3 Shr■ nkage and Growth of Stacking Faults dur■ ng
3.5 Summary
4.3.3 Suppression of OSF formation through laser― damage
ln the prё vious section′ it was shown that the laser― induced
defects caused by 1630
」/cm2 pulses on the back surface prevent
lthe formation of oxidation― induced impurity precipitates near the front surface. In this section′ it is shoWn that the laser―
■nduced defects act as continu■ ng s■nks for ■mpur■ tieS and suppress OSF formation dur■ ng the subsequent ox■ dation of the wafer.
Wafers were prepared in the same manner as the wafer ■n Fig. 4.6. After removing the oxide film′ the wafers were
re―oxidized at 1200° C for 30 min in a dry oxygen ambient. Figure 4.9 shows optical micrographs taken in the tWO areas in the front surface corresponding to the nondamaged and damaged areas on the back surface after Sirtl etching for 3 m■ n.
Many OSFls were formed in the area corresponding to the non―
damaged arear as is shown in the lower― left part of Fig. 4.9.
Most of these OSFls originated at the oxidation― induced impurity precipitates formed during the first oxidation. A detailed TEM
analysis showed that the precipitates around which a strong
strain field existed had a needle― like shape. The strain field
is presumed to be caused by a compFesSiOn of the neighboring
s■licon matr■ x which must em■ t s■licon self― ■
nterstitials
■ntothe adjoining lattice during the second oxidation. Augmented by
these self― ■nterstitials′ the self― ■nterstitials caused by the second ox■ dation can condense ■nto Frank loops and can grow OSFis.
On the other hand′ no defect was observed in the area of the front surface corresponding to the damaged area in the back
surfacer as is shown in the lower― right part of Fig. 4.9. This means that the laser― ■nduced lattice defects can act as cOntinu―
■ng s■nks for ■mpur■ties dur■ ng the second ox■ dation.
In this way′ using the laser― induced lattice defects′ we can suppress OSF formation by preventing the formation oif oxidation―
induced impurity precipitates.
4.3.4 工]mprovement of generation lifetime through laser― damage getter■ ng
Although it is well known that controlled damage on the back surface of the wぎ fer can improve generation lifetime by removal
of point defects such as undesirable impurities′ it has not been established that laser― damage getterttng specifically improves
generation lifetime.
Results summar■ z■ng the effect of laser― damage getter■ ng on generation lifetttme are plotted in Fig. 4.10. The generation
lifetime in the gettered area was improved by about two orders of magn■tude over that in the ungettered area. Generation lifetime in the gettered area tended to level off at about 103 uSec. The
plots in the ungettered area′ however′ were spread between l and 100 uSec. TO investigate the reason for the improvement of the
lifetime′ the surface of the wafer was preferentially etched usttng the Sirtl etching technique after removal of the MOS capa―
citors. It was found that lttfetime improvement was caused by the suppression of oxidation― induced impurity precipitates and small OSFls by laser― damage gettering.
To evaluate the electr■ cal activ■ ty of these defects ■n the ungettered area′
the reciprocal of generation lifetime 1/tg Was plotted agattnst the density d of defects in Fig. 4.ll. The graph shows that the value of 1/tg iS nearly proportional to defect
density d. Therefore′ 1/tg Can be expressed by 1/tg = k ̀ d + 1/t。
where k is the proportional constant which activ■
ty of defects expressed in cm2/uSec
generation l土 fetttme expressed in uSeC24).
value of 2 × 10 6 土n Fig. 4.ll.
(4.1) defines the electr■ cal and to is background
The constant k had a
4.4 Summary
A method of gettering′ which we call laser― damage gettering′
has been developed to suppress the formation of lattice defects and to improve generation lifetime in silicon. This method uti―
lizes laser ■rradiation damage as getter■ ng s■tes for a var■ ety of po■ nt defects′ such as ■mpur■ties and self― ■nterstitials.
Summarizing the data′ the fo1lowing conclusions can be drawn.
l. High―
power laser pulses above 20 J/cm2 produced ther―
mally stable dislocation lines and dislocation clusters and 10W―
power laser pulses of about 15 J/cm2 also generated stabtte
qis10catiOn clusters and pseudo― swirl defects in the irradiated
region. Howeverr no generation of thermally stable defects was observed in the case of laser pulses of power lower‐ than 10
J/cm2.
