Oxide TFTs:  A Progress Report

For flat-panel-display backplane applications, oxide-TFT technology is the new kid on the block – recently conceived and in its early stages of commercialization.  How is it going for oxide-TFT technology as it attempts to match up with a-Si:H and LTPS technology?

by John F. Wager

OXIDE thin-film-transistor (TFT) products are a recent development.  The path toward commercialization began in the spring of 2012, with Sharp’s announcement that Gen 8 Kameyama Plant No. 2 would be retrofitted to the production of indium gallium zinc oxide (IGZO) LCD panels for tablets, high-resolution notebook PCs, and monitors.  Sharp’s first IGZO product was the Aquos Phone Zeta SH-02E, appearing (in Japan) in the fall of 2012.  In early 2013, LG Display began taking pre-orders (in South Korea) for its 55-in. OLED TV, which employed an IGZO-TFT backplane.

In retrospect, the period of early commercialization of oxide TFTs that occurred between spring 2012 and fall 2014 can be characterized as a time of high expectations, over-promising while under-delivering, skepticism, hype, over-optimistic timelines, misinformation, excitement, apprehension, giddiness, and confusion.  Thus, in all respects the introduction of oxide-TFT technology was a “situation normal” rollout.

Oxide-TFT early commercialization drama centered on four main questions.  First, did Sharp know something that no one else did?  Did it possess a manufacturing “secret sauce” that would preclude other companies from duplicating its oxide-TFT success?  Was CAAC (c-axis-aligned crystallinity) the secret?  Second, was oxide-TFT technology just another flash-in-the-pan technology that would either elude being tamed by high-volume manufacturing or would underperform when eventually productized?  Third, many observers questioned whether oxide-TFT technology could really compete with low-temperature polysilicon (LTPS), especially given LTPS’s multi-decade head start, commercial entrenchment, and perceived performance advantages.1  Fourth, rumors about Apple were flying in all directions.  Was Apple really moving toward the use of oxide TFTs?  If so, what was holding it up?

To a large extent, this era of oxide-TFT confusion and uncertainty ended mid-October 2014 with Apple’s launch of its iMac with its 5K Retina display.  This product had an IGZO-TFT backplane.

Oxide TFTs or IGZO

Some readers may be confused: If this article pertains to oxide TFTs, why is IGZO the only oxide being discussed?  Aren’t there any other oxides of relevance out there?  Good question.

Oxides are an attractive class of materials for commercialization for several reasons.  First, from an anion classification perspective, oxides are the most numerous compounds on the planet, by far.  Thus, oxides offer a rich palette of possibilities.  Second, oxides tend to be thermodynamically stable.  This attribute often leads to excellent product lifetime and reliability.  Third, oxides are air processable.  This fortuitous property offers a possible path toward lower manufacturing costs.

Oxides are not a materials panacea, however.  For an oxide to be useful, it must possess requisite attributes for the application of interest.  For flat-panel-display backplane applications, a candidate oxide should have (at least) the following properties: (1) possess an electron mobility significantly greater than that of hydrogenated amorphous silicon (a-Si:H), i.e., μ_n >> 0.5 cm²/V-sec, (2) exhibit superior TFT stability compared to that of an a-Si:H TFT, and (3) be capable of scaling to large areas (Gen 10+) at a price competitive to that of a-Si:H.

IGZO meets all of these flat-panel-display backplane requirements.  Currently, IGZO is the only oxide that can.  Thus, at this time, oxide TFTs and IGZO can be considered to be synonymous and interchangeable terms, at least from the perspective of commercialization.

Oxide-TFT Products

Product assessment is one approach for answering the question of whether oxide-TFT technology is “real.”  Table 1 is a list of IGZO-based products that were either available (blue) or had been announced (red) as of mid-October 2015.

Table 1 is surprising, and very encouraging.  Adopting the usual perspective that oxide-TFT technology is in its infancy, the number of IGZO products already available via a simple Internet search seems astonishing, especially given IGZO’s current lack of manufacturing capacity, as discussed in the next section of this article.  Additionally, the product breadth found in Table 1 is auspicious – phones, monitors, tablets, laptops (even gaming laptops), desktops, and TVs, i.e., products with small, medium, large, and very large areas.  These considerations argue against the assertion that IGZO is a one-trick-pony technology, capable of addressing only a limited and perhaps even a niche market.

---

Oxide-TFT Manufacturing

As alluded to in the discussion of Table 1, it appears likely that the oxide-TFT product market is currently limited by the lack of high-volume manufacturing capacity, as well as by the newness of the technology itself.  Table 2 is an attempt to estimate current and near-term oxide-TFT production activities.

