15

92

PerspectiveReceived: 7 December 2012 Revised: 4 February 2013 Accepted article published: 10 June 2013 Published online in Wiley Online Library: 11 July 2013

(wileyonlinelibrary.com) DOI 10.1002/jctb.4141

Radioactive liquid waste treatment atFukushima DaiichiPaul Sylvester,a∗ Tim Milnera and Jesse Jensenb

Abstract

The earthquake and tsunami on 11 March 2011 severely damaged the Fukushima Daiichi nuclear plant leading to the mostsevere nuclear incident since Chernobyl. Ongoing operations to cool the damaged reactors at the site have led to the generationof highly radioactive coolant water. This is currently treated to remove Cs-137 and Cs-134 and then passed through a reverseosmosis (RO) unit to reduce the salinity before being cycled back to the reactors. Because only the Cs isotopes are removed, theRO reject water cannot be discharged from the site and this has led to the accumulation of over 200 000 m3 (52 million gallons)of extremely contaminated water which is currently stored on site in tanks. EnergySolutions, in partnership with Toshiba, werecontracted to develop a system to reduce 62 isotopes in this waste to allowable levels. This was a significant technical challengegiven the high background salt content of the waste-water, the variation in aqueous chemistry of the radioactive isotopesand the presence of relatively high concentrations of non-active competing ions (e.g. Ca and Mg) which inhibit the removal ofisotopes such as Sr-89 and Sr-90.c© 2013 Society of Chemical Industry

Keywords: Fukushima; radionuclides; cooling water; ion exchange; waste treatment

INTRODUCTIONThe Fukushima Daiichi Nuclear Power plant is located on theeastern coast of Japan and consists of six boiling water reactors(BWRs). A massive earthquake (measured at magnitude 9.0 on theRichter scale) at 2.46 pm on 11 March 2011 severed off-site powerto the plant and triggered the automatic shutdown of the threeoperating reactors – Units 1, 2, and 3. The control rods in those unitswere successfully inserted into the reactor cores, ending the fissionchain reaction. The remaining reactors – Units 4, 5, and 6 – hadpreviously been shut down for routine maintenance purposes.When the offsite power was cut, backup diesel generators beganproviding electricity to pumps circulating coolant water to the sixreactors. As a result of the earthquake, a tsunami occurred whichhit the Daiichi site approximately 46 min after the initial quake andwas followed by a second tsunami wave roughly 8 min later. Thesewaves were estimated to have been 15 m high when they impactedthe site. This resulted in flooding of the rooms containing theemergency diesel generators which consequently ceased workingcausing power loss to the pumps that circulate coolant water in thereactor. This caused the reactors to overheat due to the high rate ofheat generation due to radioactive decay that normally continuesfor a short time even after a nuclear reactor shut down. Significantmelting of the core elements, cladding and control rods is thoughtto have occurred in units 1 to 3. During the emergency, sea waterwas initially pumped into the reactors to maintain cooling, thoughthis has now been replaced by deionized water. This coolingwater is continually being exposed to the reactor core as well asother damaged reactor components resulting in the generationof a large volume of radioactively contaminated waste-water.1 Todate, this water has just been treated to remove Cs-137 and Cs-134(via an adsorption-based treatment system) and passed througha reverse osmosis (RO) unit to reduce the total dissolved solids(TDS) content. The reject from this RO treatment is now stored on

site in tanks at the Daiichi site and requires treatment. The currentCs-134/137 removal system is incapable of removing the othernuclides present, necessitating the development of a completelyseparate plant to reduce the levels of up to 62 radionuclides toor below the allowable limit. Additional waste-water from the ROunit is continually being generated as a result of the ongoingcore cooling activities and is adding to the wastes stored on site.Decontaminated water that has been treated by the plant stillcontains tritium and will be temporarily stored in tanks on sitewhile a decision is made regarding whether it will be released orsubjected to additional treatment.

BACKGROUNDThe RO reject wastes currently stored at the Daiichi site are variablein composition but in general have the following characteristics:

• Close to neutral pH.• The isotopes responsible for the bulk of the radioactivity are Sr-

89 and Sr-90/Y-90, but multiple other radioisotopes are presentat lower levels.

• The wastes have a high TDS content (up to 20 000 mg L-1), withthe primary ions being Na+, Cl-, SO4

2-, K+, Ca2+ and Mg2+.• Concentrations of Ca and Mg are at levels measured in hundreds

of mg L-1, complicating Sr-89/90 removal.

