関西学院大学リポジトリ

全文

(1)STUDIES OF OXIDATION― INDUCED STACKING FAULTS IN SttLICON. by Yoshinori Hayafu]■. B. Sc′ Kwansei Gakuin University′. 1967. Mo Sc′ Kwansei Gakuin University′. 1969. ｀・・. 11‐. Thesis submitted to the Faculty ︱. of Sc■ ence′ Kwanse■ Gaku■ n University in Candidacy for. the Degree of Doctor of Science 事. September. ′ 1983.

(2) ABSTRACT The aim of this thesis is to Study the behavior of pOint ■n s■ licon dur■ ng defects and ox■ dat`■ on― ■nduced stacking faults thermal n■ tr■ dation of s■ licon and ox■ dized― s■ licOn. This thes■ s also presents an investigation of the methods of COntro1 0f stacking faults dur■ ng thermal ox■ dation of S■ liCOn′ as well as the mechan■ sms of these methods. Thermal n■ tr■ dation of s■ licon and ox■ d ized― s■ liCOn ■n an ammonia ambient forms thin′ amorphous silicon nitride filmS. In the nitridation of silicon′ silicon interstitials migrate from the s■ licon― s■ licon n■ tr■ de ■nterface to the s■ liCOn n■ tr■ de surface to react with ammonia or its derivatives. In the nitridation of oXidized― silicon′ the reaction of ammOnia with Which iS then silicon dioxide decomposes the silicOn dioxide′ converted to silicon oxynitride. During the course of Silicon dioxide decomposition′ free silicon atoms are generated′. injected into silicon as excess silicon self―. then. interstitials.. ■n nk in s■ liCOn and groW stacking faults shr■ oxidized― silicon during the nitridation of SiliCOn and oxidiZed― This thes■ s expla■ ns the behav■ or of oX■ dation― ■nduced s■ licon.. ox■ dation― ■nduced. stacking faults during the nitridation by a model based On the licon ■nterstitials near the S■ li― degree of saturation of the s■ con surface. Three new methods of controlling stacking faults during thermal oxidation of silicon are preSented. The first method′ ngr Suppresses the generatiOn Of called laser damage getter■ ng ox■ dation Of S■ liCOn. ox■ dation― ■nduced stacking faults dur■ This suppress■ on ■s attr■ buted to the action of the stress field nt ■nduced defects whiCh getters po■ around thermally stable laser― defects. In the second method′ viCinal surfaces of (001)Orien― tation are used to ann■ hilate ox■ dation― ■nduced staCking faults during the oxidation of siliconI The faults are annihilated by the unfaulting reaction of Frank partial dislocation to perfect dislocation. In the third method′ laser beam irradiation removes the oxidation― induced stacki′ ng faults in the surface region of silicon. The mechanism of this method is based On the liquid phase epitaxial regrowth of the area melted by laSer irradiation..

(3) ZUSAMMENFASSUNG Ziel dieser Dissertation ist das Studttum des Verhaltens von punktformigen Fehlstellen und von durch Oxydation eingelettteten Stapelfehlern in Silttzium wahrё nd des thermischen Nitridierens von Silizium und oxydiertem Silizium. Sie untersucht auch. Kontrollmethoden von Stapelfehlern wahrend der therhischen Oxydation von Sttlizium sowie das Prinzip dieser Methoden. Thermisches Nitridieren von Silizium und oxydiertem Silizium in Ammoniak… Atmosphare bildet dunne amorphe Silizium― Nitrid― Filme. Beim Nitridieren von Silizium wandern Zwischengitter― Silizium― Nitrid― Grenzschicht zur siliziumatome von der Silizium― Siliziumnitridoberfliche und reagieren lrlit Ammoniak oder seinen um zersetzt die Der■ vaten. Be■ m Nitr■ dieren von oxydiertem Siliz■ Oxynitride. Reaktion lnit Ammoniak das Siliziumdioxyd zu Silizium―. wahrend der Siliziumdioxyd― Zersetzung werden freie Siliziumatome geschaffen′ welche anschliessend in die Siliz■ umstruktur als Ueberschuss― Zwischengitteratome eingelagert werden. Oxydations― bedingte Stapelfehler in Silizium fuhren zu verdiChtungF in oxydiertem Silizium fuhren sie beim Nitridieren von Silizium zum Gitterwachstum. In dtteser These wird das Verhalten der oxydationsbedingten Stapelfehler wahrend der Nitr■ dierung durch ein Modell erklart′ welches auf dem shttigungsgrad der Silizium― Zwischengitteratome nahe der Siliziumoberflache beruht. Dre■ neue Mathoden zur Kontrolle der Stapelfehler wahrend der thermischen Oxydation von Silizium werden beschrieben. Die erste Methode′ Laser― Fehlstellen― Sammeln (Gettering)genannt′ verhutet die Bildung oxydationsbedingter Stapelfehler whhrend der Oxydation von Silizium. Diese Tatsache kann der Aktion des Spannungsfeldes zugeschrieben werden′ welches um thermostabile laser■ nduz■ erte Fehlstellen entsteht.. ―. ■ ■■. 一.

(4) In der zweiten Methode werden Nachbarflttchen zu eineビ (001) Or■ entierung verwendet′ um oxydations― ■nduz■ erte Stapelfehler wahrend der Oxydation von Siliz■ um zu annullieren. Diese Fehler werden durch dtte`Franklsche Reaktion zwischen Teilversetzungen und perfekten Versetzungen geheilt. In der dritten Methode werden in der Siliziumoberflache durch Laserbestrahlung oxydationsbedingte Stapelfehler entfernto Der Mechan■ smus dieser Methode basiert auf der epitaxischen Flussigphasen― Fl五 chenre― generation der durch die Laserbestrahlung geschmolzenen Oberflache.. ―. ■V. ―.

(5) ACKNOWLEDGEMENTS The authOr would like to express his sincere appreciation to Professor R. Shintan■ of Kwanse■ Gaku■ n Un■ vers■ ty for hiis kind. and valuable adV■ Ce throughout the course of this work. This work yas done at Sony Corporation Research Center. gu■ dance. l. The author wishes to express special thanks to Dr. M. Kikuchi′ director of the Research Center′. Dr. N. Watanabe′ deputy direc― tor′ the late Dr. Z. Ishii′ and Mr. S. Usui′ head of the thin fillm sem■ conductor research laboratory for the■ r continuous support and valuable advice to accomplish this work. The author is very grateful to Mr. s. wakayama′ general manager of patent div■ s■ On′ and Dr. S. Kawado′ head of the mater■ al character― ization laboratory for bringing this field of research to his attention.. The cooperation of Mr. K. Kajiwara in the study of nitrida― tion of s■ licon and ox■ dized s■ licon ■s acknowledged w■ th grati― tude. The author wishes to acknowledge motivating discussions on defects in silicon crystal with Drs. Ae Shibata and Y. Aoki′ and Messrs. T. Yanada′ U. sato and 」 . Aoyama. An express■ on of thanks is also due to Mrs. S. Kumagai for her diligent technical assistance throughout this work. Thanks are also due to Mr. H. Hayashi fOr his help ■n fabricating the silicon devices and to Miss E. Kamogawa for her skillful technique in Auger electron spectroscopic measurement. Finally′ the author wishes to thank his wife′ Yoshiko′ his sons′ Yoshiyuki and 」unji and his parents′ Tadao and Yasue for the■ r continuous encouragement and support.. V. ―.

(6) PAPERS REPRODUCED IN THIS THESttS. Chapter 2. Y. Hayafuji and K. Kajiwara: Nitridation of Silicon and Oxidized― J.Electrochem.Soc。 ′ 129 2102 (1982).. Chapter 3. Silicon′. Y.Hayafuji′ K.Kajiwara′ and S.Usui: Shrinkage and Growth of Oxidation Stacking FaultS dur■ ng Thermal Nitr■ dation of Silicon and Ox■ dized― Silicon′ J.Appl.Phys.′ 53 8639 (1982).. Chapter 4. Chapter 5. Y.Hayafu]■ ′ T.Yanada′ and Y.Aoki: Laser Damage Gettering and lts Application to Lifetime lmprovement in Silicon′ J.Electrochem.Soc.′ 128 1975 (1981).. Y.Hayafuji and S.Kawado: Stacking Fault Ann■ hilation Dependence on Surface Or■ entation ■n Silicon′ JoAppl.Phys.′ .53 1215 (1982).. Chapter 6. Y.Hayafuji′ Y.Aokir and S.Usui: Direct Measurement of the Melt Depth of Silicon during Laser lrradiation′ Appl.Phys.Lett.′ 42 720 (1983) and Y.Hayafu]■. Y.Aoki′ A.Shibata′ and S.Usui: Laser― Enhanced Nucleation of Oxidation― Induced Stacking Faults ■n Silicon′ ′ 」 .Ogawa′. 」 .Appl.Phys.′. 54 3606 (1983).. ―. V■. ―.