The primary take-away from Table 2 is that three companies appear to be actively shipping oxide-TFT panels, i.e., Sharp, Samsung, and LG Display.  All three companies have announced their intention to ramp up their oxide-TFT production volume in 2016, although Sharp has not yet specified this quantitatively.  Note that the sheets per month unit used in Table 1 is only part of the production volume story, given areal differences associated with fab size.  For example, a Gen 5 sheet of glass has dimensions of 680 × 880 mm (0.6 m²), whereas a Gen 8 sheet of glass is 1870 × 2200 mm (4.1 m²).  The three other players listed in Table 2, i.e., BOE, CPT, and CEC Panda, have publicly announced that they expect to be shipping oxide-TFT products soon.  However, it is not yet clear if they will be able to hit their late-2015 timelines, or if product introduction will be delayed until sometime in 2016.

---

Total worldwide oxide-TFT capacity was estimated as ~65M m² in 2014 and is projected to be ~125M m² (2015) and ~195M m² in 2016.³  This same flat-panel-display production capacity assessment predicts that 2016 oxide-TFT capacity will overtake that of LTPS, which is projected to be ~180M m² in 2016.

Commercial adoption of oxide-TFT technology requires the availability of appropriate equipment for large-area high-volume manufacturing.  Figure 1 is an example of the type of tool innovation that is required before a new technology such as IGZO is indeed ready for commercialization. The large-area coater shown in Fig. 1 is capable of producing uniform IGZO films over a 2200 × 2500 mm (Gen 8.5) substrate.  The use of rotary targets improves target lifetime and production yield, reduces particle generation, and lowers manufacturing costs compared to that obtainable using planar targets.⁷  (For more about this system, see the article, “A Process for Using Oxide-TFTs over LTPS-TFTs for Better OLED-TV Manufacturing” in the November/ December 2015 issue.)

Fig. 1:  This rotary target array consists of 12 Gen 8.5 IGZO sputter targets (a subset is pictured at left) installed in an Applied Materials PIVOT^®-sputtering system.  (For a sense of scale, note the person standing in front of the system.)  Image courtesy Applied Materials, Inc.

Tools such as the one discussed in Fig. 1 are capable of fabricating IGZO TFTs with excellent properties and with acceptable uniformity even over Gen 8.5 dimensions.  Figure 2 summarizes typical IGZO-TFT performance results.  Note that a mobility of 12 cm²/V-sec is 24 times larger than that typically obtained for an a-Si:H TFT.  Also notable is the large drain current on-to-off ratio of 10⁸.  This is several orders of magnitude better than that obtainable using a-Si:H or LTPS TFTs.  This translates into a lower power advantage for IGZO TFTs, an important and potentially game-changing distinction.^1,8  Finally, the IGZO-TFT stability specs reported in Fig. 2 in terms of positive- and negative-bias temperature stress (PBTS and NBTS) and negative-bias illumination stress (NBIS) are all quite encouraging, especially when it is realized that this result is obtained using Gen 8.5 tools.

Fig. 2:  IGZO-TFT performance obtained using Gen 8.5 manufacturing tools is shown at left and right.  Image courtesy Applied Materials, Inc.  PBTS = Positive-Bias Temperature Stress.  NBTS = Negative-Bias Temperature Stress.  NBIS = Negative-Bias Illumination Stress.

NBIS is a topic that has been extensively discussed in the literature.  It was originally considered to be a possible show-stopper for the commercialization of IGZO technology.  It appears that the industrial solution to circumventing the deleterious effects of NBIS often involves the use of light shielding.⁹

Oxides Other Than IGZO

As mentioned previously, IGZO and oxide TFTs are different terms used to refer to essentially one and the same thing, at least in the context of products, since IGZO is the only oxide currently being commercialized for flat-panel-display backplane applications. This does not mean that IGZO is the only oxide that will or can ever be employed in a future display or non-display product.

Then why is IGZO the only oxide currently being marketed?  In a word, it comes down to standardization.  For a market as large as that of the flat-panel-display industry, panel manufacturers have to work closely with tool vendors, materials suppliers, and others in order to successfully implement a new technology.  This is a vast undertaking.  Many different partners, as well as competitors, have to agree to cooperatively explore a new technology in order for innovation relevant to high-volume manufacturing to occur.  These kinds of co-operative undertakings occur all the time in the semiconductor integrated-circuit industry, giving rise to the formulation of roadmaps that are useful for guiding R&D activities.

