This content has been downloaded from IOPscience. Please scroll down to see the full text.
Download details:
IP Address: 140.113.38.11
This content was downloaded on 25/04/2014 at 16:07
Please note that terms and conditions apply.
Fabrication of Whitely Luminescent Silicon-Rich Nitride Films by Atmospheric Pressure
Chemical Vapor Deposition
View the table of contents for this issue, or go to the journal homepage for more 2008 Jpn. J. Appl. Phys. 47 4696
(http://iopscience.iop.org/1347-4065/47/6R/4696)
Fabrication of Whitely Luminescent Silicon-Rich Nitride Films
by Atmospheric Pressure Chemical Vapor Deposition
Chia-Hung LIN, Wu-Yih UEN, Yen-Chin HUANG, Zhen-Yu LI2, Sen-Mao LIAO, Tsun-Neng YANG1, Shan-Ming LAN1, and Yu-Hsiang HUANG1
Department of Electronic Engineering, College of Electrical Engineering and Computer Science, Chung Yuan Christian University, Chungli 32023, Taiwan
1Institute of Nuclear Energy Research, P.O. Box 3-11, Lungtan 32500, Taiwan
2Department of Photonics and Institute of Electro-Optical Engineering, National Chiao Tung University, 1001 TA Hsueh Road, Hsinchu 30010, Taiwan
(Received October 21, 2007; accepted February 4, 2008; published online June 13, 2008)
Silicon-rich nitride (SRN) films that can exhibit an intense white-light emission were fabricated by atmospheric pressure chemical vapor deposition. SRN films were deposited on Si substrates using gaseous SiH2Cl2(DCS) and NH3as the source
materials for Si and N, respectively. The deposition temperature was kept at 850C, and H
2was used as the carrier gas with its
flow rate modulated to maintain chamber pressure at 1 atm during the deposition. The optical properties of films obtained at various deposition times from 15 to 60 min were examined by photoluminescence (PL) measurement. An intense luminescence band (1.5 – 3.5 eV) was observed by the naked eye for all as-deposited samples. Besides, time-resolved PL exhibited a short radiative lifetime of about 1 ns for SRN films. Moreover, high resolution plan-view transmission electron microscopy demonstrated the existence of Si dots in SRN films with the dot sizes ranging from 2 to 6 nm and a dot density of about 4 1012/cm2. On the basis of the results obtained, we considered that the related luminescence mechanism for SRN films is connected to crystalline Si dots produced therein. [DOI:10.1143/JJAP.47.4696]
KEYWORDS: SiNx films, AP-HCVD, Si-QDs, HRTEM
1. Introduction
In recent years, Si-based light-emitting diodes (LEDs) have become very popular owing to their low cost of fabrication and easy integration with conventional IC technology compared with the present high-efficiency III– Vs-based LEDs. This naturally makes Si-based light-emit-ting materials some of the best for realizing optoelectronic integrated circuits (OEICs). The discovery of bright photo-luminescence (PL) from electrochemically etched porous silicon (PSi) has led to much research on fabricating efficient Si-based LEDs.1–4)However, the application of PSi has been limited by its unstable optical properties and physical fragility. Accordingly, other approaches to fabricating Si-based LEDs on silicon oxide (SiOx) have gradually been
developed, including the use of native SiOx (NSO)5–8) and
Si-rich oxide (SRO),9,10) doping or ion-implanting SiO
2
with various atoms, such as Si, Ge, Er3þ, and others,11–13) inserting multiple quantum wells into a SiOx film,14) and
growing Si/SiO2 superlattices.15) However, for LEDs
fab-ricated from a SiOxmaterial, the applied working voltage is
too high, and will be an issue for practical use.
Besides the materials mentioned above, silicon nitride (SiNx) is also an alternative for Si-based LEDs. In particular,
the potential barrier at a Si/Si3N4 interface is much lower
than that at a Si/SiO2 interface.16)Therefore, for Si/SiNx,
conditions for carrier injection can be improved, so SiNxhas
greater potential for fabricating white-light LED.17)
In using a SiNxmaterial for white-light LEDs, basically a
silicon-rich nitride (SRN) film with crystalline or amorphous Si quantum dots (QDs) embedded therein must be formed. So far, almost all these types of film have been fabricated by plasma-enhanced chemical vapor deposition (PECVD).17–23) The main advantage of the PECVD method is its relatively low deposition temperatures (usually 400C). Compared
with PECVD, atmospheric pressure CVD (APCVD) is another conventional deposition method for SiNx films but
can be a simpler fabrication process and cost-competitive for device applications of the SRN structure. Additionally, higher deposition temperatures in APCVD can be expected for achieving better material quality, such as lower pinhole density and better adhesion to the substrate. Nevertheless, the use of APCVD for fabricating whitely luminescent films has seldom been studied and relevant works are scarce.24) Hence, in this paper, we present the results of our investigation on SRN films fabricated by APCVD. In particular, to our knowledge, this is the first time to recognize that APCVD can be used to fabricate SRN films with crystalline Si QDs embedded therein.