(7) OTHER PAPERS RELATED TO THE THESIS. l). Y.Hayafuji′ s.Kawado′ and Z.工 shi: Knoop Hardness Of Phosphorus― Diffused Silicon Single Crystals′ Jpn.」 .Appl.Phys. ll 1389 (1972).. 2). Y.Hayafuji and s.Kawhdo: The Relationship_between Surface Damage due to Scratching and Knoop Hardness in Phosphorus― Diffused Silicon Crystals′ Denki Kagaku 41 508 (1973).. 3). s.Kawado′ Y.Hayafuji″ and T.Adachi: Observation of Lattice Defects ■n Silicon by Scann■ ng Electron MicroscOpy Utilizing Beam lnduced Current Generated in Schottky Barriers′ Jpn.J.Appl.Phys. 14 407 (1975).. 4). Y.Hayafuji′ Toshimada′ and S.KaWado: DislocatiOn configuration under Periodic Stress Field Caused by a Poly― Silicon Film on Silicon′ Silicon 1977 (The ElectrocheIIlical Society′ p.750.. Semiconductor Princeton′ 1977). 5). Y.Hayafuji′ T.Yanada′ S.Usui′ S.Kawado′ A.Shibata′ N.Watanabe′ M.Kikuchi′ and K.E.Williams: Laterally Seeded Regrowth of silicon over Si02 through Strip Electron― Beam ttrradiation′ Appl.Phys.Lett. 43 (1983).. 6). Y.Hayafuji′ S.Kawado′ and z.Ishii: Knoop Hardness change in Phosphorus― Diffused Silicon Single Crystal′ The 2nd SympOsium On Serrliconductors and lntegrated Circuits ・ Of the Electrochemical Society of Japan′ Apri1 1972′ p.63′ E17.. ―. V■. ■. ….

(8) 7). Y. Hayafuji′ T. Yanada and S. Kawado: Contro1 0f Diffusion― Induced Defects by Lapping the Back Surface of silicOn wafers′ The 6th sympOsium on SeFniCOnductors and lntegrated Circuits of the Electrochemical Society of Japan′ Apri1 1974′ p. 37′ E10.. 8). s. Kawado′ Y. Hayafuji and T. Adachi: SEM Observation of Dislocations in Phosphorus Diffused Silicon′ The Spring Meeting Of the Japan Society of Applied Physics′ Apri1 1975′ 4p― B-10。. 9). S. Kawado′ Y. Hayafuji and T. Adachi: Observation of Lattice Defects ■n Silicon Us■ ng a New SEM― EBIC Technique′ Extended Abstracts′ Electrochem■ cal Soc■ ety′ Vol. 75-2′ 1975′ Abstract No. 184 (Electrochemical Society′ Princeton′ 1975).. 10). Y. Hayafu」. i′. T. Yanada and s. Kawado:. Orientattton Dependence of stacking Fault Annihilation in Silicon′ The Fall Meeting Of the Japan Society of Applied Physics′ October 1977′ 15a… W-4.. 11). T. Yanada′ Y. Hayafuji and S. Kawado: GeneratiOn of Microprecipitates in silicon during Thermal Ox■ dation′. The Fall Meeting Of the. 」apan. Society of Applied Physics′. October 1977′ 15a― W… 3.. 12). Y. Hayafuji′ Y. Aoki and S. Kawado: Defect cOntrol and Lifetime lmprovement in Silicon through Laser Damage Getter■ ng′ The Fall Meeting Of the,Japan Society of Applied Physics′ November 1978′ 3p一 M-12.. V■. ■■. ―.

(9) 13). Y. Aoki′ Y. Hayafuji and S. Kawado: Role of Oxygen in silicon on Defect Generation and MOS Character■ stics′ Extended Abstracts′ Electrochemical Socttety′ Vol.° 78-2′ 1978′ Abstract No. 207 (Electrochemical Society′. Princeton′. 1978). 14). Y. Ichida′ Y. Hayafu]■ ′ S. Kawado and T. Shimada: Electr■ cal Character■ stics of Lattice Defects ■n Silicon′ The Spring Meeting of the 」apan Society of Applied Physics′ March 1978′ 27p― T-6.. 15). Y. Hayafuji′ T. Yanada and Y. Aoki: Laser Gettering and lts Application to Low Temperature Process′ Extended Abstracts′ Electrochemical Society′ Vol. 79-2′ 1979′ Abstract No. 485 (Electrochemical Society′ Princeton′ 1979).. 16). Y. Hayafuji′ T. Yanada and Y. Aoki: Laser Gettering and lts Application to Low Temperature Process′. The spring Meeting of the Japan sOciety of Applied Physics′ March 1979′ 27p― X-9.. 17). J, ogawa′ Y. Sator Y. Hayafuji and Y. Aoki: EBIC Observation of Defects in Laser lrradiated Silicon′ The Spring Meeting of the Japan Society of Applied Physics′ Apri1 1980′ la― C-9.. 18). Y. Aoki′. J. Ogawa′ Y. Sato and Y. Hayafuji: ■n The Generation and Ann■ hilation of Stacking FaultS Silicon by Laser Annealing′ The Sprttng Meeting of the Japan Society of Applied Physics′ Apri1 1980′ la― C-10.. ―. ■X. ―.

(10) 19). Y. Hayafuji and K. Kajiwara: NitridatiOn of oxidized silicon′ The Fall Meeting of the Japan Society of Applied Physttcs′ October 1981′ 7p― W-17.. 20). K. Kajiwara′ T. Yanada and Y. Hayafuji: The Enhancement Effect of lon― Implanted Atoms on Direct NitridatiOn of silicon′ The Fall Meeting of the Japan Society of Applied Physics′ October 1981′ 8a― W-1.. 21). Y. Hayafuji′ T. Yanada′ S. Usui′ S. Kawado′ A. Shibata′ N.Watanabe′ M. Kikuchi and K. E. Williams: Lateral Epitaxial Growth of Si over Si02 through Strip Electron― Beam lrradiation′ Extended Abstracts′ Electrochemical Society′ Vol. 83-1 1983′ Abstract No. 374 (Electrochemical Society′. Princeton′. 1983). 22). Y. Hayafuji′ T. Yanada′ s. Usui′ S. Kawada′ A. Shibata′ N.Watanabe′ M. Kttkuchi′ H. Hayashi and K. E. Williams: Recrystallization of Polycrystalline si over Si02 through Strip Electron― Beam lrradiation′ The lst U.s.― 」apan Seminar on Solid Phase Epitaxy and lnterface Kinetics.. X. ―.

(11) CONTENTS Page l. ユ.l. l.2. ●●●●●●●●●●●●●●●●●●●●●●●●● General lntroduction .......● Background e....● ●●●●●00● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●● Scope of This Thesis ........● References ......● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●. l l 4 7. [PART ONE] KINETICS OF SILICON ttNTERSTITIALS AND STACKttNG FAULTS DURING THERMAL NITRIDATION 2. 2.l 2.2 2.2.l 2.2.2 2.3 2.3.l 2.3.2 2.3.3 2.3.4 2.4 2.4.1 2.4.2 2.5. Thermal Nitridation of Silicon and Oxidized― Silicon .........● ●●●●●●●●●●●●●●●●●●●●●●●●●●● Introduction ........● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●● Exper■ mental Procedure .....● Wafer treatment .... Measurements ........● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● 。........● ●●●●●●●●●●●●●●●●●●●●●●● Exper■ mental Results ●●●●●●●●●●●●● Growth of silttcon nitride films ........● Chemical composition of silttcon nitride films .......・ Etching character■ stics of s■ licon n■ tr■ de films ..... Ox■ dation res■ stance of s■ licon n■ tr■ de films .....● ●●. 12 12 13 13. 14 15 15 16. 17 18. Discuss■ on ....... 19. ●●●●●●●● Mechanism of nitridation of siliCOn .........● Kinetics of n■ tr■ dation of ox■ dized― s■ licon .......● ●● Summary ...... References ......● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● Tables .........● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● Figures .........● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●. 19. 25 26 27. ng Shr■ nkage and Growth of Stacking Faults dur■ Thermal Nitridation .....● ●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●Q● ●●●●●●● 3.l lntroduction ......● 3.2 Exper■ mental Procedure .....● ●●●●●●●●●●●●●●●●●●●●●。●●●● ●●●●●●●●●●●●●●●●●●●●●●● 3.3 Exper■ mental Results .........● ●●●●●●●●●●●●●●●●●●●●●●●●●● 3.3.l Shrinkage of OSFis ........● ●●●●●●●●●●●●●●●●●● 3.3.2 Growth of stacking faults .........● ●● 3.3.3 Effects of oxide thickness on NSF growth .........●. 34 34 35 37 37 38 39. 23 24. 3. o●. ―. X■. ―.

(12) Page 3.3.4 3.3.5 3。 4 3.4.l 3.4.2 3.4.3. Effect of ammonia partial pressure on oSF behavior ... 39 Auger analys■ s of n■ tr■ ded films ......。 .● ●●●●0● ●●●●●● Discussion `.....● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● shrinkage of OSFls ........● ●●●●●●●●●●●●●●●●●●●●●●●●●● Growth of stacking faults .........● ●●●●●●●●●●●●●●●●●0 Effect of oxide thickness of NSF growth .........● ●●●●. 3.4.4 3.5. ●●●●●●●●●●●● Effect of ammon■ a partial pressure ......● Summary ........● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● References ......● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● Figures .........● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●. 40 41 42 43 46 47 47 49 50. INDUCED STACKttNG FAULTS. [PART TWO] CONTROL OF OXIDATION― 4. suppression of Stacking Fault Generation through Laser― Damage Getter■ ng .......● ●●●●●●●●●●●●●●●●●●●●●●● 4.l lntroduction ......● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● 4.2 Experimental Procedure .....● ●●●●●●●●●●●●●●●●●●●●●●●●● 4.3 Results and Discuss■ on ......● ●●●●●●●●●●●●●●●●●●●●●●●● 4.3.l Lattice defects caused by laser irradiation ........● ● 4.3.2 suppression of oxidation― induced precipitate. 56 56 57 59 59. formation by laser damage gettering .........● ●●●●●●●● 61 4.3.3 Suppression of OSF formation through laser― damage gettering ......● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● 63 4.3.4 1mprovement of generation lifetime through laser― damage getter■ ng .........●. 4.4. Summary ........● References ......● Figures .........●. ●●●●●●●●●●●●●●●●●●●●●. ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●. Vic■ nal Surfaces ... 64 64 66 68. 77. 5. Ann■ hilation of Stacking Faults. 5.l. 1ntroduction ........●. 5.2 5.3 5.4 5.4.l 5.4.2. Experimental Procedure .....● ●●●●●●●●●●●●●●●●●●●●●●●●● 78 Exper■ mental Results .........● ●●●●●●●●●●●●●●●●●●●●●●● 78 Discuss■ on ......● ●●●●′●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● 79 Annihilation of OSFls on the (l11)and (lll)planes .. 79 Annihilation of OSF]s on the (l11)and (l11)planes .. 81. ■n. ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●. ―. X■. ■. ―. 77.