∗ Correspondence to: Paul Sylvester, EnergySolutions, 100 Center Point Circle,Suite 100, Center Point II, Columbia, SC 29210, USA.Email: psylvester@energysolutions.com

a EnergySolutions, 100 Center Point Circle, Suite 100, Center Point II, Columbia,SC, 29210, USA

b EnergySolutions, 2345 Stevens Drive, Suite 240, Richland, WA, 99354, USA

J Chem Technol Biotechnol 2013; 88: 1592–1596 www.soci.org c© 2013 Society of Chemical Industry

15

93

Radioactive liquid waste treatment at Fukushima Daiichi www.soci.org

Table 1. Major radioactive isotopes in the Fukushima waste

Isotope Half Life

RO Reject (1)

Concentration

Bq mL-1

RO Reject (2)

Concentration

Bq mL-1

Cs-134 2.07 years 2.50E + 00 1.49E + 01

Cs-137 30.2 years 3.90E + 00 2.04E + 01

Co-58 70.9 days 1.20E + 00 < 7.00E−01

Co-60 5.27 years 1.40E + 01 8.63E−01

Mn-54 312.1 days 4.50E + 01 1.08E + 00

Ru-103 39.3 days 5.10E−01 <1.07E + 00

Ru-106 1.02 years 7.80E + 00 <8.53E + 00

Sb-125 2.76 years 1.40E + 02 1.03E + 02

Sr-89 50.52 days 1.10E + 04 1.24E + 04

Sr-90 29.1 years 1.60E + 04 1.13E + 05

Y-90 2.67 days 1.60E + 04 1.13E + 05

Total beta / 4.30E + 04 3.01E + 05

Total alpha / 4.60E−04 Not Measured

Some of the shorter lived isotopes (e.g. Co-58, Ru-103) initially detectedin the wastes during preliminary trials have mostly decayed away andare currently present at much lower levels than indicated in the table.However, significant activities of many other isotopes still remainwhich means the wastes pose a radiological hazard and need to betreated with caution.

In addition to the Sr-89/90, numerous other radioisotopes (bothfission and neutron activation products) have been detected inthe wastes that need to be removed by the treatment system. Theradiological compositions of two of the actual wastes that wereused in preliminary laboratory testing are given in Table 1. ROReject (1) was used in December 2011 trials while RO Reject (2)was used in later trials in May 2012.

LABORATORY TESTINGPrior to designing the treatment system, extensive testing wasperformed using actual waste samples from Fukushima Daiichito identify suitable ion exchange media and adsorbents and todemonstrate the efficacy of the system design. Media were initiallyscreened using simple batch trials to determine distributioncoefficients, Kds. This was accomplished by simply contactinga known mass of media with a specified volume of solution for apredetermined amount of time (usually 24 h) using a mineralogicalroller (the type that is typically used for polishing gemstones). Testsamples were sealed in plastic bottles and packed in the drum inplace of the usual pebbles and grit. The drum was then placed onthe rollers and allowed to rotate for a predetermined time. Themixture was finally filtered and the activity remaining in solutionanalyzed. Kd values were then calculated according to Equation (1):

kd = Ia − Fa

Fa.

v

m(1)

where Ia = initial activity of isotope, Fa = final activity of isotope,v = volume of solution (mL), m = mass of adsorbent (g).

The values obtained were only used as a screen to identify mediafor further testing. Media that exhibited low Kds (<100 mL g-1) werenot tested further.

During the testing, it soon became clear that pre-treatment ofthe wastes was necessary to remove the bulk of the radioactivity

and competing ions that would otherwise interfere with the per-formance of the ion exchange materials and adsorbents. Followingthe pre-treatment stages, multiple ion exchange media were thenused to polish the pre-treated liquids to acceptable levels. Therequired activity levels in the final treated water were very chal-lenging and were close to the limits of detection for many isotopes.

