Direct method for the quantitative analysis of surface contamination on ultra-low background materials from exposure to dust
M. L. di Vacri, I. J. Arnquist, S. Scorza, E. W. Hoppe, Jeter Hall
DDirect method for the quantitative analysis of surfacecontamination on ultra-low background materials fromexposure to dust.
M.L. di Vacri a, ∗ , I.J. Arnquist a , S. Scorza b,c , E.W. Hoppe a , J. Hall b,c a Pacific Northwest National Laboratory, Richland, WA 99352, USA. b SNOLAB, Lively, ON P3Y 1N2, Canada. c Laurentian University, Department of Physics, Sudbury, ON P3E 2C6, Canada.
Abstract
In this work we present a method for the direct determination of contaminantfallout rates on material surfaces from exposure to dust. Naturally occurringradionuclides K, Th,
U and stable Pb were investigated. Until now,background contributions from dust particulate have largely been estimatedfrom fallout models and assumed dust composition. Our method utilizes avariety of low background collection media for exposure in locations of interest,followed by surface leaching and leachate analysis using inductively coupledplasma mass spectrometry (ICP-MS). The method was validated and appliedin selected locations at Pacific Northwest National Laboratory (PNNL) andthe SNOLAB underground facility. A comparison between data obtained fromdirect ICP-MS measurements and those estimated from current model-basedpredictors is also performed. Discrepancies of one order of magnitude or higherare observed between estimated and directly measured accumulation rates.
Keywords:
Surface contamination, ultralow background materials, dustradioactivity, rare event physics, ICP-MS. ∗ [email protected] Preprint to be submitted to Nucl. Instr. and Meth. A June 24, 2020 a r X i v : . [ phy s i c s . i n s - d e t ] J un . Introduction Dust is composed of fine particles resulting from the grinding and breakingof materials in the local environment. It includes soil and minerals from rocks,small amounts of pollen, human and animal hair and skin cells, fibers and de-bris from the materials found in the environment. The chemical compositionof dust generally reflects the local chemical composition of soil and rocks, butit can vary depending on local anthropogenic activities and the environment(e.g. indoor vs outdoor or ordinary environment vs cleanrooms). Interest ininvestigating dust composition is generally related to air quality and health riskassessments, especially in indoor and outdoor environments where the concen-tration of dust is significantly high and/or there is a risk associated with thepresence of heavy metals, as reported in [1], [2] and [3]. Dust is also a primaryconcern for a variety of applications. Studies have been performed to investigateand control the effects of dust particulate deposition on artworks, as reportedin [4]. The semiconductor industry is also well known to be susceptible to par-ticulate contamination [5][6]. When there is a need to reduce the impact ofparticle deposition on critical materials and surfaces, operations are typicallyconducted in cleanrooms. Though designed to maintain extremely low levels ofparticulates, cleanroom facilities are not totally free from particulate fallout [7].While still limited research has been conducted to predict the contribution ofparticulate contamination in cleanrooms for very critical applications, the workin [8] reports one of the most accepted models for particle fallout predictions incleanrooms.In the realm of rare event physics detectors, for which the purity of allmaterials included in the detector is a very critical aspect, dust fallout on ma-terial surfaces and its contribution to radioactive background is of significantconcern. The construction of successful ultralow background (ULB) detectorsimplies extensive ultrasensitive assay campaigns to select the purest materialsmeeting the extremely stringent radiopurity levels required for rare event de-tection [9]-[13]. The major contributors to material radioactive backgrounds2re the naturally-occurring primordial radionuclides K, Th and
U andtheir daughters [14]. Limits for these elements in ULB materials can be verydemanding, in the µ Bq · kg − regime or even lower, with greater importancegiven to material closer to the active target and the larger the mass incorpo-rated into the detector. Isotope Pb, the relatively long-lived ( λ of 22 years)progeny in the U decay series, is oftentimes also a concern and can be sig-nificantly out of secular equilibrium from
U. For example, levels of
Pb inmodern, refined lead can be significantly higher (six orders of magnitude) thanlevels assumed from the
U decay series [15][16]. For this reason, and the factthat large amounts of Pb shielding are used in ULB experiments, we chose tomonitor stable Pb fallout rates from dust. Much effort is being dedicated topredict and control the background contribution from dust to material contam-ination in cleanroom facilities [17]-[19]. Predictions typically rely on previouslydeveloped models for dust fallout, such as those described in [8] and [17]-[19],and assumed dust elemental composition and other physical variables. Thiswork demonstrates a method for the direct determination of radiocontaminantaccumulation rates on material surfaces from dust.The method includes the exposure of a material to dust, followed by dis-solution of deposited contamination in 5% nitric acid and solution analysis viaICP-MS. Accumulation rates (e.g., ngday − · cm − ) or the analytes of interestare determined based on the measured quantity of analyte, exposed materialsurface area and time of exposure. Accumulation rates in terms of radioactivity(e.g., µ Bq · day − · cm − ) are calculated from the specific activities (Bqg − ) ofthe analytes. It should be noted that ICP-MS analysis does not detect radioac-tive decays, but, instead, detects the atoms (more specifically the ions) afterbeing mass resolved from concomitant ions in the sample. ICP-MS is a usefultechnique for the ULB physics community in reaching µ Bq/kg sensitivities forradionuclides with very long half-lives (e.g. > years, like Th,
U, and K). Due to the tiny quantities of isotope
Pb, there are just not enoughatoms to efficiently detected this radionuclide directly by ICP-MS. In this work,only stable Pb is directly measured via ICP-MS. Accumulation rates of
Pb3re inferred from measured stable Pb accumulation rates, assuming a specificactivity of 200 Bq
Pb per kg of modern, refined lead from a previous work[20]. It is worth pointing out that we are only studying the
Pb contaminationfrom dust falling out of the air and depositing on material surfaces. This work isnot meant to investigate
Pb contaminations on surfaces from radon progenyimplantation.A variety of locations for studying radiocontaminant accumulation ratesfrom dust were investigated at Pacific Northwest National Laboratory (PNNL,Richland, Washington) and SNOLAB (Ontario, Canada), a deep undergroundclass 2,000 cleanroom, excavated at a depth of 6,800 ft in the working Creightonnickel mine, hosting neutrino and dark matter experiments [32]. This studyprovides a straightforward, ultrasensitive pathway to assess background contri-butions from particulate dust fallout.
