- Email: [email protected]
fluorine-passivation effects in amorphous silica fiber
Accepted Manuscript Research paper Hydrogen-/fluorine-passivation effects in amorphous silica fiber Zhixing Peng, Baonan Jia, Jie Zhang, Binbin Yan, You Wang, Bin Yang, Pengfei Lu PII: DOI: Reference:
S0009-2614(18)30752-8 https://doi.org/10.1016/j.cplett.2018.09.027 CPLETT 35939
To appear in:
Chemical Physics Letters
Received Date: Revised Date: Accepted Date:
14 July 2018 25 August 2018 11 September 2018
Please cite this article as: Z. Peng, B. Jia, J. Zhang, B. Yan, Y. Wang, B. Yang, P. Lu, Hydrogen-/fluorine-passivation effects in amorphous silica fiber, Chemical Physics Letters (2018), doi: https://doi.org/10.1016/j.cplett.2018.09.027
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Hydrogen-/fluorine-passivation effects in amorphous silica fiber Zhixing Penga, Baonan Jiaa, Jie Zhanga, Binbin Yana, You Wangb, Bin Yangc，* ,Pengfei Lua,*
a
State Key Laboratory of Information Photonics and Optical Communications, Beijing University of Posts and
Telecommunications, POB 72, Beijing 100876, China b
Southwest Institute of Technical Physics, POB 432, Chengdu 610041, Sichuan, China
c
High-Tech Research and Development Center, Ministry of Science and Technology, Beijing 100044, China
* [email protected], [email protected]
Abstract: We present first-principles combined GW and Bethe-Salpeter equation on passivation effect between ODC(I) and H/F atoms. The injection of H or F atoms help to form the robust Si-H or Si-F bond which increase the optical band gap, erase the defect state of ODC(I) at 0.74 eV and improve radiation toughness. The reaction of ODC(I) defect with a H atom or F atoms is barrierless, the Si-O length near the ODC(I) sites are all gradually getting shorter and bond energy are getting stronger. Coupled with the atomic structures and electro-optic properties, we found out that F is the better passivator.
Keywords: first principles, silica fiber, ODC defect, passivation
1. Introduction Silica has been one of the key materials in many technological fields and widely used in microelectronic or optoelectronic devices such as metal-oxide-semiconductor (MOS) devices, silica-based optics and optical fiber waveguides [1,2]. Because of its outstanding properties, the study above the nature of silica or silica-based optical fiber has attracted numerous attention [3,4]. However, in laser irradiation environment, some intrinsic defects like oxygen vacancy (ODC(I)) and silicon dangling bond (E’ center) would degrade the performance of optical fiber and decrease the transmission efficiency [5-7]. Therefore, how to eliminate or passivate these intrinsic defects in silica fiber could be critical to the application and development in telecommunication industry. In pure silica or silica fiber, oxygen deficient centers (ODC) including E’ centers and ODC(I) are playing an important role in the radiation-induced degradation of the opto-electronic properties and decrease of bandwidth. The ODC(I) could be observed by means of optical absorption and photoluminescence spectrum which its unique absorption band at 7.6 eV and emission bands at 4.4 eV, 2.7 eV [8-10]. Similarly, the E’ center leads to an absorption band at 5.8 eV [11,12]. In order to eliminate the affection caused by these intrinsic defects, hydrogen is introduced as a defect passivator during the manufacturing process [13]. The injection of hydrogen could effectively reduce the attenuation at visible to near-ultraviolet wavelengths and improve the depth of the deep ultraviolet region (DUV) [10-12]. Hydrogen-loading could help to suppress the intrinsic defects by saturating dangling bonds or strained bonds in silica or silica-based optical fiber [11-13]. El-Sayed et al. found that hydrogen atoms interact with strained Si-O bond, and formed two distinct defect structures, a [SiO4/H]0 center and a hydroxyl E’ center [14,15]. Lopez et al. found H2 molecules could react with ODC(I) or E’ center to formed Si-H bonds and interstitial H atoms [16]. In addition, there are some researches proved that fluorine-doping could promote the structural relaxation and improve the optical transparency in the vacuum-ultraviolet region [17-19]. In general, fluorine was added to a silica-based preform by vapor-phase axial deposition or a sol-gel method [20,21]. Moderate fluorine-doping is effective in decreasing the strained Si-O bonds or Si-Si bonds and enhancing the resistance of laser radiation hardness [18,20]. As fluorine has a high chemical activity, it usually exists in the form of Si-F bonds in modified silica [17]. K.
Saito et al. found fluorine-doping had an effect to break a 3-/4-membered rings in silica and formed with non-bridging Si-F bond [22]. J. E. Huheey et al. noted that the bond energy of Si-F bond (5.8 eV) was stronger than Si-O bond (4.7 eV) [23]. Even though plenty of works focus on the properties of the intrinsic defects or H-/F-related configuration in silica or silica fiber, while the nature of passivation process between the intrinsic defects, for instance ODC(I) and typical non-metallic element dopants like H/F, is still under debate. In this paper, we address the comparison passivation effect between ODC(I) and H/F atoms, and we analyze the geometry structure and formation energy of corresponding derivative defects. We calculate the electronic structures and optical properties of all these defect configurations. Our paper is organized as follows. In Section 2, the computational methods and models are presented. Our results and discussion are given in Section 3. Finally, we give a brief summary in Section 4. Our work not only aims at displaying a better understanding of the dopants’ passivation effect with the intrinsic defect, but also provides a better guidance to the poor performance of optical fiber in strong radiation environments.
