- Email: [email protected]
SiO2 multilayers-based p-i-n structure
Effect of resonant tunneling on electroluminescence in nc-Si/SiO 2 multilayersbased p-i-n structure D.Y. Chen, Y. Sun, Y.Y. Wang, Y.J. He, G. Zhang PII: DOI: Reference:
S0040-6090(14)01294-2 doi: 10.1016/j.tsf.2014.12.035 TSF 33999
To appear in:
Thin Solid Films
Received date: Revised date: Accepted date:
4 June 2014 1 November 2014 19 December 2014
Please cite this article as: D.Y. Chen, Y. Sun, Y.Y. Wang, Y.J. He, G. Zhang, Effect of resonant tunneling on electroluminescence in nc-Si/SiO2 multilayers-based p-i-n structure, Thin Solid Films (2014), doi: 10.1016/j.tsf.2014.12.035
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.
ACCEPTED MANUSCRIPT Effect
of
resonant
tunneling
on
electroluminescence
in
nc-Si/SiO2
multilayers-based p-i-n structure D. Y. Chen1,2*, Y. Sun1, Y. Y. Wang1, Y. J. He1, G. Zhang1 Nanjing University of posts and Telecommunications, Nanjing 210046, China
T
1* 2
Nanjing National Laboratory of Microstructures and Key Laboratory of Advanced Photonic and Electronic
[email protected]
Abstract
SC R
tel: +86-18936005695
IP
materials, Department of Physics, Nanjing University, Nanjing 210093, China
P-i-n structures with SiO2/nc-Si/SiO2 multilayers as intrinsic layer were prepared in conventional enhanced
chemical
vapor
deposition
system.
NU
plasma
Their
carrier
transport
and
electroluminescence properties were investigated. Two resonant tunneling related current peaks
MA
with current dropping gradually under forward bias were observed in the current voltage curve. Non-uniformity of the interfaces might be in responsible for the gradually dropping of the current.
D
Electroluminescence intensity of the device under bias of 7 V which is near the resonant tunneling
TE
peak voltage of 7.2 V was weaker than that under 6.5 V. According to the Gaussian fitting results of the spectra, the intensity of the sub-peak of 650 nm originating from recombination of injected
CE P
electrons and holes was decreased the most. When resonant tunneling conditions are met, it might be that most of injected electrons participate in resonant tunneling and fewer in Pool-Frenkel tunneling which is the main carrier transport mechanism to contribute to electroluminescence
AC
intensity.
Keywords: resonant tunneling; electroluminescence; SiO2/nc-Si/SiO2 multilayers; Poole-Frenkel tunneling;
1. Introductions Resonant tunneling has been reported and studied extensively in GaAs/AlGaAs [1-3], Si/SiGe [4-5], single-crystalline-Si/amorphous-SiO2 system [6-7], nc-Si embedded in dielectric matrices such as SiO2 or SiNx [8-14]. And SiO2/nc-Si/SiO2 system has some advantages such as process compatibility with microelectronic technology, higher recombination possibility for the small distance between wave functions of electrons and holes than that in bulk silicon, and so on.
ACCEPTED MANUSCRIPT SiO2/nc-Si/SiO2 system has been studied for its potential applications in light emitting diode (LED), optoelectronic integrated circuits [15–19] and other areas. For example, SiO2/nc-Si/SiO2 system was deposited on patterned substrate to increase carrier injection possibility [20] and
IP
T
achieve higher electroluminescence (EL) efficiency. In another way, SiO2/nc-Si/SiO2 system was used as intrinsic layer to construct p-i-n structure to obtain higher efficient LED [21]. Besides,
SC R
SiO2/nc-Si/SiO2 system has also been studied for its potential application in high speed switch circuits [22-23] for their high peak to valley current ratio (PVCR) of 60. The thicknesses of the nc-Si and SiO2 sub-layers and film quality both play important roles in resonant tunneling devices,
NU
which affect both the peak voltage and PVCR values. But, what will happen to the EL in SiO2/nc-Si/SiO2 multilayers-based p-i-n structure if resonant tunneling takes place? In this work,
MA
resonant tunneling and the effect on EL in the p-i-n structures will be studied in details. 2. Experimental Details
D
P-i-n structures were prepared in a conventional plasma enhanced chemical vapor deposition
TE
system with radio frequency (r.f.) of 13.56 MHz. The r.f. power and substrate temperature were kept at 50 W and 250 ℃, respectively. Amorphous Si/SiO2 (α-Si/SiO2) multilayers of six periods
CE P
were prepared by alternating α-Si deposition and in situ plasma oxidation processes. Silane was used to deposit α-Si with flow rate of 5 sccm. And oxygen was used to oxidize the α-Si with flow rate of 20 sccm. The deposition and oxidation time is 40 s and is 4 min, respectively. Phosphine
AC
and Borane were used as doping precursor of the n-type and p-type α-Si in the p-i-n structure, the corresponding conductivities are about 3x10-3 S·cm-1 and 5x10-5 S·cm-1, respectively, measured using Travelling Wave Method. The thickness of n-type and p-type α-Si are around 30 nm and 50 nm, respectively, according to the transmission electron microscope (TEM) image of the structure in figure 1. The as-deposited p-i-n structures were dehydrogenated at 450 ℃ for 1 h firstly, then were rapid thermal annealed at 1000 ℃ for 50 s and finally were furnace annealed at 1100 ℃ for 1h to form SiO2/nc-Si/SiO2-based p-i-n structure. All the treatments were carried out in N2 gas. In this work, single crystal p-Si wafers with resistivity of 2~8 Ω·cm were used as substrates. After thermal treatments, aluminum (Al) pad was evaporated as electrodes on both sides of the devices, and ring-like mask was used in top electrode evaporating process to form ring-like Al
ACCEPTED MANUSCRIPT pad as shown in our previous work [21] for light extracting in EL measurement. And EL spectra
NU
SC R
IP
T
were recorded by FluoroMax-2 under forward bias only.