2. Experimental
Ultralow background perfluoroalkoxy alkane (PFA) screw cap vials from Sav-illex (Eden Prairie, MN) were used as containers for sample preparations, prepa-ration of reagent solutions and as an exposed dust collection media. Squaresilicon coupons (22 mm per side) were cut from a 100 mm diameter, 500 µ mthick, bare silicon wafer from Virginia Semiconductor (Fredericksburg, VA).The wafer was polished on one side. The polished side of the coupons wasused as the exposed surface. Silicon wafers have been utilized in the past fordust deposition studies in cleanrooms [5]. They are typically low backgroundmaterials [21]. Labware rinsing and preparation of reagent solutions were per-formed using 18.2 MΩ · cm deionized water from a MilliQ system (Merck Milli-pore GmbH, Burlington, MA, USA). Optima grade nitric and hydrochloric acid(Fisher Scientific, Pittsburg, PA, USA) were used. Standard solutions of Thand
U (Oak Ridge National Laboratory, Oak Ridge, TN, USA) were used toquantify Th and U using an external calibration curve. Non-natural isotopeswere chosen as calibration standards to avoid carry over effects from natural Th4nd U standard solutions, given the extremely low signals to be detected. Stan-dard solutions of K, Ca, Pb (Inorganic Ventures, Christiansburg, VA, USA) andFe (High Purity Standards, North Charleston, SC, USA) were used for quan-titation of these elements through an external calibration curve. All labwareinvolved in the study vials, bottles, pipette tips, tongs and materials exposedto dust underwent cleaning and validation before use. A preliminary cleaningwas performed in a 2%v/v Micro90 TM detergent (Cole-Parmer, Vernon Hills,IL, USA) solution, followed by multiple rinsing with MilliQ water. Leachingin 3M HCl and 6M HNO solutions preceded a validation to ensure sufficientlylow backgrounds. The validation step consisted of pipetting a small volumeof 5%v/v HNO into each container, closing, shaking and allowing to incubateat 80 ◦ C for at least 12 hours. Tongs, pipette tips and silicon wafer couponswere soaked in 5%v/v HNO using validated PFA vials and containers. Theleachate was then analyzed via ICP-MS. Any labware failing validation under-went additional cycles of leaching and validation tests until meeting backgroundrequirements. Leachates of Si coupons and PFA vials measured prior to surfaceexposure were used as process blanks. Determinations of K, Ca, Fe, Pb, Thand U were performed using an Agilent 8900 ICP-QQQ-MS (Agilent Technolo-gies, Santa Clara, CA), equipped with an integrated autosampler, a microflowPFA nebulizer and a quartz double pass spray chamber. For lead, thoriumand uranium determinations, plasma, ion optics and mass analyzer parameterswere optimized based on the instrumental response from a standard tuning so-lution from Agilent Technologies containing ca. 0.1 ng · mL − of Li, Mg, Co,Y, Ce, Tl. The instrumental response from Tl was used as a reference signal,in order to maximize the signal to noise ratio in the high m/z range. Oxideswere monitored and kept below 2% based on the m/z=156 and m/z=140 ra-tio (CeO + /Ce + ) from Ce in the tuning solution. Potassium, calcium and irondeterminations were performed in cool plasma with NH reaction mode. Instru-mental parameters were optimized based on the instrumental response from asolution containing ca. 1 ng · g − K and 0.1 ng · g − Ca and Fe, in-house dilutedfrom a stock solution. Detection limits (DLs) were calculated as 3 · StdDev of5nalyte DLK [fgg − ] 30.0Ca [pgg − ] 0.82Fe [pgg − ] 0.66Pb [fgg − ] 70.8Th [fgg − ] 0.61U [fgg − ] 0.79 Table 1: ICP-MS detection limits for K, Ca, Fe, Pb, Th and U, measured as 3 · StdDev of n=3PBs, for each analyte. n=3 process blanks (PBs). Detection limits for each analyte are reported inTable 1. All sample solutions were measured above detection limits.Surfaces were exposed over a time period of about a month or longer. Afterthat, vials were filled with 1.5 mL of 5%v/v HNO3 solution, recapped, shakenand left filled over a few hours for contaminant dissolution. Extreme care wastaken during this operation not to disturb any deposition on the surface. Novisible dust was noticed on the collected media. Silicon coupons were carefullytransferred into validated clean PFA vials. During the transfer, coupons werehandled only from the very end edges using validated ultraclean tongs, makingsure the exposed surface was facing upward during the transfer. Vials containingthe coupons were then filled with 5 mL of 5%v/v HNO3 solution and capped.The coupons were left fully submerged in the nitric acid solution for a few hours.Vials were often shaken, in order to collect and dissolve all the contaminationdeposited on the silicon surface. Solutions were analyzed with ICP-MS for K,Ca, Fe, Pb, Th and U and accumulation rates for these elements were calculated.The total surface from both vials and cap was considered for an exposed surfacefor one PFA vial sample. For the silicon surface, each coupon represented asample and only the upward exposed surface area was considered. We assumedno dust deposited on the face of the coupon in contact with the clean wipe anddid not include contributions from the negligible side surface of the coupons6500 µ m thickness).
