Tritium Fuel Cycle: Challenges and Advances for Fusion Power Plants

Remi Delaporte-Mathurin

Half-life: 12 years

\mathrm{T} \rightarrow \mathrm{He} + \mathrm{e}^-

☢

\mathrm{n} + ^6\mathrm{Li} \rightarrow \mathrm{T} + \mathrm{He} + 4.8 \ \mathrm{MeV}
\mathrm{n} + ^7\mathrm{Li} \rightarrow \mathrm{T} + \mathrm{He} + \mathrm{n} - 2.5 \ \mathrm{MeV}
\mathrm{TBR} = \mathrm{\frac{tritium \ produced}{tritium \ consumed}} > 1

❓How ❓

Lithium is used to breed tritium

\mathrm{D} + \mathrm{T} \rightarrow \mathrm{He} + \mathrm{n}
\mathrm{n} + ^6\mathrm{Li} \rightarrow \mathrm{T} + \mathrm{He} + 4.8 \ \mathrm{MeV}
\mathrm{n} + ^7\mathrm{Li} \rightarrow \mathrm{T} + \mathrm{He} + \mathrm{n} - 2.5 \ \mathrm{MeV}
\mathrm{D} + \mathrm{T} \rightarrow \mathrm{He} + \mathrm{n}

Magnet

Breeding blanket

\mathrm{n} + \mathrm{Li} \rightarrow \mathrm{T} + \mathrm{He}

The breeding blanket

Plasma

The breeding blanket is only one component of the fuel cycle

Samuele Meschini et al 2023 Nucl. Fusion 63 126005

The fuel cycle can be simulated with the residence time model

\frac{{dI_i}}{{dt}} = \sum_{{j \neq i}} \left( f_{j \to i} \frac{{I_j}}{{\tau_j}} \right) - \left(1 + \epsilon_i\right) \left( \frac{{I_i}}{{\tau_i}} \right) - \lambda I_i + S_i

Component \(1\)

\(\tau_1\), \(I_1\), \(S_1\)

\( f_{1 \rightarrow 2} \frac{{I_1}}{{\tau_1}}\)

\( f_{1 \rightarrow 3} \frac{{I_1}}{{\tau_1}}\)

\( f_{0 \rightarrow 1} \frac{{I_0}}{{\tau_1}}\)

All inputs

Outputs + losses

Radioactive decay

Source term

M. Abdou et al. Fusion Engineering and Design 100 (2015) 2–43

Samuele Meschini et al 2023 Nucl. Fusion 63 126005

Let's play with a simple fuel cycle model

Burns tritium

Breeds tritium (TBR)

Tritium Extraction System

Breeding Blanket

Plasma

Storage

neutrons

TBR

(constrained by technology)

Doubling time

(driven by economics)

Startup inventory

(constrained by safety)

Source: remdelaportemathurin.github.io/interactive-plots/fuel_cycle/

Startup inventory

Tritium Extraction System

Breeding Blanket

Plasma

Storage

Samuele Meschini et al 2023 Nucl. Fusion 63 126005

ARC will require ~1 kg of start-up tritium inventory

Doubling time ~1-2 year(s)

The most important parameter for TBR is the Tritium Burn Efficiency (TBE)

Samuele Meschini et al 2023 Nucl. Fusion 63 126005

TBR Sensitivity index

Most important parameters

Samuele Meschini et al 2023 Nucl. Fusion 63 126005

Required TBR

❓How hard is it to achieve TBR ~ 1.1 ❓

The most important parameter for TBR is the Tritium Burn Efficiency (TBE)

We are here
TBR ~ 1.1

Plasma improvements

What are the different blanket concepts?

Water Cooled Lithium Lead

Arena et al Energies 2023, 16(4), 2069

Tritium Breeding Ratio

Water Cooled Lithium Lead

Arena et al Energies 2023, 16(4), 2069

Tritium Breeding Ratio

James Dark et al 2021 Nucl. Fusion 61 116076

They just made it worse!

⚠️Pipes generate a build up of tritium

Tritium transport FESTIM simulation (2021)

Dual Cooled Lithium Lead

Palermo et al Fusion Engineering and Design 109–111 (2016) 13–19

Tritium Breeding Ratio

Helium Cooled Pebble Bed

Zhou et al Appl. Sci. 2023, 13(4), 2081

Tritium Breeding Ratio

ARC Liquid Immersion Blanket

Tritium Breeding Ratio

 Jin Whan Bae et al 2022 Nucl. Fusion 62 066016

Sorbom et al, Fus Eng Design, Volume 100, November 2015, Pages 378-405

⚛️Breed tritium

🛡️Shield from neutrons

🔥Extract heat

Liquid Immersion Blanket

FLiBe

ARC power plant

Sorbom et al, Fus Eng Design, Volume 100, November 2015, Pages 378-405

Tritium Breeding Ratio

How much do we trust these numerical results?

design target +9.5 %

DEMO requirement 1.05

Fischer et al Fusion Engineering and Design 2020 10.1016/j.fusengdes.2020.111553

Problem: current reactors don't breed T

Breeding Blanket design

Fusion Power Plant

will test...

is needed for...

The EU-DEMO plan

Step 1. Build ITER [pause for dramatic effect]

Step 2. Plug some mockups in the Tritium Breeding System (TBS)

Step 3. Pick the best one and build a FULL SCALE prototype for DEMO

Step 4. Hope scales well 🤞

Plan B

There's no plan B.

ITER

TBS

DEMO

The LIBRA plan:

let's test it before ARC

What is the smallest breeding blanket that can demonstrate a TBR of 1?