2. Computational methods and models The ODC(I) configuration used in this work are obtained by a non-defective silica fiber model which quenching from a 2×2×2 crystal silica supercell and containing 32 Si atoms and 64 O atoms. Classical molecular dynamics (MD) method is performed and the density of silica supercell is in good agreement with the experimental value of 2.20 g/cm3 [24]. The Tersoff potential and the Langevin thermostat are used to control atomic interaction and system temperature [25,26]. A 2×2×2 k-point mesh for Brillouin zone integration is employed and the value of Gaussian broadening is 0.1 eV which has the minimal entropy of the supercell. The lattice structure is fully relaxed with the threshold of a maximum force of 0.01 eV/Å and optimized with high accuracy that the ground state electronic convergence limit is at 10-5 eV. The first-principle calculations presented in this work are based on density functional theory (DFT) by using plane wave-pseudopotential code VASP (Vienna ab initio simulation package) [27,28]. The projected augmented wave (PAW) [29,30] method has been employed to treat the interaction between core electrons and valence electrons. Many-body perturbation theory techniques have been performed to calculate the optical absorption spectra and quasi-particle
energies [31]. Electronic properties are calculated by the GW (where G stands for one particle Green function and W refers to screened Coulomb potential) approximation with a GW0 scheme and optical properties are obtained by the Bathe-Salpeter-equation (BSE) method based upon the quasi-particle scGW0 calculations [32,33]. The attraction between quasi-electron and quasi-hole could be solved by the BSE method. And the precision of GW-BSE calculation is set up within 0.1 eV. The potential lowest energy configuration is searched along the reaction path and a convergence criterion of 0.01 eV/Å was applied.
3. Results and discussion 3.1 Geometry structures. The amorphous SiO2 network is constructed by n-membered rings (n=3-9), where n referred to the number of Si atoms in a ring and the vast majority is distributed between 4 to 7 [34,35]. Figure 1 shows the ODC(I) defect configurations which by removing a O atom in 4MR, 5MR, 6MR and 7MR sites, respectively,. For convenience, these ODC(I) defect configurations are defined as VODC_nMR (n=4-7) . And after relaxation, it clearly shows that the VODC_nMR configurations remain at a stable structure.
(a) (a) VODC-4MR
(b) VODC-5MR
(c) VODC-6MR
(d) VODC-7MR
Figure. 1. ODC(I) defect in 4~7MR,. (a) ODC(I) defect in 4MR, (b) ODC(I) defect in 5MR, (c) ODC(I) defect in 6MR, (d) ODC(I) defect in 7MR, Si is shown in red, O is shown in yellow, and H is shown in white.
To compare relative stability of ODC(I) in different size of rings, we calculated the average formation energies of VODC_nMR (n=4-7) by the equation below.
(1) where
and
respectively.
are total energies of VODC_nMR and the non-defective silica fiber model is the energy of a single O atom. The calculated formation energies are given in
Table 1. Our calculations illustrate that the average formation energies of VODC_nMR are ranging from 5.14 eV to 5.75 eV, which are almost in accordance with the result of Pacchioni et al [36]. We find that VODC_5MR has the smallest formation energy which means the ODC(I) defect centers are more likely to occur in the 5MR site. And the average formation energy of VODC_5MR is closest to the experimental value [37] among all the displayed sites. All these results indicate that VODC_5MR is most energetically favorable among all the configurations and we conduct further calculation based on this model. Table 1. Calculated average formation energy of VODC_nMR
(n=4-7). aReference [38], bReference [37]
Configuration
VODC_4MR
VODC_5MR
VODC_6MR
VODC_7MR
Theorya
Exptb
(eV)
5.63(0.32)
5.14(0.21)
5.75(0.42)
5.45(0.27)
5.21(0.14)
5.05(0.10)
3.2 Atomic structures. In general, silica fiber as one of the typical amorphous materials, its properties are usually changed with physical disorder and external element doping. The physical disorder is related to the number of tetrahedral SiO4 units and variations in bond length. Figure 2(a) shows a typical ODC(I) defect configuration and four configurations which modified with fluorine or hydrogen atoms. The local structures of modified configurations are displayed in Fig. 2(b). Out of 32 different oxygen vacancy sites of random network structures are generated and modified with hydrogen or fluorine atoms.
(a)
one H atom doped
two H atoms doped
One F atom doped
two F atoms doped
(b)
Figure 2. (a) Typical ODC(I) defect configuration, (b) Local structure of H-/F-induced defect configurations, the top are one H-doped structure and two H-doped structure; the bottom are one F-doped structure and two F-doped
structure. Si atoms are yellow balls, O atoms are red balls, H atoms are white balls, F atoms are green balls.
After geometry optimization, we found that hydrogen-doping or fluorine-doping lead to structural relaxation and accompanied with minor variations of bond length or bond angle. Especially it has more significant influence on the two tetrahedral units around the ODC(I) defect site which marked with dashed black circles in Fig. 2(a). Add up to the 32 ODC(I) configurations relaxed with the DFT method, the statistics show that the Si-O bonds are ranging from 1.60-1.70 Å and average value is around 1.65 Å, which is accordance with experimental value [39]. The reaction of ODC(I) defect with only a H atom or F atoms are barrierless, and the barrier energy for ODC(I) interacts with H2 is 1.63 eV [14]. We calculate the Si-O bond length of the two tetrahedral units in Fig. 2(a) because atomic structure of the units is more likely affected by the doping of H/F atoms. In Fig. 3 we compare the distribution of Si-O bond length of ODC(I) defect and the H/F doped configurations. In Fig. 3(a), the initial Si-O bond length of ODC(I) defect is distributed in 1.62-1.66 Å and with an average value of 1.646 Å. When doped with a H atom, the bond length is varying from 1.61-1.66 Å as shown in Fig. 3(b) and has a smaller average value of 1.641 Å. Furthermore when doped with two H atoms, the average value of Si-O bond is around at 1.638 Å. As for Si-H bond, the average value is 1.49 Å which is in accordance with El-Sayed’s report [15]. In Fig. 3(d), when doped with a F atom, the bond length of Si-O is mainly distributed in 1.61-1.64 Å and the mean value is 1.624 Å. Similarly, after doping with two F atoms in Fig. 3(e), the value is mainly distributed in 1.60-1.64 Å, the mean value is approximately at 1.615 Å. case %%of of case
60
(a) ODC(I)
40 20
case %%of of case
600
(b) H-doped
40 20
%%ofofcase case
40
%%ofofcase case
40
% %ofofcase case
600
40
(c)Two H-doped
20 600
(d) F-doped
20 600
(e)Two F-doped
20 0
1.60
1.61
1.62
1.63
1.64
1.65
1.66
1.67
1.68
BondLength length(Å)(Å) Bond
Figure. 3. The distribution for Si-O bond length of different geometrical structures (a) ODC(1) defect structure; (b)
one H atom doped structure ; (c) two H atoms doped structure; (d) one F atom doped structure; (e) two F atoms doped structure.
