

[0075] a first electrode and a second electrode, wherein, in use, one or more particles flowing along the microfluidic channel are detected by the particle sensor and the passage of the one or more particles is recorded as a pulse (e.g. a resistance pulse).

[0076] The particle sensor can comprise a flow path for particles. The particle sensor or flow path can extend at an obtuse angle, at an acute angle or orthogonally to the axial flow path. In one embodiment, the particle sensor extends at a right angle to the microfluidic channel.

[0077] Fluid flow through the microfluidic channel of the device can be driven by a pressure applied to the fluid (e.g. by a pump), by an electrical potential difference applied between the first and second electrodes, or by a combination of pressure and electrical potential difference. In some embodiments, fluid flow is driven primarily by pressure applied to the fluid, and to a lesser extent by the electrical potential difference between the first and second electrodes, or vice versa.

[0078] In some embodiments (e.g. in use), the microfluidic channel is filled with an electrolyte solution.

[0079] The first particle sensor comprises a detection region at a point along the longitudinal axis of the microfluidic channel and / or the axial flow path between the first and second electrodes. The detection region can comprise a nanopore or constriction in the channel. Passage of a particle in the fluid sample through the detection region can be recorded as a pulse.

[0080] In some embodiments, a second particle sensor can be provided. The base of the first particle sensor comprises at least one port for receiving a further sensor, e.g. a second particle sensor. In embodiments, a plurality of second sensors can be provided.

[0081] In some embodiments, the first particle sensor is configured to detect particles having a size of 1 pm to 100 pm, 5 pm to 100 pm, 10 pm to 80 pm or 20 pm to 50 pm. In some embodiments, the first particle sensor is configured to detect particles of 5 pm to 50 pm. Thus, particles smaller than 1 pm will not be detected by the first particle sensor.

[0082] It will be appreciated that the cross-sectional area of the microfluidic channel or of the detection region therein can be selected according to the size of the particles to be detected. Thus, in some embodiments, the microfluidic channel or the detection region therein has a cross-sectional area of less than 10 4 pm 2 .

[0083] Thus, the first particle sensor can be capable of detecting, characterising and / or counting biological particles (such as cells (e.g. bacterial cells, fungal cells, algal mammalian cells, human cells, blood cells, cancer cells, stem cells), exosomes, vesicles, proteins, protein-protein or protein-nucleic acid complexes, nucleic acids, antibodies, colloids), organic particles (such as polymeric particles and drug particles), inorganic particles (such as metal particles, gas bubbles and emulsions).

[0084] Optionally, the device comprises at least one second particle sensor comprising a nanopore.

[0085] The second particle sensor can further comprise at least one electrode. The second particle sensor can be configured such that when a fluid sample comprising at least one particle passes through the nanopore, the electrode detects the at least one particle in the fluid sample and records a pulse (e.g. a resistance pulse or a conductance pulse).

[0086] In some embodiments, the nanopore of the second particle sensor is a solid-state nanopore. As known to the skilled person, a solid-state nanopore typically comprises a hole (e.g. 1 to 1000 nm in diameter) formed in a membrane. The membrane can be formed from any suitable material. In some embodiments, the membrane is formed from a material selected from the group consisting of silicon nitride, silicon dioxide, glass, graphene, plastic (e.g. polyurethane or polyester).

[0087] In some embodiments, the nanopore is provided in a nanotube. As known in the art, a nanotube is a class of nanopore that can be used to detect and analyse single molecules in solution. Nanotubes can be readily manufactured with a highly controllable pore size, making them an economically viable alternative to traditional solid-state nanopores. Detecting and analysing single molecules using nanotubes relies on resistance pulse sensing. To this end, the nanotube is filled with an electrolyte, and the tip is immersed in the electrolyte, and a voltage is applied between an electrode inside the nanotube and an electrode outside, to generate an electric field at the tip. This field drives molecules of interest through the nanotube pore, resulting in a detectable pulse.

[0088] The nanotube can be formed from any suitable material, such as a metal, a polymer, glass, quartz, an organic material (e.g. graphene) or an inorganic material (e.g. boron nitride). In some embodiments, the nanotube is made from glass or quartz.

[0089] The nanotube can be manufactured using methods known to the skilled person. Typically, a mechanical pipette puller is used to pull a nanotube from a capillary tube (e.g. quartz).

[0090] It will be appreciated that the diameter of the nanopore of the second particle sensor can be selected according to the size of the particles to be detected. In some embodiments, the second particle sensor is configured to detect particles having a size of 1 nm to 100 pm, 5 nm to 50 pm, 10 nm to 20 pm, 20 nm to 10 pm, 30 nm to 10 pm, 40 nm to 5 pm, 50 nm to 2 pm or 100 nm to 1 pm.

[0091] Accordingly, in some embodiments, the nanopores have a diameter of 1 nm to 100 pm, 5 nm to 50 pm, 10 nm to 20 pm, 20 nm to 10 pm, 30 nm to 10 pm, 40 nm to 5 pm, 50 nm to 2 pm, or 100 nm to 1 pm. It will be appreciated that the nanopores are not necessarily circular. Accordingly, herein, the "diameter" of a nanopore refers to the average dimension of the pore. It will also be appreciated that reference to the diameter of a nanopore refers to the internal diameter.

[0092] Accordingly, the second particle sensor can be capable of detecting, characterising and / or counting particles such as microorganisms (e.g. bacterial cells, fungal cells, algae), viruses, nucleic acids (e.g. DNA, RNA), peptides, proteins, polymeric particles, inorganic particles, metallic particles, gas bubbles and emulsions.

[0093] Conveniently, the second particle sensor can be configured to detect particles of a different size to the particles that the first sensor is configured to detect. The second particle sensor can be configured to detect a different range of particle sizes to the range of particle sizes that the first particle sensor is configured to detect. The second particle sensor can be configured to detect smaller particles than the particles that the first sensor is configured to detect. Accordingly, the second particle sensor can comprise nanopores having a diameter that is smaller than the diameter (or largest dimension) of the constriction or nanopores of the first particle sensor. Accordingly, the second particle sensor can be capable of measuring particles that the first particle sensor is not capable of measuring. The range of particle sizes that can be detected by the first and second particle sensors respectively can or can not overlap. For example, the first particle sensor can be configured to detect particles falling within the range 2 pm to 100 pm, whereas the second particle sensor can be configured to detect particles falling within the range 1 nm to 2 pm. For example, a population of particles flowing through the device or along the microfluidic channel can be polydisperse and / or have a wide size distribution. The first particle sensor can be configured to detect a first subset of particle sizes within the wide particle size distribution, wherein the second particle sensor can be configured to detect a second, different subset of particle sizes within the wide particle size distribution. Optimally, the first particle sensor can detect a relatively larger first subset of particle sizes, whereas the second particle sensor can detect a relatively smaller first subset of particle sizes. The first and second subsets of particle sizes can or can not overlap.

[0094] It will be appreciated that, as used herein, reference to a particle size refers to the largest lateral dimension of a given particle. Particles that can be detected by the sensors and devices of the application are not necessarily spherical, but can be elongate or irregularly shaped. It will be appreciated that the present application allows the volume of a particular particle to be determined.

[0095] Advantageously, the use of first and second particle sensors configured to detect different sized particles enables the apparatus as a whole to detect particles spanning a larger size range, for example from nanometer to micrometer sized analytes. It will be appreciated that the first and second sensors can be independently configured to detect particles within a desired size range. The tunability of each sensor means that the apparatus can be tailored for a given application, for example to the type of fluid being analysed.

[0096] In some embodiments, the apparatus comprises two or more second particle sensors. For example, the apparatus can comprise three, four, five, six or more second particle sensors.

[0097] In some embodiments, one of the second particle sensors is configured to detect particles of a different size (i.e. falling into a different size range) to one or more of the other second particle sensors. In some embodiments, each of the second particle sensors is configured to detect particles of a different size or different size range. In some embodiments, one of the second particle sensors comprises a nanopore having a different diameter to the nanopore within another of the second particle sensors. In some embodiments, each of the second particle sensors comprises a nanopore having a different diameter to each of the nanopores within the other second particle sensors.

[0098] The nanopore of the second particle sensor can be positioned off-axis relative to the longitudinal axis of the microfluidic channel. By “off-axis” it is to be understood that the second particle sensor is offset from the main direction of fluid flow within the microfluidic channel. For example, the second particle sensor can extend at an angle relative to the microfluidic channel. In one embodiment, the second particle sensor can provide a flow path for the particles. The flow path can extend at an obtuse angle, at an acute angle or orthogonally to the microfluidic channel or axial flow path.

[0099] The nanopore can be located within a second channel extending from the (main) microfluidic channel. Thus, the second channel provides an additional flow path along which fluid can flow past the nanopore of the second particle sensor. The nanopore can be spaced apart (i.e. set back from) the fluid flow within the microfluidic channel.

[0100] The second channel can extend perpendicularly (i.e. at a 90° angle) from the microfluidic channel. In such embodiments, the nanopore can be positioned parallel to the longitudinal axis of the microfluidic channel (i.e. parallel to the main direction of fluid flow). The arrangement of the second channel (or each second channel, in embodiments where multiple second particle sensors are provided) extending perpendicularly from the main microfluidic channel can facilitate manufacture.

[0101] Alternatively, the second channel can extend from the microfluidic channel at a non-perpendicular angle, i.e. at an angle greater or less than 90°. In further embodiments, the second channel can curve as it extends from the microfluidic channel. It will be appreciated that in such embodiments, the nanopore can not be parallel to the longitudinal axis of the microfluidic channel, but can be angled with respect thereto.

[0102] The second particle sensor can be positioned upstream of the detection region of the first particle sensor (i.e. between the inlet and the detection region), or it can be positioned downstream of the detection region or the first particle sensor (i.e. between the detection region and the outlet). The second channel at which the second particle sensor can be located (or can be provided) can be connected to the microfluidic channel at a point upstream or downstream of the detection region of the first particle sensor. In embodiments in which the device comprises two or more second particle sensors, one or some of the second particle sensors can be located upstream of the detection region, and the remaining second particle sensors can be located downstream of the detection region. Alternatively, all of the second particle sensors can be located upstream of the detection region, or all of the second particle sensors can be located downstream of the detection region.

[0103] In some embodiments, the second particle sensor comprises a holder which houses a nanopore (i.e. a solid-state nanopore comprising a membrane or a nanopipette) and an electrode.

[0104] The second channel can be present in the holder, with the nanopore being located inside the second channel. When the second particle sensor is connected to the base of the first particle sensor, the second channel connects (i.e. forms a flow path with) the microfluidic channel. For example, the holder can be in the form of a plug or screw which is configured to be releasably inserted into a port in the base of the first particle sensor, thereby forming a fluidic connection between the microfluidic channel and the second channel. Fluid is thus able to flow from the main microfluidic channel and into the second channel, such that it passes through the nanopore.

[0105] In some embodiments, the nanopore (e.g. a nanopipette or a solid-state nanopore membrane) does not extend across the entire width of the second channel, such that fluid is able to flow past rather than through the nanopore. The second channel can extend through the holder to its surface, and can have an open end, thereby providing an exit for fluid which has been flowed through and / or around the nanopore.

[0106] In some embodiments, the second particle sensor further comprises a tube. The tube can be in fluid communication with the second channel. At least a portion of the tube can be housed within the holder. In some embodiments, a portion of the tube extends from the holder. In some embodiments, the tube provides a further fluid outlet from the device. Liquid that flows around (and / or through) the nanopipette can flow into the tube and out of the device. The presence of the tube has been found to aid fluid flow in the device and facilitate set-up of the device by enabling removal of air bubbles.

[0107] Conveniently, the second particle sensor can be reversibly connected to the base of the first particle sensor. This enables a device to be provided in which the second particle sensor (or some or all of the second particle sensor) is interchangeable. Thus, a second particle sensor can be selected that has the most suitable nanopore size for detecting small particles. In this way, the sensitivity of the device can be adjusted according to the required use or the nature of the fluid being analysed. When a second particle sensor is present, the sensor can be filled with an electrolyte solution between the nanopore and the electrodes (e.g. between the membrane and the electrodes).

[0108] In some embodiments, the (or each) second particle sensor can comprise or be connected to a flow regulator (e.g. a pump). The flow regulator can be used to control the flow of fluid through the second sensor. In embodiments in which the second particle sensor does not comprise or is not connected to a flow regulator, the flow of fluid through the second sensor will be driven primarily or exclusively by the potential difference between the working electrode and the ground electrode of the second particle sensor.

