Here, we present a protocol on how to build and test a simple but efficient low-cost particle detector.
As particles with a size of 1 µm or smaller pose a severe health risk to the human body, the detection and regulation of particle emissions are of great importance. A large share of particulate emissions are emitted by the transport sector. Most of the commercially available particle detectors are bulky, very expensive, and need additional equipment. This paper presents a protocol to build and test a standalone particle detector that is small and cost-efficient.
The focus of this paper lies in the description of the detailed construction manual with video and the sensor evaluation procedure. The computer-aided design model of the sensor is included in the supplemental material. The manual explains all the construction steps, from 3D printing to the fully operational sensor. The sensor can detect charged particles and is therefore suitable for a wide range of applications. A possible field of application would be soot detection from power plants, wildfires, industries, and automobiles.
Inhalation of particles with a size of 1 µm or smaller poses a high risk of adverse health effects on the human body. With increasing environmental pollution from combustion processes, respiratory diseases are growing in the population1,2,3. To promote health and counteract pollution, it is necessary to first identify the sources of pollution and quantify the degree of pollution. This can be done with existing particle detectors. However, these are large and very often far too expensive for private or citizen science purposes.
Many of the commercially available particle detectors are bulky, very expensive, and require additional equipment to be operated4. Most of them also need several aerosol-conditioning steps. For example, dilution is needed for detectors that use light scattering as their measurement principle, and the measurement range is limited by the wavelength5,6,7. Particle detectors that use laser-induced incandescence as a detection principle need both high energy laser sources and an energy-consuming cooling system8.
Particle detectors that use condensation particle counters are normally used as the gold standard for particle concentration measurement; these need preconditioning, dilution, and working fluids (e.g., butanol)9,10,11. The advantages of an electrostatic sensor lie in the simple and compact design and the low fabrication costs. However, in comparison to condensation particle counters, significant deductions have to be made regarding accuracy.
An electrostatic sensor represents an alternative to these methods. Electrostatic sensors can be robust, light, inexpensive to manufacture, and can be operated without supervision. The simplest form of an electrostatic sensor is a parallel plate capacitor with a high electric field between its plates. As aerosol is conveyed into the high voltage region between the two copper electrodes, naturally charged particles deposit on the electrodes of different polarity12 (Figure 1).
Dendrites form on the surface of the electrodes in the direction of the field lines of the applied high voltage between the electrodes, and are charged via contact charging. Fragments of these dendrites eventually break off the electrodes and redeposit on the electrode with opposite polarity, transferring their charge. These fragments carry a high number of charges. Because the electrode is grounded, the deposited charge generates a current leading to a voltage drop at the internal resistance of the bench multimeter. The more often this happens per unit of time, the higher the current, and consequently, the higher the voltage drop (Figure 2).
Owing to the high voltage induced by the charge deposition of the fragments, no further amplifier electronics are needed. The formation of dendrite break-off particles and the subsequent charge release of these particles represents a natural signal amplification12. The resulting sensor signal is proportional to the particle mass concentration. This signal can be detected with an off-the-shelf bench multimeter.
Figure 1: Sensor schematics. Aerosol flows into the aerosol inlet, is propagated through the left flow channel, and then reaches the gap between the high-voltage electrode (inner electrode) and the measuring electrode (outer electrode). There, the particles contribute to dendrite growth and, as previously explained, break-off, thus generating the sensor response. Afterward, the particles flow further through the right flow channel and leave the sensor at the aerosol outlet. Please click here to view a larger version of this figure.
Figure 2: Physical principle. Positively and negatively charged particles, as well as neutral particles, enter the gap between the electrodes of opposite polarity. They are diverted by the electric field lines to the electrode of opposite polarity and deposit their charge there. Then, they become part of a dendrite and take over the charge of the respective electrode. The field density is highest at the dendrite tip, where more particles are trapped. When the drag force exceeds the binding forces, segments of the dendrites break off, which in turn strike the opposite electrode and deposit their charges. Please click here to view a larger version of this figure.
With a cylindrical design, as in Warey et al.10, the probability of soot bridges forming can be minimized. Further information about the sensor geometry, applied voltage, gas flow velocity, and particulate matter concentration can be found there. They suggest correlation of the sensor signal to particulate matter streaming through the sensor (equation 1).
