Health risks from radioactive particles on Cumbrian beaches near the Sellafield nuclear site

A monitoring programme, in place since 2006, continues to recover radioactive particles (<2 mm diameter) and larger objects from the beaches of West Cumbria. The potential risks to members of the public using the beaches are mainly related to prolonged skin contact with or the inadvertent ingestion of small particles. Most particles are classified as either ‘beta-rich’ or ‘alpha-rich’ and are detected as a result of their caesium-137 or americium-241 content. Beta-rich particles generally also contain strontium-90, with 90Sr:137Cs ratios of up to about 1:1, but typically <0.1:1. Alpha-rich particles contain plutonium isotopes, with Pu:241Am α ratios usually around 0.5–0.6:1. ‘Beta-rich’ particles have the greatest potential to cause localised skin damage if held in stationary contact with the skin for prolonged periods. However, it is concluded that only particles of >106Bq of 137Cs, with high 90Sr:137Cs ratios, would pose a significant risk of causing acute skin ulceration. No particles of this level of activity have been found. Inadvertent ingestion of a particle will result in the absorption to blood of a small proportion of the radionuclide content of the particle. The subsequent retention of radionuclides in body organs and tissues presents a potential risk of the development of cancer. For ‘beta-rich’ particles with typical activities (mean 2 × 104 Bq 137Cs, Sr:Cs ratio of 0.1:1), the estimated committed effective doses are about 30 µSv for adults and about 40 µSv for 1 year old infants, with lower values for ‘alpha-rich’ particles of typical activities. The corresponding estimates of lifetime cancer incidence following ingestion for both particle types are of the order of 10−6 for adults and up to 10−5 for infants. These estimates are subject to substantial uncertainties but provide an indication of the low risks to members of the public.


Introduction
Discharges of liquid effluent from the Sellafield nuclear site to the Irish Sea have occurred since the 1950s and are monitored both by Sellafield Ltd and the Environment Agency (Cefas 2022). Considering potential doses and risks to members of the public, the main radionuclides discharged from the Sellafield nuclear licensed site have been caesium-137, strontium-90, plutonium isotopes and americium-241. Dispersion in the sea leads to some accumulation of discharged radionuclides in seafood and consequently to radiation doses to consumers. Doses are assessed for the critical group of high-rate consumers or the representative person-representative of higher rate consumers (ICRP 2006). The estimated annual effective doses to local seafood consumers peaked around the mid to late 1970s at around 2 mSv (Hunt 1997), principally as a consequence of the actinide content of shellfish. Discharges were subsequently substantially reduced with the introduction of the site ion exchange effluent plant in the mid-1980s and the enhanced actinide removal plant in 1994, with consequent reductions in doses assessed for high-rate fish and shellfish consumption. While annual doses to local seafood consumers are currently around 0.3-0.4 mSv, the main contributors to dose are now enhanced levels of naturally-occurring radionuclides, including polonium-210, from past non-nuclear industries, with only a small contribution from current Sellafield discharges (Cefas 2022).
Prior to 1985, the environmental discharge permit for the Sellafield nuclear site allowed liquid waste to be discharged to the marine environment with or without solid matter. The permit was changed in January 1985 with a variation which stated that liquid discharges should be 'as far as is practicable free of solids' . Any particulates discharged into the marine environment in earlier years would have been dispersed and, depending on tides and currents, could have returned to local beaches (Atkins Ltd 2018). Woodhead et al (1985) presented results for analyses of radioactive items, including particles, recovered on the foreshore following an incident in 1983 when solvent containing 90 Sr, 137 Cs and ruthenium-106 was released via the sea pipeline and resulted in the contamination of local beaches. Since 1983, monitoring of the beaches has been part of the routine environmental monitoring programme at Sellafield (Sellafield Ltd 2021). In addition to suspended particles being discharged via the sea pipeline in earlier years, the dismantling of redundant discharge pipelines in the early 2000s is likely to have led to dispersion to sea of substantial quantities of contaminated 'silts' (Atkins Ltd 2018).
The recognition of the presence of large numbers of radioactive particles and other contaminated items on Cumbrian beaches has been relatively recent. In 2003, a particle was detected containing 24 kBq of 90 Sr (2 Bq 137 Cs) during routine monitoring along the strandline. The detection of this unusual particle by its beta-particle emissions from 90 Sr/ 90 Y, rather than gamma-radiation emitted by 137 Cs, prompted the Environment Agency to require Sellafield Ltd to undertake a review of the potential sources of particles and of the methodology for monitoring large areas of beach (Sellafield Ltd 2021). From 2006, an extensive monitoring programme was introduced using a motorised detection system. A total of 2711 radioactive particles (<2 mm diameter) and 728 larger objects (including gravel and stones) were recovered between 2006 and March 2022 with detected activities of particles of up to around 300 kBq of either 137 Cs or 241 Am and larger objects of up to 3.7 MBq 137 Cs and 618 kBq 241 Am (Sellafield Ltd 2022). Note that two particles with 241 Am activities >300 kBq were recovered in 2007 (357 kBq and 634 kBq) but all other particles have had 241 Am activities <300 kBq. The other main radionuclides present in or on these particles and larger objects were 90 Sr and Pu isotopes.
Radioactive particles have also been recovered from the foreshore at the Dounreay Nuclear Power Development Establishment in Scotland, and on the nearby Sandside Bay. Like the particles on Cumbrian beaches, the Dounreay particles also contain 90 Sr, 137 Cs, Pu isotopes and 241 Am, although with generally lower activities of the actinide isotopes. Detected activities of 137 Cs have been in the range of 1 kBq-100 MBq, most typically around 100 kBq (COMARE 1999, SEPA 2012, Byrnes et al 2020. Although the origins of the Dounreay and Sellafield particles are quite different, the potential risks to the health of beach users are dominated in both cases by the possibility of either prolonged skin contact and localised damage or their inadvertent ingestion and the associated cancer risks (Charles et al 2005, Charles andHarrison 2007). Brown and Etherington (2011) initially provided an assessment of the potential health risks of particles from Sellafield and Oatway et al (2020aOatway et al ( , 2020b updated and extended these initial findings. This paper focusses on risks to health for individuals who might either experience prolonged skin contact with or ingest a Sellafield particle, referring to Oatway et al (2020aOatway et al ( , 2020b for assessments of the probability of such exposures occurring. The inhalation of particles and entry into a wound are also considered briefly. Risks associated with exposure to larger objects were not pursued in detail as they can be expected to be substantially less, either for stationary contact with skin or for inadvertent ingestion (Oatway et al 2020a(Oatway et al , 2020b).