[0109] As mentioned above, the first particle sensor comprises a first electrode (ground electrode) and a second electrode (working electrode). The application of a potential difference between the first sensor and the second sensor enables the first particle sensor to detect particles and can also aid in driving the flow of fluid through the microfluidic channel. The first electrode and the second electrode thus form a first electrode set.

[0110] The second particle sensor can comprise at least one electrode. In some embodiments, the second particle sensor comprises a single electrode as a working electrode. The first electrode (ground electrode) of the first particle sensor can also be used as a ground electrode for the second particle sensor. Thus, in some embodiments, a second electrode set is formed from the first electrode (ground electrode) of the first particle sensor and the working electrode of the second particle sensor. For example, in embodiments in which the device comprises two sensors (a first particle sensor and a second particle sensor), the device can comprise three electrodes, comprising a common ground electrode and a working electrode for each particle sensor.

[0111] In embodiments in which a plurality of second particle sensors are provided, each second particle sensor can comprise a working electrode, and each of the first and second particle sensors can share a common ground electrode (first electrode).

[0112] Alternatively, the (or each) second particle sensor can be provided with its own ground electrode in addition to the working electrode. Thus, the or each second particle sensor can comprise a set of electrodes.

[0113] In some embodiments, the apparatus comprises one or more third sensors. The third sensor can be configured to measure a parameter of the fluid, such as oxygen content, pH, ionic strength, temperature or viscosity. Thus, the third sensor can comprise a probe for measuring the required parameter, such as a pH probe, a viscosity probe, an oxygen probe and an ionic strength probe or a thermometer. In some embodiments, the apparatus can comprise a plurality of third sensors. Each of the third sensors can be configured to detect a different parameter of the fluid.

[0114] In some embodiments, the third sensor comprises a holder to house the probe. Similar to the holder of the second particle sensor, the holder of the third sensor can take the form of a plug or screw which can be inserted into a port in the base of the first particle sensor such that the probe is able to contact the fluid flowing through the microfluidic channel.

[0115] Thus, in another aspect, the present application provides a kit for characterising one or more particles in a fluid sample, the kit comprising:

[0116] - an apparatus comprising a first particle sensor as described herein; and

[0117] - at least one further sensor for connecting to the first particle sensor (e.g. for connecting to the base of the first particle sensor). As described herein, the further sensor can be a second particle sensor and / or a third sensor.

[0118] Conveniently, the further sensor can be releasably connectable to the base of the first particle sensor. In some embodiments, the base of the first particle sensor comprises at least one port for receiving the further sensor.

[0119] In some embodiments, the second particle sensor comprises a holder to house a nanopore and an electrode, the holder having a second channel therein in which the nanopore is located, and wherein the holder is configured to be releasably inserted into a port in the base of the first particle sensor.

[0120] In some embodiments, the kit comprises at least one second particle sensor and at least one third sensor.

[0121] In some embodiments, the kit includes two or more second particle sensors, wherein one of the second particle sensors includes a nanopore having a diameter that is different than a diameter of a nanopore in another of the other second particle sensors.

[0122] For example, one of the second particle sensors can include a first nanopore having a first diameter, and another of the second particle sensors can include a second nanopore having a second diameter, wherein the first diameter is different than the second diameter.

[0123] In some embodiments, the kit includes a plurality of second particle sensors, e.g., 2, 3, 4, 5, 6, or more second particle sensors. In some embodiments, each of the second particle sensors includes a nanopore having a diameter that is different than a diameter of a nanopore in each of the other second particle sensors.

[0124] The kit can include at least one third sensor for connection to the first particle sensor (e.g., a base for connection to the first particle sensor). In some embodiments, the kit includes two or more third sensors (e.g., 2, 3, 4, or more third sensors).

[0125] Accordingly, the present invention conveniently provides a modular system in which different sensors can be interchangeably connected to a first particle sensor including a microfluidic channel.

[0126] The apparatus can further include a lid configured to seal the microfluidic channel of the apparatus.

[0127] In some embodiments, the lid can also create a constriction within the microfluidic channel in order to adjust the sensitivity of the first particle sensor.

[0128] The lid can include a main ridge that fits into the microfluidic channel and is configured to seal the microfluidic channel. In some embodiments, the lid includes a main ridge that is configured to be received within the microfluidic channel when the lid is placed on the base, thereby reducing the volume of the channel. The main ridge can have a depth that is less than a depth of the microfluidic channel. The main ridge can have a length that is substantially the same as a length of the microfluidic channel (length being a dimension measured parallel to a longitudinal axis of the base / microfluidic channel). The main ridge can have a width that is substantially the same as a width of the microfluidic channel.

[0129] Optionally, the main ridge further includes a secondary ridge extending from the main ridge. The secondary ridge can be used to create a constriction within a portion of the microfluidic channel or at a particular point within the channel. The secondary ridge can have a width that is substantially the same as a width of the main ridge.

[0130] Optionally, the secondary ridge comprises a conduit that allows fluid sample to flow through the secondary ridge when the lid seals the microfluidic channel. In some embodiments, the primary ridge and the secondary ridge together span the height and width of the microfluidic channel when the lid is placed on the base. In such embodiments, the secondary ridge comprises a conduit that allows fluid to flow through the secondary ridge when the lid seals the microfluidic channel.

[0131] In some embodiments, the lid comprises a layer (e.g. a polymer layer) on the lid surface that contacts the base when the lid is placed on the base so as to seal the microfluidic channel. The layer (e.g. the polymer layer) thus acts as a gasket.

[0132] The layer (e.g. the polymer layer) can cover substantially the entire surface or part of the surface. For example, in embodiments in which the lid comprises a primary ridge and optionally a secondary ridge, the layer (e.g. the polymer layer) can not cover the primary ridge and / or the secondary ridge.

[0133] The layer (i.e. the gasket) can be formed from any suitable deformable (e.g. compressible) material. Suitable materials include synthetic or natural rubber, silicone, cork, cellulose, foam, nitrile and fibre. In some embodiments, the layer is a polymer layer. The polymer layer can be formed from a deformable polymer such that the shape of the polymer layer can be altered upon the application of a force. Suitable polymers include polydimethylsiloxane (PDMS).

[0134] In some embodiments, the lid is attached to the base with one or more screws. The lid can be attached to the base by a plurality of screws (e.g. 2, 3, 4, 5, 6 or more screws).

[0135] In some embodiments in which the lid comprises a polymer layer, the polymer layer can deform and be forced into the microfluidic channel when the screws that attach the lid to the base are tightened, thereby reducing the volume of the channel.

[0136] It will therefore be appreciated that the internal volume (i.e. shape and / or size) of the microfluidic channel, and hence the sensitivity of the first sensor, can be conveniently controlled by: (i) the structure of the lid, via the provision of a primary ridge and optionally a secondary ridge, and / or (ii) the provision of a deformable polymer layer that can be compressed into the microfluidic channel.

[0137] Optionally, the apparatus comprises a polymer layer between the lid and the base, wherein the polymer layer is configured to seal the apparatus, preferably wherein the polymer layer is polydimethylsiloxane (PDMS). Optionally, the polymer layer can be configured to adjust the sensitivity of the apparatus.

[0138] In some embodiments, one or more components of the device are 3D printed. For example, the base, lid and / or holder of the second sensor can be 3D printed. In some embodiments, every component of the device is 3D printed.

[0139] The base can further comprise a groove comprising an O-ring configured to prevent leakage from the microfluidic channel.

[0140] Optionally, the device is configured on a chip, preferably the chip is for online processing.

[0141] According to the present invention, there is provided a method of characterising one or more particles in a fluid sample comprising one or more devices as described herein, wherein the method comprises: passing a fluid sample comprising at least one particle through the fluid inlet, along the microfluidic channel, past the first and second electrodes, and out of the fluid outlet; and wherein each of the at least one particle present in the fluid sample is recorded as a resistance pulse in the presence of an electrical potential difference between the first and second electrodes.

[0142] Preferably, the method further comprises the step of using a predictive logistic regression model to characterise the resistance pulse to determine information about the size, shape and flow rate of the particle.

[0143] The method can be used to characterise one or more of:

[0144] (i) cells in a bodily fluid

[0145] (ii) organic compounds, proteins, peptides, cells, bacteria, fungi, algae, viruses, nucleic acids (e.g. DNA), exosomes, colloids, polymeric particles (e.g. microplastics) and nanomedicines

[0146] (iii) inorganic materials, such as metal particles.

[0147] The method of the present invention can be used for online sensing of one or more fluid samples for high-throughput processing.

[0148] The base of the device can be composed of a polymer and / or resin and can be manufactured using a moulding / manufacturing technique capable of creating three-dimensional shapes. Three-dimensional printing techniques that can be used in the present invention are well known to those skilled in the art. Alternatively, the device can be manufactured using any other convenient manufacturing process, such as moulding or micro-injection.

[0149] Resistance pulse sensors (RPS) provide detailed characterization of particles from small molecules to nanomaterials on an individual particle basis. They provide information about particle size, shape, concentration, and charge, and also provide some information about particle shape. Importantly, the low cost and high throughput (tens to hundreds of particles per second) of RPS makes it suitable for manufacturing workflows. Recently, progress has also been made in applying the technique to high-throughput characterization of object deformation. Thus, RPS provides a multi-faceted non-destructive characterization of nanomaterials.

[0150] The signal in an RPS experiment can reveal information about the shape of the material. Resistance pulse sensing can also be combined with a predictive logistic regression model (called RPS-LRM) to rapidly characterize the size, aspect ratio, shape, and concentration of nanomaterials when rod-like and spherical mixtures exist in the same solution. RPS-LRM can be applied to characterize nanoparticles over a wide size range and different aspect ratios, and can distinguish between nanorods when they possess an aspect ratio greater than 2 on nanospheres. RPS-LRM can rapidly measure the ratio of nanospheres to nanorods in solution in a mixture, regardless of their relative size and ratio, i.e. many large spherical particles do not interfere with the characterization of smaller nanorods.

[0151] Here we use high-aspect ratio nanopores to identify the shape of individual nanoparticles in solution. The signals recorded for spheres and rods are sufficiently different to allow their classification on an individual basis. The analysis is fast, with hundreds of particles analyzed in a few seconds. Furthermore, the differences in the shape of the pulses observed experimentally for rods and spheres correspond to a computational model that predicts the pulse shape, which allows for a more extensive characterization of nanoparticle shape using large-pore RPS coupled with statistical methods. Calibrating the response of the nanopore method makes the process repeatable over several nanopores and days, and allows for accurate measurement of the ratio of spheres to rods in solution.

[0152] During the RPS process, particles pass through a channel or pore that conducts ions, and the ion current is monitored as a function of time. The change in current during each translocation, also known as a "pulse", depends on the ratio of particle and channel size. Information about particle size, concentration, and velocity can be measured even in turbid solutions that flow at high speed.

[0153] RPS can be made from materials such as graphene, polymers, silicon nitride, and glass. Their sensitivity can be varied by changing the size of the channel. Transport through the pore or channel can be controlled by adjusting the potential difference, the charge on the pore wall, the electrophoretic mobility of the analyte, the supporting electrolyte concentration, and induced convection. Maintaining sensitivity while preserving high count rates has been a challenge in the past, as pulse frequency is directly related to pore diameter.

[0154] The inventors have found that by using additively manufactured flow channels, different sensors can be included, designed and customized for different applications. By combining additively manufactured parts with precision ion-drilled nanopores, multiple sensors can be included in the same flow channel. Preferably, the flow channel has more than one sensor. More preferably, there are two or more sensors. However, the number of sensors can be more than one, more than two, more than three, more than four. Optionally, the number of sensors can be greater than 10. Preferably, the sensors are nanopore sensors. The number of sensors and the size of the flow channel can be customized according to the intended application of the device and the particles to be detected.

[0155] The use of flow allows fast counting and observation of particles at a rate of one event per second at low particle concentrations (e.g., 1 x 10 -3 Particles / ml). The sample flows / infuses into the sensor through a fluidic inlet. As the sample passes through the microfluidic channel, it passes through a narrow constriction and particles are characterized by one or more sensors.