Sensor (V) = 5.7 × 10-5 C V0 e0.62V × (1)
C is the mass concentration of the particulate matter, V0 is the applied voltage, V is the exhaust velocity, L is the electrode length, and S is the electrode gap13.
Bilby et al. focused on the detailed study of the underlying physical effect of the electrostatic sensor9. These studies included an optically accessible setup and a kinetic model to explain the signal amplification of the dendrite-based sensor (see equations 2 and 3).
(2)
(3)
S represents a stack of soot discs of 10-100 soot agglomerates with a size of 50-100 nm; Dn represents a dendrite with n disks; Br denotes a break-off fragment composed of f disks; S and ki are rate constants12.
This paper presents a protocol on how to build and test a simple but efficient low-cost particle detector that can be used for high particle concentrations without further equipment. Previous work on this type of electrostatic sensor has mostly focused on exhaust measurements. In this work, laboratory-generated soot particles are used as test aerosols. The described sensor is based on `previous work from Warey et al. and Bilby et al12,13.
The sensor body consists of a stereolithography-based 3D printed body, coaxial electrodes cut from copper tubes, a vacuum gasket, and a vacuum clamp. Materials such as the vacuum gasket, cable, copper tubes, and 3D resin for one sensor cost less than €40. The additional equipment needed is a high-voltage source, a USB bench multimeter, and a soldering station. To evaluate the sensor, a defined aerosol source and a reference instrument are also required once (see Table of Materials). The size of the sensor described in this protocol is 10 cm x 7 cm. This size was chosen specifically for the experiment and can still be reduced significantly (see modifications/sensor dimensions in the discussion).
This protocol describes how to build, test, and use a simple low-cost particle sensor. A schematic of the protocol is shown in Figure 3-beginning with the 3D print of the sensor hull and the electrode manufacturing, the assembly of the sensor, as well as testing and an example of field application of the sensor.
Figure 3: Schematic for the method. The protocol is divided into four major steps. First, all parts for the sensor housing are printed. Then, the electrodes are manufactured. In the third step, the 3D printed sensor housing with the electrodes and the vacuum gasket are assembled. In the last step, the sensor performance is evaluated. Please click here to view a larger version of this figure.
The most important steps of the 3D printing process are shown in Figure 4. At first, the right slicer settings for the print are chosen. Afterward, the most important parts of the print and the preprocessing of the 3D printed model are discussed. For this step, a resin 3D printer with an isopropanol bath and UV hardening device and a straight grinder are needed.
Figure 4: Schematic of 3D print. (A) The slicer 3D model is depicted; (B) the printer during the printing process. Postprocessing steps: (C) flushing and (D) UV-hardening. Please click here to view a larger version of this figure.
Figure 5 shows the most important steps of electrode manufacturing: the form shaping of the electrodes as well as the soldering of the contact to the electrodes. For this step, two copper tubes with different diameters, a caliper, a pipe cutter, a straight grinder, a vice, a soldering station and soldering tin, isolated cables with two different colors, thermal protective gloves, and a wire cutter are needed.
Figure 5: Electrode manufacturing. (A) Measuring, (B) cutting, (C) deburring, and (D) soldering of the electrodes. Please click here to view a larger version of this figure.
The assembly section in the protocol explains how the sensor is assembled. The most important sensor parts are depicted in Figure 6, namely the outer electrode holder, the flow channel, and the inner electrode holder. Figure 7 shows the most important steps in the sensor assembly. For this step, epoxy glue, protective clothing, a vacuum seal, a vacuum clamp, safety goggles, and gloves are needed.
Figure 6: Sensor parts. (A) The outer electrode holder, (B) flow channel, and (C) the inner electrode holder. Please click here to view a larger version of this figure.
Figure 7: Sensor assembly. All steps of the sensor assembly are shown. A–E shows the assembly of one half of the sensor. (A) The inner electrode holder is glued to the flow channel. (B) The inner electrode is placed onto the inner electrode holder. (C) The outer electrode is placed into the outer electrode holder. (D) The outer electrode holder is glued onto the flow channel + inner electrode holder assembly. (E) The vacuum sealing snaps into the outer electrode of one sensor half and then snaps into (C), the identical second outer electrode of the other sensor half. Please click here to view a larger version of this figure.