Monitoring and recovery of radioactive particles
Nuvia Ltd is responsible for the beach monitoring program under contract to Sellafield Ltd A vehicle-mounted detection system is used for all accessible areas, with updates from the Groundhog Evolution 2 to Groundhog Synergy system in 2009 and Synergy 2 in 2014 (Sellafield Ltd 2021). Walked surveys using portable equipment have been used to monitor less accessible areas. The main beach areas monitored are shown in figure 1, from St Bees in the North to Drigg in the South, along with the intermediate beaches of Braystones, Sellafield and Seascale. In addition, more distant beaches have been monitored less frequently, including Allonby beach on the North Cumbrian coast.
Items found are classified as either particles (<2 mm diameter), similar in size to grains of sand, or larger objects that include contaminated pebbles and stones. The initially detected radionuclides are principally either 137 Cs or 241 Am, with a small number of finds in which the detected radionuclide is cobalt-60. The majority of particles are classified as either 'alpha-rich' or 'beta-rich' , referring to their content of predominantly either alpha particle of beta particle emitting radionuclides. For a total of approaching 2300 alpha-rich particles recovered between 2006 and March 2022, measured mean and maximum 241 Am activities were about 30 kBq and 300 kBq, respectively (see above for two exceptions recovered in 2007). Analyses of a sub-set of these particles showed that they also contain alpha particle emitting Pu isotopes, principally 238 Pu and 239 Pu, with a range of Pu: 241 Am α ratios centring around 0.5-0.6:1. They also contain the beta emitter, 241 Pu (typically in excess of 241 Am), and can also contain 90 Sr and 137 Cs. The alpha-rich particles are composed mostly of iron and iron oxide and are typically around a few 100 µm in diameter. For the total of around 400 beta-rich particles recovered between 2006 and March 2022, mean and maximum 137 Cs activities were about 20 kBq and 300 kBq, respectively. Analyses of a sub-set of these particles showed that they generally also contain 90 Sr, with widely varying 90 Sr: 137 Cs ratios centring around 0.07:1 and with a maximum ratio for the majority of particles of around 1:1. They may also contain actinide isotopes. The beta-rich particles are of heterogeneous composition, with constituents including graphite, metals and rock minerals.
Larger objects are mainly beta-rich with only six alpha-rich objects having been found to date (Sellafield Ltd 2021, 2022. In general, these larger objects are rock fragments, with 137 Cs activities up to about 4 MBq, but low 90 Sr: 137 Cs ratios, centring around 0.001:1. A total of 14 particles and four larger objects have been detected by measurement of 60 Co activity, with a maximum of 24 kBq.

Skin structure and radiation effects
Radioactive particles in contact with skin present a potential risk in terms of the development of short-term acute ulceration and the later occurrence of skin cancer. Acute damage due to direct cell-killing occurs only at higher doses and has been termed a tissue reaction (previously, deterministic effect), occurring above 'dose thresholds' which relate to specific end-points and below which no effects are seen (ICRP 2007(ICRP , 2012a. Cancer risks are regarded as stochastic and assumed to apply across the dose range with low risks at low doses defined by the assumption of a linear non-threshold (LNT) dose-response relationship (ICRP 2007, NCRP 2018, Shore et al 2018. Different cell types are recognised as the primary targets for radiation-induced tissue reactions and stochastic effects in skin (Charles 1990, Hopewell 1990, NCRP 1999, Charles and Harrison 2007, ICRP 2015b. The skin consists of two distinct layers, the outermost epidermis and the underlying dermis (figure 2). The epidermis is continually renewed, with new cells being produced from stem cells in the basal layer, and dead cells being sloughed from the outer layer of the stratum corneum at the skin surface, with a total turnover time of the order of 4-6 weeks. The dermis has an upper papillary layer, comparable in thickness to the epidermis, with a much thicker underlying reticular region. The papillary dermis is highly vascularised while the reticular dermis is characterised by the organisation of collagen fibres that give the skin its mechanical strength and flexibility (Hopewell 1990).
The epidermis varies in thickness according to body site (ICRP 2002). The basal cell layer extends around the skin appendages, notably the shaft and base of the hair follicles that project deep into the dermis to the dermal/subcutaneous fat interface. At many sites on the body, over 50% of the basal layer stem cells may be associated with the lining canals of the hair follicles. However, it has been suggested (ICRP 2015b) that the origin of basal cell carcinoma, the main type of skin cancer induced by ionizing radiation, may be predominantly a small proportion of intra-follicular basal cells located in the 'rete-pegs' , the bases of the undulations in the basal layer (see figure 2). The depth of the undulating intra-follicular basal layer is between 20 µm and 100 µm over most of the body, although the soles of the feet and the palms of the hands are exceptional sites with epidermal thicknesses in excess of 500 µm. The International Commission on Radiological Protection (ICRP) and International Commission on Radiation Units and Measurements (ICRU) use a nominal average value of 70 µm for adults (ICRP 1977, 1991a, 1991b, ICRU 1997, generally interpreted for dosimetric purposes as a depth of 50-100 µm. ICRP (2002) provides reference data for nominal epidermal depths as a function of age as: 45 µm for newborn and at age 1 year and 5 years, 50 µm at age 10 years, 60 µm at age 15 years, and 70 µm for adults. However, it has recently been concluded that these data will not be used directly to specify target depths for skin cancer in the development of ICRP dosimetric phantoms. Recognising the substantial variation in epidermal thicknesses with body site and that hair follicles may also contain a substantial proportion of the epidermal stem cells, a simplified scheme has been adopted for the most recent ICRP computational phantoms in which the 50-100 µm depth for adults (nominal 70 µm) will also be used at age 15 years, and a slightly wider band of 40-100 µm will be used at age 10 years and younger (ICRP, in preparation).