[0156] Translocation of nanoparticles or analytes through these microfluidic channels can be monitored by measuring the ionic current. Each translocation event causes a change in the conductance of the channel known as a resistance pulse, which is related to the physical properties of the analyte. Thus, resistance pulse sensors are extremely attractive sensing platforms with single particle / analyte resolution. Information about the size, concentration, and charge of the analyte can be measured quickly, reliably, and at a high level of sensitivity. The microfluidic channel size can be created to fit the size of the particles flowing through the channel.

[0157] The microfluidic channel can be formed by using three-dimensional printing and can allow detection of particles with a size of 1 to 100 microns, preferably 5 to 50 microns, and optionally 10 to 25 microns. However, any desired channel size can be created. The microfluidic channel can have multiple sensors to allow characterization of the size, shape, and concentration of one or more particles present in the fluid sample.

[0158] The particles to be characterized are present in a fluid sample. Fluid sample refers to particles in solution. The fluid sample can include any electrically conductive liquid including, but not limited to, fluids such as blood, urine, buffer solutions, or salt-containing solutions such as seawater. The one or more particles can be selected from, but not limited to, organic compounds, proteins, DNA, cells, bacteria, viruses, exosomes, colloids, and nanodrugs.

[0159] To adjust the sensing range of the device of the present invention, instead of having to manufacture different sized constrictions, a standard flow microfluidic channel is retained and the device further comprises a lid which both seals the device (and allows easy access for cleaning) but also adjusts the sensitivity of the device by inserting protrusions into the channel, changing the size of the constrictions and thus adjusting the sensitivity. By keeping the same flow channel and changing the lid, the sensitivity of the sensor can be adjusted. The shape, size and number of protrusions inserted from the lid into the flow can contribute to measuring particle size and shape and can be changed accordingly.

[0160] The lid provides a reusable seal which allows for easy assembly / disassembly and cleaning. The lid can comprise a main ridge which can be an adjustable insert to increase the sensitivity of the device. The lid can be any suitable size or shape. The main ridge can be one or more ridges. The main ridge can be manufactured to correspond to the size of the microfluidic channel. The main ridge can further comprise a secondary ridge extending therefrom. The secondary ridge can be any size but is preferably smaller in size than the main ridge and can further comprise a conduit which allows liquid to pass through the lid.

[0161] According to a preferred aspect of the present invention, there is provided a first particle sensor comprising a base comprising a microfluidic channel, wherein the microfluidic channel comprises a first electrode and a second electrode positioned along the microfluidic channel, wherein the first particle sensor is configured such that when a fluid sample comprising at least one particle passes along the microfluidic channel and past the first electrode and the second electrode, each of the at least one particle is recorded as a resistance pulse in the presence of an applied potential difference between the first electrode and the second electrode.

[0162] The present invention can further comprise a second particle sensor comprising a holder housing a membrane and an electrode, wherein the membrane comprises at least one aperture, wherein the second particle sensor is configured such that when a fluid sample comprising at least one particle passes through the at least one aperture of the membrane, the electrode detects the at least one particle in the fluid sample and records a resistance pulse.

[0163] Figure 1 A preferred embodiment of the assembled device (1) is shown, the main components of which are the lid (2) and the device base (3). The assembled device further comprises threaded holes (4) which allow the second particle sensor (5) and electrode (6) to be screwed into the device (1). The lid (2) and base (3) are secured together with screws (7) and nuts (8). A top view of the assembled device is shown in Figure 2 . A bottom view of the assembled device is shown in Figure 3 . In Figure 1 and Figure 3 , the outer threaded holes (4A and 4B) accommodate the fluid inlet and fluid outlet (13) respectively.

[0164] Figure 4 An embodiment of the device (1) is shown without the lid (2). This embodiment shows the main microfluidic channel (12) in which the liquid flows in at the inlet (4A), along the microfluidic channel (12) and out of the device (1) at the outlet (4B). In this embodiment, the microfluidic channel (12) and the flow of fluid from the inlet (4A) up to the surface of the base unit and then along the top surface. The flow of fluid follows this path, past each electrode (6) and out of the outlet (4B). Figure 4

[0165] To ensure that no liquid leaks from the device (1), between the lid (2) and the base unit (3), an O-ring is housed in a groove (10) (as shown in Figure 4 ), which represents the first configuration of the lid. The groove (10) and O-ring follow the path of the microfluidic channel (12) along the base unit surface.

[0166] In the second lid configuration, a polymer layer can be used instead of an O-ring as a seal. The polymer can be any suitable polymer. Preferably, the polymer is polydimethylsiloxane PDMS (16) and this layer is placed and sandwiched between the device base (3) and the lid (2) and is held in place by the screw (7) and nut (8).

[0167] Figure 5 is a top view of the microfluidic channel (12), electrodes (6), O-ring groove (10) and inlet / outlet (13) when the lid (2) is removed.

[0168] Figure 6 A preferred embodiment of the lid (2) is shown in its first configuration. It has a protrusion (main ridge) (14) that runs the entire length of the microfluidic channel. When the device (1) is sealed as shown in Figure 1 , the depth of the main ridge (14) determines the total volume of liquid required to fill the channel (12). Figure 7 A top view and a bottom view of the lid (2) are shown.

[0169] Along the main ridge is a secondary ridge (15) which can be any length but is preferably the width of the channel (14) and the height of the channel (14) when the device is sealed. Within the secondary ridge is a conduit (11) which allows fluid to flow through the device when it is sealed with the lid in this configuration.

[0170] ​The size of the first particle sensor (11) determines the sensitivity of the sensor. The channel (12), the sensor (11) and the two electrodes (6) record the data from the first particle sensor. When a particle passes through the sensor channel (12) filled with electrolyte, each particle is recorded as a resistance pulse between the two electrodes (6) in the presence of an applied potential difference. The pulse size and shape depend on the channel and pore size, and the pulse reveals information about the particle size, shape and flow rate.

[0171] The first particle sensor (11) can be used independently and does not require the presence of the second particle sensor (5). All particles in the solution flow through the first particle sensor (11).

[0172] Figure 8 A preferred embodiment of the lid (2) in its second configuration is shown. Between the lid (2) and the base unit (3) a polymer layer, such as a PDMS layer (16) is inserted. The lid can be flat or have a primary ridge (14) and a secondary ridge (15) as described previously. The PDMS layer (16) has two functions, first, it provides a seal that will not cause leaks. Second, the volume of the channel and the sensitivity of the first particle sensor (11) can be changed by the PDMS layer (16). The PDMS polymer can be forced into the microfluidic channel upon the application of pressure caused by the tightening of the screw and the force of the lid being applied down onto the PDMS. The liquid volume in the microfluidic channel (12) is reduced by increasing the pressure (tightening) on the screw (7). As the volume of the channel decreases, the sensitivity of the first particle sensor (11) increases. The tightening of the screw (7) affects the sensitivity of the first particle sensor (11). The tightening of the screw has no effect on the second particle sensor (5).

[0173] The second particle sensor (5) can be in a separate housing that can be connected to the base plate of the microfluidic channel (12). The fluid sample flows over the top of the second particle sensor (5) without passing through the sensor.

[0174] Figure 9A second particle sensor (5) is shown, which includes a small hole in a membrane (9). The membrane can be composed of silicon nitride, polyurethane or polyester membrane. The membrane can be fixed in place using 3D printed threads (5) which also include electrodes in the threaded holes (4). Between the electrodes (6) and the membrane is an electrolyte solution. Fluid in the microfluidic channel (12) flows over the top of the sensor (5). An electric field is applied between the two electrodes (6). Charged particles in the microfluidic channel (12) move along the electric field gradient via electrophoresis and convection processes, while neutral particles move by convection. The particles pass from the microfluidic channel (12) through the membrane (9) housed in the second particle sensor (5). As the particles pass through the hole in the membrane, the electrodes (6) record a resistance pulse and characterise the particle shape, size and charge.

[0175] The location of the second particle sensor (5) can be anywhere along the microfluidic channel (12), but not underneath the subridge (15). There can be more than one second particle sensor (5).

[0176] The second particle sensor (5) can be used independently and does not require the presence of the first particle sensor (11). The second particle sensor (5) does not require a flowing solution, so the microfluidic channel (12) can be filled with sample and the flow can be turned off, and the second particle sensor (5) will still characterise particles that can move through the hole in the membrane (9) via electrophoresis. In this way, the second particle sensor (5) can be used both when there is flow and when there is not. The first particle sensor (11) will only work when there is flow. Figure 10 The interaction between fluid flow and one or more sensors is shown.

[0177] Reference is made to Figure 11 (a) and Figure 12 Fluid travels along the microfluidic channel within the first particle sensor as indicated by the arrows. The first particle sensor includes a constriction 120 that provides a detection region. The constriction 120 is located between a first electrode 122 and a second electrode 124 that form a set of electrodes. As Figure 11 (b) shows the passage of a particle through the detection region is detected as a current pulse.

[0178] Figure 13 An image showing an example of a lid (2) according to the present application is shown, the lid (2) including a main ridge (14) of dimensions 1 x 2 mm (H x W) that stretches the length of the microfluidic channel (12). The microfluidic channel (12) and corresponding lid (2) can of course be made to any suitable dimensions. Preferably, the lid (2) provides a seal. When the lid (2) includes a ridge, this allows the sensitivity of the sensor to be tuned. Furthermore, the ridge can be made such that it allows interchangeability between different sensors without the need to change the core flow components.

[0179] Figure 14 The lid (2) is shown including additional secondary ridges (15) that are extensions of the primary ridges (14). By increasing the size of the ridges, the size of the sensor is reduced, which allows for the characterization of smaller particles (such as Figure 14 shown). Figure 15 The results in show that particles can be detected in a microfluidic channel (showing a pulse from an 80 micron particle in a 150 micron channel).

[0180] Maintaining sensitivity while preserving high count rates is a difficult challenge because the pulse frequency is directly related to the aperture size. Therefore, to be able to measure smaller particles in solution, the channel diameter can be reduced.

[0181] The second particle sensor (5) can be a second nanopore made from ion-drilled silicon nitride membrane (9). The second particle sensor (5) can be positioned parallel to the flow and after the first particle sensor. In one example of the device (shown in Figure 1 ), fluid flows through the first particle sensor (11) and past the second particle sensor (5). This prevents larger particles measured in the first particle sensor (11) from clogging the second particle sensor (5). Since the second particle sensor (5) is parallel to the flow, the RPS can characterize particles of any size without worrying about clogging. The second particle sensor (5) (as well as any number of additional sensors) can be screwed in, changed, and selected for the user's sample / application.

[0182] To add additional sensors, additional 3D printed threads can be added along the base (shown in Figure 17 ). The device can also include additional solid-state nanopore holders that house smaller apertures (9), expanding the analytical range of the device (1).

[0183] The second particle sensor (5) can also house the lower fluidic unit and working electrode (6). The holder can be installed using threads located at the bottom of the flow device and a set of O-rings can be used to provide a seal. A CAD image of the holder is shown in Figure 17 .

[0184] To be able to characterize particles of various sizes simultaneously, the second particle sensor (5) will allow for the detection and characterization of particles as low as 1 nm or suitable for sequencing DNA.

[0185] When the device (1) is used in a manufacturing process to produce high concentrations of particles in a high flow rate reactor, to aid the device (1), two identical devices can be used simultaneously, such as Figure 1 or Figure 18Two identical devices are shown in the middle). The main flow / sample is diverted so that the volume and flow rate through the second device is different to the first. This allows each device to be able to characterise different properties / particles in the same sample.

[0186] Figure 19 The operation of the device is illustrated schematically. In Figure 19 In A, small particles of size 1 nm to 2 pm enter the microfluidic channel. The first particle sensor, configured to detect larger particles, does not detect the small particles. The small particles flow through the nanopore of the second particle sensor, generating a signal. In Figure 19 In B, large particles of size 2 to 100 pm enter the microfluidic channel. The large particles are detected by the first particle sensor, generating a signal. In Figure 19 In C, a fluid comprising both large and small particles flows into the microfluidic channel. The large particles are detected by the first particle sensor, while the small particles are detected by the second particle sensor.

[0187] Figure 20a A device 30 is shown comprising a first particle sensor (Sensor 1) and a plurality of second particle sensors (Sensor 2,... Sensor n). The plurality of second particle sensors can be referred to as third, fourth, fifth, etc. particle sensors. The first particle sensor comprises a microfluidic channel 32 having an inlet 34, an outlet 36 and a constriction in which a detection region 38 is provided. The device 30 further comprises a first ground electrode 40 and a second working electrode 42 which together form a first electrode set 44 of the first particle sensor.

