Tritium Transport, Permeation, and Control

Paul Humrickhouse

Idaho National Laboratory

Fusion Safety Program

4th IAEA DEMO Programme Workshop

Karlsruhe Institute of Technology

November 15, 2016

Overview

• Tritium in DEMO: some magnitudes

• Fundamental transport phenomena

– Permeation, regimes

– Isotope effects

– Trapping

• Mitigating tritium losses

– Extraction systems

– Permeation barriers

• Are they necessary?

• Perspectives from the FNSF study

– Parameter uncertainties

Tritium in DEMO

• In DEMO it will be necessary to breed tritium at the same rate it is consumed (55.6 kg/GWf-year)

• The tritium production rate in DEMO will need to be ~104-105 times that of ITER or fission reactors:

• Losses from the blanket and heat transfer systems must be small (~1 g/yr) in order to meet environmental and safety requirements- dose to public from normal operations2 < 0.1 mSv/yr (10 mrem/yr)

• More efficient power generation requires higher temperatures, but these facilitate tritium permeation, which must be mitigated:

• DEMO losses must be <10-5 of the produced tritium

PWR1 CANDU1 Gas-cooled

reactor1

Molten salt

reactor1

ITER DEMO

(2-3 GWfus)

T generated (kg/y) 0.000075 0.1 0.002 0.09 ~0.004 110 - 170

1H. Schmutz, INL/EXT-12-26758, 2012 2DOE-STD-6002-96, “Safety of Magnetic Fusion Facilities: Requirements”

Fundamental Transport Phenomena

• Implantation (FW)

• Diffusion

• Trapping

• Recombination

• Adsorption/dissociation

• Solubility in solids and liquids also plays a key role

CT2,Bulk

2

2 2T,FWC KΓ rT

Wall diffusion

x

C -DΓ m

m

2

rT2 FW1T,C K

Molecular

Recombination Flux

22 TdT p KΓ

Implantation Flux

~ 1x1020 m-2-s-1

Plasma

t

Ct

Bulk Trapping

Gas Driven Permeation • Three things happening:*

• At each surface:

– Molecules dissociating:

– Atoms recombining:

• In the bulk solid:

– Diffusion:

• If surface dissociation/recombination are fast relative to diffusion (diffusion limited):

• Say C2 is small…

• If diffusion is fast relative to surface dissociation/recombination (surface limited):

1 2

C1

Jd,1

Jr,1

Jr,2

Jd,2

C2

JD

2Tdd PKJ

2CKJ rr

x

CCDJ 12

irid JJ ,, isi PKC

r

ds K

KK

Sieverts’ Law,

where is the solubility

0CCi 0DJ 1,1,2,2

1drr JJJ

Permeation Regimes

• When diffusion is rate limiting (W’ >> 1):

• Define permeability:

• When surface effects are rate limiting (W’ << 1):

• A dimensionless number* tells us where the transition is:

• Diffusion-limited (~√P) transport is usually assumed if parameters unknown

• Note that it only makes sense to talk about permeability in this particular regime

• When surface-limited, fluxes are lower than expected from diffusion-limited theory

*Ali-Kahn et al JNM 76/77 (1978) 337-343.

x

PDKJ s

PKJ d

sDK

PxKW d

x

PJ

Perkins and Noda JNM 71 (1978) 349-364.

Isotope effects

• Hydrogen (or deuterium) can recombine with tritium, and their presence alters tritium permeation rates

• Define

• “Surface-limited” and “Diffusion-limited” apply to the sum of all isotopes:

HT

Ts

PPx

PDKJ

Td PKJ2

1

Surface Limited Diffusion Limited

1W 1W

T

H

s PPx

DKJ

Ts P

x

DKJ

“Isotope-limited”

DK

PxKW

s

dDK

PPxKW

s

HTd

Flux unchanged

but inventory in

solid suppressed!

