Investigation of Lattice Defects in GaAsN Grown by Chemical Beam Epitaxy Using Deep Level Transient Spectroscopy

With only 3 % of N and 9 % of In, InGaAsN with a band gap of 1.04 eV was obtained and could be lattice matched to GaAs and Ge. This dilute nitride semiconductor has been selected as a promising candidate for high efficiency multijunction tandem solar cells (Geisz and Friedman, 2002). However, the diffusion length of minority carriers and the mobility are still lower than of that in GaAs or InGaAs and showed a considerable degradation with increasing the N concentration. These electrical properties are insufficient to insure the current matching in the multijunction solar cell structure AlInGaP/GaAs/InGaAsN/Ge (Friedman et al., 1998). An obvious reason of such degradation is the high density of Nrelated lattice defects that can be formed during growth to compensate for the tensile strain caused by the small atomic size of N compared with that of arsenic (As) and to the large miscibility of the gap between GaAs and GaN. These defect centers are expected to act as active recombination and/or scattering centers in the forbidden gap of the alloy (Zhang & Wei, 2001). However, no experimental evidence has yet been reported. On the other hand, the conductivity of undoped p-type InGaAsN or GaAsN and their high background doping (Friedman et al., 1998; Kurtz et al., 1999; Moto et al., 2000; Krispin et al., 2000) prevent the design of wide depletion region single junction solar cell and the fabrication of intrinsic layer to overcome the short minority carrier lifetime. This serious problem was expected in the first stage to the density of unintentional carbon in the film (Friedman et al., 1998; Kurtz et al., 1999; Moto et al., 2000). However, the carrier density in some InGaAsN semiconductors was found to be higher than that of carbon (Kurtz et al., 2002). Furthermore, the high density of hydrogen (up to 1020 cm−3) and the strong interaction between N and H in InGaAsN to form N-H related complex were confirmed to be the main cause of high background doping in InGaAsN films (Li et al., 1999; Janotti et al., 2002, 2003; Kurtz et al., 2001, 2003; Nishimura et al., 2007). In addition, N-H complex was found theoretically to bind strongly to gallium vacancies (VGa) to form N-H-VGa with a formation energy of 2 eV less than that of isolated VGa (Janotti et al., 2003). These predictions were supported experimentally using positron annihilation spectroscopy results (Toivonen et al., 2003).


