Franc¸ois TEMPLIER†a), Julien BROCHET†, Bernard AVENTURIER†, David COOPER†, Alexey ABRAMOV††, Dmitri DAINEKA††, and Pere ROCA i CABARROCAS††, Nonmembers
SUMMARY Hydrogenated polymorphous Silicon allows to fabricate TFTs with very interesting characteristics including better threshold volt-age stability than a-Si TFTs, lower leakvolt-age current than μc-Si:H TFTs and excellent uniformity. Investigation of threshold voltage shift mechanisms of pm-Si:H TFTs has shown a specific semiconductor material degradation with different activation energies compared to a-Si:H TFTs. TEM analysis has evidenced for the first time a significant structural difference between pm-Si:H and a-Si:H materials, in the TFT device configuration. Pm-Si:H appears to be very suitable for low cost and high performance AM-OLED fabrication.
key words: thin-film transistors, polymorphous, silicon, OLED 1. Introduction
Amorphous silicon (a-Si:H) thin film transistors (TFTs) are extensively used as switching element for active-matrix liq-uid crystal displays. However, one of the most serious problems of such devices is the low reliability evidenced by important shift of the threshold voltage, which makes these devices not suitable for active matrix OLED dis-plays, where a high stability is required for the driving TFT. For many years, microcrystalline (μc-Si:H, also sometimes called nanocrystalline silicon) has been evidenced as a good alternative to a-Si:H, providing much more stable devices, and slightly higher mobility [1], [2]. However, it has been shown that μc-Si:H TFTs can exhibit high leakage current, and that part of this leakage is intrinsically related to the smaller bandgap of μc-Si:H (Eg = 1.1 eV) compared to
a-Si:H (Eg = 1.8 eV) [3]. Some years ago, hydrogenated
polymorphous silicon (pm-Si:H), which consist of silicon nanocrystals embedded in a relaxed amorphous silicon ma-trix, has been proposed for solar cells but also for TFTs [2]. Later, it was shown that pm-Si:H TFTs can exhibit improved threshold voltage (Vth) stability compared to a-Si:H
coun-terparts, together with low drain leakage current [4]. Last year we presented first results on voltage shift mechanisms of pm-Si:H, unveiling specific behaviour [5]. In this paper, we will investigate Si:H TFTs, by describing i) the
pm-Manuscript received April 21, 2010. Manuscript revised May 21, 2010.
†The authors are with CEA-LETI Minatec, 17, rue des
Mar-tyrs, 38054 Grenoble, France.
††The authors are with LPICM, Ecole Polytechnique, CNRS,
91128 Palaiseau Cedex, France. a) E-mail: [email protected]
DOI: 10.1587/transele.E93.C.1490
Si:H depositions conditions, ii) the Vthshift mechanisms, iii)
the structural properties of the film in the TFT stack and iv) the corelation between structure and electrical properties of the TFTs.
2. Experimental
2.1 Deposition of pm-Si:H TFTs
Polymorphous silicon, a nanostructured thin film material is obtained under plasma conditions where secondary reac-tions take place in the plasma bulk. This usually happens when trying to increase the deposition rate by increasing the total pressure or by increasing the RF power. An exam-ple of the effect of the total pressure on the deposition rate and species present in a silane (SiH4) and hydrogen (H2)
RF glow discharge is shown in Fig. 1 [6]. With increasing pressure, radicals and molecules start to polymerize form-ing polysilane chains and clusters which become nanopar-ticles. Further increase of pressure results in coagulation of the nanoparticles and forming of agglomerates and then powder with size up to few microns. While particle
forma-Fig. 1 The effect of total pressure on the deposition rate and species in a SiH4+H2discharge.
Fig. 2 Cross section of BG-BCE pm-Si:H TFTs.
tion is often though to be harmful for material properties and device efficiency our studies show that this is not always the case. It was shown by several groups that, in contrast to ag-glomerates and powders which are negatively charged and, therefore, trapped in the plasma, those small particles with size up to few nanometers can be neutral or even positively charged and contribute to deposition [7].
As a result, a nanostructured material, called polymor-phous silicon (pm-Si:H), which contains nanocrystals with typical size about 2 nm embedded in amorphous matrix, can be deposited at the conditions just before agglomeration oc-curs. The extensive study of this material in the last decade shows that it has low defect density and improved transport properties [8].
2.2 pm-Si:H TFT Processing
A standard four-level mask process has been used for the fabrication of bottom gate (BG) TFTs with a back-channel etch (BCE) technology (Fig. 2).