Pre-treatment Stage 1To reduce the activity of a large number of radionuclides, iron(III)chloride was added to the wastes to provide a source of iron forthe generation of an iron(III) hydroxide floc in situ. The iron(III)chloride was added as a 40% solution, causing a drop in pH andnecessitating the addition of sodium hydroxide to raise the pH tonear neutral. The floc was then filtered through a 0.45 µm filter inthe laboratory (later testing used a laboratory scale 20 nm ceramiccross-flow filtration system) and the aqueous phase passed to thesecond pre-treatment stage. The addition of sodium hypochloritewas also found to enhance the removal of some radionuclides,especially Mn-54, presumably by generating higher oxidationstate species more amenable for removal by co-precipitationwith the iron hydroxide floc or adsorption on the surface of thefloc particles. This treatment process is well known in the watertreatment industry and forms the basis of the Enhanced ActinideRemoval Plant (EARP) that has been operating successfully for overtwo decades at the Sellafield plant in the United Kingdom.2,3 Inaddition to Mn-54, the majority of Co-58/60, Sb-125, Y-90, andactinides were removed along with lesser amounts of Sr-89/90and Ru-103/106.

This Stage 1 pre-treatment was subjected to a large series ofparametric experiments to evaluate the effects of iron dosing,hypochlorite dosing and final pH on isotope removal efficiencies.The final parameters selected represent a compromise betweenisotope removal efficiencies, chemical consumption and wastegeneration.

Pre-treatment Stage 2The second pre-treatment stage was designed to remove thebulk of the Ca, Mg and Sr ions that would otherwise impact anySr-selective ion exchange media. Ion exchange media are mostapplicable to the removal of trace species from aqueous wastestreams and the presence of hundreds of mg L-1 concentrationsof competing ions will greatly impact media life (even for theselective Sr removal media) and effectiveness leading to excessivemedia consumption. This latter issue is of particular importancewhen using Sr-selective ion exchange media since they tend tocost many times more than traditional ion exchange resins andgranular media.

Several approaches were considered to remove Ca, Mg andSr from the wastes via the precipitation of insoluble salts. Thesolubilities of a number of salts of these metal ions are shown inTable 2.4

Based upon the solubility data in Table 2, carbonate wasselected as the anion to precipitate the bulk of the Mg, Ca and Srfrom the wastes. Sodium carbonate was added in excess of thestoichiometry of the combined Mg, Ca and Sr, and the pH raisedusing sodium hydroxide to ensure the presence of carbonate(CO3

2-) rather than hydrogen carbonate (HCO3-) anions, thus

minimizing residual alkaline earth metal concentrations in solutionfollowing the precipitation. Using this approach, it was possibleto reduce Mg and Ca levels from hundreds of mg L-1 to less than5 mg L-1 and to remove in excess of 90% of the activity of Sr-89

J Chem Technol Biotechnol 2013; 88: 1592–1596 c© 2013 Society of Chemical Industry wileyonlinelibrary.com/jctb

15

94

www.soci.org P Sylvester, T Milner, J Jensen

Table 2. Solubility of selected metal salt

Solubility (g of salt per 100 g of water)

Metal ion (M) MCO3 MSO4 M(OH)2 M3(PO4)2

Mg2+ 0.18 35.7 0.00069 0.00009

Ca2+ 0.00066 0.205 0.16 0.00012

Sr2+ 0.00034 0.0135 2.25 0.00001

and Sr-90. Precipitation of the Group II metal phosphates was alsoconsidered but was eventually eliminated since excess phosphatecould interfere with the removal of certain radionuclides, such asSb-125, in the downstream adsorption system.

Analysis of the precipitate indicated that it consistedpredominantly of magnesium hydroxide and calcium carbonateas opposed to solely the carbonates of these two cations. This waspresumably due to the very low solubility of magnesium hydroxidein comparison with magnesium carbonate.

Ion exchanger and adsorbent selectionAlthough the two pre-treatment stages were found to remove inexcess of 95% of the total radioactivity, a number of ion exchangemedia and adsorbents were still required to get the activity ofmany isotopes down to the required levels. The selection of ionexchange media was influenced by a number of factors including:

• Media availability. Although many selective ion exchange mediahave been described in the scientific literature, relatively feware commercially available in the quantities required for thetreatment of the waste liquids stored at the Fukushima site.

• Selectivity. The RO waste streams requiring treatment containhigh background levels of inactive ions, such as sodium, chlorideand sulfate. Suitable media must show a high selectivity of targetradioisotopes over these background ions. Conventional strongacid cation exchange resins, strong base anion exchange resinsand zeolites do not have sufficient selectivity to be of use. Inparticular, only inorganic compounds are known to be suitablefor Sr-89/90 under these conditions.5,6

• Physical attributes. Any resin selected must have sufficientphysical integrity to be used in column systems. Althoughpolymeric ion exchange resins generally have good physicalstability, many inorganic media exist as fine powders whichare unsuitable for dynamic column operations. Only inorganicmedia available in pelletized or granular forms were consideredfor testing.