🎯Demonstrate T self-sufficiency with DT neutrons
 

🎯Validating neutronics and tritium transport models
 

🎯Investigate tritium extraction from liquid breeders

LIBRA experiment

500L FLiBe

14 MeV neutron source

Inconel

double wall

Li + n → T + He

Neutron multiplier

The LIBRA experiment

Tritium transport

Transport mechanisms:

  • Diffusion
  • Advection

Release pathways:

  • Release gas/liquid interface
  • Permeation through walls

The LIBRA experiment

He

Tritium detection

The LIBRA experiment

A staged approach to \(\mathrm{TBR} \approx 1\) ...

Tritium Breeding Ratio

LIBRA full-scale

The BABY programme studies tritium breeding at a small scale

  1. \(14 \ \mathrm{MeV}\) neutrons generated
     
  2. tritium created from nuclear reactions
     
  3. tritium transport in the salt
     
  4. tritium released into the gas phase
     
  5. tritium collection and accountancy

Molten salt @ 700C

neutron generator

Tritium collection

reentrant heater

How to measure TBR?

\mathrm{TBR} = \mathrm{\frac{T \ produced}{T \ consumed}}
= \mathrm{\frac{T \ produced}{n \ produced}}

We need to measure these two numbers!

The BABY 100 mL experiment

The BABY experiment was modelled in OpenMC

ClLiF salt gives the highest TBR at the 100 mL scale

Neutrons are detected with a combination of techniques

Niobium activation foils

Diamond detector

\( (n,\alpha)\) peak provides real-time data

Bjornstadt et al The Current Status and Future Trends on Radioisotope Application in Industry (2014) 10.13140/RG.2.2.10990.00322

Neutrons are detected with a combination of techniques

A8 Diamond proton recoil telescope

cividec.at/neutron_diagnostics-A8.html

supported by Cividec

DD peak

(2 MeV)

DT peak

(14 MeV)

Neutron detection

Tritium was measured using

Liquid Scintillation Counting

HTO, TF and TCl

(soluble in water)

HT, T2

(insoluble)

Liquid Scintillation Counting

Counts between 0-18.6 keV for tritium detection

The TBR measurement somewhat  agreed with neutronics simulations 

Hypothesis: tritium was lost to permeation

\mathrm{TBR} = \frac{1.17\times10^{10} \ \mathrm{T}}{3.35\times10^{13}\ \mathrm{n} }
= {3.57\times10^{-4}}

Modelled TBR (OpenMC)

Measured TBR

V \frac{d c_\mathrm{salt}}{dt} = S - \textcolor{#438343}{Q_\mathrm{wall}} - \textcolor{#2a7eb8}{Q_\mathrm{top}}
Q_i = A_i \ k_i \ (c_\mathrm{salt} - c_\mathrm{external}) \\ \approx A_i \ k_i \ c_\mathrm{salt}
S = \mathrm{TBR} \cdot \Gamma_n

A transient 0D model is used to simulate the tritium release

\(k\) mass transport coefficient

neutron rate

100 mL

salt

Top release

Wall release

  • \( \mathrm{TBR} = 4.71 \times 10^{-4} \) (from OpenMC)
  • \( \Gamma_n = 3.88 \times 10^8 \) n/s (from measurement)
  • Mass transport coeffs. \( k_i \) (fitted)

100 mL

salt

Top release

Wall release

A transient 0D model is used to simulate the tritium release

Cumulative release (Bq)

Neutron generators on

Reproducibility

Varying the mass transfer coefficients only affects the dynamics

Redox potential affects the release dynamics

Cumulative release (Bq)

Redox potential affects the tritium speciation

Total tritium release

Upgrade: BABY 1 L

1 L of salt

Top release gas sweep

Outer-vessel for permeated tritium

New crucible

Only one neutron generator below the crucible

One bubbler per gas line

×6

New world record! 🥇

The previous TBR record has been broken again!

Tritium Breeding Ratio

Now we need ×500...

The BABY 1L results also agree with the model

But:

  • Is the 0D assumption valid?
    • Tritium production not homogeneous
    • Temperature not homogeneous
       
  • Why is there no permeation?
     
  • Is this reproducible?

BABY 1L run 2 (as of 24 Feb 2025)

Where will we go from here?

LIBRA will help validate FESTIM models

Delaporte-Mathurin et al, International Journal of Hydrogen Energy 63, 2024, 786-802

Temperature

(steady state)

Velocity

(steady state)

Tritium concentration

github.com/festim-dev/FESTIM

The FESTIM model highlights a qualitative discrepancy 

  • Very sensitive to diffusivity
     
  • Wall release > Top release
    Different than measured

Hypotheses

  • Advection not correctly taken into account?
  • Permeation barrier: oxide layer?
  • Complex chemistry?

Top release

Wall release

Diffusivities of FLiBe and FLiNaK

OpenMC model

Is our neutronics model an accurate representation?

4 lead bricks → +11 % TBR

?

BABY will inform the design and operation of LIBRA-Pi (100 L)

Take aways

  • Tritium self sufficiency is the next big risk item on the fusion roadmap
    • \( Q_\mathrm{plasma} = 40+ \) is useless without fuel for the reaction
    • Safety factor is 9.5 %, but the real uncertainty may be larger...
       
  • Up to now, R&D was mostly (only?) focused on computational studies
     
  • Some initiatives give us hope

Will it be soon enough to build a successful FOAK fusion power plant by 2030-2040?

Thank you!

Any question?

✉️   remidm@mit.edu