A novel TFT-OLED integration for OLED-independent pixel programming in
`amorphous-Si AMOLED pixels
`
`Bahman Hekmatshoar
`Alex Z. Kattamis
`Kunigunde Cherenack
`Sigurd Wagner
`James C. Sturm
`
`Abstract— The direct voltage programming of active-matrix organic light-emitting-diode (AMOLED)
`pixels with n-channel amorphous-Si (a-Si) TFTs requires a contact between the driving TFT and the
`OLED cathode. Current processing constraints only permit connecting the driving TFT to the OLED
`anode. Here, a new “inverted” integration technique which makes the direct programming possible
`by connecting the driver n-channel a-Si TFT to the OLED cathode is demonstrated. As a result, the
`pixel drive current increases by an order of magnitude for the same data voltages and the pixel data
`voltage for turn-on drops by several volts. In addition, the pixel drive current becomes independent
`of the OLED characteristics so that OLED aging does not affect the pixel current. Furthermore, the
`new integration technique is modified to allow substrate rotation during OLED evaporation to improve
`the pixel yield and uniformity. The new integration technique is important for realizing active-matrix
`OLED displays with a-Si technology and conventional bottom-anode OLEDs.
`
`Keywords—a-Si,activematrix,AMOLEDdisplay,TFT-OLEDintegration,bottom-anodeOLEDs.
`
`Introduction
`1
`Superior properties of organic light-emitting diodes (OLEDs)
`such as high-speed response, emissivity, wide viewing angle,
`simple structure, and anticipated low fabrication cost make
`them very appealing for display applications.1 Integrating
`OLEDs with TFTs in the form of active matrices is required
`for achieving low power consumption in mid- and large-
`sized displays.2,3 Although amorphous-Si (a-Si) technology
`is low in cost, in widespread production, and very suitable
`for large-area deposition especially on flexible substrates,
`low-temperature poly-Si (LTPS) has been the first material
`of choice for TFT backplanes since the introduction of
`AMOLED displays. The advantages of LTPS over a-Si are
`(i) higher TFT mobility, (ii) higher TFT stability, and (iii)
`availability of p-channel TFTs.2,4 Although using a-Si for
`
`AMOLED applications has been demonstrated5,6 and com-
`plete a-Si AMOLED displays have been realized by indus-
`try,7,8 the commercial production of AMOLED displays
`requires that weaknesses of a-Si be resolved or effectively
`compensated. The low field-effect mobility in a-Si may be
`compensated for by developing high-efficiency OLED’s
`which require low driving currents.9 The instability of
`threshold voltage in a-Si TFTs is especially serious when
`they are fabricated at low process temperatures compatible
`with the typical flexible clear plastic substrates. The reliabil-
`ity of a-Si TFTs can be improved by using clear plastic sub-
`strates which allow higher process temperatures10 or by
`circuits which compensate for threshold voltage shift.11
`Using more-efficient OLEDs also alleviates the a-Si TFT
`stability problem, because the threshold voltage shift is
`lower at lower driving currents and lower gate voltages.9 A
`
`FIGURE 1 — Circuit schematic of 2-TFT AMOLED pixels: (a) conventional structure with p-channel TFTs (low-temperature poly-Si), (b)
`conventional structure with n-channel TFTs (a-Si), and (c) new “inverted” structure with n-channel TFTs (a-Si).
`
`The authors are with Princeton University, Princeton Institute for the Science and Technology of Materials (PRISM) and the Department of Electrical
`Engineering, Olden St., Princeton, N.J. 08540; telephone 609/258-6624, fax –1840, e-mail: hekmat@princeton.edu.
`© Copyright 2008 Society for Information Display 1071-0922/08/1601-0183$1.00
`
`Journal of the SID 16/1, 2008
`
`183
`
`SAMSUNG, EXH. 1010, P. 1
`
`