Inset is the high resolusion image. 3. Results and Discussions
MA
Fig. 1. TEM image of the p-i-n structure with SiO2/nc-Si/SiO2 multi-layers as the intrinsic layer.
D
The deposition parameters of the intrinsic SiO2/nc-Si/SiO2 multilayers were the same as that
TE
of the sample C in our previous work [23], the thickness of nc-Si and SiO2 were about 6 nm and 5.2 nm, respectively. Effect of p-i-n structure to the EL of the SiO2/nc-Si/SiO2 multilayers has
CE P
been studied in our previous work [21], with turn-on voltage lowered and EL efficiency enhanced due to the reduced carrier tunneling barrier. EL properties of the present p-i-n structure are shown in figure 2 (a). The turn-on voltage is 6.5 V, which is higher than that in our previous work [21]
AC
for its thicker SiO2 layers. The EL spectra of the device are Gaussian fitted with three sub-peaks. Figure 2 (b) gives the fitting results of the EL spectrum under bias of 9 V. One sub-peak is around 800 nm. Its peak position and intensity changed slightly with bias. The origination may be the Si=O double bond [[24-25]. As for the other two sub-peaks, they are around 535 nm and 650 nm, respectively. The originations may be hot electron relaxation [26-27] for the former proposed by Pavesi group and recombination of injected electrons and holes in nc-Si/SiO2 multilayers within nc-Si and/or via the luminescence centers near Si/SiO2 interfaces [28-29] for the latter. The evolution of the two sub-peaks’ positions and intensities with bias are plotted in figure 2 (c) and (d). As bias increased from 7 V to 9 V, both the positions of the two sub-peaks changed slightly, yet their intensities increased continuously. When bias is below 7V and above 9 V, both the sub-peaks’ positions shifted obviously. The origination of the shift will be investigated in future. Here we might conclude that the shift might have nothing to do with resonant tunneling, and may
ACCEPTED MANUSCRIPT be related to the applied bias. From the plots of the intensities evolution in figure 2 (d), a sharp dip can be observed in the plot (black dot line) of the sub-peak of 650 nm, and the corresponding bias is 7 V which is near the resonant tunneling peak voltage of around 7.2V as shown in the
IP
T
current-voltage (IV) curve in figure 3. Slightly decreasing can be observed in the plot of the other sub-peak of 535 nm at the same bias. Thus the sub-peak of 650 nm could answer for the effect of
CE P
TE
D
MA
NU
SC R
resonant tunneling on EL intensity.
Fig. 2. EL spectra of the p-i-n structure (a); Gaussian fitting results of the EL spectra under bias of
AC
9 V (b); plots of the two sub-peaks’ positions with bias (c) and plots of intensities with bias (d).
Fig. 3. The measured IV curve (solid line) and the approximate background IV curve (dash line) of the p-i-n structure. Two broad resonant tunneling current peaks can be seen in the IV curve in figure 3. The peak
ACCEPTED MANUSCRIPT voltages are around 7.2 V and 12.8 V, the corresponding PVCR values are about 2 and 1.03, respectively. In our previous work we have reported on resonant tunneling in SiO2/nc-Si/SiO2 system. Here we could analyze the current peak using energy band diagram as shown in figure 4. It is well known that the resonant tunneling current could be approximately expressed as the
IP
I E lf En
T
following formular [22]:
SC R
Where, I is the resonant tunneling current, and E lf is the Fermi level in the emitter,
En (n=1, 2……) is the energy level in the quantum well, here in this work is nc-Si grains. The
NU
value of ( E lf En ) is in direct proportion to the applied bias. Thus I should be in linear relation
MA
with the bias. This is conformed in our experiment by the dot lines of b-c and d-e in the measured IV curve in figure 3. When energy level En is lower than Ecl (conduction band bottom of the emitter), resonant tunneling stops immediately, resulting in current dropping sharply as in our
D
previous work [23]. While in this work, current drops gradually with bias, and the IV curve is
TE
nearly symmetric, which looks similar as the band to band tunneling in Esaki’s diode [1]. In order to analyze carrier transport in the device, we consider a simplified structure with
CE P
two SiO2 barriers and one nc-Si well layer to construct the band diagram as in figure 4. Where, Ecl and Ec are conduction band bottoms of the emitter and bulk silicon, respectively;
E1 and E2
are quantized energy level in nc-Si quantum well. Panel (a) is under thermal equilibrium condition
AC
(without considering the interface states), corresponding to point “a” in IV curve in figure 3. As bias increases, E lf is pulled up to be equal to and then greater than E1 as in panel (b), corresponding to the first resonant tunneling process, from point “b” to “c” in figure 3. When bias increases further, Ecl is greater than E1 as in panel (c), the first resonant tunneling stops, and current drops with bias. If bias increases further on, the second resonant tunneling will occur and induces the second current peak of d-e-f as shown in figure 3. In the p-i-n structure here, the interfaces are non-uniform as can be seen clearly in the TEM image as shown in figure 1, which might lead to the quantized energy levels in the every quantum dot of nc-Si grain in the same quantum well layer being unequal to one another, resulting in forming of sub-band-like energy level, therefore, resonant tunneling in the structure behaves like band-to-band tunneling as in Esaki’s diode. Thus, as shown in figure 3 from “c” to “d” and “e” to “f”, the current dropped gradually rather than sharply when bias is higher than the peak voltage.