[0188] The device 30 further comprises second particle sensors 46. The second particle sensors 46 comprise a second electrode set having a ground electrode and a working electrode. In some embodiments, the ground electrode of the second electrode set 47 can be shared with the first working electrode set. That is, the plurality of particle sensors can comprise n+1 electrodes, where n is the number of particle sensors, and a shared ground electrode is used. In other embodiments, at least some of the particle sensors can comprise a working electrode set comprising a ground electrode and a working electrode. In the illustrated embodiment, the second particle sensor 46 comprises a working electrode 48 which forms a second electrode set 50 with the first ground electrode 40. A further second particle sensor 52 is provided which comprises a working electrode 54 which forms a third electrode set 56 with the first ground electrode 40. Each of the second particle sensors 46, 52 is provided with a flow regulator 55, 56.

[0189] The electrical signals output by each working electrode set can be referred to as providing a channel of pulse data associated with the corresponding particle sensor. The pulse data indicates the current between the electrodes of the electrode set. The current is caused by a particle passing through the corresponding particle sensor. Thus, each current pulse represents a particle passing through the particle sensor. Since each sensor is independent, the pulse data of multiple channels can be output from the sensors at least partially simultaneously. That is, the pulses representing current can be at least partially simultaneous among the multiple channels. Thus, an apparatus for use with a particle sensor according to embodiments of the application can store the pulse data for subsequent analysis, or can include multiple apparatuses that receive and analyze the pulse data simultaneously.

[0190] Figure 20b A second particle sensor (60) according to a further embodiment of the application is shown. The second particle sensor (60) includes a body (62) having external threads (64) for connecting the second particle sensor (60) to a base of the apparatus. The body has an internal channel (66) in which a nanoliter pipette (68) is located. When the second particle sensor (60) is connected to the base of the apparatus, the internal channel (66) is connected to the main microfluidic channel of the first particle sensor, thereby forming a flow path between them. A tube (70) is also housed with the body (62) and is in fluid communication with the internal channel (66). A portion of the tube (70) extends out of the body (62). The tube (70) can be used as the primary fluid outlet of the apparatus.

[0191] For example, in some embodiments, the flow rate through the primary fluid outlet of the microfluidic channel, and optionally each tube present in the series of second particle sensors, can be independently controlled or stopped. Thus, the system can be configured so as to allow all of the sample to flow out of a single outlet, such as the tube (70).

[0192] Figure 20c The passage of fluid through the apparatus is shown. A fluid sample including large particles (80) and small particles (82) passes through the microfluidic channel of the first particle sensor, as indicated by arrow A. Some of the fluid enters the second particle sensor (84). The second particle sensor (84) includes a nanoliter pipette (86) within a housing (88). The nanoliter pipette (86) has a hole (90) formed therein that is sized so that only the small particles (82) are able to pass through the hole (90) and be detected. The large particles (80) travel around the nanoliter pipette in an external flow stream, as indicated by arrow B, and exit the apparatus as waste.

[0193] Larger particles are prohibited from entering the nanopore (90) of the second particle sensor, such that the fluid flow rate through the nanopipette (86) ("internal flow") is lower than the flow rate that bypasses the nanopipette ("external flow"). In this way, the nanopore (90) of the second particle sensor is constantly flushed by the liquid flow, which pushes larger particles away from the nanopore, thus preventing clogging of the nanopore. The flow rate of liquid through the nanopipette also contributes to increased sensitivity. If the flow rate through the nanopipette (86) is increased, more particles can be detected. Likewise, if the flow rate through the nanopipette (86) is stopped, the second particle sensor will remain inactive until the flow is switched back on.

[0194] Figure 20d - g shows how the fluid flow through the second particle sensor can be adjusted. In Figure 20d , fluid from the main microfluidic channel is allowed to flow into the second sensor and through the nanopipette at the flow rate shown by arrow Fl. Fluid that does not pass through the nanopore flows around the nanopipette, as shown by arrow Wl.

[0195] Figure 20e shows how the flow rate through the nanopipette can be increased using a flow regulator (not shown), as shown by arrow F2. In this case, the fluid flow around the nanopipette, rather than through it, is reduced, as shown by arrow W2.

[0196] In Figure 20f , the fluid flow through the nanopipette is reversed (arrow F3), thus preventing the translocation of particles through the nanopore. In Figure 20g , all flow through the second particle sensor is stopped, such that no liquid is travelling through or around the nanopipette.

[0197] The workflow is as shown in Figure 21 . The device (1) can also be used without flow, for example, when the user wants to inject a sample / material into the sensor for analysis, for example, blood, milk, or nanomaterials / nanomedicines. The measurements can be used to characterise the physical properties of the material, or the properties of the material are related to the sample matrix, and the physical properties of the particles act as a diagnostic / analytical signal.

[0198] The analysis package described below can be implemented as a computer program configured to perform one or more of the steps described above when run on a processor. Typically, such a computer program can be configured to perform at least the following when executed by a computer processor: receiving as input data representative of electrical resistance pulses corresponding to particles, the electrical resistance pulses being obtained by a particle sensor as described herein; comparing the received data with pre-stored characterising shape and size data representative of known shape and size particles; based on the comparison, determining one or more of the shape and size of at least one particle; and outputting the determined one or more of the shape and size of the particle. It will be appreciated that examples of the computer program disclosed herein can be implemented in hardware, software, or a combination of hardware and software. Any such software can be stored in the form of volatile or non-volatile storage, for example, storage devices such as ROM, whether erasable or rewritable or memory, for example, RAM, memory chips, device or integrated circuits or storage can be stored in the form of optical or magnetic media, for example, CD ROM, DVD or magnetic disk or tape. It will be appreciated that the storage devices and storage media are embodiments of non-transitory machine-readable storage that are suitable for storing a program or programs implementing a method embodying the disclosure and that the results provided by such a program can be output to an external device for example, visual display or stored in the form of a computer readable storage medium for further processing or storage. It will be appreciated that examples disclosed herein can provide a program comprising code for implementing any of the methods disclosed herein, and a machine-readable storage storing such a program. Still further, examples disclosed herein can be conveyed electronically via any medium (e.g. communication signal) over a wired or wireless connection and examples appropriately encompass these.

[0199] In another aspect of the application, there is provided a computer implemented method comprising:

[0200] receiving as input pulse data representative of electrical current pulses corresponding to particles;

[0201] comparing the received pulse data with pre-stored characterising shape data representative of particles of known shape;

[0202] based on the comparison, determining the shape of the at least one particle; and

[0203] outputting an indication of the determined particle shape.

[0204] In some embodiments, the electrical current pulses can be obtained from a particle sensor as defined herein or a device as defined herein. However, in other embodiments, the pulse data can be obtained from other types of resistance pulse sensor (RPS).

[0205] In some embodiments, the pre-stored characterising shape data comprises one or more spline coefficients representative of particles of known shape.

[0206] In some embodiments, the one or more spline coefficients are one or more spline coefficients of a b-spline representative of the known shape of the particle.

[0207] In some embodiments, the one or more spline coefficients are a predetermined set of spline coefficients representative of the known shape of the particle.

[0208] The pre-stored shape data representative of shape data can be for normalised particle sizes, wherein the computer program is configured to normalise the received data.

[0209] In some embodiments, the pre-stored shape data comprises an indication of a width of the resistance pulse.

[0210] In some embodiments, the pre-stored shape data comprises an indication of a maximum depth of the resistance pulse.

[0211] The data representative of the resistance pulse corresponding to the particle can comprise data corresponding to a particle passing through a sensor. In some embodiments, the data representative of the resistance pulse corresponding to the particle comprises data corresponding to a particle passing through a nanopore.

[0212] In another aspect, there is provided a computer implemented method comprising:

[0213] receiving, as input, pulse data representative of current pulses corresponding to particles of a known shape;

[0214] determining a model from the received data indicative of particles of a known shape.

[0215] In some embodiments, the current pulses can be obtained from a particle sensor as defined herein or a device as defined herein. However, in other embodiments, the pulse data can be obtained from other types of resistance pulse sensor (RPS).

[0216] In another aspect, the present invention provides a computer implemented method of characterising a nanoparticle comprising:

[0217] receiving, as input, pulse data representative of current pulses corresponding to particles of a known shape;

[0218] classifying the pulse data as corresponding to one of a plurality of predetermined nanoparticle types using a pulse shape model.

[0219] Figure 34A method 3400 according to embodiments of the application is shown. The method 3400 is a method of processing pulse data according to embodiments of the application. The pulse data can be received from a particle sensor according to embodiments of the application. The method 3400 can be used to pre-process the pulse data. The pre-processed pulse data can be used with methods according to other embodiments of the application, such as described below with reference to Figure 35

[0220] In block 3410, pulse data is received from a particle sensor, such as described at least with reference to Figure 20a The pulse data can be received from each of one or more particle sensors, such as a first particle sensor and a second particle sensor of a particle sensor. In some embodiments, the pulse data is received from the particle sensor in a plurality of channels. The pulse data can be stored in a memory of a device, such as a computer, for receiving the pulse data. The pulse data can be in the form of raw signal pulse data as shown in the upper portion of Figure 24

[0221] In block 3420, pulses are extracted from the pulse data. In block 24, one or more individual pulses are extracted from the raw pulse data. Each pulse can be identified according to a significant deviation from a baseline of the raw pulse data. Each pulse can include a predetermined number of data points or data samples.

[0222] In block 3430, each pulse is aligned to include a predetermined number of data samples. That is, the predetermined number of data samples of the pulse are aligned to capture the predetermined number of data samples. In one embodiment, each pulse includes 301 data samples, although it should be appreciated that other numbers of data samples can be used. In particular, an odd number of data samples can be used so that the number of data samples on either side of a middle data sample can be the same, i.e., data sample 151 can be selected as the middle data sample and the pulse includes 150 data samples on either side. The number of data samples is shown in Figure 24

[0223] In block 340, the pulse data is de-trended. De-trending means removing any trends from the sampled pulse data of each pulse, such as an increase or decrease in the amplitude of the raw pulse data.

[0224] ​​​In some embodiments, the method 3400 includes normalizing the amplitude of each pulse. Normalizing the amplitude means that each pulse is determined to have a predetermined amplitude, such as a normalized amplitude or depth of 1. In this way, the size of each pulse is removed from the pulse data. Normalizing the pulse data allows the shape information of the pulses to be separated from the size information. By using normalized pulse data, each pulse and the corresponding particle can be classified according to its shape, as will be explained.

[0225] In block 3460, the pulse data is output. The pulse data can be output by being stored in a data storage medium, such as a memory of a device that performed the method 3400.

[0226] Embodiments of the invention include a method of creating a classifier for classifying particles. Pulse data can be received from a particle sensor according to embodiments of the invention. However, the method can be used on pulse data obtained from other sensors. Embodiments of the method 3500 can be performed by a computer system, such as the computer system 1000 shown in Figure 35 .

[0227] The method includes a block 3510 of receiving pulse data. The pulse data can be received from a particle sensor, such as the particle sensor 1000 shown in Figure 34 The method can be received in the form of pre-processed pulse data. The pulse data can be received from a memory of a computer system, such as the computer system 1000 shown in

[0228] In block 3520, the pulse data is quantified. Quantifying means determining one or more quantities or measures of each pulse. The pulse data can be quantified by determining one or more statistics of each pulse. As shown in Figure 24 one or more of the following: an indication of the pulse amplitude of each pulse pi, p2... p n an indication of the width of each pulse and one or more measures of the shape of each pulse.

[0229] The indication of the width of each pulse or the lock width can be determined as one or more fractions of the height of each pulse, such as 0.75, 0.5, 0.25 of the pulse height, as shown in Figure 24 The one or more measures of the shape of each pulse can include fitting a spline to the pulse data. In some embodiments, the pulse data is normalized before fitting the spline. The spline can be a quadratic spline. In some embodiments, the spline is a quadratic spline with 25 fixed knots and three fixed points (24 coefficients). The spline coefficients provide a mathematical description of the denoised pulse shape. In the case of fitting a spline to normalized pulse data, the spline indicates only the shape of each pulse.

