Size Effect on the Short Range Order and the Crystallization of

Subscriber access provided by MEMORIAL UNIV

Article

Size Effect on the Short Range Order and the Crystallization of Nano-sized Amorphous Alumina Leonid Bloch, Yaron Kauffmann, and Boaz Pokroy Cryst. Growth Des., Just Accepted Manuscript • DOI: 10.1021/cg500580c • Publication Date (Web): 10 Jul 2014 Downloaded from http://pubs.acs.org on July 13, 2014

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Crystal Growth & Design is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 23

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Crystal Growth & Design

Size Effect on the Short Range Order and the Crystallization of Nano-sized Amorphous Alumina Leonid Bloch,† Yaron Kauffmann,† and Boaz Pokroy∗,†,‡ Department of Materials Science and Engineering, Technion – Israel Institute of Technology, Haifa, Israel, and Russell Berrie Nanotechnology Institute, Technion – Israel Institute of Technology, Haifa, Israel E-mail: [email protected]

Abstract Drawing inspiration from nature, where some organisms can control the short range order of amorphous minerals, we successfully manipulated the short range order of amorphous alumina by surface and size effects. By utilizing the Atomic Layer Deposition (ALD) method to grow amorphous nanometrically thin films, combined with state-of-the-art electron energy loss spectroscopy (EELS) and X-ray photoelectron spectroscopy (XPS), we showed experimentally that the short range order in such films is strongly influenced by size. This phenomenon is equivalent to the well-known size effect on lattice parameters and on the relative stability of different polymorphs in crystalline materials. Additionally, direct measurements of crystallization temperature have shown a dramatic increase in amorphous to crystalline phase transition temperature for thinner amorphous Al2 O3 films. Finally, we show that the short range order changes while still in the amorphous phase, before the amorphous to crystalline transformation takes place. ∗

To whom correspondence should be addressed Department of Materials Science and Engineering, Technion – Israel Institute of Technology, Haifa, Israel ‡ Russell Berrie Nanotechnology Institute, Technion – Israel Institute of Technology, Haifa, Israel †

1 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Many organisms use crystallization through an amorphous precursor as the route for biologically controlled mineralization. 1 This approach has several advantages, such as lowering the activation energy for crystal formation, control over the final shape of the crystal, controlled incorporation of impurities, and control over the resulting polymorph. 2 The latter is achieved by manipulation of the short range order in the amorphous phase, to resemble that of the succeeding crystalline phase. 3 This control is achieved through different additives, such as organic molecules, water, magnesium, or phosphorus. 4 This strategy used by organisms, if emulated synthetically, would have a tremendous impact, as many aspects of science and technology rely on amorphous materials 5 and require specific crystalline structures in crystalline materials. 6 The crystalline structure can be controlled mainly through temperature and pressure manipulation (thermodynamics) and by epitaxial growth. 7 However, to the best of our knowledge there is no study or technology in which a polymorph has been stabilized via control of the short range order of its amorphous precursor in a non-biological system, although transition between different amorphous phases and the phenomenon of polyamorphism were studied before. 8,9 In our quest to emulate this biological strategy, we note the well-known feature of nanometric scale structures, namely that the influence of surface properties on the bulk is very significant. 10,11 It was shown, for example, that size influences the lattice parameters of crystalline nanoparticles 12,13 and changes the comparative stability of different phases. 14,15 This occurs due to the increasing influence of surface stress and surface energy as the particle size decreases. This influence does not require the particle to be crystalline, but should be present in any solid material, and is distinct from the phenomenon of varying structures on liquid surfaces, as in a solid material not only the surface energy plays an important role, but the surface stress does as well. 16 In the present study we investigated whether the short range order of an amorphous material can indeed be manipulated solely via size. We chose atomic layer deposition (ALD) 2 ACS Paragon Plus Environment

Page 2 of 23

Page 3 of 23

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


as our material deposition method, since it is a technique that can provide extremely precise, sub-nanometric, thickness control and can deposit conformal and pinhole-free amorphous films of various materials. 17,18 One of the most common processes performed using an ALD is the growth of thin, amorphous, Al2 O3 films, used for nanofabrication in a vast variety of applications. 19,20 Therefore, our material of choice in this study was amorphous alumina. An additional advantage of using Al2 O3 is that it has many metastable polymorphs at room temperature, 21 which may also suggest an ease of switching between different short range order states.