SC R
IP
T
ACCEPTED MANUSCRIPT
CE P
TE
D
MA
NU
Fig. 4. Simplified Conduction band diagram of the structure under different bias conditions.
Fig.5. Plot of ln( Ibackground ) versus V0.5, dot line is for linear fitting.
AC
In order to investigate the effect of resonant tunneling on the EL behavior, the background current (defined as current originating from other than resonant tunneling mechanisms) was estimated by dash-lining the valley current points of “a”, “d” and “f” smoothly as shown in figure 3. Plot of ln( Ibackground ) versus V0.5 was given in figure 5. A linear relationship can be found when the bias is larger than 2.5 V from the straight dot line in figure 5, suggesting that Poole-Frenkel tunneling (PF tunneling) is the main transport mechanism [30-31] besides resonant tunneling, i.e., PF tunneling contributes to the background current and EL property of the device. When bias is 7 V, which is near the resonant tunneling peak voltage of 7.2 V, most of the carriers injected from the emitter may participate in the fast process of resonant tunneling, thus fewer in PF tunneling, resulting in weak EL intensity. When higher bias is applied, carrier energy in emitter misaligns with the quantized energy level E1 in nc-Si layer, resonant tunneling stops and more
ACCEPTED MANUSCRIPT and more carriers can take part in PF tunneling to contribute to EL. On the other hand, The EL sub-peak of 650 nm under 7 V is affected the most from the dip in figure 2 (d). This sub-peak originates from recombination of injected electrons and holes in nc-Si/SiO2 multilayers within
IP
T
nc-Si and/or via the luminescence centers near Si/SiO2 interfaces. Resonant tunneling is a fast process, the tunneling time is less than 1ps [22], therefore electrons that participate in resonant
SC R
tunneling have not enough time to recombine with holes and thus make no contribution to EL at sub-peak of 650nm, leading to weaker EL intensity under resonant tunneling peak voltage. 4. Conclusions
NU
Carrier transport mechanisms and EL in SiO2/nc-Si/SiO2 multilayer-based p-i-n structure were studied. Resonant tunneling under forward bias of 7.2 V and 12.8 V, with PVCR value of 2
MA
and 1.03, respectively, can be observed in the IV curve. EL intensity under bias of 7 V is dramatically decreased in comparison with that of 6.5 V. The EL spectra was Gaussian fitted with
D
three peaks of 800 nm, 650 nm and 535 nm, and the intensity of the one of 650 nm is decreased
TE
the most, which originates from recombination of injected electrons with holes in multi-layers. When resonant tunneling takes place, most of the injected electrons from the emitter take part in
CE P
the fast process of resonant tunneling, thus fewer in PF tunneling to recombine with holes to contribute to EL. Therefore EL intensity under bias near resonant tunneling peak voltage is decreased greatly. On the other hand, the non-uniformity of the interfaces make the quantized
AC
energy level in the every nc-Si dot unequal to one another in the same well layer, leading to sub-band-like energy level forming and resulting in current dropping gradually in IV curve . ACKNOWLEDGMENTS This work was supported by NSFC (Nos. 61036001, 10874070), “973” project (2007CB613401), NSF of Jiangsu Province (BK2010010, 14KJB510025) and the Fundamental Research Funds for the Central Universities (1112021001).
ACCEPTED MANUSCRIPT References [1] R. Tsu and L. Esaki, R. Tsu and L. Esaki, Tunneling in a finite superlattice, Appl. Phys. Lett., 22 (1973) 562 [2] T. C. L. G. Sollner, W. D. Goodhue, P. E. Tannenwald, C. D. Parker, and D. D. Peck,
T
Resonant tunneling through quantum wells at frequencies up to 2.5 THz, Appl. Phys. Lett., 43
IP
(1983) 588
[3] M. Tsuchiya and H. Sakaki, Dependence of resonant tunneling current on well widths in
SC R
AlAs/GaAs/AlAs double barrier diode structures, Appl. Phys. Lett., 49 (1986) 88 [4] H. C. Liu, D. Landheer, M. Buchanan, and D. C. Houghton, Resonant tunneling in Si/Si1−x Ge x double‐barrier structures, Appl. Phys. Lett., 52 (1988) 1809
barrier diodes, Appl. Phys. Lett., 59 (1991) 973
NU
[5] K. Ismail, B. S. Meyerson, and P. J. Wang, Electron resonant tunneling in Si/SiGe double
[6] Y. Ishikawa, T. Ishihara, M. Iwasaki, and M. Tabe, Negative differential conductance due to
MA
resonant tunnelling through SiO2/single-crystalline-Si double barrier structure, Electron. Lett., 37 (2001) 1200
[7] H. Ikeda, M. Iwasaki, Y. Ishikawa, and M. Tabe, Resonant tunneling characteristics
D
in double-barrier structures in a wide range of applied voltage, Appl. Phys. Lett., 83 (2003) 1456
TE
