---
title: Chlorine Defects in 4H-SiC
url: https://www.emergentmind.com/topics/chlorine-based-defects-in-4h-sic
type: topic
---

# Chlorine Defects in 4H-SiC

Chlorine-based defects in 4H-SiC are extrinsic point defects and defect complexes containing chlorine, now studied primarily as telecom-band color centers and spin-active centers in the 4H polytype of silicon carbide. Their recent literature differs sharply from the established deep-level literature of 4H-SiC: a 2025 review of electrically active defects detected by DLTS, Laplace-DLTS, and MCTS reports that chlorine and other halogens are not mentioned at all in the mainstream junction-spectroscopy treatment of 4H-SiC, whose dominant electrically active centers remain \(V_\mathrm{C}\), \(V_\mathrm{Si}\), \(C_i\), \(C_\mathrm{Si}\!-\!V_\mathrm{C}\), and boron-related defects [2502.14426]. Chlorine-based defects therefore emerged not from the classical DLTS taxonomy, but from first-principles screening, implantation-based photoluminescence, ODMR, and subsequent wavefunction-level modeling of telecom-emitting centers in Cl-implanted 4H-SiC [2304.14525].

## 1. Position within the 4H-SiC defect landscape

In the conventional electrically active defect literature, 4H-SiC is organized around intrinsic and common impurity-related centers such as \(Z_{1/2}\), EH1, EH3, EH4/5, EH6/7, S1/S2, shallow boron, and the D-center. In that framework, chlorine is absent: the 2025 review explicitly states that “chlorine (Cl) is not mentioned at all,” that no halogens are named anywhere, and that there is no discussion of Cl-related deep levels, halogen complexes, or the effects of chlorinated CVD chemistries such as SiCl\(_4\) or SiHCl\(_2\) on defect formation [2502.14426].

That absence is technically significant because DLTS and MCTS are chemically blind but electronically sensitive. The review notes DLTS sensitivity down to \(\sim 10^9\ \text{cm}^{-3}\), and it further states that halogen-related centers, if they produced deep levels in the bandgap at sufficient concentration, would be detectable as new peaks. A plausible implication is that chlorine-based defects were either below the DLTS detection threshold, electrically inactive in the surveyed materials, or simply outside the process space emphasized in the reviewed studies. The same review therefore functions as a baseline map: any chlorine-related center in 4H-SiC must be distinguished from the already well-mapped intrinsic and boron-related background [2502.14426].

## 2. Microscopic models and theoretical discovery

The defining modern theoretical construct is the chlorine-vacancy center, “ClV,” introduced through high-throughput first-principles screening of 52,600 extrinsic defects in 4H-SiC. In that framework, the most important chlorine-containing defect is a silicon vacancy plus substitutional chlorine on a neighboring carbon site, \(V_\mathrm{Si}\mathrm{Cl}_\mathrm{C}\), described as an NV-like center. The four nearest-neighbor configurations are \(hh\), \(kk\), \(hk\), and \(kh\); \(hh\) and \(kk\) are on-axis with approximate \(C_{3v}\) symmetry, while \(hk\) and \(kh\) are off-axis with \(C_{1h}\) symmetry. The same work reports that placing Cl on a Si site is disfavored by about \(4.45\ \text{eV}\), that the nearest dissociation \(V_\mathrm{Si}+\mathrm{Cl}_\mathrm{C}\) leaves a binding energy exceeding \(7\ \text{eV}\), and that the relevant quantum-defect charge state is \(\mathrm{ClV}^{+}\), whose ground state is a triplet \(S=1\) for all four configurations [2304.14525].

That theoretical study also gives explicit thermodynamic charge-transition levels, in eV above the valence-band maximum, for the four configurations:
\[
\begin{array}{c|cccc}
\text{Transition} & hh & kk & hk & kh \\
\hline
(2^+|+) & 1.07 & 1.05 & 1.19 & 1.14 \\
(+|0)   & 1.76 & 1.73 & 1.73 & 1.69 \\
(0|-)   & 2.00 & 1.87 & 1.92 & 1.76 \\
(-|2^-) & 3.01 & 2.96 & 2.99 & 3.01
\end{array}
\]
and uses the standard defect-formation expression
\[
E^{q,S}_\text{form}(E_F)=E^{q,S}_\text{defect}-E_\text{bulk}-\sum_i n_i\mu_i+qE_F+E_\text{corr}^q.
\]
Later benchmarking with SCAN and \(r^2\)SCAN retained the same NV-like motif and reported neutral ClV formation energies of \(6.82\ \text{eV}\) (PBE), \(7.13\ \text{eV}\) (SCAN), \(7.44\ \text{eV}\) (\(r^2\)SCAN), and \(7.82\ \text{eV}\) (HSE), while emphasizing that the triplet ground state and telecom-range optical transition survive across functionals [2501.07289].