These results show that after doped with H/F atom, the Si-O length both gradually gets shorter. As for Si-O bond, the shorter bond length usually means a higher bond energy. Thus it is less likely to break the bond and form with a defect structure which means that the doped configurations would be more stable [40,41]. In Ref. [17], Funabiki et al. noted that Si-F bond (bond energy; 5.8 eV) is much stronger than Si-O bond (bond energy; 4.7eV) [17]. Combined with the above calculation, we find the Si-O length with F-doped configuration is much smaller than the value in H-doped configuration, which means F is the better passivator than H.
3.3 Electron density distribution For the sake of better understanding passivation process and effects of the conversions, we compared the relative differential partial charge density of ODC(I) defects with H/F-doped configurations in Fig. 4. For the one H-doped defect in Fig. 4(a), the doping of H breaks Si-Si bond. With the addition of another H atom which shows in Fig. 4(b), the effective charge near the Si atoms decreases and the electrons transfer to the H atoms. As for the one F-doped defect in Fig. 4(c), similarly, F atom could break Si-Si bond and absorbs more effective electrons because of its better electronegativity property. In Fig. 4(d), with the addition of another F atom and forms robust Si-F groups, there is almost no charge interaction between two F atoms, while there is a slight charge interaction between the two H atoms. Compared with these defect configurations, the independence between the two F atoms is stronger and the formed Si-F bonds could be more robust.
(a)
(b)
0.5720 0.4745
Si
0.5373
0.3770
Si
0.2795
0.4305
H
0.3237
0.1820
Si
0.2170
0.08450
H
0.1102
Si
-0.01300 -0.1105
0.003500
H
-0.1032
-0.2080
(c)
-0.2100
(d)
0.3650 0.2650
Si
Si
-0.03500
0.4200 0.3094
0.1650 0.06500
0.6440
0.1989 F
F
0.08833 -0.02222
-0.1350
Si
-0.1328 -0.2350 -0.2433
-0.3350
F
Si
-0.4350 -0.5350 -0.6350
F
-0.3539 -0.4644 -0.5750
Figure. 4. Differential charge density calculations for the H-/F-induced defect structure. (a). Differential charge density of one H atom doped structure. (b). Differential charge density of two H atoms doped structure. (c). Differential charge density of one F atom doped structure. (d). Differential charge density of two F atoms doped structure.
3.4 Electronic structure and Optical property. Figure 5 shows the calculated total density of states (TDOS) for the five different defect structures, the calculated band gap of the stable ODC(I) configuration is 9.18 eV. It is known that DFT method may underestimate true value [42], so the GW+BSE method is performed and its’ major advantage is the quasi-particle eigenvalues could be obtained as a first-order perturbation of a Kohn-Sham, set of eigenvalues. For the stable ODC(I) configuration, Figure 5(a) shows an unoccupied defect state appears at 8.21eV and an occupied defect state locates at 0.74 eV, which is missing after doping with H/F. In Fig. 5(b) and Fig. 5(d), there is a defect state near 5 eV after doped with a single H or F atom and it is caused by E’ center. By comparing the Fig. 5(c) with Fig. 5(e), there is an occupied defect state around at Fermi energy in the Fig. 5(c) while it is not observed in Fig. 5(e). It is proved that F has a better effection of improving the electronic properties in silica fiber.
Figure 5. Calculated DOS of ODC(I) defect in black solid line and H/F-doped defect in red solid line and blue solid line, respectively.
To better comprehend the variation of optical properties after doping with H/F, we compared the optical absorption spectrum of the ODC(I) configuration and its modified configurations in Fig. 6. In Fig. 6(a), the ODC(I) defect causes an absorption band near 7.7 (0.5) eV which is in line with Girard’s et al result [10]. By comparing the Fig. 6(b) with Fig. 6(d), we find both configurations exist an absorption band during 5.0-6.4 eV after doping with a single H or F, and these should be caused by the E’ center defect. While after doping with two H/F atoms, both two configurations do not exhibit optical absorption bands within band gap of the silica fiber as shown in Fig. 6(c) and Fig. 6(e). Both elements could help to improve electronic properties, increase optical band gap of silica fiber and improve the optical transparency in DUV region.
Figure 6. Calculated optical absorption spectra (top) of ODC(I) defect in black solid line and H/F-doped defect in red solid line and blue solid line, respectively.
4. Conclusion In summary, classical MD and GW/BSE methodologies are used to explore the atomic structures, electronic and optical properties of the ODC(I) defect as well as its’ H/F-doped defects. We analyzed the different ODC(I) sites in 4-7MR and figured out the most energetically favorable site is VODC_5MR which the formation energy of 5.14 eV. Then the variation of atomic structure after H/F-doped is discussed, the Si-O length near the ODC(I) site would gradually get shorter which means a stronger bond energy after doping with H/F atoms. The Si-O length with F-doped configuration is even smaller than the value of H-doped configuration and the former configuration might be more stable. We analyze the electronic structures and optical properties of these configurations, the injection of H or F atoms could help to form robust Si-H or Si-F bond, it also improves the electro-optic properties and radiation toughness as well as vacuum-ultraviolet optical transmission performance. These results suggest that F is the better passivator in front of laser or radiation irradiation. Our calculations provide precise analysis and useful instructions to interaction in silica fiber and other silica-based devices, which is of remarkable significance to improve the radiation hardness of silica fiber. Acknowledgements This work was supported by the National Natural Science Foundation of China (61675032), the
Open Program of State Key Laboratory of Functional Materials for Informatics, and the Beijing Computational Science Research Center (CSRC). We acknowledge the computational support from the National Key Research and Development Program of China (No.2017YFB0405100). Reference [1]
H.S. Nalwa, Silicon-Based Materials and Devices, Academic Press, San Diego (2001).