[8] Y. Ishikawa, H. Ikeda, and M.Tabe. Potential-well-roughness-induced transition from resonant tunneling to single-electron tunneling in Si/SiO2 double-barrier structure, Appl. Phys. Lett., 86 (2005) 013508
CE P
[9] B. Berghoff, S. Suckow, R. Rölver, B. Spangenberg, H. Kurz, A. Dimyati, and J. Mayer, Resonant and phonon-assisted tunneling transport through silicon quantum dots embedded in SiO2, Appl. Phys. Lett., 93 (2008) 132111
AC
[10] Z. Pei, A. Y. K. Su, H. L. Hwan, and H. L. Hsiao, Room temperature tunneling transport through Si nanodots in silicon rich silicon nitride, Appl. Phys. Lett., 86 (2005) 063503 [11] M. Fukuda, K. Nakagawa, S. Miyazaki, and M. Hirose, Resonant tunneling through a self-assembled Si quantum dotAppl. Phys. Lett., 70 (1997) 2291 [12] A. Surawijaya, H. Mizuta, and S. Oda, Observation and Analysis of Tunneling Properties of Single Spherical Nanocrystalline Silicon Quantum Dot, Jpn. J. Appl. Phys., 45 (2006) 3638 [13] G. R. Lin, The Structural and Electrical Characteristics of Silicon-Implanted Borosilicate Glass, Jpn. J. Appl. Phys., 41 (2002) L1379-L1382 [14] G. R. Lin, C. J. Lin, C. K. Lin, L. J. Chou and Y. L. Chueh, Oxygen defect and Si nanocrystal dependent
white-light
and
near-infrared
electroluminescence
of
Si-implanted
and
plasma-enhanced chemical-vapor deposition-grown Si-rich SiO2, J. Appl. Phys., 97 (2005) 094306 [15] C. Delerue, G. Allan and M. Lannoo, Theoretical aspects of the luminescence of porous silicon,Phys. Rev. B, 48 (1993) 11024 [16] M. Zacharias, J. Heitmann, R. Scholz and U. Kahler, Size-controlled highly luminescent
ACCEPTED MANUSCRIPT silicon nanocrystals: A SiO/SiO2 superlattice approach, Appl.Phys. Lett., 80 (2002) 661 [17] B. Averboukh, R. Huber, K. W. Cheah, Y. R. Shen, G. G. Qin, Z. C. Ma and W. H. Zong, Luminescence studies of a Si/SiO2 superlattice, J. Appl. Phys., 92 (2002) 3564 [18] J. Xu, L. Yang, Y. J. Rui, J. X. Mei, X. Zhang, W. Li, Z. Ma, L. Xu, X. F. Huang and K. J.
T
Chen, Photoluminescence characteristics fromamorphous SiC thin films with various structures
IP
deposited at lowtemperature, Solid State Commun., 133 (2005), 565-568
[19] R. Huang, K. Chen, P. Han, H. Dong, X. Wang, D. Chen, W. Li, J. Xu, Z. Ma and X. Huang,
SC R
Strong green-yellow electroluminescence from oxidized amorphous silicon nitride light-emitting devices, Appl. Phys. Lett., 90 (2007) 093515
[20] D. Chen, Y. Liu, J. Xu, D. Wei, H. Sun, L. Xu, T. Wang, W. Li, K. Chen, Improved emission electroluminescent
device
containing
nc-Si/SiO2
multilayers
by using
NU
efficiency of
nano-patterned substrate, Opt. Exp., 18 (2010) 917
[21] D. Y. Chen, D. Y. Wei, J. Xu, P. G. Han, X. Wang, Z. Y. Ma, K. J. Chen, W. H. Shi and Q.
MA
M. Wang, Enhancement of electroluminescence in p–i–n structures with nano-crystalline Si/SiO2 multilayers, Semicond. Sci. Technol., 23 (2008) 015013 [22]L. Du, Y. Q. Zhuang, Nanoelectronics, first ed., publish house of electronics industry, Beijing,
D
2004, pp. 277
TE
[23] D. Y. Chen, Y. Sun, Y. J. He, L. Xu, and J. Xu, Resonant tunneling with high peak to valley current ratio in SiO2/nc-Si/SiO2 multilayers at room temperature, J. Appl. Phys., 115 (2014)
CE P
043703
[24] M. Luppi and S. Ossicini, J. Appl. Phys. 94 (2003) 2130 [25] M. Kulakci, U. Serincan and R. Turan, Semicond. Sci. Technol. 21 (2006) 1527 [26] Z.Gaburro, G. Pucker, P. Bellutti, and L. Pevesi;Solid State Commun., 114 (2000) 33
AC
[27] G. Pucker, P. Bellutti, M. Gazzanelli, Z.Gaburro and L. Pevesi, Opt. Mater.; 17 (2001)27 [28] G. R. Lin, C. J. Lin, and C.K. Lin, J. Appl. Phys., 97 (2005) 094306 [29] Y. Kanemitsu, Phys. Rep., 263 (1995) 1 [30] E. A. Morais, L. V. A. Scalvi, S. J. L. Ribeiro, and V. Geraldo, Poole–Frenkel effect in Er doped SnO2 thin films deposited by sol-gel-dip-coating, Phys. Stat. Sol. (a), 202 (2005) 301–308 [31] W.R. Harrella, J. Freyb, Observation of Poole-Frenkel effect saturation in SiO2 and other insulating , Thin Solid Films, 352 (1999) 195-204
ACCEPTED MANUSCRIPT Figure captions:
T
Fig. 1. TEM image of the p-i-n structure with SiO2/nc-Si/SiO2 multi-layers as the intrinsic layer.
IP
Fig. 2. EL spectra of the p-i-n structure (a); Gaussian fitting results of the EL spectra under bias of
SC R
9 V (b); plots of the two sub-peaks’ positions with bias (c) and plots of intensities with bias (d).
Fig. 3. The measured IV curve (solid line) and the approximate background IV curve (dash line) of
MA
NU
the p-i-n structure.
D
Fig. 4. Simplified Conduction band diagram of the structure under different bias conditions.
TE
Fig.5. Plot of ln( Ibackground ) versus V0.5 from background current voltage curve, dot-line is
AC
CE P
linear fitting.