Introduction
With only 3 % of N and 9 % of In, InGaAsN with a band gap of 1.04 eV was obtained and could be lattice matched to GaAs and Ge. This dilute nitride semiconductor has been selected as a promising candidate for high efficiency multijunction tandem solar cells . However, the diffusion length of minority carriers and the mobility are still lower than of that in GaAs or InGaAs and showed a considerable degradation with increasing the N concentration. These electrical properties are insufficient to insure the current matching in the multijunction solar cell structure AlInGaP/GaAs/InGaAsN/Ge ). An obvious reason of such degradation is the high density of Nrelated lattice defects that can be formed during growth to compensate for the tensile strain caused by the small atomic size of N compared with that of arsenic (As) and to the large miscibility of the gap between GaAs and GaN. These defect centers are expected to act as active recombination and/or scattering centers in the forbidden gap of the alloy (Zhang & Wei, 2001). However, no experimental evidence has yet been reported. On the other hand, the conductivity of undoped p-type InGaAsN or GaAsN and their high background doping Kurtz et al., 1999;Moto et al., 2000;Krispin et al., 2000) prevent the design of wide depletion region single junction solar cell and the fabrication of intrinsic layer to overcome the short minority carrier lifetime. This serious problem was expected in the first stage to the density of unintentional carbon in the film Kurtz et al., 1999;Moto et al., 2000). However, the carrier density in some InGaAsN semiconductors was found to be higher than that of carbon (Kurtz et al., 2002). Furthermore, the high density of hydrogen (up to 10 20 cm −3 ) and the strong interaction between N and H in InGaAsN to form N-H related complex were confirmed to be the main cause of high background doping in InGaAsN films (Li et al., 1999;Janotti et al., 2002Janotti et al., , 2003Kurtz et al., 2001Kurtz et al., , 2003Nishimura et al., 2007). In addition, N-H complex was found theoretically to bind strongly to gallium vacancies (V Ga ) to form N-H-V Ga with a formation energy of 2 eV less than that of isolated V Ga (Janotti et al., 2003). These predictions were supported experimentally using positron annihilation spectroscopy results (Toivonen et al., 2003).
On the other hand, similar electrical properties were obtained in InGaAsN grown by metalorganic chemical vapor deposition (MOCVD) and molecular beam epitaxy (MBE) despite the large difference in the density of residual impurities, which excludes them as a main cause of low mobilities and short minority carrier lifetimes. For that, lattice defects, essentially related to the N atom, were expected to be the main reason of such degradation. Several theoretical and experimental studies have investigated carrier traps in InGaAsN films. Theoretically, using the first principles pseudo-potential method in local density approximation, four N-related defects were proposed: (As Ga -N As ) nn , (V Ga -N As ) nn , (N-N) As , and (N-As) As ( Zhang & Wei, 2001). While the two first structures were supposed to have lower formation probabilities, the two split interstitials (N-N) As , and (N-As) As were suggested to compensate the tensile strain in the film and to create two electron traps at around 0.42 and 0.66 eV below the conduction band minimum (CBM) of InGaAsN with a band gap of 1.04 eV, respectively (Zhang & Wei, 2001). Experimentally, the ion beam analysis provided a quantitative evidence of existence of N-related interstitial defects in GaAsN Ahlgren et al., 2002;Jock, 2009). Furthermore, several carrier traps were observed in GaAsN and InGaAsN using deep level transient spectroscopy (DLTS). A deep level (E2/H1), acting as both an electron and a hole trap at 0.36 eV below the CBM, was observed (Krispin et al., 2001). Other electron traps in GaAsN grown by MBE were recorded: A2 at 0.29 eV and B1 at 0.27 eV below the CBM of the alloy (Krispin et al., 2003). In addition, a well known electron trap at 0.2  0.3 eV and 0.3  0.4 eV below the CBM of p-type and n-type GaAsN grown by MOCVD were observed, respectively (Johnston et al., 2006). Although the importance of these results as a basic knowledge about lattice defects in GaAsN and InGaAsN, no recombination center was yet experimentally proved and characterized. Furthermore, the main cause of high background doping in p-type films was not completely revealed. Chemical beam epitaxy (CBE) has been deployed (Yamaguchi et al., 1994;Lee et al., 2005) to grow (In)GaAsN in order to overcome the disadvantages of MOCVD and MBE. It combines the use of metal-organic gas sources and the beam nature of MBE. (In)GaAsN films were grown under low pressure and low temperature to reduce the density of residual impurities and to avoid the compositional fluctuation of N, respectively. Furthermore, a chemical N compound source was used to avoid the damage of N species from N 2 plasma source in MBE. Although we obtained high quality GaAsN films gown by CBE, the diffusion length of minority carriers is still short (Bouzazi et al., 2010). This indicates that the electrical properties of GaAsN and InGaAsN films are independent of growth method and the problem may be caused by the lattice defects caused by N. Therefore, it is necessary to investigate these defects and their impact on the electrical properties of the film. For that, this chapter summarizes our recent results concerning lattice defects in GaAsN grown by CBE. Three defect centers were newly obtained and characterized. The first one is an active non-radiative N-related recombination center which expected to be the main cause of short minority carrier lifetime. The second lattice defect is a N-related acceptor like-state which greatly contributes in the background doping of p-type films. The last one is a shallow radiative recombination center acceptor-like state.

Deep level transient spectroscopy
To characterize lattice defects in a semiconductor, several techniques were used during the second half of the last century. Between these methods, we cite the thermally stimulated current (TSC) (Leonard & Grossweiner, 1958;Bube, 1960), the admittance spectroscopy (Losee, 1974), the increase or decay curves of photoconductivity (Rose, 1951;Devore, 1959), the optically stimulated conductivity (Lambe, 1955;Bube, 1956), and the analysis of spacecharge-limited currents as function of applied voltage (Smith & Rose, 1955;Rose, 1955;Lampert, 1956). The basic concept of using the change of capacitance under bias conditions by the filling and emptying of deep levels was already anticipated fifty years ago (Williams, 1966). The thermally stimulated capacitance (TSCAP), which gives the temperature dependence of junction capacitance (Sah et al., 1978;Sah & Walker, 1973), was used. By dressing the properties of all these techniques, D. V. Lang found that they lacked the sensitivity, the speed, the depth range of recorded trap, and the spectroscopic nature to make them practical for doing spectroscopy on non-radiative centers. For that, in 1974, he proposed DLTS as a characterization method of lattice defects that can overcome the disadvantages of the other methods (Lang, 1974). DLTS is based on the analysis of the change of capacitance due to a change in bias condition at different temperatures. It can be applied to Schottky contacts and p-n junctions. DLTS has advantages over TSC due to its better immunity to noise and surface channel leakage currents. It can distinguish between majority and minority carrier traps, unlike TSC, and has a strong advantage over admittance spectroscopy, which is limited to majority-carrier traps. Comparing with TSCAP, DLTS has much greater range of observable trap depths and improved sensitivity. Despite the success of optical techniques such as photoluminescence to characterize superficial levels, they are rarely used in the study of non-radiative deep levels. Furthermore, such experiences must be done in the infrared domain. However, sensors are less sensitive than in the visible domain. Thus, we need a technique, which can separate between minority and majority traps and evaluate their concentrations, their energies, and their capture cross sections.