[2]
J. Crisp, B. Elliott, Introduction to Fiber Optics (Third Edition), 2005.
[3]
R.P. Gupta, Physical Review B 32 (1985) 8278.
[4]
J. Sarnthein, A. Pasquarello, R. Car, Physical Review Letters 74 (1995) 4682.
[5]
A. Salleo, S.T. Taylor, M.C. Martin, W.R. Panero, R. Jeanloz, T. Sands, F.Y. Génin, Nature Materials 2 (2002) 796.
[6]
A. Salleo, T. Sands, F.Y. Génin, Applied Physics A 71 (2000) 601.
[7]
F.R. Wang, Acta Physica Sinica 59 (2010) 5122.
[8]
G. Pacchioni, G. Ieranò, Journal of Non-Crystalline Solids 216 (1997) 1.
[9]
S. Girard, Y. Ouerdane, G. Origlio, C. Marcandella, A. Boukenter, N. Richard, J. Baggio, P. Paillet, M. Cannas, J. Bisutti, IEEE Transactions on Nuclear Science 55 (2008) 3473.
[10]
S. Girard, N. Richard, Y. Ouerdane, G. Origlio, A. Boukenter, L. Martin-Samos, P. Paillet, J.P. Meunier, J. Baggio, M. Cannas, IEEE Transactions on Nuclear Science 55 (2008) 3508.
[11]
S. Agnello, G. Buscarino, F. Gelardi, R. Boscaino, Physical Review B 77 (2008) 195206.
[12]
G. Buscarino, R. Boscaino, S. Agnello, F. Gelardi, Physical Review B 77 (2008) 155214.
[13]
D.M. Fleetwood, R.D. Schrimpf, Defects in microelectronic materials and devices, CRC press, 2008.
[14]
A.-M. El-Sayed, Y. Wimmer, W. Goes, T. Grasser, V.V. Afanas' ev, A.L. Shluger, Physical Review B 92 (2015) 014107.
[15]
A.-M. El-Sayed, M.B. Watkins, T. Grasser, V.V. Afanas’ev, A.L. Shluger, Physical Review Letters 114 (2015) 115503.
[16]
N. Lopez, F. Illas, G. Pacchioni, The Journal of Physical Chemistry B 104 (2000) 5471.
[17]
F. Funabiki, T. Kamiya, H. Hosono, Journal of the Ceramic Society of Japan 120 (2012) 447.
[18]
K. Kajihara, M. Hirano, L. Skuja, H. Hosono, Frenkel defect process in amorphous silica, Damage to VUV, EUV, and X-ray Optics III. International Society for Optics and Photonics, 2011, p. 80770R.
[19]
L. Skuja, M. Hirano, H. Hosono, K. Kajihara, physica status solidi (c) 2 (2005) 15.
[20]
M. Kyoto, Y. Ohoga, S. Ishikawa, Y. Ishiguro, Journal of Materials Science 28 (1993) 2738.
[21]
S. Shibata, T. Kitagawa, M. Horiguchi, Journal of Non-Crystalline Solids 100 (1988) 269.
[22]
N. Shimodaira, K. Saito, A.J. Ikushima, Journal of Applied Physics 91 (2002) 3522.
[23]
J. Huheey, INORGANIC CHEMISTRY : PRINCIPLES STRUCTURE AND REACTIVITY, HARPER & ROW, 1983.
[24]
Z.A. Weinberg, G.W. Rubloff, E. Bassous, Physical Review B 19 (1979) 3107.
[25]
R.W. Godby, M. Schlüter, L.J. Sham, Physical Review B Condensed Matter 37 (1988) 10159.
[26]
L. Hedin, Physical Review 139 (1965) 663.
[27]
G. Kresse, J. Hafner, Physical Review B Condensed Matter 47 (1993) 558.
[28]
G. Kresse, J. Hafner, Physical Review B Condensed Matter 49 (1994) 14251.
[29]
P.E. Blöchl, Physical Review B 50 (1994) 17953?17979.
[30]
G. Kresse, Physical Review B 59 (1999) 1758.
[31]
L. Martin-Samos, G. Bussi, Computer Physics Communications 180 (2009) 1416.
[32]
J.P. Perdew, K. Burke, M. Ernzerhof, Physical Review Letters 77 (1996) 3865.
[33]
G. Onida, L. Reining, A. Rubio, Rev.Mod.Phys 74 (2002) 601.
[34]
M. Sitarz, W. Mozgawa, M. Handke, Journal of Molecular Structure s 511–512 (1999) 281.
[35]
A. Pasquarello, R. Car, Physical Review Letters 80 (1998) 5145.
[36]
G. Pacchioni, R. Ferrario, Physical Review B 58 (1998) 6090.
[37]
L. Skuja, Journal of Non-Crystalline Solids 239 (1998) 16.
[38]
N. Adelstein, D. Lee, J.L. Dubois, K.G. Ray, J.B. Varley, V. Lordi, Aip Advances 7 (2017) 2179.
[39]
R.L. Mozzi, B.E. Warren, Journal of Applied Crystallography 2 (2010) 164.
[40]
W. Wang, P. Lu, L. Han, C. Zhang, L. Wu, P. Guan, R. Su, J. Chen, Journal of Non-Crystalline Solids 434 (2016) 96.
[41]
W. Wang, P. Lu, L. Han, C. Zhang, C. Yang, R. Su, J. Chen, Applied Physics A 122 (2016) 929.
[42]
L.M. Pratt, N.V. Nguỹên, B. Ramachandran, Journal of Organic Chemistry 70 (2005) 4279.
Graphical abstract
Hydrogen-/fluorine-passivation effects in amorphous silica fiber
(a)
one H atom doped
two H atoms doped
One F atom doped
two F atoms doped
(b)
Highlights
.
The injection of H or F atoms would help to form the robust Si-H or Si-F bond which could increase the optical band gap, erase the defect states in band gap , improve radiation toughness and vacuum-ultraviolet optical transmission performance.
