なかのにっき

1. A8:G208 が装置の吐く csv データの貼り付け場所。 実際には 1 点目と 201 点目はかけ離れた値が出ることが多く、 解析には使用しない。 B 列が電圧、E 列が電流。F 列は E 列の Error(%) だが、 いまのところ解析には用いていない。 C,D 列は RF 測定用の補正で、うちの現在の装置では関係なし。
2. J-L 列でイオン飽和電流の直線フィット。 最初は 9 行目から 10 行分の I-V データを対象に LINEST() でフィット、 その後 1 行ずつデータを追加。r2 値が最大になったところで 傾きと切片を決定。r2 値が最大になる行のインデックスが N3 セル。 傾き・切片が N4:N5。
3. 遷移領域のフィット。電流測定値からイオン飽和電流の直線関数 (M 列) を差し引き、結果を N 列へ。 この値が負となる最大電圧の計測点を拾う (R3 セル)。 フィッティングの低圧側開始点は、この R3 のセル+何点かシフトさせた 点にする。これは人間がグラフを目視して判断。この値は V2 セルに入力する。
   
   あとはイオン飽和電流と同様に LOGST() でフィッテイング (R-T 列)。 r2 が最大になるインデックスが V3、そのときの I=b*m^V の m が V4, b が V5。 結果の関数が U 列。
4. 電子飽和電流。実測電流からイオン飽和電流のフィッティング関数を 引いたデータ (N 列) を対象に、LINEST() で合わせる。 V3 のインデックス値に何 V かを加え、フィッティング領域の 最大値とする。これは人間が AC2 セルに入力。 そこから低圧側に向かってフィット範囲を広げていき、r2 の最大値を拾う。 ただし低圧側の最小値は V3 セルに対応する値まで (←この項を今回追加)。
5. フローティング電圧は、イオン飽和電流と遷移領域の fit 関数の合計が 0 になる値 (絶対値が最小になる値) としてソルバーで解く。結果は AF10 セル。
6. プラズマ電位は、遷移領域と電子飽和電流が等しくなる値 (差の絶対値が最小になる値) としてソルバーで解く。結果は AF13 セル。

We studied the properties of (Nb0.7,Ti0.3)N films deposited by reactive magnetron sputtering in an atmosphere of argon and nitrogen at ambient substrate temperature, with a particular focus on the technological factors that determine film texture. The texture in the nitrides of transition metals determines many processes, including the wear resistance of tool coatings, diffusion in microelectronic devices, and the rate of chemical etching. Thus, since our goal is to use (Nb0.7,Ti0.3)N films in superconducting microelectronic devices, texture control is an essential element of our technology. We find that increasing the total gas pressure, while keeping the film chemical composition constant, results in a decrease in the ratio of the [200] and [111] x-ray diffraction (XRD) line intensities on θ-2θ Bragg-Brentano scans. Similar changes in XRD patterns are observed as the nitrogen injection increases for a constant sputtering pressure. In addition, XRD examination shows that some samples have in-plane texture developed due to self-shadowing during growth. Transmission electron microscopy reveals that all of the films consist of textured, elongated grains. Analyzing the experimental data, it is concluded that the thermalization of the sputtering yield determines the process of texture formation in the experiment with pressure variation, with an increase in adatom energy resulting in a change in texture from [111] to [100]. However, adatom energy is not the only determining factor--the nitrogen concentration in the sputtering gas also has a strong impact on the film texture. In particular, despite the fact that an increase in nitrogen injection results in an increase in adatom energy, the film texture is driven toward [111].

The improvement of the diffusion barrier performance for Cu metallization, by inserting a thin Al layer between two TiN layers, has been clearly demonstrated and reported by us. The key idea behind our scheme is "stuffing" of grain boundaries of columnar TiN films by Al2O3. It has been also found that the barrier property is at its best when the Al thickness is 1 nm, but above this value, the barrier performance degrades drastically when the upper TiN film is not preannealed. In this study, why the barrier breaks down at above 1 nm of Al interlayer thickness is investigated. High-resolution transmission electron microscopy, scanning transmission electron microscopy, and energy dispersive spectroscopy analyses revealed that the fast diffusion of Cu in the presence of the free Al is the main reason for the failure of the present diffusion barrier scheme. These results are discussed on the basis of the differences between the movements of Al and Cu through TiN film, and the differences between the solid solubilities of Al in Cu and in Si. Our results show that both Al interlayer thickness and the oxygen content in TiN film should be properly controlled to take full advantage of the present multilayer diffusion barrier scheme.

I = A + B n

    I - A
n = -----
      B

      |δA|   |I - A    |   |δA|   |n δB|
δn = |---| + |----- δB| = |---| + |-----|
      | B |   |  B^2    |   | B |   |  B  |

      Σn_i/(δn_i)^2
n_m = ---------------
       Σ 1/(δn_i)^2

        Σδn_i/(δn_i)^2    Σ1/(δn_i)
δn_m = ----------------- = --------------
          Σ 1/(δn_i)^2    Σ 1/(δn_i)^2

version:      3.3
author(s):    Mark F. Komarinski, <markk@linuxdoc.org>
              Jorge Godoy, <godoy@metalab.unc.edu>
              David C. Merrill, <dcmerrill@mindspring.com>
updateed:     November, 2000

Lists the tools, procedures, and hints to get LDP authors up
to speed and writing.

• Linux Frequently Asked Questions with Answers
  Robert Kiesling rkiesling@mainmatter.com
  * updated

• BogoMips mini-Howto
  Wim van Dorst, baron@clifton.hobby.nl
  v30, 2000-11-09
  * updated
  
  
• C++ Programming HOW-TO
  Al Dev (Alavoor Vasudevan) alavoor@yahoo.com
  v23.0, 12 Nov 2000
  * updated
  
  
• The Clock Mini-HOWTO
  Ron Bean, rbean@execpc.com
  v2.1, November 2000
  * updated
  
  
• FBB packet-radio BBS mini-HOWTO
  Miroslav "Misko" Skoric, YT7MPB, m.skoric@eunet.yu
  v1.3, 12 November 2000
  * updated
  
  
• LILO, Linux Crash Rescue HOW-TO
  Al Dev (Alavoor Vasudevan) alavoor@yahoo.com
  v4.0, 12 Nov 2000
  * updated
  
  
• Lilo mini-HOWTO
  Miroslav "Misko" Skoric, m.skoric@eunet.yu
  v3.3, 12 November 2000
  * updated
  
  
• Linux+WindowsNT mini-HOWTO
  Miroslav Skoric, m.skoric@eunet.yu
  v2.4, 12 November 2000
  * updated
  
  
• Linux/MIPS HOWTO
  Ralf B=E4chle, ralf@gnu.org
  November 12, 2000
  * updated
  
  
• Database-SQL-RDBMS HOW-TO document for Linux (PostgreSQL
  Object Relational Database System)
  Al Dev (Alavoor Vasudevan) alavoor@yahoo.com
  v37.0, 12 Nov 2000
  * updated