2. Thermally stable lattice defects acted as s■ nks fOr
po■nt defects dur■ ng subsequent ox■ dation. LaSer― damage getter―
ing prevented the formation of oxidation― induced impurity preci―
pitates. An energy density of at least 15 J/cm2 was necessary to
produce the laser damage which produced effective getter■ ng.
3. Laser― induced thermally stable lattice defects
suppressed the formation of OSF:s by preventing the formation of oxidation― induced impurity precipitates.
4. Laser― damage gettering improved generation lifetime in the gettered area by two orders of magnitude over generation lifetime in the ungettered area. This lifetime improvement was
caused by the suppress■ on of defect formation by laser― damage getter■ ng.
REFERENCES
l) D.J.D.Thomas: Phys.status Solidi′ 3 2261 (1963).
2) H.」 .Quisser and P.G.G.Van Loon: J.Appl.Phy. 35. 3066 (1964).
3) G.R.Booker and R.Stickler: Ph ilos.Mag. 12 1303 (1965).
4) w.AoFisher and 」.A.Am ick: J.Electrochem.Soc。 113 1054 (1966).
5) M.L.Joshi: Acta Metall. 14 1157 (1966).
6) S.M.Hu: 」.Appl.Phys. 15 1567 (1974).
7) K.Taniguchi′ D.A.Antoniadis and Y.Matsushita: Appl.Phys.
Lett. 42 961 (1983).
8) T.Y.Tan and U.G6sele: Appl.Phys.Le tt. 19 86 (1981).
9) A.M.Lin′ R.W.Dutton′ D.A.Antoniadis and W.A.Tiller:
J.Electrochem.Soc. 128 1121 (1981).
10) D.工 .Pomerantz: J.Electrochem.Soc. 119 255 (1972).
11) J.E.Lawrence: J.Appl.Phys. 10 360 (1969).
12) J.Matsui and T.Kawamura: Jpn.J.Appl.Phys. l1 197 (1972).
13) T.Y.Tan and W.K.Tice: Ph ilos.Mag. 32 615 (1976).
14) D.Pomerantz: J.Appl.Phys. 38 5020 (1976).
15) G.A.RozgOnyi′ P.M.Pe troff′ and M.H.Read: J.Electrochem.Soc.
122 1725 (1975).
16) P.M.PetrOff′ G.A.RozgOnyi′ and T.T.Sheng: J.Electrochem.Soc.
125 565 (1976).
17) C.M.Hs ieh′ J.R.Mathews′ H.D.Se idel′ K.A.Pickar′ and C.M.Drum′ Appl.Phys.Le tt. 22 238 (1973).
18) G.H.Schwuttke′ K.Yang′ and H.Kappert′ Phys.Status Solidi A 42 553 (1977).
19) Y.Hayafuj i′ Y.Aokir and S.Kawado: Paper 3p― M‑12 presented at the Fall Meeting of the Japan Society of Applied Physics′
Osaka′ Japan′ Nov. 3‑5′ 1978.
20) C.w.Pearce and V.J.Zaleckas′ J.Electrochem.Soc. 12■ 1436 (1979).
21) Y.Hayafuj i′ T.Yanada′
and Y.Aoki: Paper 485 presented at the
Electrochemical Society Meeting′ Los Angeles′ California′
Oct. 14‑19′ 1979.
22) L′ E.Katz′ C.w.Pearcer a,nd P.F.Schmidt: Paper 486 presented at the Electrochemical Society Meeting′ Los Angeles′
California′ oct. 14‑19′ 1979.
23) F.P.He iman′
IEEE Trans.Electron Devices′ ed‑14 781 (1967).
24) Y.工
chidar T・Yanadar and s.Kawado: ASTM STP 721′ American
Society for Testing and Materials′ p.107 (1980).
Fig. 4.1. Transm■ss■On electron m■
crographs sho tions
■nduced at laser pulses w■
th anOf (a)10 J/cm2′ (b)15
」/cm2 and (c)
w■ng disloca―
energy dens■ ty
22 J/cm2.
Transmission electron micrographs of sample oxidized after irradiation by 15 J/cm2 pulses. Fig. 4.3(o) and (c)are magnified views which correspond to 00d00 and 00s00 in Fig. 4.3(a)′ respectively.
Fig。 4.3.
(111)
[1121
中 [1101
Laser Traces
[0111
１一 一１
２
x―ray topograph by 1630 J/cm2′
from the laser
グ
[1211
Of oxidized wafer irradiated that dis10CatiOns moved far and formed netWOrks.