Coupled with the calculations of atomic structures and electro-optic properties, we found out that F is the better passivator.
S0009-2614(18)30752-8 https://doi.org/10.1016/j.cplett.2018.09.027 CPLETT 35939
To appear in:
Chemical Physics Letters
Received Date: Revised Date: Accepted Date:
14 July 2018 25 August 2018 11 September 2018
Please cite this article as: Z. Peng, B. Jia, J. Zhang, B. Yan, Y. Wang, B. Yang, P. Lu, Hydrogen-/fluorine-passivation effects in amorphous silica fiber, Chemical Physics Letters (2018), doi: https://doi.org/10.1016/j.cplett.2018.09.027
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Hydrogen-/fluorine-passivation effects in amorphous silica fiber Zhixing Penga, Baonan Jiaa, Jie Zhanga, Binbin Yana, You Wangb, Bin Yangc，* ,Pengfei Lua,*
a
State Key Laboratory of Information Photonics and Optical Communications, Beijing University of Posts and
Telecommunications, POB 72, Beijing 100876, China b
Southwest Institute of Technical Physics, POB 432, Chengdu 610041, Sichuan, China
c
High-Tech Research and Development Center, Ministry of Science and Technology, Beijing 100044, China
* [email protected], [email protected]
Abstract: We present first-principles combined GW and Bethe-Salpeter equation on passivation effect between ODC(I) and H/F atoms. The injection of H or F atoms help to form the robust Si-H or Si-F bond which increase the optical band gap, erase the defect state of ODC(I) at 0.74 eV and improve radiation toughness. The reaction of ODC(I) defect with a H atom or F atoms is barrierless, the Si-O length near the ODC(I) sites are all gradually getting shorter and bond energy are getting stronger. Coupled with the atomic structures and electro-optic properties, we found out that F is the better passivator.
Keywords: first principles, silica fiber, ODC defect, passivation
1. Introduction Silica has been one of the key materials in many technological fields and widely used in microelectronic or optoelectronic devices such as metal-oxide-semiconductor (MOS) devices, silica-based optics and optical fiber waveguides [1,2]. Because of its outstanding properties, the study above the nature of silica or silica-based optical fiber has attracted numerous attention [3,4]. However, in laser irradiation environment, some intrinsic defects like oxygen vacancy (ODC(I)) and silicon dangling bond (E’ center) would degrade the performance of optical fiber and decrease the transmission efficiency [5-7]. Therefore, how to eliminate or passivate these intrinsic defects in silica fiber could be critical to the application and development in telecommunication industry. In pure silica or silica fiber, oxygen deficient centers (ODC) including E’ centers and ODC(I) are playing an important role in the radiation-induced degradation of the opto-electronic properties and decrease of bandwidth. The ODC(I) could be observed by means of optical absorption and photoluminescence spectrum which its unique absorption band at 7.6 eV and emission bands at 4.4 eV, 2.7 eV [8-10]. Similarly, the E’ center leads to an absorption band at 5.8 eV [11,12]. In order to eliminate the affection caused by these intrinsic defects, hydrogen is introduced as a defect passivator during the manufacturing process [13]. The injection of hydrogen could effectively reduce the attenuation at visible to near-ultraviolet wavelengths and improve the depth of the deep ultraviolet region (DUV) [10-12]. Hydrogen-loading could help to suppress the intrinsic defects by saturating dangling bonds or strained bonds in silica or silica-based optical fiber [11-13]. El-Sayed et al. found that hydrogen atoms interact with strained Si-O bond, and formed two distinct defect structures, a [SiO4/H]0 center and a hydroxyl E’ center [14,15]. Lopez et al. found H2 molecules could react with ODC(I) or E’ center to formed Si-H bonds and interstitial H atoms [16]. In addition, there are some researches proved that fluorine-doping could promote the structural relaxation and improve the optical transparency in the vacuum-ultraviolet region [17-19]. In general, fluorine was added to a silica-based preform by vapor-phase axial deposition or a sol-gel method [20,21]. Moderate fluorine-doping is effective in decreasing the strained Si-O bonds or Si-Si bonds and enhancing the resistance of laser radiation hardness [18,20]. As fluorine has a high chemical activity, it usually exists in the form of Si-F bonds in modified silica [17]. K.
Saito et al. found fluorine-doping had an effect to break a 3-/4-membered rings in silica and formed with non-bridging Si-F bond [22]. J. E. Huheey et al. noted that the bond energy of Si-F bond (5.8 eV) was stronger than Si-O bond (4.7 eV) [23]. Even though plenty of works focus on the properties of the intrinsic defects or H-/F-related configuration in silica or silica fiber, while the nature of passivation process between the intrinsic defects, for instance ODC(I) and typical non-metallic element dopants like H/F, is still under debate. In this paper, we address the comparison passivation effect between ODC(I) and H/F atoms, and we analyze the geometry structure and formation energy of corresponding derivative defects. We calculate the electronic structures and optical properties of all these defect configurations. Our paper is organized as follows. In Section 2, the computational methods and models are presented. Our results and discussion are given in Section 3. Finally, we give a brief summary in Section 4. Our work not only aims at displaying a better understanding of the dopants’ passivation effect with the intrinsic defect, but also provides a better guidance to the poor performance of optical fiber in strong radiation environments.
