Covalent bonding and JJ mixing effects on the EPR parameters of Er3 + ions in GaN crystal
Chai Rui-Peng†, , Li Long, Liang Liang, Pang Qing
Department of Physics, Xi'an University of Architecture and Technology, Xi'an 710055, China

 

† Corresponding author. E-mail: chairuipeng2005@163.com

Project supported by the Foundation of Education Department of Shaanxi Province, China (Grant No. 16JK1461).

Abstract
Abstract

The EPR parameters of trivalent Er3+ ions doped in hexagonal GaN crystal have been studied by diagonalizing the 364×364 complete energy matrices. The results indicate that the resonance ground states may be derived from the Kramers doublet Γ6. The EPR g-factors may be ascribed to the stronger covalent bonding and nephelauxetic effects compared with other rare-earth doped complexes, as a result of the mismatch of ionic radii of the impurity Er3+ ion and the replaced Ga3+ ion apart from the intrinsic covalency of host GaN. Furthermore, the JJ mixing effects on the EPR parameters from the high-lying manifolds have been evaluated. It is found that the dominant JJ mixing contribution is from the manifold 2K15/2, which accounts for about 2.5%. The next important JJ contribution arises from the crystal–field mixture between the ground state 4I15/2 and the first excited state 4I13/2, and is usually less than 0.2%. The contributions from the rest states may be ignored.

1. Introduction

In recent years much attention has been paid to the investigation of rare-earth (RE) doped GaN semiconductors, due to their potential in micro- and optoelectronics devices applications.[111] Of particular interest has been the electrical and optical properties of trivalent Er3+ in GaN host,[511] since the 1.54 μm luminescence center arising from the 4I13/24I15/2 transition is in the region of minimal loss in optical fibers.

It is well known that the open 4f shell of the trivalent 4f11 ion Er3+ is spatially protected from the 5s and 5p shells due to a greater radial extent. Accordingly, the central RE ions are less affected by the surrounding ligand ions when doped in a certain host, thus resulting in a larger spin–orbit interaction than the crystal field if compared with 3dN transition-metal complexes.[1216] However, the case may be different for the RE doping from one host to another because of the distinct features between the central ion and the ligand ion, e.g., ionic radius, charge, mass, etc.

To gain a better insight into the local structures and optical properties of the RE materials, electronic paramagnetic resonance (EPR) is regarded as a suitable tool to investigate the interrelation between the electronic structure and molecular structure. For instance, Palczewska et al. has performed the EPR measurement on Er3+-doped single GaN crystal, and the corresponding EPR parameters g = 2.861 ± 0.003, g = 7.645 ± 0.003, A = (110 ± 5) × 10−4 cm−1, and A = (290 ± 5) × 10−4 cm−1 were obtained.[17] It should be noted that a rule of average g-factor gav = 1/3(g + 2g), obtained from experiment, may be applied to predict the assignment of the lowest Kramers doublet by comparing with the theoretical one.[1820] Namely, if the average g-factor gav ≈ 6.0, where the contributions from all high excited states are ignored and only the pure 4I15/2 manifold is taken into account, the ground resonance states may be assigned to Kramers doublet Γ7 in Td symmetry, while the lowest states may be attributed to Γ6 if gav ≈ 6.8 for the same site symmetry.[21] The rule is appropriate for an axial C3v or C4v symmetry.[22,23] Consequently, Palczewska et al. assigned the ground states to Γ7 for the tetrahedral Er3+ center in GaN epilayers based on the fact that the experimental average g-factor (gav ≈ 6.05) is approximately equal to the expected gav of the doublet Γ7.[17] In order to account for these EPR parameters, Maâlej et al. recently carried out a theoretical study on the EPR parameters of the Er3+ ions in GaN crystal by means of the first-order perturbation method,[24] where only the ground manifold 4I15/2 is included. They also claimed that the lowest Kramers doublet belongs to the Γ7 based on the calculated average g-factor gav ≈ 6.14. However, to the best of our knowledge, the experimental gav of the most Er3+-doped complexes is generally less than the expected value 6.0 or 6.8 for the ground Kramers doublet Γ7 or Γ6, respectively.[19] Hence, the average g-factor gav for the hexagonal Er3+ center in GaN seems to be somewhat abnormal if the ground doublet is Γ7. For that matter, it is necessary to perform a detailed theoretical analysis on the EPR parameters of GaN:Er3+ system to examine the actual assignment associated with the ground Kramers doublet.