The test section explains how to set up the experiment to compare the newly built sensor with a reference instrument. For this step, a bench multimeter, vacuum pump, high-voltage supply, aerosol generator, dilution bridge, aerosol tubes, Y-fitting, one mass flow controller (MFC), an aerosol mixer, a reference instrument, and a cotton swab are needed.
1. 3D printing
2. Electrode manufacturing
3. Assembly
4. Tests
5. Field application
Figure 8: Sensor setup. A diagram of the sensor setup. Aerosol flows through the sensor. The sensor is connected to the voltmeter and a high-voltage supply. The voltmeter is controlled by a control unit that logs the sensor data. Please click here to view a larger version of this figure.
Figure 9: Experimental plan for sensor evaluation. A stable aerosol source is used to mimic a particle source. The outflowing particle stream is split into path (A), sensor setup; and path (B), ventilation, enters the dilution bridge, and is further distributed to an aerosol mixer. After the mixer, the aerosol stream is split between a reference instrument path (D), which measures parallel to the sensor. This reference instrument needs dilution air, which is distributed through path (C). Path (E): an MFC draws air through the sensor. This MFC is protected from the aerosol stream with a HEPA filter. Abbreviations: MFC = mass flow controller; HEPA filter = high-efficiency particle absorbing filter. Please click here to view a larger version of this figure.
Figure 10: Field test: the experimental plan. In this setup, an aerosol source is measured. The outflowing particle stream is split into path A) sensor setup and path B) ventilation and then enters the sensor. In this setup, an MFC with a HEPA filter upstream sucks the aerosol through the sensor. Abbreviations: MFC = mass flow controller; HEPA filter = high-efficiency particle absorbing filter. Please click here to view a larger version of this figure.
The exact correlation of the sensor signal to particulate mass varies based on particle charge distribution and size distribution, as well as the aerosol composition. Therefore, the sensor must be calibrated to a particular application with a reference instrument. This section explains how to compare the newly built sensor with a reference instrument.
The starting phase of the sensor takes approximately 5-10 min, depending on the chosen particle concentration. Within the starting phase, the sensor signal significantly increases while the sensor is exposed to a constant particle concentration. After the starting phase, the sensor signal stabilizes. At that stage, an equilibrium state for accumulation and fragmentation of dendrites is reached and the sensor signal is then proportional to the incoming soot concentration. After this initialization phase, the sensor is ready to measure any changes in aerosol concentration.
The measurement data shown in Figure 11 starts from the moment the sensor is in the above-mentioned equilibrium state. To calculate the sensor current in amperes, the collected data in volts must be divided by the value of internal resistance to obtain the correct current value.
The vertical axis shows the sensor signal in amperes and the horizontal axis shows the aerosol concentration measured by the reference instrument in mg/m3. A linear fit with its representative parameters is also given in the plot. The high uncertainty of the measured data is due to the high dynamics when adjusting the concentration with the dilution bridge. The linear fit parameters are an R2 value of 0.80, an intercept of -0.53 nA, and a slope of 2.80 nAm3/mg with a standard deviation of 1.4 nA.
Figure 11: Positive results. The sensor signal is plotted on the vertical axis in amperes, whereas the particle concentration measured by the reference instrument in mg/m3 is plotted on the horizontal axis. In addition, a linear fit with the most important parameters is added to the plot. The linear fit parameters are an R2 value of 0.80, an intercept of -0.53 nA, and a slope of 2.80 nAm3/mg. Please click here to view a larger version of this figure.
There is also the possibility that particles clog the path between the electrodes, in which case conductive soot bridges form between the electrodes. Because soot is a conductive material, these soot bridges form a short circuit between the electrodes. The measured signal rises rapidly with increasing thickness of the conductive path, up to the point where the voltage becomes so high that the voltmeter might be damaged. An example for an experiment with forming soot bridges can be seen in Figure 12. The signal rises in very steep jumps/steps and does not stop or flatten out. Dendrites are also no longer formed, and the sensor is no longer in a state of equilibrium. In this case, the high-voltage source must be switched off immediately, the sensor has to be cleaned, and a new measurement has to be started.
