Electron-beam induced crystallization transition in self-developing amorphous AIF3 resists

Article abstract of DOI:10.1063/1.117361

Transmission electron microscopy is used to investigate electron-induced crystallization of thermally evaporated amorphous AlF₃(α-AlF₃). It is shown that this material undergoes a very complicated crystallization process with three crystalline substances (Al, AlF₃, and Al₂O₃) formed as the dose increases. The sequence of the crystallization is highly sensitive to the presence of water, which inhibits radiolytic dissociation of α-AlF₃ into Al and fluorine, reduces the dose required for the crystallization of α-AlF₃, and causes the transformation of AlF₃ into Al₂O₃.

Full text of DOI:10.1063/1.117361

Electronbeam induced crystallization transition in selfdeveloping amorphous AlF3
resists
G. S. Chen, C. B. Boothroyd, and C. J. Humphreys
Citation: Applied Physics Letters 69, 170 (1996); doi: 10.1063/1.117361
View online: http://dx.doi.org/10.1063/1.117361
View Table of Contents: http://scitation.aip.org/content/aip/journal/apl/69/2?ver=pdfcov
Published by the AIP Publishing
Articles you may be interested in
Electron beam induced crystallization of a GeAu amorphous film
J. Appl. Phys. 80, 6170 (1996); 10.1063/1.363691
Electronbeam induced etching of resist with water vapor as the etching medium
J. Vac. Sci. Technol. B 14, 4262 (1996); 10.1116/1.588587
Enhanced resistance to metal induced crystallization of amorphous Ge in contact with Bi
Appl. Phys. Lett. 69, 2039 (1996); 10.1063/1.116872
Ionbeam synthesis of amorphous SiC films: Structural analysis and recrystallization
J. Appl. Phys. 79, 6907 (1996); 10.1063/1.361514
Crystallization of amorphous silicides by energy beam annealing
AIP Conf. Proc. 50, 533 (1979); 10.1063/1.31712
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
132.174.255.116 On: Tue, 02 Dec 2014 15:11:41

