WorksPhysics and Astronomy
Substitutional platinum as an efficient nonradiative recombination center in silicon
Abstract
Platinum (Pt) is widely used for carrier-lifetime control in silicon power devices, yet the microscopic nonradiative recombination mechanism of the substitutional platinum () dopant remains debated. Using first-principles calculations combined with nonradiative multiphonon theory, we systematically investigate the electronic structures and carrier capture dynamics of . Our results show that both the donor () and acceptor () levels of exhibit large capture cross sections for electron and hole carriers, thereby making an effective recombination center. Notably, the calculated capture cross sections are sensitive to the symmetry-equivalent defect configurations with different Jahn-Teller distortions. By accounting for two different configurations of neutral during transitions properly, our calculated carrier capture cross sections align well with experimental values. This work provides a microscopic picture of the carrier capture processes induced by and emphasizes the importance of symmetry-equivalent configurations in defect-assisted nonradiative recombination.
Nonradiative recombination at deep-level defects strongly influences the performance of semiconductor devices[1, 2, 3]. In photovoltaic cells, defects that act as strong recombination centers limit carrier lifetimes and reduce the power conversion efficiency. In contrast, for power devices like insulated gate bipolar transistors (IGBTs), the precise introduction of recombination centers is essential to realize high switching speed and low turn-off loss. Therefore, a quantitative understanding of the defect-assisted nonradiative recombination is important for the optimization of device performance.
Platinum (Pt) impurities in silicon (Si) have long attracted considerable attention due to their application as carrier lifetime controllers. Compared to alternatives like gold, Pt offers superior high-temperature stability, better turn-on performance, and lower leakage currents[4, 5], which are fundamentally governed by the deep levels Pt introduces into the band gap. Specifically, donor levels within 0.26–0.36 eV above the valence band maximum (VBM) and acceptor levels within 0.19–0.28 eV below the conduction band minimum (CBM) have been widely observed in Pt-doped Si[6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17]. Although some studies suggested that the two levels may originate from distinct defects[12, 18], these levels are widely attributed to substitutional Pt ()[9, 11, 19]. This assignment is further supported by the identification of a double donor level near the VBM, which is also assigned to the same platinum center[16, 19].
While early studies[10, 11, 20] proposed that the donor and acceptor levels of serve as the dominant lifetime-controlling states in p-type and n-type Si, respectively, this view has been challenged by subsequent studies. Lisiak et al.[9, 21] argued that both levels of would not be effective in recombination as they are far from the middle of the band gap. They proposed instead that a level at eV, possibly associated with a Pt-related complex, is more favorable to dominate the carrier lifetime. This interpretation was supported by Kumar et al.[22], who observed a compensation effect induced by Pt doping and argued that an effective acceptor-type recombination center must be located near or below the intrinsic Fermi level of Si. Therefore, the acceptor level, lying close to the CBM, was considered unlikely to be the dominant recombination level. Furthermore, Ewaraye et al.[23] proposed that another level at eV in Pt-doped Si measured by deep-level transient spectroscopy (DLTS) is a good recombination center while the acceptor level is not. These differing experimental interpretations therefore motivate a further examination of whether can act as an effective recombination center.
Based on the Shockley-Read-Hall (SRH) theory[24], assessing whether a defect acts as an effective recombination center requires determination of both majority and minority carrier capture cross sections. However, experimental data for , primarily from DLTS, provide only limited information on minority carrier capture and exhibit discrepancies of 2–3 orders of magnitude across different reports[1, 8, 9, 16, 17, 23, 25], hindering the definitive identification of recombination centers. Therefore, systematic theoretical evaluations of carrier capture cross sections are essential for clarifying the role of , yet such a study is still lacking.