`

`serious issue with a-Si TFT pixel circuits is the direct pro-
`gramming of the pixel current by the data voltage, which is
`not conventionally possible in a-Si technology due to the
`lack of p-channel TFTs.2,4 Addressing this issue is the focus
`of this paper.
`Figure 1(a) shows the circuit schematic of a conven-
`tional 2-TFT AMOLED pixel fabricated in LTPS technol-
`ogy using conventional TFT-OLED integration with
`p-channel poly-Si TFTs. The pixel cross section is shown in
`Fig. 2. The conventional integration sequence is dictated by
`three constraints: (i) the OLEDs must be evaporated after
`the TFT fabrication process because the TFT process severely
`damages the OLEDs; (ii) the best OLEDs are deposited
`from anode to cathode, i.e., the anode (e.g., ITO) is depos-
`ited first, followed by the organic layers and then the cath-
`ode (bottom-anode OLEDs); and (iii) patterning the
`organic layers and cathode by photolithography is generally
`not feasible without damaging the OLEDs. As a result, the
`driver TFT is connected to the OLED anode rather that the
`
`OLED cathode (Fig. 2). With p-channel TFTs [Fig. 1(a),
`LTPS AMOLED pixels], the TFT terminal connected to the
`OLED is the drain, and therefore the gate-source voltage of
`the driver TFT is determined directly by the data voltage
`(Vdata) and is independent of the OLED characteristics.
`This is because the TFT current in saturation is controlled by
`VGate – VSource, or in this case Vdata – VSS (VSS is a fixed voltage).
`If the conventional integration (Fig. 2) is used for a-Si
`technology where only n-channel TFTs are available, the
`TFT terminal connected to the OLED will be the TFT
`source [Fig. 1(b), conventional a-Si AMOLED pixels]. There-
`fore the data voltage is split across the OLED and the gate
`source of the driving TFT [Vdata = VGS(driver) + VOLED].
`This is not desirable for two reasons: (i) higher data voltages
`are required for programming the pixel to obtain the same
`pixel currents [requiring the same VGS (driver)], because a
`part of the data voltage drops across the OLED rather than
`dropping entirely across the gate source of the driving TFT;
`and (ii) the voltage drop across the gate source of the driving
`TFT and thus the pixel current depends on the OLED char-
`acteristics, which may vary device to device in manufactur-
`ing and vary with time during device operation. Therefore,
`direct programming of a-Si TFTs requires a new technique
`for connecting the driver TFT to the OLED cathode instead
`of the OLED anode, as shown in Fig. 1(c), so that the data
`voltage may be transferred directly to the gate source of the
`driver TFT. Such an integration technique is presented in
`this work.
`
`Fabrication
`2
`The schematic cross section of an a-Si AMOLED pixel fab-
`ricated with the inverted integration process is shown in
`Fig. 3. The cross section corresponds to the circuit sche-
`
`FIGURE 2 — Schematic cross section of the fabricated conventional
`AMOLED structure [used in the pixel circuits of Figs. 1(a) and 1(b)]
`during the evaporation of (a) the organic layers and (b) cathode.
`
`FIGURE 3 — Schematic cross section of the fabricated new “inverted”
`AMOLED structure of Fig. 1(c), during the evaporation of (a) the organic
`layers and (b) cathode.
`
`184
`
`Hekmatshoar et al./ A novel TFT-OLED integration
`
`SAMSUNG, EXH. 1010, P. 2
`
`

