Edited* by Karl K. Turekian, Yale University, North Haven, CT, and approved July 18, 2011 (received for review June 10, 2011)
A recent earthquake and the subsequent tsunami have extensively damaged the Fukushima nuclear power plant, releasing harmful radiation into the environment. Despite the obvious implication for human health and the surrounding ecology, there are no quantitative estimates of the neutron flux leakage during the weeks following the earthquake. Here, using measurements of radioactive 35S contained in sulfate aerosols and SO2 gas at a coastal site in La Jolla, California, we show that nearly 4 × 1011 neutrons per m2 leaked at the Fukushima nuclear power plant before March 20, 2011. A significantly higher 
activity as measured on March 28 is in accord with neutrons escaping the reactor core and being absorbed by the coolant seawater 35Cl to produce 35S by a (n, p) reaction. Once produced, 35S oxidizes to 
and 
and was then transported to Southern California due to the presence of strong prevailing westerly winds at this time. Based on a moving box model, we show that the observed activity enhancement in 
is compatible with long-range transport of the radiation plume from Fukushima. Our model predicts that 
, the concentration in the marine boundary layer at Fukushima, was approximately 2 × 105 atoms per m3, which is approximately 365 times above expected natural concentrations. These measurements and model calculations imply that approximately 0.7% of the total radioactive sulfate present at the marine boundary layer at Fukushima reached Southern California as a result of the trans-Pacific transport.
The Fukushima nuclear power plant, situated in northeastern Japan, is one of the largest nuclear power stations in the world, consisting of six boiling water reactors. A recent massive earthquake (9.0 magnitude) (http://earthquake.usgs.gov/earthquakes/eqinthenews/2011/usc0001xgp/) on March 11, 2011, followed by a deadly tsunami has extensively damaged the power plant by disabling the reactor cooling system, which led to nuclear radiation leaks and activation of a 20-km evacuation zone surrounding the plant (http://www.iaea.org). Three weeks subsequent to the earthquake, evidence surfaced of a partial nuclear meltdown in units 1, 2, and 3, visible explosions in units 1 and 3, and a possible uncovering of spent fuels pool associated with units 1, 3, and 4 (http://www.iaea.org). It is suspected that the explosions at unit 3 may have damaged the primary containment vessel. The events at units 1, 2, and 3 have been rated at level 7 (major release of radioactive material with widespread health and environmental effects requiring implementation of planned and extended countermeasures) on the International Nuclear Event Scale (http://www.iaea.org). The earthquake triggered the automatic shutdown of the three operating reactors—units 1, 2, and 3—and the control rods were completely inserted to terminate the nuclear fission reaction occurring within the fuel core. The details of a nuclear reactor design and operations are described by ref. 1. Residual heat produced at the reactor core due to the radioactive decay of fission products requires extended time periods to cool. Heat must be removed by cooling the system to prevent the fuel rods from overheating and causing a core meltdown, which results in injection of neutrons and other fission products into the atmosphere. Spent fuel that has been removed from a nuclear reactor generates intense heat and is continuously cooled to provide protection from its radioactivity. As reported, maintaining sufficient cooling to remove the decay heat from the reactor was crucial; consequently, the spent fuel pool was the main challenge at the affected reactor site after the March 11 tsunami.