3. Results
The method of dust collection and analysis was preliminarily validated ex-posing both dust collection media (PFA vials and silicon wafer coupons) todifferent class cleanroom settings at PNNL, including non-cleanrooms. Valida-tion aimed at assessing the dependence of the method on the material surface,as well as the reliability of obtained results. Sets of three PFA vials and threesilicon coupons were exposed to air in a class 10,000 cleanroom at PNNL inadjacent locations, as shown in Figure 1.
Figure 1: Left: PFA vials exposed to dust in a class 10,000 cleanroom at PNNL. In orderto maximize the exposed surface, both vial and cap were facing upward. The total exposedsurface for each vial (vial and cap) is ca. 19 cm . Right: silicon coupons exposed to dust inthe same class 10,000 cleanroom, in an adjacent location. The total exposed surface for eachcoupon, only accounting for the upward exposed face, is ca. . Results from a 29-day exposure in a class 10,000 cleanroom at PNNL areshown in Figure 2 and listed in Table 2. Vials and coupons were exposed invery adjacent locations, at same height. Results are reported, as the averageand standard deviation of the three replicates of each set (PFA and Si), in termsof radioactivity accumulation rates in µ Bq · day − · cm − . Natural potassium andstable (natural) lead were measured by ICP-MS; their contributions in termsof radioactivity were estimated based on their specific activities. A value of200 Bq Pb · kg − of modern commercial lead was assumed from a previousstudy [20]. Results show no significant difference between the accumulationrates on the two types of surfaces for Th, U and Pb. A minor discrepancy7of a factor of ≈ Figure 2: Measured accumulation rates in µ Bq per day per square cm of K, Pb,
Thand
U on PFA and Si surfaces. Results are reported as the average and standard deviationof three independent replicates. Natural potassium and stable lead were measured with ICP-MS. Accumulation in terms of radioactivity for these elements were calculated based on theirspecific activities. Activities from
Pb were inferred assuming 200 Bq Pb · kg − of modern,refined Pb [20]. Results were compared to data reported by the U.S. Geological Survey(USGS) [22] for the relative content of K, Th and U in the soil of the considered8 ocation Accumulation rate [ µ Bq · day − · cm − ] K Pb Th UPFA vial (3.3 ± · − (2.4 ± · − (7.6 ± · − (2.1 ± · − Si Coupon (1.5 ± · − (2.1 ± · − (2.0 ± · − (3 ± · − Table 2: Accumulation rates values, in µ Bq per day per square cm, measured for K, Pb,
Th and
U on PFA and Si surfaces. Results are reported as the average and standarddeviation of three independent replicates. Natural potassium and stable lead were measuredwith ICP-MS. Accumulation in terms of radioactivity for these elements were calculated basedon their specific activities. A value of 200 Bq Pb · kg − Pb was considered for Pb [20]. area (Pacific Northwest, PNW). According to the USGS, potassium content inthe PNW area (level of 10 ppm) is ca. four orders of magnitude higher than thatof thorium and uranium, whose concentrations are of the same order of magni-tude (ppm levels). Data for Pb are not reported. Dust deposited in cleanroomslooks to be slightly higher in K and/or lower in U/Th relative to distributionsseen in surface soil in the area. Potassium contaminations in cleanrooms aremore likely introduced by particulate from human skin cells of cleanroom users,fibers from their garb and grinding of polymers and other materials handledin the space (instrumentation, labware, etc.) rather than particulate from soil.The minor discrepancy between accumulation rates of potassium on PFA and Sisurfaces could be explained with a different source, and static effect, of partic-ulate introducing K contaminations compared to particulate carrying Th andU. Particle count was performed regularly, twice per week, in the cleanroomduring the exposure time. While nominally a class 10,000 cleanroom, resultsfrom particle count indicated an average cleanroom class better than 1,000.A set of six PFA vials was also exposed to dust in a class 10 laminar flow hoodat PNNL over a 30-day period. Concentrations of K, Pb, Th and U in solutionswere measured within instrumental background levels, showing no significantcontamination was introduced by dust. Data are therefore not plotted.A second test was conducted comparing accumulation rates for Th and
U in a class 10,000 cleanroom and in a non-cleanroom environment (e.g., an9 igure 3: Measured accumulation rates in µ Bq per day per square cm of
Th and
U onPFA surfaces in an office space and in a class 10,000 cleanroom at PNNL. Results are reportedas the average and standard deviation of eight independent replicates. office space). A set of eight PFA vials was exposed to dust in an office spaceat PNNL for 30 days and compared to a set of eight PFA vials exposed in aclass 10,000 cleanroom at PNNL over the same period. Accumulation rates for
Th and
U are shown in Figure 3. They were measured at (1.0 ± · − and (2.7 ± · − µ Bq · day − · cm − for Th and U respectively in the officespace. Accumulation rates for Th and