The dermis is generally between 1-4 mm thick, depending on body site. The well vascularised papillary dermis, closest to the epidermis, is an important target for radiation-induced tissue reactions, resulting from high dose, direct cell-killing (Hopewell 1990, ICRP 1991a, NCRP 1999. The most important tissue reaction that can be produced by radioactive particles, with radiations energetic enough to irradiate the dermis of the skin, is acute dermal ulceration. This is distinctly different from secondary ulceration resulting from the failure of moist desquamation to heal following the irradiation of large areas of skin, which can be seriously debilitating and may require a skin graft or skin flap to affect a repair. Localised damage caused by hot particles will be more readily repaired by cell migration from the periphery of the small very high dose region. Acute ulceration will occur within two weeks of irradiation and heal over a period of several weeks, perhaps leaving a small scar (Hopewell 1990). For less penetrating radiations, such as low energy beta particles (e.g. from cobalt-60 or promethium-147; see figure 2) and alpha particles, damage will be limited to direct cell-killing in the supra-basal epidermis, which can result in acute epidermal necrosis, occurring within two weeks and healing within a few days to leave no scar. This effect is not unique to hot particles but results from superficial damage that is rapidly repaired by normal cell division within the epidermis. While acute dermal ulceration is the main concern following contact with radioactive particles, moist desquamation is the main concern for larger objects (see section 3.2 below).
In considering doses delivered by radioactive particles and the potential for localised tissue damage, it is necessary to specify an area of irradiation as well as the depth of the target tissue. Published data refer to different areas and depths. Early particle studies on pig skin used dose measurements over areas of 1.1 mm 2 at a depth of 16 µm, the so-called skin surface dose, made with a specific instrument (extrapolation chamber). As reviewed by NCRP (1999), Reece et al (1994) showed that localised pig skin exposures at doses in the range of several 100s-1000s Gy (1.1 mm 2 , 16 µm) from 3 mm diameter 90 Sr/ 90 Y disc sources produced acute ulcerations that developed fully within about 2-3 weeks of exposure, produced a scab after 4-5 weeks, and finally healed with a visible scar at about 12-16 weeks. Hopewell et al (1986) showed that pig skin exposures using 1 and 2 mm diameter 90 Sr/ 90 Y sources at doses between ∼100 and 1000 Gy (1.1 mm 2 , 16 µm) produced acute ulcerations within a few days at higher doses, which healed within 1.5-3 weeks. The ED 50 (the dose to produce an effect in 50% of individuals) was estimated to be about 350 and 180 Gy for 1 and 2 mm particles, respectively (NCRP 1999). Limited information on effects on human skin show similar responses to those observed in pigs (Dean et al 1970, Kaurin et al 2001a, 2001b. ICRP (1991aICRP ( , 1991bICRP ( , 2007 and NCRP (1999) have standardised the control of local skin doses using estimates of the average dose over the most exposed 1 cm 2 of skin at a depth of 70 µm, although this dose distribution has little biological significance in relation to the acute ulceration endpoint (see below). The available data show good consistency in ED 50 values for acute ulceration for radionuclides emitting beta particles over a wide range of energies (ICRP 1991a, NCRP 1999, Charles and Harrison 2007, with values of 6, 6, 11, and 8 Gy for 175 Yb (E max ∼ 0.47 MeV), 170 Tm (E max ∼ 0.97 MeV), fissioned 235 UC 2 microspheres (E max ∼ 1.8 MeV), and 90 Sr/ 90 Y (E max ∼ 2.3 MeV), respectively. Similarly, for acute epidermal necrosis resulting from irradiation by lower energy beta particles, ED 50 values of 15, 8, and 12 Gy were obtained for 147 Pm (E max ∼ 0.25 MeV), 60 Co (E max ∼ 0.32 MeV), and 46 Sc (E max ∼ 0.35 MeV), respectively. These values were for 1-2 mm diameter particles and high dose rates. It remains uncertain how such effects, resulting from acute cell death, are influenced by dose rate.
The averaging of doses over 1 cm 2 has no direct biological significance in relation to acute ulceration from small particles occurring over areas of a few mm 2 . The 1 cm 2 averaging is designed primarily for the control of larger area, localised irradiation, for which the end-point of concern is moist desquamation rather than acute ulceration. It has also been suggested that doses should be calculated at a depth of 100-150 µm in evaluating 'hot particle' exposures and depths of 300-500 µm for dermal tissue reactions more generally (ICRP 1991a). While such judgements may be scientifically correct, ICRP (1991bICRP ( , 2007, ICRU (1997) and NCRP (1999) have chosen to calculate and measure doses at the nominal depth of the epidermal basal layer of 70 µm (50-100 µm) in the control of both tissue reactions and stochastic effects. While there is practical merit in this simplified approach, applied to radioactive particles and wider field irradiation, it conceals the substantial differences in reactions of concern and associated differences in depths of target cells, and the mode of cell death. With these caveats, skin doses are evaluated primarily as an average over 1 cm 2 at a depth of 70 µm (50-100 µm) in the following section.

Skin doses and risks from particles
Considering the possibility of radioactive particles on Cumbrian beaches being held in contact with the skin of beach users, the greatest risk of acute ulceration is from beta-rich particles. The maximum 137 Cs activity of the finds to date is about 3 × 10 5 Bq, although the mean 137 Cs activity is an order of magnitude less. Measured 90 Sr: 137 Cs ratios are a maximum of around 1:1 for all but one particle recovered since 2006, with most measurements below 0.1:1. The exception was a particle detected in 2014 which contained 1.7 × 10 5 Bq 90 Sr and 8 × 10 3 Bq 137 Cs; there was also an earlier find in 2003 of a particle with 2.4 × 10 4 Bq 90 Sr and 2 Bq 137 Cs.