Figure 12: Negative result. A short circuit has occurred during the measurement. The sensor signal in amperes is plotted on the vertical axis and the measurement time is plotted on the horizontal axis. The sensor signal continues to increase without restriction. Please click here to view a larger version of this figure.
If a flat line is displayed and the sensor current does not rise at all to a value above 1 nA, follow the troubleshooting directions in the discussion section. The sensor must be in the equilibrium state at all times to measure the entering aerosol accurately; therefore, a sufficiently high initial aerosol concentration has to be provided at the beginning of the experiment.
Supplemental File 1: This file represents the computer-aided design (CAD) file to print out the flow channel depicted in Figure 7A with holes for the cable. Please click here to download this File.
Supplemental File 2: This file represents the CAD file to print out the flow channel depicted in Figure 7A without holes. Please click here to download this File.
Supplemental File 3: This file represents the CAD file to print out the inner electrode holder depicted in Figure 7A. Please click here to download this File.
Supplemental File 4: This file represents the CAD file to print out the outer electrode holder depicted in Figure 7C (right). Please click here to download this File.
Supplemental File 5: This file represents the CAD file to print out the flow channel without holes depicted in Figure 7C (left). Please click here to download this File.
Supplemental File 6: This file represents the CAD file to print out the electrode spacer. Please click here to download this File.
Critical steps
Print post-processing
Almost any step in this protocol can be paused or postponed, except for post-processing of the freshly printed 3D parts (protocol step 1.5). If the UV protection screen of the printer is opened, the post-processing should begin immediately, otherwise the small cable channels, as well as the cavity for the seal, will clog. The precision fit of the cavity ensures that the sensor can be sealed airtight. This is important because the sensor is very sensitive to flow fluctuations. The hardening process is also important (protocol step 1.4); if the temperature is set too high, the material becomes too brittle and can break under the forces exerted by the clamp onto the outer electrode holder.
Electrode manufacturing
Careful cutting and deburring (protocol steps 2.2-2.3) of the electrodes is very important because irregularities in the electrode gap cause perturbations in the electrical and velocity fields, which leads to poor sensor performance. In the worst-case scenario, a strong irregularity can cause the electrodes to come so close that the breakdown voltage is exceeded, and a short circuit occurs. From this point on, no statement can be made about the measurement signal and the measurement electronics are prone to damage.
Assembly
Assembly of the sensor (protocol steps 3.4-3.6) is crucial, as this creates the electrode gap. As mentioned above, the distance between the electrodes is very important; this gap must be uniformly 1 mm over the entire length. These steps are important because they can change the electric field in the sensor drastically. The overall deposition behavior, as well as dendrite formation, can be influenced by the change in the electric field. Thus, it can no longer be guaranteed that the sensor response is linear to the incoming aerosol. The worst-case scenario of a short circuit also applies here.
Modifications
3D printing
Other possible modifications are the use of different 3D printing resins. There are many different resins on the market that can change the density, flexibility, temperature resistance, and strength of the sensor housing.
Sensor dimensions
The first design criterion for the sensor is a safety configuration. The dielectric strength of air between the electrodes is 3 mm/kV. This length must not be undercut in any case. The higher the electric potential, the more particles are deposited, and these deposited particles are then prone to form dendrites. The dimensions of the electrodes were chosen so that easily available standard components can be used. Designs of similar sensors known to the authors used the following dimensions for a flat sensor: 9 mm width, 2 mm length, 1 mm gap, and 15 mm length, with a diameter of 8.5 mm and gap of 1.3 mm for a cylindrical design12,13. In addition, it should be ensured that the sensor can be manufactured by hand in a normal workshop. A 1 mm gap is the absolute minimum gap that still allows the sensor to be cleaned manually. Here, 1 kV was used as a good compromise of safety and efficient particle deposition, as well as availability of voltage sources in this range.
Electrodes
Since the exact distance of 1 mm between the sensor electrodes is so crucial for performance, even more development work can be put into this step. For example, the 3D printed fixture can be made even more accurate, or a lathe can be used instead of a simple pipe cutter for cutting and deburring, if the equipment is available. Another option is to use a saw instead of a pipe cutter. In this case, the edges of the saw must be ground afterward. This method causes less deformation than the pipe cutter, but takes longer. In comparison to epoxy glue, silicone gives the cables more room to move, and it becomes easier to respace the electrodes. However, since the cables have more room to move, it is more difficult to seal the sensor. Instead of the vacuum clamp, which is easier to open at once, a self-made design is also feasible. Here, only holes for some screws and a cavity for the sealing cord must be altered in the 3D design.
