Electron emission and capture by oxygen-related bistable thermal double donors in silicon studied with junction capacitance techniques

It has been recently suggested that oxygen-related bistable thermal double donors (BTDDs) are responsible for the reduction of minority carrier lifetime and conversion efficiency of novel amorphous-crystalline Si heterojunction solar cells with their base formed from n-type Czochralski-grown (Cz) silicon [M. Tomassini et al., J. Appl. Phys. 119, 084508 (2016)]. To test this hypothesis, we have studied processes associated with carrier emission and capture by BTDDs in p+-n and Schottky barrier diodes on n-type Cz-Si materials with the use of junction capacitance techniques. By means of deep level transient spectroscopy, we have detected electron emission signals from the deep donor state of the BTDD-0 and BTDD-1 centers. The values of activation energy for electron emission (Eem) have been determined as 1.01 ± 0.01 and 0.91 ± 0.01 eV for the BTDD-0 and BTDD-1 centers, respectively. Such high Eem values are very unusual for defects in Si. We have carried out measurements of electron capture kinetics and associated shallow donor–deep donor transformations for the BTDD-0 and BTDD-1 defects at different temperatures in the diodes with different doping levels. Energy barriers for the capture-transformation processes have been determined. It is argued that BTDDs are responsible for carrier trapping in n-type Cz-Si crystals but are not effective recombination centers.It has been recently suggested that oxygen-related bistable thermal double donors (BTDDs) are responsible for the reduction of minority carrier lifetime and conversion efficiency of novel amorphous-crystalline Si heterojunction solar cells with their base formed from n-type Czochralski-grown (Cz) silicon [M. Tomassini et al., J. Appl. Phys. 119, 084508 (2016)]. To test this hypothesis, we have studied processes associated with carrier emission and capture by BTDDs in p+-n and Schottky barrier diodes on n-type Cz-Si materials with the use of junction capacitance techniques. By means of deep level transient spectroscopy, we have detected electron emission signals from the deep donor state of the BTDD-0 and BTDD-1 centers. The values of activation energy for electron emission (Eem) have been determined as 1.01 ± 0.01 and 0.91 ± 0.01 eV for the BTDD-0 and BTDD-1 centers, respectively. Such high Eem values are very unusual for defects in Si. We have carried out measurements of electron capture kinetics and ass...


I. INTRODUCTION
It has been known since the 1950s that heat-treatments of oxygen-rich Si crystals in the temperature range 350-500 °C result in the formation of a family of defects with shallow donor levels. 1,2It has been further found that the defects are helium-like centers with the donor levels E(0/+) and E(+/+2) located in the energy ranges from 0.05 to 0.07 eV below the conduction band edge E c -(0.05-0.07)eV and E c -(0.12-0.16)8][9] Therefore, the defects are usually referred to as oxygen-related thermal double donors (TDDs). 8The TDD defect family consists of up to 17 species (from-TDD-0 to TDD-16) differing by the number of oxygen atoms incorporated. 8,10,11It is thought that oxygen atoms in TDDs are arranged in linear chains lying in <110> planes.6][17][18][19][20] Emission or capture of electrons or holes by the BTDDs in one of the configurations can trigger a transition of the defects to other configurations.6][17][18][19][20][21][22] Figure 1 shows a configuration coordinate diagram for the BTDD defects which we have constructed from the published data, and Table I presents energy parameters for all three species of the BTDD family, which were determined in previous works.
3][14] It should be noted, however, that practically all experimental information about the electronic properties of BTTDs has been obtained from observations of the defects in the shallow donor configuration.No signals from the deep donor configuration of BTDDs have been detected by spectroscopic techniques, such as electron spin resonance, deep level transient spectroscopy (DLTS), or infrared absorption spectroscopy.An exception is a weak infrared absorption line assigned to a local vibrational mode of the BTDD-1 and BTDD-2 centers in the deep donor state. 23he recent growing use of n-type Czochralski (Cz) silicon materials for photovoltaic (PV) applications (Refs.7][28][29][30] Furthermore, it has been speculated in a recent study (Ref.29) that BTTDs in the deep donor state are the effective recombination centers and their presence can cause the reduction in conversion efficiency of novel amorphous-crystalline Si heterojunction solar cells with the base from n-type Czochralski-grown silicon.20]27 The existence of such trapping centers can result in some errors in measurements of the minority carrier lifetime by photoconductance decay and quasi-steady-state photoconductance techniques. 27,31t is possible to take into account the effects of the trapping centers on the minority carrier lifetime measurements, but this requires information about the electronic characteristics of the centers.So a deeper understanding of electronic and recombination properties of BTDDs is required.