In this study, we employ first-principles calculations to systematically investigate the electronic structure and nonradiative recombination mechanisms of in Si. By calculating carrier capture cross sections based on the nonradiative multiphonon theory, we show that both donor () and acceptor () levels can facilitate rapid electron-hole recombination. The calculated capture process is sensitive to the choice of symmetry-equivalent defect configurations. By considering the configurations of the neutral defect during transitions properly, we obtain capture cross sections in good agreement with experimental results. These results provide a microscopic picture to understand the nonradiative carrier capture processes induced by and underscore the importance of symmetry-equivalent configurations when investigating defect-assisted recombination.
First-principles calculations are performed within density functional theory (DFT) using the Heyd-Scuseria-Ernzerhof (HSE) hybrid functional[26] as implemented in the Vienna Ab Initio Simulation Package (VASP)[27]. The projected augmented-wave (PAW) pseudopotentials[28] are used with a plane-wave cutoff energy of 370 eV, and a single -centered -point mesh is used to sample the Brillouin zone. Spin polarization is included in all cases. The fraction of Hartree–Fock exchange is set to 0.25. The calculated band gap is 1.20 eV at the point, in good agreement with the experimental value of 1.17 eV[29]. During structural relaxation, all atoms in the supercell are fully relaxed until the force on each atom falls below 0.01 eV/Å. Although Pt is a heavy element, test calculations show that spin-orbit coupling (SOC) has negligible effects on the electronic structures. Therefore, SOC is neglected hereafter (see the Supporting Information, SI).
The formation energy for a defect in charge state is determined as follows[2]:
where and are the total energies of the supercell with and without the defect, respectively. denotes the chemical potential of element , referenced to the energy of the stable elemental phase, and is the number of atoms removed from () or added to () the supercell to form the defect. is the Fermi level referenced to the VBM of the host supercell. Finally, is the finite-size correction for charged defects. We employ the correction scheme proposed by Freysoldt, Neugebauer, and Van de Walle (FNV)[30], using a static dielectric constant of 11.9. Formation energies and transition levels are calculated within 512-atom supercells. Considering the computational cost and the fact that nonradiative capture is mainly driven by local lattice distortion and electron–phonon coupling in the vicinity of the defect, carrier capture cross sections are calculated in a 216-atom supercell (size-convergence verified in the SI). We use the nonradiative multiphonon theory to calculate the capture cross section following the methodology described in Ref. 31, which explicitly considers the change of phonon modes due to lattice relaxation.
We first examine the equilibrium atomic structures of the defects in different charge states. In a perfect Si crystal, a substitutional atom preserves symmetry. Figure 1 shows that this symmetry is maintained only for the double-positive state (), where the Pt atom remains at the lattice site center and the four Si–Pt bond lengths increase from 2.35 Å in bulk Si to 2.37 Å. For the other three charge states, we observe significant structural distortions driven by the Jahn-Teller effect. The neutral state () exhibits symmetry, whereas the ground-state structures of single-positive () and single-negative () charge states both adopt symmetry.

Figure 1. Equilibrium atomic structures of the platinum substitutional defect in the (a) double-positive, (b) single-positive, (c) neutral, and (d) single-negative charge states. Platinum atom is shown in orange and silicon atoms are in blue, with each unique Si-Pt bond length labeled. The corresponding electron energy level diagrams at the point with character symmetry labels are shown in the bottom panel.
The observed structural distortions can be understood within the vacancy model proposed by Watkins[32]. The defect levels in the band gap originate mainly from vacancy-like orbitals of the neighboring Si atoms, while the Pt states are located deep in the valence band and remain fully occupied. The partially occupied levels in the gap are susceptible to Jahn–Teller distortions, driving the symmetry lowering observed for the +, 0, and − charge states.
In the double-positive case, the level is empty, preserving the symmetry and pushing the nearest-neighbor Si atoms slightly outward. In the single-positive state, the single occupied level is unstable, which in turn induces spontaneous Jahn-Teller distortion to symmetry and yields a spin . This distortion results in the elongation of one pair of opposite Si–Si distances and shortening of the other pair.
In the case of neutral state , two electrons occupy the level, which splits into a fully occupied state and unoccupied doubly degenerate states under symmetry, giving a low-spin configuration . For the single-negative state , the ground-state configuration has and symmetry. Compared with , the Pt atom exhibits a larger displacement here, reaching 0.29 Å. This structure was experimentally confirmed by electron paramagnetic resonance (EPR) in the 1960s[33].