ACCEPTED MANUSCRIPT Highlights
T IP SC R NU MA D TE
CE P
Two resonant tunneling peaks with current dropping gradually were observed. The EL intensity of the structure under resonant tunneling peak voltage is weakened. P-F tunneling is the main transport mechanism besides resonant tunneling.
AC
S0040-6090(14)01294-2 doi: 10.1016/j.tsf.2014.12.035 TSF 33999
To appear in:
Thin Solid Films
Received date: Revised date: Accepted date:
4 June 2014 1 November 2014 19 December 2014
Please cite this article as: D.Y. Chen, Y. Sun, Y.Y. Wang, Y.J. He, G. Zhang, Effect of resonant tunneling on electroluminescence in nc-Si/SiO2 multilayers-based p-i-n structure, Thin Solid Films (2014), doi: 10.1016/j.tsf.2014.12.035
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.
ACCEPTED MANUSCRIPT Effect
of
resonant
tunneling
on
electroluminescence
in
nc-Si/SiO2
multilayers-based p-i-n structure D. Y. Chen1,2*, Y. Sun1, Y. Y. Wang1, Y. J. He1, G. Zhang1 Nanjing University of posts and Telecommunications, Nanjing 210046, China
T
1* 2
Nanjing National Laboratory of Microstructures and Key Laboratory of Advanced Photonic and Electronic
[email protected]
Abstract
SC R
tel: +86-18936005695
IP
materials, Department of Physics, Nanjing University, Nanjing 210093, China
P-i-n structures with SiO2/nc-Si/SiO2 multilayers as intrinsic layer were prepared in conventional enhanced
chemical
vapor
deposition
system.
NU
plasma
Their
carrier
transport
and
electroluminescence properties were investigated. Two resonant tunneling related current peaks
MA
with current dropping gradually under forward bias were observed in the current voltage curve. Non-uniformity of the interfaces might be in responsible for the gradually dropping of the current.
D
Electroluminescence intensity of the device under bias of 7 V which is near the resonant tunneling
TE
peak voltage of 7.2 V was weaker than that under 6.5 V. According to the Gaussian fitting results of the spectra, the intensity of the sub-peak of 650 nm originating from recombination of injected
CE P
electrons and holes was decreased the most. When resonant tunneling conditions are met, it might be that most of injected electrons participate in resonant tunneling and fewer in Pool-Frenkel tunneling which is the main carrier transport mechanism to contribute to electroluminescence
AC
intensity.
Keywords: resonant tunneling; electroluminescence; SiO2/nc-Si/SiO2 multilayers; Poole-Frenkel tunneling;
1. Introductions Resonant tunneling has been reported and studied extensively in GaAs/AlGaAs [1-3], Si/SiGe [4-5], single-crystalline-Si/amorphous-SiO2 system [6-7], nc-Si embedded in dielectric matrices such as SiO2 or SiNx [8-14]. And SiO2/nc-Si/SiO2 system has some advantages such as process compatibility with microelectronic technology, higher recombination possibility for the small distance between wave functions of electrons and holes than that in bulk silicon, and so on.
ACCEPTED MANUSCRIPT SiO2/nc-Si/SiO2 system has been studied for its potential applications in light emitting diode (LED), optoelectronic integrated circuits [15–19] and other areas. For example, SiO2/nc-Si/SiO2 system was deposited on patterned substrate to increase carrier injection possibility [20] and
IP
T
achieve higher electroluminescence (EL) efficiency. In another way, SiO2/nc-Si/SiO2 system was used as intrinsic layer to construct p-i-n structure to obtain higher efficient LED [21]. Besides,
SC R
SiO2/nc-Si/SiO2 system has also been studied for its potential application in high speed switch circuits [22-23] for their high peak to valley current ratio (PVCR) of 60. The thicknesses of the nc-Si and SiO2 sub-layers and film quality both play important roles in resonant tunneling devices,
NU
which affect both the peak voltage and PVCR values. But, what will happen to the EL in SiO2/nc-Si/SiO2 multilayers-based p-i-n structure if resonant tunneling takes place? In this work,
MA
resonant tunneling and the effect on EL in the p-i-n structures will be studied in details. 2. Experimental Details
D
P-i-n structures were prepared in a conventional plasma enhanced chemical vapor deposition
TE
system with radio frequency (r.f.) of 13.56 MHz. The r.f. power and substrate temperature were kept at 50 W and 250 ℃, respectively. Amorphous Si/SiO2 (α-Si/SiO2) multilayers of six periods
CE P
were prepared by alternating α-Si deposition and in situ plasma oxidation processes. Silane was used to deposit α-Si with flow rate of 5 sccm. And oxygen was used to oxidize the α-Si with flow rate of 20 sccm. The deposition and oxidation time is 40 s and is 4 min, respectively. Phosphine
AC
and Borane were used as doping precursor of the n-type and p-type α-Si in the p-i-n structure, the corresponding conductivities are about 3x10-3 S·cm-1 and 5x10-5 S·cm-1, respectively, measured using Travelling Wave Method. The thickness of n-type and p-type α-Si are around 30 nm and 50 nm, respectively, according to the transmission electron microscope (TEM) image of the structure in figure 1. The as-deposited p-i-n structures were dehydrogenated at 450 ℃ for 1 h firstly, then were rapid thermal annealed at 1000 ℃ for 50 s and finally were furnace annealed at 1100 ℃ for 1h to form SiO2/nc-Si/SiO2-based p-i-n structure. All the treatments were carried out in N2 gas. In this work, single crystal p-Si wafers with resistivity of 2~8 Ω·cm were used as substrates. After thermal treatments, aluminum (Al) pad was evaporated as electrodes on both sides of the devices, and ring-like mask was used in top electrode evaporating process to form ring-like Al
ACCEPTED MANUSCRIPT pad as shown in our previous work [21] for light extracting in EL measurement. And EL spectra
NU
SC R
IP
T
were recorded by FluoroMax-2 under forward bias only.