(13) Page 5.5. 6 6.l 6.2 6.2.l 6.2.2 6.2.3 6.3 6.3.l 6.3.2 6.3.3 6.4 6.4.l 6.4.2 6.5. 7. 7.l 7.2. Summary ........● References ......● Figures .........●. ●●●●●●●●●●:● ●●.....● ●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●0● ●●●●●●. 83 84 85. Removal of Stacking Faults by Laser Melting and Recrystallization .........● ●●●●●●●●●●●●●●●●●●●●●●●●●● 90 lntroduction ......● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● 90 Exper■ mental Procedure .....● ●●●●●●●●●●●●●●●●●●●●●●●●● 91 Measurement of melted depth .........● ●●●●●●●●●●●●●●●● 91 92 Generation and removal of stacking fault nuclei ...... Removal of stacking faults ........● ●●●●●●●●●●●●●●●●●● 92 Exper■ mental Results .........● ●●●●●●●●●●●●●●●●●●●●●●● 93 Depth of melted region .......● ●●●●●●●●●●●●●●●●●●●●●●● 93 93 Generation and removal of stacking fault nuclei ...... Removal of stacking faults ........● ●●●●●●●●●●●●●●●●●● 94 Discussion ......● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● 95 Depth Of melted region .......● ●●●●●●●●●●●●●●●●●●●●●●● 95 Mechan■ sm of laser― enhanced generation of stacking faults .........● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● 96 Summary ........● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● 98 References ......● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● 99 Figures .........● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● 101 Conclusions .........● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● 106 Kinetics of Silicon lnterstitials and Stacking Faults during Thermal Nitridation .....● ●●●●●●●●●●●●●● 106 ●●● 107 Control of Oxidation― Induced Stacking Faults .....●. ―. X■. ■■. ―.

(14) CHAPTER l l.l. GENERAL INTRODUCTttON. Backgrownd. ln the past decade′ crystallographic defects in single― crystal s■ licon have rece■ ved much attention because of the importance of the defectsl 3)to the electronics industry and the 5) physical signttficance of the defects to sOlid state physics.4′ crystal Recent developments in the growth of perfect single― scale integrated (VLSI) silicon enable us to fabricate very large― circuits.6′ 7) Also′ recent developments in methods of cOntrol― ling the ■ntroduction of well― defined crystallographic defects ― ■nto perfect s■ ngle― crystal s■ licon enable us to study the phys■ cal and electrical properties of a specific defect.8-ll) In order to fabricate VLSI circuits in polished silicon treatments′ wafers′ it is necessary to carry out several heat― such as thermal oxidation of silicon′ diffusion of dopant impurity into silicon′ epitaxial growth of thin film silicon on the sub― strate silicon′ or chemical vapor deposition of the insulating film. These thermal processes can generate crystallographic for defects ■n s■ licon. Such crystallographic defects are′ example′ oxidation― induced stacking faults (hereafter called OSFOs)′ 12-23)diffusion― induced dislocations24-28)and epitaxial induced defects OSFOs stacking faults.29-32) Among these process― vely′ because of the■ r ■mportance have been studied most extens■ 21)and because of their in the study of point defects in silicon20′ marked effect on the electr■ cal character■ stics of s■ licon devices.8′ 33-35) 。xidation― induced stacking faults in silicon are known to be extr■ ns■ c ■n nature′ that is′ they are composed of extra planes of atoms bounded by Frank partial dislocations of. a Burgers vector of the type a/3′ {lll}.36). 。sFls. are usually. generated near the surface region of silicon wafers so they have 8′ 33-35) They a marked influence on the deviceOs characteristics。 act as s■ tes of the carr■ er generation― recombination which causes the excess leakage current in p― n junclionS′ 8′ 37-40)the lowering. of junction breakdown voltage′. 8′. 34)and the reduction of carrier. lifetimes in silicon.34′ 41) '。 xidation― induced stacking faults degrade the electr■ cal character■ stics of s■ lttcon dev■ ces and decrease the production ytteld of VLSI circuttts. The degradation. - 1 -.

(15) of the devicels characteristics is especially noticeable in photosens■ tive mosa■ c dev■ ces such as cCD imagers′ where the OSFls produce generation sites which are responsible for bright 43) For these reasons′ spots in video displays.42′ much effort has been spent On ways to suppress the oSF generation in silicon. Ox■ dation― ■nduced. stacking faults are generated_■ n the near― surface region of wafers during the thermal oxidation of silicon at elevated temperatures (typically 900 to 1250° C)in an oxidiz― ing ambient.12-23) Fault generation may occur in a mechanically 44-46)。 r in a damaged area caused by an abrasion or scratch′ 16′ grown― in swirl defect′ 32′ 45)。 r in an impurity precipitate。 47-49) The OsFls grow in oxidizing ambients and shrink in non― oxidizing ambients.50-53) The rate of growth of the OSFls. depends on the rate of abSOrption and the rate of shr■ nkage on the rate of emission of silicOn self― interstitials by the Frank 54-57) Based・ partial dislocation surrounding the faults.19-21′ these findings′ a number of methods for suppressing OSF genera― tion have been developed Over the past decadec Chemical etching 44)and thermal annealing of of the mechanically damaged areas15′ the as― grown wafers ■n non― ox■ diz■ ng ambients can prevent OSF nucleation。 58′ 59) Gettering is most commonly used for suppress― ■ng OSF generation. Three getter■ ng,mechan■ sms′ on which three different gettering techniques are based′ have been identified:. 。n. (1)physical or Cottrel gettering′ (2)solubility enhancement gettering′ and (3)point defect― ion pair chemical gettering. Among these three gettering mechanisms′ physical gettering is the most dominant.61) Physical gettering technique uses crystal imperfectiOns that fOrm in electrically inactive regions of wafers and act as s■ nks fOr unwanted impur■ ties. The backs■ de―. damage gettering technique uses dislocations and/or Stacking faults on the backsurface to getter s■ licon self― ■nterstitials and impurities62-64)and the diffusion gettering technique utilizes misfit dislocatiOns formed by dopant diffusion to getter. impurities.65-67) Other majOr physical gettering techniques are the chlorine oxidation gettering technique′ 37′ 68-70)the i6n― 71′ 72)the intrinsic ilmplantation damage gettering technique′ gettering technique using sio2 precipitates′ 73-75)and the film depositiOn gettering technique using nitride or poly― silicon. - 2 -.

(16) film.40′ 65) At the present time′. however′ a method for elimi― nating OsFOs completely has not been developed. At present′ silicOn diOxide film grown by thermal oxidation of s■ licon ■s the mater■ al most often used for gate ■nsulators metal― insulator― semiconductor field effect transistors (MISFETs) and for capacitors in VLSI circuits. The recent development of submicrometer MIsFETs for VLSI circuits has underlined the need. ■n. for very thin ■nsulators less than 200 A thick in the near future.76) The use of very thin silicon dioxide film as gate insulators has many disadvantages′ such as low resistivity to ■lmpur■ ty diffus■ On′ a strong tendency to react w■ th electrodes and a high interface trap density. A goOd candidate to take the place of silicon dioxide film is silicon nitride film prepared by thermal nitridati9n of sili― con and ox■ dized― s■ licon. Thermally grown s■ licon n■ tr■ de offers many advantages over s■ licOn diox■ de ■n films less than 200 Å thick′ such as a higher barrier against impurity diffusion′. higher chemical stability and a higher dielectric constant.77-79) Unfortunately′ at this timer the growth mechanism of silicon nitride films and the physical and chemical properties of the films is not well underst00d. Researchers are still investigat― ing how tO produce thermal silicon nitride films. The crysta1lo― graphic defects generated during thermal nitridation of silicon and oxidized― silicon have not been investigated so far. Very recently′ it was found by the author and his co― wOrkers that OSFls in silicOn shrink during thermal nitridation of silicon′ more importantly′ that nitridation of oxidized silicon with 80)aS silicon dioxide film generates stacking faults in silicon′ ■s descr■ bed in chapter 2. subsequently′ Mizuo confirmed our results On the behav■ Or of stacking faults dur■ ng thermal n■ tr■ ― dation of silicon and oxidized― silicon.81) These nitridation― induced stacking faults (hereafter called NSFls)′ as well as OSF's′ may have adverse effects on the characteristics and the performance Of vLSI circuits. Because knowledge of the kinetics and′. of s■ licOn― ■nterstitials and vacanc■. es dur■ ng n■ tr■ dation. ■s. very. important in the study of poin′ t defects in silicon′ it is necessary tO clarify the fOrmatiOn mechanism′ thermal behavior′ and physical nature of the osF:s.. - 3 -.

