Journal of the Korean Physical Society, Vol. 54, No. 1, January 2009, pp. 549553
TFT Backplane Technologies for AMLCD and AMOLED Applications
Jae Beom Choi, Young Jin Chang, Cheol Ho Park, Beom Rak Choi and Hyo Seok Kim
OLED Lab., Samsung Electronics Co., Gyeonggi 449-711
Kee Chan Park
Department of Electronic Engineering, Konkuk University, Seoul 143-701
(Received 24 January 2008)
We thoroughly investigated low-temperature polycrystalline silicon (LTPS) thin- lm transistor
(TFT) backplane technologies based on (1) a melt-mediated crystallization process with laser sys-
tems, (2) a solid phase crystallization process with advanced annealing systems and (3) a single-
crystalline Si layer transferred onto a large glass substrate for
at-panel-display applications. Ex-
tensive micro-structural analyses of the silicon lms, comparison of the TFT performances and
evaluation of the image quality of the displays enabled us to choose the competitive technologies for
large-area active-matrix liquid-crystal display (AMLCD) and active-matrix organic light-emitting
diode (AMOLED) applications.
PACS numbers: 85.60.Pg, 73.61.Cw, 81.05.Cy, 81.10.Fq, 81.10.Jt
Keywords: Low-temperature polycrystalline silicon (LTPS), Excimer laser annealing (ELA), Sequential lat-
eral solidi cation (SLS), Solid phase crystallization (SPC), Nanocap-assisted crystallization (NAC), Silicon
on glass (SiOG)
I. INTRODUCTION
Large-area active-matrix liquid crystal display
(AMLCD) TVs are based on amorphous silicon (a-
Si) thin- lm-transistor (TFT) backplanes. However,
small-area AMLCDs for premium mobile devices are
based on low-temperature polycrystalline silicon (LTPS)
TFT backplanes because the conventional a-Si TFT
backplane cannot meet speci cations such as high
aperture ratio and low power consumption. In addition,
active-matrix organic light-emitting diode (AMOLED)
displays with LTPS TFT backplanes have been adopted
not only for the mobile devices but also for TVs with
diagonal sizes larger than 10 inches.
Unlike the AMLCD which utilize an a-Si:H TFT back-
plane for large size and a LTPS backplane for small
size, there are still many research activities to nd the
optimum TFT backplane technology for a high-quality
AMOLED display. There are four major TFT backplane
technologies for AMOLEDs: (1) a-Si:H TFTs, (2) LTPS
TFTs obtained by using a melt-mediated crystallization
process, (3) LTPS TFTs obtained by using a solid phase
crystallization process and (4) a single crystalline silicon
layer transferred onto a glass substrate. Among these,
the a-Si:H TFT technology is best established for mass
production and the displays with a-Si:H TFT backplanes
show excellent uniformity over large areas [1]. However
E-mail: keechan@konkuk.ac.kr; Fax: +82-2-3437-5235
the threshold voltage shift under continuous positive bias
stress is a critical limitation to AMOLED applications
[2]. In this paper, we will review the other three di er-
ent technologies to obtain a suitable backplane for both
active-matrix display applications.
II. MELT-MEDIATED CRYSTALLIZATION
PROCESS
The melt-mediated crystallization process includes (1)
melting of the precursor a-Si lm and (2) subsequent
solidi cation of the liquid a-Si, resulting in a polycrys-
talline silicon (poly-Si) lm of various microstructures,
depending on the process parameters. The phase trans-
formation scenarios of the melt-mediated crystallization
process include (1) partial melting, where only the sur-
face of the a-Si lm is melted and solidi cation takes
place in the vertical direction from the bottom, resulting
in small grains, (2) near complete melting, where the a-Si
lm is melted to the bottom with a small number of un-
melted Si clusters remaining discontinuously and lateral
solidi cation takes place producing grains much larger
than the lm thickness and (3) complete melting, where
the entire a-Si lm is melted and nucleation-triggered
solidi cation starts under super-cooled condition, again
resulting in ne grains.