As discussed in section 3.1, ICRP interprets the 70 µm nominal epidermal depth as a depth of 50-100 µm for dosimetric purposes, using this for its reference phantoms of 15 years of age and adults and a slightly wider band of 40-100 µm at 10 years of age and younger. Dose rates calculated using two codes, VARSKIN (Hamby et al 2021) and MCNP (Monte Carlo N-Particle Transport code; Goorley et al 2012), gave values of 3 × 10 −6 Gy Bq −1 h −1 for 90 Sr (+ 90 Y) and 1 × 10 −6 Gy Bq −1 h −1 for 137 Cs (+ 137m Ba) for both 50-100 µm and 40-100 µm depths, giving a total of 4 × 10 −6 Gy Bq −1 h −1 for a 1:1 90 Sr: 137 Cs particle. This value does not allow for self-attenuation within the particle which could result in small reductions in dose rates. Figure 3 shows MCNP results for 137 Cs alone and for 90 Sr + 137 Cs (1:1) and measured dose rates for selected beta-rich Sellafield particles, distinguishing those with higher 90 Sr: 137 Cs ratios (0.3-1.7) from those with lower 90 Sr: 137 Cs ratios (<0.06). Maximum measured values are consistent with calculated values. The estimated dose rate from measurements made on a particle recovered in 2018 with 1.3 × 10 5 Bq 137 Cs (Sr:Cs Oatway et al (2011) compared measured skin dose rates for Sellafield beta-rich particles with values for Dounreay particles published previously (Charles et al 2005. There are two main types of Dounreay particle recovered on the Dounreay foreshore and nearby Sandside Bay, containing fragments of fuel from the materials test reactor (MTR) and the Dounreay Fast Reactor (COMARE 1999). While different in origin from Sellafield particles, they are similar to Sellafield beta-rich particles in containing 90 Sr and 137 Cs and presenting a potential risk of acute dermal ulceration. The maximum measured dose rates for Sellafield particles were lower than measured values for Dounreay particles but with similar dose rates per Bq 137 Cs as for MTR particles for which a 90 Sr: 137 Cs ratio of 0.9:1 was assumed.
The data reviewed in section 3.1 suggest that risks of acute ulceration may be evaluated assuming an ED 50 of 10 Gy and a threshold for the observation of any effects of 2 Gy (dose averaged over 1 cm 2 at 70 µm depth in skin); these values were used in previous assessments of Dounreay particles (Charles et al 2005. Applying a dose rate of 4 × 10 −6 Gy h −1 Bq −1 for a 10 5 Bq 137 Cs Sellafield particle (1:1 90 Sr: 137 Cs), the threshold exposure time for acute ulceration would be reached in 5 h and the ED 50 would require an exposure time of around 25 h. As concluded previously with regard to Dounreay MTR particles, it is unlikely that a particle of this activity would remain in a position of stationary contact with skin for sufficient time to result in ulceration, recognising that even small movements of the particle against the skin would reduce the possibility of damage occurring. It was further concluded that a significant risk of observable effects would apply to Dounreay particles with activities of ⩾10 6 Bq 137 Cs, with dose rates of ⩾3 Gy h −1 . Similarly, for beta-rich particles from Sellafield, observable skin ulceration could occur with particles of ⩾10 6 Bq 137 Cs, with higher 90 Sr: 137 Cs ratios (around 1:1). No particles of this level of activity have so far been found around Sellafield.
Larger objects contaminated with 137 Cs will generally result in lower dose rates to skin than particles of the same 137 Cs activity because of their lower 90 Sr: 137 Cs ratios, the effects of self-attenuation by the object, and because the radionuclides may be spread over more than 1 cm 2 . Oatway et al (2011Oatway et al ( , 2020b reported a dose coefficient of 9 × 10 −7 Gy h −1 Bq −1 cm −2 from measurements of dose rates from pebbles, although with a wide variation. The end-point of concern for larger objects is moist desquamation rather than acute ulceration, with ED 50 and threshold values for exposures over several hours of around 30-40 Gy and 20-30 Gy, respectively Lloyd 1996, ICRP 2012a). However, this radiation reaction is highly dose-rate dependent, as illustrated in studies in pig skin (Millar and Hopewell 2007). Note also that the higher threshold and ED 50 values for moist desquamation than acute ulceration are due to the arbitrary averaging of doses over 1 cm 2 in each case; see discussion in section 3.1 that refers to much higher doses in the immediate vicinity of particles. For the most active object found to date of 4 × 10 6 Bq 137 Cs, a stone of approximately 7 cm × 5.5 cm, if it is conservatively assumed that all activity is present on a 10 cm 2 surface in contact with skin, the dose rate could be around 0.4 Gy h −1 (assuming 9 × 10 −7 Gy h −1 Bq −1 cm −2 ), and the ED 50 and threshold as specified above would be reached after about 5 d and 3 d, respectively, and hence would also be impacted by repair of sublethal damage over these extended periods of exposure. It is inconceivable that a large object would be held in stationary contact with the skin for such extended periods.
Estimated dose rates from alpha particles as a function of depth into the epidermis, calculated using MCNP6, are illustrated in figure 4. The energies of alpha particles emitted by Pu isotopes and americium-241 are in the range from 5.1 MeV ( 239 Pu) to 5.5 MeV ( 238 Pu, 241 Am). Associated photon dose rates are low. Alpha particle irradiation from alpha-rich particles containing 241 Am and Pu isotopes will not be effective in causing acute dermal ulceration since their penetration through skin tissue is only to a maximum depth of around 35-40 µm within the epidermis. Although mean epidermal thickness is greater than 45 µm, even for neonates and infants (ICRP 2012a), there is variation between body sites and between individuals. The most recent ICRP assumption for the calculation of skin dose for younger children is that sensitive cells of the epidermal basal layer are at a depth of 40-100 µm (see above). On this basis, epidermal doses resulting from alpha particle irradiation from Pu isotopes and 241 Am can be regarded as being unlikely to be sufficient to cause acute epidermal necrosis which is, as discussed above, a less severe reaction than acute ulceration.