Because both on- and off-site power to the plant was disabled, seawater mixed with boric acid (to reduce the neutron flux in the core and thus slow down the nuclear reaction) was continuously pumped into the reactor vessels of units 1, 2, and 3 from March 13 to March 26, 2011 (http://www.iaea.org). On March 17, seawater was sprayed by helicopters and by concrete pumps on reactor unit 3 and spent fuel containment. The steam generated due to overheating was released through the relief valve into the atmosphere to avoid explosive hydrogen buildup within the reactor vessel. Approximately a few hundred tons of seawater were used as a coolant before switching to fresh water. A consequence is that salts and minerals present in seawater become radioactive by reaction with thermal neutrons emitted from the reactor. For example, nuclear detonation in seawater produces 35S by slow neutron capture by 35Cl via the (n, p) reaction (2). As described by ref. 3, mainly two isotopes, 1H and 35Cl, present in seawater absorb the slow neutrons. 35Cl, which is highly abundant in seawater (0.55 mol/kg), absorb neutrons to produce radioactive 36Cl and 32P via 35Cl[n,γ]36Cl and 35Cl[n,α]32P reaction, respectively. 35Cl is also converted to radioactive 35S: 
35Cl[n,p]35S has a substantial cross-section of 0.44 ± 0.01b for thermal neutrons (4, 5). 35S is also produced via 34S(n,γ)35S, but approximately 200 times less than 35Cl(n,p)35S (2). The chemical state of 35S produced after neutron capture is not known, but it is expected that 35S is oxidized into 
and 
(6). We assumed that all 35S produced in seawater is transferred to the atmosphere along with the water steam generated at the reactor core due to excessive heating.
Radioactive 35S (β decay to 35Cl, half-life approximately 87 d) is naturally produced in the atmosphere by cosmic ray spallation of 40Ar (7). Once produced, 35S rapidly oxidizes to 
(approximately 1 s), which is removed from the atmosphere by wet and dry deposition. 
may undergo gas and aqueous phase oxidation to produce 
aerosols, which are removed from the atmosphere. 35S is the only radioactive isotope that simultaneously exists in both gas and aerosol phases and has a suitably short half-life to detect air motions, which renders it a sensitive tracer in understanding boundary layer chemistry and air mass transport on short time scales (8, 9). 35S has recently been measured in sulfate aerosols in the Antarctic atmosphere to understand the boundary layer chemistry and stratospheric-tropospheric air mass exchange (10). 35S measurements in sulfate collected from snow and water samples have been measured to determine the source and age of the ground water and also to understand the melting of glaciers and their contribution to lake and stream water (11).
The simultaneous gas and particle speciation and short half-life render 35S a unique isotope to clock gas to particle transformation and transport. Any 35S produced at the Fukushima nuclear plant as a result of earthquake-related damage and seawater-based cooling, combined with its speciation into both the particle and gas phases, could serve as a point source of artificially made 35S and provides an excellent opportunity to understand gas-to-particle conversion and long-range transport over the Pacific.
We report 35S measurements of atmospheric SO2 and sulfate aerosols collected at Scripps Institution of Oceanography (SIO) Pier, La Jolla (32.8N, 117.3W, 10 m amsl) between March 9 and April 12, 2011. As shown in Fig. 1 (Table 1), 
activity typically varies between 180 and 475 atoms m-3. A significant increase in 
activity (1,501 atoms m-3) was observed on March 28, which is the highest activity ever recorded at this sampling site. Similarly, 
concentrations increased to 120 atoms m-3 on March 28 from its background activity of 30–40 atoms m-3.
35S activity measured in SO2 gas and sulfate aerosol collected at SIO, La Jolla. A significant increase in 
concentration was observed on March 28, 2011, from the natural background 
activity. The higher activity observed at SIO is due to the presence of higher concentration of 
(365 times higher than the natural abundance) at Fukushima and the radiation plume being transported over the Pacific.
The measurements of 35S activity contained in SO2 (gas) and sulfate aerosol collected at Scripps, La Jolla, California, during March–April 2011
We have continuously been measuring 
and 
at SIO since February 2009, and thus the background 35S activity is well characterized at the sampling site before the Fukushima nuclear plant tragedy transpired. The annual average activity in 
and 
is 458 ± 157 and 72 ± 61 atoms m-3, respectively. Brothers et al. (9) reported a similar background activity at the Scripps Pier during 2008. At SIO, 
(
) values fluctuate between 300 and 500 (100–150) atoms m-3 throughout the year except a few days when 
(
) activity spikes to 950 (430) atoms m-3, respectively. This sudden spike in activity is due to stratospheric intrusion events (spring) and the Santa Ana wind events (easterly high pressure drive winds during late fall–winter) (that lead to the mixing of higher altitude air containing higher 35S activity into the marine boundary layer (MBL). The average 
activity observed during spring is 523 ± 190 (Mar–Apr 2009) and 557 ± 156 (Mar–Apr 2010), whereas for 
is 72 ± 30 atoms m-3. The 
activity observed on March 28 is higher (> factor of 2) than the spring average, whereas 
values lie within 2σ deviation. Even though the polar region has the highest 35S production rate, the 
activity measured at a coastal site in Antarctica (Dumont D’Urville station) is significantly lower (900 atoms m-3) than the activity observed on March 28.