Among the various types of laser systems that can be
used for the melt-mediated crystallization process, the
-549-
-550- Journal of the Korean Physical Society, Vol. 54, No. 1, January 2009
Fig. 1. TS-SLS process: (a) laser irradiation and (b) re-
sulting microstructure.
excimer laser annealing (ELA) system has been widely
utilized for the LTPS TFT backplanes for AMLCDs [3].
The crystallization is carried out by scanning narrow
laser pulses (e.g., 465-mm long and 0.5-mm wide) over
the a-Si lm on a large glass substrate. The process
window of the laser energy density in the near-complete-
melting condition is rather narrow; thus, the resulting
microstructure of the poly-Si material is sensitive to
uc-
tuations in the laser energy. Therefore, the ELA system
should have highly uniform laser intensity pro le, both
the long and the short axes and shot-to-shot consistency.
In particular, even a small variation in the laser inten-
sity can be easily perceived in case of the AMOLED dis-
play because the brightness is directly associated with
the current
owing through the driving transistor. The
mura in the scanning direction is associated with the
nonuniformity of the laser intensity along the long axis
and the mura perpendicular to the scanning direction
is associated with the shot-to-shot nonuniformity of the
laser intensity.
On the other hand, the sequential lateral solidi cation
(SLS) process with a patterned laser beamlet has also
been utilized for AMLCD production [4, 5]. The SLS
process is composed of (1) laser irradiation through a
patterned mask to provide an abrupt temperature pro-
le at the edge of the irradiated area, leading to con-
trolled super lateral growth (C-SLG), (2) translation of
the substrate by a precisely controlled distance and (3)
repetition of (1) and (2) resulting in complete crystal-
lization of the lm. Compared to the conventional ELA
systems, the advantage of the SLS process is (1) a wider
process window in laser intensity, (2) controllability of
the grain size and (3) scalability of the substrate.
However, the process time of the original SLS process
is several times longer than that of the ELA. To improve
the throughput of the SLS processes to be even higher
than ELA, the two-shot (TS) SLS process has been de-
veloped and is being utilized in the mass production as
Fig. 2. (a) 300 VGA AMLCD and (b) 1400 WXGA
AMOLED display fabricated on the TS-SLS TFT backplane.
the most competitive technique [6]. Figure 1 illustrates
the TS-SLS process including (a) the laser irradiation
scheme and (b) the resulting microstructure. \L" is the
line width of the open area on the mask, \S" is the space
between the open areas and the grain size is determined
as (L + S)/2.
With the LTPS TFT backplanes obtained by us-
ing the TS-SLS process, we could make high-quality
AMLCD products and a 1400 WXGA (1280 RGB
768) AMOLED display without any mura associated
with the laser crystallization process, as shown in Fig-
ure 2. We adopted a voltage-addressed compensation
circuit with six TFTs and a capacitor in each pixel for
the 1400 AMOLED display.
III. SOLID PHASE CRYSTALLIZATION
(SPC)
The solid phase crystallization (SPC) process is the
simplest and the lowest-cost process to obtain poly-Si
lms on large-area glass substrates. In this process, un-
like the melt-mediated crystallization process, the a-Si
lm is directly transformed to the crystalline structure
via nucleation and grain growth process at a tempera-
ture around 600 C. The SPC process can be catego-
TFT Backplane Technologies for AMLCD and AMOLED { Jae Beom Choi et al. -551-
Fig. 3. 2.200 qVGA AMOLED display employing the SPC
TFT backplane.
rized into two groups: (1) simple SPC process and (2)
metal-induced-crystallization (MIC) process where the
crystallization temperature is reduced to below 500 C
by employing metal catalysts. As for the process equip-
ment, magnetic- eld-aided rapid thermal annealing may
be utilized in addition to a conventional furnace. The
magnetic eld induces an eddy current in the heated a-
Si lm and, thus, produces a poly-Si lm in a reduced
process time [7].
