Journal of the Korean Physical Society, Vol. 48, January 2006, pp. S27 S31 Uniformity Improvement of the Ion Implantation System for Low Temperature Poly-Silicon TFTs Hirohiko Murata, Masateru Sato, Eiji Isobe, Kazuya Yoshida, Tatsuo Nishihara and Kouji Inada Sumitomo Eaton Nova Corp., Ehime 799-1362, Japan Toshiharu Suzuki Sumitomo Eaton Nova Corp., Tokyo 158-0097, Japan (Received 12 October 2005) Performance of the an ion-implantation system for low-temperature poly-si TFT was improved to meet various requirements for fabrication of high-quality LCD and OLED displays. With the unique inductively coupled plasma ion source, the deviation of ion beam current density was reduced to less than 3 %; 1σ over 730 920 mm substrate. The mini -uniformity in a display panel was also improved by optimizing the ion beam extraction electrodes. PACS numbers: 85.40.Ry Keywords: Flat panel display, Low-temperature poly silicon, Impurity doping, Ion implantation, Dose uniformity I. INTRODUCTION Small-to-medium medium-size liquid-crystal display (LCD) panels integrating many circuits for mobile phones, digital still cameras and PDAs are prevailing widely, employing low-temperature poly-si (LTPS) thinfilm-transistor (TFT) arrays as back-planes. The LTPS TFT is expected to form the technology to realize integration of many functional circuits on glass, that is, system on glass (SOG). Improvement in the uniformity of TFT performance such as I ds and V th on substrate has always been required to obtain high production yield. Recently, the establishment of high short-range uniformity has also been required in order to realize highquality LCD pictures with high yield. For organic lightemitting diode (OLED) displays, stricter short-range uniformity of TFT performance over adjacent pixels in a panel is essential because OLED is a current-driven device, where the brightness of each pixel in an OLED display is directly influenced by V th and current drive ability of the back-plane TFT. The impact of dispersion in TFT performance on picture quality can be barely mitigated by using compensation circuits in pixel TFTs of an OLED display [1,2]. As the compensation cirucuits necessitate a number of TFTs in a pixel, it makes the circuit design complicated, tends to decrease the production yield, and reduces the aperture ratio in a bottom- E-mail: Suzuki tsh@senova.co.jp; Tel.: +82-3-549-7801; Fax.: +82-3-549-7804 -S27- emission display. The compensation effects by these circuits are not perfect for reducing the dispersion of pixel brightness. It is considered that the ultimate solution must be brought about through the maturing of process technology. The key process technologies to influencing the uniformity of TFTs in the LTPS-TFT fabrication process are crystallization and impurity doping. Ion implanters without mass analyzer have been used for the doping from low dose for channel region to high dose for source and drain regions. Implanters of this kind, i.e. the ones without mass analyzer, are usually equipped with the a hot-cathode-type ion source with a number of filaments to extract a ribbon-shaped ion beam. This type of ion source has the difficulty in establishing stability in a lowdose condition and to keeping overall uniformity of a substrate, especially for a long operation period, because of the unequal erosion between source filaments. We have developed an ion implanter without mass analyzer for LTPS TFT fabrication, which has a unique inductively-coupled plasma (ICP) ion source and a beam shutter system in order to handle both high- and lowdose conditions as shown in Figure 1 [3]. With these unique mechanisms, this system can implant impurities with excellent stability and uniformity. Many of these ion implanters without mass analyzer have an ion source with plasma electrodes that have slits with rows of holes [4, 5]. This electrode structure often brings about mini -ripples in the beam profile which disturb short-range uniformity in a product display panel