We analyzed two different possible natural scenarios that might cause 
concentrations to deviate from the background to the level observed on March 28.
During stratospheric intrusion, stratospheric air masses containing higher 
and 
are mixed to the upper free troposphere and are subsequently downward transported to the lower altitude. A 4-Box model, as described by ref. 10, with the model parameters in Table S1 was used to determine the steady state concentration of 
and 
at Scripps (Table S2) and to calculate the volume fraction of stratospheric air mass mixing into the free troposphere during the stratospheric intrusion event. The model calculation (see SI Text) shows that nearly 40% of stratospheric air mass must mix into the free troposphere to increase the 
activity to 1,500 atoms m-3 as observed on March 28. This is highly unlikely because most of the stratospheric air mass (90%) returns back to the stratosphere within 6 h of cross-tropopausal exchange (12), and only a small fraction (1–10%) is entrained into the free troposphere (10, 12, 13). In addition, relative humidity and surface ozone concentrations do not exhibit any noticeable change at SIO during the sampling time period typically associated with stratospheric incursions. Consequently, it is regarded as unlikely that a stratospheric intrusion event occurring at Scripps or any region lying in the downwind direction from Fukushima caused the observed enhancement in 
and 
on March 28.
Long-range transport of gases and aerosol from east Asia to Southern California is most intense during springtime (14, 15) because of strong midlatitude westerly winds (16). The Fukushima nuclear plant served as a point source of 
and 
between March 13 and 26, and a radiation plume from Fukushima would be expected to reach Southern California (Scripps) approximately between March 20 and April 1 if the transport time is 6–7 d (17, 18). A 10-d back trajectory for air masses arriving at Scripps on March 28 using a HYSPLIT (19) model reveals that the air mass originated at Japan (Fig. 2) and the long-range transport occurred in the boundary layer with a mean transit altitude of 0.9 km. A moving box model, where a well-mixed box moves along the air mass back trajectory (Fig. 3), was developed to calculate 
and 
concentrations at La Jolla due to long-range transport of the radiation plume from Fukushima. The transit time and corresponding transit altitude of the air mass were calculated from a HYSPLIT model for the last 10 d before it arrived at SIO (Table S3). On day 8, the box passes over the region near Fukushima and subsequently spends 4 d in the marine boundary layer before reaching La Jolla on March 28. As shown in Fig. 3, 
aerosol and gaseous 
change their concentration through various processes such as dilution due to eddy diffusion, dry deposition over the ocean, oxidation, and the radioactive decay during long-range transport. The model parameters are described in detail (10) (see SI Text and Tables S1 and S3). On day 10, the initial concentration of 
(
) in the box is taken to be 913 (391) atoms m-3, the same as the average activity in the free troposphere (Table S2). The bidirectional air mass exchange from the box occurs depending on the dilution lifetime (τdil) of 4.9 d (17). The air mass flux into the box is a ratio of the background concentration of 35S, i.e., 
and 
present outside the box at that particular altitude (z) to the dilution lifetime. In the model, we assumed that 35S was released mainly in the 
phase at the reactor core. The model sensitivity test (Table S4) shows that the model results are not sensitive to the dimension of the moving box but is sensitive to the dilution lifetime (τdil). Other parameters, such as oxidation lifetime (τox) and the initial concentration of 
and 
in the box had a very small effect on the model results.