Figure 2(a) shows the calculated formation energies of defect in Si. As can be seen, three defect transition levels are introduced in the band gap. The (2+/+) and (+/0) donor levels are located at 0.14 eV and 0.39 eV above the VBM, respectively. The (−/0) acceptor level is 0.27 eV below the CBM. The calculated transition levels above are in good agreement with experimental results, as shown in Figure 2(b).

Figure 2. (a) Calculated formation energies of the Pt defect as a function of the Fermi level (). Solid lines represent the ground-state configuration for each charge state (, , , and ), while dashed lines denote higher-energy metastable configurations. The vertical dashed arrows mark the thermodynamic transition levels. (b) Comparison between the calculated transition levels (this work) and experimental values[1, 8, 12, 16, 17, 19, 23, 25]. All energies are referenced to the VBM or CBM as indicated.
To determine whether defect can serve as effective recombination centers, we calculate the nonradiative capture cross sections of electrons and holes for different charge states of the defect. We first analyze the nonradiative recombination process associated with the acceptor level (0/−). The configuration coordinate (CC) diagrams for these transitions are presented in Figure 3, where the potential energy surfaces (PESs) of

Figure 3. (a) CC diagram for hole capture (Pt + h → Pt) to the neutral state. (b) Calculated as a function of temperature, showing the average value (solid line) and contributions from three degenerate VBM states (dotted lines) compared with experimental data[1, 17, 20, 25] (stars). (c) CC diagram for electron capture (Pt + e → Pt) from the neutral state. (d) Calculated compared with experimental data[1, 8, 12, 17, 23, 25]. Solid and dotted lines show the average and individual contributions from six CBM states.
Pt and Pt are shown as functions of the generalized configuration coordinate within the harmonic approximation. In each diagram, corresponds to the equilibrium configuration of the final state, and denotes the displacement between the equilibrium configurations of the initial and final states. The effective phonon energy of the final state is represented as the average of the phonon energies in different modes, weighted by the projection to lattice relaxation,
where is the energy of -th phonon in the final state, and is the corresponding phonon mode vector. The effective phonon energy of the initial state based on the final-state phonon mode can be expressed as
where is the Duschinsky matrix between final state phonon mode and initial state mode .
When Pt () captures a hole from the VBM, it transits into Pt (). In contrast, when Pt () captures an electron from the CBM, the defect returns to the negatively charged state Pt (). Within a semiclassical picture, carrier capture proceeds through the crossing point of the two PESs, where a thermal activation barrier may arise. As can be seen in the CC diagram for hole capture process shown in Figure 3(a), the transition involves an energy barrier of 0.08 eV. Since the VBM of Si is triply degenerate, we calculate the hole capture cross sections () for each branch separately, together with their average value. As shown in Figure 3(b), exhibits a non-monotonic temperature dependence. Below 100 K,
decreases with increasing temperature due to the reduction of the Sommerfeld factor[34], as the Coulomb attraction between Pt and the hole becomes less effective at higher carrier thermal velocities. As the temperature rises above 100 K, increases as more phonons become available to promote the system toward the crossing point. This behavior is consistent with the temperature trend reported by Miller et al.[20]
We now turn to the electron capture process. Due to the indirect band gap of Si, the CBM consists of six equivalent valleys along the directions. In the supercell calculation, these valleys are folded onto the point, giving six nearly degenerate conduction band states. We therefore calculated the electron capture cross sections () for each conduction band state and the averaged value.