The microscopic picture is not yet fully converged across the literature. Most DFT and implantation papers interpret the active center as \(\mathrm{Cl}_\mathrm{C}\!-\!V_\mathrm{Si}\) [2304.14525], and later experimental engineering work uses the same Cl\(_\mathrm{C}\)–\(V_\mathrm{Si}\) description [2510.25008]. By contrast, one multireference wavefunction study denotes \(\mathrm{ClV}^+\) as a substitutional chlorine atom on a silicon site, Cl\(_\mathrm{Si}\), adjacent to a silicon vacancy, \(V_\mathrm{Si}\), while still recovering an NV-like active manifold consisting of one \(a_1\) and one doubly degenerate \(e\) set with four electrons and a triplet ground state [2511.21965]. This discrepancy is itself part of the present subject.

## 3. Optical spectroscopy and telecom-band emission

Theoretical prediction placed the main \(\mathrm{ClV}^+\) zero-phonon lines between about \(1330\) and \(1590\ \text{nm}\), depending on configuration: \(0.93\ \text{eV}\) (\(\sim 1330\ \text{nm}\)) for \(hh\), \(0.86\ \text{eV}\) (\(\sim 1440\ \text{nm}\)) for \(kk\), \(0.83\ \text{eV}\) (\(\sim 1490\ \text{nm}\)) for \(hk\), and \(0.78\ \text{eV}\) (\(\sim 1590\ \text{nm}\)) for \(kh\), with HSE radiative lifetimes of \(28.4\), \(30.9\), \(24.3\), and \(28.1\ \text{ns}\), respectively [2304.14525]. The SCAN/\(r^2\)SCAN benchmarking later reported configuration-resolved lifetimes in the \(31.0\!-\!42.91\ \text{ns}\) range, again keeping ClV in the telecom-emitter class [2501.07289].

Experimental realization then established chlorine-related telecom emission in implanted and annealed 4H-SiC. One study reports four Cl-related configurations, Cl1–Cl4, with ZPLs at \(1539.2\), \(1533.3\), \(1396.7\), and \(1351.1\ \text{nm}\), respectively. Cl1 and Cl2 lie in the C-band, have linewidths of \(\sim 0.3\ \text{meV}\) at low temperature, and show local vibrational mode replicas at energy separations of \(11.5\) and \(12.4\ \text{meV}\); temperature-dependent measurements further reveal Cl1′ at \(1526.7\ \text{nm}\) and Cl2′ at \(1520.9\ \text{nm}\), each separated from the corresponding ground ZPL by \(\sim 6.7\ \text{meV}\). Polarization-resolved PL shows efficient excitation only for \(E_\text{exc}\perp c\), while emission is \(E_\text{PL}\perp c\) for Cl1, Cl2, and Cl4, but \(E_\text{PL}\parallel c\) for Cl3 [2602.08854].

A separate experimental study uses the nomenclature ClV1–ClV4 and assigns the observed ensemble lines to the four crystallographic configurations \(hh\), \(kk\), \(hk\), and \(kh\). In that assignment, ClV1 is at \(1350\ \text{nm}\), ClV2 at \(1472\ \text{nm}\), ClV3 at \(1532\) and \(1538\ \text{nm}\), and ClV4 at \(1556\) and \(1561\ \text{nm}\), thereby spanning the O-, S-, and C-bands. The off-axis configurations are reported as split doublets, with splittings of \(3.2\ \text{meV}\) for ClV3 and \(2.6\ \text{meV}\) for ClV4 [2510.25008].

The vibronic metrics are unusually favorable in experiment but not in early theory. For Cl1 and Cl2, one PL/ODMR study estimates a Debye-Waller factor of \(22\!-\!25\%\) [2602.08854]. A later quantum-memory study reports a Debye-Waller factor of \(39\%\) for ClV1 at \(1348.7\ \text{nm}\), with phonon sideband peaks at \(1387.9\) and \(1431.1\ \text{nm}\) corresponding to a vibrational energy of about \(25\ \text{meV}\), and an excited-state lifetime of \(450\pm10\ \text{ps}\) at \(10\ \text{K}\) [2605.03717]. Those values differ sharply from early ClV\(^+\) calculations, which predicted Debye-Waller factors of \(0.95\!-\!2.88\%\) or, in the wavefunction-based reformulation, \(\sim 1\!-\!3\%\) [2304.14525; 2511.21965]. This mismatch is central to current interpretation.