Inset is the high resolusion image. 3. Results and Discussions
MA
Fig. 1. TEM image of the p-i-n structure with SiO2/nc-Si/SiO2 multi-layers as the intrinsic layer.
D
The deposition parameters of the intrinsic SiO2/nc-Si/SiO2 multilayers were the same as that
TE
of the sample C in our previous work [23], the thickness of nc-Si and SiO2 were about 6 nm and 5.2 nm, respectively. Effect of p-i-n structure to the EL of the SiO2/nc-Si/SiO2 multilayers has
CE P
been studied in our previous work [21], with turn-on voltage lowered and EL efficiency enhanced due to the reduced carrier tunneling barrier. EL properties of the present p-i-n structure are shown in figure 2 (a). The turn-on voltage is 6.5 V, which is higher than that in our previous work [21]
AC
for its thicker SiO2 layers. The EL spectra of the device are Gaussian fitted with three sub-peaks. Figure 2 (b) gives the fitting results of the EL spectrum under bias of 9 V. One sub-peak is around 800 nm. Its peak position and intensity changed slightly with bias. The origination may be the Si=O double bond [[24-25]. As for the other two sub-peaks, they are around 535 nm and 650 nm, respectively. The originations may be hot electron relaxation [26-27] for the former proposed by Pavesi group and recombination of injected electrons and holes in nc-Si/SiO2 multilayers within nc-Si and/or via the luminescence centers near Si/SiO2 interfaces [28-29] for the latter. The evolution of the two sub-peaks’ positions and intensities with bias are plotted in figure 2 (c) and (d). As bias increased from 7 V to 9 V, both the positions of the two sub-peaks changed slightly, yet their intensities increased continuously. When bias is below 7V and above 9 V, both the sub-peaks’ positions shifted obviously. The origination of the shift will be investigated in future. Here we might conclude that the shift might have nothing to do with resonant tunneling, and may
ACCEPTED MANUSCRIPT be related to the applied bias. From the plots of the intensities evolution in figure 2 (d), a sharp dip can be observed in the plot (black dot line) of the sub-peak of 650 nm, and the corresponding bias is 7 V which is near the resonant tunneling peak voltage of around 7.2V as shown in the
IP
T
current-voltage (IV) curve in figure 3. Slightly decreasing can be observed in the plot of the other sub-peak of 535 nm at the same bias. Thus the sub-peak of 650 nm could answer for the effect of
CE P
TE
D
MA
NU
SC R
resonant tunneling on EL intensity.
Fig. 2. EL spectra of the p-i-n structure (a); Gaussian fitting results of the EL spectra under bias of
AC
9 V (b); plots of the two sub-peaks’ positions with bias (c) and plots of intensities with bias (d).
Fig. 3. The measured IV curve (solid line) and the approximate background IV curve (dash line) of the p-i-n structure. Two broad resonant tunneling current peaks can be seen in the IV curve in figure 3. The peak
ACCEPTED MANUSCRIPT voltages are around 7.2 V and 12.8 V, the corresponding PVCR values are about 2 and 1.03, respectively. In our previous work we have reported on resonant tunneling in SiO2/nc-Si/SiO2 system. Here we could analyze the current peak using energy band diagram as shown in figure 4. It is well known that the resonant tunneling current could be approximately expressed as the
IP
I E lf En
T
following formular [22]:
SC R
Where, I is the resonant tunneling current, and E lf is the Fermi level in the emitter,
En (n=1, 2……) is the energy level in the quantum well, here in this work is nc-Si grains. The
NU
value of ( E lf En ) is in direct proportion to the applied bias. Thus I should be in linear relation
MA
with the bias. This is conformed in our experiment by the dot lines of b-c and d-e in the measured IV curve in figure 3. When energy level En is lower than Ecl (conduction band bottom of the emitter), resonant tunneling stops immediately, resulting in current dropping sharply as in our
D
previous work [23]. While in this work, current drops gradually with bias, and the IV curve is
TE
nearly symmetric, which looks similar as the band to band tunneling in Esaki’s diode [1]. In order to analyze carrier transport in the device, we consider a simplified structure with
CE P
two SiO2 barriers and one nc-Si well layer to construct the band diagram as in figure 4. Where, Ecl and Ec are conduction band bottoms of the emitter and bulk silicon, respectively;
E1 and E2
are quantized energy level in nc-Si quantum well. Panel (a) is under thermal equilibrium condition
AC
(without considering the interface states), corresponding to point “a” in IV curve in figure 3. As bias increases, E lf is pulled up to be equal to and then greater than E1 as in panel (b), corresponding to the first resonant tunneling process, from point “b” to “c” in figure 3. When bias increases further, Ecl is greater than E1 as in panel (c), the first resonant tunneling stops, and current drops with bias. If bias increases further on, the second resonant tunneling will occur and induces the second current peak of d-e-f as shown in figure 3. In the p-i-n structure here, the interfaces are non-uniform as can be seen clearly in the TEM image as shown in figure 1, which might lead to the quantized energy levels in the every quantum dot of nc-Si grain in the same quantum well layer being unequal to one another, resulting in forming of sub-band-like energy level, therefore, resonant tunneling in the structure behaves like band-to-band tunneling as in Esaki’s diode. Thus, as shown in figure 3 from “c” to “d” and “e” to “f”, the current dropped gradually rather than sharply when bias is higher than the peak voltage.
SC R
IP
T
ACCEPTED MANUSCRIPT
CE P
TE
D
MA
NU
Fig. 4. Simplified Conduction band diagram of the structure under different bias conditions.
