istics of the spectral sensitivity of each layer when the light intensity was 4 SUN and the irradiation and recovery times are, respectively, 205 and 517h, with the temperature being maintained at 50°C. It can be seen that the levels of degradation for the mid- dle and top layers are quite large compared to the degradation level of the bottom layer which is also the same as far as their recovery processes are concerned.
Fig. 3 shows the degradation and recovery characteristics of J,, for each layer under the same conditions as those for obtaining the results illustrated in Fig. 2. In this case the value of J,, is nor- malised to an initial value. The degradation of J,, for the middle layer is 12% while it is almost the same (6%) for both the top and bottom layers. The recovery of J,, for the middle layer is 6.5% followed by a value of 4% for each of the other two layers. From Figs. 1 and 3 it is evident that the nature of the curves depicting the degradation and subsequent recovery processes for both q and
Js, follow almost the same pattern: a rapid initial decrease followed by much slower advances in the case of degradation.
1.00 0 0.95 .- I
E
= 0.90 - b-
- - 6 8 0.85 2light intensity (SUN) 1225/41
Fig. 4 Optical intensity dependence of degradation characteristic of cur-
rent density after 20h at 70°C
A higher level of degradation occurs in the bottom and middle layers compared to that in the light-facing top layer. Fig. 4 shows the degradation characteristic of J,, for each layer under the fast test process with light intensities of 1,4 and 8 SUN, the exposure time being 23h. The decrease in the value of J,, is minimum for the top and highest for the middle layer. For the bottom layer, J,, decreases to a certain value until the light intensity becomes 4 SUN. But after that it becomes almost constant even when the intensity is increased to 8 SUN. The level of degradation of q cor- responds to that for the .Tqc of each layer. From Fig. 4 it is also evident that the deterioration does not depend on the light inten- sity after a certain value.
Conclusion: We have evaluated the degradation and subsequent recovery characteristics of each layer of a three-layer stacked a-Si solar cell by comparing the q and spectral sensitivity (and hence
JSJ. The degradation in spectral sensitivity and recovery for each
layer have been found to show a corresponding change in q and a strong correlation. It also establishes a normal phase barrier for the light intensity. As far as the degradation is concerned, its value is highest for the middle layer, followed by the top and bottom layers. It is evident that the degradation pattern for the bottom layer should be chosen as standard as far as the evaluation of the overall q is concerned. This is because the smallest absolute value of spectral sensitivity is obtained for the bottom layer. Moreover, solar cell degradation of the light-facing top layer occurs quickly and the degradation at this point is seen to reach a saturation level.
0 IEE 1999
Electronics Letters Online No: 19991351 D 01: IO. I049/el: 19991351
T. Kojima (1-1-4 Umezono Tsukuba-shi, Ibaraki 305-8568, Japan) 29 September I999
References
YANG, J., BANERJEE, A., and GAHA, s.: ‘Correlation of component cells with high efficiency amorphous silicon alloy triple-junction solar cells and modules’. 2nd WCEPV, Proc. PVSEC, 1998, pp. IGARI, s., NOSE, J . , NAKANO, A., SHIGEKUNI, T., and TERASHIMA, H.: ‘Output stability of various PV modules by long-term exposure test at JQA’. Tech. Dig. PVSEC-9, 1996, pp. 477478
TAKAHISA, K., KOJIMA, T., NAKAMURA, K., KOYANAGI, T., and YANGISAWA, T.: ‘Stabilized efficiency of stacked a-Si solar cells’. 2nd WCEPV, Proc. PVSEC, 1998, pp. 777-780
TAKAHISA, K., KOJIMA, T., NAKAMURA, K., KOYANAGI, T., and YANGISAWA, T.: ’Experimental model and long-term prediction of photovoltaic conversion efficiency of a-Si solar cells’, Solar Energy Mat. Solar Cells. 1997, 49, pp. 179-182
387-390
Formation of Si nanoclusters in amorphous
silicon thin films by excimer laser annealing
Jiun-Lin Yeh,
Hsuen-Li
Chen,An
Shih and Si-chenLee
It is shown that an Si nanocluster is formed in an amorphous silicon (a-Si) thin filmfollowing irradiation using a pulsed KrF excirner laser. The photoluminescence spectrum of the irradiated 70nm thick a-Si film at a power density of 180mJ/cm2 at one shot shows two luminescence bands centred at -1.31 and 1.76eV. The peak emission wavelength depends on the silicon nanocluster sue, whch is -Wnm. A mechanism for the formation of Si nanoclusters is also proposed.