Electron-beam induced crystallization transition in self-developing
amorphous AlF3 resists
G. S. Chen
Department of Materials Science, Feng Chia University, Taichung 407, Taiwan
C. B. Boothroyd and C. J. Humphreys
Department of Materials Science and Metallurgy, University of Cambridge,
Cambridge CB2 3QZ, United Kingdom
͑Received 21 February 1996; accepted for publication 30 April 1996͒
Transmission electron microscopy is used to investigate electron-induced crystallization of
thermally evaporated amorphous AlF3͑a-AlF3). It is shown that this material undergoes a very
complicated crystallization process with three crystalline substances ͑Al, AlF3, and Al2O3) formed
as the dose increases. The sequence of the crystallization is highly sensitive to the presence of water,
which inhibits radiolytic dissociation of a-AlF3 into Al and fluorine, reduces the dose required for
the crystallization of a-AlF3, and causes the transformation of AlF3 into Al2O3. © 1996 American
Institute of Physics. ͓S0003-6951͑96͒01128-X͔
Direct electron beam writing on low-molecular weight,
ionic oxides, and halide materials is currently the most prom-
ising method for defining ultra-small nanostructures.1 Such
materials have potential applications as high-resolution elec-
tron beam resists, or for high density information storage.
Among which amorphous aluminum trifluoride (a-AlF3) is
almost the most attractive material being used as a positive
or negative self-developing resist.2,3 Several research groups
have demonstrated the potential of the subnanometer cutting
and ruling by an intense beam of electrons ͑SCRIBE͒ ‘‘hole-
drilling’’ technique because well-defined nanometer-scale
holes and lines can be cut through about 50 nm of a-AlF3 in
only about a few milliseconds.4–7 However, because
a-AlF3 is so sensitive to beam irradiation, its exposure
mechanism is still not fully understood and contradictory
results have even been presented.6,7 In addition, that water
can play a role during the SCRIBE/hole drilling process has
been widely suspected, but its effects in modifying the drill-
ing mechanisms have rarely been quantitatively studied.8,9
We have noted that previous researchers always assume that
the exposure behavior of a-AlF3 is unaffected by the humid-
ity of the films. However, a-AlF3 films are normally porous
and contain varying amounts of water, some of which is
chemically bonded to the AlF3 and some of which is ab-
sorbed on the surface and in the pores.10 We have first dem-
onstrated that a-AlF3 films undergo a series of very compli-
cated crystallization transition processes, which are highly
sensitive to the presence of water, if irradiated in conven-
tional transmission electron microscopes ͑TEMs͒.11 In this
letter, we use electron microscopy to present a means of
semiquantitatively analyzing the effect of the water on the
crystallization process of this material.
Figure 1 shows a dark field image, recorded from the
a-AlF3 diffuse ring ͑inset͒ and with minimum dose, which
exhibits speckle contrast characteristic of amorphous materi-
als. A diffraction pattern from a briefly damaged a-AlF3 film
is shown in Fig. 2 ͑inset͒. It can be seen that the a-AlF3
undergoes a crystallization transition upon irradiation form-
ing Al and crystalline AlF3(c-AlF3). After prolonged irra-
diation a strong 0.14 nm diffraction ring of Al2O3 also
emerges. Dark field images of the irradiated crystalline prod-
ucts ͑Al and AlF3) are shown in Fig. 2 from which it can be
seen that the Al microcrystallites are randomly oriented,
while the c-AlF3 has more variable grain sizes. Selected area
diffraction instead of imaging technique was performed to
investigate the process of the crystallization transition since
the films were so sensitive to beam damage that it was dif-
ficult to record a few images without causing serious dam-
age.
In order to assess the effect of water on the crystalliza-
tion of a-AlF3, we studied two types of thermally evaporated
films. For the dry films, anhydrous AlF3 was heated at
400 °C for 10 h to remove water prior to evaporation. For the
wet films, AlF3•3H2O was evaporated without preheating.