Fundamental concept of DLTS 2.1.1 Capacitance transient
To fully understand DLTS, it is worth to have a basic knowledge of capacitance transients arising from the SCR of Schottky contacts or p + -n/n + -p asymmetric junctions. If a pulse voltage is applied to one of these device structures that is originally reverse-biased, the SCR width decreases and the trap centers are filled with carriers (majority or minority depending on the structure). When the junction is returned to reverse bias condition, the traps that remains occupied with carriers are emptied by thermal emission and results in a transient decay. The capacitance transients provide information about these defect centers. Here, we restrain our description to a p + -n junction where the p-side is more much heavily doped than the n-side, which gives the SCR almost in the low doped side. The causes of change in capacitance depend on the nature of applied voltage. In case of reverse biased voltage, the junction capacitance, due to the change in SCR width, is dominant. However, when the applied voltage is forward biased, the diffusion capacitance, due to the contribution of minority carrier density, is dominant. The basic equation governing the capacitance transient in the p + -n junction is expressed by where A is the contact area, V b is the built-in potential,  0 is the permittivity of the semiconductor material, and e is the elementary charge of an electron. C 0 , N T , N D , and  www.intechopen.com denote the junction capacitance at reverse bias, the density of filled traps under steady state conditions, the ionized donor concentration, and the time constant that gives the emission rate, respectively. The change in capacitance after the recharging of traps is given by In most cases of using transient capacitance, the trap centers form only a small fraction of the SCR impurity density, i.e., N T << N D . Hence, using a first-order expansion of Eq.
(2) gives where W R is the total SCR at reverse bias voltage V R , L 1 = W R -, L 2 = W p -, and where W p , E F , and E T denote the SCR at V p , the Fermi level, and the trap energy level.

Thermal emission of carriers from deep levels
The emission rates for electrons and holes are given, respectively by where  n , N c , and v thn are the thermal capture cross section, the density of states, and the thermal velocity of holes, respectively.  p , N v , and v thp are the same parameters for holes. E CBM , E VBM , and E T are the energy levels of the conduction band minimum, the valence band maximum, and the trap, respectively.

Other DLTS related techniques
The isothermal capacitance transient spectroscopy (ICTS) and the double carrier pulse DLTS (DC-DLTS) are two DLTS related methods. They are used to obtain the density profiling of lattice defects and to check whether they act as recombination centers or not, respectively. www.intechopen.com

Isothermal capacitance transient spectroscopy
ICTS is used to analyze the profiling of lattice defects in the SCR of the semiconductor. It can be done through three different methods. The first one is obtained by fixing V R and varying V p to build difference of transients among the SCR of the device. The second method is evaluated by measuring at constant V p and varying V R . The last option is obtained by varying V R and V p , where the profiling analysis is also possible without building difference of transients. Using the first method, the medium trap density T ij Nx () at a point ij x is given by where C ij is the amplitude difference of the two capacitance transients.