Single

Isotope

Two

Isotopes

HT PP

HT PP

HTTT PPP21

2 HTHH PPP

21

2

Surface-limited and

“Isotope-limited” fluxes have same

dependence on PT

but different

proportionality

constants

Humrickhouse JNM (in preparation)

Tritium Trapping

• Tritium inventory in the bulk material can be affected by traps, which:

– Increase the energy required to move from site to site (i.e. alter the effective diffusion

– Potentially increase the tritium inventory of the material

• Trapping sites result from:

– Impurities and structural irregularities arising from cold work

– Precipitation of alternate phases including gas bubbles

– Radiation damage

trmttt CαCfα

t

C

Ct – Trapped concentration (m-3)

αt – Trapping rate coefficient (s-1)

ft – Probability of landing in a trap site (-)

Cm – Mobile concentration (m-3)

αr – Release rate coefficient (s-1)

kT

E

N

Ccf

D tor

t

o

ttt exp;;

2

D – Tritium diffusion coefficient (m2-s-1)

λ – jump distance or lattice constant (m)

cto – Trap site concentration (m-3)

N – Bulk material atom density (m-3)

o – Debye frequency (s-1)

Et – Trap energy (eV)

Tritium Trapping (2)

• Trap properties are being measured for neutron irradiated (HFIR) materials exposed to tritium plasmas (TPE, ion fluence: 1025-1026 m-2)

• Trap density measured by nuclear reaction analysis (NRA) – shows some deuterium trapped in tungsten even at 500 C

• Thermal desorption reveals a distribution of trap energies for irradiated tungsten

* M. Shimada et.al., J. Nuc. Mat., (2011) in press

TMAP simulation of TDS spectrum

for 0 dpa W, 200C

TMAP simulation of TDS spectrum for

0.025 dpa W, 200C

Tritium flows and loss paths

Breeding

Zone

Breedin

g Zone

Breedin

g Zone

Coolan

t

Coolant

Tritium

Extraction

System

Coolant

Purification

System

Sec

on

dar

y

Pri

mar

y

Back to Blanket

Back to Blanket

• Losses:

– From breeding zone to coolant (permeation through structure)

– From breeder and coolant pipes to building (permeation through pipe walls)

– From primary to secondary coolant (permeation through HX walls)

To Tritium Plant

To Tritium Plant

Minimizing tritium permeation losses

• Our two primary strategies for managing tritium permeation losses are:

– Efficient tritium extraction systems

• Efficient extraction keeps circulating inventories low, reducing the concentration gradients that drive permeation

– Application of permeation barriers to structural materials

• Blocks permeation, potentially increases circulating inventory

• Extraction concepts attempt to do the following:

– Provide a medium (purge gas, getter, etc.) where tritium will preferentially accumulate, relative to structural materials

– Maximize the contact area of the breeder/coolant with this medium

– Minimize the transport distance through the breeder/coolant to reach this medium

– Maximize the residence time in the extraction system (i.e. reduce the flow rate)

Tritium extraction concepts

• Immersed getters

• Liquid getter/cold trap

• Release to purge gas or vacuum

– Bubblers, spray/droplets, extraction columns

• Vacuum permeator

Immersed Getters

• Getters such as U, Zr(Co), Pd, Ti, etc. are commonly used to remove tritium from gases (Example: a 60g U bed can hold 2g tritium)

• Immersed getters (V, Nb, Ta) have been proposed in the past to remove tritium from PbLi

• Principle demonstrated at small scale with V in PbLi, though not to saturation1

L. Sedano, Ciemat report, 2007.

1H. Feuerstein, Fusion Technology

(14th SOFT), 1986, p. 646

• Issues:

– Integrity of material (getter beds for gases are reduced to fines and require ceramic filters)

– Lifetime of beds under cyclic loading (e.g. daily)

– Deleterious effects of impurities including oxygen

Liquid metal getter with cold trap

• Concept investigated at KIT in the 1980s

– Intermediate NaK loop proposed, which acts as a tritium getter

– NaK cooled so as to precipitate solid hydrides

– Tritium removed from solid hydrides by vacuum pumping

• Might also take the form of a thin film between concentric HX tubes

– Processing rate required for fusion may imply large heat loss

– Li is a possible alternative to Na/NaK but subsequent extraction is more difficult

• High saturated concentration

• Less chemically reactive than Na/NaK

Primary

Intermediate/

Secondary

Alkaline metal film

(Na, NaK, or Li)