The simple SPC process does not require any addi-
tional process other than the thermal annealing. The
precursor material and the annealing temperature deter-
mine the average grain size and the crystallinity of the
completed lm. Since the scale of crystalline irregularity
in the SPC poly-Si is much smaller than the device di-
mension, SPC TFTs have uniform characteristics over a
large area [8,9]. As shown in Figure 3, we could obtain a
2.200 qVGA (240 RGB 320) AMOLED display by us-
ing the SPC TFT backplanes without any compensation
circuit in the pixel.
For AMOLED TV applications, a carrier mobility of
1 cm2/Vs is high enough to drive the OLED current
in each pixel because the TFT channel width can be
expanded to hundreds of micrometers for a pixel den-
sity below 100 ppi (pixels per inch) to meet the current
requirements, which is the case in a normal TV. How-
ever, the crystallinity must be increased by reducing the
defect density in the SPC poly-Si lm in order to im-
prove the mobility up to a level that can be used for
Fig. 4. 21.300 UXGA AMOLED display based on the NAC
TFT backplane.
high-performance AMLCDs because more and more cir-
cuits need to be integrated on recent value-added display
panels. In addition, the process temperature should be
further lowered below 550 C in order to prevent the
glass warpage problem frequently observed in large glass
substrates.
In the MIC process, metal catalysts are utilized to
promote the crystallization process at reduced tempera-
ture. For example, when Ni is used as the catalyst ma-
terial, the Ni atoms in/on the a-Si lm can form nickel
silicide at temperatures lower than the intrinsic crystal-
lization temperature (600 C) [10, 11] and the NiSi2
propagates through the a-Si matrix, leaving a needle-
shaped crystalline Si region even at 484 C [12]. Since
the individual Si grains obtained when using the MIC
process tend to have textures with a certain orienta-
tion, further treatment can provide better structured
poly-Si lms with reduced defect density at the grain
boundaries [13,14]. With improved crystallinity, the MIC
TFTs can have a mobility higher than that of the SPC
TFTs [12, 15{20]. However, the leakage current associ-
ated with Ni contamination is rather higher compared
with that of the simple SPC TFTs. In order to solve
the Ni contamination problem, a gettering process, in
which the Ni atoms in the crystallized Si lm di use to
the phosphorus-implanted region during the dopant ac-
tivation process and a nanocap-assisted crystallization
(NAC) process [21], in which a thin SiO2 capping layer
is formed on top of the a-Si lm prior to the deposition
of Ni in order to control the Ni contents, have been de-
veloped. With the LTPS TFT backplanes obtained by
using the NAC process, we fabricated a 21.300 UXGA
(1600 RGB 1200) AMLCD as shown in Figure 4. In
order to exploit these MIC materials for AMOLED ap-
plications, we found that the nonuniformity of the grain
size and the nonuniformity of the Ni distribution along
the domain boundaries should be controlled.
Although the uniformity and the current-driving ca-
pability of the SPC, including MIC, TFTs are sucient
as AMOLED backplanes, the stability is not satisfactory
-552- Journal of the Korean Physical Society, Vol. 54, No. 1, January 2009
Fig. 5. Schematic diagram of the SiOG process comprising
(a) hydrogen implantation, (b) bonding, (c) separation and
(d) thinning of the Si wafer transferred on the glass substrate.
due to high trap density. The poor stability is revealed
as hysteresis in the TFT characteristics and causes im-
age sticking in the display [22,23]. The microcrystalline
silicon (c-Si) TFTs that have recently attracted much
attention also su er from the same instability problem,
though they are much better than the a-Si TFTs [8,24].
Poor stability due to the interface states is observed even
for the badly fabricated ELA LTPS TFTs [25]. However
it is remarkably improved by using a SLS backplane ow-
ing to the higher crystallinity that is characteristic of the
SLS process.