In this work, we will perform a quantitative theoretical study on the JJ mixing effects, arising from the high-lying excited states (such as the terms 4I13/2, 2K15/2, 2L15/2, etc.), on the EPR parameters based on a complete energy matrix method, where all excited states are included. Furthermore, the relation between the EPR g-factors and the orbit reduction factor k reflecting the covalent bonding and nephelauxetic effects is studied systematically. The plausible abnormal g-factors particularly for the average g-tensor gav of the Er3+ ion in GaN are reasonably expounded.

2. Theoretical method

In general, the perturbation Hamiltonian for a 4f11 configuration ion Er3+ in an axial crystal field can be expressed as[2529]

where the former six terms on the right represent the spherically symmetric free-ion parts, and the latter term represents the crystal field potential function. The operators and parameters appearing in Eq. (1) are written and defined according to the conventional meaning.[29] The photoluminescence excitation studies suggest that the Er3+ ion occupies the substitutional Ga3+ site with approximate C3v site symmetry when doped in the hexagonal GaN host,[5] and the local structure of the studied system has been plotted in Fig. 1. The crystal field potential for the trigonal C3v symmetry is as follows:

To study the EPR parameters without spoiling the overall coupling of the various interactions, the actual Zeeman operator has been included in Eq. (1) instead of the approximate operator-equivalent technique, when it is subjected to an external magnetic field H. The general matrix elements of the Zeeman operator ĤZEE in the intermediate coupling scheme |γSLJMJ⟩ may be expressed as[30]

Likewise, the magnetic hyperfine interaction ĤHF, which arises from the interaction between the unfilled electrons and nucleus, may be written as

where , β and βN are the Bohr magneton and nuclear magneton, respectively; μN is the value of the nuclear magnetic moment in nuclear magnetons, and I represents the nuclear spin.[31] The constants required for the following calculations are I = 7/2, μN = −0.5647, and ⟨ r−3⟩ = 10.6 a.u. for the 167Er3+ isotope.[32] By recoupling these angular momenta, the Ni may be expressed as

where the sum runs over all the unfilled electrons. The EPR parameters of the trigonal Er3+ center in GaN can be studied by using spin Hamiltonian[33]

where S = 1/2 is taken up. By now, the correlation between the EPR parameters and the perturbation Hamiltonian has been established by complete energy matrices.

Fig. 1. Local structure of hexagonal (GaN4)9− cluster in the GaN crystal. The central cation Ga3+ is replaced by the Er3+ ion with approximate C3v symmetry when Er3+ is embedded in GaN host.
3. Calculations and discussion