Results and Discussion Investigating the short range order of amorphous materials is more complex than studying the crystalline structure of crystalline materials. In this study we chose to use transmission electron microscopy (TEM) and in particular the electron energy loss spectroscopy (EELS) method, as well as X-ray photoelectron spectroscopy (XPS), to probe the short range order. We deposited various nanometric thicknesses of amorphous alumina directly onto ‘holey’ amorphous carbon and amorphous Si3 N4 membrane TEM grids. After deposition the samples were transferred directly into the TEM, and EELS was performed without any further sample manipulation or preparation. An example of such a film deposited on a holey amorphous carbon membrane can be seen in Figure 1. It is apparent that the film, as expected, is very conformal and void-free. The deposited films were amorphous, which could be verified by electron diffractions (Figure S1). EELS reference measurements were performed on sub-µm powders of crystalline α-Al2 O3 and γ-Al2 O3 , at the aluminum L2,3 -edge (Figure 2A). The resulting spectra conformed well with literature reports. 22,23 For γ-Al2 O3 the edge onset was at about 75.5 eV, and the main maximum at 78.6 eV. The corresponding values for α-Al2 O3 were 77.3 eV and 78.5 eV. It should be noticed that a pre-edge shoulder feature (∼77.3 eV) at the aluminum



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Al2 O3 film Amorphous carbon

Figure 1. Cross-section of coated TEM grid. This scanning electron microscope (SEM) micrograph of Al2 O3 film deposited on a holey carbon TEM grid shows the conformity of the film. The area seen in the image is an edge of one of the holes in the carbon film. L2,3 -edge can be observed only in the case of γ-Al2 O3 . This is because of the presence of 4-coordinated (tetrahedral) Al sites, 22,24 which are present only in the γ-Al2 O3 structure, and not in that of α-Al2 O3 , in which all the Al sites are rather 6-coordinated (octahedral). Similar measurements were performed on the ALD-deposited amorphous films of different thicknesses (Figure 2B). A clear trend could be observed in these measurements, namely that the prominent 4-coordinated Al feature (shoulder appearance at the Al L2,3 -edge, at ∼74 eV) is more pronounced in the case of thinner films, whereas for the thicker ones (above approximately 20 nm) the shoulder that appears in the spectrum is much weaker. This means that thicker amorphous films demonstrate, on average, a short range order in which fewer tetrahedral Al sites are present, while thinner amorphous films demonstrate a structure richer in these sites. This finding is striking, as the only difference between these films is their thickness (deposition parameters were kept constant for all samples). It should be noticed, however, that the O-K edge, that should have also been affected by these structural changes, could not be examined using this method because the films were deposited on oxygen containing substrates, which would have interfere with the oxygen edge 4 ACS Paragon Plus Environment

Page 4 of 23

Page 5 of 23

α-Al2O3 γ-Al2O3

4-coordinated Al

Intensity (normalized)

A

75

80

85

90 95 Energy (eV)

100

105

110

115

44.4 nm 17.5 nm 9.5 nm 4-coordinated Al

B Intensity (normalized, shifted)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


70

80

90

100 110 Energy (eV)

120

130

Figure 2. Comparative EELS measurements, Al L2,3 -edge. All spectra are background subtracted and are normalized by integrated intensity. The arrow indicates the shoulder feature caused by the presence of tetrahedral Al sites. (A) α and γ-Al2 O3 nano-powder references. (B) Measurements on films of three different thicknesses. The spectra are shifted slightly to emphasize the differences (hence “shifted” on the energy axis label). of the samples. To further verify the EELS measurements, we used the XPS technique. A reference measurement was performed on the same powders of crystalline α and γ alumina as those on which the EELS calibration measurements were performed. Figure 3A shows the Al2p



peaks acquired by this measurement. Peak broadening can be seen, as well as a slight shift towards lower energies in the case of γ-Al2 O3 (center at 74.35 eV, width at half maximum of 3.03 eV), relatively to the peak of α-Al2 O3 (center at 74.6 eV, width at half maximum of 2.26 eV), in good agreement with the EELS results (see Figure 2A). This is attributable to the presence in γ alumina of tetrahedral Al sites, which have lower binding energy than the octahedral Al sites. 25,26 A