Fig.5. Plot of ln( Ibackground ) versus V0.5, dot line is for linear fitting.
AC
In order to investigate the effect of resonant tunneling on the EL behavior, the background current (defined as current originating from other than resonant tunneling mechanisms) was estimated by dash-lining the valley current points of “a”, “d” and “f” smoothly as shown in figure 3. Plot of ln( Ibackground ) versus V0.5 was given in figure 5. A linear relationship can be found when the bias is larger than 2.5 V from the straight dot line in figure 5, suggesting that Poole-Frenkel tunneling (PF tunneling) is the main transport mechanism [30-31] besides resonant tunneling, i.e., PF tunneling contributes to the background current and EL property of the device. When bias is 7 V, which is near the resonant tunneling peak voltage of 7.2 V, most of the carriers injected from the emitter may participate in the fast process of resonant tunneling, thus fewer in PF tunneling, resulting in weak EL intensity. When higher bias is applied, carrier energy in emitter misaligns with the quantized energy level E1 in nc-Si layer, resonant tunneling stops and more
ACCEPTED MANUSCRIPT and more carriers can take part in PF tunneling to contribute to EL. On the other hand, The EL sub-peak of 650 nm under 7 V is affected the most from the dip in figure 2 (d). This sub-peak originates from recombination of injected electrons and holes in nc-Si/SiO2 multilayers within
IP
T
nc-Si and/or via the luminescence centers near Si/SiO2 interfaces. Resonant tunneling is a fast process, the tunneling time is less than 1ps [22], therefore electrons that participate in resonant
SC R
tunneling have not enough time to recombine with holes and thus make no contribution to EL at sub-peak of 650nm, leading to weaker EL intensity under resonant tunneling peak voltage. 4. Conclusions
NU
Carrier transport mechanisms and EL in SiO2/nc-Si/SiO2 multilayer-based p-i-n structure were studied. Resonant tunneling under forward bias of 7.2 V and 12.8 V, with PVCR value of 2
MA
and 1.03, respectively, can be observed in the IV curve. EL intensity under bias of 7 V is dramatically decreased in comparison with that of 6.5 V. The EL spectra was Gaussian fitted with
D
three peaks of 800 nm, 650 nm and 535 nm, and the intensity of the one of 650 nm is decreased
TE
the most, which originates from recombination of injected electrons with holes in multi-layers. When resonant tunneling takes place, most of the injected electrons from the emitter take part in
CE P
the fast process of resonant tunneling, thus fewer in PF tunneling to recombine with holes to contribute to EL. Therefore EL intensity under bias near resonant tunneling peak voltage is decreased greatly. On the other hand, the non-uniformity of the interfaces make the quantized
AC
energy level in the every nc-Si dot unequal to one another in the same well layer, leading to sub-band-like energy level forming and resulting in current dropping gradually in IV curve . ACKNOWLEDGMENTS This work was supported by NSFC (Nos. 61036001, 10874070), “973” project (2007CB613401), NSF of Jiangsu Province (BK2010010, 14KJB510025) and the Fundamental Research Funds for the Central Universities (1112021001).
ACCEPTED MANUSCRIPT References [1] R. Tsu and L. Esaki, R. Tsu and L. Esaki, Tunneling in a finite superlattice, Appl. Phys. Lett., 22 (1973) 562 [2] T. C. L. G. Sollner, W. D. Goodhue, P. E. Tannenwald, C. D. Parker, and D. D. Peck,
T
Resonant tunneling through quantum wells at frequencies up to 2.5 THz, Appl. Phys. Lett., 43
IP
(1983) 588
[3] M. Tsuchiya and H. Sakaki, Dependence of resonant tunneling current on well widths in
SC R
AlAs/GaAs/AlAs double barrier diode structures, Appl. Phys. Lett., 49 (1986) 88 [4] H. C. Liu, D. Landheer, M. Buchanan, and D. C. Houghton, Resonant tunneling in Si/Si1−x Ge x double‐barrier structures, Appl. Phys. Lett., 52 (1988) 1809
barrier diodes, Appl. Phys. Lett., 59 (1991) 973
NU
[5] K. Ismail, B. S. Meyerson, and P. J. Wang, Electron resonant tunneling in Si/SiGe double
[6] Y. Ishikawa, T. Ishihara, M. Iwasaki, and M. Tabe, Negative differential conductance due to
MA
resonant tunnelling through SiO2/single-crystalline-Si double barrier structure, Electron. Lett., 37 (2001) 1200
[7] H. Ikeda, M. Iwasaki, Y. Ishikawa, and M. Tabe, Resonant tunneling characteristics
D
in double-barrier structures in a wide range of applied voltage, Appl. Phys. Lett., 83 (2003) 1456
TE
[8] Y. Ishikawa, H. Ikeda, and M.Tabe. Potential-well-roughness-induced transition from resonant tunneling to single-electron tunneling in Si/SiO2 double-barrier structure, Appl. Phys. Lett., 86 (2005) 013508
CE P