Concentric

HX Tubes

J. Reimann, Fusion Technology

(14th SOFT), 1986, p. 1579

Droplets in vacuum

• Concept: spray coolant as small droplets into a purge gas or vacuum

• Small droplets provide high surface area and small transport distance

• “Vacuum Disengager” proposed for HYLIFE-II IFE design study1

• Analytical solution and numerical models suggested 99.9% efficiency; no experiments performed

• Now under investigation for PbLi (as Vacuum Sieve Tray)2

• Different analytical solution and numerical models suggest 70% efficiency achievable

• Measured extraction was lower than predicted by 10x, but models depend on (uncertain) solubility and diffusivity

1T. Dolan, Fus. Tech. 21 (1992) 1949-1954

2F. Okino, FED 87 (2012) 1014-1018

Compact Mass Extractor • Gas/Liquid Contactor – planned for HCLL (TBM and

DEMO)

• Structured packing disperses PbLi flow and creates a large interfacial area between PbLi and gas

• 30% efficiency for single column as tested in MELODIE loop1

• HCLL DEMO (14 inventory re-circulations per day) requires at least 80% efficiency2 even with permeation barriers (with ~100x reduction factor)

• Larger scale tests, optimization planned at TRIEX loop (ENEA)

Sulzer Column

20 cm

Structured

packing

2O. Gastaldi et al. FED 83 (2008) 1340–1347

1N. Alpy et al. FED 49-50 (2000) 775-780

1B. Merrill, 2005/06/15 ARIES meeting

Vacuum permeator

• For a DCLL blanket, higher flow rates must be processed than for an HCLL

– Simple scaling from most efficient MELODIE tests indicates that for a DCLL blanket (~470 inventory re-circulations/day), 240,000 extraction columns would be required1 (!)

• DCLL flow rates are much higher (~470 inventory re-circulations/day required)

• Similar performance in a much smaller device is potentially achievable with a vacuum permeator

• Concept: a shell-and-tube mass exchanger with tritium-laden primary, vacuum secondary, and high-permeability, thin-walled tubes

• Required efficiency depends on the reactor design (losses), but:

– What is achievable?

– How does it scale?

Extraction – Vacuum Permeator

• Analytical solution based on axial and radial transport mechanisms reveals rate-limiting phenomena

– Efficiency:

– Dimensionless numbers:

– : Diffusion through membrane is rate-limiting

– : Mass transport to membrane is rate-limiting

– : Tritium swept through before permeating

• Optimization (minimum size) based on required efficiency, other design constraints is straightforward

• Device can be compact with very high permeability membrane materials (e.g. group V metals: Nb, V, Ta)

• Concept not yet demonstrated experimentally but a number of experiments are planned around the word

Tantalum

650 ˚C

η 0.95

Tubes (#) 687

Tube length (m) 12.4

Velocity (m/s) 3.30

Volume (m3) 0.89

ζ 80.0

1exp1

t =2KTL

vri ioilT

s

rrrKK ln

P. W Humrickhouse and Brad J. Merrill, Fusion Science and Technology 68 (2015) p. 295-302.

Example:

Optimized for FNSF (1 of 4)

1

1

t >>1

Group 5 Metals - oxidation

• Group 5 metals have very high tritium permeabilities and from that standpoint are promising tube materials

• They are compatible with PbLi, but the oxygen partial pressure on the vacuum side must be kept below 10-10

Pa to prevent oxidation1

• Application of a Pd coating can prevent this, and commercial hydrogen purifiers based on this concept are available

• They have a very narrow range of operation around 400 ˚C

– At lower temperatures, hydrides form and embrittle the structure

– At higher temperatures, Pd and substrate diffuse together, reducing (irreversibly) the tritium permeability

1R. Kurtz, 2005 ITER TBM meeting

V. Alimov, International

Journal of Hydrogen Energy

36 (2011) 7737-7746

Potential solutions

• Inter-diffusion of Pd and substrate can be prevented by an intermediate layer that separates them

– Such composites are being actively investigated in the hydrogen energy research community

– These are typically ceramics with some porosity so as not to prevent tritium permeation

– Al2O3, Nb2C, HfN, YSZ, etc. mentioned in literature

• Alloys?