[0230] In some embodiments, the method 3500 includes a block 3530 of checking consistency of the data. To verify correct operation, pulse data is acquired for predetermined particles of known size, shape, and composition. In block 3530, statistics corresponding to pulses of predetermined particles are checked against expected values. If the statistics are outside one or more thresholds for predetermined particles, the method 3500 can stop. Otherwise, the method proceeds to block 3540.

[0231] In block 3540, a classifier is constructed to distinguish between particles having different characteristics. For example, the classifier can be constructed to distinguish between particles of different shapes, such as spheres and rods, although other kinds of particle shapes are contemplated. In other examples, the classifier can be constructed to distinguish between particles of different sizes and / or materials. To construct the classifier, pulse data is obtained for pulses corresponding to a group of particles corresponding to particles that need to be distinguished as calibration pulse data. For example, a first group of pulse data can be obtained for spherical particles, and a second group of pulse data can be obtained for rod-shaped particles. Given the group of pulse data, a predictive classification model is constructed using a regression method. The regression method can be a penalized regression method. In one embodiment, the model uses the maximum depth of the pulse, in another embodiment, the model uses the pulse width, in another embodiment, the method uses spline coefficients for the pulse data, which can also include the maximum depth of the pulse, in another embodiment, the model uses spline coefficients for normalized pulse data, where the spline coefficients can be spline coefficients of a b-spline, and in another embodiment, the model uses a fixed set of spline coefficients from normalized pulse data. The fixed set of coefficients can be coefficients 9, 10, 16, 17, 18, and 19, which have been found to be particularly effective, although it will be appreciated that other coefficients can be used. In this way, the classifier is constructed to be able to distinguish between different classes of particles, such as particles of different shapes, based on their pulse data output from the particle sensor, which can be in accordance with an embodiment of the invention.

[0232] In block 3550, the constructed classifier is output, such as by being stored in a memory of a computer system that executes the method 3500. The classifier is stored for later use in classifying particles according to their respective pulse data.

[0233] Figure 36 A method is shown that uses a classifier according to an embodiment of the invention. The classifier can be the one produced by the method 3500 of Figure 35 . The method 3600 is a method of characterizing particles from received pulse data. The method 3600 can be performed by an apparatus such as the one described below with reference to Figure 37 .

[0234] Method 3600 includes a block 3610 of receiving pulse shape data from a particle sensor. The particle sensor can be a particle sensor according to embodiments of the application as described above. The pulse shape data can be transmitted as an electrical signal received at an interface of a device performing method 3600. As described above with reference to Figure 20a The pulse shape data can be received in one of a plurality of data channels. The pulse shape data can include a plurality of data points at various points in time indicative of current between electrodes of an electrode set, such as 301 data points as described above.

[0235] Method 3620 includes a block 3620 of determining a shape of a particle. Block 3620 can include classifying the received pulse data 3625 as corresponding to one of a predetermined plurality of nanoparticle types using a pulse shape classifier provided by method 3500, method 3500 being stored in a memory of the device. In some embodiments, block 3620 includes comparing the received pulse data 3625 to pre-stored characterization shape data representative of particles of known shape. Further, in some embodiments, block 3620 includes determining, based on the comparison, a shape of at least one particle corresponding to the received pulse shape data. The classifier can output an indication of the pulse data corresponding to one of a predetermined number of particle classes, such as particles having a predetermined shape or a predetermined shape and / or size.

[0236] In block 3630, an indication of the determined shape of the particle is output. Block 3630 can include storing the indication in a memory of the device performing the method.

[0237] Figure 37 A system 3705 according to embodiments of the application is shown. System 3705 includes a control unit 3710 and a particle sensor 3720. Particle sensor 3720 can be a particle sensor according to embodiments of the application, such as described above. In particular, particle sensor 3720 can include a plurality of particle sensors, such as a first particle sensor and a second particle sensor. Each particle sensor can output data on a respective data channel 3725, 3726, as Figure 37 shown, including two data channels.

[0238] Control unit 3710 is arranged to receive pulse data 3725, 2726 indicative of current between a first electrode and a second electrode of a working electrode set associated with a microfluidic channel when a particle traverses the microfluidic channel.

[0239] The control unit 3710 comprises a processing device 3730 and a memory 3740. The processing device 3730 is arranged to execute computer readable instructions which can be stored in the memory 3740. The processing device 3730 is arranged to execute a method according to embodiments of the application as defined by the computer readable instructions. The memory 3740 can be arranged to store data used in the method. Received pulse data can be stored in the memory 3740. In particular, the memory 3740 can store pulse data 3725 received from the particle sensor 3720. The memory 3740 can also store one or more models 3750 or classifiers used in the method. The one or more classifiers 3750 can be one or more pulse shape classifiers 3750, such as described above, for classifying the pulse data 3725 as corresponding to one of a plurality of predetermined particle types. The control unit 3710 can be arranged to execute a method 3600 according to embodiments of the application, such as Figure 36 is shown.

[0240] The presence of the analyte causes the release of particles into the solution. The number of released particles is related to the concentration of the analyte. Each analyte causes the release of nanoparticles of a specific shape or size. The released particles in the solution are identified and counted by the sensor which provides fast quantification.

[0241] The surface of a material (e.g. a glass slide, a cellulose membrane, etc.) is functionalized with DNA molecules. DNA 2 is complementary to DNA 1 immobilized on the particle. Via the formation of double stranded DNA between DNA 1 and DNA 2, the particle is immobilized on the surface. The presence of analyte-1 in the solution causes the disruption of the interaction between DNA 1 and DNA 2 (where DNA 1 or DNA 2 can be an aptamer for analyte-1). The interaction between analyte-1 and DNA 1 / 2 causes the release of the particle into the solution. The number of particles in the solution is calculated via the RPS sensor.

[0242] As the surface can be loaded with many different sizes or shapes of particles, multiple analytes can be quantified simultaneously. Each particle is held on the surface via the formation of dsDNA. Here DNA 1 and DNA 2 are complementary and there is no interaction between DNA 3 and DNA 4. Likewise, DNA 3 and DNA 4 are complementary and particle 2 is held on the surface via the interaction of DNA 3 and DNA 4. In the presence of analyte 1, there is a specific interaction between analyte 1 and DNA 1 / 2, causing particle 1 to be released into the solution. In the presence of analyte 2, there is a specific interaction between analyte 2 and DNA 3 or DNA 4, causing particle 2 to be released into the solution. The RPS sensor of the present application can measure and identify the number of different particles.

[0243] The presence of flow, and the ability to characterize a wide range of particle sizes, enhances diagnostic applications and is a unique combination of diagnostic chips and RPS sensors.

[0244] All of the features disclosed in this specification (including any accompanying claims, abstract and drawings), and / or all of the steps of any method or process specified in this specification, can be combined in any combination, except combinations where at least some of such features and / or steps are mutually exclusive.

[0245] Unless otherwise indicated, each feature disclosed in this specification (including any accompanying claims, abstract and drawings) can be replaced by alternative features that are functionally equivalent or similar in some way. Thus, unless otherwise indicated, each feature disclosed is only an example of a generic series of equivalent or similar features.

[0246] The application is not restricted to the details of any foregoing embodiments. The application extends to any novel one, or any novel combination, of the features disclosed in this specification (including any accompanying claims, abstract and drawings), or to the steps of any method or process so disclosed.

[0247] Throughout the description and claims of this specification, the words "comprise", "contain", "include" and "comprising", "containing", "including" and the like are to be construed in an inclusive sense as opposed to an exclusive or exhaustive sense, and are intended to mean that items include the listed items but not to the exclusion of other items. Throughout the description and claims of this specification, unless the context requires otherwise, the singular will include the plural and vice versa. In particular, where used in the specification and / or claims, the terms "comprise", "comprising", "include", "including" and the like are taken to specify the presence of stated features, integers, steps and / or components but not to the exclusion of others. Throughout the description and claims of this specification, the term "about" preceding a number or other such quantifier means that the number or other such quantifier can vary from the recited number or other such quantifier by up to 1%, 5%, 10%, 15%, 20%, 25%, 30%, 35%, 40%, 45%, 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85%, 90%, 95%, or 100%.

[0248] All of the features disclosed in this specification (including any accompanying claims, abstract and drawings), and / or all of the steps of any method or process specified in this specification, can be combined in any combination, except combinations where at least some of such features and / or steps are mutually exclusive. The application is not restricted to the details of any foregoing embodiments. The application extends to any novel one, or any novel combination, of the features disclosed in this specification (including any accompanying claims, abstract and drawings), or to the steps of any method or process so disclosed.

[0249] Example 1 : Shape analysis

[0250] Materials and methods

[0251] Materials: Three types of nanoparticles were used in this study: carboxylated polystyrene particles (200 nm in diameter, denoted as CPC200, from Izon Science, Christchurch, NZ), carboxylated polystyrene particles (158 nm in diameter, denoted as PS150, purchased from Bangs Laboratories, Inc, Indiana, USA), and nanorods purchased from CMD Ltd. Iron oxide nanorods were provided by CMD Ltd, Cardiff, UK.

[0252] Particle preparation: Carboxyl groups were added to iron oxide nanorods using PEI and PAAMA (poly(acrylic acid-co-maleic acid)) (PEI), Mw 750000 g mol -1 Analytical standard, 50% wt., P3143, poly(acrylic acid-co-maleic acid) (PAAMA), Mw ~ 3000 g mol -1 50% wt., 416053, purchased from Sigma Aldrich, UK. Reagents were prepared in purified water with a resistance of 18.2 MW cm. Particles were taken from a stock solution (50 pL) and suspended in PEI (1 mL, 5% in H20). The solution was left on a roller for 30 minutes. The solution was centrifuged at 10000 rpm for 5 minutes and the PEI solution was removed from the particles and replaced with water. The sample was vortexed and sonicated until the particles were fully dispersed. This washing step was repeated twice to ensure that all excess PEI had been removed. The PEI coated particles were suspended in PAAMA (5% in 50 mM NaCl) for 30 minutes and left on a roller. The same method was used to remove excess PEI. The particles were then stored in water at 2-4 °C. Carboxyl polystyrene particles (158 nm) from Bangs Laboratories, US and CPC200 (200 nm) from Izon Science, NZ were used without modification. The particles were diluted using a 50 mM potassium chloride solution (KCI, >99%, P / 4240 / 60 purchased from Fisher Scientific, UK).

[0253] Method validation: To develop the model, we used comparable volumes of particles, i.e. PS150 and carboxyl-coated nanorods. For calibration of the model, we used samples of pure nanorods and (small, PS150) nanospheres. Then, we recorded >500 events with solutions containing mixtures of nanorods to (large, CPC200) nanospheres. To generate the mixtures, we first diluted the nanorod and nanosphere particle solutions such that each sample had a comparable particle count rate, i.e. number of pulses per unit time. Thus, when mixed in equal amounts, the signal ratio of nanorods to nanospheres was equal to 1. We do not assume here that the concentrations of nanorods and nanospheres are equal, only that the number of translocations of each type of particle is comparable. Different volumes of stock solutions were mixed and then the known pulse ratios were obtained.

[0254] RPS setup: All measurements were performed using a qNano (Izon Science, NZ) in combination with tunable nanopores (NP150s) and the data capture and analysis software Izon Control Suite v.3.1. The lower fluidic chamber contained the electrolyte (75 μί). The particles were suspended in the same electrolyte and placed in the upper fluidic chamber (40 μί). Before analysis, all samples were vortexed and sonicated for 30 s. After each sample run, 40 μί of electrolyte was placed in the upper fluidic chamber and the system was washed several times with different pressures to ensure that there were no residual particles and thus no cross-contamination between samples. Since multiple pores were needed for the whole set of experiments, we made sure that they had comparable pore sizes as much as possible. For this, we used pores of the same size provided by the manufacturer. Due to the polyurethane material and the manufacturing process, the size was expected to vary. To compensate for this, we matched the baseline current within 5% of 110 nA and ran control samples, blanks and calibration beads to allow comparison between data sets. In the absence of information about the exact size of the PU pores, it is possible to calibrate the response of the nanopores using particles of known size and volume. Here we used polystyrene particles with an average diameter of 235 nm. Since the RPS blockade amplitude Δί p The signal is linearly related to the particle volume, so a single point calibration is sufficient.

