We have two different technologies in development, alert systems built around optical particle monitors, and passive PM monitors to meet EPA’s proposed (supplementary monitoring) (link pm-monitoring #categories) goals for citizen science.

Verifying Passive PM monitors

Our goals are to check the correspondence between the passive PM monitors we’re replicating and other PM measurement tools. Our short-term goal is to replicate a a correspondence to a Federal Reference Method of +/-20% with a stretch goal of reaching +/-10%. [link to PM-monitoring]

This can be achieved by co-locating the monitors with FRM monitors at one of the ~900 EPA certified test stations across the US. @GretchenGehrke has tentative plans to do so in North Carolina. We will build up to this test with calibration testing at Public Lab and Chicago State (CSU).

Problem Definition

Material choices in the manufacture of passive monitors can change the particles that are captured. Some materials can generate a static charge that may attract and repel particles depending on the particles’ charge. Our goal is that home DIYers, secondary schools, and universities could potentially replicate our tools and techniques making small quantities of monitors. We have picked materials as close as possible to the literature while also in line with the expected production capabilities of target users.

We have built monitors to the specifications given in the literature, substituting 3D printed ABS for the cap and laser-cut acrylic for the housing. The original monitors were made on a CNC machine and cut from stainless steel, with an aluminum housing. CNCing services are less available to us, to DIYers, and to future universities or secondary schools that may use this monitoring methodology. CNCing services for stainless steel are prohibitively expensive. We therefore had caps milled out of aluminum.

We have to test: 1) The consistency of low-run production units, comparing data from monitors of the same materials in the same location. 2) the effects of material choice on particle count, comparing the data from monitors of of different materials in the same location. 3) the correspondence between our monitors and reference measurements.

Internal testing:

We’ll calibrate our SEM and Visible Light microscopy with NIST-certified spheres of a known size, following Ott et al. so we can test the same image analysis on two different imaging and hardware configurations: SEM: Acrylic body & graphite & carbon tape stub
Aluminum body & graphite & carbon tape stub Visible Light Microscopy: Acrylic body and aluminum & glass stub Aluminum body & aluminum & glass stub

We will then collect data on several 7-day co-deployments of passive PM monitors against our own reference instruments for PM10-2.5. We will get a reference reading for PM10-2.5 by collecting PM10 without gravimetric calibration using a Dylos 1100 monitor, and PM2.5 data with a 24-hour gravimetric calibration using a PDR-1500 with a PM2.5 impactor and filter. This data will be transmitted and collected with Open Pipe Kit, and the filters analyzed at Chester LabNET in Tigard, OR.

colocated testing:

When we are confident in our optical monitors, we will send either the aluminum or acrylic passive PM monitors (based on the performance of each in our internal testing) to an EPA FRM monitoring station at the primary EPA air quality research and development facility, in North Carolina.

Next Steps

@AmberWise, @Damarquis, and @Paul at Chicago State will be: comparing co-located visible light samples and SEM samples taken in Chicago verifying ImageJ counts against manual particle counts examining the accuracy of thresholding in ImageJ using samples taken in Wisconsin Imaging samples located at Public Lab for comparison to DYLOS & PDR-1500 concentrations Imaging samples from FRM colocation.

Looking over my recent notes and experimental data, these are the plans for A. Wise and students in the coming couple of months: Damarquis (writing his thesis/analyzing some of the images from last fall). Thesis will focus on comparing the optical images and SEM images of the samples taken from Mari’s house in November - specifically, both glass and carbon tape samplers were co-located, so we’d like to see if the two different types of imaging methods gives us similar ImageJ output/concentration values to each other. Paul - some image analysis fine-tuning: 1) comparing various threshold settings in ImageJ to see what effect on concentration calculation it might have and 2) manually count and measure some of the particles in (how many?) images to compare and check if ImageJ is getting a close-to-correct count collecting more samples?

Analysis Automation

The analysis protocol we are using in ImageJ should be scripted so that the data tables for our spreadsheet are automatically generated.

Imaging Automation:

As we verify the viability of our analysis, we also should look at options for low-cost automation of the microscopy. This can be conducted with either an automated slide stage or SEM microscope automation. Automated SEM services are available for ~$100 a sample from the RJ Lee Group. It may also be possible Automated slide stages for visible light microscopy are commercially available, but several thousands of dollars. Two open source automated microscope projects exist: Oxford Open Lab Tools microscope: ~$900 in parts plus assembly and 3D printing Trend Africa 3D printed microscope: price unknown. less expensive optics and more 3D printing

Silica Speciation with Passive monitors

Speciation can be useful for an indication of a pollutant’s presence, but might not provide an accurate count of speciated particles, as ~300 particles are needed in a sample to get an accurate reading(Ott et al. 2008).

It is possible to differentiate between crystalline and amorphous silica, which have different health outcomes, using SEM. Currently respirable crystalline silica, but not respirable amorphous silica, is listed as a hazardous air pollutant in six states and is regulated for occupational exposure. The SEM at Chicago State, like many in academic labs, is capable of some automation of X-ray spectrometry and SEM.

It may be possible to speciate silica with polarized light microscopy -- a technique frequently used for identifying silica in lung tissue.

The 'bifringent silicate crystals' in this image is visible as red dots among silicosis fibrules.

Citation: Darrin K. Ott, William Cyrs, Thomas M. Peters. Passive measurement of coarse particulate matter, PM10-2.5. Aerosol Science 39 156 – 167 (2008)

Optical Monitoring, Alerts, and tie-ins to visual observation

Goals

Alerts of potentially high PM concentrations serve to assist community members two ways: understanding in real-time when their health may be at risk, and identifying situations when violations are likely to be observed visually..

Next Steps

We will reach out to the PetCokeAlerts team and the Public Lab Chicago chapter to see how their existing wind-speed based alerts system could work with visual observation and reporting.

Public Lab has been working with the Open Pipe Kit team to develop a framework for simple, field-maintainable nodes sending data from electronic sensors to web services. Following the release (January 2016) of Open Pipe Kit’s latest stable version with email and text message alert system, we are ready to integrate it with existing optical monitors. to add PM-based information to alerts.

We will deploy a prototype version of this in Portland to use in calibrating passive PM monitors.

POWER TAG LINK