Fig, Pos■tiOn of dis10cation Clusters.
Fig. taken
show■ng grooves
Fig. 4.5. Etch patterns Of
■rradiation and 10 」
/cm2 and 15
the oxidized samples (b)and (c)with irrad J/cm2′
respectively.
(a)without
‑ 71 ‑
iation by
『織畿
Fig. 4.6. optical micrographs Of
diation by 1630 J/cm2′
suppress■ on of defect
wafer oxidized after irra―
show■ng laser̲damage getter■ ng formation.
Impuritie,
θ θ
θ ノ〆 : 11
Defect D:siomtions Impurities
Ox:dation
:ons
laser damage gettering by low― power
of about 15 J/cm2.
ａｍ
Ｂｅ ｌ ｌ Ψ ｒ ｅ Ｓ ａ ｌ ｌ 上 Ｖ Ｌ
● θ
ｌ ｌ ｌ ｌ
■
▼
Fig. 4.7. Mechanism of laser pulses
Prec:pitates Disiocat:ons
S:02
Dlstocat Precipltates
ヽ :
θ θ θ θａｍ ｌ
Ｂｅ
ｉ
■ Ｖ
ｅｒ
Ｓ Ｌａ
←
︱
︱
︱
︱ 上 Ｖ
Impuritles
ｌ ｌ ｌ ｌ ｌ
←
▼
Impurities
ｉ ｌ ｌ ｌ
←
▼
Oxidation
Si02
orks
Fig. 4。 8. Mechanism of laser damage gettering by high― power
laser pulses above 20 J/cm2.
Strained Reg:on
′ ′ ′ θ C
θ θ 。 ′ , Dislocations
Fig. 4.9. Optical micrograph showing the effect of laser― d amage getter■ ng. The upper half was ox■ dized oncer and
the lower half was oxidized twice. The right half corresponds to the damaged area on the back surface′
and the left half to the nondamaged area.
‑ 75 ‑
1。4
10 5 0 5̲ 10
Pos:t:on(mm) ungettered+Cettered
Fig. 4.10. Distribution of generation lifetime in the gettered and ungettered areas.
０３
︵ ｏ ｏ ｎ ュ
︶
０ ２
♂
♂
● ＦＥ ０ Ｊ Ｅ
〇 一 ０
﹂Ｏ Ｃ０ ０
0。6
0 1。 0 2。O X105 Defect Density (cm 2)
Fig. 4.ll. 1/tg VS. d curve. The tangent of thiS Curve
indicates the electrical activity Of defects.
０
．
０
︒
０
︒
０
︒
０
︒
０
︵Ｔυ●●■︶ ＯＣ﹂一●ヒＪ Ｃ〇
一一
﹂Φ⊂００ ︸０
一 ●００﹂Ｑ
一 〇∝０
●
●
●
●●
● ｎ ｕ
●
●■
●
●
●
●
●
●
●
●
●
●●
●
●●
●
●
●
●●
● ８
■ １
●
CHAPTER 5 ANNIHILATION OF STACKING FAULTS IN VICttNAL SURFACES 5.l Introduction
Under ordinary c■ rcumstances′ ox■datiOn―■nduced stacking faults (osFls)in silicon crystal are very stable thermally and
mechanically. However′ there are several circumstances where the
pre―existing OSFls can be annihilated from the crystal during thermal oxidation of silicon. Three mechanisms have been
reported for annihilating OSFOs during oxidation: Mechanism I′
proposed by Hsiehl)et al.′ 。perates when the OSF]s are consumed by the advancing oxide― silicon interface. Mechanism II′ reported by Ravi2)and Hu′ 3)。 perates when there is a decrease in the
chemical driving force which grows the OSFls. This decrease can be brought about by lowering the oxidation rate. Mechanism III operates when there is an unfaulting reaction.2′ 4) There are drawbacks in both Mechanisms l and II. Mechanism l is capable Of annihilating only sma1l osFOs or OSF:s whose habit planes are of a sha1low angle to the surface and it only can operate when the growth rate of the ox■ de ■s greater than the growth rate of the OSFls. Hu has reported that in Mechanism ttl at temperatures
above about 12000Cr the growth of OSFls does not fo1low Eq.(3.2)
and a retrogrowth of osFls′ i.e.′ a shrinkage of OSFls′ ocCurS.