2. Experiment
The samples were prepared using a home-made one-flow APCVD system with a vertical chamber. The substrates used were 2-in. (111)-oriented, 1– 2 cm p-type Si wafers. Before loading, the substrate was cleaned by boiling it in H2SO4: H2O2¼3 : 1 for 15 min and then dipped in HF
solution (HF : H2O ¼ 1 : 10) for 15 s to obtain a stable
hydrogen-passivated surface. Dichlorosilane (SiH2Cl2, DCS)
and ammonia (NH3) were used as the source materials for Si
and N, respectively, and H2 was the carrier gas. After
loading, the substrate was initially heated to 1100C in H2
ambient for 7 min, to remove the passivated layer on the surface. Then, SRN films were grown at a constant temper-ature of 850C and at a constant flow ratio of ½NH
3=
½DCS ¼ 1:2 in an H2environment. The film deposition time
was varied from 15 to 60 min.
The optical properties were examined by room-temper-ature photoluminescence (PL) measurement. PL spectra were excited with the 325 nm line of a He–Cd laser at excitation power densities of 0.1– 0.25 kW/cm2. The
lumi-nescence was dispersed by a 1 m monochromator and detected using a water-cooled GaAs photomultiplier.
More-
E-mail address: [email protected]; [email protected] #2008 The Japan Society of Applied Physics
over, radiative lifetime was determined from time-resolved photoluminescence (TRPL) measurement at room temper-ature. In this case, the signal was excited by a GaN semiconductor laser ( ¼ 396 nm) with a pulse width of about 0.15 ns and a power density of 0.2 W/cm2. The thickness of SRN films was determined by scanning electron microscopy (SEM). The microstructure of the films was studied by plan-view high resolution transmission electron microscopy (HRTEM) using a Philips Tecnai F20 micro-scope operated at 200 keV. Specimens used for HRTEM were first chemically etched, then mechanically polished to <5 mm and finally ion-milled to achieve electron trans-parency.
3. Results and Discussion
Figure 1 shows the room-temperature PL spectrum of a typical SRN film deposited for 40 min (referred to below as sample SRN-A). As can be observed, a strong white luminescence band covers a range from 1.5 to 3 eV with three subpeaks at 2.08, 2.45, and 2.81 eV, individually. Two intervals among these peaks are 360 and 370 meV. Consid-ering these large energy deviations, it is plausible that the aforementioned subpeaks did not result from Si QDs of different sizes nor the transitions between different energy levels in the same Si QD. The origin of these subpeaks will be analyzed in detail after Fig. 4 is presented. Now, to the spectrum shown in Fig. 1 a Gaussian distribution is fitted over the 1.5 – 3.5 eV range with the peak at 2.5 eV. Obviously, the luminescence peak energy is blue-shifted
from that of crystalline Si bulk (1.12 eV) by 1.38 eV. This result is considered to be related to the quantum confinement effect of Si QDs embedded in the matrix of amorphous SiNx with a band-gap energy of approximately 5 eV,25) as
examined from both theoretical and experimental points of view described below.