[0255] In the PU setup, there are two main modes of transport. First is electrophoresis. Since all particles have carboxyl surface chemistry, they will travel towards the anode on the side of the membrane opposite to the sample. The second is convection, caused by fluid flow under the influence of gravity. Here, the pores have a vertical orientation and the sample is placed on top of the membrane.

[0256] Electron microscopy setup: Polystyrene nanosphere-shaped nanoparticles and iron oxide nanorods were diluted in deionized water and 5 μΐ of sample was dropped onto a copper plate and allowed to evaporate at room temperature.

[0257] Mathematical model approach: Data obtained from the instrument software, including resistance data and lockout output files for the entire run, were analysed in R 3.4.1 using the RStudio interface. First, a method was developed to repeatedly extract each pulse. This involved extracting the 1001 time points containing the detected lockout event, identifying the lockout minimum, and then extracting 301 time points with the pulse minimum at the 151 point. This resulted in 150 data points before and after the pulse amplitude. To account for any baseline drift and noise variation, the first 50 and last 50 time points were used to detrend each pulse.

[0258] From each run, a file was created consisting of all extracted pulses, as well as a file of extracted pulses normalised to a depth of 1. Finally, the extracted pulses and normalised extracted pulses were each approximated by a quadratic b-spline with fixed knots (using the R package cobs), and the coefficients saved in two additional datasets.

[0259] The model was built in the R base package (glm) and the package glmnet (for lasso penalised regression). Two calibration runs were used to train the model using either nanosphere or nanorod solutions alone. From the calibration data, two test solutions were run, again containing either nanorods or nanospheres alone. This process was repeated for each nanopore, and a total of 5 PU nanopores were tested and their ability to predict the shape of each nanoparticle recorded in the run according to their shape using the R package pROC.

[0260] Simulation Method: The finite element method (FEM) was used to predict the pulse shape caused by nanospheres and nanorods passing through a conical aperture on an axial trajectory. Commercial software Comsol Multiphysics 5.2 was used to address the underlying electrostatic problem governed by the Laplace equation Δφ = 0. It was assumed that the channel walls and the boundary conditions of the particles were insulating, meaning that the electric field component perpendicular to the boundary must vanish at the surface. The aperture was assumed to be conical with a length of 250 μm, a macroaperture diameter of 52 μm, and a microaperture diameter of 666 nm. The aperture length and macroaperture diameter were extracted from SEM data, while the microaperture diameter was calculated from the baseline current (applied voltage 1.46 V, solution conductivity 0.667 S / m) using a model described elsewhere. The nanorods were 450 nm long and 90 nm wide, both extracted from SEM data. Nanospheres of 158 nm were selected to match the volume of the nanorods. Simulations of nanorods and nanospheres were repeated at a sufficient number of points along the pore axis to extract the current as a function of the particle's center position. Since the primary transport mechanism of the parameters used in the experiments is the hydrodynamic flow generated by the applied pressure head, and it can be assumed that the particles follow the fluid flow, the position-current relationship can be directly scaled to a time-current relationship.

[0261] Results and discussion

[0262] We have developed a method for analyzing nanoparticles that allows us to classify individual particles based on their shape, thereby enabling us to determine the ratio of particles of different shapes in a solution. This method, called the Resistance Pulse Sensing Logistic Regression Model (RPS-LRM), links resistance pulse data with signal processing and shape prediction statistical algorithms to classify individual particles according to their shape as they pass through a pore. Two sets of experiments were conducted.

[0263] 3.1 Using RPS signals to identify particle shapes

[0264] The first experiment involved two types of nanoparticles, nanorods ( Figure 22 bi), and nanospheres ( Figure 22 (bii). To develop this method, we chose materials with approximately equal volumes. While efforts were made to precisely match the volumes of the nanorods and nanospheres, it should be noted that they are not exactly equal. Interestingly, excellent agreement was observed between the RPS and S / TEM data. This simple observation is important because previous RPS studies of nanorods showed that the tumbling rate of smaller nanorods affects the blocking amplitude, leading to an overestimation of the nanorod size. Here, we do not rule out the ability of nanorods to tumble in solution, but the calculated number of rotations is <0.5 rpm when they pass through the sensing region of the nanopore. Therefore, their direction of approach to the pore may be the same as their direction of departure from the sensing region.

[0265] A computational model was established to determine the theoretical pulse shape of nanospheres and nanorods as a function of particle position within the pore. Figure 23 a shows the simulated normalized pulse shapes of nanorods and nanospheres as they translocate the nanopore. A clear difference in the current-position relationship of nanorods and nanospheres is observed as the particle approaches the sensing region on the left-hand side. The predicted difference is smaller when the particle is present in the sensing region on the right-hand side. Note that while we have modeled nanorods aligned with the axis of the nanopore (see experimental section), it is not possible to predict the preferred orientation of the nanorod as it approaches the opening. To compare, Figure 23 b shows the average measured pulse shapes of >500 nanospherical and nanorod particles. Note that the difference in pulse sharpness between the predicted and observed pulses is due to the difference in the horizontal axis and the number of data points. However, in terms of the comparison between the pulses of the two particles, the average measured pulses show similar trends to the simulations, which indicates that it should be possible to determine which type of particle is detected by using information about the pulse shape. To further illustrate the sensitivity of this technique, nanorods of different aspect ratios were run in the entire setup. Figure 23 c shows the average measured pulse shapes of >500 particles of different aspect ratios. In the current setup, nanorods with an aspect ratio lower than 2 cannot be distinguished from nanospheres.

[0266] 3.2 Measuring the proportion of nanorods in a mixed solution.

[0267] However, RPS is a single-particle analysis technique, capable of size and concentration analysis, and this is also applicable to nanorods. A signal is recorded as each nanoparticle passes through the first or second particle sensor. Therefore, the use of an average of hundreds of pulses for shape classification can be misleading. If a sample contains both nanospherical and nanorod-shaped particles, the average signal can lead to a false classification of the material or a false positive, depending on the ratio of nanorods to nanospheres. Having shown that the PU pore can detect the average difference between nanorods and nanospheres, the next step is to apply an analysis and classification process to each pulse and particle that moves through the nanopore. For each experiment, two runs were performed for each type of nanomaterial. Using the analysis package described below, individual pulses of 301 time points centered on the pulse minimum were extracted from the raw data files (see Figure 24). Each pulse was then aligned and detrended. The 301 time points were chosen to balance the desire to have a stable baseline on either end of the pulse in order to detrend and align the signals, but to avoid capturing two pulses in most cases. The pulses were then normalized to a depth of 1 in order to separate the size information from the shape information. This helps to illustrate the versatility of the technique, as any size / volume of particle can be shape classified in future applications.

[0268] Shape data in the form of spline coefficients for the normalized pulses was obtained using a quadratic spline with 25 fixed knots and three fixed points (24 coefficients). The spline coefficients provide a mathematical description of the denoised pulse shape. Note that standard Fourier denoising methods are not applicable to this setup, as the pulse length is the same order of magnitude as the length scale of the noise oscillations. Figure 24 A schematic of this is shown, but note that the spacing of the junctions is not uniform as shown; more junctions are used in areas where the current changes rapidly than in areas where it is most stable at the ends.

[0269] Figure 24 Recorded pulse data is shown. Each pulse, containing 301 data points, was isolated. Pulse analysis was done by one of five models. A) Pulse amplitude, B) lock-in width at fractions of 0.75, 0.5, 0.25 of the pulse height. C) Non-normalized spline fit and D-E) spline fit of the normalized pulse amplitude.

[0270] Figure 25 Visualization of key spline segments that identify the particle shape as determined by a multiple t-test with Bonferroni correction is shown.

[0271] Figure 26 (a) shows a plot of the ratio of rods to spheres predicted using data model D versus the known ratio.

[0272] Five different types of logistic regression models can be applied to the data from one pure rod run and one pure sphere run (training data). To determine the performance of each model in the development analysis, we built each model from data from five holes and tested the particle classification from the second pair of runs (test data) from each hole. The first model A uses only the maximum depth of the pulse, which has been shown to be approximately proportional to the particle volume. The second model B uses the vector of blocked data (pulse width) described elsewhere. The third model C uses spline coefficients on the extracted signal. This also includes the size data from model A. The fourth model D uses spline coefficients of a b-spline to fit the extracted signal after the extracted signal has been normalized to a depth of 1. This model only examines the shape of the resistive pulse because the size (depth) data has been removed. Models B, C, and D are applied to each hole on the training data using lasso penalized regression (glmnet package in R). This means that the variables (i.e. coefficients non-zero) selected for these models are different for different holes. The final model E uses a fixed set of spline coefficients from the normalized signal data based on the results of a multiple t-test (coefficients 9, 10, 16, 17, 18, and 19) and fits the model using standard logistic regression. The signal processing and modeling methods are shown in Figure 24

[0273] The performance of the models is compared using the maximum sensitivity + specificity of the ROC curve along the test data for each model. Here sensitivity is the percentage of nanospheres correctly classified and specificity is the percentage of nanorods correctly classified and so the maximum score is 200. For each set of data (two runs of each particle type), there are four choices of training data and test data (two choices of nanosphere runs and two choices of nanorod runs). We refer to the models built and tested in these four different ways as replicates. The scores of the models for these four replicates across the 5 nanoholes are recorded in Table 1. It is worth noting that for all models, each hole must be calibrated before the particle shape can be characterized, i.e. a sample of pure nanorods or nanospheres is first run through each hole. It is also worth noting that the calibration on one hole does not predict the particle shape on another hole, data not shown. We attribute this to differences in the hole structure and reproducibility of the manufacturing process, as well as its stability over the course of the experiment.

[0274] To determine which parts of the normalized spline model are key to identifying the particle shape, we performed a multiple t-test with Bonferroni correction on each spline coefficient for all pairs of rod and sphere data for each hole (4 pairs per hole). Figure 25 ​Key spline segments were visualized, i.e. the parts of the curve where the shape display spheres and rods showed significant differences. For all pores, the current drop (i.e. when the particle enters the pore) had splines (green segments) that displayed important features of the particle shape. In some cases, pores 2, 3, and 4, the second part of the pulse (i.e. the part recorded as the particle traverses the pore) also showed significant ability to measure particle shape. This also suggests that pores of different shapes can have better overall ability to distinguish particle shape than conical pores.

[0275] We have shown that model C, which involves spline coefficients of non-normalized signals, performs better than models involving data traditionally extracted from RPS depth and pulse width. Specifically, for model A (depth data only), the overall average score for each combination of training and test data for the five pores was 144 ± 15, with a standard deviation of 15, and approximately 74% of spherical particles were identified as spheres and 70% of rod particles were identified as rods. Model B had an average score of 129 ± 20. Interestingly, this was worse than the model using depth alone, which suggests that pulse width data is not an effective way to capture shape data, and that the additional width data can actually cause the model to overfit, rather than improve predictive performance. The high standard deviation also suggests this, which indicates that the width data collected by the instrument is not a reliable measure of shape.

[0276] Table 1. Summary sensitivity and specificity values for each model and nanopore tested

[0277]

[0278]

[0279] In contrast, model C had an average score of 149 ± 13 at the optimal classification point. Thus, we see that, on average, a classification model that includes pulse shape produces a better classification model of nanoparticle shape than depth alone, and the standard deviation is also slightly lower, which suggests that the method of building the model is more reliable. Model C also has the strong advantage that it removes the need for any prior knowledge of particle size in the material analysis.

[0280] It is worth noting that even when size information is removed by pulse normalization, shape information is still retained. In particular, model E had an average score of 141 ± 14, corresponding to approximately 75% sensitivity (correct sphere classification) and 66% specificity (correct rod classification). Overall, in all cases, the specificity was lower than the sensitivity at the maximum of the summary, which we attribute to the variation in rod orientation producing a larger range of possible pulse shapes, and thus making it more difficult to characterize with a dichotomous model than spheres.

[0281] While particle shape identification of pure samples is very powerful, the ultimate demonstration of the RPS-LRM functionality is to mix together different ratios of spheres and rods and ask the model to predict the ratio of rods to spheres in the mixture. Although its performance is slightly worse (average 133 ± 13), Model D is used here instead of Model E to illustrate that without knowledge of the key spline vectors, any user can download the R code here and apply it to RPS experiments. The choice of different sized particles in this experiment has two reasons. First, this allows us to be able to use size to determine whether each given pulse corresponds to a rod or a sphere, thus enabling us to obtain a basic fact about the particle shape. Recall that Model D uses only shape data from the depth-normalized pulses, so we can compare the performance of this model for particle shape detection to the basic fact given by size. Thus, the particle ratio predicted using size and the classification ratio by Model D should be comparable. The second reason is to demonstrate that a model fit using one size of spheres can be used to determine the shape of a different size of spheres, thus further demonstrating that shape can be studied independently of size.