Double carrier pulse DLTS
DC-DLTS is used in asymmetric n + -p or p + -n junctions (Khan et al., 2005). It aims to check whether a trap is a recombination center or not. As shown in Fig. 1, two pulsed biases are applied to the sample, in turn, to inject majority and minority carriers to an electron trap. At the initial state, the junction is under reverse bias, and the energy level E T of the trap is higher than the Fermi level (E Fn ).When the first pulse voltage is applied to the sample, E Fn is higher than E T , which allows the trap to capture electrons. During the second reverse biased pulse, with a duration t ip , holes are injected to the SCR from the pside of the junction. After the junction pulse is turned off, electrons and holes are thermally emitted. The amount of trapped carriers can be observed as a change in the DLTS peak height of the trap. If the trap captures both electrons and holes, the DLTS maximum of the corresponding level decreases compared with that in conventional DLTS. Such a decrease is explained by the electronhole (eh) recombination process, which indicates that the level is a recombination center. Fig. 1. Basic concept of capture and thermal emission processes from an electron trap located at an energy level E T in p + -n junction. A saturating injection pulse is applied to the reverse biased junction to fill the trap with holes.
To formulate the recombination process, we consider the same notation in § 2.1.2, with assuming that n  N D . The relationship between the total density of recombination centers and that only occupied by electrons in the n-side of the junction can be expressed by where p is the average of injected holes. As a solution of Eq. (10), we have where n T () = (pc p + e n )/( -1 N T ),  -1 = pc p + N D c n + e n + e p ), and t ip is the width of the injected pulse. Considering the I DLTS and I DC-DLTS the peak heights of the recombination center in conventional and DC-DLTS, respectively. Equation (11) can be rewritten properly as Similarly for hole trap in the p-side of a n + -p junction, which acts as recombination center, we obtain where n is the average of injected electrons.

Experimental procedure
All GaAsN films were grown by CBE on high conductive n-or p-type GaAs 2 off toward [010] substrate using Triethylgallium ((C 2 H 5 ) 3 Ga, TEGa), Trisdimethylaminoarsenic ([(CH 3 ) 2 N] 3 As, TDMAAs), and Monomethylhydrazine (CH 3 N 2 H 3 , MMHy) as Ga, As, and N sources, respectively. The flow rates TEGa = 0.1 sccm and TDMAAs = 1.0 sccm were considered as conventional values. The growth temperatures of 420 C and 460 C were used for p-type and n-type GaAsN, respectively. Concerning the doping, p-type GaAsN films are unintentionally doped. The n-type alloys were obtained using a silane (SiH 4 ) source or by growing the films under lower MMHy and high growth temperature. Three different device structures are used in this study: (i) n-and p-type GaAsN schottky contacts, (ii) n + -GaAs/p-GaAsN/p-GaAs, and (iii) n-GaAsN/p + -GaAs hetro-junctions. The N concentration in all GaAsN layers was evaluated using XRD method. Aluminum (Al) dots with a diameter of 0.5/1 mm were evaporated under vacuum on the surface of each sample. Alloys of Au-Ge (88:12 %) and Au-Zn (95:05 %) were deposited at the bottom of n-type and p-type GaAs substrates for each device, respectively. Some samples were treated by postthermal annealing under N 2 liquid gas and using GaAs cap layers to avoid As evaporation from the surface. The temperature and the time of annealing will be announced depending on the purpose of making annealing. The background doping and the doping profile in the extended depletion region under reverse bias condition were evaluated using the capacitance-voltage (C-V) method. The leakage current in all used samples ranged from 0.3 nA to 10 A for a maximum reverse bias voltage of -4 V. A digital DLTS system Bio-Rad DL8000 was used for DLTS and C-V measurements. The activation energy E t and the capture cross section  n,p were determined from the slope and the intercept values of the Arrhenius plot, respectively.

Lattice defects in GaAsN grown by CBE
In this section, the distribution of electron and hole traps in the depletion region of GaAsN grown by CBE will be dressed using DLTS and related methods.

