Skip to content
TheChatArchiveThe Chat Archive
Coherecommand-a-03-2025· CohereHugging Face LMArena Human Preference 140K

Amorphous In-Ga-Zn-O (a-IGZO) thin-film transistors (TFTs) are considered a key…

From a Chatbot Arena battleThe same person sent this prompt to two anonymous models at once. Compare it with chatgpt-4o-latest-20250326’s answer.
  1. User

    Amorphous In-Ga-Zn-O (a-IGZO) thin-film transistors

    (TFTs) are considered a key technology for next-generation displays and

    electronics, owing to their excellent electrical properties and potential for

    large-area uniformity [1], [2]. Among various device architectures, the

    top-gate self-aligned (TG-SA) structure is particularly favored as it offers

    significant advantages in reducing parasitic capacitance and enhancing

    scalability, which are critical for high-performance circuits [3], [4].

    However, a significant bottleneck limiting the

    performance and reliability of TG-SA a-IGZO TFTs is their inherent

    vulnerability to hydrogen (H). The unwanted incorporation of

    H often leads to severe

    operational instability, most notably a significant negative shift in the

    threshold voltage (VTH​)

    [5], [6], [7], swing slope (S.S) degradation, and hump phenomenon. While extensive research has addressed H diffusion from

    various sources like the gate insulator or etch-stop layers [8], [9], [10], a

    critical and often-overlooked pathway in the TG-SA architecture is the lateral

    diffusion of hydrogen through the source/drain (S/D) metal contacts. This lateral diffusion pathway becomes acutely

    problematic during necessary post-fabrication thermal treatments, such as

    forming gas anneal (FGA) or chemical vapor deposition (CVD) processes.  The issue is exacerbated

    as the channel length (LG​) is scaled down, which increases the proximity of

    the hydrogen-rich S/D contacts to the active channel area [11], [12].

    Therefore, suppressing this lateral hydrogen diffusion without compromising the

    benefits of thermal annealing is a critical challenge for securing the intended

    electrical characteristics in scaled TG-SA a-IGZO TFTs [13], [14].

    In this letter, we demonstrate that a tungsten nitride
    

    (WN) diffusion barrier integrated into the S/D contacts is a highly effective

    solution to this issue. By inserting a WN layer, the lateral diffusion of

    hydrogen is successfully suppressed, allowing the device to withstand a 300°C

    FGA process without any degradation in its electrical properties. This result

    highlights a viable pathway to achieving both stable operation and high

    performance in scaled TG-SA a-IGZO TFTs. 내용을 살리면서 절반정도로 분량을 줄여줄래?

  2. Cohere Assistant

    Amorphous In-Ga-Zn-O (a-IGZO) thin-film transistors (TFTs) are pivotal for next-generation displays and electronics due to their superior electrical properties and large-area uniformity [1], [2]. The top-gate self-aligned (TG-SA) structure is particularly advantageous for reducing parasitic capacitance and enhancing scalability, critical for high-performance circuits [3], [4]. However, TG-SA a-IGZO TFTs are susceptible to hydrogen (H) incorporation, leading to operational instability, such as negative threshold voltage (VTH) shifts, swing slope (S.S) degradation, and hump phenomena [5]–[7]. While H diffusion from gate insulators or etch-stop layers has been studied [8]–[10], lateral H diffusion through source/drain (S/D) metal contacts, especially during post-fabrication thermal treatments like forming gas anneal (FGA), remains a critical yet overlooked issue. This problem intensifies with channel length (LG) scaling, as H-rich S/D contacts approach the active channel [11], [12].

    To address this, we introduce a tungsten nitride (WN) diffusion barrier into the S/D contacts, effectively suppressing lateral H diffusion. This innovation enables TG-SA a-IGZO TFTs to withstand 300°C FGA without electrical degradation, offering a robust solution for achieving stable and high-performance scaled devices [13], [14].

Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.