Notably, we find that the electron capture process is sensitive to the Jahn-Teller distortion orientation of the initial and final states. Since both Pt () and Pt () can exhibit anisotropic distortions, multiple symmetry-equivalent orientations exist within the and distorted lattices. The hole capture transition [Figure 3(a)] to the neutral configuration exhibits a large lattice displacement ( amu Å), involving a reorientation of the distortion direction relative to the configuration (compared in Fig. S2 of the SI). If the same configuration is used for electron capture, the substantial structural reorganization would create a large barrier ( eV) in the PESs, leading to a very small (see Fig. S4 of the SI). In contrast, if a symmetry-equivalent configuration, denoted as [Figure 3(c)], which shares a compatible distortion direction with the state, is considered, the transition via this configuration reduces the lattice displacement ( amu Å) and leads to a nearly barrier-less transition, enabling a much more efficient capture pathway.
The calculated is shown in Figure 3(d). The values have almost the same order of magnitude as that of . As the temperature rises from 0 K to 200 K, decreases from cm to cm. Above 200 K, plateaus at the order of cm. This negative temperature dependence is characteristic of barrier-less transitions, where thermal activation only yields limited increase in transition probability, causing the cross section to decline as the carrier thermal velocity increases ().
It is important to note that the and configurations are thermodynamically degenerate, representing the same Jahn-Teller distortion along different axes. Climbing Image Nudged Elastic Band (CI-NEB) [35] calculations reveal a low reorientation barrier of 0.12 eV between them (see Fig. S3 of the SI), enabling rapid thermal reorientation on a sub-nanosecond timescale (– Hz) at room temperature. Consequently, carrier capture kinetically funnels through the nearly barrier-less pathway, which justifies evaluating the cross sections via this optimized configuration.
The large cross sections (– cm) for both electron and hole captures are in good agreement with experimental data, indicating that the acceptor level enables highly efficient recombination and further confirming that identifying the correct symmetry-equivalent configuration of Jahn-Teller distortion is essential for reproducing the microscopic capture dynamics of .
For the donor level (+/0), the nonradiative capture process also depends on the choice of symmetry-equivalent configuration. As shown in Figure 4(a), electron capture by into the neutral configuration proceeds via a nearly barrier-less transition, resulting in large electron capture cross sections ( cm). Governed by Coulombic attraction, decreases monotonically with increasing temperature [Figure 4(b)], consistent with the trend reported in Ref. 20.

Figure 4. (a) CC diagram for electron capture () to the neutral state. (b) Calculated as a function of temperature, showing the average value (solid line) and contributions from six degenerate CBM states (dotted lines) compared with experimental data[1, 17, 20] (stars). (c) CC diagram for hole capture () from the neutral state. (d) Calculated compared with experimental data[1, 8, 12, 17, 23, 25] (stars). Solid and dotted lines show the average and individual contributions from three VBM states.
The subsequent hole capture process starting from the symmetry-equivalent configuration of involves a substantial activation barrier, as illustrated in Figure 4(c). The resulting capture cross section falls within a narrow range on the order of cm [Figure 4(d)]. Despite the presence of the barrier, exhibits only a weak temperature dependence. This behavior reflects the quantum-mechanical nature of multiphonon capture, where phonon-assisted tunneling effectively reduces the activation barrier inferred from the CC diagram[36, 37]. Overall, these large magnitudes of and (reaching cm) confirm that the donor level, similar to the acceptor level, functions efficiently in capturing both types of carriers.
To explicitly connect our microscopic calculations to macroscopic performance, we evaluated the defect-assisted recombination efficiency of at 300 K. Employing the recombination model based on the Sah-Shockley statistics[38], we find the recombination coefficient reaches s at a defect density of cm, yielding a carrier lifetime below 100 ns. These results definitively validate that acts as a highly efficient nonradiative recombination center. Further details are provided in the SI.
In summary, we have elucidated the microscopic nonradiative recombination dynamics of in silicon using first-principles calculations combined with nonradiative multiphonon theory. Our results show that introduces (2+/+), (+/0), and (0/−) levels within the band gap, consistent with experimental observations. The calculated nonradiative capture cross sections indicate that both the donor (+/0) and acceptor (0/−) levels can efficiently capture electron and hole @page 4/5 starts_mid=1 ends_mid=1 carriers, establishing as an effective recombination center. We further show that the carrier capture behavior is sensitive to the choice of symmetry-equivalent defect configurations with different Jahn-Teller distortions. These findings provide a microscopic understanding of carrier recombination at and highlight the importance of selecting symmetry-equivalent configurations in quantitative descriptions of defect-assisted nonradiative capture process.