α-Al2O3 γ-Al2O3

B ~74.85


~74.35

80

78

C

76

~531.9

74 72 Binding energy (eV)

~531.45

70

68

78

α-Al2O3 γ-Al2O3

~74.35 14.0 nm, 75° 14.0 nm, 25° ~74.45 2.37 nm, 25°

3λ

3λ


~74.6

25°

75°

77

76


73

72

71

14.0 nm, 75° 14.0 nm, 25°

D

~531.45


~531.75 Intensity (normalized)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 6 of 23

536

534


528

526

536

535

534

533 532 531 Binding energy (eV)

530

529

528

Figure 3. XPS measurements, Al2p and O1s peaks. The values near the peaks denote the position of the maximum. (A) Al2p spectra of crystalline Al2 O3 references. A shift to lower energies can be seen for γ-Al2 O3 , due to the presence of 4-coordinated Al. (B) Al2p spectra from two different photoelectron take-off angles of a 14 nm thick amorphous Al2 O3 film, and a 2.37 nm thick amorphous film. A noticeable shift to lower energies can be seen for the lower take-off angle and the thin sample. (C) O1s spectra of crystalline Al2 O3 references. A shift to lower energies can be seen for γ-Al2 O3 . (D) O1s spectra from two different photoelectron take-off angles of a 14 nm thick amorphous Al2 O3 film. A shift to lower energies is visible for the lower take-off angle. To investigate the size effect on amorphous thin films we had to take into consideration 6 ACS Paragon Plus Environment

Page 7 of 23

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


that in contrast to the EELS method, in which the signal originates from the entire sample thickness, in XPS the signal originates only from the top several nanometers below the surface. We therefore performed angle-resolved XPS (ARXPS), which enabled us to analyze the sample with depth resolution, by sampling photoelectrons from different takeoff angles. The sampling depth from which 95% of the signal intensity in ARXPS comes can be roughly estimated as 3λ sin θ, where λ is the attenuation length for photoelectrons in the sample, and θ is the takeoff angle of the photoelectrons. 27 Thus, the sampling depth increases when probing at higher takeoff angles (see inserts in Figure 3B). For this procedure we used thin amorphous alumina films similar to those used for the EELS measurements, but these films were deposited on silicon wafers, rather than on the TEM grids. We used two film thicknesses, 2.37 nm and 14 nm. Two takeoff angles were chosen, 25° and 75°, which, for alumina, correspond to sampling depths of roughly 3.6 nm and 8.1 nm, 28 respectively. We first investigated the Al2p peak of the 14 nm film at both takeoff angles. In Figure 3B, a shift to lower binding energies from the shallower takeoff angle (25°, ∼3.6 nm) is seen, suggesting that the less energetic, 4-coordinated Al sites are more abundant closer to the film surface. We then investigated the Al2p peak of the 2.37 nm film, at the 25° takeoff angle only, as at this angle the signal already originates from the entire film thickness. In both of the 25° measurements, on the thicker film as well as on the thinner one, the majority of the signal is derived from roughly the same sample thickness, and it appears that there is practically no shift in the binding energies between them. Unlike with the reference samples, no significant broadening was detected between the peaks. The reason for this might be that while the different crystalline polymorphs have a fixed fraction of each coordination site, in the amorphous samples this fraction changes gradually, with size. These observations, like the EELS measurements, indicate that the fraction of tetrahedral Al sites is greater near the surface of amorphous films and hence also in thinner amorphous films. It should be noticed that although the existence of an amorphous surface phase that