U in the class 10,000 cleanroomwere (7.6 ± · − and (2.1 ± · − µ Bq · day − · cm − , respectively. Anaccumulation rate scaling of a factor of 100 was observed between a class 10,000cleanroom and an office space at PNNL. Non-cleanroom environments are typ-ically considered of class 1,000,000 [7], although particle concentration highlydepends on geographic location, daily weather conditions, ongoing activities andbuilding ventilation.Measurements of dust accumulation rates have been performed at the SNO-10AB underground research facility, for selected areas. Dust collection was alsoperformed in an office space at the SNOLAB surface building. SNOLAB main-tains a series of fixed air particulate meters (MET ONE 6000P) at various placesin the laboratory. The data from these sensors are made available to the as-troparticle physics community and SNOLAB users. In addition, a set of twelvewitness plates are located around the laboratory at an average height of about8 feet. Tape lifts are used to measure dust levels on the witness plates with aninterval between lifts of about a month [23]. For the study presented herein,whenever possible, we have investigated the same locations as the fixed dustmonitors and witness plates with preference for the most sensitive locations,close to the current and future experiments sites. Figure 4 shows a map of theSNOLAB underground facility along with the selected locations tagged with ared star and a letter notation. Figure 4: Map of the SNOLAB undergroundfacility. Investigated locations are tagged witha red star and a letter notation.
A: South Drift LBL . B: Room 127 . C: Drift F . D: Room 141 . E: SNO+ control room . F: Room 104 . G: Room 123 . I: Drift F/J . J: Room 132 . K: Room 137 .A brief description of the underground locations is listed below [24]:
A: South Drift LBL . 11t is a space in the mezzanine level in the south drift, where the lowbackground screening facilities are located (HPGe detectors, XRF, alphacounters).
B: Room 127 .Ladder lab, between the CUTE and SuperCDMS SNOLAB experimentlocations.
C: Drift F .Main laboratory hallway, close to PICO-50 experimental area.
D: Room 141 .Bottom part of the cryopit which will be hosting the next generationneutrinoless double beta decay experiment.
E: SNO+ control room . F: Room 104 .Transition area (dirty carwash) between the mine drift and the class 2,000cleanroom facility (non-cleanroom environment).
G: Room 123 .Junction area right outside the refuge station lunchroom.
I: Drift F/J .Main hallway, close to SENSEI and DAMIC experimental areas.
J: Room 132 .Chemistry laboratory.
K: Room 137 .Deck on top of the DEAP water tank.All locations are class 2,000 cleanrooms, except location F, a transition areafrom the mine drift to the clean area where the cleaning of materials occursbefore access to the laboratory. An office space on the third floor of the surface12uilding (non-cleanroom, labelled as location H in Table 3) was also investi-gated. Sets of four PFA vials were used for dust collection in each location andexposed over a 42-day period, except locations D and K, where the exposurelasted 50 days. Exposure was conducted during the 20-day Creighton mineshutdown in August 2018, when very limited activities were being carried on inthe laboratory. Measured accumulation rates for K, Pb,
Th and
U,in µ Bq · day − · cm − are plotted in Figure 5 and reported in Table 3. As statedpreviously, a specific activity of 200 Bq · kg − was assumed for stable Pb [20].An average of the four replicates measured values along with its standard devi-ation for each element and location are quoted. In Figure 5, the above groundnon-cleanroom location is represented with an upward black triangle, while theunderground non-cleanroom location is marked with a downward black trian-gle. Underground class-2,000 cleanroom areas are indicated with circles. Emptygrey circle markers refer to underground cleanroom areas where some activitiesduring the exposure time might have triggered higher accumulation rates for K, Th and
U (locations B, G and J, later explained) compared to all theother underground class 2,000 cleanroom locations, indicated with a full greycircle marker. Full black circle markers in Figure 5 show the average accumu-lation rates along with their standard error of the mean (SEM) considering theclass 2,000 cleanrooms shown as full grey circle markers.Among the cleanroom locations, accumulation rates in locations B, G andJ were measured at 10 − µ Bq · day − · cm − for K and 10 − µ Bq · day − · cm − for Th and
U, about one order of magnitude higher compared to thosemeasured in the other cleanroom locations (A, C, D, E, I and K) for the threeradionuclides. In location E, only accumulation rate for
U matched those ob-served in B, G and J, while K and
Th depositions are comparable to all theother cleanroom locations. Differences in measured accumulation rates of con-taminants reflect the location of the cleanroom in the underground facility, localactivities and activities carried out in adjacent areas. An event of dust rate sat-uration was registered by the dust monitor in location B on one day during thevial exposure. Location G is situated between two non-2,000-class-cleanroom13 igure 5: Measured accumulation rates in µ Bq per day per square cm for K, Pb,