II. EXPERIMENTAL DETAILS
We have used p + -n-n + and Schottky barrier diodes on n-type Czochralski-grown Si materials to study the BTDD defects.P + -n-n + diodes were formed by implantation and subsequent thermal activation of 60 keV boron ions (front side) and 60 keV phosphorus ions (back side) in the phosphorus-doped (ρ ≈ 5 Ω cm) commercial electronic grade Cz-Si wafers.The area of the p + -n-n + diodes was 6 mm 2 and the leakage current at 10 V reverse bias (RB) was about 10 −6 A. The concentration of interstitial oxygen (O i ) in these diodes was estimated as (9 ± 1) × 10 17 cm −3 from the rates of capture of interstitial carbon atoms by the O i atoms determined from annealing experiments on the diodes irradiated with alpha particles at 260 K. 32,33 Schottky barrier diodes (1 mm in diameter) were prepared by thermal evaporation of gold through a shadow mask on samples from a few n-type Cz-Si crystals with initial resistivity from 0.7 to 5 Ω cm.Concentrations of interstitial oxygen and substitutional carbon atoms in the crystals have been determined from infrared absorption spectra measured at room temperature.The oxygen concentration in the crystals was in the range from 7 × 10 17 cm −3 to 1 × 10 18 cm −3 and the carbon concentration was below 2 × 10 16 cm −3 .Some of the samples have been treated in the temperature range from 350 to 450 °C to generate TDDs.The heat-treatments resulted in increases in shallow donor concentration and, therefore, the capacitance of the diodes but did not result in significant changes in their leakage current and ideality factor.
Current-voltage and capacitance-voltage (C-V) measurements have been carried out in order to evaluate the quality of the diodes and to determine the uncompensated shallow donor concentration and width of the probed depletion regions.Deep level transient spectroscopy and high-resolution Laplace DLTS (L-DLTS) techniques have been used for the detection of electron emission from deep defect states. 34Measurements of changes in the total capacitance of the diodes resulted from filling in the depletion regions with electrons have been used to monitor transformations of the BTDD defects from the shallow donor state to the deep donor state.