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

differs from the bulk was suggested in glasses before, 29 here we observed such a phenomenon consistently, on precisely thickness-controlled samples. We performed similar experiments, on the same samples, to investigate the oxygen binding energy, by probing the O1s peak. In Figures 3C and 3D, a similar trend can be observed for this peak. The signal from shallower depth is shifted slightly to lower binding energies, similarly to the shift of the O1s signal from γ-Al2 O3 , relatively to α-Al2 O3 . This provides further evidence that the short range order near the surface is more “γ-like” than “α-like”. The 2.37 nm thin film sample could not be examined for binding energy of oxygen sites, as the oxygen from the silica substrate would interfere. On the other hand, we made sure that no such interference was present in the 14 nm sample, by verifying that no Si peaks were visible in the spectra from either of the takeoff angles (Figure S3). Once it became clear that the short range order does indeed depend on the thickness of the amorphous film, and having noticed that some of these amorphous films crystallized after being exposed for some time (about 15 minutes) to the electron beam in the TEM, we wanted to further investigate how the short range order evolves, while still in the amorphous state, during exposure to the electron beam and just before the amorphous to crystalline transformation. It should be noticed that electron-beam induced crystallization of amorphous materials was used before in bio-inspired materials studies. 30 To this end, time-dependent EELS measurements were performed on amorphous alumina films of varying thicknesses (Figure 4). A constant area of each sample was exposed to a focused electron beam (with the same parameters between different samples) and EELS measurements on Al L2,3 -edge were performed with one minute intervals between them, using the same beam. The TEM calibration was achieved after several trial and error attempts: too concentrated beam caused damage to the samples, and a beam too dispersed caused extremely slow changes. In the conditions we worked, the typical values for the beam current and the dose were ∼2.7×10−8 A and 2×105 Am−2 , respectively. Additionally, we would like to emphasize


Page 8 of 23

Page 9 of 23

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


that once the desired parameters have been reached, no changes at all were made between the different compared samples. After measurements on two different zones of each sample, the following results were observed: for the thinnest sample (4.16 nm), Figure 4A clearly shows that fast rising and some smoothing of the initially extremely prominent shoulder feature occurred, but afterwards, this feature continued oscillating around the same point, and overall, there were hardly any changes in other features of the spectrum. Even after 20 minutes of exposure and measurements, the short range order did not stabilize on a specific state, but kept fluctuating back and forth. After the time-resolved measurement we switched to diffraction mode, and found that the sample remained in its amorphous state, and did not crystallize (see insert in Figure 4A). In another film, which was just ∼3 nm thicker (7.39 nm; Figure 4B), the behavior was different: smoothing of the shoulder feature was continuous, and simultaneously, another peak started appearing and increasing in intensity at approximately 84.7 eV. Interestingly, it can be seen that the spectrum continuously moves towards resembling that of the γAl2 O3 , until at some point it overlaps it perfectly. Here we want to stress the remarkable finding that the short range order evolved continuously towards that of the γ-Al2 O3 , even before crystallization took place. This behavior was observed in both of the zones that were measured. At the end of this measurement, we again switched to diffraction mode, which took several minutes while the beam was still focused on the same area, and noticed a very strange pattern (see insert in Figure 4B). No diffraction rings could be seen, however at one azimuthal angle a clustered scatter of points appeared. This is very surprising, as it implies that the sample crystallized under the beam at a highly preferred orientation. Even more surprising is the fact that this pattern has a wide range of d-spacings. At this point we cannot fully explain this phenomenon, but we can state that at some point we probably did, locally, crystallize the film. Nevertheless, it is obvious that this diffraction pattern does not represent γ-Al2 O3 clearly,



A γ-Al2O3 ref. 0 min. 1 min. 2 min. 3 min. 4 min. 5 min. 6 min. 7 min. 8 min. 9 min. 10 min. 11 min. 12 min. 13 min. 14 min.


4.16 nm

75

76

77 78 Energy (eV)

79

80

75

80

85 90 Energy (eV)

95

100

B γ-Al2O3 ref. 0 min. 1 min. 2 min. 3 min. 4 min. 5 min. 6 min. 7 min. 8 min.


7.39 nm

78 78.5 79 79.5 80 80.5 81 Energy (eV)

75

80

85 90 Energy (eV)

95

100

C α-Al2O3 ref. γ-Al2O3 ref. 0 min. 1 min. 2 min. 3 min. 4 min. 5 min. 6 min. 7 min. 8 min. 9 min. 10 min.