Localised irradiation of skin should also be assumed to pose a risk of skin cancer. While localised skin damage is assessed in terms of dose to the most exposed 1 cm 2 , the risk of skin cancer is assessed for irradiation of the whole skin, an area estimated as 1.9 m 2 for the adult male and 0.48 m 2 for the 1 year-old infant (ICRP 2012a). There have been suggestions that skin cancer risk from ionising radiation is likely to be confined to areas that are also exposed to UV radiation (ICRP 1991a, Little et al 1997 but more recent studies have produced conflicting conclusions and the issue of an interaction of effects is considered to be unresolved (ICRP 2015b). For a hypothetical beta-rich particle that delivered a dose of 10 Gy (1 cm 2 ; 70 µm), the ED 50 for acute ulceration, the average dose over all skin would be 0.53 mGy for the adult male and 2.1 mGy for the infant. The population-averaged risk of skin cancer incidence is assumed by ICRP (2007) to be 10 −1 per Gy. There is evidence, from follow-up studies of the A-bomb survivors at Hiroshima and Nagasaki and from radiotherapy patients, that the risks are substantially higher following exposure at younger ages (ICRP 2015b). A dose of 0.53 mGy implies a risk of lifetime incidence of skin cancer of 5 × 10 −5 , applying the risk factor of 10 −1 Gy −1 . If it is assumed that the risk is ten times greater for young children, a dose of 2.1 mGy corresponds to a risk of lifetime incidence of skin cancer of 2 × 10 −3 . The lethality factor assumed by ICRP (2007) is 2 × 10 −3 , referring to factors of 10 −4 for the more common basal cell carcinoma and 10 −2 for squamous cell carcinoma (ICRP 1991a(ICRP , 2007. Estimated risks of fatality are then around 1 × 10 −7 in adults and 4 × 10 −6 in infants.

Dose estimates
Intakes of radionuclides into the body occur primarily by either ingestion or inhalation, although intake through wounds can also be important in some circumstances, principally in certain occupational situations (ICRP 2007(ICRP , 2015a. For exposures to Sellafield particles, the primary concern is the inadvertent ingestion of either alpha-rich or beta-rich particles, the subsequent absorption of radionuclides into blood, retention in body organs and tissues, and the associated risks of cancer (Oatway et al 2020a(Oatway et al , 2020b, noting that doses will be orders of magnitude too low to cause tissue reactions. Following ingestion of alpha-rich particles, only a small fraction of the radionuclides associated with particles are absorbed into the blood. Oatway et al (2011) summarised experimental results obtained for the absorption of 241 Am and Pu isotopes in rats following ingestion of Sellafield particles. Results for ten particles showed fractional uptake to blood ranging from about 2 × 10 −7 to 2 × 10 −5 for Pu and 4 × 10 −8 to 2 × 10 −5 for Am (the fraction of ingested radionuclide absorbed and entering the bloodstream). For the majority of particles (8 of 10), uptake of both elements was in the range of 10 −7 -10 −6 . In general, comparisons have shown that measurements of uptake in rats and other mammalian species are reasonable predictors of uptake in humans. The use of an uptake factor of 3 × 10 −5 by Oatway et al (2011Oatway et al ( , 2020b) is likely to be conservative; a value of 10 −6 is used here.
Pu and Am absorbed into the blood are distributed to body organs and tissues with long-term retention in the skeleton and liver (ICRP 2015a, 2019). For radiological protection purposes, ICRP uses biokinetic and dosimetric models to calculate effective doses from intakes of radionuclides for different ages at intake by ingestion or inhalation (ICRP 2007(ICRP , 2015a(ICRP , 2019. Doses are integrated over a 50 year period from the time of intake for adults and to age 70 years for children, and referred to as committed effective doses. Using the most recent ICRP methodology and models (ICRP 2007(ICRP , 2015a(ICRP , 2019, and assuming fractional absorption to blood of 10 −6 of the ingested activity, adult dose coefficients are 2.2 × 10 −10 Sv Bq −1 for 238 Pu, 2.4 × 10 −10 Sv Bq −1 for 239/240 Pu, 1.2 × 10 −10 Sv Bq −1 for 241 Am and 2.3 × 10 −12 Sv Bq −1 for 241 Pu. These values are two orders of magnitude lower than those used by Oatway et al (2011Oatway et al ( , 2020aOatway et al ( , 2020b. Values for children have not yet been published using updated methodology but data based on earlier methodology (ICRP 1991b(ICRP , 2012b show that Pu dose coefficients for 10 year-old children are about 10% higher and values for 1 year-old infants are about 70% higher than the corresponding values for adults. Using the most recent methodology, committed effective doses are directly proportional to absorption to blood. ICRP (2012b) also provides dose coefficients for newborn infants, with values about ten times higher than those for the 1 year-old infant because of greater assumed uptake to blood in the immature intestine. It is considered unrealistic to include this age-group in this assessment, both in terms of the extremely low probability of ingestion of a particle and the conservatism of applying a higher value of intestinal absorption.
Committed effective doses provide single values for the control of radiation exposures but conceal, as steps in their calculation, the contributions made by doses to individual organs and tissues and the time-course of dose delivery. For isotopes of Pu and 241 Am, the liver, bone surfaces and red bone marrow are important contributors to overall dose, contributing a total of 86% of the committed effective dose in the example of 239 Pu ingested by adults. Because of the long-term retention of Pu and Am in body tissues and the long half-lives of their radioisotopes, doses are delivered throughout the integration period, although losses by excretion result in decreasing doses. In the example of ingestion of 239 Pu by adults, 50% of the committed effective dose is delivered in 20 years and 80% in 35 years (Tracy Smith UKHSA; personal communication). For the 1 year-old infant, retention times during childhood are shorter and the corresponding figures are 50% in 8 years and 80% in 33 years. For a high activity particle of 300 kBq 241 Am, and assuming a Pu: 241 Am α ratio of 0.6:1 which is a generally typical value considering all recovered particles, the dose coefficient for adults discussed above results in a committed effective dose following ingestion of about 80 µSv. Greater intestinal absorption of the actinides by a factor of 10 (10 −5 instead of 10 −6 ) would result in a proportionate increase in estimated committed effective dose to 800 µSv. For ingestion by a 1 year-old infant, the corresponding estimates of committed effective doses are about 140 µSv and 1.4 mSv, respectively. For particles with more typical activities (mean 3 × 10 4 Bq 241 Am), estimated doses are a factor of 10 lower at about 8 µSv for adults and 14 µSv for 1 year-old infants (ratio of 0.6:1; fractional intestinal absorption of 10 −6 ).