• Capacity. Frequent media change outs will generate excessiveamounts of radioactive waste for disposal and will also causesystem downtime resulting in a decreased waste throughput.Media with either a demonstrated high capacity for the isotopeof interest or predicted high capacities were selected.

Media testingMedia identified from the scientific literature and manufacturers’technical data sheets that appeared to offer some selectivity forthe key isotopes in the waste solutions were extensively testedas described previously. This was achieved using simple batchtechniques and small column experiments using the facilities ofthe nearby (but undamaged) Fukushima Daini plant. A photographshowing part of the treatment column used in the preliminarytesting is shown (Fig. 1).

Figure 1. Column trials used to develop the waste treatment process.

Figure 1 shows part of the series of columns used to treat thewastes after the precipitation pre-treatment steps. Media usedin these columns had previously been screened using batch Kd

determinations. The data obtained from these preliminary trialswere used to refine the selection of the media and parameterssuch as flow rate and pH optimized to maximize the performanceof the media. Over the course of the testing, several iterationsof columns were assessed before the final arrangement of mediacolumns was selected. In the photograph, 50 mL ion exchangecolumns are being used for the testing and the water is passedthrough from left to right at a flow rate of approximately 10 bedvolumes (BVs) per hour using a peristaltic pump.

SYSTEM ENGINEERINGEnergySolutions was responsible for conceptual design of thetreatment system. This process, which would normally be doneover a 6 month time frame, was condensed down to a periodof 6 weeks. The overall system design consists of three distinctprocess steps as discussed previously.

Stage 1 pre-treatmentThe Stage 1 iron hydroxide floc pre-treatment consists of twoconical-bottomed vessels for the ferric hydroxide precipitationand chemical feed tanks. These are operated in sequence withone vessel filling while the other undergoes the precipitation stepfollowed by solid/liquid separation. Stage 1 pre-treatment oper-ates in batch mode, but has to allow a net continuous throughputof 250 m3 day-1 (46 gallon min–1) per train. This is the reasonfor the two large 26.25 m3 (6935 gallon) precipitation vessels.One vessel is filling over a period of 150 min while the effluentin the second vessel is being dosed, settled, decanted and finallyfiltered. This batch treatment cycle takes approximately 100 minto accomplish. After settling, the bulk of the clear supernatant isdecanted and sent to the Stage 2 pre-treatment while the settledprecipitate is passed to the cross flow ultrafiltration (XUF) unitwhere it is concentrated and stored for ultimate disposal. Filtratefrom the XUF unit is also passed to Stage 2.

Stage 2 pre-treatmentStage 2 pre-treatment consists of a single precipitation vesseland the associated chemical feed tanks. Unlike Stage 1, Stage 2

wileyonlinelibrary.com/jctb c© 2013 Society of Chemical Industry J Chem Technol Biotechnol 2013; 88: 1592–1596

15

95

Radioactive liquid waste treatment at Fukushima Daiichi www.soci.org

Figure 2. Photograph of the Fukushima Daiichi coolant water treatment plant (courtesy of TEPCO).

pre-treatment operates in continuous mode with a throughputof 250 m3 day-1 (46 gallon min-1) per train. Filtrate from Stage1 pre-treatment flows continuously into the precipitation vessel,where it is dosed with sodium carbonate and the pH raised toapproximately 12 using sodium hydroxide. This vessel is vigorouslyagitated and has a residence time of approximately 15 min. Slurryoverflows continuously from this vessel into a second ultrafiltrationrecirculation vessel, which feeds into the ultrafiltrationrecirculation loop. Concentrated slurry is intermittently dischargedfrom this loop such that the dilute slurry in-flow is matched bythe filtrate and concentrated slurry out-flow. Each Stage 2 pre-treatment train is capable of processing a minimum of 250 m3 day-1

(46 gallon min-1) in normal operation per train.

Cross-flow ultrafiltration (XUF) unitsThere are two XUF units associated with each treatment system.The first unit produces a ferric hydroxide slurry with a pH of around8 while the second unit generates a calcium carbonate/magnesiumhydroxide slurry with a pH around 12. These wastes are dischargedto separate containment vessels for storage. Both units useceramic membranes with a nominal 20 nm pore size operatedin recirculation mode.