Cooperation leads to standardization.  About 5 years ago, IGZO was informally selected as a consensus best-of-class candidate oxide for the pursuit of commercialization.  Many other oxides were being investigated 5–10 years ago for display applications.  Although oxide-TFT technology then looked like an attractive flat-panel-display backplane option, panel manufacturers, tool vendors, and materials suppliers were unwilling to seriously devote their time and resources to an unproven oxide-TFT technology until a viable contender material was identified.  IGZO was chosen.  Recent oxide-TFT technology trends indicate that IGZO was an excellent choice.

Which oxides other than IGZO are likely to emerge in future applications?  The answer to this question is not entirely obvious.  Here are some thoughts.  First, a viable candidate oxide should have an amorphous microstructure.  a-Si:H and IGZO are scalable to Gen 8.5+ sizes for one primary reason; they are amorphous.  Second, an IGZO replacement must offer some sort of compelling manufacturing and/or performance advantage.  Consider two examples.  I think that it is highly likely that a future oxide-TFT material will contain tin as a constituent element, since incorporation of tin will allow oxide wet etchability to be dramatically engineered over a broad range of selectivity.¹⁰  In this respect, recent industrial R&D involving ITZO¹¹ and IGZTO^12,13 is intriguing.  Next, and most importantly, the flat-panel-display industry seems to have an insatiable desire for ever-increasing TFT-channel-layer electron mobility (as well as hole mobility, if it could only get it).  Alternative channel layers such as ITZO¹¹ may prove capable of satisfying next-generation oxide-TFT electron- mobility requirements.

An alternative approach for improving TFT-channel-layer electron mobility is to employ a dual-layer stack.  Typically, amorphous oxides with the highest mobilities also possess very high electron-carrier concentrations.  This makes it difficult to fabricate enhancement-mode TFTs with turn-on voltages near zero.  Rather, a negative gate voltage is required to deplete electrons in the channel in order to turn off the current.  This results in depletion-mode TFT behavior.  A dual-layer channel TFT employs a very thin high-mobility oxide at the interface to the gate insulator, which is covered by a thicker more insulating oxide.  This approach allows for concomitant optimization of mobility and threshold voltage.

Figure 3 illustrates results achieved using the dual-layer TFT approach for an IZO/IGZO TFT. The 62 cm²/V-sec mobility reported in Fig. 3 for a 10-nm-thick IZO TFT suggests that oxide TFTs with mobilities 100 times greater than that of a-Si:H are likely to be a commercial reality in the not too distant future.  Note also that preliminary TFT stability data is very encouraging for these dual-layer IZO/IGZO TFTs.

Fig. 3:  The above charts reveal high-mobility IZO/IGZO dual-layer TFT performance.  Image courtesy Applied Materials, Inc.

Right Material, Wrong Reason

The scientific rationale for exploring amorphous- oxide semiconductors, such as IGZO, was first articulated by Hosono and co-workers, way back in 1996.¹⁴  They pointed out that large ionic radius, spherically symmetric s-orbitals that form the conduction bands in an amorphous oxide semiconductor offer the potential for higher mobility than that obtainable in a-Si:H.  More simply, a larger orbital overlap promotes better electronic exchange between atoms, leading to higher mobility.  Although this simple picture has been universally adopted and promulgated, by me, among many others, I no longer believe it to be true.

To see this, consider the following simple equation¹⁵:

μ_drift = [n/(n + n_T)]μ₀.       (1)

This equation says that the drift mobility, μ_drift, i.e., the relevant mobility responsible for channel current in a TFT, is equal to a trap-free mobility, μ₀, times the fraction of electrons in the channel that are free, not trapped, where n is the free-electron carrier concentration and n_T is the trapped electron concentration.  Formulation of a physics-based model for carrier mobility in an amorphous semiconductor then is used to demonstrate that the trap-free mobilities for a-Si:H and IGZO are rather similar.¹⁵  Thus, differences in mobility between a-Si:H and IGZO are found to be a consequence of dissimilar trapping tendencies.  Electron trapping is more pronounced in a-Si:H than in IGZO.  In turn, these electron-trapping tendencies are found to be related to the nature of disorder in each material, i.e., bond-angle variability for a-Si:H and cation-sublattice disorder for IGZO.

Without getting any more deeply enmeshed in details associated with the chemistry and physics of amorphous semiconductors, here is the point.  If the picture proposed by Hosono and co-workers is accurate, mobility differences between a-Si:H and IGZO should arise from differences in μ₀ in Eq. (1).  However, this is not found to be the case.  This new perspective is perhaps somewhat disappointing since the chemical insight inherent in the Hosono picture provided a very compelling motivation for the development of amorphous- oxide semiconductors as a new class of materials.