Hu suggested that the onset of the shrinkage of oSF's is depen―
dent on surface orientation and that the annealing out of OSFOs occurs in the final stage of retrogrowth. In the operation of Mechanism III′ hOwever′ Ravi2)and Kawado4)have observed that the elimination of osFls is caused by an unfaulting reaction which may be caused by the nucleation and expansion of two Shockley partials during successive oxidation processes. Among these three mechanisms′ Mechanism III is the most useful because it can be easily triggered in usual oxidation conditions and
because of its suitability in studying the interaction of OSFls
w■th crysta1lographic defects.
OSF annihilation by the Mechanism III′ however′・is not yet well understood. In this chapter′ we study the effect of
Mechanism III on oSF annihilation on vicinal surface of (001) orientation in silicon and propose two kinds of the unfaulting reactiOns which lead to the conversion of OSFls to perfect
dislocation loops. A statistttcal treatment of OSF annihilation
■s presented and compared w■ th the exper■ mental results.
5.2 Experimental Procedure
Silicon wafers about 300 μm thick w■ th different or■ enta―
tions from O° to 6° off the [001]axis along the [110]zone were
prepared from p― type′ 40 to 60 Ω―cm′ dislocation― free Czochralski crystals with the growth axis in a [001]orientation. The front
surface of the wafers was mechanically and chemically polished′
while the back surface was chemically etched. Thermal oxidation was carried out in dry 02 at 1150° C for O.l to 200 hr. After an
ox■de layer was etched off in a dilute HF solution′ the wafers were etched in a Wright etchant5)to measure length of OSF:s.
The detattls of the defect structure were examined by x― ray diffraction topography.
5.3 Exper■mental Results
Figure 5.l shows the etch patterns of three O.9° ―off― (001) wafers ox■ dized for 20′ 50′ and 200 hr′ respectively. Two kinds of defects ■mages′ A and B′ were observed in the photographs.
X―ray analysis showed that image A corresponded to a OSF bounded
by a a/3<lll> Frank partial dislocation and image B to a perfect dislocation loop with a Burgers vector of a/2く 110> type. With
increasing oxidation time′ the density of OSFis along the [110]
direction′ lying on the (111)and (111)planes′ gradually decreased and the density of perfect loops along the [110]direc―
tion increased. The density of OSFls (DI)On the (111)and (l11) planes for different or■ entations were measured and were plotted in Fig. 5.2.
On the other hand′ osFOs along the [110]direction′ lying on the (111)and (111)planes′ disappeared quicker than those along
the [110]direction. All of the OSFIs and perfect 100ps along
the [110]direction disappeared with an oxidation time of 200 hr.
The variation of the density (DII)with oxidation time for OSFOs on the (111)and (111)planes is shown in Fig. 5.3. A marked
surface orientation dependence on the fault density was apparent′
while the density of osFls on the (111)and (l11)planes was
independent of the surface orientation. In an extreme case′ in a
3.7°―off― (001)wafer′ all the OSFOs dttsappeared within 5hr. In
all cases′ the density of OSFls on the (l11)and (111)planes could be expressed as
DII = DIIoexp(― αtnl)′ (5.1)
dependent surface value of zero for a where DIIo was the initial
the surface or■ entation and
or■entation and found to be 3.5 hr 0・ 4 for a 3.7° ―off― (
(001)Wafer.
dens■ty′ α
was a conStant nl was independent of the
%0.4。
α
showed a maximum 001)Wafer and was nearly
On
5.4 Discuss■ on
The result suggests that the ann■ hilation mechan■ sm for OSFls on the (111)and (111)planes is different from that for
OSFIs on the (111)and (l11)planes′ although the nucleation and growth of OSFIs embryos occur in the same manner in four′ {lll・} planes at the beginning of oxidation.
5.4.l Annihilation of OSFls on the (111)and (l11)planes The anntthttlation of OSF:s on the (111)and (111)planes is caused by the impingement of two Frank partial dislocations on
these planes′ as in image B in Fig. 5.1(c). We statistica■ ■y evaluated the impingement of Frank partials′ i.e.′ fault annihi―
lation. The rate of the impingement (dR1/dt)iS related to the growth rate of OSFls as
dR1/dt
〓 27TDILcosOdL/dt′where DI is the density of OSFls on the (111 0 is the angle between、 [001]and either [lll the OSF length. The growth rate of OSFls is
ox■dation time t′ temperature T′ and oxygen an oxidizing ambient.6′ 7) The fault length
(5.2) )and (lll)planes′
]or [llll and L is known to depend on partial pressure p in L can be oexpressed as L 〓 Alpmtnexp(‑00/kT)′
≡ βtn′ (5.3)
where m and n are number exponents′ Ql the actiVation energy associated with the growth of OSFls′ k the Boltzmann cOnstant′
and Al a cOnstant dependent on the surface orientation. If we introduce the probability PI that an OSF remains by the time t′
then the density DI can be expressed as DI = DIol ′
where DIO is the density The relation between the
■s expressed as dP1/dt = Combining Eqs.(5。 2)
bility as
(5.4) in the initial stage of oxidation.
probability and the rate of impingement
‥
PIdR1/dt.