Molybdenum deposited by sputerring (200 nm) was used for Gate, Source and Drain metal electrodes. Silicon nitride (SiNx) deposited by PECVD was used for Gate
di-electric and passivation with thicknesses of 300 and 200 nm, respectively. The active pm-Si:H layer (150 nm) was cov-ered by n+-doped a-Si:H layer (50 nm) for S/D contacts. 2.3 pm-Si:H Structural Characterization
Transmission Electron Microscopy (TEM) was performed on pm-Si:H material to observe the presence of Si nanocrys-tals. Specimens were prepared for examination using Fo-cused Ion Beam milling. In a first set of experiments, sam-ples were cleaned and thinned down to 100 nm using 5 KV Ga ions. A first observation was made, together with Fast Fourier Transform (FFT) of images to identify any crys-tallinity in the materials. In a second set of experiment, sam-ples were thinned again to around 10 nm using low energy ion milling and observed under high resolution TEM.
3. Results and Discussion
3.1 Performance of pm-Si:H TFTs
Transfer characteristics of pm-Si:H TFTs (Fig. 3) have been
Fig. 3 Transfer characteristics of typical pm-Si:H TFTs. Table 1 Comparison of extracted parameters for a-Si: H, pm-Si: H and μc-Si: H TFTs [5].
shown to be comparable to a-Si:H ones in terms of mobil-ity, threshold voltage, subthreshold swing and ON/OFF ratio [4], [5].
Also, a very interesting feature is that they exhibit very low leakage current in comparison to μc:Si:H counterparts, but the later have lower Vthshift. A summary of comparison
is given (Table 1).
Another important parameter for large area electron-ics is on-plate homogeneity. This is particularly the case in AM-OLEDs, where any local Vthfluctuation of the
driv-ing TFT leads to visible variation of luminance. Therefore, we have measured transfer characteristics over the plate (Fig. 4).
As it can be seen, transfer characteristics of pm-Si TFTs are really superimposed on each plate, which indicates very good on-plate homogeneity.
3.2 Threshold Voltage Shift Mechanisms
The advantage of pm-Si:H TFTs over a-Si:H ones concern-ing the Vthstability has been demonstrated in the past years:
for example,: the Vth shift can be reduced by a factor of 2
under standard operating conditions [4], [5]. However, Vth
shift mechanisms for pm-Si:H-based devices are not well known so far.
We modelled the Vthshift with a power-time dependent
law of the form:
Fig. 4 On-plate distribution of transfer characteristics of typical pm-Si:H TFTs 200 mm wafer glass plate. Measurements at Left side/Center/Right side on a horizontal line, on TFTs with W= 400 μm /L = 20 μm; at VDS = 10 V.
Where:
• ΔVthis the difference between the threshold voltage Vth
at a given time and the initial threshold voltage Vth0,
• t is the stressing time,
• β is a coefficient used to distinguish the different degra-dation mechanisms.
Then, by plotting (1) as a function of time in a logarith-mic scale, one can extract β parameter. Stress measurements were performed at 3 temperatures for a-Si:H and pm-Si:H TFTs (Fig. 5).
Case of a-Si:H TFTs:
Two regimes can be distinguished for each temperature. The first regime is temperature dependent: β=0.38, 0.41 and 0.49 at T=22◦C, T=60◦C and T=90◦C, respectively. This regime is generally associated with the defect state creation (dan-gling bonds) in the a-Si:H material, as the dominant degra-dation mechanism [9]. In the second regime, the slope re-mains almost the same, so it is not temperature-dependant, which seems to indicate that charge trapping in the SiNx
in-sulator is the dominant mechanism responsible of the Vth
degradation [9]. Case of pm-Si:H TFTs:
Two regimes are also observed for each temperature. The first regime has parameters β=0.43, 0.48 and 0.51 at
T=22◦C, T=60◦C and T=90◦C, respectively. These val-ues of β are higher than those of a-Si:H. This might reflect a slightly different defect creation mechanism in pm-Si:H TFTs. Regarding the second regime, a behaviour similar to that in a-Si:H TFTs is observed, indicating that charge trap-ping in the gate dielectric is the dominant mechanism.
As a summary, by comparing a-Si:H and pm-Si:H TFTs; it appears that (i) both devices exhibit behaviour con-sistent with defect creation in the semiconductor material,
Fig. 5 Vthshift as a function of stress time for a-Si:H TFT (top) and pm-Si:H TFT (bottom). Stress conditions: VG=20 V, VDS=0.1 V, T=22◦C (), 60◦C (♦) and 90◦C ().
and (ii) defect creation in pm-Si:H might be different from the one in a-Si:H devices.
