Radiation damage in silicon exposed to high-energy protons

Photoluminescence, infrared absorption, positron annihilation, and deep-level transient spectroscopy (DLTS) have been used to investigate the radiation damage produced by $24\mathrm{GeV}∕c$ protons in crystalline silicon. The irradiation doses and the concentrations of carbon and oxygen in the samples have been chosen to monitor the mobility of the damage products. Single vacancies (and self-interstitials) are introduced at the rate of $\sim 1{\mathrm{cm}}^{-1}$, and divacancies at $0.5{\mathrm{cm}}^{-1}$. Stable di-interstitials are formed when two self-interstitials are displaced in one damage event, and they are mobile at room temperature. In the initial stages of annealing the evolution of the point defects can be understood mainly in terms of trapping at the impurities. However, the positron signal shows that about two orders of magnitude more vacancies are produced by the protons than are detected in the point defects. Damage clusters exist, and are largely removed by annealing at $700\text{to}800\mathrm{K}$, when there is an associated loss of broad band emission between 850 and $1000\mathrm{meV}$. The well-known $W$ center is generated by restructuring within clusters, with a range of activation energies of about $1.3\text{to}1.6\mathrm{eV}$, reflecting the disordered nature of the clusters. Comparison of the formation of the $X$ centers in oxygenated and oxygen-lean samples suggests that the $J$ defect may be interstitial related rather than vacancy related. To a large extent, the damage and annealing behavior may be factorized into point defects (monitored by sharp-line optical spectra and DLTS) and cluster defects (monitored by positron annihilation and broadband luminescence). Taking this view to the limit, the generation rates for the point defects are as predicted by simply taking the damage generated by the Coulomb interaction of the protons and Si nuclei.

Figure 1

Infrared absorption measured at $20\mathrm{K}$ in oxygen-rich samples with high carbon (upper curve, displaced by $+0.1{\mathrm{cm}}^{-1}$ for clarity) and low carbon (lower curve). Radiation-induced peaks are labeled by their chemical origin, where for clarity the oxygen and carbon interstitials are simply labelled O and C. $I$ is a self-interstitial, O2 the oxygen dimer, and $2I$ two self-interstitials. The ‘CI’ peak is a doublet. Other peaks are caused by oxygen interstitials and oxygen dimers.

Figure 2

The average lifetimes of the positrons as functions of the measurement temperature in the oxygenated (diamonds) and nonoxygenated (squares) samples, in the as-irradiated state.

Figure 3

Lower figure shows the average lifetime of the positrons, measured at $20\mathrm{K}$, in the oxygenated (diamonds) and nonoxygenated (squares) samples, following annealing at $50\mathrm{K}$ stages. The upper plot shows the defect-related lifetimes derived from the measured data.

Figure 4

Preirradiation, both materials show identical luminescence above $1000\mathrm{meV}$ from the phosphorus donors (“Ph”) plus weak free-exciton emission. The oxygenated material, labeled “Pre,” also contains weak dislocation bands at 808 and $874\mathrm{meV}$. After irradiation, the high-energy luminescence is quenched, and both materials show a broadband between about 750 and $1000\mathrm{meV}$, plus weak $W$ emission at $1018\mathrm{meV}$. In the oxygenated sample $(“+\mathrm{O}”)$, carbon interstitials have been trapped preferentially at oxygen, forming the C center $\left(789\mathrm{meV}\right)$, and in the nonoxygenated sample (“No O”) preferentially at substitutional carbon, creating the $G$ center $\left(969\mathrm{meV}\right)$. The sharp absorption lines and the dip near $800\mathrm{meV}$ are from atmospheric and instrumental absorption. The spectra shown here were measured at $4.2\mathrm{K}$ and at $60\mu \mathrm{eV}$ resolution.