2. Computational methods and models The ODC(I) configuration used in this work are obtained by a non-defective silica fiber model which quenching from a 2×2×2 crystal silica supercell and containing 32 Si atoms and 64 O atoms. Classical molecular dynamics (MD) method is performed and the density of silica supercell is in good agreement with the experimental value of 2.20 g/cm3 [24]. The Tersoff potential and the Langevin thermostat are used to control atomic interaction and system temperature [25,26]. A 2×2×2 k-point mesh for Brillouin zone integration is employed and the value of Gaussian broadening is 0.1 eV which has the minimal entropy of the supercell. The lattice structure is fully relaxed with the threshold of a maximum force of 0.01 eV/Å and optimized with high accuracy that the ground state electronic convergence limit is at 10-5 eV. The first-principle calculations presented in this work are based on density functional theory (DFT) by using plane wave-pseudopotential code VASP (Vienna ab initio simulation package) [27,28]. The projected augmented wave (PAW) [29,30] method has been employed to treat the interaction between core electrons and valence electrons. Many-body perturbation theory techniques have been performed to calculate the optical absorption spectra and quasi-particle
energies [31]. Electronic properties are calculated by the GW (where G stands for one particle Green function and W refers to screened Coulomb potential) approximation with a GW0 scheme and optical properties are obtained by the Bathe-Salpeter-equation (BSE) method based upon the quasi-particle scGW0 calculations [32,33]. The attraction between quasi-electron and quasi-hole could be solved by the BSE method. And the precision of GW-BSE calculation is set up within 0.1 eV. The potential lowest energy configuration is searched along the reaction path and a convergence criterion of 0.01 eV/Å was applied.
3. Results and discussion 3.1 Geometry structures. The amorphous SiO2 network is constructed by n-membered rings (n=3-9), where n referred to the number of Si atoms in a ring and the vast majority is distributed between 4 to 7 [34,35]. Figure 1 shows the ODC(I) defect configurations which by removing a O atom in 4MR, 5MR, 6MR and 7MR sites, respectively,. For convenience, these ODC(I) defect configurations are defined as VODC_nMR (n=4-7) . And after relaxation, it clearly shows that the VODC_nMR configurations remain at a stable structure.
(a) (a) VODC-4MR
(b) VODC-5MR
(c) VODC-6MR
(d) VODC-7MR
Figure. 1. ODC(I) defect in 4~7MR,. (a) ODC(I) defect in 4MR, (b) ODC(I) defect in 5MR, (c) ODC(I) defect in 6MR, (d) ODC(I) defect in 7MR, Si is shown in red, O is shown in yellow, and H is shown in white.
To compare relative stability of ODC(I) in different size of rings, we calculated the average formation energies of VODC_nMR (n=4-7) by the equation below.
(1) where
and
respectively.
are total energies of VODC_nMR and the non-defective silica fiber model is the energy of a single O atom. The calculated formation energies are given in
Table 1. Our calculations illustrate that the average formation energies of VODC_nMR are ranging from 5.14 eV to 5.75 eV, which are almost in accordance with the result of Pacchioni et al [36]. We find that VODC_5MR has the smallest formation energy which means the ODC(I) defect centers are more likely to occur in the 5MR site. And the average formation energy of VODC_5MR is closest to the experimental value [37] among all the displayed sites. All these results indicate that VODC_5MR is most energetically favorable among all the configurations and we conduct further calculation based on this model. Table 1. Calculated average formation energy of VODC_nMR
(n=4-7). aReference [38], bReference [37]
Configuration
VODC_4MR
VODC_5MR
VODC_6MR
VODC_7MR
Theorya
Exptb
(eV)
5.63(0.32)
5.14(0.21)
5.75(0.42)
5.45(0.27)
5.21(0.14)
5.05(0.10)
3.2 Atomic structures. In general, silica fiber as one of the typical amorphous materials, its properties are usually changed with physical disorder and external element doping. The physical disorder is related to the number of tetrahedral SiO4 units and variations in bond length. Figure 2(a) shows a typical ODC(I) defect configuration and four configurations which modified with fluorine or hydrogen atoms. The local structures of modified configurations are displayed in Fig. 2(b). Out of 32 different oxygen vacancy sites of random network structures are generated and modified with hydrogen or fluorine atoms.
(a)
one H atom doped
two H atoms doped
One F atom doped
two F atoms doped
(b)
Figure 2. (a) Typical ODC(I) defect configuration, (b) Local structure of H-/F-induced defect configurations, the top are one H-doped structure and two H-doped structure; the bottom are one F-doped structure and two F-doped
structure. Si atoms are yellow balls, O atoms are red balls, H atoms are white balls, F atoms are green balls.
After geometry optimization, we found that hydrogen-doping or fluorine-doping lead to structural relaxation and accompanied with minor variations of bond length or bond angle. Especially it has more significant influence on the two tetrahedral units around the ODC(I) defect site which marked with dashed black circles in Fig. 2(a). Add up to the 32 ODC(I) configurations relaxed with the DFT method, the statistics show that the Si-O bonds are ranging from 1.60-1.70 Å and average value is around 1.65 Å, which is accordance with experimental value [39]. The reaction of ODC(I) defect with only a H atom or F atoms are barrierless, and the barrier energy for ODC(I) interacts with H2 is 1.63 eV [14]. We calculate the Si-O bond length of the two tetrahedral units in Fig. 2(a) because atomic structure of the units is more likely affected by the doping of H/F atoms. In Fig. 3 we compare the distribution of Si-O bond length of ODC(I) defect and the H/F doped configurations. In Fig. 3(a), the initial Si-O bond length of ODC(I) defect is distributed in 1.62-1.66 Å and with an average value of 1.646 Å. When doped with a H atom, the bond length is varying from 1.61-1.66 Å as shown in Fig. 3(b) and has a smaller average value of 1.641 Å. Furthermore when doped with two H atoms, the average value of Si-O bond is around at 1.638 Å. As for Si-H bond, the average value is 1.49 Å which is in accordance with El-Sayed’s report [15]. In Fig. 3(d), when doped with a F atom, the bond length of Si-O is mainly distributed in 1.61-1.64 Å and the mean value is 1.624 Å. Similarly, after doping with two F atoms in Fig. 3(e), the value is mainly distributed in 1.60-1.64 Å, the mean value is approximately at 1.615 Å. case %%of of case
60
(a) ODC(I)
40 20
case %%of of case
600
(b) H-doped
40 20
%%ofofcase case
40
%%ofofcase case
40
% %ofofcase case
600
40
(c)Two H-doped
20 600
(d) F-doped
20 600
(e)Two F-doped
20 0
1.60
1.61
1.62
1.63
1.64
1.65
1.66
1.67
1.68
BondLength length(Å)(Å) Bond
Figure. 3. The distribution for Si-O bond length of different geometrical structures (a) ODC(1) defect structure; (b)
one H atom doped structure ; (c) two H atoms doped structure; (d) one F atom doped structure; (e) two F atoms doped structure.