′ (5.3)′ and (5.5)′ we can express the proba―
(5.6)
l =(2π
DIoβ2t2ncosO+1)‑1.
Substituting Eq.(5.6)into Eq.(5.4)′ we obtain 1/Di ‑ 1/PI。 = 27Tβ
2t2ncosO.
(5.5)
(5.7)
Plotting the OSF length versus timer the values on n and
βwere
■
ndependent of the surface or■ entation and found tO be about O.67 and 13
μm hr‑0・ 67′ respectively′which is in good agreement with the values reported in the literature。
8′9) The curve calculated from Eq.(5.7)is shOWn in Fig. 5.4。 The measured values of 1/DI
‑ 1/DIo were also plotted against tl・
34 in Fig. 5.4′where the
values of DIo were nearly equal to 103 cm 2 for the different
surface or■ entations. The calculated curve ■s ■n good agreement
w■th the exper■ mental results.
We propose a model to explain how perfect dttslocation loops are annealed out of a crystal after unfaulting reaction. The model is schematically illustrated in Fig. 5.5. At stage (b)in
Fig. 5.5 the dislocation loops have an identical Burgers vector
of a/2[110]parallel to the surface′ which was determined by
image contrast analysis of x― ray topographs. The unfaulting reaction may be caused by the nucleation and the passage of two
Shockley partial dislocations (one above and one below the extra plane)across the fault.2) A possible unfaulting reaction for
the faulted loop of the Burgers vector a/3[111l can be expressed by a/3[llll + a/6[12=]+ a/6[211]→
a/2[110]′and for the
a/3[111] faulted loop′ a/3[llI]+ a/6[=21]+ a/6[211]
→a/2[110].
5.4.2 Annihilation of OSFIs on the (111)and (I11)planes Equation (5.1)suggests that the annihilation of OSF's On the (111)and (111)planes is not caused by the impingement of two Frank partials′ though the result shown in Figs. 5.1(a)and
5。 1(b) indicates the existence of an unfaulting reaction in the
ann■hilation mechan■ sm. In this caser the unfaulting reaction ■s undoubtedly related to the familiar terrace― ledge―kink (TLK)
structure on a slightly misoriented surface. Because linear
steps ■n the TLK structure ■ntroduce lattice stress near the bulk
surfacer these steps can be regarded as surface dislocations.10′ 11) The annihilation of OSFls on the (111)and (l11)planes may be
caused by the shrinkage of dislocation loops converted by the interaction of OSFls with surface dislocations. Let uS derive the density of oSF:s (DII)on the (l11)and (111)planeS Statis―
tically. We assume that osF annihilation is caused by the impingement of osF:s on surface dislocations. The rate of the
impingement (dRI1/dt) iS related to the growth rate of OSFls′
expressed by Eq.(5.3)′ as
(5.8) where d is the dens■ ty of surface dislocations′ which depends on
surface orientation and time. In the same manner as the probabi―
lity PI was derived′ we can obtain the probability PII that an OSF will remain by time t｀ as:
PII = exp(―nβ (5.9)
The density of surface dislocations (d)iS usually given by dRI1/dt = dnβtn 1′
ホ
tn ldt)・d = tanψ/h′ (5.10)
wheFe
ψis the angle of the surface orientation from the [001]
direction and h is the height of a monatomic linear step.12) Howeverr the stress induced by the surface dislocations cannot be enough to convert̀osFIs to perfect dislocations′ because each
surface dislocation ■s regarded as a source of lattice stress ■n a pseudomorphic monolayer.10) Thereforer instead of d′ we have to use in Eq.(5.9)the effective density of the surface disloca―
tions deff WhiCh Can convert OSFls tO dislocation loops. Tan et al.13)reported that a tensile plane stress T exists in silicon near the ■nterface layer dur■ ng thermal ox■ dation. They expressed the stress as:
T ∝ (dL/dt)1/m′
(5.11)
where m is a constantr usually between l and 2.14) This stress can be concentrated at the surface dislocations.15) so that the stress concentration ■s such that the stress around the surface dislocations becomes large enough to convert OSF:s to dislocation loops. Hencer the effective density of the surface dislocations deff Can be expressed as:
deff = f(d)T′
where f(d)iS a function of Eqs.(5.10)′ (5.ll)′ and (5.