Thand U U on PFA surfaces in selected locations at the SNOLAB facility. Results arereported as the average and standard deviation of four independent replicates. Activities from
Pb were inferred assuming 200 Bq Pb · kg − of modern, refined Pb [20]. areas: the already mentioned transition area between the mine and the clean lab(location F, or dirty carwash) and a refuge room. Moreover, location G is themain entrance for all materials introduced in the laboratory. After 20 days ofreduced activity at SNOLAB during the Creighton Mine shutdown, operationsresumed at SNOLAB for the rest of the vial exposure. Drilling work on the wallopposite to the vials was performed in location J during a week of the exposuretime; it might have affected the results from this location. Excluding locationsB, G and J for the aforementioned reasons, average accumulation rates and theirstandard deviations were calculated for K, Th and
U as (5.7 ± · − ,(6.3 ± · − and (6.5 ± · − µ Bq · day − · cm − , respectively. A spread ofabout one order of magnitude in deposition rates is noticed for K, Th and14 ocation Accumulation rate [ µ Bq · day − · cm − ] K Pb Th UA (4.6 ± · − (3.2 ± · − (2.0 ± · − (5.9 ± · − B (1.3 ± · − (2 ± · − (1.3 ± · − (2.6 ± · − C (1.1 ± · − (2.4 ± · − (1.8 ± · − (4 ± · − D (1.3 ± · − (2 ± · − (5.5 ± · − (2 ± · − E (1.6 ± · − (1.9 ± · − (3.2 ± · − (1.5 ± · − F (2.9 ± · − (2.9 ± · − (1.3 ± · − (2.5 ± · − G (4.9 ± · − (5 ± · − (3.5 ± · − (5.8 ± · − H (1.83 ± · − (3.0 ± · − (1.3 ± · − (8.0 ± · − I (8 ± · − (1.9 ± · − (2.1 ± · − (6 ± · − J (4.1 ± · − (9.9 ± · − (2.4 ± · − (1.8 ± · − K (2 ± · − (6 ± · − (9 ± · − (6 ± · − Table 3: Accumulation rates values, in µ Bq per day per square cm, measured for K, Pb,
Th and
U on PFA surfaces in selected locations at the SNOLAB facility. Results arereported as the average and standard deviation of four independent replicates. Accumulationin terms of radioactivity for these elements were calculated based on their specific activities.A value of 200 Bq Pb · kg − Pb was considered for Pb [20].
U and three orders of magnitude for
Pb. Levels measured in these loca-tions, as listed in Table 3, are compatible with those obtained in the cleanroomat PNNL which was measured to be better than class 1,000. Indeed, the mon-itoring of particulate counts via the laser dust monitor at SNOLAB showedcounting rates compatible with a cleanroom class better than 1,000 during theexposure time. Highest levels of K, Th and
U accumulation rates weremeasured in the two non-cleanroom locations, F and H.The wide variability of
Pb accumulation rates observed among clean-room locations reflects the lead material storage/handling history undergroundat SNOLAB. The highest rate was measured in location A, where, during theexposure period, bricks of lead were stored and handled for the construction ofa lead shielding of a HPGe detector. The overall spread in cleanroom locationsspans over four orders of magnitudes, 10 − to 10 − µ Bq · day − · cm − . At non-cleanroom locations, values were recorded at 10 − and 10 − µ Bq · day − · cm − .15mong the four investigated radioisotopes, K contributed the most radioac-tivity from dust, as also observed at PNNL (Figure 2).While the direct detection data employing the ICP-MS method describedabove showcases some baseline accumulation rates for PNNL and SNOLABduring moments of normal (PNNL) and minimal (SNOLAB) activity, our dustinvestigation method could be employed during a wide variety of activities andin different locations. For example, rotating exposures of witness vials could beemployed to understand accumulation rates during specific moments of instal-lation, setup, existence, etc. Depending on the sensitivity required, exposuretimes could range from a single day to many hundreds of days. Rotating ex-posure vial investigations could be setup to disentangle contamination stepsduring complex, multistep installation processes, or understand contributionsfrom dust during moments of high traffic or unusual activity. Moreover, thesame method outlined here for the analysis for K, Pb, Th, and U, could alsobe employed to assess contributions from nearly the entire periodic table. Suchan approach may help ascertain the major source contributors to the dust (e.g.soil, concrete, machinery, etc.), or understand impurities that enter processesthat may compromise performance (e.g., Ba-tagging applications). Given theaccumulation rates above from an unoccupied SNOLAB environment, let us as-sume we are installing a large detector component, such as the 12 m diameteracrylic vessel from the SNO+ detector [25], in one of the monitored locations,for example location C. The intrinsic
Th and
U in the acrylic has beenmeasured in the µ Bq/kg or lower [23] for each isotope, resulting in a total ac-tivity of the order of the order of 10 µ Bq from
Th and
U for the totalmass of the vessel ( ≈