Electron traps in GaAsN grown by CBE 4.1.1 DLTS spectra and properties of a N-related electron trap
The DLTS spectrum of Fig. 2(a) shows an electron trap (E2) at 0.69 eV below the CBM of GaAs. After rapid thermal annealing at 720C for 2 min, E2 disappears completely whereas a new electron trap (E3) appears at 0.34 eV below the CBM. From the Arrhenius plots of Fig. 2(d), the capture cross sections of E2 and E3 are calculated to be  E2 = 8.1 × 10 -15 cm 2 and  E3 = 7.5 × 10 -18 cm 2 , respectively. Based on previous results about native defects in n-type GaAs, E2 and E3 are independent of N and considered to be identical to EL2 and EL3, respectively (Reddy et al., 1996). In order to focus only on N-related lattice defects, these two energy levels will be excluded from the DLTS spectra of Ncontaining n-type GaAsN. The addition of a small atomic fraction of N to GaAs leads to the record of a new electron trap (E1), at an average activation energy 0.3 eV below the CBM of GaAsN. The DLTS spectra of as grown and annealed n-type GaAs 0.998 N 0.002 are given in Figs. 2. (b) and (c), respectively. The activation energies (E E1 ) and the capture cross sections ( E1 ) of E1 for N varying GaAsN samples are given Fig. 3 (a) and (b), respectively. The fluctuation of E E1 from one sample to another can be explained by the effect of PooleFrenkel emission, where the thermal emission from E1 is affected by the electric field (Johnston and Kurtz, 2006). As illustrated in Fig. 3(c), with increasing the filling pulse duration, the DLTS peak height of E1 saturates rapidly. This behavior is explained by the large value of  E1 , compared with that of E2, E3, and other native defects in GaAs. The adjusted densities of E1 (N E1 ) in as grown and annealed samples are plotted in Fig. 3(d). N E1 increases considerably with increasing [N] in the film and persists to post thermal annealing. This indicates that E1 is a N-related and a stable electron trap. The defect center E1 was not observed previously in N free GaAs grown by CBE despite the existence of N species in the chemical composition of the As source. This can be explained through three possible scenarios. First, the absence of tensile strain in GaAs prevents the formation of E1. Second, the N atom in the atomic structure of E1 comes from the N source. Finally, the N atom comes from the N and As compound sources and in presence of tensile strain E1 can be formed. The tensile strain was reported in most theoretical and experimental studies. As given in Fig. 3(e) and using the ICTS, this idea is supported by the uniform distribution of N E1 in the bulk of GaAsN. This indicates that E1 is formed during growth to compensate for the tensile strain in the GaAsN films caused by the small atomic size of N compared with that of As. Furthermore, the properties of E1 are identical to that of the famous electron traps reported by Johnston et Kurtz (Johnston and Kurtz, 2006) and Krispin et al. (Krispin et al., 2003) in MOCVD and MBE grown ntype GaAsN, respectively. As illustrated in Fig. 3(d), the densities of these traps are approximately similar to N E1 despite the large difference in the density of residual impurities between the three growth methods. Therefore, the atomic structure of E1 may be free from impurities. Furthermore, by carrying out DLTS measurements for minority carriers in undoped p-type film, E1 was also observed. This indicates clearly that E1 is independent of doping atoms (see Fig. 3(f)).

Nature of the electron trap E1
Two methods are used to verify whether E1 is a recombination center or not. The first method is indirect, in which the activation energy of deep levels is correlated with that of the reverse bias current in the depletion region of n-type GaAsN Schottky junction and n + -GaAs/p-GaAsN heterojunction. These two device structures are selected because the current is due mainly to electrons. The second method is direct, in which DC-DLTS is used to show the behavior of the electron traps in simultaneous injection of majority and minority carriers in the depletion region.

Origin of reverse bias current in GaAsN
The temperature dependence of the reverse bias current in the depletion region of n-type GaAsN Schottky junction and n + -GaAs/p-GaAsN is shown in Fig. 4(a) for reverse bias voltages of 0.5 and -0.5 V, respectively. At lower temperature, the dark current changes slowly in the two structures, then fellows an Arrhenius type behavior. As shown in Fig. 4(b), the same result is obtained by applying reverse bias voltages of 1 and -1 V. Under these conditions, the reverse bias current I d (T) can be expressed by where I  , E, k, and T denote the limit of the high-temperature current, the thermal activation energy of the reverse bias current, the Boltzmann constant, and the temperature, respectively. The I-V characteristics deviate in the two samples from the thermionic emission. This is explained by the fact that supplying the p-n junction under reverse bias conditions decreases the product of excess carriers to less than the square of intrinsic carriers. Hence, the Shockley-Read-Hall (SRH) generation mechanism is activated to increase the product of excess carriers to assure the balance of charge. The generated carriers are swept to the transition regions by the electric field in the depletion region. Therefore, an SRH center, with a thermal activation energy arround 0.3 eV, is the origin of the dark current in the SCR of GaAsN. The thermal activation energies are measured with respect to majority and minority carriers in n-type GaAsN schottky junction and n + -GaAs/p-GaAsN heterojunction, respectively. This corresponds to the conduction band in the two structures. By correlating the conduction mechanism and DLTS measurements, the thermal activation energy of the reverse bias current and the activation energy of the Nrelated electron trap E1 are typically identical. Therefore, E1 is responsible for the generation/recombination current in the depletion region of GaAsN grown by CBE.