-S28- Journal of the Korean Physical Society, Vol. 48, January 2006 Fig. 1. Cross section of inductively coupled plasma ion source with beam shutter. corresponding to the slit-hole pitch. This time, both uniformity over a substrate and uniformity over adjacent pixels ( mini -uniformity) are enhanced by improvement of the ion source and plasma electrode in the ion implanter for high-quality LCDs and OLED display panels. This paper introduces measures for improving both overall uniformity and mini - uniformity and resulting performance. II. ION SOURCE AND DOSIMETRY IN THE CONVENTIONAL SYSTEM The conventional implanter having an ICP ion source has a number of features with high accuracy and stability; these differentiate it from other non-mass-analyzed implanters with hot cathode filaments. As the implanter [5] employs an ICP ion source with RF antennas as shown in Figure 1, the plasma condition is very stable without consumption of antenna for a long period of operation of more than 6 months. Users can easily control and optimize the plasma conditions by matching the impedances of antennas. This provides high dose uniformity better than 5 % 1σ for long operating periods. This implanter is also equipped with a unique beam shutter system (Figure 1) to switch over from low-dose to high-dose condition and vice versa. The beam shutter system reduces the number of slits from 7 slits for high-dose implantation to 1 slit for low-dose implantation. This structure enables the user to switch the beam current density for high dose to that for low dose in a few tens of minutes and to keep stability and uniformity in low-dose conditions as same as those in high-dose conditions. The configuration of electrodes for extraction of a ribbon-shaped beam is shown in Figure 2. The extraction electrode system consists of 4 electrodes, each having seven slits for boron implantation. Only the plasma Fig. 2. Cross-sectional view of extraction electrode structure. electrode has the slit holes of parallelogram shape as shown in the insert of the figure. Other electrodes have slits with long openings. Thanks to this structure, alignment of electrodes is easy for maximizing the beam extraction. This implanter has an accurate dose control system. Three Faraday cups can measure the ion beam current densities at the center and both ends of the ribbonshaped ion beam. A movable Faraday cup scans over the length of the ribbon-shaped beam, measuring the beam current uniformity in 1-mm pitch. It also has a small and accurate mass spectrometer. This system can accurately analyze the intensities of ion species such as hydrogen ions, monomer dopant ions and dimmer dopant ions in the beam, and calculate the dopant fraction [5] in the total ion beam. With the intrinsic stabile nature of the ICP-type ion source, this system ensures implantedimpurity concentration accuracy, even in the low-dose conditions for channel doping. III. IMPROVEMENT OF OVERALL UNIFORMITY For the enhancement of the production yield, uniformity over a substrate is very important. Two improvements were made for achieving uniformity higher than before. The ion source chamber was shortened from 118 cm to 112 cm, thus decreasing the ion source volume. For the same source gas flow rate, the gas concentration in the ion source increased, thus resulting in an increase
Uniformity Improvement of the Ion Implantation System for Hirohiko Murata et al. -S29- Fig. 4. Dose uniformity in 730 920 mm overall substrate for boron implanted into Si wafers. (a): high-dose condition, and (b): low-dose condition. Fig. 3. Boron beam current density profile at high- and low-dose condition. in plasma density. While the distance between adjacent antennas was increased by 8.3 % compared with the conventional one, the distance between an antenna and the chamber wall at both ends of the ion source was decreased. The increase in plasma density and antenna spacing optimizes the plasma conditions and homogenizes the plasma density by decreasing the fall of current density at both ends of a ribbon-shaped beam. Figures 3(a) and (b) show the boron beam current density profiles for high-dose and-low dose conditions. For comparison, a beam current profile for low-dose conditions obtained by the ion source before the improvement is also shown in Figure 3(c). The standard deviation of the beam current density is apparently reduced from about 4 % to less than 3 %. Figures 4(a) and (b) show the dose uniformities for high-dose conditions and low-dose conditions, respectively. Boron ions were implanted into five 6-inche Si wafers set on a 730 920 mm graphite carrier as shown in the Figure 5. Sheet resistivities of Fig. 5. Sample wafer setting in dose uniformity measurement. the wafers were measured after activation annealing at 1150 C for 20 sec. Uuniformities of less than 3 % are obtained for both high dose and low dose conditions.
