**4.1.3 Possible origin of the N-related recombination center E1**

It is worth remembering that the atomic structure of *E*1 may be free from impurities and doping atoms owing to the difference in the density of residual impurities in GaAsN grown with MOCVD, MBE, and CBE. Furthermore, the uniform distribution of *N*E1 in the bulk of GaAsN indicates that *E*1 is formed during growth to compensate for the tensile strain caused by the small atomic size of N compared with that of As. Therefore, the origin of *E*1 has high probability to depend only on the atoms of the alloy (N, As, Ga). To confirm these expectations, the origin of *E*1 is investigated qualitatively by considering the results of two different experiments: (*i*) the dependence of *N*E1 to the As source flow rate and (*ii*) the effect of H implantation on the distribution of lattice defects in *n*-type GaAsN.

### **4.1.3.1 Dependence of NE1 to As source flow rates**

The objective of this experiment is to clarify whether the density of *E*1 is sensitive to the As atom or not. The MMHy was supplied to 9.0 sccm and the TDMAAs was varied between and 0.7 and 1.5 sccm. As shown in Fig. 6. (a), increasing TDMMAs drops the *N* concentration in the film and tends to saturate for a flow rate higher than 1 sccm. For two emission rate *erw* = {100, 10} s-1 and filling pulse *tp* = {0.1, 5} ms values, the DLTS spectra are normalized on the junction capacitance to exclude the effect of various carrier densities in the samples. The same *N*-related recombination center *E*1 was observed in all samples. The TDMAAs flow rate dependence of the DLTS peak height of *E*1 for two settings of measurement parameters is given in Fig. 6(b). A peaking behavior at approximately TDMAAs = 0.9 sccm was obtained. As *N*E1 is uniformly distributed in GaAsN films, the incorporation of *N* atom at the growth surface affects both the incorporation of *N*As and the formation of *E*1. If *E*1 depends only on *N* atom, the decrease of [*N*] with increasing TDMMAs flow rate results in monotonically dropping of *N*E1. However, the peaking behavior of *N*E1 indicates the sensitivity of *E*1 to As atom, either than *N*.

Fig. 6. TDMAAs flow rate dependence of (a) N concentration under supplied MMHy and (b) normalized DLTS peak height of E1 for two settings of measurement parameters.

### **4.1.3.2 Effect of H implantation on lattice defects in GaAsN**

GaAsN films were treated by H implantation. This experiment was used because H bounds strongly to *N* in GaAsN films to form *N-H* complexes (Suzuki et al., 2008; Amore & Filippone, 2005). H ions with multi-energy from 10 to 48 keV were implanted into GaAsN layers with peaks concentration of 5 1018 (GaAsNHD1) and 1 1019 atom/cm3 (GaAsNHD2),

It is worth remembering that the atomic structure of *E*1 may be free from impurities and doping atoms owing to the difference in the density of residual impurities in GaAsN grown with MOCVD, MBE, and CBE. Furthermore, the uniform distribution of *N*E1 in the bulk of GaAsN indicates that *E*1 is formed during growth to compensate for the tensile strain caused by the small atomic size of N compared with that of As. Therefore, the origin of *E*1 has high probability to depend only on the atoms of the alloy (N, As, Ga). To confirm these expectations, the origin of *E*1 is investigated qualitatively by considering the results of two different experiments: (*i*) the dependence of *N*E1 to the As source flow rate and (*ii*) the effect

The objective of this experiment is to clarify whether the density of *E*1 is sensitive to the As atom or not. The MMHy was supplied to 9.0 sccm and the TDMAAs was varied between and 0.7 and 1.5 sccm. As shown in Fig. 6. (a), increasing TDMMAs drops the *N* concentration in the film and tends to saturate for a flow rate higher than 1 sccm. For two emission rate *erw* = {100, 10} s-1 and filling pulse *tp* = {0.1, 5} ms values, the DLTS spectra are normalized on the junction capacitance to exclude the effect of various carrier densities in the samples. The same *N*-related recombination center *E*1 was observed in all samples. The TDMAAs flow rate dependence of the DLTS peak height of *E*1 for two settings of measurement parameters is given in Fig. 6(b). A peaking behavior at approximately TDMAAs = 0.9 sccm was obtained. As *N*E1 is uniformly distributed in GaAsN films, the incorporation of *N* atom at the growth surface affects both the incorporation of *N*As and the formation of *E*1. If *E*1 depends only on *N* atom, the decrease of [*N*] with increasing TDMMAs flow rate results in monotonically dropping of *N*E1. However, the peaking

**0.8 1.0 1.2 1.4**

**TDMAAs (sccm)**

**en =100 s-1, tp = 0.1ms**

**en =10 s-1, tp = 5.0 ms**

**1.0**

**1.5**

**C/CR** (**x 10-3**)

Fig. 6. TDMAAs flow rate dependence of (a) N concentration under supplied MMHy and (b)

GaAsN films were treated by H implantation. This experiment was used because H bounds strongly to *N* in GaAsN films to form *N-H* complexes (Suzuki et al., 2008; Amore & Filippone, 2005). H ions with multi-energy from 10 to 48 keV were implanted into GaAsN layers with peaks concentration of 5 1018 (GaAsNHD1) and 1 1019 atom/cm3 (GaAsNHD2),

normalized DLTS peak height of E1 for two settings of measurement parameters.