30 ton). The radioactive contamination accumulated onthe entire surface of such a component exposed to dust in location C can becalculated from data in Table 3 and Figure 5, based on the vessel dimensions.For a 12 m diameter sphere in location C, accumulated radioactive contamina-tion would be about 2 mBq · month − from K, 3 · − µ Bqmonth − from Pb,2 · µ Bq · month − from Th and 5 · µ Bq · month − from U. While theseactivities seem small and not significant compared to the activity from the vast16mount of material itself, it is very much useful to have such data to validatethese assumptions and that no unusual moments of “high activity” is happen-ing. Moreover, it should be noted that these values correlate to accumulationrates taken in SNOLAB during moments of little-to-no activity. In future stud-ies, we will assess dust accumulation rates in SNOLAB during time periods of“normal” activity.A typically assumed conservative dust fallout rate is of 10 ng · hr − · cm − in aclass 1,000 cleanroom. A fallout rate of 1-7 ng · hr − · cm − was measured in 1995at SNOLAB, when the laboratory was a class 2,500 cleanroom [26]. Based onthe relative concentrations of K, Th and U in the soil provided by the USGS andthe accumulation rates measured for the three elements (Table 2), dust accu-mulation rates of (3.3 ± · − , (3.3 ± · − , (3.3 ± · − ng · hr − · cm − are obtained respectively from K, Th and U in the cleanroom at PNNL (nom-inally class 10,000, measured better than class 1,000), with a total averageof (1.3 ± · − ng · hr − · cm − , four orders of magnitude lower than the as-sumed one. Average dust accumulation rates in the cleanroom locations in-vestigated at SNOLAB are calculated, in ng · hr-1 · cm − , as (1.4 ± · − fromK, (1.1 ± ± · − from Th and (9 ± · − from U. A widespread of values is observed, due to the spread of accumulation rate data ob-tained from the various cleanroom locations. A total average dust accumulationrate from all elements in all locations of the class 2,000 cleanroom at SNOLABis (2.7 ± · − ng · hr − · cm − and (1.0 ± · − ng · hr − · cm − if excludingdata from Pb (showing a very wide spread). Final averages for the class 2,000cleanroom, including and excluding Pb data, are, respectively one and two or-ders of magnitude lower compared to the assumed deposition rate of 10 ng ofdust · hr − · cm − in a class 1,000 cleanroom. A system of witness plates is in use at SNOLAB to monitor dust particulatefallout in various locations of the underground facility. The exposed plates areanalyzed using X-Ray Fluorescence (XRF) for mine dust and surrogate elements17 ocation Fe [ng · day − · cm − ] Ca [ng · day − · cm − ]XRF ICP-MS XRF ICP-MSB 0.10 ± ± < ± < ± < ± ± ± < ± ± ± ± ± < ± < ± ± ± < ± Table 4: Accumulation rates of Fe and Ca, reported in ng per day per square cm of surfacearea, measured via XRF at SNOLAB on the witness plates and via ICP-MS at PNNL on thePFA vials over the same period of exposure. Averages and standard deviations were calculatedfrom replicates (n=4). from the shotcrete, Fe and Ca respectively, to give an idea of the radiocontami-nant contribution from dust. Dust fallout rates are inferred from determinationsof Ca and Fe over time, assuming the dust is from mine dust and/or shotcrete.A description of how Fe and Ca XRF analyses are performed is reported in [23].Briefly, based on the analysis of rock samples and shotcrete samples from theunderground laboratory previously performed [27], the relative abundance ofTh and U to (the detectable) Ca and Fe is known. Assuming collected dust iscomposed of mainly mine dust and/or shotcrete, the radionuclide fallout rate isextrapolated from the measured Ca and Fe (which are at much higher concen-trations in the dust than Th and U) fallout rates.The dust collection experiment in this work, when possible, has targetedlocations at SNOLAB also monitored by the witness plate method. In partic-ular, witness plates are located in locations B, C, E, F, G and I. Data for Caand Fe accumulation rates from XRF analysis of witness plates relative to thePFA vial exposure period were compared to ICP-MS determinations of Fe andCa accumulation rates on the PFA vials. Table 4 reports data obtained fromthe two techniques over the same exposure time frame. Comparisons between18RF and ICP-MS data are also shown in Figure 6 for Fe (left panel) and Ca(right panel). Results are reported in ng of analyte deposited per day per squarecm of exposed surface. Results from ICP-MS analyses are reported as the av-erage value and standard deviation of four independent replicates. XRF dataare reported as the results single replicates; the uncertainty associated to eachvalue was estimated based on counting statistics. DLs for XRF measurementswere calculated as 3 · StdDev of a process blank (unexposed witness plate). XRFresults for Fe were all above detection limit, 0.13 ng · day − · cm − , except forlocations C and G.Values for XRF analysis in these locations in the left panel of Figure 6represent therefore upper limits. For Ca, instead, XRF measurements were allbelow detection limit, 1.8 ng · day − · cm − , with the exception of the transitionarea between the mine and the clean laboratory (dirty carwash, location F). Allvalues for XRF measurements in the right panel of Figure 6 are upper limits;only the result reported for location F is a measured value. Figure 6: Comparison between accumulation rates measured via XRF at SNOLAB on thewitness plates and via ICP-MS at PNNL on the PFA vials in locations B, C, E, F, G and Iover the same exposure period. All accumulation rates are reported in ng per day per squarecm. Left panel: accumulation rates for iron. Right panel: accumulation rates for calcium.