Perhaps it really does not matter at this point.  In any event, I think that it is better that my good friend Professor Hosono got the right answer, although perhaps for the wrong reason, than if he had gotten the wrong answer, but for the right reason.  So with regard to the question of how’s it going with all that oxide-TFT stuff, the answer is: “Great! Thanks for asking.”

Acknowledgments

I wish to thank D. K. Yim of Applied Materials, Inc., for supplying manufacturing data and several of the figures used in this article.

References

¹J. F. Wager, “Flat-Panel-Display Backplanes: LTPS or IGZO for AMLCDs or AMOLED Displays?,” Information Display 2, 26–29 (2014).

²http://www.patentlyapple.com/patently-apple/2015/08/apple-to-lead-the-way-for-next-gen-low-powered-oxide-tft-panels-in-ipad-pro-other-mac-products.html

³http://en.ofweek.com/news/LG-Display-accelerating-on-oxide-TFT-investment-27751

⁴http://www.4-traders.com/BOE-TECHNOLOGY-GROUP-CO-13491315/news/BOE-Technology-Chongqing-85G-Production-Line-in-Operation-20022680/

⁵http://www.cens.com/cens/html/en/news/news_inner_48620.html

⁶http://en.ofweek.com/news/CEC-Panda-to-begin-mass-production-of-IGZO-panels-in-Q3-29786

⁷E. Scheer, O. Graw, K. Schwanitz, A. Gaur, A. Hosokawa, R. Lim, Y. Ye, D. K. Yim, H. You, and M. Bender, “Advances in High Performance, High productivity Metal Oxide Films,” IMID Digest, 308–309 (2012).

⁸Sharp, “IGZO: vision for the future,” http://online.wsj.com/ad/article/vision-breakthrough.

⁹C. Ha, H-J. Lee, J-W. Kwon, S-Y. Seok, C-I. Ryoo, K-Y. Yun, B-C. Kim, W-S. Shin, and S-Y. Cha, “High Reliable a-IGZO TFTs with Self-Aligned Coplanar Structure for Large-Sized Ultrahigh-Definition OLED TV,” SID Symp. Digest Tech. Papers, 1020–1022 (2015).

¹⁰J. F. Wager, B. Yeh, R. L. Hoffman, and D. A. Keszler, “An amorphous oxide semiconductor thin-film-transistor approach to oxide electronics,” J. Curr. Opin. Solid State Mater. Sci. 18, 53–61 (2014).

¹¹N. Morosawa, M. Nishiyama, Y. Ohshima, A. Sato, Y. Terai, K. Tokunaga, J. Iwasaki, K. Akamatsu, Y. Kanitani, S. Tanaka, T. Arai, and K. Nomoto, “High-mobility self-aligned top-gate oxide TFT for high-resolution AMOLED,” J. Soc. Info. Display 21, 467–473 (2014).

¹²T. Sun, L-Q. Shi, C-Y. Su, W-H. Li, X-W. Lv, H-J. Zhang, Y-H. Meng, W. Shi, S-M. Ge, C-Y. Tseng, Y-F. Wang, C-C. Lo, and A. Lien, “Amorphous Indium-Gallium-Zinc-Tin-Oxide TFTs with High Mobility and Reliability,” SID Symp. Digest Tech. Papers, 766–768 (2015).

¹³C. Ha, H-J. Lee, J-W. Kwon, S-Y. Seok, C-I. Ryoo, K-Y. Yun, B-C. Kim, W-S. Shin, and S-Y. Cha, “Electrical Characterization of BCE-TFTs with a-IGZTO Oxide Semiconductor Compatible with Cu and Al interconnections,” SID Symp. Digest Tech Papers, 853–856 (2015).

¹⁴H. Hosono, M. Yasukawa, and H. Kawazoe, “Novel oxide amorphous semiconductors: transparent conducting amorphous oxides,” J. Non-Cryst. Solids 203, 334–344 (1996).

¹⁵K. A. Stewart, B-S. Yeh, and J. F. Wager, “Amorphous semiconductor mobility limits,” J. Non-Cryst. Solids (in press).  •

---

John F. Wager holds the Michael and Judith Gaulke Endowed Chair in the School of EECS at Oregon State University, and is a SID Fellow.  He can be reached at jfw@eecs.oregonstate.edu.