– V-Ni, Pd-Cu, V-Ti, V-Cu, others mentioned in literature

• Other coatings- Pd is necessary for separation from other gases, but we only need to prevent oxidation

Edlund, Journal of Membrane

Science 107 (1995) 147-153

Permeation Barriers • Such barriers are typically low-permeability metals (e.g. aluminum) or

ceramics such as Al2O3, Cr2O3, Er2O3

• These have achieved permeation reduction factors as high as 10,000 in the laboratory

• They have not performed as well in-pile- “Barriers that can provide a significant permeation reduction in the laboratory must be essentially defect-free1”

• Performance of permeation barriers in a radiation environment, on the necessary time scales, must be demonstrated

Levchuck et al JNM 328 (2004) 103 1R. Causey et al in Comprehensive Nuclear Materials, 2012

Hollenberg et al FED 28 (1995) 190-208

Natural oxide barriers

• Nickel alloys (Incoloy 800H, Inconel 617, Haynes 230) have been investigated for use in high temperature gas-cooled reactors

• These are adopted in the EU DEMO and TBM designs for ex-vessel helium systems

• A stable, protective Cr2O3 layer is expected1, even when hydrogen impurities exceed H2O impurities by a factor of 200 to 1

• This layer forms a very effective natural permeation barrier that makes these materials potentially attractive for fusion systems

• This has been demonstrated with MANET, but with EUROFER only a modest (~30x) PRF was demonstrated at high H2O concentrations

Aiello FED 84 (2009) 385-389

Serra JNM 240 (1997) 215-220

1Wright, INL/EXT-06-11494, 2006

Permeation Barriers Under Irradiation

• Possible reasons for reduced performance:

– Degradation of layer (area defect model1):

– Radiation-enhanced permeation2 (composite diffusion model1):

• Permeation, barrier, and and getter performance under irradiation extensively studied under TPBAR program to generate tritium in LWRs

• Despite considerable advances in understanding, phenomena still not completely understood

M

M

B

B xx

PAQ

Px

AQeff

Md

1R. Causey et al in Comprehensive Nuclear Materials, 2012 2W. Luscher et al., JNM 437 (2013). D. Senor, PNNL-22873 (2013).

Do we need permeation barriers?

• Maybe, and depends on the blanket type

• HCLL DEMO loss estimates with TES efficiency of 80% and PRFs of 5 on RAFM blanket structures and 100 on steam generator tubes:

• Note the influence of PbLi solubility; 1 g/y target not met even under optimistic assumptions

F. Franza et al., Fus. Sci. Tech 64 (2013) 631-635 A. Santucci et al., IEEE Trans. Plas. Sci. 42 (2014) 1053–1057

Insights from FNSF study

• In the FNSF study (which employs a DCLL blanket) a number of design aspects help mitigate tritium permeation losses:

– SiC flow channel inserts (FCI) act as permeation barriers within the blanket

– Relatively high PbLi flow rates characteristic of the DCLL result in less tritium generation per pass and lower tritium partial pressures between the blanket and tritium extraction system

– Efficient tritium extraction from PbLi in the form of a vacuum permeator

– Co-axial piping: tritium permeating out of the hot leg enters the cold leg and is returned to the blanket

• Total losses are influenced dramatically by parameter uncertainty in:

– PbLi solubility

– HX and ex-vessel piping permeability (Incoloy 800 oxidation state)

• Depending on assumptions losses are between ~0.03 g/yr (barriers NOT needed) and ~3 g/yr (barriers needed)

Summary and Conclusions

• DEMO must breed and process an unprecedented amount of tritium, and safety requires that losses be a very small fraction of this

– High temperatures make this difficult

• Tritium transport depends on a variety of physical phenomena including diffusion, recombination, adsorption, and trapping

– Measured properties are often highly uncertain, and so are calculations based on them

• Efficient tritium extraction is essential to limiting losses and inventories

– Concepts exist, experimental validation needed (ongoing)

• Permeation barriers in the form of coatings have not performed well under irradiation to date,

– Reasons for this are not completely understood

– They may not be necessary for some design concepts, but this is still not clear in light of parameter uncertainties

– Application more credible in more accessible and maintainable locations (e.g. on the outside of ex-vessel pipes)