**3.5 4.0 4.5** **(b)**

**4.1.3 Possible origin of the N-related recombination center E1** 

of H implantation on the distribution of lattice defects in *n*-type GaAsN.

behavior of *N*E1 indicates the sensitivity of *E*1 to As atom, either than *N*.

**4.1.3.1 Dependence of NE1 to As source flow rates** 

**0.8 1.0 1.2 1.4**

**(a) MMHy = 9 sccm**

**TDMAAS (sccm)**

**4.1.3.2 Effect of H implantation on lattice defects in GaAsN** 

**0.20**

**0.25**

 **[N] (%)**

**0.30**

respectively. The depth of implantation was thought to be distributed between 110 and 410 *nm* from the surface of GaAsN by calculating the *SRIM* 2003 simulation code (Ziegler, 1985). After implantation, the samples were treated by post thermal annealing at 500 C for 10 min under N2 gas and GaAs cap layers. As plotted in Fig. 7(a) and (b), the crystal quality of GaAsN films after implantation was controlled using XRD curves and C-V measurements. DLTS spectra of implanted samples are shown in Fig. 7(c). After implantation, *E*1 was not observed; however, two new lattice defects appeared. The signature and the density of these traps are summarized in Table 1. The thermal emission from them is plotted as an Arrhenius plot in Fig. 7(d).


Table 1. Summary of Ea, , adjusted Nt-adj, and possible origin of defects in as grown and implanted GaAsN samples.

Fig. 7. (a)DLTS spectra of as grown and implanted GaAsN and (b) their Arrhenius plots.

The new electron trap (*EP*1) is located approximately 0.41 eV below the CBM of GaAsN. Its properties are identical to that of the native defect *EL*5 in GaAs (Reddy, 1996). Its atomic structure was discussed in many publications, where the common result indicated that *EL*5 is a complex defect free from impurities and dominated by As interstitials, such VGa-Asi or AsGa-VGa (Deenapanray et al., 2000; Yakimova et al., 1993). The second new defect is a hole trap (*HP*1) at approximately an average activation energy 0.11 eV above the VBM of GaAsN. Compared with majority carrier traps in GaAs grown with various techniques, no similar hole trap to *HP*1 was reported. However, in *p*-type GaAsN grown by CBE with around the same *N* concentration, *E*HP1 and HP1 are identical to that of the hole trap *HC*2 in *p*-type GaAsN grown by CBE (bouzazi et al., 2011). This defect was confirmed recently to be an acceptor state in GaAsN films (see § 4.2.3) and to be related to the N-H bond. While the H impurity was provided by implantation, the N atom can originate through two possibilities: First; examining XRD results, the N atom can originate from its ideal site. However, [N] in as grown is much higher than that in implanted samples (1019 cm-3). It is also much higher than NHP1. Second, N atom can be originated from the complete dissociation of *E*1, since the

Investigation of Lattice Defects in GaAsN

Grown by Chemical Beam Epitaxy Using Deep Level Transient Spectroscopy 503

hole trap *H*5 presents the same properties as *H*3 and *HA*5 (Li et al., 2001; Katsuhata et al., 1978). It is proposed to be the double donor state (+/++) of *EL*2 (Bouzazi et al., 2010). The hole

Traps Et (eV) p(cm2) Nt-adj(cm-3) *H*0 0.052 2.16 10-14 4.64 1016 *H*2 0.185 3.87 10-17 4.52 1015 *H*5 0.662 6.16 10-14 6.24 1015

DC-DLTS measurements were carried out to confirm whether there is a recombination center among the hole traps or not. As shown in Figs. 9(a) and (b), the DC-DLTS signal is compared to that of conventional DLTS. A decrease in the DLTS peak height of *H*0 is observed and confirmed by varying the voltage and the duration of the injected pulse.