(17) l.2. Scope of This Thesis. The main purpose of this thesis is to study the shrinkage and growth of stacking faults during thermal nitridationoof fy s■ licon and ox■ dized― s■ licon ■n an ammon■ a ambient and to clar■ their mechanism based on our knowledge of the kinetics of silicon self― interstitials during thermal nitridation. To achieve this purpose′ the thermal nitridation of silicon and oxidized― silicon using an ammonia gas was investigated. A secondary purpOse of this thesis is to study the methods of controlling osFls during thermal oxidation of silicon using: (1)the laser― damage gettering techniquer(2)the annihilation technique on vicinal surfaces of (001)Orientation′ and (3)the laser melting and recrystallization technique. The thesis is divided into two parts. The first part presents the kinetics of s■ licon self― ■nterstitials and stacking faults dur■ ng thermal n■ tr■ dation of s■ licon and ox■ dized― silicon. The second part covers the methods of controlling OSFOs dur■ ng the thermal ox■ dation of s■ licon. Chapter 2 presents a study of the n■ tr■ dation of s■ licon and oxidized― silicon under various ammonia pressures from 10 3 t。 5. kg/cm2 using a newly developed high― pressure nitridation appara― tus. The growth kinetics′ etching rate and chemical composition′ oxidation resistance of the nitride films were investigated by ellipsometry and Auger electron spectroscopy. For films formed by nitridation of silicon′ it was found that the growth kinetics and properties were independent of the ammonia partial pressure. The nitridation of silicon is explained using the logarithmic rate law by a modified Ritch― Hunt theory′ which assumes that a very slow surface reaction at the ammonia― nitride interface is the rate― determining factor. According to this modified theory′ the nitridation of silicon proceeds mainly by the cation migra― tion of silicon interstitials from the silicon― nitride iritl(〕 rface to the nitride surface. on the other hand′ it is shown that the nitridation of Oxidized― silicon depends strongly on the ammonia partial pressure. This dependence is caused by diffusion of ammonia or its derivatives through the oxide. The reaction of ammonia with silicon dioxide decomposes the silicon dioxider then converts it into silicon oxynitride. During the course of. - 4 -.

(18) s■. licon diox■ de decompos■ tion′ free s■ licon atoms are generated′. then injected into silttcon as excess sttlicon self― interstitials. These results proved to be important in our research on the growth and shr■ nkage of stacking faults dur■ ng the thermal n■ tr■ ― dation descr■ bed in Chapter 3. Chapter 3 descr■ bes our study of the shr■ nkage and growth of stacking faults dur■ ng thermal n■ tr■ dation of s■ licon w■ thout oxide film and Of Oxidized― silicon with oxide films under ammonia. partial pressures of 10 3 t。. 4 kg/cm2.. It is shown that OSF:s in. licon w■ thout ox■ de film shrank linearly w■ th n■ tr■ dation time and that stacking faults in oxidized silttcon with oxide film grew during nitridation under ammonia partial pressures above 10 1 s■. kg/cm2.. we propose a model in which the shrinkage caused by. undersaturation of the s■ licon self― ■nterstitials near the silicon surface is caused by silicon― cation migration from the silicon― nitrttde interface to the nitride surface. 工n this model′ OSF growth is caused by the supersaturation of s■ licon self― ■nterstitials which Occurs near the s■ licon surface due to the generation of free silicon atoms by the reaction of ammonia with. silicon dioxide fol10wed by the injection of silicon atoms as ■nterstitials n■ tr■ de. ■nto. the bulk of s■ licon through the s■. licon―. ■nterface.. In Chapter 4′ a new technique of suppressing OSF generation during the thermal oxidation of silicon called laser― damage gettering is presented. Lattice defects induced by laser ■rradiation′. the■ r thermal stability and the■. r function as s■ nks It is shown that thermally. for point defects were studied. stable dislocations and pseudo―. sw■ rl. defects were generated in. the damaged region by high― power laser pulses above 15 J/cm2. It is also shown that these laser― induced defects act as getter― ■ng s■ tes for po■ nt defects ■n the undamaged area and prevent the formation of precipitates in this area during the subsequent ox■ dation of s■ licon. Laser― damage getter■ ng′ which utilizes laser― ■nduced defects on the backs■ de of s■ licon wafers as gettering sites for point defects′ is applied to suppress the OSF generation on the front s■ de′ Chapter 5 deals with the annihilation of OSFOs during ther― mal oxidation of silicOn using the vicinal surfaces of (001) .. - 5 -.

(19) orientation. Two modes of fault annihilation were analized by X― ray diffraction topography and by using a Wright etching tech― nique. In the first mode′ two OSF's interact independently of the surface orientation. In the second moder an OSF interacts ledge― kink (TLK)struCture on a with a linear step in the terrace― surface slightly misoriented from the (001)Orientation. ThiS entation dependence. ■nteraction shows a marked surface or■ of these modes are caused by an unfaulting reaction by which. BOth. Frank partial dislocatiOns surrounding the stacking faults are converted to perfect dislocations. A model of the annealing out of the perfect dislocations formed by the unfaulting reaction is hilation ■s proposed. A statistical treatment of fault ann■ presented to explain the OSF densities which depend on the proba― bility of the occurrence of the unfaulting reaction. In Chapter 6′ laser melting and recrystallization of the near― surface regions of silicon is shown to be effective in removing OSF]s. A new technique of measuring the depth of the melting caused by laser irradiation was developed by observing the presence and absence of ox■ dation― ■nduced stacking faults and oxygen thermal donors. It is shown that the removal of the stacking faults occurs when the irradiated region melts at least to the depth of the stacking faults. Chapter 7 summar■ zes the exper■ mental results and conclu― sions obtained in the course of this study.. - 6 -.

(20) REFERENCES. ″. ‖ Proceedings of Third Microelectronics l) C.W.Pearce: ‖ Measurement Technology Seminar′ p.V-27 (1981). 2) J.R.Monkowski: Solid State Technology′ _24 44 (1981). 3) R.B.Swaroop: Solid State Technology′ 26. lll (1983). 4) U.Goesele and H.Strunck: Appl.Phys。 20 265 (1979). 5) B.J.Masters and E.F.Gorey: Bull.Am.Phys.SoC. _22 328 (1977). 6) K.E.Benson′ w.Lin and E.P.Martin: Semiconductor Silicon 1981 Princeton′ 1981)p.33. (The Electrochemical Society′ 7) T.Suzuki′ N.Isawar Y.Okubo and R.Hoshi: SemiconductOr Princeton′ 1981) Silicon 1981 (The Electrochemical Society′ p.90. 120 8) K.V.RaVi′ C.」 .Varker and C.E.Volk: J.Electrochem.Soc.′ 533 (1973).. 9) S.Kawado′ Y.Hayafuji and T.Ad achi: Extended Abstracts′ Electrochemical Society′ Vol.75-2′ 1975′ Abstract No.184 (Electrochemical Society′ Princeton′ 1975). .Appl.Phys. l生 407 10) S.Kawado′ Y.Hayafuji and T.Ad achi: Jpn.」 (1975). 」 .R.Patel: Appl.Phys.Le tt.′ 11) L.C.Kimerling′ 型 H.J.Leamy and 12) 13) 14) 15) 16) 17) 18) 19) 20) 21) 22) 23) 24). 217 (1977). D.J.D.Thomas: Phys. Status M.J.OueisSer and P.G.G.Van (1964). G.R.Booker and R.Stickler: W.A.Fisher and J.A.Am ick:. Solidi′ 3 2261 (1963). Loon: J.Appl.Phys. 旦 5 3066 Ph ilos.Mag. 11 1303 (1965). l13 1054 J.Electrochem.Soc.′. (1966). M.L.Joshi: Acta Metall. 14 1157 (1966). I.R.Sanders and P.S.Dobson: Ph ilos.Mag. 22 881 (1969). S.Prussin: J.Appl.Phys. 43 2850 (1972). S.M.Hu: J.Appl.Phys. 15 1567 (1974). S.P.Murarka: Phys.Rev. 唖 2849 (1977). S.P.Murarka: Phys.Rev. 哩 692 (1980). T.Y.Tan and U.Gasele: Appl.Phys.Le tt. 10 616 (1982). K.Taniguchi′ D.A.Antoniadis and Y.Matsushita: Appl.Phys. Lett. 42 961 (1983). E.Levine′ J.Washburn and G.Thomas: J.Appl.Phys. 31 81 (1967).. - 7. ….