III. EXPERIMENTAL RESULTS AND DISCUSSION
Figure 2 shows the conventional DLTS spectra in the temperature ranges 35-300 K and 300-435 K for p + -n-n + diodes fabricated on phosphorus-doped 5 Ω cm Czochralski-grown silicon.The spectra were recorded for an as-processed diode and a diode subjected to a heat-treatment at 350 °C for 10 h.The heat-treatment resulted in an increase in uncompensated shallow donor concentration from (1.05 ± 0.1) × 10 15 cm −3 to (1.65 ± 0.1) × 10 15 cm −3 .It should be noted that all the spectra were recorded upon temperature scanning from low to high values.A small peak with its maximum at about 52 K [for an emission rate (e em ) window of 50 s −1 ] occurs in the spectrum of the as-processed diode.The magnitude of this peak has increased significantly and its maximum has been shifted to about 46 K after the heat-treatment.It has been found that the magnitude of the low-temperature peak depends on the cooling down conditions before DLTS measurements.In the spectrum recorded after cooling down with the applied  reverse bias, the magnitude of the peak is significantly higher than that in the spectrum recorded after cooling down without bias [Fig. 2 Changes in the magnitude of the peak resulted from different cooling down conditions show that a part of the TDDs is the bistable TDDs species. 21Upon cooling down with the applied bias, BTDDs can be frozen in the metastable shallow donor configuration and contribute to the electron emission signal at 46 K in the DLTS spectra, while after cooling down without bias, some of BTDDs are in the minimum-energy deep donor configuration, electron emission from which does not occur at low temperatures. 21Qualitatively similar DLTS spectra have been recorded on other p + -n and Schottky barrier diodes subjected to heat-treatments in the temperature range 350-450 °C.
In the DLTS spectra recorded in the temperature range 300-435 K for the heat-treated Si diode, a previously unreported broad peak with its maximum at about 402 K [for e em = 10 s −1 and filling pulse length (t p ) of 200 ms] occurs [Fig.2(b)].The magnitude and position of this peak depend significantly on the t p value.Clear detection of the peak requires rather long filling pulses in the range of hundreds of milliseconds.Also, the magnitude of the peak decreases with the increase in the emission rate window, thereby indicating a smaller electron population of the corresponding trap with increasing temperature.This peak has not been observed in the spectra of as-processed diodes without the 350 °C heattreatment, therefore its appearance is caused by defects which form upon annealing.
The application of the high-resolution Laplace DLTS technique allows us to resolve two electron emission signals contributing to the broad peak in the conventional DLTS spectra.The L-DLTS spectrum recorded at 435 K is shown in Fig. 3(a).The peaks in the L-DLTS spectrum are sharp, thereby indicating that electron emission signals are related to well-defined point defects. 349][20] Thus, it is reasonable to suggest that the observed emission signals are related to BTDDs.It should be noted here that the concentration of BTDDs, which are the main subject in this work, in all the samples studied has not exceeded 5% of the uncompensated shallow donor concentration, so the application of FIG. 2. Conventional DLTS spectra recorded on p + -n-n + diodes from the 5 Ω cm n-type Cz-Si material.The spectra with label 1 were recorded on an as-processed diode and the spectra with labels 2 to 5 correspond to an identical diode which was subjected to a heat-treatment at 350 °C for 10 h.Measuring conditions are given in the figures.conventional methods of analysis of DLTS and L-DLTS signals due to these defects is justified.Furthermore, it should be mentioned that E em values determined from carrier emission measurements in charge space regions for defects with strong lattice relaxations can differ significantly from E em values determined from monitoring charge state transitions under charge neutrality conditions. 37or further characterization of the traps responsible for the high-temperature electron emission signals, we have carried out: (a) measurements of electron population of the traps (which is proportional to the DLTS ΔC signal) upon changes in filling pulse length, i.e., the so-called direct capture cross section measurements and (b) measurements of transitions of the traps from the ionized shallow donor state to the filled deep donor state by monitoring changes in the capacitance, C, of the reverse-biased diodes in the temperature range 230-320 K after application of long filling pulses.To observe the capacitance changes according to measurements (b), the diodes were usually cooled down under applied reverse bias (RB) from a temperature above 400 K to a measurement temperature (T m ) and then subjected to the application of multiple long filling pulses with the pulse voltage close to 0 V, which was interrupted after certain periods of time for bias capacitance measurements.The RB cooling down step used to keep BTDDs in the shallow donor configuration can be replaced for p + -n junctions by the application of a forward bias pulse at T m = 230-320 K.The forward bias pulse results in injection of minority carriers (holes) into the probed region and deep-shallow transformation of BTDDs according to the following transition (Fig. 1): A 0 + h + + e − → B + + e − .
All the measurements were carried out on a p + -n-n + diode from 5 Ω cm n-type Cz-Si material, which was annealed at 350 °C for 10 h.
Changes in the ΔC and C values according to the above (a) and (b) processes at 420 K and 280 K, respectively, are shown in Fig. 4. The solid lines in Figs.4(a) and 4(b) are calculated for mono-exponential growth and multi-exponential decay processes with the least-square fitting of exponent parameters.The fitting parameters were ΔC values and characteristic occupancy times/rates.It should be mentioned here that an analysis of the data shown in Figs.4(a) and 4(b) indicates that the ΔC values for BTDD-1 differ significantly at 420 K and 280 K (ΔC 420 K = 0.215 pF and ΔC 280 K = 2.4 pF, respectively).This difference can be explained by partial (about 10%) filling of the BTDD-1 center with electrons in the high temperature (350-430 K) range, in which the Fermi level is below the E(0/+2) = E c -0.315 eV occupancy level of the defect. 16Usually, DLTS and L-DLTS signals in diodes with the base from n-type semiconductors are recorded when the Fermi level in the bulk of the material is located higher than an energy level of a defect, so the defect level can be fully occupied with electrons upon filling pulse conditions.However, in the case of BTDDs, the activation energies of electron emission from a deep donor state are very high, so, with the equipment, which we have used, the electron emission from the BTDD-1 defect can be observed only in the range of rather high temperatures, where in the samples studied E F ≤ E(0/+2).
In Fig. 5, the temperature dependencies of the occupancy rates for both traps are plotted together with the electron emission rates.The values have been derived from an analysis of capacitance measurements described above on diodes from 5 Ω cm n-type Cz-Si, which were subjected to 10-h heat-treatments at 350 °C and 425 °C.
Occupancy kinetics for defects with negative-U properties and different structural models have been considered in Refs.18, 19, and 37.There are some characteristic features in temperature dependencies of occupancy and emission rates for defects with U < 0, which indicate on details of their electronic structure.From an analysis of such characteristic features in the temperature dependencies of emission and occupancy rates presented in Fig. 5, and taking into account positions of the Fermi level at the measurement temperatures, some preliminary information about electronic and structural properties of the traps can be obtained.First, it is found that the capture rate for both traps is proportional to the squared concentration of free electrons when the Fermi level is in the range E c -0.26 eV < E F < E c -0.18 eV.This finding clearly indicates negative-U properties of the traps studied. 18,19,37urthermore, the occupancy rates are nearly the same as the emission rates for the BTDD-1 traps when the Fermi level is lower than E c -0.36 eV.The rates do not depend on the free carrier concentration in this E F range.This observation indicates that most likely there is no single positively charged state for BTDD-1 in configuration A. 37 Both findings are consistent with structure and electronic properties of BTDDs and indicate that these defects give rise to the electron emission signals, which are observed in the conventional and Laplace DLTS spectra in the temperature range 320−435 K [Figs.Temperature dependencies of occupancy rates for defects with such electronic properties can be described by the following equations: 18,19,37 when E F < E(0/+2) and when E(0/+2) < E F < E(+/+2).In these equations, E em is the activation energy for electron emission from the neutral deep donor state, ΔE(0/+) and ΔE(+/+2) are the energy differences between the states with corresponding charges, ΔE c (+/0) is the energy barrier for transition from the positively charged shallow donor state to the neutral deep donor state, and n is the concentration of free electrons.The solid curves in Fig. 5 were calculated with the use of Eqs. ( 1) and ( 2) and with A and B coefficients and energy values as fitting parameters.The fitting parameters obtained are presented in Fig. 5 and the calculated energy values are given in Table I.
The capture rate values for BTDD-0 traps in the high temperature (400 K-440 K) region deviate from the values obtained by extrapolation of low temperature capture data and electron emission data [solid lines calculated according to Eqs. ( 1) and ( 2)].
It was concluded in Refs.18-20 that in these temperature regions, the B + →A 0 transformation of BTDD-0 is related to a hole emission from the B + state to the valence band.
Because of limitations of our experimental setup in relation to measurement temperature and filling pulse length and the relatively small concentration [(1.0-1.5)× 10 12 cm −3 ] of the BTDD-0 traps, we could not determine precise values of activation energy and power in electron concentration dependence for the hole emission process from the data obtained.
9][20] Furthermore, the values of energy barriers for transitions from the shallow donor state to deep donor state of the BTDD-0 and BTDD-1 defects, which have been derived from the analysis of the results presented in Fig. 5 and are given in Table I, are close to those obtained in Refs.15 and 18-20 for the corresponding traps from photoconductance decay measurements on heat-treated n-type Cz-Si crystals.