DC-DLTS measurements
DC-DLTS is used to confirm the recombination nature of E1 and to characterize the recombination process via this defect center. An unintentionally doped ntype GaAsN layer ( 1 m) was grown on a ptype GaAs by CBE. This structure is not commonly used for DLTS measurements. However, the absence of a p-type doping source prevented us to obtaining a p + -n junction. Here, the p-type substrate is used as source of minority carriers. As shown in Fig. 5(a), the DC-DLTS spectrum is compared with that of the conventional DLTS. A decrease in the peak height of E1 is observed by varying the voltage of the second injected pulse and also confirmed by varying its duration. The obvious reason for such reduction is the mechanism of e-h recombination at the energy level E1 in the forbidden gap of GaAsN. Hence, E1 is reconfirmed to act as a N-related recombination center. To verify the nonradiative recombination process, the temperature dependence of  E1 is obtained by varying the emission rate window e rw from 0.5 -50 s -1 . The value of  E1 is obtained from the fitting of the Arrhenius plots for each e rw . As shown in Fig. 5(b), the natural logarithmic of  E1 shows a linear increase with the reciprocal of the temperature. It can be expressed as where E cap,e = 0.13  0.02 eV, k, T, and   = 1.38 x 10 -9 cm 2 denote the barrier height for the capture of electron, the Boltzmann constant, the temperature, and the capture cross section of www.intechopen.com electrons at an infinite temperature, respectively. At room temperature,  E1 (300 K) is evaluated to  8.89 × 10 -12 cm 2 . Such a value is large enough to shorten the lifetime of electrons in p-type GaAsN. This indicates that E1 is a strongly active recombination center at room temperature and the e-h recombination process is non-radiative. In addition, from the temperature dependence of  E1 , the true energy depth of E1 can be obtained by subtracting the barrier height for electron capture from the thermal activation energy obtained from the Arrhenius plot. The recombination center E1 is localized at E a (E1) = 0.20  0.02 eV from the CBM of GaAsN. Furthermore, the average capture cross section of holes  p , at a temperature of T = 175 K, is estimated using Eq. 12 to be  p (175 K)  5.01 × 10 -18 cm 2 . The physical parameters of E1 can be summarized in a configuration coordinate diagram (CCD), in which the energy state of E1 is described as a function of lattice configuration (Q). As shown in Fig. 5(

Possible origin of the N-related recombination center E1
It is worth remembering that the atomic structure of E1 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 E1 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 E1 has high probability to depend only on the atoms of the alloy (N, As, Ga). To confirm these expectations, the origin of E1 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.

Dependence of N E1 to As source flow rates
The objective of this experiment is to clarify whether the density of E1 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 e rw = {100, 10} s -1 and filling pulse t p = {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 E1 was observed in all samples. The 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 E1. If E1 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 E1 to As atom, either than N.

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 Amore & Filippone, 2005). H ions with multi-energy from 10 to 48 keV were implanted into GaAsN layers with peaks concentration of 5  10 18 (GaAsN HD1 ) and 1  10 19 atom/cm 3 (GaAsN HD2 ), 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 N 2 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, E1 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).  The new electron trap (EP1) is located approximately 0.41 eV below the CBM of GaAsN. Its properties are identical to that of the native defect EL5 in GaAs (Reddy, 1996). Its atomic structure was discussed in many publications, where the common result indicated that EL5 is a complex defect free from impurities and dominated by As interstitials, such V Ga -As i or As Ga -V Ga (Deenapanray et al., 2000;Yakimova et al., 1993). The second new defect is a hole trap (HP1) 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 HP1 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 HC2 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 (10 19 cm -3 ). It is also much higher than N HP1 . Second, N atom can be originated from the complete dissociation of E1, since the ratio N E1 /N HP1 is less than 2. If E1 is the split interstitial (N-N) As , N HP1 must be at least equals to that of E1 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 E1 contains only one N atom in its atomic structure. Considering the results of last sub-section, E1 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).

Effect of E1 on minority carrier lifetime in GaAsN
The effect of E1 on the electrical properties of GaAsN can be evaluated through the calculation of minority carrier lifetime using the SRH model for generationrecombination (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 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.

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 H0, H2, and H5 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. The hole traps H2 and H5 were also observed in Schottky junctions and coexist in all samples. H2 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 www.intechopen.com hole trap H5 presents the same properties as H3 and HA5 (Li et al., 2001;Katsuhata et al., 1978). It is proposed to be the double donor state ( + / ++ ) of EL2 (Bouzazi et al., 2010). The hole trap H0 cannot be observed in Schottky junctions owing to the freeze-out effect.