63.1 nm Intensity (normalized)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 23

75 75.5 76 76.5 77 77.5 78 78.5 Energy (eV)

75

80

85 90 95 Energy (eV)

100

105

110

Figure 4. Evolution of EELS spectra under electron beam, Al L2,3 -edge. The times specify the duration of exposure of samples to the electron beam, not including the measurement time. All spectra are presented as obtained, without any manipulation, apart from background subtraction. Only the γ-Al2 O3 spectrum is shifted between figures in order to be aligned with the time resolved spectra. The diffraction shown in each figure was taken after all the measurements were finished, from the area on which they were performed. (A) 4.16 nm thick film: only a slight change is observed, the shoulder feature emphasized in Figure 2 rises slightly, and then fluctuates around the same point. No crystallization occurs. (B) 7.39 nm thick film: after several minutes the spectrum starts to repeat almost perfectly the shape of the reference γ-Al2 O3 spectrum. The shoulder feature caused by tetrahedral Al sites is smoothed. Some crystallization occurs. (C) 63.1 nm thick film: on this film, some α-Al2 O3 –like features begin to appear, the shoulder feature is smoothed from both directions, and obvious crystallization is noticeable after the measurements.


Page 11 of 23

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


and it was taken after more than 16 minutes of beam exposure. Moreover, the high-resolution transmission electron microscopy (HRTEM) image of the same area after the measurement presented in figure S2B shows no sign of crystallinity. Finally, we similarly investigated a considerably thicker sample (63.1 nm). In this case we again noticed a time-resolved change in the short range order: the shoulder feature decreased considerably, and the spectrum exhibited some α-Al2 O3 –like features with time (Figure 4C). It should be pointed out, however, that in this case we did not get a perfect overlap between the final spectrum and any of the reference spectra. The final spectrum was close to that of γ-Al2 O3 , despite the appearance of few α-Al2 O3 –like features. In this case, switching to diffraction mode clearly revealed polycrystalline diffraction rings attributable to γ-Al2 O3 (Figure S4), but there was no sign of crystalline α-Al2 O3 . These results confirm previous publications about electron-beam induced crystallization of alumina 31 that claimed the formation of an “excited” amorphous state before the crystallization took place. However using ALD and EELS we can observe the precisely controlled thickness dependence of this phenomenon, with high sensitivity, and the continuing evolution of the short range order beyond that of γ-Al2 O3 towards α-Al2 O3 –resembling structure without any signs of α-Al2 O3 appearing in the electron diffraction (for thicker films). In order to obtain further details on the films’ crystallization behavior, measurements of crystallization temperature were performed. Similar measurements on ALD deposited films were done before, 32 but our measurements were continuous, direct, and in-situ. Instead of annealing films of different thickness at various temperatures and using X-ray diffraction afterwards to check for crystallinity, we made use of the ability of the ALD to coat very high aspect ratio surfaces and powdery substrates, to perform deposition on mesoporous silica gel, with surface area of 300 m2 /g. This allowed us to get up to 5:1 mass ratio between the film material and the substrate, thus enabling to obtain enough mass of film material to perform direct thermoanalytical measurements on it. In order to coat a substrate with such an extreme surface aspect ratio, and to get a good penetration into pores, the flow of



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

the carrier gas in the ALD was slowed down for one minute right after the pulse of each one of the precursors, followed by one minute purge time. This slowed the deposition process considerably, but allowed much better diffusion in and out of “hard to reach” places. The powders with different thickness of coating underwent differential scanning calorimetry (DSC) measurements after the deposition. As can be seen in Figure 5A, in Al2 O3 films, the crystallization temperature can be manipulated strongly by size, and it rises with the decrease in thickness. In order to compare this behavior of alumina to that of a different material, HfO2 films were grown using the same methods as before. HfO2 was chosen because this oxide can be grown easily with ALD, and is known to be amorphous upon deposition (unlike ZnO or TiO2 , for example). The HfO2 films were examined under TEM similarly to the alumina films described previously, but similar EELS measurements could not have been performed on them because they crystallized quickly under the beam. Nevertheless, the above described DSC measurements were performed on these HfO2 films (Figure 5B). It is evident that in HfO2 the influence of size on the crystallization temperature is much weaker. This can be explained by the smaller volume of the HfO2 unit cell. While 3

the volume of HfO2 unit cell is 138˚ A , the volume of α and γ Al2 O3 unit cells is 255 and ˚3 , respectively. Larger volume of unit cell means that the initial nucleus needed for 500A crystallization, which contains at least several unit cells, is less likely to appear in thinner films. This result fits nicely with the TEM results presented here, from which it is evident that thick films crystallize under the electron beam much more willingly than the thin ones.