As was reported, the Gaussian distribution of PL spectra from Si QDs demonstrates theoretically an ensemble of Si QDs of various sizes.26) Otherwise, the effective band-gap energy of a Si QD modeled by a block with three unequal sizes, d1, d2, and d3 can be given by27)
Egap¼Egap;0þ 2h 2 2 m1 l þm1h d2 1 þm 1 t þm? 1 h d2 2 þm 1 t þm? 1 h d2 3 ! ; ð1Þ
where Egap;0 is the indirect band gap of Si bulk at 300 K. As
can be noted from the subsequent HRTEM image displaying Si QDs, it is reasonable to assume a spherical QD with an equivalent three-dimensional (3D) size d ¼ d1¼d2 ¼d3
and then eq. (1) can be simplified to Egap¼Egap;0þ
C
d2; ð2Þ
where C is the confinement parameter.18,20,21)
The curves shown in Fig. 2 show the relationship between the effective band-gap energy and the size of a spherical QD for both crystalline and amorphous Si QDs. The curve in solid line is plotted for the former using eq. (2) with Egap;0¼1:12 eV and C 8:845 determined using the bulk
parameters of Si given in ref.27. Otherwise, the curve in dash line at the bottom is the same plot for the latter, as reported in refs.18, 20, 21. Furthermore, for comparison, the plot in dot-dash line at the top is a copy of the curve fitting to the relationship between PL peak energy and the flow rate of [SiH4]/[N2] for crystalline Si QDs examined
in ref.20. In the same reference the fitting parameters of Egap;0 and C are resolved to be about 1.16 eV and 11.8,
respectively. Using these curves and the parameters obtained from the Gaussian fitting to the PL spectrum, one can analyze the size distribution of Si QDs. The range (1.5 – 3.5 eV) and peak position (2.5 eV) of the Gaussian fitting curve shown in Fig. 1 are indicated in Fig. 2 by two open
circles and one closed circle, respectively, with their measurement errors indicated. It is found that the lumines-cence energies of Si QDs in sample SRN-A can be well described by eq. (2) for the case of crystalline Si QD, from which information on a size distribution range of 2 – 6 nm for the crystalline Si QDs and that on the largest number of dots with a size of 3 nm were extracted.
On the other hand, for a direct observation of nano-structures, Fig. 3 shows a plan-view HRTEM image of
1.6 1.8 2.0 2.2 2.4 2.6 2.8 3.0 0 5000 10000 15000 20000 25000 30000 RT-PL SRN-A Gaussian PL Intensity (arb.unit)
Energy Gap (eV)
2.08 eV 2.45 eV 2.81 eV
0.37 eV 0.36 eV
2.5eV
Fig. 1. (Color online) Room-temperature PL spectrum of sample SRN-A and Gaussian fitting to it.
0 2 4 6 8 10 0 2 4 6 8 10 12 14 Dot size (nm)
Energy gap (eV)
Theoretical calculation for c-Si dots Theoretical calculation for a-Si dots Theoretical calculation by Ref.20
presented by the fitting curve in Fig.1
1.5 eV 2.5 eV
3.5 eV
Fig. 2. (Color online) Relationship between effective band-gap energy and size of Si QD.
sample SRN-A showing a wide uniform distribution of dark nanoparticles. The top left inset of the figure shows a magnified image of a dark nanoparticle, whereas the top right inset shows the transmission electron diffraction (TED) pattern of the SRN film. Evidently, the top left inset shows that the dark nanoparticles in the SRN matrix demonstrate crystalline Si QDs, which are embedded in the amorphous SRN film. This condition is further confirmed by the TED pattern where well-defined spots in multiple directions are visible in a hollow background, showing that monocrystal-line Si QDs in various orientations are generated in the SRN film. Several HRTEM images with scanning areas ranging from 50 50 to 100 100 nm2 according to different
magnifications were used to evaluate the average distribu-tion of Si QDs in the SRN film. Figure 4 shows the number distribution of Si QDs as a function of dot size. As can be observed, the Si QDs embedded in the SRN film were also found to follow the Gaussian distribution having dot sizes ranging from 1 to 7 nm. The density of Si QDs was determined to be about 4 1012/cm2. It therefore can be
said that the size distribution of Si QDs estimated theoret-ically from Fig. 2 coincides with that determined directly from HRTEM images, as described above.
As demonstrated in Fig. 1, the white luminescence band shows three subpeaks at 2.08, 2.45, and 2.81 eV. It was indicated previously that the multiple interference phenom-enon in PL spectra can be affected by film thickness.28)In the present case, these peaks were also considered to be caused by multiple interference effects in the SRN film. In order to confirm this, we kept the flow ratio of ½NH3=½DCS
and growth temperature constant at 1.2 and 850C,
respectively, but changed the growth time as 15, 30, 45, and 60 min to attain SRN films of different thicknesses, referred to below as samples SRN-B, SRN-C, SRN-D, and SRN-E, respectively. SEM images of these samples helped determine their film thicknesses to be 21.3, 269.4, 620.3, and
789.1 nm, respectively. Moreover, Fig. 5 shows the room-temperature PL spectra obtained from these films. As is shown, the appearance of subpeaks in the PL band becomes more evident as the SRN film becomes thicker. Nearly only one wide PL band is observed when the SRN film is thinner than 300 nm, whereas this wide band develops into two or three subbands with spacings of 360 and 355 meV when the SRN film is thicker than 600 nm.