The formation of nano-structural Si has attracted increasing inter- est since the discovery of an eficient light emission from porous silicon. Quantum confinement has been suggested as the cause of observed luminescence and enhanced nonlinear optical properties [l, 21. Various methods, including pulsed-laser ablation (PLD) [3], laser-induced gas phase reaction [4], size-exclusion methods [5],
ion implantation 161 and spark processing [7], have been used to prepare Si nanoclusters (nc-Si). In this Letter, we investigate the formation of Si nanoclusters in an amorphous silicon (a-Si) thin
f h
following irradiation using a pulsed KrF excimer laser. The preparation parameters such as the a-Si thickness are studied to understand their effect on the properties of Si nanoclusters, such as their size and size distribution. A mechanism for the formation of Si nanoclusters is also proposed.a
C d
Fig. 1 SEMpictures showing surface morphologies o f t h e 10, 30, SO and 70nm thick a-Si layers irradiated by KrF laser with power intensity of 180mJ/cm2 at one shot, respectively
a lOnm b 30nm c 50nm d 70nm
2058 ELECTRONICS LETTERS
I
lth NovemberI999
Vol. 35 No. 23Amorphous Si was deposited by plasma enhanced chemical vapour deposition (PECVD) at 250°C onto glass substrates using SiH4
+
H,. After deposition, the a-Si:H samples were annealed at 550°C for 5 mm in an N2 environment to remove all the hydrogen atoms from the films. The samples were then irradiated using a pulsed KrF excimer laser(A
= 248nm) at room temperature. After irradiation, nanoclusters were formed, the cluster sizes of which were characterised using scanning electron microscopy (SEM).The variation in the poly-Si cluster size was studied using a-Si layers of different thicknesses, i.e. 10, 30, 50 and 70nm. Figs. l a 4 show, respectively, the SEM pictures of the surface morphologies of the 10, 30, 50 and 70nm thick a-Si layer irradiated using a KrF laser at a power intensity of 180mJ/cm2 at one shot. It is clear that the size of the poly-Si clusters increases from 100 to 250nm as the thickness of the a-Si film increases from 10 to 30nm, then decreases from 250 to 3 5 n m as the thickness of the irradiated a-Si film is further increased from 30 to 50nm and stays almost at 35nm as the thickness of a-Si film increased from 50 to 70nm. It is clear that with a small increases in the thickness of the a-Si film of -30nm, a sharp decrease in the poly-Si cluster size occurs. We propose that when the 10 and 30nm thick a-Si layers on 7059 glass are irradiated by a single KrF laser pulse, they are melted completely by the irradiation and the film assumes a globular shape due to the surface tension. Since the a-Si films are melted completely, the thicker a-Si layer forms thicker globular nanoclus- ters. As the film thickness increases to 50nm and beyond, the heat generated by the KrF excimer laser may not be able to melt the a- Si film entirely. This leads to random nucleation of the poly-Si on the film surface, thus limiting the size of the cluster.
2
0.8 1.0 1.2 1.4 1.6 1.8 2.0 energy, eV 1214121
Fig. 2 Photoluminescence spectrum of irradiated 70nm thick a-Si film
with power density of 180mJ/cm2 ut one shot
The photoluminescence spectra were measured at 63K using the 488nm line of an argon laser as the excitation source. The laser output power was -3.7mW. Fig. 2 shows the photoluminescence spectrum of the KrF-irradiated 70nm thick a-Si film at a power density of 180mJ/cm2 at one shot. The spectrum shows two lumi- nescence bands centred at -1.31 and 1.76eV. Using theoretical cal- culation, the peak emission wavelengths are found to correspond to the respective ground and first excited state transition from conduction band to valence band within a silicon nanoparticle with a size of
-
3 4 n m . By adjusting appropriately the power den- sity and shot number of the KrF excimer laser condition, it is in theory possible to manufacture material with a peak emission tai- lored to the desired application.We propose a possible mechanism to explain the phenomenon. The heat generated by the KrF excimer laser melts only the sur- face of the 70nm thick a-Si film, thus forming a liquid-solid inter- face. The melted Si liquid is globular in shape due to the surface tension, and Si nanoparticles are also formed during the cooling- down period. The globular shape can be attributed to the fact that
the agglomeration force of the melted Si is greater than the adher- ence force to the substrate.
