Of the 180 samples from Japan, 108 were collected in Japan during 2016 while the remaining 72 samples were collected during 2011. These samples included fine surface soils, sediments from drainage areas, and soils from floor mats. Dusts were collected from bulk and surface dust accumulations; including air handling fans, residential air filters, vacuum cleaner bags, automobile air filters, public restroom ventilation fans and from surfaces at public transportation points throughout northern Japan. This study used a mix of samples submitted by volunteers and by the authors. There were controls on the volunteers’ methods used to select samples, however direct sampling oversight was limited. Fifteen scientists and volunteer citizen-scientists collected these samples in areas across Japan, but predominantly in Fukushima Prefecture and Minamisoma City. Sampling locations were in publicly-accessible areas such as bike paths, roadways, sidewalks and public buildings. Permits were received to sample in restricted areas where post-Fukushima meltdown decontamination work was in progress. A majority of these samples were collected from locations in decontaminated zones cleared for habitation by the National Government of Japan.

Sample collection

Sample collection was biased by performing a preliminary visual survey to facilitate collection from areas where fine particulate matter can accumulate, such as low spots on roads or rooftops, air handling fan blades, floor mats and rooftops. For the 2016 sample set (108 of 180 samples), an International Medcom Inspector Alert surface contamination monitor (radiation survey meter) was used to identify samples from within low lying areas and on contaminated outdoor surfaces. Screening introduces a bias to the soil sample set allowing for maximum probability of collecting particulate matter that might contain hot particles. Indoor dusts, HVAC system dusts and auto air filters were randomly selected and no survey meters were used nor were surface radiation measurements taken prior to dust sample collection.

At the time of the 2016 sampling campaign, mapped surface activity data was available from Safecast, an open citizen-led group that collected activity data via a standardized device of their own design, the bGeigie. Uncontaminated areas in Japan (as mapped by safecast.org) have bGeigie-measured activities on the order of 0.08 uSv/hr. or less. Areas of known contamination are on the order of 0.16 uSv/hr. and higher. More than 90 percent of the samples in this study come from the areas Safecast-mapped as 0.16 uSv/hr. and higher (Figure 1). This indicates that the data are more representative of the contaminated zone, rather than of Japan as a whole. Mapping via the Safecast bGeigie proceeds with a plastic alpha and beta shield around the detector element. This makes the device essentially a gamma activity monitor. Thirteen of the 2016 samples from Japan were measured by the primary instrument (Ortec NaI well gamma photon detector) and the bGeigie. With the shield present on the bGeigie so that both devices measured gamma energy only, the R2 value between the two sets was 0.97, a good fit (Figure 2). Without the shield the bGeigie also accumulated beta and alpha energy, so the fit was poorer, with R2 = 0.15. The strong correlation between the bGeigie in gamma mode and the Ortec spectrometer is added evidence that the sampling methodology provides data that is more representative of the contaminated areas in northern Japan, and less so for Japan as a whole.

For each 2016 sample collection location point, a photograph was taken to provide an image of where the sample was collected as well as a record of the GPS location. All Japanese samples were air dried at ambient temperature prior to analysis, then shipped internationally to Worcester Polytechnic Institute in Worcester, MA, USA, for analysis.

Motor vehicle engine air filters process large volumes of air. In Japan private vehicles average 65 liters of gasoline use per month. (Schipper 2009) This fuel requires approximately 638 cubic meters of air for complete combustion. This is about 30 cubic meters per day, which is in the same order of magnitude as a working adult tidal air volume of 10 to 20 cubic meters per day. It was hypothesized that engine air filters in routine use and first installed prior to March 11, 2011 would provide an approximation of the amount of radioactive dusts present in ambient air for each driving region.

Bulk dust and street dust/soil samples were air dried at ambient temperatures prior to analyses. Dust samples containing macroscopic objects and excessive pet hair were sieved to pass a 150 micron brass ASTM #100 screen. Samples were divided and separately screened by gamma spectroscopy to determine if activity was heterogeneously distributed within samples. Samples were sequentially divided if necessary, based on gamma-screening results.

Dust was mechanically removed from automobile and HVAC filters. Motor vehicle air filters varied in usage from nearly new to as much as 55,000 logged kilometers of use. Sample results for dust samples were reported on a kBq kg-1 basis. Eight filter samples with detectable activities but with dust that could not be quantitatively removed from filter media were assigned a dust mass of 1.0 g. This introduced a low bias to these samples. Nine samples had activities greater than 0.25 MBq kg-1. To ensure safe transport, these samples were mass-limited to 3 grams before shipping from Japan.

Included in this set of nine atypically-radioactive samples, was a 300 mg sample of street dust received from a location about 10 km from the Fukushima-Dai-ichi accident site in Namie-machi, Futaba-gun, Fukushima Prefecture. This is in the restricted zone, close to but just outside of the exclusion zone. A very similar sample was collected from Iitate. The particulate matter samples were analyzed by sodium iodide gamma spectrometry. One vacuum bag received from a home in Nagoya, Japan, yielded subsamples with no detectable radioactivity above background, despite a sizable amount of activity for the bag as a whole. (Nagoya is 433 km from Fukushima Dai-ichi). By sample division and hot particle isolation with a pancake detector, a single 1 cm by 2 cm glass slide was prepared, with a small number of microscopic dust particles (with activity > 1 MBq ug-1) mounted via double sided adhesive tape.

Analyses of the samples proceeded sequentially by NaI gamma spectrometry, autoradiography of high activity samples, and scanning electron microscopy / energy dispersive X-ray analysis (SEM/EDS) of individual radioactively-hot particles identified by autoradiography (Moody 2015). Autoradiographs were prepared from the highest specific activity samples using blue-sensitive X-ray film, followed by SEM/EDS analysis of autoradiographically positive portions of the sample. Air filter media that had positive gamma spectrometry results were mounted in a single layer onto double sided adhesive paper sheets. These sheets with dusts were then attached to 3 mm thick copper plates. Vehicle and HVAC air filters were prepared by cutting the filter media from their frames, and mounting the filter media on 3 mm thick copper plates. A sheet of blue-sensitive Xray film was sandwiched with the mounted filters, and exposed in a dark photographer’s box for seven days. The autoradiographs used MidSci® classic blue autoradiography film BX and D76 processing.

All gamma spectrometry data are corrected for geometry, and were standardized against a known activity of 137Cs. Gamma photon analyses used Ortec® NaI and Canberra® GeLi flat plate photon detectors. An Ortec® NaI well detector and 1K MCA were also used. Counting efficiency @ 662 keV was found to be 30% based on use of a certified calibration standard from an Eckert & Ziegler Isotope Products standard source, manufactured and certified on Sept. 12, 2011, with 40.12 nCi of 137Cs. The 137Cs in the standard used for quantitation and any Fukushima-related 137Cs in the samples were of the same approximate age, therefore no 137Cs decay correction is warranted in this five-year study. Nuclides not amenable to gamma spectrometry, such as strontium, were not analyzed in this study.

Samples with evident gamma spectral peaks for uranium, thorium or plutonium were selected for SEM/EDS analyses. All particles were mounted as a monolayer on a 25 mm OD Ted Pella, Inc., PELCO® tape tab-covered aluminum SEM stub. If necessary to improve particle conductivity, the samples were carbon or gold coated prior to SEM/EDS analysis. SEM/EDS work was tested commercially at Microvision Laboratories of Chelmsford, MA, using a Bruker® X-Flash® Peltier-cooled silicon drift detector (SDD). The electron beam current was 0.60 nAmperes, accelerated at a voltage of < 0.5 to 60 keV.