AUTHOR DECLARATIONS
Conflict of Interest
The authors have no conflicts to disclose.
Author Contributions
Data Availability
The data that support the findings of this study are available from the corresponding author upon reasonable request.
References
- [1]J. Lutz, H. Schlangenotto, U. Scheuermann, and R. De Doncker, “Semiconductor properties,” in Semiconductor Power Devices: Physics, Characteristics, Reliability (Springer International Publishing, Cham, 2018) pp. 21–99.
- [2]C. Freysoldt, B. Grabowski, T. Hickel, J. Neugebauer, G. Kresse, A. Janotti, and C. G. Van de Walle, “First-principles calculations for point defects in solids,” Reviews of Modern Physics 86, 253–305 (2014).DOI
- [3]J. S. Park, S. Kim, Z. Xie, and A. Walsh, “Point defect engineering in thin-film solar cells,” Nature Reviews Materials 3, 194–210 (2018).DOI
- [4]M. Miller, “Differences between platinum-and gold-doped silicon power devices,” IEEE Transactions on Electron Devices 23, 1279–1283 (1976).DOI
- [5]B. Baliga and E. Sun, “Comparison of gold, platinum, and electron irradiation for controlling lifetime in power rectifiers,” IEEE Transactions on Electron Devices 24, 685–688 (1977).DOI
- [6]H. Carchano and C. Jund, “Electrical properties of silicon doped with platinum,” Solid-State Electronics 13, 83–90 (1970).DOI
- [7]M. Conti and A. Panchieri, “Electrical properties of platinum in silicon,” Alta Frequenza 40, 544–546 (1971).
- [8]J. Pals, “Properties of Au, Pt, Pd and Rh levels in silicon measured with a constant capacitance technique,” Solid-State Electronics 17, 1139–1145 (1974).DOI
- [9]K. Lisiak and A. Milnes, “Energy levels and concentrations for platinum in silicon,” Solid-State Electronics 18, 533–540 (1975).DOI
- [10]M. D. Miller, H. Schade, and C. J. Nuese, “Use of platinum for lifetime control in power devices,” in 1975 International Electron Devices Meeting (IEEE, 1975) pp. 180–183.DOI
- [11]M. Pugnet, J. Barbolla, J. C. Brabant, F. Saint-Yves, and M. Brousseau, “Energy levels for platinum and palladium in silicon measured by the dark transient capacitance technique,” physica status solidi (a) 35, 533–543 (1976).DOI
- [12]S. D. Brotherton, P. Bradley, and J. Bicknell, “Electrical properties of platinum in silicon,” Journal of Applied Physics 50, 3396–3403 (1979).DOI
- [13]Y. Kwon, T. Ishikawa, and H. Kuwano, “Properties of platinum-associated deep levels in silicon,” Journal of Applied Physics 61, 1055–1058 (1987).DOI
- [14]W. Stöffler and J. Weber, “Influence of hydrostatic pressure on the platinum levels in silicon,” Physical Review B 33, 8892–8895 (1986).pubmed.ncbi.nlm.nih.gov/9938318
- [15]A. A. Gill, N. Baber, and M. Z. Iqbal, “Role of the mid-gap level as the dominant recombination center in platinum doped silicon,” Journal of Applied Physics 67, 1130–1132 (1990).DOI
- [16]H. Zimmermann and H. Ryssel, “Trivalent character of platinum in silicon,” Applied Physics Letters 58, 499–501 (1991).DOI
- [17]B. D. B. Deng and H. K. Kuwano, “Platinum as recombination-generation centers in silicon,” Japanese Journal of Applied Physics 34, 4587 (1995).DOI
- [18]J. L. A. Alves and J. R. Leite, “Self-consistent one-electron states of substitutional and interstitial 5d transition-atom impurities in silicon,” Physical Review B 34, 7174–7187 (1986).pubmed.ncbi.nlm.nih.gov/9939373
- [19]J.-U. Sachse, E. Sveinbjörnsson, W. Jost, J. Weber, and H. Lemke, “Electrical properties of platinum-hydrogen complexes in silicon,” Physical Review B 55, 16176–16185 (1997).DOI
- [20]M. D. Miller, H. Schade, and C. J. Nuese, “Lifetime-controlling recombination centers in platinum-diffused silicon,” Journal of Applied Physics 47, 2569–2578 (1976).DOI
- [21]K. P. Lisiak and A. G. Milnes, “Platinum as a lifetime-control deep impurity in silicon,” Journal of Applied Physics 46, 5229–5235 (1975).DOI
- [22]M. J. Kumar, C. Selv kumar, V. Ramamurthy, and K. Bhat, “On the dominant recombination level of platinum in silicon,” physica status solidi (a) 87, 651–655 (1985).