IV. SINGLE CRYSTALLINE SILICON
Finally, we investigated the feasibility of utilizing sin-
gle crystalline Si materials on large glass substrates for
active matrix display applications. Since the variation in
the TFT performance is detrimental to the image quality
of the AMOLED, there have been attempts to avoid the
grain boundary-related problems by transferring single-
crystalline Si layers to the glass substrates [26].
Figure 5 shows a schematic diagram of the silicon-on-
glass (SiOG) process developed by Corning Inc. [27].
The SiOG process consists of (1) hydrogen implanta-
tion to the Si wafers to form the separation zone (Fig-
ure 5(a)), (2) anodic bonding of the hydrogen-implanted
wafers to the glass substrate, resulting in strong SiOx
bonding between them (Figure 5(b)), (3) separation of
the wafers, leaving a thin silicon layer on the glass sub-
strate (Figure 5(c)) and (4) thinning the Si layer through
chemical and mechanical polishing (Figure 5(d)). The
thickness of the Si layer can be controlled within a stan-
dard deviation of several nanometers and the maximum
process temperature does not exceed 400 C. Figure 6
shows a 1.900 (320 RGB 240) AMLCD and a 2.200
(240 RGB 320) AMOLED display fabricated by us-
ing the SiOG process. F The advantages of the SiOG
process are (1) the absence of troublesome grain bound-
aries, which enables us to make a uniform AMOLED
display free from the mura associated with the grain
boundaries, (2) a lower defect density compared with
Fig. 6. (a) 1.900 qVGA AMLCD and (b) 2.200 qVGA
AMOLED display based on the SiOG TFT backplane.
the polycrystalline material which leads us to fabricate
an AMLCD with a high level of monolithic circuit inte-
gration and (3) a simpli ed TFT process with cost com-
petitiveness over the conventional LTPS process. The
ultimate goal of the SiOG technology is to obtain high-
quality AMOLED displays without any compensation
circuit in the pixel and fully-integrated AMLCDs.
V. SUMMARY
We reviewed the TFT backplane technologies for
high-performance active-matrix
at panel display ap-
plications. Although the melt-mediated crystallization
process with a laser system is widely utilized in the
mass production of AMLCDs, the laser-related device's
nonuniformity should be further improved to fabricate
AMOLED displays with enhanced production yield.
Solid phase crystallization may be an alternative for
large-area AMOLED displays if the stability of the TFT
is improved. For the present, the laser-annealed LTPS
TFT Backplane Technologies for AMLCD and AMOLED { Jae Beom Choi et al. -553-
TFT is the only possible technology for the commercial-
ization of AMOLED display because it has no critical
problem, unlike the image sticking in the SPC backplane.
The SiOG technology is expected to be utilized for small-
sized AMLCDs or AMOLED displays in the near future.
REFERENCES
[1] S. H. Kim, J. H. Hur, K. M. Kim, J. H. Koo and J. Jang,
J. Korean Phys. Soc. 48, S80 (2006).
[2] J. H. Koo, J. W. Choi, Y. S. Kim, M. H. Kang, S. H.
Kim, E. B. Kim, H. Uchike, S. W. Lee and J. Jang, J.
Korean Phys. Soc. 50, L933 (2007).
[3] T. Nishibe and H. Nakamura, SID '06 Digest, 1091
(2006).
[4] C. W. Kim, K. C. Moon, H. J. Kim, K. C. Park, C. H.
Kim, I. G. Kim, C. M. Kim, S. Y. Joo, J. K. Kang and
U. J. Chung, SID '04 Digest, 868 (2004).
[5] J. B. Choi, K. C. Park, K. C. Moon, J. H. Eom, R.
Yokoyama and C. W. Kim, J. Soc. Inf. Display 15, 931
(2007).