**0.6**

The shallow hole trap H0 is observed and reported for the first time owing to the temperature range it was recorded, which cannot be reached with standard DLTS systems. Second, its capture cross section is large enough to capture majority carriers and minority carriers. The reduction in the peak height of H0 is explained by the electron holerecombination. This implies that H0 is a shallow recombination center in p-type GaAsN grown by CBE and can also play the role of an acceptor state. To verify whether the recombination process via H0 is radiative or not, the temperature dependence of the capture cross section of electrons is obtained by varying the emission rate *erw* from 1 to 50 s-1. As shown in Fig. 10(a), the peak temperature of H0 shifts to high temperatures with increasing *erw*. The value of H0 is obtained from the fitting of the Arrhenius plots for each *erw*. It is important to note that the fitting errors of activation energy and capture cross section are relatively large owing to the instability of temperature in the range of measurements. As shown in Fig. 10(b), the capture cross section of *H*0 does not exhibit an Arrhenius behavior, which excludes the non-radiative recombination process. Its shallow energy level suggests

**0.9**

**1.2**

**1.5**

**DC-DLTS peak height (a. u.)**

**1.8**

**2.1**

**0.0 0.3 0.6 0.9 1.2**

**2nd pulse duration (ms)**

**01234 2nd pulse voltage (V)**

trap *H*0 cannot be observed in Schottky junctions owing to the freeze-out effect.

Table 2. Summary of *E*t, p, and adjusted *N*t-adj of *H*0, *H*2, and *H*5.

**4.2.2 Radiative shallow recombination center H0** 

**30 35 150 300**

**Temperature (K)**

*H***2** *H***3**

Fig. 9. (a)DC-DLTS spectra of p-type GaAsN for various second pulse voltage and (b) *H*0 DC-DLTS peak height dependence of second pulse voltage and duration.

**3.00 V**

**1.75 V**

*H***0 DLTS**

**DC-DLTS Voltage of the 2nd pulse**

**0.0**

**0.5**

**1.0**

**DLTS signal**(**a.u**)

**1.5**

**2.0**

ratio NE1/NHP1 is less than 2. If *E*1 is the split interstitial (N-N)As, NHP1 must be at least equals to that of *E*1 and one N atom remains free. This expectation is disapproved by DLTS measurements, where *N*HP1 is largely less than *N*E1. This means that *E*1 contains only one N atom in its atomic structure. Considering the results of last sub-section, *E*1 may be the split interstitial (*N-As*)As formed from one *N* and one *As* in a single *As* site. This result is supported by the theoretical calculation (Zhang et al., 2001).

### **4.1.4 Effect of E1 on minority carrier lifetime in GaAsN**

The effect of *E*1 on the electrical properties of GaAsN can be evaluated through the calculation of minority carrier lifetime using the SRH model for generationrecombination (Hall, 1952; Shockley & Read, 1952). Such parameter has been estimated to be less than 0.2 ns as a result of the calculation according to

$$
\sigma\_{\varepsilon1} = \left(\upsilon\_{\rm thm} \sigma\_{\varepsilon1} N\_{\varepsilon1}\right)^{-1} < 0.2 \text{ ns} \tag{17}
$$

Therefore, E1 is considered to be the main cause of short minority carrier in GaAsN. It is required to investigate the formation mechanism of this defect in order to decrease its density and to recover the minority carrier lifetime in GaAs.

### **4.2 Hole traps in GaAsN grown by CBE**

### **4.2.1 DLTS spectra and properties of hole traps in GaAsN**

Here, we only focus on the hole traps that coexist in all *p*-type GaAsN based Schottky junctions and *n*+-GaAs/*p*-GaAsN heterojunction. The difference between these two structures is the temperature range in which the DLTS measurements can be carried out due to the freeze-out of carriers. The DLTS spectrum of *p*-type GaAsN in the heterojunction is shown in Fig. 8(a). Three hole traps *H*0, *H*2, and *H*5 are observed at 0.052, 0.185, and 0.662 eV above the VBM of GaAsN. Their peak temperatures are 35, 130, and 300 K, respectively. The thermal dependence of emission from the hole traps is plotted as an Arrhenius plot in Fig. 8(b). The activation energy, capture cross section, and density are given in Table 2.

Fig. 8. (a) DLTS spectra of p-type GaAsN grown by CBE and (b) their Arrhenius plots.

The hole traps *H*2 and *H*5 were also observed in Schottky junctions and coexist in all samples. *H*2 is a N-related acceptor-like state and was proved to be in good relationship with high background doping in GaAsN films. These properties will be discussed later. The


hole trap *H*5 presents the same properties as *H*3 and *HA*5 (Li et al., 2001; Katsuhata et al., 1978). It is proposed to be the double donor state (+/++) of *EL*2 (Bouzazi et al., 2010). The hole trap *H*0 cannot be observed in Schottky junctions owing to the freeze-out effect.

Table 2. Summary of *E*t, p, and adjusted *N*t-adj of *H*0, *H*2, and *H*5.