[0282] These samples were analyzed. Model D gives a linear response, that is, as the rod ratio increases, the model identifies an increase with an impressive linear relationship (R-squared value of 0.9264). The gradient here is greater than 1 due to imperfect specificity and sensitivity reported previously. Note that the presence of two particles in solution does not affect the ability of the model to predict particle shape. This data can also be used to estimate the possible best sensitivity and specificity for the current experimental setup when all particles in the test and training sets are run under identical conditions. Note that for the specific data of the particular well tested, we get a model with a sensitivity of 72% and a specificity of 91% using Model D. That is, 91% of the spherical (large) particles are identified as spheres by their shape data alone and 72% of the rod-shaped (small) particles are identified as rods by their shape data alone Figure 26 b).

[0283] Example 2: Tunable 3D printed microfluidic resistive pulse sensor

[0284] Materials and Methods

[0285] Chemicals and reagents: CPC2000, 2 micron carboxylated polystyrene calibration particles were obtained from Izon Science Ltd., 10 micron and 20 micron carboxylated polystyrene calibration particles labelled CP10M and CP20M were obtained from Izon Science Ltd. 30 micron carboxylated polystyrene particles, catalogue number 84135 were obtained from Sigma-Aldrich, Potassium chloride was obtained from Fisher Scientific UK, >99% product sample number: P / 4240 / 60, RS component, Acc silicone QSil216, product sample number: 458-765, part number: QSil 216 were obtained, Algal samples, Aurantiochytrium Mangrovei (spheroids), product sample number: RCC893 and Navicula ramosissima (rods), product sample number: RCC5374 were obtained from Roscoff Culture Collection, Isopropanol was obtained from VWR.

[0286] Data analysis: Data analysis was performed within the data analysis module of the Izon control suite, pulse shape extraction was performed using Molecular Devices clampfit version 10.7 and pulse shape analysis was performed using custom R code.

[0287] Equipment assembly: To assemble the equipment, the lid was secured to the base via six machine screws located at each corner and two screws in the middle of the equipment. The screws were tightened into place using nuts. HPLC fittings were attached into each of the threads in order to house the inlet, electrode and outlet of the pump. Once fully assembled, the equipment was placed into a custom Faraday cage and the electrolyte solution was pumped into the equipment.

[0288] Equipment printing: Both the lid and base of the equipment were printed on an Asiga Pico HD27 UV using FORMlabs clear resin. The files were converted from the CAD software Siemens NX11 to STL and prepared for printing using the Asiga Composer software. Once printing was complete, the parts were cleaned and post cured using a UV light box.

[0289] PDMS gasket: The PDMS gasket was formed by mixing the A and B parts of QSil 216 in a ratio of 10: 1. The lid was placed into a petri dish with the ridge facing the bottom. Uncured PDMS was poured around the edge of the lid ensuring that the whole lid was covered by the ridge and that no large air pockets were left. The PDMS was then cured at 70 °C for one hour or until set.

[0290] SEM / EDS and optical imaging: Prior to SEM imaging, samples were sputter coated in Au / Pd for 90 seconds using a Quorum Q150T ES gold sputter coater. SEM images were captured on a Zeiss 1530 VP FEGSEM. EDS data were captured using an Oxford Instruments X-mas 80mm2detector and processed using Oxford Instruments Aztec EDS microanalysis software. Microscope images were captured using a Nixon Optiphot2 optical microscope and images were captured using a DS 5M camera with DS-L1 camera control unit.

[0291] Electrode fabrication: Electrodes were fabricated by inserting a length of silver wire (obtained from Advent Research Materials product code: AG5485) with a diameter of 0.25mm and purity of 99.99% into the tip of a pipette. A small section of wire was passed through the narrow end of the pipette and epoxy glue was used to secure the wire in place; the electrode was then allowed to dry. Quick epoxy was used to secure the wire in place; the electrode was then allowed to dry.

[0292] Sample running: Samples were loaded into a Dolomite mitos p-pump which was controlled via the flow control centre software. Once the required pressure was input into the software, the pump drove the sample into the device. The recording software was activated when a signal was detected and recorded each sample for the required time or until the required number of particles were detected. The flow rate was determined by setting the lid to the required baseline and pressure. A pre-weighed Eppendorf was then placed at the outlet for one minute, removed and re-weighed to determine the mass and volume of liquid that had flowed through the device in that time period.

[0293] Tea sample preparation: Tea samples were prepared by cutting open the bag and discarding the contents. The bag was then washed with deionised water and allowed to dry. Finally, a glass vial of deionised water was heated to 95°C and the dry tea bag was placed in the vial for 5 minutes. After 5 minutes, the solution was poured into another vial and allowed to cool. The sample was then diluted to the required concentration with electrolyte solution.

[0294] Tea sample electron microscopy preparation: Samples were prepared by vacuum filtration of the decanted solution using 0.02 micron anodised 47mm filtration membranes. The membranes were then washed five times with 15MΩ deionised water and left to dry before mounting them onto aluminium SEM stubs using carbon adhesive tabs. Pulse shape analysis: Pulses were analysed by custom R code previously reported. Pulses were extracted, de-rendered and aligned before being approximated using b-splines (using the COBS software package). The logistic model used for shape / run discrimination was built on training data using three spline coefficients that showed the maximum predictive power for shape classification (under lasso penalty regression using the glmnet software package). Classification power was then assessed on independent test data.

[0295] Results

[0296] Figure 27 a shows an image of the base unit, lid, sensing region and electrodes. The lid can have several designs; the first is a flat surface that mimics the surface of the acetic acid film. The second design has a ridge that extends 1mm out from the surface and is aligned to extend into the microfluidic channel. The third has multiple ridges that can change the shape of the channel and sensing region. The device is designed to be integrated into a flow system and therefore has printed threads to connect to a pump.

[0297] The PDMS layer acts as a sealing component and gasket to prevent any leaks. It is fixed into place by screws and the shape or size (internal volume) of the channel can be controlled via two mechanisms. The first is by changing the shape and structure of the lid. The lid can be made as a flat surface or contain protrusions designed to fit within the main channel of the base unit. The second option is to use the PDMS layer which, upon application of force via the screws, is compressed into the channel. Figure 28 A schematic of the process is shown for the flat lid, the PDMS covers its entire lower surface and as the screw tension is increased it forces the flexible PDMS gasket into the channel, resulting in a decrease in the internal volume of the channel. For the second lid design, the ridges cause the maximum change in channel volume as the screw is tightened, this is in contrast to the PDMS layer which does not extend on top of the ridges.

[0298] When the channel is filled with a conductive liquid and a potential difference is applied between the two electrodes, the size of the channel can be monitored in real time by the baseline current, I, as I is proportional to the sensing volume, so the resistance can then be related to the current by Ohm's law: R = L / σhw, where R is the resistance, L is the channel length, σ is the conductivity of the solution and h and w are the height and width of the channel respectively.

[0299] Figure 29aFlow rate vs. baseline current for a flat lid device is shown at constant applied pump pressure. Baseline current is controlled by changing the pressure on the lid with a screw. Greater force compresses the PDMS layer or ridge further into the channel and results in a smaller baseline current. Figure 29a Data in d shows the relationship between baseline current and spiral tension, illustrating the use of a compressible PDMS layer to adjust channel dimensions. As shown in Figure 28 The shape of the sensing region and its internal volume are further reduced by adding a ridge on the lid that extends into the fluid, as shown in e. The screw and PDMS compression layer allow further adjustment of the channel volume. An extreme version allows the ridge to completely seal the channel and work like a 3D printed valve.

[0300] Figure 29 also shows the relationship between applied voltage and current for the ridge lid in the presence and absence of convection. The linear relationship between current-voltage and current consistency in the case of fluid flow indicates that there is no preferential current flow / diodes or resistance problems between the electrodes. To test for any effects the lid material can have on the current, the base unit was sealed with an acetate film to mimic gen-1. Figure 29d Example baseline current traces and noise during convection are also shown for different ionic strengths. After the channel was sealed with a PDMS gasket and lid, the device was used as a resistive pulse sensor. 20 pm particles were added to the sample cell and pushed through the device using a range of flow rates.

[0301] Figure 30 a shows example current vs. time traces, and the relationship between pulse frequency and flow rate, Figure 30 b. Pulse frequency J has a linear relationship with flow rate AP and concentration Cs Figure 30 b, c), as predicted by the equation J = CsAP, where the contributions of electrophoresis and electroosmosis are negligible at high flow rates. Flow rate does not change the size of the signal, but it changes the translocation time.

[0302] The same PDMS gasket can be reused 3 times before any leaks are observed, each time placing the components together and adjusting the tension on the screw to match the baseline current. This is interpreted to create a similarly sized channel. Figure 30 d shows example data sets from the same device and PDMS gasket that was taken apart and resealed. The device can remain sealed for four days without any degradation in signal quality, but it should be noted that calibration is required each day to quantify the analyte due to some drift in the signal. Figure 30 e also shows an example of the reproducibility of the setup, which shows the clogging distribution of 20 pm particles using the same base and lid unit with different PDMS gaskets, the average pulse shape for the three different assemblies as Figure 30f. Note that the shape of the pulse here reflects the internal shape of the sensing zone. While we assumed that the signal should be a rectangular pulse with a flat and consistent current during the translation event, the pulse shape indicates a different relationship. Similar effects were seen in the previous example as well, and future work needs to model the current and behavior of the signal. We also note that the thickness of each PDMS gasket can be different, but a consistent sensing zone can be produced each time by tightening the screw to match the baseline current. The pulse shape also depends on the flow rate, as this affects the time within the channel.

[0303] The sensitivity of the RPS can be changed by varying the volume of the sensing zone. Figure 31 An example of this is shown in a, where the size of the clogging of the same size particles increases with the increase in the screw tension as the screw is tightened. With this ability, the same device printed with an initial channel size of 100 pm can measure particles from 2 to 30 pm Figure 31 b-e). This simple configuration can enable an enhanced dynamic range on a set of reusable components.

[0304] The device was designed to screen for contamination in environmental and food samples. One new threat to the global environment is microplastics, so the first test was inspired by a recent publication. Hernandez and colleagues found that certain tea bags shed billions of plastic microparticles during use. Following the protocol they published (Environ. Sci. Technol., 2019, 53, 12300-12310), we placed the bags in hot water for 5 minutes and passed the solution through our sensor. Within a few seconds, the particle displacement device, Figure 32 a. Using particles of known size and concentration, we calculated an average size of 21.9 pm at 6.52 x 10 3 particles / mL. We acknowledge that, unlike previous studies, we were unable to calculate the entire range of particle sizes, and the number of nanoplastics was not included in this value. Tea bags from several manufacturers were tested, and the total number of particles shed from one tea bag was as high as 6.52 x 10 4 The presence of particles was consistent across several tea manufacturers, and analysis via SEM confirmed that these particles were carbon-based. Changing the ionic strength of the liquid produced interesting effects, Figure 32b. At lower ionic strengths, both the direction of the pulse reverses and conductive pulses are recorded significantly. The reversal of the pulse can be caused by two factors, the first being that the polymer can contain a higher concentration of ions, making them "conductive" relative to the surrounding liquid; hydrogels have a similar effect. Alternatively, the high surface area seen in the SEM, combined with lower ionic strengths and larger double layers, can form a dense cloud of ions around each particle, increasing the conductivity of the liquid during each translocation. When the same polymer with a smooth surface was purchased and placed into the device, only resistive pulses were recorded, even at low ionic strengths. The porous silica particles found in toothpaste also predominantly produce conductive pulses at lower ionic strengths. While conductive pulses give an indication of a change in physical properties, the lack of a suitable calibrant makes it more difficult to determine the particle size using conductive pulses. For a fully automated screening device, confirming the chemical nature of the particle can require embedding additional sensors in the channel, and this is within the scope of future work. However, the direction of the pulse can give an early indication of a man-made particle.