Table 5: Concentrations of K, Ca, Fe, Pb, Th and U in rock and concrete at the SNOLABsite [27].
For Fe, XRF and ICP-MS data are overall compatible in all locations, exceptfor a discrepancy of about one order of magnitude in locations E. In this location,the vials were not adjacent to the witness plate. They might have been exposedto a different air flow causing the observed inconsistency. A large variabilityamong replicates of the same set of vials was observed in ICP-MS results forFe in locations C, G and I, which resulted in high standard deviations for thesevalues. ICP-MS data for Ca are compatible with detection limits from XRFdata. A small discrepancy (factor of ≈ F e X = F m Y · C X C Y (1)Where F e X is the estimated fallout rate for element X (K, Pb, Th or U),F m Y is the measured fallout rate of element Y (Fe or Ca) and C X and C Y arethe concentrations of elements X and Y in the rock or concrete. Estimatedfallout rates were compared to fallout rates directly measured via ICP-MS.Estimates were obtained for each of the four investigated elements (K, Pb, Thand U) from measured fallout rates of both surrogate elements Fe and Ca.Moreover, calculations were performed in both assumptions of mine dust (rock)and concrete being the sole source of dust particulate. For Fe and Ca, dataobtained from the ICP-MS analysis were used, given that ICP-MS providedactual values, not upper limits, for both elements in all considered locations.A comparison between estimated and measured data for K, Pb, Th and U wasperformed by plotting ratios of estimated over measured fallout rate for eachelement. Figure 7 shows the ratios for all elements, in the assumption of minedust as the only source of particulate. For each element both data obtainedfrom Ca measurements (blue markers) and Fe measurements (orange markers)are plotted. A black dotted line is set at 1 for the ideal case in which theestimated and the measured fallout rates match.Estimates of fallout rates for K are compatible with measured values inlocations B and I. Values estimated from Ca are compatible with measuredvalues in location E, where estimates of rates for K from Fe provides insteadvalues lower than those directly measured. In location G, data estimated from Feare compatible with measured data due to the relatively high standard deviation.In all the other locations, data inferred from both from Fe and Ca overestimateK fallout rates of about one order of magnitude compared to directly measured21 igure 7: From top to bottom: ratios of estimated over ICP-MS measured fallout rates for K,Pb, Th and U for locations B, C, E, F, G and I, assuming the mine dust being the only sourceof particulate contamination. Concentrations of elements in the rock at the SNOLAB site(Table 3) were used to estimate fallout rates for each element both from Ca (blue markers)and from Fe (orange markers). igure 8: From top to bottom: ratios of estimated over ICP-MS measured fallout rates for K,Pb, Th and U for locations B, C, E, F, G and I, assuming the concrete being the only sourceof dust particulate. Concentrations of elements in the rock at the SNOLAB site (Table 3)were used to estimate fallout rates for each element both from Ca (blue markers) and fromFe (orange markers). . Discussions and Conclusion This work provides an effective and valuable method for the direct measure-ment of contaminant accumulation rates on material surfaces after exposure todust. Dust collection is made extremely practical by the use of ultralow back-ground PFA vials, already used in a variety of ultrasensitive analyses [12] and[28]-[31], as collection media. Exposure can be easily performed in any locationof interest, only requiring an operator available for vial exposure and recapping.The possibility to recap collection media after exposure allows for transportationof the media to facilities equipped with an ultrasensitive ICP-MS laboratory,when such a facility is not available on site, without any loss of information.Any research or production facility where particulate contamination from dustis critical to activities can significantly benefit from the method we have devel-oped and proposed in this work. Given that the ICP-MS technique can reachunique sensitivities when coupled with ultraclean procedures, as also demon-strated by the comparison between data obtained at SNOLAB by XRF andICP-MS over the same exposure time, reasonable exposure times (from one dayto about one month, depending on the environment classification) are requiredto investigate dust fallout even in cleanrooms, without the need of any expen-sive dust collection equipment.While originally developed to quantify long-livedradionuclides and stable Pb, the method can potentially investigate any stableelement in the periodic table, providing a very localized elemental fingerprint todust sources that could be utilized to obtain insight into background sources ofdust. Exploiting the ICP-MS multi-element capabilities, target elements can bedirectly monitored, eliminating the need to infer data based on models and/orassumptions, which, as we have demonstrated, can result in significant inaccu-racies. Dust fallout rates and composition in cleanrooms strongly depend on thelocal ongoing activities. Data reported in this work refer to a period of regularactivity at PNNL and reduced activity at SNOLAB. New campaigns of mea-surement at SNOLAB are planned and some of them have already started. Weintend to target a larger number of elements in the future campaigns, aiming at26ocally identifying the major contributors to dust. Moreover, we intend to utilizethe method to quantitatively test dust removing procedures ( e.g., blowing withpure nitrogen, wiping or spraying material surfaces). Although developed tobenefit the design and construction of next generation ultrasensitive detectorsstudying rare events in the field of fundamental physics, the method proposedin this work can be advantageously exploited in all research or industry appli-cations where ultralow levels of contamination, stable or radioactive, from dustparticulate is critical and needs to be monitored and/or rejected [33][34].
