KAHVELab Lithium Project

The injector consists of a microwave-discharge ion source, a permanent-magnet LEBT, a diagnostics box and an 800 MHz RFQ. All elements were designed and manufactured in Türkiye.

The output is a 2 MeV proton beam at 1 µA peak current and low duty factor. Within TAGEP 1881, this beam is used to drive a compact neutron source, and the resulting neutrons are used to breed ³H in a second lithium target.

The two-stage reaction chain

The first reaction is endothermic and requires 1.88 MeV. A 2 MeV beam therefore operates 120 keV above threshold, where the emitted neutrons are soft and forward-peaked.

The second reaction is exothermic, has no threshold and has a positive Q-value of about 4.78 MeV, meaning it occurs readily with thermal and slow neutrons rather than fast ones.

• Angular escape. Neutrons are emitted over a wide cone, whereas a flat plate subtends only the forward direction. A substantial fraction never reached the copper or the paraffin.
• Lateral escape. Neutrons that did enter the paraffin drifted sideways and left through the side boundary before thermalisation was complete.

With hemispherical shells every emission angle is intercepted and lateral leakage is removed by construction.

The beam travels along the symmetry axis, entering from the left through the evacuated channel. The copper is finned, and serves as both heat removal and beam stop.

Status of the work packages

⁷Li thickness and lifetime

The thickness was scanned over 5–100 µm at 10⁷ protons per point and assessed against the 1.88 MeV threshold. 20 µm was selected and confirmed at 10⁸ protons. The 1 mm copper backing extinguishes the primary proton flux completely.

Burn-up from the RESNUCLE card gives 6.77×10⁻⁵ nuclei per proton against 6.97×10^{19 7}Li atoms in the target. With the 2% duty factor, half-consumption occurs at 4.12×10¹² s — about 1.3×10⁵ years. Burn-up is therefore not a design constraint; target life will be limited by surface oxidation and thermal cycling.

Neutron energy spectrum

The flux was scored with USRBDX at 10⁹ primaries and a Landau function was fitted in Python to extract the most probable value.

Same material, same nominal thickness, ~9 keV apart. No single cause has been confirmed; the discrepancy is reported openly and remains under investigation.

⁷Be production

The ⁷Li(p,n)⁷Be reaction produces neutrons and simultaneously leaves beryllium-7 in the target. ⁷Be has a 53-day half-life, emits a 477 keV gamma, and beryllium is chemically toxic.

Scored with RESNUCLEI: 2.30×10⁻⁵ per primary proton, which is 1.44×10⁸ per second at 1 µA and 2.87×10⁶ per second at the actual 2% duty factor. The activity is modest, but it is confined to the 20 µm layer that will eventually be removed and replaced, so it governs the handling procedure rather than the shielding design.

The flux entering the moderator is 2.714×10⁻⁵ neutrons per primary. The thickness was scanned at 10⁸ primaries per point, recording both the total flux leaving the paraffin and the thermal component (< 0.025 eV).

The thermal fraction rises monotonically with thickness, whereas the thermal flux peaks at 5 cm and then falls as absorption takes over. Production scales with flux, so 5 cm was selected.

A ⁶Li shell encloses the paraffin so that thermalised neutrons meet it radially. Capture was scored with USRBDX and the product nuclei with RESNUCLEI, at 10⁸ primaries per point.

Going from 0.01 to 50 cm raises the capture efficiency from 28% to 99.7%, a factor of 3.6, at a cost of 360 litres and 166 kg of enriched ⁶Li, which is expensive and strategically controlled. 100 µm was selected: at 1 µA with the 2% duty factor, corresponding to 1.25×10¹¹ effective protons per second, the yield is 2.384×10^{5 3}H nuclei per second.

Dose equivalent is the absorbed energy per unit mass weighted by the biological damage each radiation type causes, in sieverts; neutrons carry the higher weighting factor. It was scored with the FLUKA DOSE-EQ card at 10⁸ primaries, neutrons and photons separately, with raw output in pSv per primary converted to Sv/h. The 2.5 µSv/h limit for occupied areas follows ICRP as adopted in NDK regulation.