This would indicate a difference in the material prop-erty, which should be investigated by structural material characterization.
3.3 Structural Characterization by TEM
After a first sample preparation, Transmission Electron Mi-croscopy was performed on the whole PECVD stacks used for TFT fabrication consisting of SiNx, pm-Si:H and n+
a-Si:H (Fig. 6) or SiNx, a-Si:H and n+ a-Si:H (not shown).
A high resolution TEM was then applied on the active layers pm-Si:H (Fig. 7, left) and a-Si:H (Fig. 7, right).
At this stage, no difference could be seen between them, nor any nano-crystal. However, FFT images indi-cate that there is slight order in both specimens. Observing such FFT image without seeing nano-crystals might be due to the fact that samples are too thick (∼100 nm as described in sample preparation) compared to the size of the expected Si nano-crystals.
Fig. 6 TEM observation of PECVD stack consisting of SiNx, pm-Si:H and n+ a-Si:H deposited on glass.
Fig. 7 High-resolution TEM observation of pm-Si:H (left) and a-Si:H (right) materials, with associated Fast Fourier Transform (bottom).
Therefore, samples were thinned again to around 10 nm. Images of high resolution TEM performed in the same conditions as previously are given for the two materi-als (Fig. 8).
A first interesting result is that some order can now be seen. On a-Si:H sample (top), some very short distance or-der can be seen all over the image. However, it is only in areas appearing dark that nano-crystals can be seen, with or-der in the range of 2–5 nm (inset). This structure is the same through all the a-Si:H layer thickness. In the pm-Si:H layer (Fig. 8, bottom), nano-crystals are also present, but with a much higher density. Also, the crystalline matrix is almost continuous between the interface with SiNxand the top of
the pm-Si:H layer.
As a summary, TEM analysis has evidenced a signifi-cant structural difference between pm-Si:H and a-Si:H ma-terials. Silicon nano-crystals are present in both, with a much higher density in pm-Si:H. The presence of nanocrys-tals in a-Si:H could be a consequence of deposition condi-tions at pressure close to the onset of formation of pm-Si:H, as can be seen in Fig. 1.
Fig. 8 High-resolution TEM observation of a-Si:H (top) and pm-Si:H (bottom) materials, thinned to∼10 nm. Inset in top photographs shows a Si nano-crystal at higher magnification.
4. Conclusion
Hydrogenated polymorphous Silicon can be deposited in a standard PECVD chamber and allows to fabricate TFTs with very interesting characteristics including better Vth
sta-bility than a-Si TFTs, lower leakage current than μc-Si:H and excellent uniformity.
Investigation of threshold voltage shift of pm-Si:H TFTs has shown a specific semiconductor material degra-dation mechanisms compared to a-Si:H TFTs.
TEM analysis has evidenced for the first time a sig-nificant structural difference between pm-Si:H and a-Si:H materials. Silicon nano-crystals are present in both, with a much higher density in pm-Si:H. This is consistent with mechanism of threshold voltage shift in the two kinds of ma-terials. Overall: pm-Si:H is very interesting material, both from scientific, technical and industrial point of views. It ap-pears to be very suitable for low cost and high performance AM-OLED fabrication.
Aknowledgments
This work has been supported by the European Commis-sion, within Information and Communication Technologies, AMAzOLED n◦216815 Project.
by the layer-by-layer technique,” J. Appl. Phys., vol.86, p.7079, 1999. [3] A.T. Hatzopoulos, N. Arpatzanis, D.H. Tassis, C.A. Dimitriadis, M. Oudwan, F. Templier, and G. Kamarinos, “Study of drain leakage current in bottom-gated nanocrystalline silicon thin-film transistors by conduction and low-frequency noise measurements,” IEEE Trans. Electron. Devices, vol.54, no.5, pp.1076–1082, 2007.
[4] F. Templier, M. Oudwan, F. Sermet, A. Abramov, and P. Roca i Cabarrocas, “Improvement of threshold voltage stability with bottom-gate polymorphous silicon thin film transistors,” Proc. ITC 07 /SID-ME Spring Meeting 2007, Rome, Italy, Jan. 2007.