In this work, a refined least-squares fitting method, rather than the regular fitting routine, has been developed by simulating the EPR parameters and optical spectra simultaneously, which may reflect the macro- and microscopic optical properties of the GaN:Er3+ system, respectively. In order to reduce the number of adjustable parameters in our calculations, all free-ion parameters except the spin–orbit coupling parameter ζ are taken as the same values as the free-ion parameters of Er3+ obtained by Carnall et al,[34] in view of the fact that the small change of free-ion parameters only affects the displacement of the centre of gravity of various manifolds, and hardly affects the Stark splitting and the EPR parameters. The obtained crystal–field parameters together with the free-ion parameters are tabulated in Table 1. The comparison between the theoretical values and experimental values of the Stark energy levels of the manifolds 4I15/2 and 4I11/2 is listed in Table 2 by diagonalizing the 364×364 complete energy matrices. The calculated and experimental values of the EPR parameters including the g-factors g, g, and hyperfine structure constants A and A are projected in the Table 3. From Table 2 it can be found that the calculated Stark splittings are in good agreement with the observed values, though the third level (No. Z3) of 4I15/2 and the second level (No. A2) of 4I11/2 show a slightly large deviation (but not more than 5.0 cm−1). The errors can be expected because the spin-correlated crystal–field interaction of perturbation Hamiltonian has been ignored considering the absence of the complete optical spectra.[35,36] However, there are moderate deviations for the EPR g-factors as shown in Table 3, if the covalent bonding and nephelauxetic effects characterized by an average orbital reduction factor k are not taken into account, viz., k = 1. Just as pointed out by Maâlej et al.,[24] the JJ mixing effects between the ground states and excited states (e.g., 4I13/2, 2K15/2, 2L15/2) by the coupling of the crystal–field, spin–orbit, and the orbital angular momentum interactions should be evaluated in order to get some insight into the actual effects from the high-lying manifolds. For this purpose, the eigenfunctions of the lowest Kramers doublet comprised of the ground manifold 4I15/2 and the predominant excited manifolds 4I13/2, 2K15/2, and 2L15/2 are obtained and tabulated in Table 4. Note that the coefficients for each state satisfy the following relation: if one state of the Kramers doublet is expressed as

then its Kramers conjugate state is

Table 1.

The free-ion parameters and crystal–field parameters for Er3+ in GaN crystal. The values of all parameters are expressed in units of cm−1.

.
Table 2.

The calculated and experimental Stark levels of the hexagonal Er3+ centres in GaN.

.
Table 3.

The theoretical and experimental EPR parameters for the Er3+ centers in single GaN crystal. The hyperfine structure parameters A and A are in units of 10−4 cm−1.

.
Table 4.

The eigenfunctions of the lowest Kramers doublet for Er3+ ions in GaN crystal.

.

With the aid of the operator-equivalent technique developed by Stevens et al.,[37] the JJ mixing effects on the EPR parameters from the manifolds 4I15/2, 4I13/2, 2K15/2, and 2L15/2 have been evaluated and collected in Table 3 (lines 1–5). It is shown from Table 3 that the leading JJ contribution to the g-factors, in addition to the ground 4I15/2, arises from the excited manifold 2K15/2, which accounts for about 2.5%. However, the second important contribution comes from the crystal–field mixing effect between the ground 4I15/2 and the first excited state 4I13/2, where the ratio is less than 0.2% and the JJ contribution to g is nearly twice larger than that to g. The next JJ mixing contribution is from the state 2L15/2 which merely accounts for about 0.023% for g and 0.022% for g, respectively. Moreover, the contribution from the first excited state 4I13/2 to the EPR parameters is least compared with the above terms, which may be ascribed to the weaker crystal–field strength than the spin–orbit interaction for the RE complexes. The JJ contribution to the g-factors from the rest states, such as the terms 4I11/2, 4I9/2, 4F9/2, and 2K13/2, may be almost negligible. The case for the hyperfine structure constants A and A is similar to the EPR g-factors. It follows that the JJ mixing effects from the high manifolds indeed cannot contribute more to the final EPR parameters. Furthermore, we consider that the lowest resonance states may originate from the Kramers doublet Γ6 based on the evaluated average g-tensor gav ≈ 6.76 when the orbit reduction factor k is taken as 1.0. However, the noticeable deviation of the calculated g-factors from the measured values needs to be expounded if the above conjecture on the ground state assignment is reasonable.