[9] B. Berghoff, S. Suckow, R. Rölver, B. Spangenberg, H. Kurz, A. Dimyati, and J. Mayer, Resonant and phonon-assisted tunneling transport through silicon quantum dots embedded in SiO2, Appl. Phys. Lett., 93 (2008) 132111
AC
[10] Z. Pei, A. Y. K. Su, H. L. Hwan, and H. L. Hsiao, Room temperature tunneling transport through Si nanodots in silicon rich silicon nitride, Appl. Phys. Lett., 86 (2005) 063503 [11] M. Fukuda, K. Nakagawa, S. Miyazaki, and M. Hirose, Resonant tunneling through a self-assembled Si quantum dotAppl. Phys. Lett., 70 (1997) 2291 [12] A. Surawijaya, H. Mizuta, and S. Oda, Observation and Analysis of Tunneling Properties of Single Spherical Nanocrystalline Silicon Quantum Dot, Jpn. J. Appl. Phys., 45 (2006) 3638 [13] G. R. Lin, The Structural and Electrical Characteristics of Silicon-Implanted Borosilicate Glass, Jpn. J. Appl. Phys., 41 (2002) L1379-L1382 [14] G. R. Lin, C. J. Lin, C. K. Lin, L. J. Chou and Y. L. Chueh, Oxygen defect and Si nanocrystal dependent
white-light
and
near-infrared
electroluminescence
of
Si-implanted
and
plasma-enhanced chemical-vapor deposition-grown Si-rich SiO2, J. Appl. Phys., 97 (2005) 094306 [15] C. Delerue, G. Allan and M. Lannoo, Theoretical aspects of the luminescence of porous silicon,Phys. Rev. B, 48 (1993) 11024 [16] M. Zacharias, J. Heitmann, R. Scholz and U. Kahler, Size-controlled highly luminescent
ACCEPTED MANUSCRIPT silicon nanocrystals: A SiO/SiO2 superlattice approach, Appl.Phys. Lett., 80 (2002) 661 [17] B. Averboukh, R. Huber, K. W. Cheah, Y. R. Shen, G. G. Qin, Z. C. Ma and W. H. Zong, Luminescence studies of a Si/SiO2 superlattice, J. Appl. Phys., 92 (2002) 3564 [18] J. Xu, L. Yang, Y. J. Rui, J. X. Mei, X. Zhang, W. Li, Z. Ma, L. Xu, X. F. Huang and K. J.
T
Chen, Photoluminescence characteristics fromamorphous SiC thin films with various structures
IP
deposited at lowtemperature, Solid State Commun., 133 (2005), 565-568
[19] R. Huang, K. Chen, P. Han, H. Dong, X. Wang, D. Chen, W. Li, J. Xu, Z. Ma and X. Huang,
SC R
Strong green-yellow electroluminescence from oxidized amorphous silicon nitride light-emitting devices, Appl. Phys. Lett., 90 (2007) 093515
[20] D. Chen, Y. Liu, J. Xu, D. Wei, H. Sun, L. Xu, T. Wang, W. Li, K. Chen, Improved emission electroluminescent
device
containing
nc-Si/SiO2
multilayers
by using
NU
efficiency of
nano-patterned substrate, Opt. Exp., 18 (2010) 917
[21] D. Y. Chen, D. Y. Wei, J. Xu, P. G. Han, X. Wang, Z. Y. Ma, K. J. Chen, W. H. Shi and Q.
MA
M. Wang, Enhancement of electroluminescence in p–i–n structures with nano-crystalline Si/SiO2 multilayers, Semicond. Sci. Technol., 23 (2008) 015013 [22]L. Du, Y. Q. Zhuang, Nanoelectronics, first ed., publish house of electronics industry, Beijing,
D
2004, pp. 277
TE
[23] D. Y. Chen, Y. Sun, Y. J. He, L. Xu, and J. Xu, Resonant tunneling with high peak to valley current ratio in SiO2/nc-Si/SiO2 multilayers at room temperature, J. Appl. Phys., 115 (2014)
CE P
043703
[24] M. Luppi and S. Ossicini, J. Appl. Phys. 94 (2003) 2130 [25] M. Kulakci, U. Serincan and R. Turan, Semicond. Sci. Technol. 21 (2006) 1527 [26] Z.Gaburro, G. Pucker, P. Bellutti, and L. Pevesi;Solid State Commun., 114 (2000) 33
AC
[27] G. Pucker, P. Bellutti, M. Gazzanelli, Z.Gaburro and L. Pevesi, Opt. Mater.; 17 (2001)27 [28] G. R. Lin, C. J. Lin, and C.K. Lin, J. Appl. Phys., 97 (2005) 094306 [29] Y. Kanemitsu, Phys. Rep., 263 (1995) 1 [30] E. A. Morais, L. V. A. Scalvi, S. J. L. Ribeiro, and V. Geraldo, Poole–Frenkel effect in Er doped SnO2 thin films deposited by sol-gel-dip-coating, Phys. Stat. Sol. (a), 202 (2005) 301–308 [31] W.R. Harrella, J. Freyb, Observation of Poole-Frenkel effect saturation in SiO2 and other insulating , Thin Solid Films, 352 (1999) 195-204
ACCEPTED MANUSCRIPT Figure captions:
T
Fig. 1. TEM image of the p-i-n structure with SiO2/nc-Si/SiO2 multi-layers as the intrinsic layer.
IP
Fig. 2. EL spectra of the p-i-n structure (a); Gaussian fitting results of the EL spectra under bias of
SC R
9 V (b); plots of the two sub-peaks’ positions with bias (c) and plots of intensities with bias (d).
Fig. 3. The measured IV curve (solid line) and the approximate background IV curve (dash line) of
MA
NU
the p-i-n structure.
D
Fig. 4. Simplified Conduction band diagram of the structure under different bias conditions.
TE
Fig.5. Plot of ln( Ibackground ) versus V0.5 from background current voltage curve, dot-line is
AC
CE P
linear fitting.
ACCEPTED MANUSCRIPT Highlights
T IP SC R NU MA D TE
CE P
Two resonant tunneling peaks with current dropping gradually were observed. The EL intensity of the structure under resonant tunneling peak voltage is weakened. P-F tunneling is the main transport mechanism besides resonant tunneling.
AC