No direct measurements have been made of the intestinal absorption of 137 Cs and 90 Sr from beta-rich particles, but Oatway et al (2011Oatway et al ( , 2020aOatway et al ( , 2020b referred to gut fluid leachate studies showing low availability and dissolution of radionuclides. ICRP (2016) uses a default value of 0.25 for the fractional absorption of Sr into blood following ingestion, applying this value to soluble forms in foods, but also specifying lower values of 0.01 for Sr titanate and 0.0025 for intestinal absorption following escalation of poorly soluble particles from the lungs. The ICRP (2017) assumption for soluble forms of Cs is complete absorption (fractional absorption of 1) but a lower value of 0.1 is also specified for less soluble forms (e.g. Cs in fuel fragments). The fractional absorption values used by Oatway et al (2011Oatway et al ( , 2020aOatway et al ( , 2020b of 0.3 for 90 Sr and 1 for 137 Cs are likely to be conservative and overestimate doses and risks. The lower ICRP values of 0.1 for Cs and 0.01 for Sr appear more realistic estimates for absorption from particles encountered on beaches. Committed effective dose coefficients for ingestion by adults, calculated using the most recent ICRP models, are 1.6 × 10 −9 Sv Bq −1 for 137 Cs, assuming fractional absorption of 0.1, and 1.1 × 10 −9 Sv Bq −1 for 90 Sr, assuming fractional absorption of 0.01 (ICRP 2016(ICRP , 2017. Values for children calculated using updated methodology have not yet been published but previous data showed that values for 137 Cs are largely independent of age while values for 90 Sr are about twice and three times the adult values for intakes by 10 year-old children and 1 year-old infants, respectively (ICRP 2012b). The subsequent organ distributions of Cs and Sr are very different. Cs is distributed relatively uniformly throughout the soft tissues of the body while Sr is retained principally in the skeleton, with doses to skeletal tissues contributing about 80% of the committed effective dose in adults. Retention times for Cs are relatively short, with >90% of the committed effective dose from 137 Cs being delivered within 1 year of intake. For 90 Sr, 30% of the committed effective dose is delivered within 1 year, 50% in 4 years and 80% in 20 years (Tracy Smith UKHSA, personal communication).
The maximum 137 Cs activity of beta-rich particles recovered to date was 3 × 10 5 Bq, but with a low 90 Sr: 137 Cs ratio of 0.002:1. However, assuming a high 90 Sr: 137 Cs ratio of 1:1 and using the dose coefficients for adults given above, the committed effective dose following ingestion of such a particle is estimated to be about 800 µSv for adults. For ingestion by a 1 year-old infant, the corresponding estimate of committed effective dose is about 1.5 mSv. For particles with more typical activities (mean 2 × 10 4 Bq 137 Cs; 90 Sr: 137 Cs ratio of 0.1:1), estimated doses are about 30 µSv for adults and 40 µSv for 1 year-old infants.
The maximum activity particle detected from measurement of 60 Co was 2 × 10 4 Bq. Using the ICRP (2016) dose coefficient for insoluble oxides (absorption of 0.05) of 2.1 × 10 −9 Sv Bq −1 , this activity corresponds to a committed effective dose of 42 µSv for ingestion by an adult.

Cancer risk estimates
The main stochastic effect resulting from low-dose radiation exposure is cancer. The principal source of quantitative information on cancer risks is epidemiological studies of the Japanese survivors of the atomic bombings at Hiroshima and Nagasaki (ICRP 2007). In general, the epidemiological data show a linear dose-response relationship between cancer rates and absorbed dose from gamma rays in the dose range from around 100 mGy to a few Gy. To apply the risk estimates derived from the A-bomb survivor data to lower doses and dose rates, ICRP (2007) used a dose and dose rate effectiveness factor of two for solid cancers (Rühm et al 2016, Shore et al 2017. For leukaemia incidence, the A-bomb survivor data are consistent with the use of a linear-quadratic dose response relationship, with linearity below 0.1 Gy. ICRP uses an LNT dose-response relationship for exposures at low doses and low dose rates (<100 mGy; <5 mGy h −1 ). The consensus view for radiological protection purposes is that this LNT dose-response assumption represents a reasonable overall interpretation of current evidence on radiation-induced cancer at low doses and dose-rates (Preston et al 2003, ICRP 2007, UNSCEAR 2012, 2021, NCRP 2018, Shore et al 2018. ICRP (1991bICRP ( , 2007 specifies nominal risk coefficients as averaged values across populations, all ages and both sexes. In ICRP (2007), detriment is derived from estimates of the lifetime incidences of specific cancers, taking account of the severity of disease in terms of lethality, quality of life and years of life lost. The values used, with detriment dominated by estimates of cancer risks but with a small addition to take account of possible hereditary effects, are given in table 1. Values for the whole population (0-84 years of age at exposure) are somewhat larger than for the working age population (18-64 years of age at exposure) because cancer risks are generally greater for exposures at younger ages because of the longer life-span for expression of risk and greater sensitivity to the induction of some cancers.
For the purposes of assessment of potential stochastic risks from Sellafield particles, it is of interest to consider differences in risk between children and adults and also between lifetime risk expressed in terms of either fatality or incidence. Estimates of lifetime risks of fatality and incidence are shown in table 2 for Age group Incidence a Fatality 0-19 years b 13.5 9 20-79 years 4.5 3 a Excluding skin cancer, addressed separately for skin contact with particles. b Applied to particle intakes by 1 year-old infants and 10 year-old children.
For the purposes of this assessment, risk coefficients are presented as single values, although it should be recognised that these values are subject to substantial uncertainties, with increasing uncertainties with decreasing doses. There are also considerable differences between cancer types in their age-at-exposure responses but the overall differences reflected in table 2 remain valid.