`

`matic of Fig. 1(c). The a-Si TFT backplane is fabricated at
`temperatures up to 300°C on glass.5 The apparent (i.e., not
`corrected for contact resistance) saturation mobility and
`threshold voltage of the driving TFTs (L = 5 µm) are 0.65 ±
`0.04 cm2/V-sec and 1.7 ± 0.2 V, respectively. After processing
`the TFT backplane (including ITO as the OLED anode),
`insulating “separators” are formed by patterning a layer of
`positive photoresist using conventional photolithography.
`As shown in Fig. 3(a), the organic layers are then evaporated
`at an angle in such a way that an interconnect extension
`connected to the driving TFT is not coated with the organic
`layers, taking advantage of the separator’s shadowing effect.
`We have used 10-µm-thick photoresist separators and stand-
`
`ard TPD/ALq3 organic layers for this experiment. Then, as
`shown in Fig. 3(b), the cathode (Mg–Ag/Ag) is evaporated
`at an angle opposite to the organic evaporation angle to form
`the OLED cathode and also to contact the interconnect
`extension. Therefore, the electrical circuit of Fig. 1(c) is
`realized.
`
`TFT-OLED integration results
`3
`The measured dc characteristics of a-Si AMOLED pixels
`integrated with the conventional and inverted processes are
`compared in Fig. 4. First, in the inverted structure [Fig.
`4(b)], the pixel drive current, Ipixel, turns on at Vdata = 1.7 V
`(corresponding to the threshold voltage of the driver TFT)
`which is considerably lower than the conventional design
`
`FIGURE 4 — Measured pixel current (Ipixel) as a function of
`the
`programmed data voltage (Vdata) of (a) a conventional a-Si AMOLED
`pixel shown in the pixel circuit of Fig. 1(b) and 1(b) an inverted a-Si
`AMOLED pixel shown in the pixel circuit of Fig. 1(c). The SPICE
`simulation for Vselect = 15 V is also plotted in (b).
`
`FIGURE 5 — The effect of the drift in OLED characteristics caused by
`storing unencapsulated devices in an environment relatively high in
`oxygen and humidity, on the pixel driving current for (a) conventional
`and (b) inverted AMOLED pixels.
`
`Journal of the SID 16/1, 2008
`
`185
`
`SAMSUNG, EXH. 1010, P. 3
`
`

`

`FIGURE 6 — Schematic cross section of
`the “modified” inverted
`AMOLED pixel, during the evaporation of (a) the organic layers and (b)
`cathode.
`
`[Fig. 4(a)], where Ipixel turns on at Vdata = 4.8 V [corre-
`sponding to the threshold voltage of the driving TFT (1.7 V)
`plus the turn-on voltage of the OLED (3.1 V)]. Second, in
`the inverted structure, at typical operational current levels
`of a few microamperes, the pixel current is higher by an
`order of magnitude than the current in the conventional
`structure for the same data voltages. This is because in the
`conventional design the data voltage is split across the
`OLED and the gate source of the driving TFT, but in the
`inverted design it is converted directly to the gate-source
`voltage of the driving TFT. SPICE simulations confirm the
`experimental behavior of the inverted pixels [Fig. 4(b)].
`To further verify the independence of the pixel driving
`current from the OLED characteristics, we compared the
`drift in the output characteristics of conventional and
`inverted AMOLED pixels after storing them in a non-ideal
`environment. The AMOLED arrays, which were not encap-
`sulated, were stored in a nitrogen box with a relatively high
`oxygen content of about 100 ppm, for 6 months. The storage
`condition will not alter a-Si TFTs, but the oxygen content
`and humidity lead to considerable OLED degradation. Figure
`5(a) shows a large drop in Ipixel of conventional AMOLED
`pixels after storage. This is because the OLED degradation
`causes an increase in the voltage drop across the OLED for
`a given current, and thus a higher voltage is required to
`achieve the same VGS(driver) and the same Ipixel in the
`driver TFT. In contrast, Ipixel of inverted AMOLED pixels
`[Fig. 5(b)] is not affected by OLED degradation, an obser-
`vation verifying that Ipixel is independent of the OLED
`characteristics (provided that Vdd is high enough to ensure
`the driver TFT is still in saturation).
`
`186
`
`Hekmatshoar et al./ A novel TFT-OLED integration
`
`FIGURE 7 — (a) Optical micrograph of a modified inverted pixel (Fig.
`6) prior to OLED evaporation and (b) higher magnification of the
`TFT-OLED contact region along with schematic cross section along line
`a–a′. The non-modified inverted structure (Fig. 3) has the same geometry,
`except for the separator which lacks the overhang.
`
`4 Modified integration
`Although the integration process introduced in Fig. 3 real-
`izes the inverted structure of Fig. 1(c) and makes direct pro-
`gramming of the pixel current possible, it is prone to pixel
`yield loss and non-uniformity because it does not allow for
`substrate rotation during the evaporation of organic layers
`and the cathode. To overcome this problem, we have modi-
`fied the integration process by using insulating separators
`with an overhanging projection (Fig. 6) using a double-layer
`photoresist process. Implementation with other methods
`may be possible as well. In our experiment, we have used
`10-µm-high separators with 5 µm of overhang. As shown in
`Fig. 6(a), the organic layers are then evaporated at normal
`incidence and the substrate is rotated during organic evapo-
`ration. The overhang shadows an exposed interconnect
`
`FIGURE 8 — Comparison of the AMOLED pixels fabricated by the
`inverted process of Fig. 3 (no rotation) and the modified inverted process
`of Fig. 6 (with rotation).
`
`SAMSUNG, EXH. 1010, P. 4
`
`