These results show that after doped with H/F atom, the Si-O length both gradually gets shorter. As for Si-O bond, the shorter bond length usually means a higher bond energy. Thus it is less likely to break the bond and form with a defect structure which means that the doped configurations would be more stable [40,41]. In Ref. [17], Funabiki et al. noted that Si-F bond (bond energy; 5.8 eV) is much stronger than Si-O bond (bond energy; 4.7eV) [17]. Combined with the above calculation, we find the Si-O length with F-doped configuration is much smaller than the value in H-doped configuration, which means F is the better passivator than H.
3.3 Electron density distribution For the sake of better understanding passivation process and effects of the conversions, we compared the relative differential partial charge density of ODC(I) defects with H/F-doped configurations in Fig. 4. For the one H-doped defect in Fig. 4(a), the doping of H breaks Si-Si bond. With the addition of another H atom which shows in Fig. 4(b), the effective charge near the Si atoms decreases and the electrons transfer to the H atoms. As for the one F-doped defect in Fig. 4(c), similarly, F atom could break Si-Si bond and absorbs more effective electrons because of its better electronegativity property. In Fig. 4(d), with the addition of another F atom and forms robust Si-F groups, there is almost no charge interaction between two F atoms, while there is a slight charge interaction between the two H atoms. Compared with these defect configurations, the independence between the two F atoms is stronger and the formed Si-F bonds could be more robust.
(a)
(b)
0.5720 0.4745
Si
0.5373
0.3770
Si
0.2795
0.4305
H
0.3237
0.1820
Si
0.2170
0.08450
H
0.1102
Si
-0.01300 -0.1105
0.003500
H
-0.1032
-0.2080
(c)
-0.2100
(d)
0.3650 0.2650
Si
Si
-0.03500
0.4200 0.3094
0.1650 0.06500
0.6440
0.1989 F
F
0.08833 -0.02222
-0.1350
Si
-0.1328 -0.2350 -0.2433
-0.3350
F
Si
-0.4350 -0.5350 -0.6350
F
-0.3539 -0.4644 -0.5750
Figure. 4. Differential charge density calculations for the H-/F-induced defect structure. (a). Differential charge density of one H atom doped structure. (b). Differential charge density of two H atoms doped structure. (c). Differential charge density of one F atom doped structure. (d). Differential charge density of two F atoms doped structure.
3.4 Electronic structure and Optical property. Figure 5 shows the calculated total density of states (TDOS) for the five different defect structures, the calculated band gap of the stable ODC(I) configuration is 9.18 eV. It is known that DFT method may underestimate true value [42], so the GW+BSE method is performed and its’ major advantage is the quasi-particle eigenvalues could be obtained as a first-order perturbation of a Kohn-Sham, set of eigenvalues. For the stable ODC(I) configuration, Figure 5(a) shows an unoccupied defect state appears at 8.21eV and an occupied defect state locates at 0.74 eV, which is missing after doping with H/F. In Fig. 5(b) and Fig. 5(d), there is a defect state near 5 eV after doped with a single H or F atom and it is caused by E’ center. By comparing the Fig. 5(c) with Fig. 5(e), there is an occupied defect state around at Fermi energy in the Fig. 5(c) while it is not observed in Fig. 5(e). It is proved that F has a better effection of improving the electronic properties in silica fiber.
Figure 5. Calculated DOS of ODC(I) defect in black solid line and H/F-doped defect in red solid line and blue solid line, respectively.
To better comprehend the variation of optical properties after doping with H/F, we compared the optical absorption spectrum of the ODC(I) configuration and its modified configurations in Fig. 6. In Fig. 6(a), the ODC(I) defect causes an absorption band near 7.7 (0.5) eV which is in line with Girard’s et al result [10]. By comparing the Fig. 6(b) with Fig. 6(d), we find both configurations exist an absorption band during 5.0-6.4 eV after doping with a single H or F, and these should be caused by the E’ center defect. While after doping with two H/F atoms, both two configurations do not exhibit optical absorption bands within band gap of the silica fiber as shown in Fig. 6(c) and Fig. 6(e). Both elements could help to improve electronic properties, increase optical band gap of silica fiber and improve the optical transparency in DUV region.
Figure 6. Calculated optical absorption spectra (top) of ODC(I) defect in black solid line and H/F-doped defect in red solid line and blue solid line, respectively.
4. Conclusion In summary, classical MD and GW/BSE methodologies are used to explore the atomic structures, electronic and optical properties of the ODC(I) defect as well as its’ H/F-doped defects. We analyzed the different ODC(I) sites in 4-7MR and figured out the most energetically favorable site is VODC_5MR which the formation energy of 5.14 eV. Then the variation of atomic structure after H/F-doped is discussed, the Si-O length near the ODC(I) site would gradually get shorter which means a stronger bond energy after doping with H/F atoms. The Si-O length with F-doped configuration is even smaller than the value of H-doped configuration and the former configuration might be more stable. We analyze the electronic structures and optical properties of these configurations, the injection of H or F atoms could help to form robust Si-H or Si-F bond, it also improves the electro-optic properties and radiation toughness as well as vacuum-ultraviolet optical transmission performance. These results suggest that F is the better passivator in front of laser or radiation irradiation. Our calculations provide precise analysis and useful instructions to interaction in silica fiber and other silica-based devices, which is of remarkable significance to improve the radiation hardness of silica fiber. Acknowledgements This work was supported by the National Natural Science Foundation of China (61675032), the
Open Program of State Key Laboratory of Functional Materials for Informatics, and the Beijing Computational Science Research Center (CSRC). We acknowledge the computational support from the National Key Research and Development Program of China (No.2017YFB0405100). Reference [1]
H.S. Nalwa, Silicon-Based Materials and Devices, Academic Press, San Diego (2001).