Evaporation of the dry films was carried out at a pressure of
The crystallization process was investigated with a
JEOL 2000FX TEM operating at an accelerating voltage of
100 kV in conjunction with a Gatan parallel electron energy
loss spectrometer ͑EELS͒. All films ͑50 nm͒ were deposited
with an Edwards 306 thermal evaporator onto carbon sup-
porting films ͑ϳ5 nm͒ on standard TEM grids. The deposi-
tion rate, monitored with a quartz thickness monitor, was 0.4
nm/s.
FIG. 1. Dark field image taken using an aperture on the AlF3 diffuse ring
͑inset͒ showing the amorphous structure of the as-prepared evaporated
AlF3.
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
170 Appl. Phys. Lett. 69 (2), 8 July 1996 0003-6951/96/69(2)/170/3/$10.00 © 1996 American Institute of Physics
132.174.255.116 On: Tue, 02 Dec 2014 15:11:41

FIG. 2. Diffraction pattern ͑inset͒ showing the Al͕111͖ and AlF3͕110͖ rings.
͑a͒ and ͑b͒ show dark field images taken from the same region of an irradi-
ated AlF3 film using objective apertures on the Al͕111͖ and AlF3͕110͖ rings,
respectively.
ϳ5ϫ10Ϫ6 mbars whereas the wet films were evaporated at a
much higher pressure (ϳ1ϫ10Ϫ4 mbars) caused by the re-
lease of the hydrated water. The amount of water left in the
films was identified using transmission infrared spectros-
copy, which showed that the wet films had an enormous
absorption band at 3 ␮m ͑O–H stretch͒ while this absorption
band in the dry films was negligible. Infrared spectroscopy
also showed that the AlF3 films absorbed water rapidly when
they were exposed to air. X-ray diffractometer showed that
all types of heated powder possessed the same ͑tetragonal͒
structure. All the films were freshly prepared and transferred
to the TEM within 15 min of preparation. The beam current
was measured with a Faraday cage at the side of the speci-
mens, which was routinely checked every ϳ10 min, making
sure that any decay of beam current was compensated for.
The same area of the specimen was irradiated throughout
each experiment, with the irradiation being stopped at fre-
quent intervals for a series of diffraction patterns to be ob-
tained. Diffraction patterns from each series were then digi-
tized and radially averaged to produce electron intensity as a
function of scattering angle. Figure 3 shows typical resultant
FIG. 4. Normalized diffraction peak areas of the Al͕111͖, c-AlF3͕110͖, and
the Al2O3 0.14 nm ring as a function of dose for ͑a͒ dry and ͑b͒ wet types of
films. All patterns were taken under identical conditions, therefore, the peak
areas are comparable between the two figures. It is to be noted that the total
amount of AlF3 in the film cannot be directly judged from the c-AlF3 ͕110͖
peak area because this sharp peak is on top of a very broad diffuse peak of
a-AlF3. ͑c͒ shows the integrated counts of the O 535 eV and F 685 eV K
edges ͑measured within a 50 eV window͒ as a function of dose.
diffraction intensities of three diffraction patterns from a se-
ries for the dry type of films irradiated at a current density of
6.9ϫ103 A mϪ2, showing that the damage process is
gradual with the broad peaks of a-AlF3 disappearing and
sharp peaks of crystalline substances appearing.
For each of the two types of films, diffraction patterns
were obtained ͑strictly under identical conditions to each
other͒ over a range of doses from 105 to 108 C mϪ2. The
diffraction patterns were radially averaged as in Fig. 3 and
the areas under the most prominent peak for each crystalline
reaction product found. Figures 4͑a͒ and 4͑b͒ plot the dif-
fraction peak areas as a function of dose for the two types of
films, allowing the crystallization process to be monitored as
a function of dose. The crystallization behavior of Al for the
two types of films at low dose regime is similar. However, a
FIG. 3. Radial average of diffraction patterns from a series showing the
progress of crystallization transition from Al, to AlF3, then Al2O3.
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
Appl. Phys. Lett., Vol. 69, No. 2, 8 July 1996 Chen, Boothroyd, and Humphreys 171
132.174.255.116 On: Tue, 02 Dec 2014 15:11:41