[5] J. Brochet, M. Oudwan, D. Daineka, P. Roca i Cabarrocas, and F. Templier, “Electrical characterization of polymorphous thin-film tran-sistors for application to active-matrix OLED displays,” Proc. In-ternational Thin-Film Transistor Conference 2009 (ITC’09)/SID-ME Spring Meeting 2009, pp.157–160, Palaiseau, France, March 2009. [6] P. Roca i Cabarrocas, Th. Nguyen-Tran, Y. Djeridane, A. Abramov,
E. Johnson, and G. Patriarche, “Synthesis of silicon nanocrystals in silane plasmas for nanoelectronics and large area electronic devices,” J. Phys. D, Appl. Phys., vol.40, p.2258, 2007.
[7] S.N. Abolmasov, L. Kroely, and P. Roca i Cabarrocas, “Negative corona discharge: Application to nanoparticle detection in rf reac-tors,” Plasma Sourc. Sci. Technol., vol.18, p.015005, 2009. [8] Y.M. Soro, A. Abramov, M.E. Gueunier-Farret, E.V. Johnson, C.
Longeaud, P. Roca i Cabarrocas, and J.P. Kleider, “Device grade hydrogenated polymorphous silicon deposited at high rates,” J. Non Cryst. Solids, vol.354, p.2092, 2008.
[9] M.J. Powell, “The physics of amorphous-silicon thin-film transistors,” IEEE Trans. Electron Devices, vol.36, no.12, pp.2753–2763, 1989.
Franc¸ois Templier received the Engineer degree in Physics, and Ph.D. degree in micro-electronics all from the National Polytechnic In-stitute of Grenoble, France, in 1989 and 1992, respectively. From 1993 to 1999, he was with Thomson LCD (now Thales Avionics LCD) where he was working on new a-Si TFT pro-cesses for active-matrix liquid crystal displays. Since 1999, he has been with CEA-LETI in Grenoble, France, where his research activity is focused on flexible displays and new devices for active matrix displays. Dr. Templier is a member of Club Visu — Society for Information Display France where he is currently a member of the Board.
Julien Brochet received the M.Sc. degree in Physics and engineering — materials for micro-electronic and nanotechnologies from the Uni-versit´e Montpellier 2 — Sciences et Techniques du Languedoc-Roussillon, Montpellier, France, in 2008. He is currently a Ph.D. student in the doctoral school Electronique, Electrotechnique, Automatique, Traitement du Signal in Grenoble and work in Leti-Minatec, Grenoble. He is cur-rently working in the field of electrical degra-dation of amorphous and polymorphous silicon thin-film transistors.
of pm-Si by PECVD, and on the encapsulation and assembly of OLED displays.
David Cooper completed his B.Sc. in physics at the University of Liverpool in 1997 and then spent five years in commercial research and development focused on optical and elec-tronic engineering. In 2002 he moved to the University of Cambridge to study for a Ph.D. fo-cused on off-axis electron holography for dopant profiling jointly supervised by Paul Midgley and Rafal Dunin- Borkowski. David then moved to CEA LETI in 2006 where he has worked on de-veloping new techniques for the characterisation of semiconductors using one of the first FEI Titan TEMs.
Alexey Abramov The biography and photo are not available.
Dmitri Daineka The biography and photo are not available.
Pere Roca i Cabarrocas has twenty five years experience in the field of plasma depo-sition of silicon based thin films for large area electronic applications. He received his Ph.D. from University Paris VII in 1988. After a post-doc position in Princeton University he joined the Laboratory of Physics of Interfaces and Thin Films at Ecole Polytechnique where he holds a position as a director of research and pro-fessor. His topics cover the study of RF dis-charges, the processes of nanoparticle forma-tion, the growth of amorphous, polymorphous and microcrystalline silicon thin films through in situ techniques and the optimization of solar cells and thin film transistors. More recently he has been applying the silicon nanocrystals synthesized in the plasma as building blocs for the epitax-ial growth of silicon thin films and Si/Ge quantum wells. On the other hand, he has extended the plasma processes to the growth of vertical sil-icon nanowires for third generation solar cells and of horizontal ones for planar electronic applications; both types of wires being achieved at low temperature and in a single pump down process. He has over 300 papers in peer reviewed journals and 26 patents.
![Table 1 Comparison of extracted parameters for a-Si: H, pm-Si: H and μc-Si: H TFTs [5].](https://thumb-ap.123doks.com/thumbv2/123deta/5627606.1500468/2.892.465.815.432.592/table-comparison-extracted-parameters-si-si-μc-tfts.webp)