[2]
J. Crisp, B. Elliott, Introduction to Fiber Optics (Third Edition), 2005.
[3]
R.P. Gupta, Physical Review B 32 (1985) 8278.
[4]
J. Sarnthein, A. Pasquarello, R. Car, Physical Review Letters 74 (1995) 4682.
[5]
A. Salleo, S.T. Taylor, M.C. Martin, W.R. Panero, R. Jeanloz, T. Sands, F.Y. Génin, Nature Materials 2 (2002) 796.
[6]
A. Salleo, T. Sands, F.Y. Génin, Applied Physics A 71 (2000) 601.
[7]
F.R. Wang, Acta Physica Sinica 59 (2010) 5122.
[8]
G. Pacchioni, G. Ieranò, Journal of Non-Crystalline Solids 216 (1997) 1.
[9]
S. Girard, Y. Ouerdane, G. Origlio, C. Marcandella, A. Boukenter, N. Richard, J. Baggio, P. Paillet, M. Cannas, J. Bisutti, IEEE Transactions on Nuclear Science 55 (2008) 3473.
[10]
S. Girard, N. Richard, Y. Ouerdane, G. Origlio, A. Boukenter, L. Martin-Samos, P. Paillet, J.P. Meunier, J. Baggio, M. Cannas, IEEE Transactions on Nuclear Science 55 (2008) 3508.
[11]
S. Agnello, G. Buscarino, F. Gelardi, R. Boscaino, Physical Review B 77 (2008) 195206.
[12]
G. Buscarino, R. Boscaino, S. Agnello, F. Gelardi, Physical Review B 77 (2008) 155214.
[13]
D.M. Fleetwood, R.D. Schrimpf, Defects in microelectronic materials and devices, CRC press, 2008.
[14]
A.-M. El-Sayed, Y. Wimmer, W. Goes, T. Grasser, V.V. Afanas' ev, A.L. Shluger, Physical Review B 92 (2015) 014107.
[15]
A.-M. El-Sayed, M.B. Watkins, T. Grasser, V.V. Afanas’ev, A.L. Shluger, Physical Review Letters 114 (2015) 115503.
[16]
N. Lopez, F. Illas, G. Pacchioni, The Journal of Physical Chemistry B 104 (2000) 5471.
[17]
F. Funabiki, T. Kamiya, H. Hosono, Journal of the Ceramic Society of Japan 120 (2012) 447.
[18]
K. Kajihara, M. Hirano, L. Skuja, H. Hosono, Frenkel defect process in amorphous silica, Damage to VUV, EUV, and X-ray Optics III. International Society for Optics and Photonics, 2011, p. 80770R.
[19]
L. Skuja, M. Hirano, H. Hosono, K. Kajihara, physica status solidi (c) 2 (2005) 15.
[20]
M. Kyoto, Y. Ohoga, S. Ishikawa, Y. Ishiguro, Journal of Materials Science 28 (1993) 2738.
[21]
S. Shibata, T. Kitagawa, M. Horiguchi, Journal of Non-Crystalline Solids 100 (1988) 269.
[22]
N. Shimodaira, K. Saito, A.J. Ikushima, Journal of Applied Physics 91 (2002) 3522.
[23]
J. Huheey, INORGANIC CHEMISTRY : PRINCIPLES STRUCTURE AND REACTIVITY, HARPER & ROW, 1983.
[24]
Z.A. Weinberg, G.W. Rubloff, E. Bassous, Physical Review B 19 (1979) 3107.
[25]
R.W. Godby, M. Schlüter, L.J. Sham, Physical Review B Condensed Matter 37 (1988) 10159.
[26]
L. Hedin, Physical Review 139 (1965) 663.
[27]
G. Kresse, J. Hafner, Physical Review B Condensed Matter 47 (1993) 558.
[28]
G. Kresse, J. Hafner, Physical Review B Condensed Matter 49 (1994) 14251.
[29]
P.E. Blöchl, Physical Review B 50 (1994) 17953?17979.
[30]
G. Kresse, Physical Review B 59 (1999) 1758.
[31]
L. Martin-Samos, G. Bussi, Computer Physics Communications 180 (2009) 1416.
[32]
J.P. Perdew, K. Burke, M. Ernzerhof, Physical Review Letters 77 (1996) 3865.
[33]
G. Onida, L. Reining, A. Rubio, Rev.Mod.Phys 74 (2002) 601.
[34]
M. Sitarz, W. Mozgawa, M. Handke, Journal of Molecular Structure s 511–512 (1999) 281.
[35]
A. Pasquarello, R. Car, Physical Review Letters 80 (1998) 5145.
[36]
G. Pacchioni, R. Ferrario, Physical Review B 58 (1998) 6090.
[37]
L. Skuja, Journal of Non-Crystalline Solids 239 (1998) 16.
[38]
N. Adelstein, D. Lee, J.L. Dubois, K.G. Ray, J.B. Varley, V. Lordi, Aip Advances 7 (2017) 2179.
[39]
R.L. Mozzi, B.E. Warren, Journal of Applied Crystallography 2 (2010) 164.
[40]
W. Wang, P. Lu, L. Han, C. Zhang, L. Wu, P. Guan, R. Su, J. Chen, Journal of Non-Crystalline Solids 434 (2016) 96.
[41]
W. Wang, P. Lu, L. Han, C. Zhang, C. Yang, R. Su, J. Chen, Applied Physics A 122 (2016) 929.
[42]
L.M. Pratt, N.V. Nguỹên, B. Ramachandran, Journal of Organic Chemistry 70 (2005) 4279.
Graphical abstract
Hydrogen-/fluorine-passivation effects in amorphous silica fiber
(a)
one H atom doped
two H atoms doped
One F atom doped
two F atoms doped
(b)
Highlights
.
The injection of H or F atoms would help to form the robust Si-H or Si-F bond which could increase the optical band gap, erase the defect states in band gap , improve radiation toughness and vacuum-ultraviolet optical transmission performance.
Coupled with the calculations of atomic structures and electro-optic properties, we found out that F is the better passivator.