[0305] In the ocean, calculating plastics in the presence of biological particles is a challenge. However, Figure 32 The results in Figure 6 show that bacteria and algal particles with fairly smooth surfaces should also produce resistive pulses. This is confirmed in the data shown in Figures 7c and 7d. Here we present a series of data collected for algal particles at different ionic strengths. Figure 32 c and Figure 32 d show data for algal particles collected at different ionic strengths. Figure 32 This interesting observation can enable scientists to quickly screen for the presence of plastics in a sample.

[0306] It is well known that diatoms and algae come in a variety of shapes depending on the strain. Therefore, a description of the shape of the particle can be helpful in identification. Here we placed a rod-shaped alga (Synedra ulna) and some spherical algae (Rivularia haematites) through the device. We have previously shown how the shape of the particle can be inferred from the shape of the pulse generated by the particle. Figure 33The average depth-normalized pulse shapes of two algal strains and their corresponding calibration bead runs are shown. To quantify the differences between the pulses produced by the algae and the calibrant, the pulses were approximated by splines and the spline coefficients were compared using the modeling method described above. For these models, the spline just before the pulse minimum and the spline around the second peak in each pulse were used because they showed the greatest differences between the algae and the calibrant. For each type of algae, the model and device were calibrated to spheres run under the same assembly. When comparing the calibrant to the algae, the model was able to correctly identify 87% of the algal spheres and 86% of the algal rods. This means that most of the algae could be identified from the sphere calibrant. For comparison, when the calibration spheres were run twice under the same assembly, the model only correctly identified 66% of the runs of the spheres, which is close to the expected random 50% correct classification if the pulses from the two runs were the same. One surprising result from the data set was the ability to identify the spherical algae. While their general shape should be similar to the calibrant, the algae have flexible surfaces that can deform in the fluidic system. Deformation of biological particles has been explored for cells and exosomes, and the same feature here can help to distinguish biological particles from artificial particles. This establishes that the device can be used for shape analysis, and this can be improved with optimized run methods and models. The second unexpected and advantageous result from the analysis is that the device is able to perform shape analysis across multiple uses (e.g., days and assemblies) as long as a calibration is performed each time; this is in contrast to the current commercial system for which the model was originally developed, which permanently loses the ability to detect shape after a much shorter period of use. The flow rate used to calculate the algae also affects the quality of the pulse shape information; as shown in FIG. 6B, the pulse shapes from the two strains converge at higher flow rates. Figure 33

[0307] The invention can be defined by the following clauses:

[0308] 1. A first particle sensor, comprising:

[0309] a base comprising a microfluidic channel, wherein the microfluidic channel comprises a first electrode and a second electrode positioned along the microfluidic channel, wherein the first particle sensor is configured such that when a fluid sample comprising at least one particle passes along the microfluidic channel and flows past the first electrode and the second electrode, each of the at least one particle is recorded as a resistance pulse in the presence of an electrical potential difference between the first electrode and the second electrode.

[0310] 2. A second particle sensor, comprising:

[0311] ​a holder to house a membrane and electrodes, wherein the membrane comprises at least one aperture, wherein the second particle sensor is configured such that when a fluid sample comprising at least one particle passes through the at least one aperture of the membrane, the electrodes detect the at least one particle in the fluid sample and record a resistance pulse.

[0312] 3. An apparatus for characterising one or more particles in a fluid sample, comprising:

[0313] (iii) a first particle sensor according to clause 1 ; and / or

[0314] (iv) at least one second particle sensor according to clause 2,

[0315] wherein the apparatus further comprises a fluid inlet and a fluid outlet connected by the first particle sensor in clause 1, wherein in use, the fluid sample passes in through the inlet, along the microfluidic channel of the first particle sensor, past the first and second electrodes, and out through the fluid outlet, with an electric potential difference applied between the first and second electrodes, and wherein each of the at least one particle is recorded as a resistance pulse.

[0316] 4. The apparatus of clause 3, wherein the apparatus comprises at least one second particle sensor according to clause 2, which is located within the microfluidic channel.

[0317] 5. The apparatus of clause 3 or clause 4, wherein each of the components of the apparatus is 3D printed.

[0318] 6. The apparatus of any one of clauses 3 to 5, wherein the apparatus further comprises a lid configured to seal the microfluidic channel of the apparatus and create a constriction within the microfluidic channel in order to adjust the sensitivity of the first particle sensor.

[0319] 7. The apparatus of clause 6, wherein the lid comprises a primary ridge which fits into the microfluidic channel and is configured to seal the microfluidic channel.

[0320] 8. The apparatus of clause 7, wherein the primary ridge further comprises a secondary ridge which extends from the primary ridge, wherein the secondary ridge comprises a conduit which allows the fluid sample to flow through the secondary ridge when the lid seals the microfluidic channel.

[0321] 9. The apparatus of any one of clauses 3 to 8, wherein the base further comprises a groove which comprises an O-ring which surrounds the microfluidic channel and is configured to prevent leakage from the microfluidic channel.

[0322] 10. The device according to any one of clauses 3 to 9, wherein the device comprises a polymer layer between the lid and the base, wherein the polymer layer is configured to seal the device, preferably wherein the polymer layer is polydimethylsiloxane (PDMS). Wherein the thickness of the PDMS also regulates the channel.

[0323] 11. The device according to any one of clauses 3 to 10, wherein the microfluidic channel is filled with an electrolyte solution.

[0324] 12. The device according to any one of clauses 3 to 11, wherein the second particle sensor is filled with an electrolyte solution between the membrane and the electrode.

[0325] 13. The device according to any one of clauses 3 to 12, wherein the second particle sensor membrane (9) is selected from the group consisting of silicon nitride, polyurethane and polyester.

[0326] 14. The device according to any one of the preceding clauses, wherein the first particle sensor is configured to detect particles having a size of 5 pm to 100 pm.

[0327] 15. The device according to any one of clauses 2 to 13, wherein the second particle sensor is configured to detect particles having a size of 1 pm to 100 pm.

[0328] 16. The device according to any one of clauses 3 to 15, wherein the device is configured on a chip.

[0329] 17. A method of characterising one or more particles in a fluid sample, comprising the device according to any one of clauses 3 to 16, wherein the method comprises passing a fluid sample comprising at least one particle through the fluid inlet, along the microfluidic channel, past the first and second electrodes, and out of the fluid outlet; and wherein each of the at least one particle present in the fluid sample is recorded as a resistance pulse in the presence of an electrical potential difference between the first and second electrodes.

[0330] 18. The method according to clause 17, wherein the first particle sensor can detect particles having a size of 5 pm to 100 pm.

[0331] 19. The method according to clause 17 or clause 18, wherein the second particle sensor can detect particles having a size of 1 nm to 100 pm.

[0332] 20. The method according to any one of clauses 17 to 19, wherein the method further comprises the step of using a predictive logistic regression model to characterise the resistance pulse.

[0333] 21. The method of any one of Clauses 17 to 19, wherein the method is used to characterise one or more of:

[0334] (i) cells in a bodily fluid

[0335] (ii) organic compounds, proteins, cells, bacteria, viruses, DNA, exosomes, colloids and nanomedicines.

[0336] 22. The method of any one of Clauses 17 to 21, wherein the method is used to sense one or more fluid samples in-line for high-throughput processing.

[0337] 23. A computer program recorded on a computer readable storage medium configured to, when executed by a computer processor, perform at least:

[0338] receiving as input data representative of resistance pulses corresponding to particles, the resistance pulses obtained by a particle sensor according to any one of Clauses 1 to 3;

[0339] comparing the received data to pre-stored characterising shape and size data representative of particles of known shape and size;

[0340] determining one or more of the shape and size of the at least one particle based on the comparison; and

[0341] outputting the determined one or more of the shape and size of the particle.

Claims

1. An apparatus for characterising one or more particles in a fluid sample, the apparatus comprising: an inlet; an outlet; a microfluidic channel extending between the inlet and the outlet and providing an axial flow path for fluid flowing therealong; a first particle sensor for detecting passage of a particle moving along the axial flow path, the first particle sensor comprising: a first electrode set comprising a first electrode and a second electrode positioned along the microfluidic channel; and a detection region located between the first electrode and the second electrode; a second channel extending from the microfluidic channel; and at least one second particle sensor comprising a second electrode set, the second particle sensor being configured to detect particles of a different size range to that detected by the first particle sensor, wherein, in use, one or more particles flowing along the microfluidic channel are detected by the first particle sensor and passage of the one or more particles is recorded as a pulse, the apparatus being characterised in that the at least one second particle sensor comprises a nanopore located in the second channel, wherein the nanopore does not extend across the entire width of the second channel such that the fluid can flow around the nanopore.

2. The apparatus of claim 1, wherein, The first particle sensor is configured to detect particles of a size of 1 pm to 100 pm.

3. The apparatus of claim 1, wherein, The second particle sensor extends at an angle to the microfluidic channel.

4. The apparatus of any one of claims 1 to 3, wherein, The second particle sensor provides a flow path for particles.

5. The apparatus of claim 4, wherein, The flow path of the second channel extends at an obtuse angle, an acute angle or orthogonally to the axial flow path.

6. The apparatus of claim 5, wherein, The second particle sensor is configured to detect particles of a size of 1 nm to 1 pm.

7. The apparatus of any one of claims 1 to 3, wherein, The second particle sensor is located downstream of the detection region of the first particle sensor.

8. The apparatus of any one of claims 1 to 3, wherein, The second channel extends perpendicularly from the microfluidic channel.

9. The apparatus of any one of claims 1 to 3, wherein, The nanopore of the second particle sensor is a solid state nanopore, which is a hole formed in a membrane.

10. The apparatus of claim 9, wherein, The membrane is formed from a material selected from the group consisting of silicon nitride, silicon dioxide, polyurethane and polyester.

11. The apparatus of any one of claims 1 to 3, wherein, The nanopore is provided in a nanopipette.

12. The apparatus of any one of claims 1 to 3, wherein the second particle sensor comprises an electrode.

13. The apparatus of any one of claims 1 to 3, wherein, The second particle sensor comprises a holder housing the nanopore and electrode, wherein the holder has a second channel therein, the nanopore being located in the second channel.

14. The apparatus of claim 13, wherein, The holder is in the form of a plug or screw which is releasably insertable into a port in the base of the first particle sensor such that a fluid connection is formed between the microfluidic channel and the second channel.

15. The apparatus of any one of claims 1 to 3, wherein, The second particle sensor comprises a tube in fluid communication with the second channel, wherein the tube provides a further fluid outlet.

16. The apparatus of any one of claims 1 to 3, comprising two or more second particle sensors.

17. The apparatus of claim 16, wherein, Each of the second particle sensors is configured to detect a different size range of particles.

18. The apparatus of any one of claims 1-3, wherein, The second particle sensor comprises or is connected to a flow regulator.

19. The apparatus of any one of claims 1-3, wherein, The apparatus comprises a third sensor configured to measure a parameter of the fluid.

20. The apparatus of claim 19, wherein, The parameters are selected from temperature, pH, oxygen content, ionic strength, and viscosity.

21. The apparatus of any one of claims 1-3, wherein, One or more components of the device are 3D printed.

22. The apparatus of any one of claims 1-3, wherein, The device further comprises a lid configured to seal the microfluidic channel.

23. The apparatus of claim 22, wherein, The lid comprises a main ridge configured to be received within the microfluidic channel, thereby reducing a volume of the channel.

24. The apparatus of claim 23, wherein, The main ridge further comprises a secondary ridge extending from the main ridge, thereby creating a constriction within the channel.

25. The apparatus of claim 24, wherein, The main ridge and the secondary ridge together span a height and a width of the microfluidic channel, and wherein the secondary ridge comprises a conduit that allows fluid to flow through the secondary ridge when the lid seals the microfluidic channel.

26. The apparatus of any one of claims 23-25, wherein, The lid comprises a polymer layer on the lid surface that mates with the base to seal the microfluidic channel.

27. The apparatus of claim 26, wherein, The polymer layer is polydimethylsiloxane (PDMS).

28. The apparatus of claim 27, wherein, The lid is attached to the base by one or more screws.

29. The apparatus of claim 28, wherein, Upon tightening the screws, the polymer layer is forced into the microfluidic channel, thereby reducing the volume of the channel.

30. The apparatus of any one of claims 1-3, wherein, The base further comprises a groove comprising an O-ring that surrounds the microfluidic channel and is configured to prevent leakage from the microfluidic channel.

31. The apparatus of any one of claims 1 to 3, wherein, The device is configured on a chip.

Images