Early studies have indicated that trivalent erbium ion Er3+ will occupy the position of Ga3+ site with local C3v symmetry when implanted in GaN host.[5] However, there is considerable difference of the ionic radii for the impurity ion Er3+ and the replaced cation Ga3+ of host (rEr3+ ≈ 0.89 Å, rGa3+ ≈ 0.62 Å). More overlap of electronic cloud between the central Er3+ ion and the ligand ions N3 − in the GaN:Er3+ system can be expected as a result of the larger ionic radius of impurity Er3+, and thus produce a stronger covalent bonding effect to some extent apart from the dominant covalency of host GaN. In the paper, in order to evaluate the covalent and nephelauxetic effects characterized by an average orbital reduction factor k as a whole,[38,39] the calculated results of the EPR g-factors (g, g, gav) as a function of the orbital reduction factor k are obtained and collected in Table 5, and the dependences of EPR g-factors are also plotted in Fig. 2. As shown by Fig. 2, the EPR g-factors are very sensitive to the orbital reduction factor k. The g-factors will increase linearly along with the increase of the factor k. Furthermore, we find that the agreement of the calculated g-factors may be further improved if the orbital reduction factor k is taken as 0.84, which reflect the stronger covalent bonding effect as expected. In particular, as for the average g-tensor gav, the theoretical value is in accord with the experimental value. At the same time, it is of interest to note that the calculated hyperfine structure parameter is in excellent agreement with the experimental value if the same reduction factor k = 0.84 is used, whereas the calculated A = 130.09 × 10−4 cm−1 when the orbital reduction effect is not taken into account. It is concluded that the stronger covalent bonding and nephelauxetic effects may be mainly responsible for the large deviation of the gav from the predicted value because of the mismatch of ionic radii. Our recent studies on the Er3+ ion implanted in ScVO4 crystal also show that there is more covalent effect on the EPR parameters if the ionic radius of central ion Er3+ is bigger than the substituted ion of host, where a comparable orbit reduction factor k ≈ 0.826 has been obtained based on a more detailed theoretical analysis.[27] Of course, to evaluate the effect of the hybridization of 4f states with the states of N3 − ions on the EPR parameters with care, more theoretical studies based on the first-principle calculations are needed. However, we do not mean to undertake it in the present study, since the mechanism controlling the atomic rearrangement has not been established, and the overall optical spectra data have not been reported yet, when the Er3+ is doped in single GaN crystal.

Table 5.

The EPR g-factors (g, g, gav) as a function of the orbital reduction factor k for Er3+ ions doped in GaN.

.
Fig. 2. Dependence of the EPR g-factors (g, g, gav) on the orbital reduction factor k for the GaN: Er3+ system.
4. Conclusion

A complete theoretical method has been established by diagonalizing the 364×364 energy matrix to study the EPR parameters of the trivalent Er3+ ions doped in hexagonal GaN crystal. The results show that the ground states may be derived from the Kramers doublet Γ6 for the GaN: Er3+ system, and the strong covalent bonding and nephelauxetic effects, as a result of the mismatch of ionic radii of the impurity Er3+ and replaced Ga3+ ion, may be mainly responsible for the larger deviation of the average EPR g-tensor gav from the theoretical value (gav ≈ 6.8) as expected. Moreover, the JJ mixing contribution to the EPR parameters from the high-lying excited states has been evaluated. Differing from the previous studies, it is found that the predominant JJ mixing contribution is the state 2K15/2, which accounts for about 2.5%, and the second major JJ contribution arises from the crystal–field mixture between the ground state 4I15/2 and the first excited state 4I13/2, and is usually less than 0.2%. The next JJ mixing contribution is from the state 2L15/2 which merely accounts for about 0.023% for g and 0.022% for g, respectively. The contributions from the rest states such as the 4I13/2, 4I11/2, 4I9/2, 4F9/2, and 2K13/2 may be ignored. It follows that the JJ mixing effects from the high manifolds could not really add more contribution to the final EPR parameters, which may be due to the larger spin–orbit coupling interaction than the crystal–field interaction for the RE-doped complexes.