We conclude that the irradiation of an a-Si film using pulsed KrF excimer laser is found to form Si nanoclusters in an a-Si film. The heat generated by the KrF excimer laser melts the surface of the a-Si film, the melted a-Si film becomes globular in shape due to the surface tension, and Si nanoparticles are formed during the cooling-down period. From the photoluminescence spectrum of the 70nm thick a-Si film irradiated by the KrF laser at a power density of 180mJ/cm2 at one shot, two luminescence bands centred at
-
1.31 and 1.76eV are observed which correspond to the ground and first excited state transition of a silicon nanoparticle with a size of-
3 4 n m .Acknowledgment: This work has been supported by the National Science Council of the Republic of China under Contract No. NSC88-22 15-E-002-008.
0 IEE 1999
Electronics Letters Online No: 19991361 DOI: 10.1049/e1:1999I361
Jiun-Lin Yeh, Hsuen-Li Chen, An Shih and Si-chen Lee (Department of Electrical Engineering, National Taiwan University, Taipei, Taiwan, Republic of China)
8 September I999
References
KANEMITSU, S., OKAMOTO, s., and MITO, A.: ‘Photoluminescence mechanism of silicon quantum dots and wells’. Mat. Res. Soc. Symp. Proc. 1997,452, p. 195
HENARI, F.z., MORGENSTEM, K., BLAU, w.J., KARAVANSKII, v.A., and DNEPROVSKII, v.: ‘Third-order optical nonlinearity and all-optical switching in porous silicon’, Appl. Phys. Lett., 1995, 67, pp. 323 MAKIMURA, T., and MURAKAMI, K.: ‘Dynamics of silicon plume generated by laser ablation and its chemical reaction’, Appl. Surf Sei., 1996, 242, pp. 96-98
SMIMOV, v.v., STELMALDI, o.M., and SCHMIDT, R.: ‘Deposition and analysis of silicon clusters generated by laser-induced gas phase reaction’, J. Appl. Phys., 1995, 78, p. 5302
WILSON, w.L., SZAJOWSKI, P.F., and BRUS, L.E.: ‘Quantum confinement in size-selected, surface-oxidised silicon nanocrystals’, Science, 1993, 262, p. 1242
EHBRECHT, M., FERKEL, H., HUISKEN, F., HOLZ, L., POLIVANOV, Y.N.,
WHITE, C.W., BUDAI, J.D., WITHROW, S.P., ZHU, J.G., PENNYCOOK, S.J., ZHUR, R.A., HEMBREE,.D.M., HENDERSON, D.O., MAGRUDER, R.H., YACAMAN, M.J., MONDRAGON, G., and PRAWER, S.: ‘Encapsulated nanocrystals and quantum dots formed by ion beam sysnthesis’, Nucl. Instr. Meth. Phys. Res. Sec. B, 1997, 545, pp. 127-128 RUPP, s., QUILTY, J., TRODAHL, H.J., LUDWIG, M.H., and HUMMEL, R.E.: ‘Raman study of the relationship between nanoparticles and photoluminescence in spark-processed silicon’, Appl. Phys. Lett., 1997, 70, p. 723
Self-aligned implanted ground-plane fully
depleted
SO1
MOSFET
Weize Xiong and J.P. Colinge
A method for fabricating a back-gate ground plane underneath a t h i n - f h silicon-on-insulator (SOI) MOSFET is described. It is shown by numerical simulation that the formation of the ground plane improves the subthreshold slope and short-channel characteristics of very short-channel devices.
Introduction: It is well known that the dual-gate (top and bottom gate) silicon-on-insulator (SOI) MOSFET is the most suitable device structure for suppressing short-channel effects such as drain-induced barrier lowering (DIBL) and subthreshold slope degradation. Since such devices are difficult to fabricate, altema- tive structures where a back-gate ground plane is used to reduce short-channel effects have been proposed [l, 21. In such devices a grounded back-gate prevents the electric field lines originating at the drain from terminating under the channel region, which would cause DIBL and ultimately punch-through. In this Letter a simple, manufacturable process is proposed for fully-depleted (FD) SO1 MOSFETs with back-gate ground plane.
ELECTRONICS LETTERS 11th November 1999 Vol. 35 No. 23
