Internal Charge Amplification in Germanium at 77K and 4K: From Single-Free-Flight Bounds to a Physics-Informed Ionization Model

Abstract

Internal charge amplification (ICA) in cryogenic high-purity germanium (HPGe) can lower detection thresholds by providing gain inside the detector crystal, but reliable operation requires a predictive estimate of the avalanche-onset \emph{critical electric field} \(E_{\mathrm{crit}}\). We present a compact framework for \(E_{\mathrm{crit}}\) at 77~K and 4~K (typical HPGe operating temperatures) that bridges (i) a mobility-based single-free-flight (SFF) upper bound with (ii) a physics-informed impact-ionization model incorporating energy-dependent scattering, nonparabolic (Kane) dispersion, intervalley transfer, and the high-energy ``lucky-drift'' tail. This unified treatment yields closed-form, design-useful relations, including \(E_{\mathrm{crit}}^{(\mathrm{PI})}=B(T)/\ln[A(T)d]\), and a practical calibration workflow that maps measured low-field mobility \(μ(T)\) and gain curves \(M(V)\) (Chynoweth analysis) to device-level bias targets with propagated uncertainty bands. Example electron and hole estimates indicate that realistic transport typically lowers \(E_{\mathrm{crit}}\) relative to SFF and increases the predicted change in \(E_{\mathrm{crit}}\) between 77~K and 4~K. The resulting portable formulas connect materials/transport inputs to geometry, excess noise, and field shaping, providing design-ready guidance for stable, unipolar-favored ICA with controlled quenching in Ge and other cryogenic semiconductors.

Figures (6)

• Figure 1: Critical field $E_{\mathrm{crit}}^{(\mathrm{SFF})}$ versus mobility $\mu$ on log--log axes, comparing two thresholds $\varepsilon_i=2.96~\mathrm{eV}$ (pair-creation energy) and $\varepsilon_i=0.73~\mathrm{eV}$ (band gap), for electrons $(m^\ast=0.12\,m_0)$ and holes $(m^\ast=0.28\,m_0)$. The inverse scaling $E_{\mathrm{crit}}\propto \mu^{-1}$ and $\sqrt{\varepsilon_i}$ dependence are evident.
• Figure 2: Chynoweth plot (conceptual). Illustration of $\ln\alpha$ vs $1/E$ for two temperatures using $\alpha(E)=A\,e^{-B/E}$ with $B(4\mathrm{K})<B(77\mathrm{K})$. The slope gives $B(T)$ and the intercept $\ln A(T)$, enabling direct calibration from $M(V)$ data on uniform-field diodes Chynoweth1958OkutoCrowell1974.
• Figure 3: Conceptual cross-section of a high-purity Ge internal-charge-amplification (ICA) detector. A small p$^+$ point contact (2 mm diameter) at the bottom defines a low-capacitance anode (target $C\!\approx\!0.6$ pF), while an outer n$^+$ contact forms two bulk P--N junctions. A narrow, high-field avalanche region ($\sim 5~\mu\mathrm{m}$) is engineered beneath the point contact to provide internal charge gain. Grooves ($\sim 3~\mathrm{mm}$) around the contact mitigate surface leakage and help shape the field. Illustrative operating parameters: leakage current $<10~\mathrm{pA}$, depletion voltage $\approx +3.2~\mathrm{kV}$, operating bias $\approx +5.0~\mathrm{kV}$. The detector (footprint $\sim 8~\mathrm{cm}\times 5~\mathrm{cm}$, central slot depth $\sim 2.5~\mathrm{cm}$, top-contact separation $\sim 1~\mathrm{cm}$) is housed in electrolytic copper (E-Cu) with indium foil for thermal/mechanical coupling and surrounded by ultra-low-background (ULB) Pb shielding. A cold preamplifier is mounted close to the point contact to minimize stray capacitance and series noise. This geometry concentrates the field under the anode for stable ICA while maintaining low noise and excellent energy resolution.
• Figure 4: Shown is the electric-field distribution calculated using Julia bezanson2017julia. The highest field occurs in the region around the point contact, reaching values above 8000 V/cm.
• Figure 5: Predicted amplification $M(V)$ for the ICA concept (Fig. \ref{['fig:ica_schematic']}) at 77 K and 4 K. The top panel shows the predicted amplification at 77 K and the bottom panel shows the predicted amplification at 4 K. We use an illustrative impact-ionization parameterization $\alpha(E,T)=A(T)\exp[-B(T)/E]$ with $A_{77}=3\times10^{5}~\mathrm{cm^{-1}}$, $B_{77}=4\times10^{4}~\mathrm{V/cm}$ and $A_{4\mathrm{K}}=4\times10^{5}~\mathrm{cm^{-1}}$, $B_{4\mathrm{K}}=2.8\times10^{3}~\mathrm{V/cm}$, and a multiplication width $d=5~\mu\mathrm{m}$. The device field is mapped linearly from bias. At 77 K, we assume $E(V_{\mathrm{dep}})=0$ at $V_{\mathrm{dep}}=3.2~\mathrm{kV}$ and $E(V_{\mathrm{op}})=8\times10^{3}~\mathrm{V/cm}$ at $V_{\mathrm{op}}=5.1~\mathrm{kV}$. At 4 K, no depletion voltage is assumed; instead we map the field linearly from $V_{\mathrm{start}}=150~\mathrm{V}$ with $E(V_{\mathrm{start}})=0$ to $E(V_{\mathrm{ref}})=5.3\times10^{2}~\mathrm{V/cm}$ at $V_{\mathrm{ref}}=400~\mathrm{V}$. The smaller $B$ at 4 K (longer inelastic mean free paths) yields larger $M$ at the same bias. In practice, TCAD and measured diode $M(V)$ data will be used to refine $E(V)$ and calibrate $\{A(T),B(T)\}$.