The stochastic risk coefficients presented above (tables 1 and 2) were derived largely from epidemiological studies of the effects of external exposures to gamma rays, principally cancer incidence and mortality data for the Japanese A-bomb survivors. In this assessment, these risk estimates are applied to doses from intakes of radionuclides, including alpha particle emitters, with their highly localised irradiation of tissues. Comparisons of the effects of external and internal exposures are limited by available data but generally support the use of common risk estimates and hence the models used to calculate internal doses (Harrison and Muirhead 2003, Little et al 2007, Marsh et al 2014, ICRP 2021. Data on effects of internal emitters also provide direct information on risks of specific cancers. Inhalation of 239 Pu by workers at the Mayak plant in the Russian Federation, particularly in the early years of Pu production, has caused significant excesses of lung, liver and bone cancer (Sokolnikov et al 2008, Gillies et al 2017, ICRP 2021. However, while an excess of leukaemias would also have been expected on the basis of modelled doses and risks estimated for external exposures, an excess of these cancers was seen only in relation to external dose and not internal 239 Pu dose (Shilnikova et al 2003). This discrepancy is likely to be due to the pattern of radiation exposure of the haemopoietic bone marrow from alpha particles emitted on and in bone mineral and also to relatively low effectiveness per Gy of alpha particles compared with gamma rays (RBE of around 2 instead of the standard assumption for protection purposes of an RBE of 20) in the specific case of the induction of leukaemia (Harrison andMuirhead 2003, Harrison andDay 2008). Studies of populations exposed to radiation as a result of discharges from the Mayak plant into the Techa River have shown an excess of leukaemia consistent with estimated internal doses from 90 Sr and 137 Cs as well as external doses (Krestinina et al 2013).
While cancer risk estimates are derived largely from the follow-up studies of the Japanese A-bomb survivors exposed to external gamma-rays, risk estimates for bone cancer are based on follow-up studies of patients given radium-224 for various medical conditions (ICRP 2007(ICRP , 2015b. The risk estimate used of 5 × 10 −4 Sv −1 relates to average bone dose and would be a factor of nine lower if based on bone surface dose, as calculated in ICRP models (ICRP 2015a(ICRP , 2019. In addition, bone tumours may include types for which there is a threshold dose below which no tumours will be observed and assumption of an LNT dose-response at low doses may overestimate risks (ICRP 2015b).
Estimates of lifetime risks of the incidence of stochastic effects are shown in table 3, applying the risk estimates given in table 2 to doses from alpha-rich and beta-rich particles. For both particle types, the high activity levels in table 3 are based on the highest measured levels of 241 Am and 137 Cs (except for two high activity 241 Am particles mentioned above), together with high Pu: 241 Am and 90 Sr: 137 Cs ratios, noting that no particles with these combinations of characteristics have yet been recovered (see section 2). However, as discussed in section 4.1, dose estimates would be higher if the fractional absorption of the ingested radionuclides were higher, as assumed by Oatway et al (2011Oatway et al ( , 2020aOatway et al ( , 2020b, although they could also be lower given the range of absorption values indicated by the data. For the reasons outlined above regarding doses to skeletal tissues and risks of leukaemia and bone cancer, risk estimates may be conservative, particularly for 241 Am and Pu isotopes in alpha-rich particles, but also for 90 Sr in beta-rich particles. Despite these and other substantial uncertainties in the dose and risk estimates presented in table 3, they provide an Table 3. Estimates of lifetime risk of the incidence of stochastic effects resulting from ingestion of radioactive particles by members of the public. indication of the order of risks that could result from the ingestion of a particle. As discussed by ICRP in Publication 147 (2021), effective dose can be seen to provide 'an approximate indicator of possible health risks' . Inhalation is not considered to be an important route of intake, primarily because only particles of around 10 µm diameter or less are capable of reaching the alveolar region of the lungs where significant radionuclide absorption could take place (ICRP 2015a). Larger particles would be trapped in the nose and upper airways but would then be subject to normal transport processes. Doses delivered to the epithelial lining of the upper region of the respiratory tract are best assessed as for skin, by consideration of the possibility of local damage (i.e. average dose over 1 cm 2 , 70 µm depth).
Considering the possibility of a particle being trapped in a skin wound, committed effective dose can be estimated using ingestion dose coefficients, making adjustments for the extent of any absorption to blood from the wound. In the case of a superficial skin wound affecting the skin only, it is reasonable to assume that any contaminating particle would be removed by washing within a few hours (⩽1 d). There is limited information available on which to base quantitative estimates of short-term absorption of radionuclides to blood from insoluble materials embedded in wound sites (NCRP 2007). A reasonable starting point is to assume that absorption is equivalent to that following ingestion; that is, a central value of 10 −6 for Pu isotopes and 241 Am, 0.01 for 90 Sr and 0.1 for 137 Cs. Based on this assumption, the dose coefficients specified for ingestion can also be applied to intakes through wounds and the corresponding stochastic risks would be as shown in table 3. However, it is also possible that uptake from a wound could exceed that from the alimentary tract. Measurements of the absorption of Pu isotopes and 241 Am from Maralinga test site materials showed that absorption from subcutaneous sites could exceed that following ingestion by one or two orders of magnitude (Australian Department of Primary Industries and Energy 1990).

Discussion
Radioactive particles on Cumbrian beaches in the vicinity of the Sellafield site present a potential risk to members of the public in two main ways: the possibility of tissue reactions from particles retaining fixed contact with the skin, and the potential risk of stochastic effects, principally cancer, from the ingestion of particles. Risks from larger objects are considered to be substantially lower than from particles (<2 mm diameter).