## 4. Spin activity, ODMR, and hyperfine structure

The first-generation theoretical picture is that \(\mathrm{ClV}^{+}\) is an \(S=1\) center with zero-field splitting in the microwave range. Early HSE results gave \(D=1.93\), \(1.87\), \(1.90\), and \(1.82\ \text{GHz}\) for the \(hh\), \(kk\), \(hk\), and \(kh\) configurations, with \(E=0\) for on-axis cases and \(E=209\) or \(150\ \text{MHz}\) for off-axis cases. The same work reported a largely isotropic hyperfine interaction on the chlorine nucleus of order \(10\!-\!20\ \text{MHz}\), and simulated an electron-spin coherence time \(T_2\approx1.45\ \text{ms}\) at \(100\ \text{G}\) in natural-abundance 4H-SiC [2304.14525].

Wavefunction-level analysis then recast the same defect family as an ODMR-active NV-like center with a triplet ground state \(^3\bar{A}_2\), a telecom-bright excited triplet near \(0.85\ \text{eV}\), low-lying singlets around \(0.18\!-\!0.20\ \text{eV}\), and a higher singlet near \(0.97\ \text{eV}\). In that treatment, the zero-field splitting for the \(kk\) configuration falls in the \(1.02\!-\!1.34\ \text{GHz}\) range depending on method, the triplet photoluminescence lifetime is about \(50\ \text{ns}\), the intersystem-crossing half-lives for \(m_S=\pm1\) are \(15\) and \(34\ \text{ns}\), and the \(m_S=0\) ISC rates are about \(1000\times\) slower. This produces the standard ODMR condition of bright and dark spin channels [2511.21965].

Experiment has not converged to a single spin-Hamiltonian picture. One ODMR study on chlorine-implanted 4H-SiC observes multiple resonances between \(\sim 1000\) and \(1400\ \text{MHz}\), including peaks at approximately \(1101\), \(1147\), \(1197\), and \(1255\ \text{MHz}\), with overall \(D_\text{eff}\sim1.0\!-\!1.4\ \text{GHz}\), ODMR contrast exceeding \(5\%\) under off-resonant \(1525\ \text{nm}\) excitation, and room-temperature spin activity. However, its magnetic-field dependence shows turning points around \(2\), \(8\), and \(13\ \text{G}\), and the authors state that this pattern is incompatible with a simple \(S=1\) center and instead analogous to \(S=3/2\) silicon-vacancy-like behavior [2602.08854].

A later quantum-memory study reports a different ODMR regime: sub-GHz resonances assigned to a Cl-related \(S=1\), \(I=3/2\) system with resolved \(^{35}\mathrm{Cl}\) hyperfine structure. For the main \(\nu_1\) family, the fitted parameters are \(D=560\ \text{MHz}\), \(E=60\ \text{MHz}\), and \(A=-34\ \text{MHz}\); for a second Cl-related family \(\nu_2\), \(D=455\ \text{MHz}\), \(E=60\ \text{MHz}\), and \(A=-30\ \text{MHz}\). Zero-field lines at \(509\), \(482\), \(640\), and \(122\ \text{MHz}\) are resolved, Ramsey interferometry yields beating frequencies of \(1.6\) and \(10.1\ \text{MHz}\), \(T_1=660\pm40\ \text{ns}\), and an intrinsic ensemble \(T_2^*\approx2\pm1\ \mu\text{s}\) after modeling out charge-state quenching [2605.03717]. The coexistence of GHz-scale, sub-GHz, \(S=1\), and \(S=3/2\)-like interpretations is therefore an active point of the field rather than a settled taxonomy.

## 5. Formation routes and defect engineering

The predicted formation routes include growth with Cl-containing precursors, implantation plus annealing, and vacancy creation followed by Cl incorporation. The original ClV theory explicitly notes that chlorine is routinely present in SiC CVD growth and cites DLTS on Cl-implanted p-type SiC showing Cl-related levels near \(E_V+0.97\ \text{eV}\), matching the calculated \((2^+|+)\) transition of ClV [2304.14525]. At the same time, the broad DLTS review does not connect chlorinated growth chemistry to any specific 4H-SiC deep level, so the formation problem remains split between optical/spin and classical electrical-defect literatures [2502.14426].