MFC
The MFC determines how much of the aerosol is sucked through the sensor; the rest should be able to be drained through an overflow with a HEPA filter placed at the end of the overflow, to avoid pollution of the room. By choosing a less expensive pump instead of an MFC, higher flow fluctuations will influence the sensor signal negatively.
Dilution bridge
As seen in Figure 9, a dilution bridge can be built with a simple needle valve parallel to one or more HEPA filters. Other designs include a small vise to squeeze the tube instead of the needle valve. This design has the advantage that the tube can be cleaned more easily. The more coils such a vise has, the finer the concentration can be adjusted. This is especially important for calibration measurements, where high dynamics should be avoided.
Bench multimeter
The bench multimeter measures a voltage, which must be divided by the value of internal resistance to obtain the correct current value. Depending on the chosen measuring range (e.g., 100 V), this internal resistance value can vary (e.g., 1 MΩ). It is important to select a defined range so that the internal resistance value is the same for all measured values. If "auto range" is chosen, the internal resistance value must be tracked as well.
Troubleshooting
3D printer
If the printer stops, the tank should be checked for residues of the last print; the mixer often gets stuck. One should observe the first minutes of the printing process. If it is clogged, it is either because the correct slicer settings have not been set or the fresh print has not been stored under UV protected conditions before post-processing. In the slicer settings, no support points should obstruct the flow channel and the space between the electrodes, and the internal support structures box must be unclicked before sending the file to the printer.
Aerosol source + dilution bridge
If the aerosol source seems unstable, all the HEPA filters should be checked to ensure they are in the correct position and are not clogged. Also, the aerosol generator as well as the reference instrument should be checked to ensure they have finished their warm-up phase.
Sensor
The most common faults are caused by an insufficient power supply connection, an air leak at the sensor, or when deposited particles form soot bridges between the electrodes. First, the sensor is opened to check if soot bridges have formed between the electrodes. The power source must be turned off before disconnecting the sensor cables and opening the sensor. Soot bridges are easily visible to the naked eye and can be removed with little effort. To remove soot bridges, it is best to use an optical cleaning cloth or lint-free cotton swab.
A leak that changes the flow behavior in the sensor, as well as a lower voltage at the electrodes, can change the sensor signal. It is not possible to say in advance which of these problems is responsible for an unexpected sensor response. Therefore, it is important to check both the tightness and the voltage stability as follows. First, connection from the cable to the electrodes is checked (protocol step 4.4). Next, the voltage source is checked to see if it is delivering the expected volts. An air leak is best identified with leak spray. In addition to this, the tightness can also be checked with a vacuum pump, as described in protocol step 4.4.2.
Limitations
The limitation of an electrostatic sensor is well described by Maricq et al.14. In their work, they emphasize the importance of a stable voltage source and a stable sensor flow for the performance of the sensor. For this reason, a setup with an MFC or a pump should always be used for flow control, as described in Figure 10. In addition, the sensor needs a longer time to reach equilibrium during the first test. In further experiments, where a stable dendrite population has settled on the electrodes, the amount of time to start up the sensor is reduced. However, it should be generally noted that the sensor always needs a startup time to become operational depending on the initial concentration.
Unlike a flat design, as in Bilby et al., sensor drift is not a major problem in this cylindrical arrangement12. However, fast concentration changes at low particle concentrations are still difficult to detect with the sensor. As indicated by Diller et al. and Maricq et al., for a meaningful measurement signal, the measured value is averaged over 2-10 min, depending on how much the flow changes in the experiment14,15.
With a slope of 2.8 nAm3/mg and a standard deviation of ±1.4 nA, the deviation from the regression line in Figure 11 is high. For a better understanding of the sensor accuracy, the comparison of several experiments is recommended. For repeated experiments, the slope accounts for 3.5 nAm3/mg with a standard deviation of ±1.0 nA, and 4.9 nAm3/mg with a standard deviation of ± 0.6 nA. In addition, the sensor will give a very high reading the moment the voltage source is switched on. This start value is filtered out of the measurement data.