IV. CONCLUSION
It is reasonable to conclude from the analysis of the given data that electron emission signals from the deep donor state of the BTDD-0 and BTDD-1 defects have been detected by means of the DLTS technique in the present work.By monitoring changes in magnitudes of the detected emission signals at different bias and temperature conditions in the diodes with different doping levels, we have managed to obtain detailed information on electronic properties of the BTDD-0 and BTDD-1 centers.With the knowledge of the electronic structure of the BTDD species, it is possible to predict their effects on the minority carrier lifetime in silicon crystals.The recombination of light-or forward-bias-injection-induced charge carriers through BTDDs in n-type Si consists of the following processes (Fig. 1): (i) capture of a hole by a BTDD defect in deep donor configuration and its relatively fast transformation into the shallow donor state (A 0 + h + + e − → B + + e − ); 18,19 (ii) capture of an electron by the defect in B configuration and its transformation to the initial deep donor A configuration (B + + e − → A 0 ).The results shown in Figs. 4 and  indicate that the rate of process (ii) is not high in the temperature range of operation of solar cells around 300 K, ∼10 2 s −1 for BTTD-0 and ∼10 −1 s −1 for BTDD-1.According to the results presented in Refs.18 and 19, the rate of process (ii) is even slower for the BTDD-2 defect.The B + + e − →A 0 process is slow because of the existence of a relatively high energy barrier, ΔE c (+/0), for this transformation (Table I).So, the whole cycle of the A → B → A transformations of a BTDD defect, which can be considered as a recombination event, is relatively long, in the range of hundreds of milliseconds even for the fastest cycle through the BTDD-0 center.So, it appears that the BTDD defects can be considered as minority carrier trapping centers in n-type Cz-Si crystals but not as effective recombination centers.So, the recent suggestion by Tomassini et al. 29 that BTDDs in the deep donor state are responsible for the reduction of minority carrier lifetime and conversion efficiency of novel amorphous-crystalline Si heterojunction solar cells with the base from n-type Czochralski-grown (Cz) silicon is not valid.
It is likely that in n-type Cz-Si crystals with a high concentration of TDDs and decreased resistivity, the Fermi level is shifted to the conduction band and because of this, a significant part of the helium-like TDD species, which have the E(+/+2) level at about E c -0.15 eV, are in the singly ionized state.These singly ionized centers can effectively capture holes and become doubly positively charged.The TDD centers in the +2 state are extremely effective traps for electrons, so the rate of electron capture and the back transition to the singly ionized state is very high.Therefore, in n-type Si crystals with a high concentration of TDDs, these defects could serve as recombination centers in the shallow donor configuration.