(21) 25) S.Prussin: J.Appl.Phys. 旦2 1876 (1961). 26) W.czaja: 」.Appl.Phys. 37 3441 (1966). 27) M.L.Joshi and D.Da sh: IBM J.Res.De v. 11 271 (1967). 28) 」.c.C.Tsai: Proc.IEEE′ 57 1499 (1969). 29) G.H.Schwuttke: J.Appl.Phys. 33 1538 (1962).. 30) s.Mendelson: J.Appl.Phys. 31) 32) 33) 34) 35). 旦5. 1570 (1964).. A.G.Cullis and G.R.Booker: J.Crystal Growth′ .9 132 (1971). D.I.Pomerantz: 」 .Electrochem.Soco 119 255 (1972). K.V.Ravi and c.J.varker: Appl.Phys.Le tt. _25 69 (1974). C.J.Varker and K.V.Ravi: J.Appl.Phys. 4ュ 272 (1974). A.H.Plantinga: IEEE Trans.Electron Devices′ ed-16 394. (1969). 36) JoM.Silcock and W.J.Tunstall: Philos.Mag. 10_ 361 (1964). 37) H.shiraki: Jpn.」 .Appl.Phys. 15 1 (1976). 38) L.E.Katz and L.C.Kimerling: J.Electrochem.Soc. 125 1680 39) 40) 41). 42) 43) 44). (1978). G.A.Ro zgOnyi′ P.M.Petroff and M.H.Read: 」 .Electrochem.Soc. 122 1725 (1975). P.M.Petroff′ G.A.RozgOnyi and T.T.sheng: J.Electrochem.Soc. 123 565 (1976). Y.Ichida′ T.Yanada and s.Kawado: Lifetime Factors in Silicon′ American soc.fOr Testing and Materials′ p.107′ 1980. K.Tanikawa′ Y.Ito and H.sei: Appl.Phys.Le tt. 28 285 (1976). Y.Hokari and H.shiraki: Jpn.J.Appl.Phys. _16 585 (1977). J.E.Lawrence: J.Appl.Phys. 40 360 (1969).. 45) K.V.Ravi and c.J.varker: J.Appl.Phys. 42 263 (1974). 46) C.M.shevlin and L.J.Demer: Philos.Mag. A40 685 (1979). ‐ 47) J.Matsui and T.Kawamura: Jpn.J.Appl.Phys. l1 197 (1972). 48) D.M.Maher′ 」 .R.Pa tel: J.ネ ppl.Phys. 17 3813 A.Standinger and (1976). 49) T.Y.Tan and w.K.Tice: Philos.Mag. 30 615 (1976). 50) H.shiraki: Jpn.J.Appl.Phys. 15 83 (1976). 51) H.Hashimoto′ H.Shibayama′ H.Masaki′ and H.Ishikawa: J.Electrochem.soc. 123 1899 (1976). 52) Y.Sugita′ H.shimizu′ A.Yoshinaka and T.Aoshima: J.Vac.Sci. Technol. 14 44 (1970). 53) c.L.Claeys′ G.J.Declerck and R.J.Van Overstraeten: ヽ Appl.Phys.Le tt. 35 797 (1979).. - 8 -.

(22) 54) A.M.Lin′ R.W.Dutton′ D.A.Antoniadis and W.A.Tiller: J.Electrochem.soc. 128 1121 (1981). 55) R.B.Fair: J.Electrochem.soc。 128 1360 (1981). 56) D.A.Antoniadis: J.Electrochem.Soc. 129 1093 (1982).. 57) B.Le roy: J.Appl.Phys.. 三3. 4779 (1982).. 58) S.P.Murarka′ H.J.LeVinstein′ R.B.Marcus and R.S.Wagner: J.Appl.Phys. 48 4001 (1977). 59) S.Kishino′ s.Isomae′ M.Tamura and MoMaki: Appl.Phys.Le tt. 塁 1 (1978). 60) J.E.Lawrence: Trans.AIME 242 484 (1968). ― Microstructure 61) J.E.Lawrence and H.R.Huff: VLStt Electronics Science′ (N.G.Einspruch′ N.Y. 1982.. ed)Vol.5′. p.51′ Academic Press′. 62) s.Prussin′ s.P.Li and R.H.Cockrum: Extended Abstracts′ Electrochemical society′ Vol.78-1′ 1978′ Abstract No.261 (Electrochemical Society′ Princeton′ 1978). 63) Y.Hayafuj i′ T.Yanada and YoAoki: Extended Abstract′ Electrochemical Society′ Vol.79-2′ 1979′ Abstract No.485 (Electrochemical society′ Princeton′ 1979). 64) L.E.Katz′ C.w.Pearce and P.F.Schmidt: Extended Abstract′ Electrochemical Society′ Vol.79-2′ 1979′ Abstract No.486 (Electrochemical Society′ Princeton′ 1979). 65) G.A.RozgOnyi and R.A.KuShner: J.Electrochem.Soc. 123 570 66) 67) 68) 69) 70). 71) 72) 73) 74). (1976). G.A.RozgOnyi′ P.M.Petroff and M.H.Re ad: J.Electrochem.Soc. 122 1725 (1975). A.Armigliato′ M.se rvidor′ s.solmi and I.Vecchi3 J.Appl.Phys. 48 1806 (1977). T.Hattori: J.Electrochem.Soc. 123 945 (1976). H.shiraki: Jpn.」 .Appl.Phys. 15 83 (1976). C.L.Claeys′ E.E.Laes′ G.J.Declerck and R.J.Van Overstraeten: Semiconductor Silicon 1977 (The Electrochemical Society′ Princeton 1977)p.773. T.E.Se idel′ R.L.Meek and A.G.Cullis: J.Appl.Phys. ■ 6 600 ′ (1975). H.」 .Geipel and w.K.Tice: IBM J.Res.Develop. .24.310 (1980). W.K.Tice and T.Y.Tan: Appl.Phys.Le tt. 28 564 (1976). T.Y.Tan′ E.E.Gardner and w.K.Tice: Appl.Phys.Le tt. ■ 0 175 (1977).. …. 9 -.

(23) 75) L.E.Katz and B.w.Hill: J.Electrochem.soc. 125 1151 (1978). 76) s.K.La i: Extended Abstracts′ Electrochemical Society′ Vol.81-1′ 1981′ Abstract No.214 (Electrochemical society′ Princeton′ 1981). 77) T.Ito′ s.Hijiyar T.No zaki′ H.Arakawa′ M.shinoda and Y.Fukukawa: J.Electrochem.Soc. 125 448 (1978). 78) s.P.Murarka′ c.c.chang ムnd A.CoAd ams: J.Electrochem.Soc. 126 996 (1979).. 79) Y.Hayafuji and K.Kajiwara: J.Electrochem.Soc. 129 2102 (1982). 80) Y.Hayafuj i′. K.Kajiwara and s.usui: J.Appl.Phys.. 三3. (1982). 81) s.Mizuo′ T.Kusaka′ A.Shintani′ M.Nanba and H.Higuchi: J.Appl.Phys. 54 3860 (1983).. - 10 -. 8639.

(24) [PART ONE]. KINETICS OF SILICON ttNTERSTITIALS AND STACKING FAULTS DURING THERMAL NITRIDAT工. ON.

(25) CHAPTER 2 2。 1. THERMAL NITRIDATION OF SILICON AND OXIDIZED―. SILICON. Introduction. It is well known that the behaviOr of stacking faults depends On the presence Of po■ nt defects ■n s■ licon′ such as silicon self― interstitials and vacancies. Therefore′ before studying stacking faults in silicOn during thermal nitridatiOn Of silicOn and Oxidized― silicOn′ it is necessary to study thermal nitridation and tO clarify its mechanism with regard to the point defects. Presently′ silicon diOxide film is mOst commonly used as a gate insulator. s00n′ hOwever′ thin amorphous dielectric films ess than 200 A thick will be required fOr silicon integrated Circuit techno10gy for use as gate insulatOrs in metal― insulator― semicOnductOr field― effect― transistOrs (MISFETs)and as capaci― tors in vLSI circuitso There may be better materials than silicOn diOxide fOr use in vLSI circuits′ because of silicon dioxidels 10w resistivity to the diffusion of impurities (aS′ for example′ sOdium iOns)′ its strOng tendency to react with elec― l″. trodes′ and the uncertainty of the degree of its stability in. thin films. we can expect that much work will be done to deter― m■ ne ■f s■ licon diox■ de can be used as gate ox■ des below 200 Perhaps the best candidate to take the place of silicOn dioxide is silicOn nitride prepared by direct nitridatiOn of. Å.. licon. This mater■ a1 0ffers many advantages over s■ licon diox■ de ■n films less than 200 Å thick′ such as higher barr■ er aga■ nst diffus■ on Of impur■ ties′ less tendency to react w■ th electrodes′ higher density and a larger dielectric constant. s■. In the 19600s′ several unsuccessful attempts were made to grow uniform amorphOus silicOn nitride.l-3) subsequently′ Raider et al.4)and Kooi et al.5)showed that a dielectric layer com― posed Of e■ ther s■ licOn Oxyn■ tr■ de or s■ licon n■ tr■ de m■ xed w■ th Oxide is fOrmed near the si‥ sio2 interface during heat― treatment Of oxidized― silicOn in nitrogen and ammonia gas. ItO et al.6) found that a very thin unifOrm amorphOus silicOn nitride film less than 100 A thick is generated by direct thermal nitridation of silicOn in a nitrOgen ambient with Oxidant impurities less than l ppm. ItO et al.7′ 8)and Murarka et al。 9)reported that. - 12 -.