Implantation-based experiments provide the practical process window. One PL/ODMR study uses Cl\(^+\) implantation from \(36\ \text{keV}\) to \(2\ \text{MeV}\), producing a box profile with Cl concentration \(\sim 1\times10^{16}\ \text{cm}^{-3}\), at an implantation temperature of about \(600^\circ\text{C}\), followed by \(1600\) or \(1800^\circ\text{C}\) annealing for \(1\ \text{h}\) in Ar. The active implanted region is reported to be \(\sim 500\ \text{nm}\) beneath the surface, and the telecom PL lines above \(1300\ \text{nm}\) are observed only in Cl-implanted samples [2602.08854].

A separate engineering study systematically varies fluence and annealing conditions. It uses Cl\(^+\) implantation at \(40\) and \(150\ \text{keV}\), with fluence from \(1\times10^{11}\) to \(5\times10^{14}\ \text{cm}^{-2}\), followed by \(2\ \text{h}\) vacuum anneals at \(500^\circ\text{C}\), \(1100^\circ\text{C}\), or \(1300^\circ\text{C}\). No ClV1 emission is observed after \(500^\circ\text{C}\) annealing; the ClV1 ZPL appears after \(1100^\circ\text{C}\), and its intensity approximately doubles at \(1300^\circ\text{C}\). The best ZPL-to-background ratio for ClV1 occurs at \(\Phi=1\times10^{12}\ \text{cm}^{-2}\). Ar-implanted controls, despite creating nearly the same \(V_\mathrm{Si}\) profile according to SRIM, do not show the ClV1 line, which demonstrates that the observed telecom centers are extrinsic and chlorine-derived [2510.25008].

Temperature and power dependence further constrain device-relevant behavior. Under \(976\ \text{nm}\) excitation, the ClV1 ZPL and background intensities scale linearly with power up to \(35\ \text{mW}\), with no observed saturation, and the ZPL shows negligible reduction up to about \(30\ \text{K}\). Its thermal quenching is fitted by
\[
I_{\mathrm{ZPL}}(T)=\frac{I_0}{1+B\,e^{-E_a/(k_B T)}},
\]
with \(E_a=25\pm11\ \text{meV}\) [2510.25008].

## 6. Unresolved identification and present significance

The central unresolved issue is that “chlorine-based defects in 4H-SiC” now denotes a coherent experimental and theoretical family, but not a fully settled microscopic object. Early high-throughput and hybrid-DFT work identifies the relevant center as \(\mathrm{Cl}_\mathrm{C}\!-\!V_\mathrm{Si}\), positively charged, triplet, and NV-like [2304.14525]. Later optical engineering papers use the same assignment and map four telecom configurations to \(hh\), \(kk\), \(hk\), and \(kh\) [2510.25008]. Yet one wavefunction study adopts a Cl\(_\mathrm{Si}\)-adjacent-to-\(V_\mathrm{Si}\) model [2511.21965], one PL/ODMR paper concludes that the observed defect is likely not the originally predicted \(\mathrm{ClV}^{+}\) charge state because its Debye-Waller factors and ODMR behavior differ sharply from prediction [2602.08854], and a later quantum-memory study states explicitly that microscopic identification remains tentative and that other Cl-based complexes are not ruled out [2605.03717].

These discrepancies extend to orientation assignment. The optical-engineering literature associates ClV1 with an on-axis \(hh\) configuration at about \(1350\ \text{nm}\) [2510.25008], whereas the quantum-memory study reports a discrepancy between PL-based assignment and ODMR-based assignment for the same ZPL, with ODMR favoring an off-axis interpretation for the \(\nu_1\) family [2605.03717]. Likewise, the spin picture ranges from \(S=1\) with \(^{35}\mathrm{Cl}\) hyperfine structure and state mixing [2605.03717] to an \(S=3/2\)-like interpretation based on magnetic-field turning points [2602.08854].

Even with those open questions, the present significance of chlorine-based defects in 4H-SiC is clear. They constitute a new optical and spin-defect family with ZPLs in the O-, S-, and C-bands, room-temperature spin activity in multiple reports, and process routes compatible with ion implantation and high-temperature annealing in 4H-SiC [2510.25008; 2602.08854; 2605.03717]. A plausible implication, when viewed against the DLTS review, is that chlorine-based defects are not part of the classical set of dominant bulk lifetime-killing centers in 4H-SiC, but instead define a newer class of telecom-emitting, optically addressable defects whose electronic role is being established primarily by PL, ODMR, and advanced electronic-structure theory rather than by mainstream junction spectroscopy [2502.14426].

Source: https://www.emergentmind.com/topics/chlorine-based-defects-in-4h-sic