-S30- Journal of the Korean Physical Society, Vol. 48, January 2006 Fig. 8. Box plots of averaged mini -dispersion ratio. a: modified slits, optimized condition; b: conventional slits, optimized condition; c: conventional slits, standard condition. Fig. 6. Boron beam profiles showing mini -dispersion in low-dose condition. (a): conventional slit, standard condition; (b): conventional slit, optimized condition; (c): modified slit, optimized condition. Fig. 7. Definition of short-range dispersion. IV. IMPROVEMENT OF MINI -UNIFORMITY Precise measurement of beam current density profile for low-dose condition by movable Faraday cup revealed small-amplitude and short-wavelength mini -ripples in the beam profile as depicted in curve-(a) of Figure 6. This ripple corresponds to the pitch of slit holes which have parallelogram shape and overlap with the adjacent holes having the ratio of 30 %. The mini -dispersion is defined as the ratio of the beam-current-density difference between the adjacent peak and valley to average ion beam current as shown in Figure 7. The averaged dispersion R av of these ripples is expressed by Eq. (1): D SR = Σ n n, R av = D SR I B av, (1) where D SR is the average difference over ripples, n is the number of measured ripples, n is the difference in beam current densities between n-th adjacent peak and valley, and I B av is the averaged beam current density of the ribbon shaped beam. Although the ripples could be somewhat decreased by defocusing the extracted beam by controlling the ion source conditions: source gas flow rate, extraction voltage and RF power of antenna for example (curve-(b) of Figure 6), an essential improvement is desired. In order to reduce the ripple pitch and ripple height substantially, the central slit of the 7 slits was modified. The hole pitch of the central slit was reduced to half of that in conventional slits and the overlap ratio was increased to 46.8 %. Electric field strength just after the plasma electrode was simulated and was confirmed to be homogenized, and the dispersion due to ripples was predicted to be alleviated. In Figure 6 (curve-(c)), the measured beam current density profile is shown, and a very smooth profile can be seen for low-dose condition. Figure 8 shows the box plot of average mini -dispersion R av of the beam profile extracted by the conventional slit and that of the newly modified slit for the low-dose condition of 1 10 13 cm 2 at 15 kev. The average dispersion value is reduced from 3.5 % to 0.4 % and the deviation of the ripple is also greatly reduced. The short range mini -uniformity relating to ripples is clearly enhanced by the newly modified slit. V. LONG-PERIOD STABILITY Repeatability of the dopant ion beams is very important for establishing the production yield and for managing the process easily. Beam current repeatability of the improved implanter was confirmed by the iterative implantation of high dose and low dose for a long period. Series of implantations were carried out after the switchover from low-dose conditions to high-dose conditions and vice versa. Before each implantation, the beam currents was measured. Figure 9(a) plots the dopant beam current densities measured just after the switchover between high-dose and low-dose conditions. Standard deviations of dopant beam current repeatability of less than 3.0 % are obtained both for high-dose and
Uniformity Improvement of the Ion Implantation System for Hirohiko Murata et al. -S31- VI. CONCLUSION Improvement in both overall uniformity over substrate and mini -uniformity was made in the non-massanalyzed implanter with ICP ion source for high-quality LCD and OLED displays applications. By optimizing the dimensions and antenna configuration of the inductively coupled ion source, overall uniformity is improved to less than 3 % (1σ) for high-dose and low-dose conditions. Modification of the plasma electrode attained improvement in mini -uniformity to less than 3 % (1σ). Dopant beam current stability, which is important for efficient operation, was also confirmed to be less than 3 % and dose stability near 1.0 % was obtained. This improved ion-implantation system will be successfully applied for the fabrication of TFT back-planes of high-quality LCD and OLED displays with high yield. ACKNOWLEDGMENTS The authors would like to thank Dr. N. Takahashi and Mr. Y. Hidaka for their strong support for this work. The authors would also like to thank Mr. M. Inoue and Mr. H. Ochi for their help in the preparation of the manuscript. Fig. 9. Stability of boron implant during high-dose (80 kev, 1 10 15 cm 2 ) and low-dose (15 kev, 1 10 13 cm 2 ) switch-over. (a): repeatability of dopant current density; (b): dose repeatability. for low-dose implantation conditions, even for a long and stern switchover operation. The dose repeatability was also measured by using a 6-inch wafer on the graphite carrier for every switchover from high-dose to low-dose condition and low dose to high dose condition. Results are shown in Figure 9(b) and dose repeatability near 1 % is obtained. It is confirmed that this implanter has very high repeatability, even for iterative switchover operation for up to 60 hours. REFERENCES [1] A. Yumoto, M. Asano, H. Hasegawa and M. Sekiya, IDW 01, 1395 (2001). [2] S. W. T. Tam, M. Kimura, R. Friend, T. Shimoda and P. Migliorato, IDW 02, 243 (2002). [3] Y. Shao, J. Blake, K. Chen, A. Brailove, M. King and M. Sato, 1998 Int l Conf. Ion Implant. Technol. 251 (1998). [4] Y. Inouchi, Y. Matsuda, S. Maeno, M. Konishi, J. Tatemichi, M. Nukayama, E. Muerasaka, K. Nakao, K. Orihira and Y. Andoh, 15th Int l Conf. Ion Implant. Technol. Final Program & Abstract Book, 73(C313) (2004). [5] M. Sato, M. Maruyama, H. Sakanishi, E. Isibe, K. Yoshida, T. Nishihara, T. Ochi, K. Inada, H. Murata and A. Brailove, 14th Int l. Conf. Ion Impant. Technol. 383 (2002).