Based on the measured activities of particles recovered to date, it is only the 'beta-rich' particles, containing 137 Cs and 90 Sr, that have the potential to cause acute ulceration of the dermis of the skin, in the unlikely event of prolonged stationary contact. It is considered unlikely that 'alpha-rich' particles would be able to cause the less serious reaction of acute epidermal necrosis, which is a superficial lesion that heals rapidly. For a beta-rich particle with the highest 137 Cs content found to date, 3 × 10 5 Bq 137 Cs, assuming that it also had a high 90 Sr content ( 90 Sr: 137 Cs ratio of 1:1), a maximum dose rate of about 4 × 10 −6 Gy h −1 Bq −1 would apply, taking no account of self-attenuation within the particle. Such a particle could deliver a threshold dose of 2 Gy within about 2 h and the ED 50 in around 8 h (doses averaged over 1 cm 2 skin at a nominal depth of 70 µm). However, most particles for which data are available have lower 137 Cs activities and 90 Sr: 137 Cs ratios of <0.1:1. Given that stationary contact is required for tissue damage to occur, it appears reasonable to conclude that only particles of ⩾10 6 Bq 137 Cs with high 90 Sr: 137 Cs ratios (around 1:1) will pose a significant risk of causing localised skin damage (30 min stationary contact to reach threshold dose of 2 Gy for a 10 6 Bq 137 Cs particle with 10 6 Bq 90 Sr). The probability of encountering a particle is discussed below.
Risks of cancer following the inadvertent ingestion of particles arise from the absorption of radionuclides to blood and their retention in body organs and tissues. In the estimates presented, doses and risks from beta-rich particles are greater than from alpha-rich particles, mainly because of the assumption of low intestinal absorption of the actinides (fractional uptake to blood of 10 −6 ) based on experimental in vivo measurements (Oatway et al 2011). For the ingestion of beta-rich particles of typical activities (mean 2 × 10 4 Bq 137 Cs; 90 Sr: 137 Cs ratio of 0.1), estimated doses were about 30 µSv for adults and 40 µSv for 1 year-old infants. Corresponding lifetime risks of cancer were about 1 × 10 −6 for adults and 5 × 10 −6 for infants, subject to the assumption that an LNT dose-response relationship applies to these increments on background doses and subject to substantial uncertainties as discussed in section 4.2. Doses from individual particles, assuming encounter and ingestion, vary by several orders of magnitude according to their radionuclide content, and could potentially be up to around 1 mSv as illustrated in table 3, although no particles have yet been recovered with a combination of the highest 137 Cs activity (3 × 105 Bq) and high 90 Sr: 137 Cs ratio (1:1). Oatway et al (2011Oatway et al ( , 2020aOatway et al ( , 2020b) estimated the probability of members of the public encountering radioactive particles on the beaches, taking account of habit surveys and estimates of numbers of particles, their activities, and their distribution on beaches. The maximum values, for inadvertent ingestion of beta-rich particles, were probabilities of around 10 −10 yr −1 -10 −9 yr −1 , with substantially lower average values. In assessing overall risk, it is considered legitimate to multiply the risk of ingestion by the risk of health effects following ingestion; ICRP (2007ICRP ( , 2021 refers to potential exposures. On this basis, overall risks of around 10 −16 -10 −14 yr −1 apply to beta-rich particle of average activities, assuming maximum encounter probabilities. While estimated doses and risks from alpha-rich particles of average activities are lower than from average beta-rich particles, greater estimates of the probability of encountering an alpha-rich particle lead to greater overall estimates of risk, of around 10 −15 -10 −12 yr −1 (Oatway et al 2020a(Oatway et al , 2020b. Uncertainties associated with estimates of the probability of encounter are likely to be greater than those associated with dose and risk estimates. Future projections of possible increases in beach usage do not affect the conclusion that these are very low risks. ICRP (2007) recommends a dose constraint for potential exposures of members of the public of 1 × 10 −5 yr −1 .
Inhalation was not considered to be an important route of intake, being limited to very small particles that are likely to be of low activity. The contamination of a skin wound could potentially result in greater absorption to blood and doses to internal organs than from ingestion. Oatway et al (2011Oatway et al ( , 2020aOatway et al ( , 2020b estimated the probability of wound contamination with a particle as around 10 −11 -10 −9 yr −1 on the basis of pessimistic assumptions regarding the likelihood of Cumbrian beach users having an open cut or abrasion. Localised irradiation of skin should be assumed to present a risk of skin cancer as well as local tissue damage (see section 3.2). As discussed above, a beta-rich particle with the highest measured 137 Cs activity of 3 × 10 5 Bq and a 90 Sr: 137 Cs ratio of 1:1 in contact with the skin for 10 h would deliver a dose of about 10 Gy (1 cm 2 ; 70 µm depth). Acute ulceration would then depend on whether the particle was held in stationary contact for this time while cancer risk would not depend on localisation of dose in this way. Because risk estimates for skin cancer induction apply to the uniform irradiation of the whole skin, it is necessary to average the local dose over the total area of the skin (1.9 m 2 in the adult) (Charles et al 2003, ICRP 2007, giving an average skin dose of 0.53 mGy. The population-averaged risk of skin cancer incidence is assumed by ICRP (2007) to be 1 in 100 per Gy. The lifetime risk of skin cancer incidence from the particle is then 5 × 10 −6 , compared with an estimate of lifetime cancer incidence of 4 × 10 −5 for ingestion of such a particle by an adult (table 3). The lethality factor for skin cancer assumed by ICRP (2007) is 2 × 10 −3 .
Considering all potential exposures, the overall conclusion is that the risk to the public from radioactive particles on Cumbrian beaches is very low. The requirements for monitoring of radioactive particles on beaches and reporting results to the Environment Agency are set out in the Sellafield Ltd Site Environmental Permit and the associated Compilation of Environment Agency Requirements (Environment Agency 2021a, 2021b). The Environment Agency has also set out notification and intervention criteria that include the reporting of unusual finds and increases in find rates and/or activities (Environment Agency 2022). Monitoring was reduced in 2020 from a total area of 150 ha yr −1 -105 ha yr −1 , in response to the conclusions of the report from Oatway et al (2020aOatway et al ( , 2020b. The current arrangements expire in 2024 and Sellafield Ltd is currently considering the future programme and whether a further reduction in monitoring might be regarded as being consistent with best available techniques. Even though the overall risks to members of the public are low, continued monitoring is required to ensure that find-rates do not increase and/or that higher activity particles are not present. The design of the monitoring programme will need to ensure that temporal trends in the different areas can be assessed adequately as particles continue to be recovered.

Data availability statement
All data that support the findings of this study are included within the article (and any supplementary files).