`

`Summary and conclusion
`5
`In summary, we have demonstrated the direct programming
`of a-Si TFT AMOLED pixels using a new integration tech-
`nique that connects the OLED top contact (cathode) to the
`underlying TFT. We have shown that by using a new “inverted”
`integration process the drive current of the fabricated pixels
`becomes essentially independent of the OLED charac-
`teristics and therefore is not affected by OLED aging. Fur-
`thermore, as a result of direct programming, the data
`voltages required for typical pixel operation currents (on the
`order of 1 mA/cm2) drop from about 15 V to about 5 V. The
`pixel yield is increased and the uniformity is improved by
`introducing a modified version of the inverted integration
`process which allows substrate rotation during OLED
`evaporation. This integration approach to the direct pro-
`gramming of a-Si AMOLED pixels may be important for the
`realization of AMOLED displays with a-Si TFT backplanes.
`
`Acknowledgment
`The authors would like to thank the Dupont Company for
`technical collaboration. This work is sponsored by the U.S.
`Display Consortium through the project on 300°C Amor-
`phous TFT Display Backplane Processes on Clear Plastic
`Substrates.
`
`References
`1 R. Dawson, Z. Shen, D. A. Furst, S. Connor, J. Hsu, M. G. Kane, R.G.
`Stewart, A. Ipri, C. N. King, P. J. Green, R. T. Flegal, S. Pearson, W.
`A. Tang, S. Van Slyke, F. Chen, J. Shi, M. H. Lu, and J. C. Sturm, “The
`impact of the transient response of organic light emitting diodes on the
`design of active matrix OLED displays,” Tech. Dig. IEEE Electron Dev.
`Meeting, 875–878 (1998).
`2 M. Hack, J. J. Brown, J. K. Mahon, R. C. Kwong, and R. Hewitt,
`“Performance of high-efficiency AMOLED displays,” J. Soc. Info. Display
`9, No. 3, 191–195 (2001).
`3 R. Dawson, M. G. Kane, Z. Shen, D. A. Furst, S. Connor, J. Hsu, R.
`G. Stewart, A. Ipri, C. N. King, P. J. Green, R. T. Flegal, S. Pearson,
`W. A. Barrow, E. Dickey, K. Ping, S. Robinson, C. W. Tang, S. Van Slyke,
`F. Chen, J. Shi, J. C. Sturm, and M. H. Lu, “Active matrix organic light
`emitting diode pixel design using polysilicon thin film transistors,”
`Conf. Proc. Laser Electron Opt. Soc. Annual Meet. LEOS 1, 128–129
`(1998).
`4 J. Lih, C. Sung, C. Li, T. Hsiao, and H. Lee, “Comparison of a-Si and
`poly-Si for AMOLED displays,” J. Soc. Info. Display 12, No. 4,
`367–371 (2004).
`5 C. C. Wu, S. D. Theiss, G. Gu, M. H. Lu, J. C. Sturm, S. Wagner, and
`S. R. Forrest, “Integration of organic LEDs and amorphous Si TFTs
`onto flexible and lightweight metal foil substrates,” IEEE Electron Dev.
`Lett. 18, No. 12, 609–612 (1997).
`6 M. H. Lu, E. Ma, J. C. Sturm, and S. Wagner, “Amorphous silicon TFT
`active-matrix OLED pixel,” Conf. Proc. Laser Electron Opt. Soc. An-