formula for the proり ability
(5.12)'
d′ independent of t. By using
12)′ Eq.(5.9)gives the fo1lowing PII:
PII = eXp[―nβ g(d)tn‑1/2m]′ (5.13)
or■entation.
planes can
(5.14) where g(d)is a function of dr dependent on surface
Then the density DItt of OSFls on the (111)and (111) be expressed as:
DII = DIIoexp[― nβ g(d)tn 1/2m]′
where DIIO is the initial density at t = 0. This equation has the same formulation as Eq.(5.1). P10tting the OSF length versus timer the value of n was found to be about O.70. Stress relaxa―
tion experiments using various silicate glasses often give the m
valu9 of 2.14) substituting these values of n and m into
Eq.(5.14)′ we obtain O。
45 as the exponent of t. This is in good agreement with the experimental result of O.4. The surface
orientation dependence of
αin Eq.(5.1)comes from g(d)in
口
q.(5.14). At present′
however′we cannOt compare the experimen―
tal values of
αwith the theoretical values of nβ
g(d)′because g(d) is an unknown function of d.
5.5 Summary
The results obtained can be summarized as fo1lows:
(1) There are two modes of annihilation caused by unfault―
ing reaction for OSF.s in silicon wafers with an orientation of
from O° to 6°
off the [001l axiS along the [110]ZOne. one mode is independent of the surface orientation and the other depen―
dent.
(2) The density of osFts on the (l11)and (l11)planes is
given by 1/DI ‑ 1/DIO
∝tl.34 and is independent of the surface
or■entation. A fault ann■ hilation of this type ■s caused by the impingement of two Frank partial dislocations.
(3) After unfaulting reaction′ two perfect dislocation loops move on to the (110)plane perpendicular to the surface′
then leave the crystal by climb motion.
(4) The density of osFls on the (lll)and (111)planes is given by DII = DttIO exp(― αtO・ 4)′ where α is dependent on the surface orientation. A fault annihilation of this type is caused by the impingement of osFls against linear steps of the TLK
structure in the slightly misoriented crystal surface.
REFERENCES
l) C.M.Hsieh and D.M.Maher: J.Appl.Phys. ■4 1302 (1973).
2) K.V.Ravi8 Phìlos.Mag. 30 1081 (1974).
3) S.M.Hu: Appl.Phys.Le tt. 27 165 (1975).
4) s.Kawado: Jpn.J.Appl.Phys. 19 815 (1980).
5) M.wright Jenkins: J.Electrochem.Soc。 124 757 (1977).
6) s.P.Murarka: Phys.Rev. B16 2849 (1977).
7) s.P.Murarka8 Phys.Rev. B21 692 (1980).
8) M.conti′ G.Corda′ R.Matteucci′ and c.Ghezzi: J.Mater.Sci. 10 705 (1975).
9) S.P.Murarka: J.Appl.Phys. 48 5020 (1977).
10) D.L.Rode: Phys.status solidi A32′ 425 (1975).
11) C.A.B.Bal18 Phys.status solidi 42 357 (1970).
12) w.K.Burton′ N.Cabrera″ and F.C.Frank: Philos.Trans.Roy.soc.
London′ 243A 299 (1951).
13) T.Y.Tan and U.Gざ
sele: Appl.Phys.Le tt. 旦9 86 (1981).
14) G.M.Bartenev: The structure and Mechanical Properties of lnorganic Glasses (wolters―
Noordhoff′ Groningen′ 1970)′p。 123.
15) J.Aleck: J.Appl.Mech. 16 118 (1949).
(b)
一１
Ｆ Ｉ
Ｃ Ｌ ノ Ｅ ヽ
Fig. 5.l Etch patterns On O.9° ―Off― (001)Wafers
l150° C fOr 20 h (a)′ 50 h (b)and 200 0SF and image B tO A COrresponds tO a
dislocation loOp.
[170]
Ox■dized at h(C). Image a perfect