FIG. 1 .
FIG. 1. Configuration coordinate diagram for the bistable thermal double donors in silicon.
(a)].A comparison of the DLTS spectra in Fig. 2(a) with those reported in the literature for heat-treated n-type Cz-Si samples (Refs.21, 35, and 36) indicates that the low-temperature peak can be associated with electron emission from the second donor level of oxygen-related TDDs.

Figure 3 (
b) shows Arrhenius plots of the T 2 -corrected emission rates for both traps.The values of activation energy for electron emission (E em ) have been determined as 0.91 ± 0.01 eV and 1.01 ± 0.01 eV.Such high E em values are very unusual for defects in silicon detected by DLTS.

FIG. 3 .
FIG. 3. (a) Laplace DLTS spectrum recorded at 435 K on a p + -n-n + diode from the 5 Ω cm n-type Cz-Si material, which was subjected to a heat-treatment at 350 °C for 10 h.Measurement conditions are given in the figure.(b) Arrhenius plots of T 2 -corrected electron emission rates measured with the use of Laplace DLTS technique for two electron traps, whose emission signals are shown in (a).

FIG. 4 .
FIG. 4. (a) Changes in magnitudes of the L-DLTS signals due to the BTDD-0 and BTDD-1 centers upon changes in the filling pulse length.L-DLTS measurements were carried at 420 K with the following measurement conditions: U b = −11 V, U p = −4 V.The solid lines are calculated for the mono-exponential growth process with least-square fitting values of ΔC max and characteristic growth rates.(b) Changes in compensated diode capacitance (C d -C b ) at U b = −11 V and T m = 280 K.The changes were induced by the application of multiple filling pulses with U p = −4 V. Further details of the measurement are described in the text.The solid line is calculated for a three-exponent decay process with least-square fitting values of ΔC n (n from 1 to 3) and characteristic decay rates.For the improvement of the signal-to-noise ratio in these measurements, we have used a backing off capacitor of C b = 200 pF to compensate the capacitance of the diode studied (C d ).

TABLE I .
Energy parameters (in eV) of bistable thermal double donors in silicon.The values given in brackets have been determined in the present work.