Reference
1Gruber J BBurdick G WWoodward N TDierolf VChandra SSardar D K 2011 J. Appl. Phys. 110 043109
2Dammak MKammoun SMaalej RKoubaa TKamoun M 2007 J. Alloys Compd. 432 18
3Metcalfe G DReadinger E DShen HWoodward N TDierolf VWraback M 2009 J. Appl. Phys. 105 053101
4Dorenbos Pvan der Kolk E 2006 Appl. Phys. Lett. 89 061122
5Stachowicz MKozanecki AMa C GBrik M GLin J YJiang H XZavada J M 2014 Opt. Mater. 37 165
6Al tahtamouni T MStachowicz MLi JLin J YJiang H X 2015 Appl. Phys. Lett. 106 121106
7Svane AChristensen N EPetit LSzotek ZTemmerman W M 2006 Phys. Rev. 74 165204
8Yang YPitzer R M 2010 J. Phys. Chem. 114 7117
9George D KHawkins M DMclaren MJiang H XLin J YZavada J MVinh N Q 2015 Appl. Phys. Lett. 107 171105
10Ugolini CFeng I WSedhain ALin J YJiang H XZavada J M 2012 Appl. Phys. Lett. 101 051114
11Nepal NZavada J MDahal RUgolini CSedhain ALin J YJiang H X 2009 Appl. Phys. Lett. 95 022510
12Chai R PKuang X YLi C GZhao Y R 2011 Chem. Phys. Lett. 505 65
13Mishra IKripal RMisra S 2012 Chin. Phys. Lett. 29 037601
14He L MJi YWu H YXu BSun Y BZhang X FLu YZhao J J 2014 Chin. Phys. 23 077601
15Dwivedi PKripal RShukla S 2010 Chin. Phys. Lett. 27 017601
16Wang L XKuang X YLi H FChai R PWang H Q2010Acta Phys. Sin.596501(in Chinese)
17Palczewska MWolos AKaminska MGrzegory IBockowski MKrukowski SSuski TPorowski S 2000 Solid State Commun. 114 39
18Lea K RLeask M J MWolf W P 1962 J. Phys. Chem. Solids 23 1381
19Ammerlaan C A Jde Maat-Gersdorf I 2001 Appl. Magn. Reson. 21 13
20Chai R PKuang X YLiang LYu G H 2015 J. Phys. Chem. Solids 80 1
21Watts R KHolton W C 1968 Phys. Rev. 173 417
22Wu S YDong H N 2008 J. Alloys Compd. 451 248
23Wu S YDong H N2006J. Lumin.119517
24Maâlej RKammoun SDammak MKammoun M 2008 Mater. Sci. Eng. 146 183
25Newman D JNg B2000Crystal Field HandbookCambridgeCambridge University Press6436–43
26Newman D JNg B 1989 Rep. Prog. Phys. 52 699
27Chai R PHao D HKuang X YLiang L 2015 Spectrochim. Acta Part A 150 829
28Gruber J BSardar D KRussell lll C CYow R MZandi BKokanyan E P 2003 J. Appl. Phys. 94 7128
29Carnall W T 1992 J. Chem. Phys. 96 8713
30Judd B R1963Operator Techniques in Atomic SpectroscopyNew YorkMcGraw-Hill Book Co254625–46
31Wybourne B G1965Spectroscopic Properties of Rare EarthsNew YorkJohn Wiley & Sons, Inc.39
32Vishwamittar Puri S P 1974 Phys. Rev. 9 4673
33Abragam ABleaney B1986Electron Paramagnetic Resonance of Transition IonsNew YorkDover Publications, Inc.277341277–341
34Carnall W TFields P RSarup R 1972 J. Chem. Phys. 57 43
35Glukhanyuk VPrzybylińska HKozanecki AJantsch W 2006 Opt. Mater. 28 111
36Glukhanyuk VPrzybylińska HKozanecki AJantsch W 2006 Opt. Mater. 28 746
37Stevens K W H 1952 Proc. Phys. Soc. London, Sect. 65 209
38Jørgensen C K1962Absorption Spectra and Chemical Bonding in ComplexesOxford, London, New York, ParisPergamon Press296
39Reisfeld R CJørgensen C K1977Laser and Excited State of Rare EarthsBerlin, Heidelberg, New YorkSplinger-Verlag123