At 1 mA the dose falls to 2.35×10⁻⁷ at 60 cm, so shielding becomes mandatory. The shield has three layers: 1 mm Al stops low-energy secondary charged particles and beta radiation at source; 15 cm borated paraffin, where the hydrogen moderates the fast neutrons and the boron then absorbs them; and 40 cm concrete for the primary gammas and the capture gammas produced by the boron.

Thermal analysis: from FLUKA to ANSYS

FLUKA was used to observe the deposited power, but it cannot model conduction, fluid flow or chiller systems, and pure lithium melts at about 180 °C in under two seconds. Resolving the full 300 s was computationally impractical, so an equivalent 1 Hz cycle was used at the same mean power and duty factor. Longer pulses with higher per-pulse energy provide a safe upper limit: if the 1 Hz model is safe, our 200 Hz machine is safe.

The layout was fitted into a strict 4 m distance from the RFQ exit to the laboratory garden door, modelled in SOLIDWORKS, in the order RFQ → steerer → quadrupole triplet → steerer → dipole → target assembly. The dipole bend angle is fixed at 10° by the physical room dimensions. Optics were computed with DemirciPRO, our in-house code, and cross-checked with TRAVEL: the beam reaches the target at 0.3 mm RMS with zero particle loss.

RFQ work started at TAEK, now TENMAK NÜKEN, at 352 MHz. KAHVELab designed and built an 800 MHz cavity in two modules — the world's highest-frequency proton RFQ. The higher frequency enables a smaller cavity, delivering 2 MeV under 1 m for full transportability. It runs at 1 mA peak current, ~1% duty factor and ~30% injected-proton capture, and also supports portable PIXE for materials analysis, archaeology and museum applications.

The quadrupole triplet, the steerers and the dipole have all been delivered. On the bench at 13.3 A the triplet measured 1727 G against 1727.3 G predicted, and the PT100 thermal test reached ~55 °C, matching the steady-state model. The steerers give 4.4 mrad deflection at 3 A, homogeneity ΔB/B < 0.2% across ±1 cm, at 12.6 W and ΔT ≈ 1.4 K, so no cooling is needed.

For the scintillator screen, three criteria were applied before any simulation: high light yield, slow degradation of the light output under irradiation, and no afterglow. CsI(Tl), NaI(Tl) and P46 (YAG:Ce) were scanned in Geant4 from 10 to 50 µm. CsI(Tl) and NaI(Tl) retain their yield per MeV as the film is thinned, whereas P46 loses roughly half of it. Based on the analysis outputs we selected P46 (10 µm), P43 (50 µm) and NaI(Tl) (10 µm), with an aluminium backing to distribute the generated heat.

The Faraday cup was modelled in CST as three layers: a copper collector 5.0 cm deep and 5.5 cm wide, a 0.3 cm PTFE insulator, and a 0.5 cm copper guard ring held at −200 V. Leakage is zero under all three emission models tested. The design and analysis processes for the Faraday cup are not yet complete, but work is continuing.

Completed this period

• Target geometry redesigned to nested hemispherical shells.
• ⁷Li thickness fixed at 20 µm, paraffin at 5 cm, ⁶Li at 100 µm, giving 2.384×10^{5 3}H nuclei per second at 1 µA and 2% duty.
• Radiation shield designed; the field outside is 10 000× below the legal limit.
• Thermal analysis completed in ANSYS; quadrupole triplet, dipole and both steerers delivered.

Planned for the next period

• Delivery and testing of the RFQ; initiation of the measurement box and remote detector control.
• WP6: construction of the shielding structure and acquisition of the NDK licence or exemption.
• WP8: target manufacture and commissioning of the gas collection system.
• Scintillator analysis is complete; currently working on the arrival of the scintillator screens.