Figure 6 shows the room-temperature TRPL spectra of the samples SRN-B, SRN-C, SRN-D, and SRN-E monitored at an energy of 2.48 eV (500 nm). Notably, normalized inten-sity is plotted on a semilogarithmic scale versus time on a linear scale. Evidently, the decay is not single exponential. The transient can be well fitted by a stretched exponen-tial,29–31) which is given by the following equation.
IðtÞ ¼ I0exp t " # ; ð3Þ
where IðtÞ is the PL intensity at time t and I0Ið0Þ. The
decay time constant is the average lifetime, and the dispersion factor varies between zero and unity, providing information on the recombination mechanism. Table I shows the average lifetime and dispersion factor extracted from the fitting of eq. (3) to the PL decay spectra shown in Fig. 6. It can be observed that lifetime increases from 0.56 to
Fig. 3. Plan-view HRTEM image of sample SRN-A. The top left insert shows a magnified view of a crystalline Si QD and the top right inset shows the transmission electron diffraction pattern of the SRN film.
2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 2.9 3.0 3.1 3.2 3.3 3.4 0 50000 100000 150000 200000 Intensity (arb.unit)
Energy Gap (eV)
RT-PL SRN-B SRN-C SRN-D SRN-E
Fig. 5. (Color online) Room-temperature PL spectra of samples SRN-B, SRN-C, SRN-D, and SRN-E. 1 2 3 4 5 6 7 0 5 10 15 20 25 30 Number of Si QD Dot size (nm)
Gaussian fitting peak: 4.036 nm
1.12 ns with increasing film thickness, whereas ranges between 0.5 and 0.6. In particular, the lifetime approaches 1 ns when the film thickness is higher than 2600 A˚ . The magnitude scale of the average lifetime and the stretched exponential decay behavior of PL intensity might also suggest the luminescence mechanism of the SRN film grown in the present study to be the quantum confinement effect.32) Additionally, it is worth noting that the lifetime seems to be influenced by film thickness. This is possibly because different sizes of Si QDs are formed when the SRN film is grown to different thicknesses. Nevertheless, to elucidate this in further detail, further study in this area must be conducted.
4. Conclusions
Silicon-rich-nitride films with Si QDs embedded inside were successfully deposited on the Si substrate by APCVD. PL measurements indicated an intense white luminescence band (1.5 – 3.5 eV) that can be observed by the naked eye and HRTEM observations demonstrated the existence of Si QDs in the grown films with dot sizes ranging from 2 to 6 nm and a dot density of about 4 1012/cm2. The white PL
band was considered to be basically associated with the quantum confinement effect of Si QDs; however, the PL spectral feature was also influenced by the multiple interference effects of the SRN film as the film was grown to over 600 nm. Additionally, a short radiative lifetime of about 1 ns was exhibited by time-resolved PL. The establish-ment of the experiestablish-mental procedure for obtaining such a strong white light emission may lead to a significant
progress towards the application of APCVD-grown SRN films to light-emission Si-based devices.
Acknowledgements
The authors would like to thank the National Science Council of Taiwan, for financially supporting this research under Contract NSC 95-2623-7-033-002-NU. The authors Uen and Lin are also grateful to the Institute of Nuclear Energy Research (INER) for technical assistance.
1) N. Koshida and H. Koyama:Appl. Phys. Lett. 60 (1992) 347. 2) B. Gelloz, T. Nakagawa, and N. Koshida:Appl. Phys. Lett. 73 (1998)
2021.
3) J. Linnros and N. Lallic:Appl. Phys. Lett. 66 (1995) 3048. 4) S. Lazarouk, P. Jaguiro, S. Katsouba, G. Mashini, S. La Monica, G.
Maiello, and A. Ferrari:Appl. Phys. Lett. 68 (1996) 1646.
5) G. G. Qin, Y. M. Huang, B. Q. Zong, L. Z. Zhang, and B. R. Zhang: Superlattices Microstruct. 16 (1994) 387.
6) G. F. Bai, Y. Q. Wang, Z. C. Ma, W. H. Zong, and G. G. Qin:J. Phys.: Condens. Matter 10 (1998) L717.
7) Y. Q. Wang, T. P. Zhao, J. Liu, and G. G. Qin:Appl. Phys. Lett. 74 (1999) 3815.
8) J. Yuan and D. Haneman:J. Appl. Phys. 86 (1999) 2358.