Conclusions These results clearly show that the short range order in an amorphous material does change as a function of size. This is of great importance, among other reasons because the short range order in an amorphous material influences the crystalline structure succeeding after


Page 12 of 23

Page 13 of 23

-1.5

2.7 nm 4.3 nm 9.7 nm 13.9 nm Fitting

A

980

990

1000

1015.9°C 1020.4°C

-3

-3.5 970

1007.6°C

-2.5

997.6°C

Heat flow (µV)

-2

1010 1020 T (°C)

1030

1040

1050

1060

0.5

B 0

-0.5

-1

-1.5

-2

4.5 nm 13.7 nm 15.7 nm 20.5 nm Fitting

241.6°C

~233°C

Heat flow (µV)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


-2.5 150

200

250 T (°C)

300

350

400

Figure 5. Crystallization temperature of amorphous Al2 O3 and HfO2 films, as measured by DSC. The exothermic reaction of crystallization is represented by the upward peak in the heat flow. In thinner films, the peak becomes less prominent, simply because there is less film material. The fitting is to a Lorentzian profile. (A) Al2 O3 crystallization temperature is strongly size dependent. (B) Much weaker dependence is observed for HfO2 films. it’s crystallization. The EELS analysis coincides with the XPS data, which show that the surface of amorphous alumina possesses a different short range order than the average in it’s bulk, and the thinner, or smaller, the amorphous solid is, the more it’s short range order resembles that near the surface. This phenomenon is due to the stronger effect that the 13 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 14 of 23

surface exerts on the bulk as the size decreases, simply because the surface-affected areas take up a more significant fraction of the bulk. The surface structure we observed in our study agrees with what was predicted by molecular dynamics simulations. 33 This study demonstrates that the short range order in amorphous materials can indeed be altered via size effects, just as the lattice parameters of crystalline materials can be altered by them. We believe that these findings will open up new areas of research, as well as drive towards the ability to control the short range order of amorphous materials, where this is a cardinal issue that could not be previously addressed, due to the lack of possibility of such control.

Methods Atomic Layer Deposition (ALD): The ALD device used for these experiments was R-200 (Picosun, Finland). The growth process of Al2 O3 films utilized trimethylaluminum (TMA) and H2 O as precursors for deposition on Si wafers and TEM grids. O3 was used as an oxidizer instead of H2 O for deposition on silica gel.

For growth of HfO2 films,

Tetrakis(dimethylamido)hafnium and H2 O were used. N2 (99.999% pure) was used as the carrier gas in all processes. The thickness of the films was determined by ellipsometry, directly on the Si wafer sample or, for other substrates, on a piece of Si wafer that was placed close to the sample during the growth process for this purpose. Same growth parameters (precursors, growth temperature, pulse and purge times) were used to grow each batch of compared samples. Pulse time for all liquid precursors was 0.1 s., while for gaseous O3 it was 1 s. Purge time varied between 4 and 5 seconds for different batches of flat samples, and was 1 min. in slow flow conditions, followed by 1 min. of regular flow for porous samples (silica gel). The growth temperature was 80℃ for the TEM grid samples, 100℃ for the silica gel samples, and 120℃ for the XPS samples. Additionally, the silica gel samples were left in the ALD reaction chamber at ∼1 torr and ∼115℃ for at least


Page 15 of 23

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


2 hours, for dehydration, before the process began. The samples were prepared as close as possible to the measurements, in order to minimize their exposure to the environment. The TEM samples were exposed for less than two hours and the XPS samples were exposed for approximately 24 hours. Although repeating measurements in the TEM that took place more than 6 months after the sample preparation showed similar results.

Scanning Electron Microscopy (SEM): The image in Figure 1 was obtained with the Zeiss Ultra Plus High Resolution FEG-SEM (Zeiss, Germany), operated at 4.0 kV, using an in-lens secondary electrons (SE) detector.