(26) the use of ammon■ a gas or ammon■ a plasma ■nstead of n■ trogen can reduce the temperature of nitridat主 ono More recently′ Ito et. al.10)have shown that the conversion of silicon dioxide to sili― con oxynitride near the surface of the silicon dioxide film can be accomplttshed by thermal nitridation of oxidized― silicon in ammon■. a.. In this paper′ the ammonia partial pressure dependence on thermal n■ tr■ dation of s■ licon and ox■ dized― s■ licon ■s studied to clarify the mechanism o「 f the nitridation of silicon and oxidized― silicon. The chemical composition′ etching rate and oxidation resistance of the resultant films were investigated. 2.2 2.2.l. Exper■ mental Procedure. Waf.er treatment. Silicon wafers with (001)orientation were prepared from a p― type dislocaton― free Czochralski― grown crystal. The doping impurity was boron and the reslstivity range was 10 to 15 Ω―cm. The front surface of the wafers was mechano― chemically polished and the back surface was chemically etched. The wafers were about 400 μm thick and 75 mm in diameter. Some of the wafers were oxidized at 9500C tO grow a silicon dioxide film about 100 A thick. Before nitridation′ silicon had a naturally― grown oxide of 9 A and oxidized― silicon had a 102 A silicon dioxide. Nitri― dation of silicon and oxidized― silicon was carried out in a high pressure furnace at temperatures between 900 and l150° C under ammonia partial pressures of 10 3 t。 5 kg/cm2 by using high― purity ammon■ a and ammon■ a― n■ trogen m■ xtures w■ th ox■ dant impur■ ties less than l ppm. Figure 2.l shows the high pressure nitridation system which consists mainly of a stainless steel pressure vessel′ a quartz tube′ and heaters and controllers of pressure′ temperature and gas flow. A micro― processor is used to operate the systemo To ma■ nta■ n established industiral standards of MOS cleanliness′ the total pressure in the quartz tube is kept higher. than that in the pressure vessel by O.l to O.5 kg/cm2. system can be operated at a maximum temperature of 1200°. Th ios C and a. maximum pressure of 25 kg/cm2 and has a flat zone of at least 760 mm with a temperature variance of up to 200 4-inch wafers per load.. - 13 -. ±1°. C.. It can accommodate.

(27) 2.2。 2. Measurements. The film thickness was measured by ellipsometry. The ellipsometer system`used is the Applied Materials AME-500 with a mercury light sOurce (λ =5461 Å ). This system orients the inci… dent and reflected light beams at a fixed angle of 70° with respect tO the normal direction of the sample. To calculate the fillm thickness from the ellipsometric measurements′ we used 4.051 - 0.028i as the refractive index of the silicon substrate. We assumed the index Of the films to be l.45′ which is that of thermal oxide′ because of the difficulty of determining the. refractive index of very thin films less than 50. Å thick by ellipsOmetry and because of the inhomogeneity of the chemical compositiOn of the films. The chemical composition of the films was determined using Auger electron spectroscopy. we used a JEOL JAMP-10 Auger spectrometer w■ th a cylindr■ cal m■ rror analyzer. In measur■ ng the depth profiles′ we used a primary electron beam with a cur― rent of l.6 μA at 5 keV and an argon ion gun of 15 mA at l kev. For quantitative Auger analys■ s′ we used the der■ vative spectra Of the KLL transitions fOr nitrogen′ oxygen and silicon of approximately 382′ 508′ and 1618 eV′ respectively′ and Auger sen― sitivity factors determined as follows. silicon oxynitride films about 3000 Å thick were deposited on silicon substrates using the CVD method under var■ Ous conditions. The compos■ tion of these films was determined by electron probe microanalyzer (EPMA)′ after which the samples were analyzed by Auger electron spectro― scopy. compar■ ng the EPMA measurement data w■ th the data of the Auger analys■ s′ the sens■ tiv■ ty factors for n■ trogen′ oxygen and silicon were determined to be l.20′ 2.99 and l.00.′ respectively. The nitridation process was as fo1lowso Silicon wafers were ■nserted into the central zOne of the cold quartz tube and the tube was then c10sed with an end cap. High― purity nitrogen with a flow rate of 10 1iters/min was passed through the tube to purge ox■ dant impur■ ties and the ox■ dant was then mon■ tored. As soon as the Oxygen content in the quartz tube reached less than l ppm′ a continuous flow of ammonia or ammonia― nitrogen mixture was sub― stituted for the n■ trogen flow′ after which the furnace tempera―. ture was raised.. The pressure was raised about l atm/min when. - 14 -.

(28) the system was Operated above l kg/cm2 and the temperature was raised about 80° C/min in all runs. After ammonia treatment′ the ammonia flow was changed back to nitrogen flowr after which the heaters were turned off. The wafers were then pulled out of the tube and placed in a vessel filled w■ th n■ trogen to cool. When n■ tr■ dation time was less than 2 hr′ however′ we ■nserted the wafers directly into the high temperature quartz tube through which ammonia or ammonia― nitrogen mixture flowed at a rate of. 20 1iters/min. The etching rate of the films was measured us■ ng a D― P etchant′ 4′ 11)consisting of 15 parts of HF′ 10 parts of HN03 and 3000 parts of de― ionized water at 20° C. ned by The ox■ dation res■ stance of the films was determ■ exposing the samples to dry oxygen at 1000° Ce The film thickness was then monitored as a function of exposure time. We defined oxidation resistance as the time rapid oxidation started. 2.3 2.3.l. Exper■ mental Results Growth of silicon nitride films. tr■ dation We studied the growth of films caused by direct n■ of silicon. Figure 2.2 shows the growth curves of the films grown under ammonia pressures of 10 3 and l kg/cm2. These curves show an initial rapid growth followed by a marked slowing down. until′ at the cr■ tical thickness of about 30′. 40 and 50 A for nitridation at 900′ 1000 and 1100° C′ respectively′ growth nearly stops. This behavior is only slightly influenced by the ammonia partial pressure. In Fig. 2.3′ the dependence of the film thick― ness on the ammonia partial pressure is given for growth tempera― tures of 900° ′ 1000° ′ and l100° C for 5 hr′ and was found to be very small′ that is′ the film thickness increased only by 20t when the ammon■ a partial pressure was chanded from lo 3 t。 5 kg/cm2. The log of the thickness plotted aga■ nst the ■nverse of the temperature showed nearly straight lines and the plot obeyed the Arrhen■ us equation for n■ tr■ dation under ammon■ a partial. pressures of l and 10 3 kg/cm2 for 5 hr.. We found an activation. energy of about O.35 eV under both conditions. This value somewhat larger than the O.23 eV obtained by Murarka et al.9). - 15 -. ■s.

(29) Nitridation of oxidized― silicon resulted in a small increase ■n the thickness of the films′ the amount of whiCh was dependent a partial on n■ tr■ dation timer n■ tr■ dation temperature and ammon■ pressure. The oxide 102 Å thick grew to l12′ l14′ l18′ and 122 A during nitridation at 11000C for O.5′ l′ 2′ and 5 hr′ respec― tively′ and 106′ l13′ and 122 Å for 5 hr at 900° ′1000° ′and 1100° Cr respectively′ in a pure ammonia ambient of l kg/cm2. The ammonia pressure dependence of the resultant film thiCkness did not show a clear tendency′ and was rather complicated. Although the relation between ammonia partial pressure and film thickness ■s not clear from these results′ we can say that the thickness of the film increased from 102 Å to about 110 Å in ammonia partial pressures below 10 2 kg/cm2 and to abOut 120 Å in. partial pressures above 10 l kg/cm2 during nitridation at 1100° C for 5 hr. 2.3.2. Chemical composition of silicon nitride films. The chemical composition of films formed by nitridation was examined by Auger electron spectroscopy. The AES analysis revealed the presence of carbon′ n■ trogen′ oxygen and s■ licon at the surface of the nitrided samples. The carbon peaks′ however′ disappeared from all AES spectra after ■on sputter■ ng for several m■. .. nutes′ so these peaks must have been caused by adsorptive. carbon atoms at the sample surfaceo Figure 2.4 shows typical AES spectra observed in the interior of the nitrided samples. Specta silicon′ respectively′ (a)and (b)are for silicon and oxidized― after nitridation at 1100° C for 5 hr under a pure ammonia pres―. sure of l kg/Cm2.. since the films contained only nitrogen′. oxygen and silicon′ we can designate them as silicon oxynitride filmso The ratio of the peak― to― peak height of nitrogen to the peak― to― peak height of oxygen is larger in the nitrided silicon sample than in the nitrided oxidized― silicon sample. It iS i]mportant to note that oxygen is present in the nitrided silicon sample and that n■ trogen ■s present in the n■ tr■ ded ox■ dized― silicon sample. The presence of oxygen in the nitrided silicon sample can be attr■ buted to trace ox■ dants ■n the ambient and the presence of n■ trogen ■n the n■ tr■ ded ox■ dized― s■ licon sample can be attr■ buted to diffus■ on of ammon■ a or ■ts der■ vatives ■nto the oxidized― silicon. Figure 2.5 shows typical chemical depth - 16 -.

(30) profiles in the nitrided silicon used in Fig. 2.4(a). Here a 48 A thick film ■s sputtered through′ and the peak― to― peak heights of the siLVV′ NKLL′ OKLL and SiKLL lines are plotted as a function of the sputtering time? These profiles show that the ■s. s■ licon oxyn■ tr■ de. Us■ ng the sens■ tiv■ ty factors determined experimentally′ the stoichiometry_is found to be SiNl.700.3 at the middle of the film′ and it appears to be more oxygen― rich in the near― surface region and to be more nitrogen― rich in the deeper region. Although we were unable to detect any significant dependence on ammOnia partial pressure of either the stoichiometry or the chemical depth profiles of films grown by nitridation of silicon′ ammonia partial pressure was found to have a significant effect on the stoichiometry and the chemical depth profiles of the films formed by nitridation of oxidized― silicon. Figure 2.6 shows the chemical depth prOfiles of a film about 100 A thick formed under fil]m. ■nhomogeneous. an ammonia pressure of l kg/cm2 at 1100° C for 5 hr.. We can see. that the film is silicon oxynitride and that the nitrogen content in the film is high and fairly uniform. If we express nitrided oxidized― silicOn film as siNxoy′ the values for x and y under various ammonia partial pressures can be given as in Table 2.l′ after correction by the sensitivity,factors. A higher ammonia partial pressure results in a higher nitrogen content and a reduced oxygen content. This phenomenon may be caused by the conversion Of silicon dioxide to silicon oxynitride during nitri― dation. 2.3.3. Etching character■ stics of s■ licon n■ tr■ de films. The etchability of silicon nitride films was examined using a D― P etchant at 200C. For the films grown by direct nitridation of s■ licon′ the plots of film thickness aga■. nst etching time revealed ammon■ a partial pressure ■ndependence and showed an almost linear decrease ■n film thickness w■ th etchingetime.. The films had nearly the same etching rate of about O.8 A∠. min′. even though they were grown under ammonia partial pressures of from 10-3 t。 5 、g/cm2. This etching rate is smaller by about one order of magn■ tude than the etching rate of 6.6 八/m■ n fOr thermal ox■ de.. - 17 -.