The advantage of the method presented here lies clearly in the simplicity, but also in the versatile possibilities to adapt the sensor shape to different needs. Therefore, in addition to soot, the sensor can detect a large variety of charged particles and is suitable for a wide range of applications, for example, particulate matter detection from power plants, wildfires, industries, and automobiles. This paper should be an incentive to agencies, companies, research teams, citizen scientists, and anyone interested in the detection of particulate matter to reproduce this simple sensor construction manual and build their own particle detector.
The authors have nothing to disclose.
This work was funded by the COMET Centre "ASSIC-Austrian Smart Systems Integration Research Center". ASSIC is co-funded by the BMK, the BMDW, and the Austrian provinces of Carinthia and Styria within the COMET-Competence Centres for Excellent Technologies programme of the Austrian Research Promotion Agency (FFG).
Equipment | |||
3D printer | Formlabs | Formlabs 3 | |
Aerosol Mixer | ESSKA | 304200812095 | 95 mm, diameter 8 mm |
Aerosol soot generator | Jing Aerosol | Model 5201 Type C miniCAST | |
Benchmultimeter | Keysight | KEYSIGHT 34465A, 0 – 100 V range, 1 MΩ internal resistance | |
Dilution Bridge | Custom built | Needel valve and HEPA filter in parallel | |
High voltage power supply | Stanford Research Systems | PS350, 5000 V – 25 W | |
Mas flow controller | Vögtlin | GSC-C3SA-BB26 | Red-y for gas flow, flow range: 0-10 L/min |
Refence Instument | AVL | MSSplus – AVL Micro Soot Sensor | |
Material | |||
Aerosol tygon tubes | Saint Gobain Fluid Transfer | AAG00012 | Diameter 7 mm |
Bidirectional flow control valves series RFO | CAMOZZI | RFO 383-1/8 | P max 10 bar |
Copper tube 12 mm | Obi | 1996602 | Diameter 12 mm |
Copper tube 18 mm | Obi | 1499441 | Diameter 18 mm |
Copper tube 22 mm | Obi | 1996628 | Diameter 22 mm |
Cotton swab | Chemtronics | 48042F | 50 m, 1 mm tip |
Epoxy glue | RS components | 132605 | RS quick set epoxy |
Hepa Nylon Einweg-Inline-Filter | Parker | 9933-05-BQ | Flüssigkeit 5.4SCFM 1/4Zoll, mit G1/4 Anschluss 8,1 bar |
Isolated electrical cable | Nexans | Diameter 2 mm, two different colors red and black | |
Photopolymer Resin | Formlabs | 851976006196 | 1 L Cartridge – Transparent (Clear) |
Soldering tin | Stannol | 574108 | |
Steckverbinder reduziert mit Stecknippel | ESSKA | IQSG120H6000 | |
Tefen polymer Y – fitting | TEFEN | TEF-8357-06-00 | |
Thermal protection gloves | As One | ||
Vacuum clamp | MISUMI | FRNWC40 | Clamp |
Vacuum seal | MISUMI | FRNWR40 | Centering ring with O-ring seal |
Tool | |||
Caliper | Starrett | DW990 | |
Deburrer | Ruko | ||
Gloves | BM Polyoo | ||
Isopropanol bath | Formlabs | FK-F3-01 | Form 3 finish kit |
PCB vice | RS components | 221-7531 | |
Pipe cutter | Rigid | 35S | |
Safety goggles | 3M | ||
Sand paper | Mirka | Different sandpaper thicknesses 40 – 200 | |
Soldering station | Ersa | Ersa i-CON 2, 400 °C, 2.2 mm soldering rod | |
Straight grainder | Dremel | F013400046 | Dremel 4000 |
UV Hardening device | Formlabs | FH-CU-01 | Form cure |
Vacuum pump | Mityvac | MV8000 | Automotive Tune-up and Brake Bleeding Kit |
Vise | Proxxon | NO 28 132 | MS4, Jaw height 10 mm, Max. Clamping width 34 mm |
Wire cutter | KNIPEX | 7712115 | |
Software | |||
MFC software | Vögtlin | Get red-y | |
Reference Instument Software | AVL | Supplied with the device: MSSplus | |
Slicer software | Formlabs | Preform Download Link: https://formlabs.com/de/software/ |