Ten-day back trajectory obtained for the air mass reaching La Jolla, California, on March 28, 2011, from HYSPLIT model indicates that the air mass originated at Japan.
Schematic description of a moving box model developed to calculate the concentration of 
and 
at Scripps pier, La Jolla (California) due to the long-range transport of radiation plume from Fukushima. The atmosphere is divided into four layers: MBL, BuL, free troposphere (FT), and lower stratosphere (LS). τdill is the dilution lifetime of the radiation plume. τox is the oxidation lifetime of 
, whereas τc and τd are the cloud scavenging and dry depositional lifetime of 
, respectively. τr is the removal lifetime of 
(see SI Text). The dotted line (red) represents the mean transit altitude at which air mass transport occurred; it was calculated from HYSPLIT model. The model parameters in the box were changed according to its transit altitude (Table S3).
The model predicts that a 
concentration of 2 × 105 atoms m-3 was present in the buffer layer at Fukushima (Table S5 and Fig. S1). Table S5 shows the effect on the concentration of 
and 
arriving at Scripps due to different proportions of 
and 
concentrations in the buffer layer (BuL) at Fukushima. Even considering the same concentration of 
and 
in the BuL at Fukushima, the model does not show any increase in 
concentrations at La Jolla. This is because the air mass spent 55% of its transit time in the marine boundary layer where 
is rapidly lost due to its oxidation to 
with dry deposition over the ocean (18). It is estimated that nearly 24% of 
is oxidized to 
in the buffer layer. In the marine boundary layer, 66% of 
is lost due to higher dry deposition over the ocean and oxidation to 
, which is in agreement with ref. 20. Based on the model calculation and the measurements, it is determined that nearly 0.7% of the total 
(present at the marine boundary layer at Fukushima) has been transported to the Scripps Pier, La Jolla, California, in a trans-Pacific transport.
The model calculated 
concentration in the MBL (2 × 105 atoms m-3) at Fukushima is higher than the stratospheric concentration (Table S2) and suggests an additional source of 35S production due to the 35Cl(n,p)35S reaction. We calculated the total number of neutrons that leaked from the reactor core to account for our observations. We use leakage for sea water irradiated by the neutrons produced inside the core. Because the reactor core was melted, the neutron releasing outside is termed as leak. The reaction cross-section of 35Cl(n,p)35S was taken from ref. 4. The attenuation length of neutrons in water at room temperature is 2.8 cm and increases at higher temperatures (21). Because of the high absorption cross-section of 35Cl, seawater has more attenuation. The value of the attenuation length of the neutrons in seawater at temperatures higher than 1,000 °C is not known. For simplicity, the attenuation length was taken to be 2.8 cm. The concentration of 
at the source (reactor core) was assumed to be 10 times higher than the model-calculated 
concentration in the marine boundary layer. Considering all the possible reactions of neutrons with seawater (3), we estimate that a total of 4 × 1011 neutrons per m2 were released before March 20 in which a fraction of 2 × 108 neutrons per m2 reacted with 35Cl to make 35S.
We report the observation of radioactive 35S produced by secondary reactions within the reactor core at the Fukushima power plant and estimate the total neutron leakage after the earthquake. The present work also provides a previously undescribed estimate of depositional and oxidation time scales of SO2 and sulfate during trans-Pacific transport due to a singular strong and well-defined 35S source. The sulfur data are unique because of the coexistence as gas and solid and adds previously undescribed insight into sulfur environmental transformational rates.
We thank two anonymous reviewers of the manuscript for their very insightful comments that improved this manuscript. A.P. thanks Elena Coupal for collecting filter papers at Scripps.
Author contributions: A.P. and M.H.T. designed research; A.P. performed research; G.D. contributed new reagents/analytic tools; A.P. analyzed data; and A.P., G.D., and M.H.T. wrote the paper.
The authors declare no conflict of interest.
*This Direct Submission article had a prearranged editor.
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1109449108/-/DCSupplemental.