(31) On the other hand′ the plots of film thickness against etching timer as can be seen in Fig. 2.7′ were very complicated tr■ dation of ox■ dized ■n the s■ licon oxyn■ tr■ de films formed by n■ silicon′ but the plots clearly revealed ammonia partial pressure dependence. Figure 2.7 shows that the etching rates at the near― surface layer′ at the ■nter■ or and at the near― ■nterface layer are different′ in spite of the homogeneity of the chemical compo― ■n the s■ tion of the films. We can see that the etching rate near― surface layer and the near― interface layer is smaller than ■n the the etching rate ■nter■ or′ which is nearly equal to the interface layer′ the etching rate of thermal oxide. In the near― fillns have nё arly the same etching rate as that of the film grown by nitridation of silicon. We can also see from Fig. 2.7 that silicon decreases as the the etchability of the nitrided oxidized― ammonia partial pressure increases. This can be attributed to ■s the ■ncreas■ ng n■ trogen content as ammon■ a partial pressure increased′ as described previously. It is also interesting to note that higher ammonia pressure results in a thicker near― ■nterface layer. 2.3.4. Oxidation resistance of silicon nitride films. The oxidation resistance of the film was determined by measuring the film thickness as a function of exposure time to a dry oxygen ambient at 1000° c. The curves shown in Fig. 2.8(a) and (b)are plots of our experimental data for silicon and ox■ dized― s■ licon n■ tr■ ded under var■ ous ammon■ a partial pressures at 1100° C for 5 hr. We can see in Fig. 2.8(a)that the oxidation resistance of the nitrided silicon does not vary significantly with the ammonia partial pressureo The oxidation resistance is found to be about 1000 min for each film. That the oxidation ■s con― res■ stance ■s ■ndependent of the ammon■ a partial pressure sistent with the before― mentioned results that the stoichiometry and the etching rate of the nitrided silicon do not depend on ammonia partial pressure in the growth process. On the other hand′ the ox■ dation res■ stance of the n■ tr■ ded oxidized― silicon is strongly affected by ammonia partial pres― sure. we can easily see a large difference of a factor of ten between the oxidation resistances shown in Fig. 2.8(b).. - 18 -.

(32) The log― log p10ts Of the oxidation resistance (R)in minutes against the ammonia partial pressure (P) in kg/Cm2 can be expressed as R = A PB. (2.1). Where A and B are constants. O.35. 2.4. The value of B was found to be. Discussion. ln this section′ we w■ ll discuss the kinetics of the n■ dation of silicon and oxidized― silicon and propose nitridation mechan■ sms for s■ licon and ox■ dized― s■ licon. 2.4.l. tr■ ―. Mechanism of nitridation of silicon. The standard free energy of the reaction was calculated us■ ng the equation 3Si + 4NH3. →. Si3N4 + 6H2. (2.2). between 1000 to 1600° K′ 12)as shown in Fig. 2.9. These free energy changes ■ndicate that the n■ tr■ dation of s■ licon ■n an ammonia ambient is thermodynamica1ly feasible in this temperature range and′ in fact′ we were able to verify that the reaction does occur. The growth kinetics for the oxidation of silicon can best be characterized by the linear― parabolic rate model proposed by Deal et al.13) In this mOdel′ the rate constants are very sensitive to oxidant content and increase linearly with the content. We cannot apply this model to‐ nitridation of silicon because there is an anomalously high、 nitridation rate in the early stages of nitridation fo1lowed by a very small rater and also because this behav■ or ■s ■nsens■ tive to ammon■ a partial pressure. Because similar phenomena have been frequently observed in the oxidation. of metals other than silicon′ we propose applying the ratte laws for the oxidation of metals to the nitridation of silicon. We must first determine the relationship between film thick― ness and nitridation time. The following five laws have been used to describe the oxidation of metal to form thin oxide14). - 19 -.

(33) linear. t=A+Bx. (2.3). parabolic. t〓 A tt B x2. (2.4). logarithmic. x=A+B ln t. (2.5). inverse log.. 1/x = A + B ln t. (2.6). c ub ttc. x=A+B t1/3. (2.7). where x ■s film thickness and t is n■ tr■ dation time. A and B are coeffic■ ents which are determ■ ned by the linear least― square method. combinations of two or more of these relationships in a s■ ngle ox■ dation― time curve′ such as a linear― parabolic law′ are also quite common. we exclude these combinations from our con― s■ deration′ however′ becauser even though they may fit the data well′. their three or more adjustable coefficients make it more. difficult to determine which factor is dominant. The values of A and B are given in Table 2.2 with the coefficient of determina― tion (R2)′ the mean square error (μ )′ the median absolute devia― tion (MAD)and the range′ for nitridation under an ammonia. pressure of l kg/cm2 at 1100° C.. The rate laws in Table 2.2 are. ordered so that the law giving the best fit is listed first. Plots of thickness against time in Fig. 2.10 also show that the , logarithmic rate law gives the best fttt of the five rate laws and gives a good fit for all times. The linear and parabolic rate laws give a very poor fit to the data′ while the inverse logari― thmic and cubic rate laws give fairly good fits. A theory giving the inverse logarithmic rate law has been developed for oxidation of metals by Cabrera et al.16) In CabreraOs theory an upper temperature limit was given. This C for the upper temperature limit has been claculated to be 500° oxidation of silicon. For the nitridation of silicon′ we esti― C because mated the upper temperature limit to be less than 500° of the higher density of si3N4 as compared to Si02 and the smaller electronegativity of nitrogen than that of oxygen. For this reason′ the ■nverse logar■ thm■ c rate law appears to the. present authors to be invalid for nitridation of silicon at about 1000° C. Theories giving thё cubic rate law have been reported by several authors.16′ 17) According to Ritchie et al.′ 18)the cubic oxidation rate law essentially depends on the pressure. If this. - 20 -.

(34) is truer it is likely that the cubic rate law is also invalid because of disagreement between the experttmental and theoretttcal results Of pressure dependence. For the above reasons′ .we will discuss the n■ tr■ dation kinetics of s■ licon ■n terms of the .logarithmic rate law. In attempting to understand the nitridation kinetics of s■ licon ■t is essential tO cOns■ der that the n■ tr■ dation reaction is pressure― independent. Ritchie et al.18)found that the slow surface reaction at the oxygen― nes ox■ de ■nterface which determ■ the oxidation rate can give the pressure― independent logarithmic rate law for the oxidation of metals. Their theory was based on the fol10wing four assumptions: (1)that a slow electron trans― fer to a s■ ngly― charged oxygen spec■ es absorbed on an ox■ de sur― face determines the oxidation rateF (2)that the oxttde surface is saturated with field― producing ionsF (3)that the electron concen― tration across the oxide is a Boltzman distributionF and (4)that the ox■ de can be regarded as a thin parallel plate capac■ tor. Now′ let us modttfy the Ritchie― Hunt theory so that it can be applied to the nitridation of silicon. From assumption (1)′ the rate― determ■ n■ ng reaction at the ammon■ a― n■ tr■ de ■nterface′. N 2 単q = N 3. (2.8). gives the rate equation 〓. ｘｔ一ｄｄ. K[N… 2]n. (2.9). Where K is the rate constant′. [N 2]a surface charge concentra― tion and n the electron dens■ ty at the n■ tr■ de surface. Assump― tion (2)makes [N 2]equal to the total number of surface sites N. In assumption (3)′ the electron density in the nitride has a Boltzman distribution′ so that the electron density at the nitride surface can be given by n = noexp(― qV/kT)′ ＳＳ. ＯＶ. ｏ■ ・■. where n face.. (2.10). the electron dens■ ty at the n■ tr■ de― s■ licon ■nter― the potential drop across the nitride by the surface. - 21 -.

(35) charge concentratiOn [N 2]and would be given by V = 87Tqx IN 2]/ε. (2.11). from assumption (4)′ where c is the dielectric constant of the n■ tr■ de and q the electron charge. Combin■ ng these equations gives the nitridation rate equation as. 士 =KNnoexp(― 篭か■). (2.12). This rate equation can be integrated to give a pressure― ■ndependent logar■ thm■ c rate law for n■ tr■ dation of s■ licon as. X= %. 貴 Ht. As we can assume t >> to in this work′ above equation.. we can exclude to from the. Now′ let us calculate the values of the total number of sur―. face sites N and the strength of electric field V/x. SinCe we have already determined the coefficient B in the logarithmic law to be 5.74 × 10 8 [cm]for nitridation under an ammonia pressure. of l kg/cm2 at 1100° c′ we get the following relation for εkT/87T q2N 〓中. ×. H‐ 0回. This equation gives the value of N as 2.95. (2.14) ×. 1011 [cm 2].. By × 106. using this value of N′ we can calculate V/x to be 2.06 [V/Cm]. We must cohpare these values with the values derived by oxidation of silicon because of the absence of data for nitrida― tion of silicon. several authors have found that N is less than × 1012 [cm 2]for the oxidation of silicon from or equal to 3 room temperature to 950° C.17-19) with respect to the potential difference v′ it has been repOrted that V is of the order of about a volt.19′ 20) This value for V leads to a value of about. 106 [v/cm] for v/x′ which is in agreement with the value obtained in this wOrk. with respect to temperature dependence of N and. - 22 -.