Electron Energy Loss Spectroscopy (EELS): An FEI Titan 80–300 S/TEM (FEI, Eindhoven, The Netherlands) was operated at 300 KeV and was equipped with an image Cs corrector. A post-column energy filter (Tridiem 866 ERS, Gatan, USA) was used to obtain the EELS measurements. The measured energy spread of the beam was 0.6 eV, and a dispersion of 0.03 eV/channel was used. In order to avoid the influence of sample exchange on the spectra, several coated grids were cut and placed on a single sample holder for the comparative measurements.

X-ray Photoelectron Spectroscopy (XPS): XPS measurements were performed in UHV (2.5 × 10−10 Torr base pressure) using the 5600 Multi-Technique System (PHI, USA). Samples were irradiated with an Al Kα monochromated source (1486.6 eV) and the outcome electrons were analyzed by a spherical capacitor analyzer using a slit aperture of 0.8 mm. Sample charging during measurements was compensated by means of a neutralizer, with additional mathematical shift used when necessary (C1s at 285 eV was taken as an energy reference for all the measured peaks). The high resolution measurements presented in figure 3 were taken in a low energy range window with pass energy (PE) of 11.75 eV and 0.05 eV/step.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Differential Scanning Calorimetry (DSC). The DSC measurements were performed on Labsys 1600 (Setaram, France). Heating rate of 5℃/min. was used.

Acknowledgement The research that yielded these results received funding from the European Research Council under the European Union’s Seventh Framework Program (FP/2007–2013)/ERC Grant Agreement no [336077]. The authors thank Dr. L. Burstein from the Wolfson Applied Materials Research Center, Tel Aviv University, for the XPS measurements, Dr. I. Levin, Prof. W. Kaplan, Dr. O. Kreinin, and Dr. K. Jorissen for helpful discussions.


Page 16 of 23

Page 17 of 23

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Supporting Information Available A

C

B

t=4.16 nm

t=7.39 nm

t=63.1 nm

Figure S1. Initial electron diffractions from the three samples used in the timeresolved EELS measurements that are presented in Figure 4. The initial diffractions could not be taken from the same spots on which the time-resolved measurements were performed, as this would affect the results. Instead they were taken later, after all measurements were completed, from a nearby spot on the same samples. (A) 4.16 nm thick sample. (B) 7.39 nm thick sample. (C) 63.1 nm thick sample. It can be clearly seen that the as-deposited Al2 O3 on all samples is amorphous.

A

C

B

t=4.16 nm

t=7.39 nm

t=63.1 nm

Figure S2. HRTEM images taken after the time-dependent EELS measurements from the three samples that are presented in Figure 4. Images of each zone before the measurement could not have been taken, in order to avoid additional beam exposure. (A) 4.16 nm thick sample. No clue for crystallinity can be seen, as in the diffraction presented in Figure 4A. (B) 7.39 nm thick sample. Despite the diffraction presented in Figure 4B, no crystallinity can be seen in the image of the same zone, although the diffraction was taken from a larger area. (C) 63.1 nm thick sample. Some crystalline areas can be seen.


180

Page 18 of 23

2.37 nm, 25° 2.37 nm, 75° 14.0 nm, 25° 14.0 nm, 75°

160

140


Al2p

Si2p

Al2s


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Si2s


80

60

Figure S3. XPS survey scan in the 180–50 eV range on the samples used in Figure 3. The Si peaks can be clearly seen in the spectra of the thin (2.37 nm) sample from both takeoff angles, and are absent from the thick (14 nm) sample spectra. From this, we deduce that no signal is coming from the SiO2 /Al2 O3 interface of the thick sample, whereas it does come from the thin sample, and therefore the oxygen signal from the thin alumina film would be convoluted with the oxygen signal from the underlying silica.