9) D. J. DiMaria, J. R. Kirtley, E. J. Pakulis, D. W. Dong, T. S. Kuan, F. L. Pesavento, T. N. Theis, J. A. Cutro, and S. D. Brorson:J. Appl. Phys. 56 (1984) 401.
10) G. G. Qin, A. P. Li, and Y. X. Zhang:Phys. Rev. B 54 (1996) R11122. 11) Y. Liu, T. P. Chen, Y. Q. Fu, M. S. Tse, J. H. Hsieh, P. F. Ho, and
Y. C. Liu:J. Phys. D 36 (2003) L97.
12) X. M. Wu, M. J. Lu, and W. G. Yao:Surf. Coat. Technol. 161 (2002) 92.
13) D. Pacifci, A. Irrera, G. Franzo´, M. Miritello, F. Iacona, and F. Priolo: Physica E 16 (2003) 331.
14) M. Zacharias, J. Bla¨sing, P. Veit, L. Tsybeskov, K. Hirschman, and P. M. Fauchet:Appl. Phys. Lett. 74 (1999) 2614.
15) M. Zacharias, J. Heitmann, R. Scholz, and U. Kahler:Appl. Phys. Lett. 80(2002) 661.
16) V. A. Volodin, M. D. Efremov, V. A. Gritsenko, and S. A. Kochubei: Appl. Phys. Lett. 73 (1998) 1212.
17) N. M. Park, T. S. Kim, and S. J. Park:Appl. Phys. Lett. 78 (2001) 2575.
18) N. M. Park, C. J. Choi, T. Y. Seong, and S. J. Park:Phys. Rev. Lett. 86 (2000) 1355.
19) K. S. Cho, N. M. Park, T. Y. Kim, K. H. Kim, G. Y. Sung, and J. H. Shin:Appl. Phys. Lett. 86 (2005) 071909.
20) T. Y. Kim, N. M. Park, K. H. Kim, G. Y. Sung, Y. W. Ok, T. Y. Seong, and C. J. Choi:Appl. Phys. Lett. 85 (2004) 5355.
21) T. W. Kim, C. H. Cho, B. H. Kim, and S. J. Park:Appl. Phys. Lett. 88 (2006) 123102.
22) L. D. Nergo, J. H. Yi, L. C. Kimerling, S. Hamel, A. Williamson, and G. Galli:Appl. Phys. Lett. 88 (2006) 183103.
23) M. Wang, D. Li, Z. Yuan, D. Yang, and D. Que:Appl. Phys. Lett. 90 (2007) 131903.
24) V. V. Vasilev, I. P. Mikhailovskii, and K. K. Svitashev:Phys. Status Solidi A 95 (1986) K37.
25) S. M. Sze: Physics of Semiconductor Devices (Wiley, New York, 1981) 2nd ed., p. 852.
26) F. Iacona, G. Franzo, and C. Spinella:J. Appl. Phys. 87 (2000) 1295. 27) J. B. Khurgin, E. W. Forsythe, G. S. Tompa, and B. A. Khan:Appl.
Phys. Lett. 69 (1996) 1241.
28) D. C. Marra, E. S. Aydil, S. J. Joo, E. Yoon, and V. I. Srdanov:Appl. Phys. Lett. 77 (2000) 3346.
29) K. Yamaguchi, K. Mizushima, and K. Sassa:Appl. Phys. Lett. 77 (2000) 3773.
30) Pavesi L:J. Appl. Phys. 80 (1996) 216.
31) A. Y. Kobitski, K. S. Zhuravlev, H. P. Wagner, and D. R. T. Zahn: Phys. Rev. B 63 (2001) 115423.
32) X. Chen, B. Henderson, and K. P. O’Donnell:Appl. Phys. Lett. 60 (1992) 2672. 0 5 10 15 20 25 30 35 0.00674 0.01832 0.04979 0.13534 0.36788 1 ln(I(t)/Io) SRN-B SRN-C SRN-D SRN-E Lifetime (ns)
Fig. 6. (Color online) Room-temperature TRPL spectra of samples SRN-B, SRN-C, SRN-D, and SRN-E monitored at an energy of 2.48 eV. Curves are normalized and plotted on a semilog vs linear scale.
Table I. Average lifetimes and dispersion factors extracted from fitting of eq. (3) to data shown in Fig. 6.
Sample Thickness (A˚ ) (ns) SRN-B 213 0.56 0.5 SRN-C 2694 0.98 0.5 SRN-D 6203 1.07 0.6 SRN-E 7891 1.12 0.6