(36) V/X′. suitability of this model was also confirmed for nitridation. of silicon. Since it is possible to calculate reasonable values for N. and V/X frOm the data presented ttn logarithmic form′. it appears. that the Ritchie― Hunt theory is applicable to the nitridation of silicon. It is probable′ then′ that the nitridation of silicon proceeds mainly by cation migration under a constant electric field. nitride interface to the Cation migration from the silicon― li― n■ tr■ de surface may be the cause of the undersaturation of s■ con self― ■nterstitials near the s■ licon surface which lead to the shrinkage of OSFOs. 2.4.2. Kinetics of n■. tr■. dation of ox■ dized― s■ licon. In the si― N― O system′ it is well known that the four solid phases of Si′ Si02′ Si2N20 and Si3N4 are stable.21) Hence it. would appear that the following two reactions occur in the nitri― dation of ox■ dized― s■ licon ■n the ammon■ a ambient. 3Si02. 4NH3. Si3N4 + 6H20. (2.15). 2Si02. 2NH3. Si2N20 + 3H20. (2.16). The standard free energy changes of these reactions are found tO be positiver as can be seen in Fig. 2.9.12) AlthOugh these posi― ■ndicate that n■ tr■ dation of ox■ dized― s■ licon tive energy changes is not thermodynamically favorabler the second reaction above may in fact be observable because:(1)there is a very small′ posi―. tive energy change of about 15 Kca1/mOl even at l100° CF and (2) the higher thermodynamic stability of Si2N20 as compared to Si3N4 under the very small amount of oxidants which are the products of the reaction of ammonia with the oxide film｀ licOn It is known that n■ trogen can diffuse through s■ dioxide films at high temperature.22) In the nitridation of oxidized― silicon′ ammonia and/or aCtiVe derivatives of ammonia .. diffuse into the oxider then occupy interstitial sites uAiformly in the oxide film′ after which they slowly exchange their inter― This pro― stitial sites for substitutional sites to grow Si2N20・ silicon cess takes place faster at the surface and at the oxide―. - 23 -.

(37) interface than in the interior of the oxide because of the loca― lized electr■ c field caused by the contact potentttal at the sur― ■n these exchange face and at the ■nterface. The difference rates leads to a difference in etching rates in spite of the uniform composition of the films′ as previously mentioned. In a w■ th n■ tr■ dation of ox■ dized― s■ licon′ the reaction of ammon■ sttlicon dioxide decomposes the silicon dioxide′ then converts it into silicon oxynitride. Free silttcon atoms are generated during the course of silicon dioxide decomposition′ and are then. injected into silicon as excess silicon self―. interstitials to. feed the growth of stackttng faults. Since n■ tr■ dation of ox■ dized― s■ licon ■s caused by diffus■ on of ammon■ a or ■ts active der■ vatives through the ox■ de film′ the formation of S12N2° of ammonia. 2.5. iS Strongly dependenヒ. on the partial pressure. Summary The n■ tr■ dation of s■ licon and ox■ dized― s■ licon was studied. under various ammonia partial pressures. After the chemical com― pos■ tion′ etching rate and ox■ dation res■ stance of the resultant f主 lms were character■ zed us■ ng Auger electron spectroscopy and ellipsometry′ it was found that the nitridation reaction of sili― independent′ while the reaction of con with ammonia was pressure― oxidized… silicon with ammonia depended strongly on the ammonia partial pressure。 On the bas■ s of this ammon■ a pressure depen― of s■ licon and ox■ dized― sttlicon are proposed. The nitridation of silicon involves a slow a― n■ tr■ de ■nterface and is well surface reaction at the ammon■ expla■ ned by the logar■ thm■ c rate law presented by Ritchie et al. The nitridation of oxidttzed― silicon may be caused by the diffu― sion of ammonia or its active derivatives through the oxide. In the models for the n■ tr■ dation of s■ licon and ox■ dized― s■ licon the undersaturation of s■ licon self― ■nterstitials near the s■ li― con surface occurs during the nitridation of silicon′ and the interstitials near the silicon supersaturation of silicon self― surface occurs dur■ ng the n■ tr■ dation of ox■ dized― s■ licon.. dency′ models for the n■. tr■ dation. - 24 -.

(38) REFERENCES l) A.N.Knopp and R.Sticker′ Electrochem.Te chnol.′ 3′ 84 (1965). 2) L.V.Gregor′ Interim Report′ Air Force Avionics′ AF33(615)-5386′ 15 November 1966. 3) R.G.Freser′ J.Electrochem.Soc.′ 115′ 1092 (1968). 4) S.I.Raider′ R.A.Gdula and J.R.Petrak′ Appl.Phys.Le tt.′ 21′ 150 (1975). 5) E.Kooi′ J.G.Van Lierop and 」 .A.Appels′ J.Electrochem.Soc. ll17 (1976). T.Ito′ S.Hijiyar T.Nozaki′ H.Arakawa′ MoShinodar and Y.Fukukawa′ J.Electrochem.Soc.′ 125′ 448 (1978). T.Itor T・ Nozaki′ H.Arakawar and M.Shinoda′ Appl.Phys. Le tt.′ 32′ 330 (1978). T.Ito′ 工.Kato′ T.Nozaki′ T.Nakamura′ and H.Ishikawa′ Appl.Phys.Lett.′ 38′ 370 (1981). S.P.Murarka′ C.c.chang′ and A.C.Ad ams′ J.Electrochem.Soc.′ 126′ 996 (1979). T.Ito′ T.Nozaki′ and H.Ishikawa′ J.Electrochem.Soc。 ′ 127′ 2053 (1980). W.A.Pliskin and R.P.Gnall′ 111′ 872 J.Electrochem.Soc.′ 123′. 6) 7) 8) 9) 10) 11) 12) 13) 14) 15) 16) 17) 18) 19). (1964). JANAF Interim Thermochemical Tables′ The Dow Chemical Co.′ Midland′ Michigan. B.E.Deal and A.S.Grove′ J.Appl.Phys.′ _36′ 3770 (1965). 0.Kubaschewski and B.E.Hopkins′ Ed itors′ :00xidation of ‖ Metals and A1loys′ p.35′ Butterworths′ London (1962). P.J.Burkhardt and L.V.Gregor′ 236′ Trans.Metall.Soc.AIME′ 299 (1966). N.Cabrera and N.F.Mottr Rep.Prog.Phys.′ _12′ 163 (1949). F.Lukes′ Surface、 sci.′ 30′ 91 (1972). I.M.Ritchie and G.L.Hunt′ Surf.Sci.′ 15′ 524 (1969). Y.Kamigaki′ oyobutsuri (in Japanese)′ 41′ 872 (1977).. 20) F.P.Fehlner′ J.Electrochem.Soc.′ 21) w.R.Ryall and A.Muan′ 22) R.M.Barrer′. 119′. 1723 (1972).. Sc ience′ 165′ 1363 (1969).. J.chem.Soc.′ 136′ 378 (1934).. - 25 -.

(39) Table 2.l.. Atomttc ratios x for nitrogen and y for oxygen to silicon′ based on the formula SiNxoy′ fOr oxidized― silicon nitrided at l100° C fOr 5 hours under various ammon■ a partial pressures.. Sample No.. Ammonic Pressure (Kgノ Cm2). Nx. Si. 10-3. 0。. 5. 1。. 8. 2. 10-2. 0.7. 1。. 7. 3. 10 1. 1。. 2. 1。. 2. 1。. 5. 0。. 8. 1。. 9. 0。. 6. 4 5. Table 2.2.. Oy. 5. Values of A and B determ■ ned by the least― square method with the coefficient of determination (R2)′ the mean square error (口 )′ the median absolute deviation (MAD)′ and the rOnge for nitridation at 1100° C under an ammonia partial pressure of l kg/cm2 for the five rate laws. ThiCkness was measured in Å ′ MAD′ and the range) Residuals (口 were also calculated in Å。00 Lo‖ iS for the logarithmic′ 00 1nv.‖ for the inverse logarithmic′ ‖ cub.・・ for the cubic′ 00 Pa.‖ for the paraboliC and ‖ Li.‖ for the linear rate lawS.. and times in seco. A. Lo. INV。. CuB。. PA. LI. -9.642. 0528 26。 320 -39620 -92760 0。. R2. B 5。. 745. 00316 0.7332. -0。. 30。. 05. 2559. - 26 -. 0。. 987. 976 0。 956 0。 828 0.790. 0。. μ. MAD. RANGE. 8. 0。. 7. 1。. 8. 1.3. 1。. 0. 4。. 4. 4. 1。. 3. 4。. 1. 0。. 1。. 2.8 3。. 4. 2.6 3。. 0. 8。 1. 10.5.

(40) Quartz ttube. Pressure Vessel Couplご. Front Door. Fig. 2.l.. Wafers Gas lntet End cap MQin Heater N2,NH3 Cool:ng Jacket. Schematic drawing of the high― apparatus.. - 27 -. pressure nitridation.