5. Acknowledgments
Pacific Northwest National Laboratory (PNNL) is operated by Battelle forthe United States Department of Energy (DOE) under Contract no. DE-AC05-76RL01830. This study was supported by the DOE Office of High EnergyPhysics Advanced Technology R&D subprogram. The authors would like tothank SNOLAB and its staff for support through underground space, logisti-cal and technical services. SNOLAB operations are supported by the CanadaFoundation for Innovation and the Province of Ontario Ministry of Researchand Innovation, with underground access provided by Vale at the Creightonmine site.
References β decay, Physical Review C 97.6 (2018)065503.[14] NCRP., Radiation exposure of the US population from consumer productsand miscellaneous sources. 1986, NCRP Report No. 95.2815] Orrell, J.L., et al., Assay methods for U, Th, and
Pb in lead andcalibration of
Bi bremsstrahlung emission from lead, Journal of Radio-analytical and Nuclear Chemistry 309 (2016) 1271-1281.[16] M.E. Keillor et al., Recent Bremsstrahlung-based assays of
Pb in leadand comments on current availability of low-background lead in NorthAmerica, Applied Radiation and Isotopes 126 (2017) 185-187.[17] B. Mount et al., LUX-ZEPLIN (LZ) technical design report, 2017.[18] D. Tiedt, Radioactive Background Simulation and Cleanliness StandardsAnalysis for the Long Baseline Neutrino Experiment Located at the SanfordUnderground Research Facility, 2013, South Dakota School of Mines andTechnology, Rapid City.[19] E. D. Hallman and R. G. Stokstad, Establishing a Cleanliness Programand Specifications for the Sudbury Neutrino Observatory, Tech. Rep. SNO-STR-91-009 (Sudbury Neutrino Observatory, Department of Physics, Stir-ling Hall, Queens University at Kingston, Kingston, Ontario, Canada K7L3N6, 1991).[20] A. Alessandrello et al., Measurements of internal radioactive contaminationin samples of Roman lead to be used in experiments on rare events, NuclearInstruments and Methods in Physics Research Section B: Beam Interactionswith Materials and Atoms 142 (1998) 163-172.[21] G. Heusser, Low-radioactivity background techniques, Annual review ofNuclear and Particle science 45 (1995) 543-590.[22] USGS. Available from: https://pubs.usgs.gov/of/2005/1413/maps.html.[23] J. Boger et al., The Sudbury neutrino observatory, Nuclear Instrumentsand Methods in Physics Research Section A: Accelerators, Spectrometers,Detectors and Associated Equipment 449 (2000) 172-207.[24] SLDO-UGL-AR-0101-02, 03, 04, 07 SNOLAB drawings, unpublished.2925] A. Bialek et al., A rope-net support system for the liquid scintillator detec-tor for the SNO+ experiment, Nuclear Instruments and Methods in PhysicsResearch Section A: Accelerators, Spectrometers, Detectors and AssociatedEquipment 827 (2016) 152-160.[26] Jagam, P., Technical Report No. SNO-STR-95-050, unpublished.[27] I.T. Lawson, Analysis of Rock Samples from the New Laboratory, STR-2007-003 SNOLAB Technical Report, 2007.[28] I.J. Arnquist et al., Mass spectrometric assay of high radiopurity solidpolymer materials for parts in radiation and rare event physics detectors,Nuclear Instruments and Methods in Physics Research Section A: Accel-erators, Spectrometers, Detectors and Associated Equipment 943 (2019)162443.[29] I.J Arnquist and E.W. Hoppe, The quick and ultrasensitive determinationof K in NaI using inductively coupled plasma mass spectrometry, NuclearInstruments and Methods in Physics Research Section A: Accelerators,Spectrometers, Detectors and Associated Equipment 851 (2017) 15-19.[30] B.D. LaFerriere et al., A novel assay method for the trace determination ofTh and U in copper and lead using inductively coupled plasma mass spec-trometry, Nuclear Instruments and Methods in Physics Research SectionA: Accelerators, Spectrometers, Detectors and Associated Equipment 775(2015) 93-98.[31] I.J. Arnquist, M.L. di Vacri and E.W. Hoppe, An automated ultra cleanion exchange separation method for the determinations of
Th and238