Ion exchange systemAfter the pre-treatment stages, the liquid passes to the AdvancedLiquid Processing System (ALPS) multi-media ion exchange trainwhere it passes through a total of 14 × 1 m3 ion exchange columnscontaining a series of media designed to remove specific isotopesof interest, followed by 2 × 3 m3 disposable columns for finalpolishing. The identification of the media used in the process isproprietary but includes inorganic media with high selectivitiesfor Cs, Sr, Sb and other ions, granular activated carbons (GACs) forthe removal of colloidal species and organics, chelating organicresins for the removal of polyvalent ions and transition metals anda hybrid organic/inorganic media for final effluent polishing.

During the passage through the ion exchange system there isalso a pH change by controlled HCl addition to reduce the pH inorder to optimize media performance. Two booster pumps are alsorequired to maintain the required flow rates through the treatmenttrain. A picture of the completed plant with some specific areasidentified is shown in Fig. 2.

CONCLUSIONSStored wastes from the RO unit used to treat the coolingwaters from the damaged Fukushima Daiichi reactors havebeen successfully decontaminated by a hybrid treatment processconsisting of multiple precipitation, ion exchange and adsorptionsteps to reduce the activities of multiple radioactive isotopesfrom activities as high as 10E+05 Bq mL-1 to levels below theallowable limit. The individual treatment steps, ion exchangemedia and adsorbents were identified using small scale laboratoryexperiments and the process successfully demonstrated at thelaboratory scale using actual waste solutions. These laboratorytrials were then used to design and engineer a full scale systemat the Fukushima Daiichi plant. Three separate treatment trainshave been constructed at the Fukushima site with each capableof treating a flow rate of up to 250 m3 (approximately 66 000 USgallons) of waste per day. The system became operational in April2013.

Wastes generated at the plant include spent ion exchangeresins and sludges generated in the pretreatment steps. For theimmediate future, these wastes will be placed in approved highintegrity containers (HICs) and stored in a purpose built vault whilea decision is made on the final disposal option.

ACKNOWLEDGEMENTSWe would like to thank Masahiko Kobayashi, Junichi Takagi,Mitsuyoshi Sato, Yoshie Akai, Rie Aizawa, Sho Suzuki and TaketoshiSatoh from Toshiba for their vital contributions to this project.We would also like to thank our colleagues at EnergySolutions,especially Paul Townson, Scott Poole, Jennifer Ruffing, Tom Gillen,Steve Liebenow, Colin Austin, Chad McMillan and Joe Jacobsenfor their assistance with this project.

REFERENCES1 http://www.world-nuclear.org/info/fukushima_accident_inf129.html2 Townson P and Gribble N. The Treatment of Radioactive Effluent

Streams by Crossflow Ultrafiltration, Filtration and Submicron ParticleRemoval From Industrial Liquids, Working Party Filtration & Separa-tion, European Federation of Chemical Engineering and Czechoslo-vakian Hydroseparation Working Party, Krkonose, Czechoslovakia(1992).

J Chem Technol Biotechnol 2013; 88: 1592–1596 c© 2013 Society of Chemical Industry wileyonlinelibrary.com/jctb

15

96

www.soci.org P Sylvester, T Milner, J Jensen

3 Townson P, The Treatment of Radioactive Effluent Streams Prior toSea Discharge by Crossflow Ultrafiltration at the BNFL Sellafield (UK)Facility. North American Membrane Society 14th Annual Meeting,Jackson Hole (2003).

4 Lide DR (ed), Handbook of Chemistry and Physics, 83rd edn. CRC PressLLC, Boca Raton, FL (2002).

5 Sylvester P, Behrens EA, Graziano GM and Clearfield A, Inorganic ionexchange materials for the removal of strontium from simulated

Hanford tank wastes, in Advances in Ion Exchange for Industry andResearch, Williams PA and Dyer A (eds). RSC Special Publication No.39,MPG Books Ltd, Bodmin, UK (1999).

6 Sylvester P. Strontium from nuclear wastes: ion exchange, inEncyclopedia of Separation Science, Wilson ID, Adlard ER, Cooke Mand Poole CF (eds). Academic Press, London, 9, 4261–4267 (2000).

wileyonlinelibrary.com/jctb c© 2013 Society of Chemical Industry J Chem Technol Biotechnol 2013; 88: 1592–1596