[6] J. B. Choi, Y. J. Chang, C. H. Park, Y. I. Kim, J. Eom,
H. D. Na, I. D. Chung, S. H. Jin, Y. R. Song, B. Choi,
K. Park, C.W. Kim, J. Souk, Y. S. Kim, B. H. Jung and
K. C. Park, SID '08 Digest, 97 (2008).
[7] K. H. Kang, S. J. Lee, S. E. Nam and H. J. Kim, Mat.
Sci. Forum 449, 513 (2004).
[8] T. Arai, N. Morosawa, Y. Hiromasu, K. Hidaka, T.
Nakayama, A. Makita, M. Toyota, N. Hayashi, Y.
Yoshimura, A. Sato, K. Namekawa, Y. Inagaki, N.
Umezu and K. Tatsuki, SID '07 Digest, 1370 (2007).
[9] S. K. Hong, B. K. Kim and Y. M. Ha, SID '07 Digest,
1366 (2007).
[10] N. K. Song, M. S. Kim, Y. S. Kim, S. H. Han and S. K.
Joo, J. Korean Phys. Soc. 51, 1076 (2007).
[11] Y. S. Kim, N. K. Song, M. S. Kim, S. J. Lee and S. K.
Joo, J. Korean Phys. Soc. 51, 1156 (2007).
[12] C. Hayzelden and J. L. Batstone, J. Appl. Phys. 73, 8279
(1993).
[13] Y. Hirakata, M. Sakakura, S. Eguchi, Y. Shionori, S.
Yamazaki, H. Washio, Y. Kubota, N. Makita and M.
hijikigawa, SID '00 Digest, 1014 (2000).
[14] S. K. Kim, J. H. Oh, J. H. Cheon and J. Jang, J. Korean
Phys. Soc. 48, 1526 (2006).
[15] S. Y. Yoon, N. Y. Young, P. J. Zaag and D. McCulloch,
IEEE Electron. Dev. Lett. 24, 22 (2003).
[16] J. C. Kim, J. H. Choi, S. S Kim, K. M. Kim and J. Jang,
Appl. Phys. Lett. 83, 5068 (1993).
[17] M. Kim, K. B. Kim, K. Y. Lee, C. H. Yu, H. D. Kim and
H. K. Chung, J. Appl. Phys. 103, 044508 (2008).
[18] N. Kubo, N. Kusumoto, T. Inushima and S. Yamazaki,
IEEE Trans. Electron Dev. 41, 1876 (1994).
[19] Z. Jin, H. S. Kwok and M. Wong, IEEE Elec. Dev. Lett.
20, 167 (1999).
[20] S. Zhang, R. Han, J. K. O. Sin and M. Chan, IEEE Elec.
Dev. Lett. 22, 530 (2001).
[21] Y. J. Chang, Y. I. Kim, S. H. Shim, S. Park, K. W. Ahn,
S. C. Song, J. B. Choi, H. K. Min and C. W. Kim, SID
'06 Digest, 1276 (2006).
[22] D. H. Nam, H. K. Lee, S. H. Jung, T. J. Ahn, C. Y. Kim,
C. D. Kim and I. J. Chung, ECS Trans. 3, 57 (2006).
[23] S. H. Jung, H. K. Lee, C. Y. Kim, S. Y. Yoon, C. D. Kim
and I. B. Kang, SID '08 Digest, 101 (2008).
[24] T. Tsujimura, W. Zhu, S. Mizukoshi, N. Mori, K. Miwa,
S. Ono, Y. Maekawa, K. Kawabe and M. Kohno, SID '07
Digest, 84 (2007).
[25] B. K. Kim, O. Kim, H. J. Chung, J. W. Chang and Y.
M. Ha, Jpn. J. Appl. Phys. 43, L482 (2004).
[26] J. B. Choi, Y. J. Chang, S. H. Shim, I. D. Chung, K.
W. Park, K. C. Park, K. C. Moon, H. K. Min and C.-W.
Kim, SID '07 Digest, 1378 (2007).
[27] J. G. Couillard, K. P. Gadkaree and J. F. Mach, US
Patent 7176528, 2007.