Radiative shallow recombination center H0
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 H0 is observed and confirmed by varying the voltage and the duration of the injected pulse. 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 e rw from 1 to 50 s -1 . As shown in Fig. 10(a), the peak temperature of H0 shifts to high temperatures with increasing e rw . The value of  H0 is obtained from the fitting of the Arrhenius plots for each e rw . 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 H0 does not exhibit an Arrhenius behavior, which excludes the non-radiative recombination process. Its shallow energy level suggests that H0 plays the role of an intermediate center in the recombination process, with the exception that the recombination is quite often radiative. Furthermore, the capture cross section of electrons can be estimated using Eq. 13 and by the reduction of the peak height of H0, which follows from the injection of minority carriers. By varying the injected pulse voltage at fixed duration, the average capture cross section of electrons  n , at a temperature T = 35 K, is estimated to be  n  3.64 × 10 -16 cm 2 . However, by varying the width of the injected pulse at fixed pulse voltage, it is estimated, at the same temperature, to be  n  3.05 × 10 -16 cm 2 . These two values are nearly identical and indicate that the capture cross section of electrons and holes of H0 are approximately the same. To identify this radiative recombination, photoluminescence (PL) measurements were carried out at low temperature on the same GaAsN sample. The PL spectra at 20, 30, 40, and 50K are shown in Fig. 10 (c). Three different peaks P1, P2, and P3 can be distinguished from PL spectra fitting. The temperature dependence of their energies is plotted in Fig. 10 (d). The three peaks were sufficiently discussed in many N-varying GaAsN samples and they are proposed to be: (i) P1 is the result of band-to-band transition, (ii) P2 is caused by free exciton or related to shallow energy level, and (iii) P3 originates from the transition between a neutral donor and a neutral acceptor pair (DAP) (Inagaki et al., 2011). Therefore, the peak P2 is suggested to be in relation with H0. The band diagram of such recombination is shown in Fig. 11, where the transition occurs between H0 and the CBM and/or a donor-like defect (E t ). The transition between free electrons in the CBM and charged H0 is called free-tocharged bound transition. Fig. 11. Band diagram of radiative recombination through H0, where the transition occurs between H0 and the CBM and/or a donor-like defect (E t ).
Concerning the origin of the acceptor-like hole trap H0, more experiments are required to discuss it. However, considering its density and the distribution of shallow acceptors in GaAs, it can be suggested that H0 is a carbon-related acceptor, where the reported ionization energy from PL and Hall Effect measurements is 0.026 eV and its density in the 10 -15 cm -3 range (Baldereschi & Lipari, 1974).