Figure 5

Intensities of the $W$ and $C$ zero-phonon lines, measured at $4.2\mathrm{K}$. Squares and circles are for the $W$ line in non-oxygenated and oxygenated samples, respectively. The triangles are for the $C$ line in the oxygenated samples. The relative uncertainties in the intensities are comparable to the heights of the symbols.

Figure 6

Full widths at half height of the $W$ and $C$ zero-phonon lines, measured at $4.2\mathrm{K}$. Squares and circles are for the $W$ line in nonoxygenated and oxygenated samples, respectively. The triangles are for the $C$ line in the oxygenated samples. The data were obtained at a spectral resolution of $20\mu \mathrm{eV}$.

Figure 7

After $425\mathrm{K}$ annealing, both materials are developing strong, and identical, $W$ bands. The nonoxygenated sample shown here, bottom spectrum, has the $G$ $\left(969\mathrm{meV}\right)$ and $C$ $\left(789\mathrm{meV}\right)$ lines; the oxygenated has a negligible $G$ line and the $C$ line is 4.5 times stronger. By $675\mathrm{K}$, the strongest features are the $J$ lines in the nonoxygenated (second spectrum from bottom), and the $1096\mathrm{meV}$ line, labeled 1, and the $P$ line in the oxygenated material (top spectrum). The sharp absorption lines and the dip near $800\mathrm{meV}$ are from atmospheric and instrumental absorption. The spectra shown here were measured at $4.2\mathrm{K}$ and at $60\mu \mathrm{eV}$ resolution.

Figure 8

Intensities of the luminescence from zero-phonon lines as functions of annealing in the nonoxygenated samples (a) and the oxygenated samples (b). Data are shown by squares for the $G$ line, circles for the $T$ line, triangles for the $X$ line, + for the $M$ line, inverted triangles for the $991\mathrm{meV}$ line, × for the $P$ line, and ⋆ for the $1096\mathrm{meV}$ line. The relative uncertainties in the intensities are comparable to the heights of the symbols.

Figure 9

Spectra are for the oxygenated material; very similar spectra are observed in the nonoxygenated material, with the intensities of the $M$, $P$, $H$, and $T$ lines depending on the carbon and oxygen content (Sec. 5). By $725\mathrm{K}$, top spectrum, the $X$ $\left(1049\mathrm{meV}\right)$ and $991\mathrm{meV}$ 9 lines are present equally in both materials, but the $1101.2\mathrm{meV}$ line 1101 only occurs in the oxygenated material. At $925\mathrm{K}$ the broad background luminescence has been reduced mainly to a sharper peak, apparently centered on $800\mathrm{meV}$, but distorted by the low-energy cutoff of the spectrometer system. New lines are seen at $1138\mathrm{meV}$ from the $B_{71}^1$ system, labeled $B$. The phosphorus “Ph” emission is again detectable, and other point-defect features are reducing, to disappear by $1125\mathrm{K}$ (bottom spectrum). The line “Cu” at $1014\mathrm{meV}$ is from copper contamination introduced during the heat treatment. The sharp absorption lines and the dip near $800\mathrm{meV}$ are from atmospheric and instrumental absorption. The spectra shown here were measured at $4.2\mathrm{K}$ and at $60\mu \mathrm{eV}$ resolution.

Figure 10

Variation of the intensity of the broad background band measured at $4.2\mathrm{K}$, integrated from $680\text{to}930\mathrm{meV}$, in the oxygenated (diamonds) and nonoxygenated (squares) samples, following annealing in $50\mathrm{K}$ stages.

Figure 11

Calculations, using the simple model of Sec. 6 for the damage created in a cube of side $1\mu \mathrm{m}$ by irradiation with ${10}^{14}{\mathrm{cm}}^{-2}$ protons, of the number of damage events which generate different total numbers of vacancies. There are about 100 damage events creating monovacancies, about 50 creating a total of two vacancies, and so on. The maximum energy of the PKA considered is $1500\mathrm{eV}$, when the probability of occurence is only 0.01% of the probability of a nucleus being displaced with $15\mathrm{eV}$.