`nual Meet. LEOS 1, 130–131 (1998).
`7 T. Tsujimura, Y. Kobayashi, K. Murayama, A. Tanaka, M. Morooka, E.
`Fukumoto, H. Fujimoto, J. Sekine, K. Kanoh, K. Takeda, K. Miwa, M.
`Asano, N. Ikeda, S. Kohara, S. Ono, C. Chung, R. Chen, J. Chung,
`C.-W. Huang, H. Guo, C. Yang, C. Hsu, H. Huang,W. Riess, H. Riel,
`S. Karg, T. Beierlein, D. Gundlach, S. Alvarado, C. Rost, P. Muller, F.
`Libsch, M. Mastro, R. Polastre, A. Lien, J. Stanford, and R. Kaufman,
`“A 20-inch OLED displays driven by super-amorphous silicon technol-
`ogy,” SID Symposium Digest Tech. Papers 34, 6–9 (2003).
`8 J. Lih and C. Sung, “Full-color active-matrix OLED based on a-Si TFT
`technology,” J. Soc. Info. Display 12, No. 4, 367–371 (2004).
`
`Journal of the SID 16/1, 2008
`
`187
`
`FIGURE 9 — (a) QVGA checkerboard demonstration of the AMOLED
`pixels fabricated by the modified inverted process presented in Figs. 6(a)
`and 6(b) the drive signals.
`
`which is connected to the driver TFT. The cathode is evapo-
`rated next at an angle while the substrate is being rotated
`[Fig. 6(b)], and therefore the OLED cathode is connected
`to the exposed interconnect and the inverted structure of
`Fig. 1(c) is realized. The cathode may be also evaporated at
`multiple angles for maximum step coverage.
`Figure 7 shows an optical micrograph of a modified
`inverted pixel prior to organic and cathode evaporation. The
`ITO area, excluding its passivated edges, defines the pattern
`of emission. AMOLED test arrays fabricated using the
`inverted integration process shown in Fig. 3 and the modi-
`fied inverted process shown in Fig. 6 are compared in Figs.
`8(a) and 8(b), respectively. It is observed that the modified
`inverted process results in a higher pixel yield and better
`uniformity. A quarter video graphics array (QVGA) checker-
`board demonstration of a 12 × 12 AMOLED test array fab-
`ricated using the modified inverted integration is presented
`in Figs. 9(a) and 9(b) as a proof of high pixel yield and uni-
`formity.
`
`SAMSUNG, EXH. 1010, P. 5
`
`

`

`9 J.-J. Lih, C-F. Sung, M. S. Weaver, M. Hack. and J. J. Brown, “A
`phosphorescent active-matrix OLED display driven by amorphous sili-
`con backplane,” J. Soc. Info. Display 11, No. 1, 14–17 (2003).
`10 K. Long, A. Kattamis, I.-C. Cheng, H. Gleskova, S. Wagner, and J. C.
`Sturm, “Stability of amorphous-silicon TFTs deposited on clear plastic
`substrates at 250°C to 280°C,” IEEE Electron Dev. Lett. 27, No. 2,
`111–113 (2006).
`11 A. Nathan, A. Kumar, K. Sakariya, P. Servati, K. S. Karim, D. Striak-
`hilev, and A. Sazonov, “Amorphous silicon back-plane electronics for
`OLED displays,” IEEE J. Sel. Top. Quantum Electron. 10, No. 1, 58–69
`(2004).
`
`188
`
`Hekmatshoar et al./ A novel TFT-OLED integration
`
`SAMSUNG, EXH. 1010, P. 6
`
`