Deep N-H related acceptor state H2
The ionized acceptor density (N A ) is found to be in good linear dependence with N concentration in p-type GaAsN samples (see Fig. 12 (a)). As given in Figs. 12(b) and (c), the junction capacitance (C j ) showed a N-related sigmoid behavior with temperature in the range 70 to 100 K. This behavior has not yet been observed in GaAs and n-type GaAsN grown by CBE. It was recorded at 20 K in silicon p-n junction and explained by the ionization of a shallow energy level (Katsuhata, 1978;1983). Hence, the N dependence of N A and C j is explained by the thermal ionization of a N-related acceptor-like state. The thermal ionization energy of this energy level was estimated in the temperature range 70 to 100 K to be between 0.1 and 0.2 eV. It is in conformity with the theoretical calculations, which suggested the existence of a N-related hole trap acceptor-like defect with an activation energy within 0.03 and 0.18eV above the VBM of GaAsN (Janotti et al., 2003;Suzuki et al., 2008). Experimentally, a deep acceptor level, A2, was confirmed in CBE grown undoped GaAsN with ionization energies of E A1 = 130 ± 20 meV . On the other hand, the properties of H2 in N-varying GaAsN schottky junctions are cited below: The peak temperature of H2 is within the temperature range of increase of C j . This means that the electric field at this temperature range fellows the same behavior of C j and depends on N concentration. Hence, the emission of carriers from the charged traps is affect by the Poole-Frenkel emission (Johnston and Kurtz, 2006). This is confirmed by the fluctuation of E H2 from one sample to another depending on N concentration (see Table 3). However, the average of E H2 is within the energy range of the acceptor level obtained from theoretical prediction and identical to E A2 Janotti et al., 2003). Furthermore, as given in Fig. 12 (d), N H2-adj is in linear dependence with N concentration. Therefore, H2 is proved to be the N-related hole trap acceptor-like state, which thermal ionization increased C j and drops the depletion region width. The contribution of H2 in the background doping of ptype GaAsN films grown by CBE can be evaluated from the ratio between the real N H2-Cj calculated from the change of C j and N A at room temperature (Bouzazi et al., 2010). As shown in Fig. 12 (e) and (f), This ratio comes closer to the unit for a N concentration greater than 0.2%. Thus, H2 is the main cause of the high background doping in p-type GaAsN. To investigate the origin of H2, it is worth remembering some previous results about carrier concentration and the density of residual impurities in undoped GaAsN grown by CBE, obtained using Hall Effect, Fourier transform infra-red (FTIR), and second ion mass spectroscopy (SIMS) measurements. On one hand, under lower Ga flow rates (TEGa), N A and N H2 sowed a rapid saturation with [N], despite the increase of [N] (see Fig. 13 (a)). This means that the atomic structure of H2 depends on other atoms, either than N. In addition, the densities of C and O was found to be less than free hole concentration, which excludes these two atoms from the origin of H2. On the other hand, using SIMS measurements, the ratio [H] TEGa = 0.02 /[H] TEGa = 0.1 was evaluated to be 0. 6 (Sato et al., 2008). Furthermore, the free hole concentration at room temperature showed a linear increase with the density of N-H bonds (Nishimura et al., 2006). This means that N A depends strongly on [H] and the saturation of N A under lower TEGa can be explained by the desorption of H from the growth surface, since the growth rate in our films was found to be in linear dependence with TEGa. Hence, the structure of H2 is related to the N-H bond. However, the N-H bond may not be the exact structure of H2 because the slope of the linear relationship between N A and [N-H] increased with increasing growth temperature (T G  [400, 430] C). This indicates that N A is determined by both the number of N-H and another unknown defect, which concentration increased with increasing T G . The binding energy of this unknown defect can be determined from Arrhenius plot. Furthermore, the formation energy of (N-H-V Ga ) -2 was found to be lower than (N-V Ga ) -3 , (H-V Ga ) -2 , and isolated V Ga -3 (Janotti et al., 2003). This means that the unknown defect may be V Ga . These predictions were experimentally supported using positron annihilation spectroscopy results (Toivonen et al., 2003). Hence, H2 may be related to the N-H-V Ga structure. [N](%) E H2 (eV)  H2 (cm 2 ) N H2-adj (cm -3 ) E max (V/cm) N H2-Cj (cm -3 )N H2-Cj / N A 0.  Table 3. Summary of E H2 ,  H2 , N H2-adj , E max , N H2-est , and the ratio N H2-Cj /N A for CBE grown undoped p-type GaAsN Schottky junctions.

Conclusion
Three defect centers, related to the optoelectronic properties of GaAsN, were identified and characterized using DLTS and some related methods: -The first defect is a N-related non-radiative recombination center (E1), located approximately 0.33 eV below the CBM of GaAsN grown by CBE. E1 is a stable defect and exhibits a large capture cross section at room temperature. Its activation energy comes closer to the midgap by increasing the N concentration in the film, which increases the recombination rates of carriers. In addition, the density profiling of E1 was found to be uniformly distributed, which may indicates that this defect was formed during growth to compensate for the tensile strain caused by N. Using the SRH model for generation-recombination, the lifetime of electrons to E1 was evaluated to be 0.2 ns. Therefore, E1 is suggested to be the main cause of poor minority carrier lifetime in GaAsN films. Its origin was suggested to be the split interstitial formed from one N and one As in a single arsenic site. - The second defect is a radiative shallow hole trap acceptor-like state (H0), at 0.052 eV above the VBM of GaAsN. it was observed and reported for the first time in GaAs and GaAsN films. The radiative recombination process through H0 is confirmed to be a freeto-charged bound transition. This defect may be in relationship with carrier density and the optical properties of the film. - The last defect is a N-related hole trap acceptor-like state (H2) located approximately 0.15 eV above the VBM of GaAsN. It was newly observed in GaAsN films. H2 was proved to have a good relationship with the carrier density and to be the main origin of high background doping in GaAsN films, essentially for relatively high N concentration. H2 is strongly suggested to be related with N-H bond. Its formation limits the junction depth and minimizes the contribution of the depletion region in the quantum efficiency of GaAsN based solar cells. In conclusion, the results obtained in this study are very useful for scientific understanding of defects in III-V-N materials and to improve GaAsN and InGaAsN qualities for realizing high efficiency multi-junction solar cells.

Acknowledgment
Part of this work was supported by the New Energy Development Organization (NEDO) under the Ministry of Economy, Trade and Industry, Japan.