much lower dose (ϳ2ϫ105 C mϪ2) is needed for the
a-AlF3 to crystallize in the wet films than in the dry films
(ϳ1ϫ106 C mϪ2). In addition, if the dry films are exposed
to air for a few days then their behaviors become more like
those of the wet films. For a dry AlF3 film, measurements
were made of the areas under the oxygen and fluorine EELS
K edges as a function of dose ͓Fig. 4͑c͔͒. It can be seen that
up to a dose of 1ϫ106 C mϪ2, fluorine decreases rapidly,
then more slowly and at a uniform rate above this dose. 1
ϫ106 C mϪ2 corresponds to the dose when a-AlF3 starts to
crystallize. Thus, up to 1ϫ106 C mϪ2, fluorine is being lost
and the AlF3 remains amorphous. At 1ϫ106 C mϪ2, the
a-AlF3 begins to crystallize and is more stable to damage, so
the rate of loss of fluorine decreases.
Al2O3 crystallites are observed in the two types of films
at high dose regime. Figures 4͑a͒ and 4͑b͒ show that, over
the range of doses from 8ϫ106 to 5ϫ107 C mϪ2, where the
Al2O3 is visible, the rapid increase of the Al2O3 signal is
consistent with the fast decrease of the c-AlF3 signal, but the
intensity of Al declines only slightly. Energy loss spectra
reveal that the oxygen 535 eV K edge is present in all as-
prepared films and increases a little during irradiation at low
doses, even though no crystalline Al2O3 is observed at these
doses. The oxygen ͑presumably water picked up from the
microscope vacuum, ϳ1ϫ106 mbars͒ is incorporated into
gether with Al colloids as a result of beam irradiation. This
could have a significant effect on the hole-drilling behavior
of this material. It is known that the crystallinity of a mate-
rial plays a crucial role in deciding the transport properties of
species produced during hole-drilling.16–18 For a-AlF3, an
abundance of data have already confirmed that it is a ‘‘pop-
ping’’ type of drilling material, which exhibits an abrupt
mass-loss behavior and tends to be associated with displace-
ment of dissociated produce ͑Al͒ from irradiated volume and
anion aggregation to form bubbles of fluorine gas. It requires
a threshold dose of у105 C mϪ2 for the bubbles to ‘‘pop.’’
For the wet films, the dose required to begin to form Al is
approximately the same as that for which crystallization of
the remaining a-AlF3 occurs, and is about the same magni-
tude as the threshold dose for hole drilling. As c-AlF3 is
more resistant to radiolysis than a-AlF3, and given the de-
pendence of hole drilling on the structure ͑crystalline or
amorphous͒ of a material already discussed by many re-
searchers, clearly the transition of a-AlF3 to c-AlF3 will al-
ter the subsequent hole-drilling behavior of AlF3 and thus
retard the drilling process.
In summary, we have shown that the electron-beam
damage of dry a-AlF3 is a very complicated process,
whereby crystalline Al is formed first as fluorine is lost, fol-
lowed by the crystallization of a-AlF3 and the formation of
Al2O3. It is important to note that the water content of the
films can greatly alter the doses required for each substrate to
crystallize.
the films as the fluorine is lost. Above
1
ϫ107 C mϪ2, Al2O3 suddenly begins to form, correspond-
ing to an abrupt increase in the oxygen signal at this dose
and a sudden decay in the c-AlF3. Above 5ϫ107 C mϪ2
,
1 C. J. Humphreys, T. J. Bullough, R. W. Devenish, and D. M. Maher, Inst.
Phys. Conf. Ser. 119, 319 ͑1991͒.
the concentration of oxygen levels off when all the c-AlF3
has reacted, and the Al2O3 itself may be decomposing due to
irradiation.12 We have also noted that, in some ͑electron-
beam deposited͒ a-AlF3 films, the intensity of Al was very
weak within doses from ϳ1ϫ106 to 5ϫ107 C mϪ2, but the
signals of c-AlF3 and Al2O3 were similar to those of Figs.
4͑a͒ and 4͑b͒. The intensity of the Al2O3 kept increasing
even though there was no detectable Al left after the dose
2 A. Muray, M. Scheinfein, and M. Isaacson, J. Vac. Sci. Technol. B 3, 367
͑1985͒.
3 E. Kratschmer and M. Isaacson, J. Vac. Sci. Technol. B 5, 369 ͑1987͒.
4 C. A. Walsh, Philos. Mag. A 59, 227 ͑1989͒.
5 Y. Ito, A. L. Bleloch, S. J. R. Granleese, and L. M. Brown, Inst. Phys.
Conf. Ser. 138, 507 ͑1993͒.
6 R. M. Allen, S. J. Lloyd, and C. J. Humphreys, Inst. Phys. Conf. Ser. 138,
87 ͑1993͒.
7 V. I. Nikolaichik, Philos. Mag. A 68, 227 ͑1993͒.
8 P. M. Mankiewich, H. G. Craighead, T. R. Harrison, and A. H. Dayem,
Appl. Phys. Lett. 44, 468 ͑1984͒.
exceeded ϳ1ϫ107 C mϪ2 13 As the Al is almost negligible,
.
it is thus unlikely that the Al2O3 is directly transformed
from reactions of the Al with any compound. From the
above results, we believe that the Al2O3 is transformed
from reaction of ͑amorphous and crystalline͒ AlF 3 with
an oxygen-related compound ͑presumably water͒, and there-
fore propose a possible chemical reaction mechanism:
2AlF3ϩ3H2O→Al2O3ϩ6HF.
9 R. W. Devenish, D. J. Eaglesham, D. M. Maher, and C. J. Humphreys,
Ultramicroscopy 28, 324 ͑1989͒.
10 J. D. Targove, B. G. Bovard, L. J. Lingg, and H. A. Macleod, Thin Solid
Films 159, L57 ͑1988͒.
11 G. S. Chen, C. B. Boothroyd, and C. J. Humphreys, Inst. Phys. Conf. Ser.
119, 325 ͑1991͒.
12 C. J. Morgan, S. J. Bailey, A. R. Preston, and C. J. Humphreys, Inst. Phys.
Conf. Ser. 119, 503 ͑1991͒.
13 G. S. Chen, Ph.D. thesis, University of Cambridge, 1994.
14 W. Heitmann, Thin Solid Films 5, 6 ͑1970͒.
Previous investigations into the possibility of using ther-
mally evaporated AlF3 as an optical film have shown that its
structure is porous with a packing density of ϳ60% of bulk
AlF3 and water can be easily incorporated into the film.14,15
This study has found that the ‘‘wet’’ AlF3 is prone to crys-
tallization at the low dose regime and c-AlF3 is formed to-
15 A. P. Bradford, G. Hass, and M. McFarland, Appl. Opt. 11, 2242 ͑1972͒.
16 R. M. Allen, Ph.D. thesis, University of Cambridge, 1992.
17 R. Zanetti, A. L. Bleloch, M. Grimshaw, J. H. Paterson, and G. A. C.
Jones, Inst. Phys. Conf. Ser. 138, 67 ͑1993͒.
18 C. J. Morgan, Ph.D. thesis, University of Cambridge, 1994.
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
172 Appl. Phys. Lett., Vol. 69, No. 2, 8 July 1996 Chen, Boothroyd, and Humphreys
132.174.255.116 On: Tue, 02 Dec 2014 15:11:41