- [23]A. O. Ewmaraye and E. Sun, “Electrical properties of platinum in silicon as determined by deep-level transient spectroscopy,” Journal of Applied Physics 47, 3172–3176 (1976).
- [24]W. Shockley and W. T. Read, “Statistics of the recombinations of holes and electrons,” Physical Review 87, 835 (1952).
- [25]S. Brotherton and P. Bradley, “Measurement of minority carrier capture cross sections and application to gold and platinum in silicon,” Journal of Applied Physics 53, 1543–1553 (1982).DOI
- [26]A. V. Krukau, O. A. Vydrov, A. F. Izmaylov, and G. E. Scuseria, “Influence of the exchange screening parameter on the performance of screened hybrid functionals,” The Journal of chemical physics 125 (2006), 10.1063/1.2406663.pubmed.ncbi.nlm.nih.gov/17176133
- [27]G. Kresse and J. Furthmüller, “Efficient iterative schemes for ab initio total-energy calculations using a plane-wave basis set,” Physical Review B 54, 11169 (1996).pubmed.ncbi.nlm.nih.gov/9984901
- [28]G. Kresse and D. Joubert, “From ultrasoft pseudopotentials to the projector augmented-wave method,” Physical Review B 59, 1758 (1999).DOI
- [29]W. Bludau, A. Onton, and W. Heinke, “Temperature dependence of the band gap of silicon,” Journal of Applied Physics 45, 1846–1848 (1974).
- [30]C. Freysoldt, J. Neugebauer, and C. G. Van de Walle, “Fully ab initio finite-size corrections for charged-defect supercell calculations,” Physical Review Letters 102, 016402 (2009).pubmed.ncbi.nlm.nih.gov/19257218
- [32]G. D. Watkins, “Deep levels in semiconductors,” Physica B+C 117, 9–15 (1983).DOI
- [33]H. H. Woodbury and W. G. Ludwig, “Spin resonance of pd and pt in silicon,” Physical Review 126, 466–470 (1962).DOI
- [34]R. Pässler, “Relationships between the nonradiative multiphonon carrier-capture properties of deep charged and neutral centres in semiconductors,” physica status solidi (b) 78, 625–635 (1976).DOI
- [35]G. Henkelman, B. P. Uberuaga, and H. Jónsson, “A climbing image nudged elastic band method for finding saddle points and minimum energy paths,” The Journal of Chemical Physics 113, 9901–9904 (2000).DOI
- [36]A. Alkauskas, Q. Yan, and C. G. Van de Walle, “First-principles theory of nonradiative carrier capture via multiphonon emission,” Physical Review B 90, 075202 (2014).DOI
- [37]D. Bartolo, Advances in nonradiative processes in solids (Science+Business Media New York, 1991).
- [38]S. Wang, M. Huang, S.-H. Wei, X.-G. Gong, and S. Chen, “Generalized a (n) bc recombination model in semiconductors with multilevel defects,” Journal of the American Chemical Society 147, 36825–36836 (2025).DOI