Page 19 of 23

A

(440) (400) (311) γ−Al2O3

440 444

222

311

400

B Intensity (a.u.)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


(444)

1

1.5

2

2.5

3

d (Å)

Figure S4. Electron diffraction obtained from the 63.1 nm thick sample after the time-resolved measurements on it were completed. Several minutes after completing the EELS measurements, this diffraction was taken from the spot on which they were performed. (A) Miller indices corresponding to the crystalline planes of γ-Al2 O3 are presented. (B) X-ray diffraction (XRD) spectrum of γ-Al2 O3 powder, the same powder as the one used for EELS and XPS reference spectra (XRD was performed at the ID31 beamline, European Synchrotron Radiation Facility (ESRF), Grenoble, France). Gray vertical stripes represent the rings from the electron diffraction above. Good resemblance can be seen, both to the peak locations and to their intensities. A slight shift to smaller d-spaces is evident in the electron diffraction. This might be due to a size effect on the lattice parameters (there are probably size differences between the crystallites in the powder and on the sample), and/or due to inaccuracy in the microscope calibration that was used to calculate the d-spaces from the electron diffraction. The darker gray stripes correspond to brighter rings, and the lighter stripes – to the barely visible rings that appear on the diffraction image. This material is available free of charge via the Internet at http://pubs.acs.org/.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

References (1) Weiner, S.; Addadi, L. Annual Review of Materials Research 2011, 41, 21–40. (2) Aizenberg, J.; Addadi, L.; Weiner, S.; Lambert, G. Advanced Materials 1996, 8, 222– 226. (3) Addadi, L.; Raz, S.; Weiner, S. Advanced Materials 2003, 15, 959–970. (4) Gower, L. B. Chemical reviews 2008, 108, 4551–4627. (5) Drabold, D. A. The European Physical Journal B 2009, 68, 1–21. (6) Zolotoyabko, E. Basic Concepts of Crystallography; Wiley-VCH: Weinheim, Germany, 2011. (7) Braga, D., Grepioni, F., Orpen, A. G., Eds. Crystal Engineering: From Molecules and Crystals to Materials; Springer, 1999. (8) Poole, P. H.; Grande, T.; Angell, C. A.; McMillan, P. F. Science 1997, 275, 322–323. (9) Franzese, G.; Malescio, G.; Skibinsky, A.; Buldyrev, S. V.; Stanley, H. Nature 2001, 409, 692–695. (10) Berry, R. Nature 1998, 393, 212–213. (11) Navrotsky, A. ChemPhysChem 2011, 12, 2207–2215. (12) Mays, C.; Vermaak, J.; Kuhlmann-Wilsdorf, D. Surface Science 1968, 12, 134–140. (13) Solliard, C.; Flueli, M. Surface Science 1985, 156, Part 1, 487–494. (14) McHale, J. M.; Auroux, A.; Perrotta, A. J.; Navrotsky, A. Science 1997, 277, 788–791. (15) Garvie, R. C. The Journal of Physical Chemistry 1978, 82, 218–224.


Page 20 of 23

Page 21 of 23

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


(16) Fischer, F.; Waitz, T.; Vollath, D.; Simha, N. Progress in Materials Science 2008, 53, 481–527. (17) George, S. M. Chemical Reviews 2010, 110, 111–131, PMID: 19947596. (18) Leskel, M.; Ritala, M. Thin Solid Films 2002, 409, 138 – 146, Proceedings of the 2nd Asian Conference on Chemical Vapour Depostion. (19) Puurunen, R. L. Journal of Applied Physics 2005, 97, –. (20) Chen, P.; Mitsui, T.; Farmer, D. B.; Golovchenko, J.; Gordon, R. G.; Branton, D. Nano Letters 2004, 4, 1333–1337. (21) Levin, I.; Brandon, D. Journal of the American Ceramic Society 1998, 81, 1995–2012. (22) Levin, I.; Berner, A.; Scheu, C.; Muellejans, H.; Brandon, D. In Modern Developments and Applications in Microbeam Analysis; Love, G., Nicholson, W., Armigliato, A., Eds.; Mikrochimica Acta Supplement; Springer Vienna, 1998; Vol. 15; pp 93–96. (23) Edlmayr, V.; Harzer, T. P.; Hoffmann, R.; Kiener, D.; Scheu, C.; Mitterer, C. Journal of Vacuum Science & Technology A: Vacuum, Surfaces, and Films 2011, 29, 041506– 041506–8. (24) Brydson, R. Journal of Physics D: Applied Physics 1996, 29, 1699. (25) González-Elipe, A. R.; Espinós, J. P.; Munuera, G.; Sanz, J.; Serratosa, J. M. The Journal of Physical Chemistry 1988, 92, 3471–3476. (26) Ebina, T.; Iwasaki, T.; Chatterjee, A.; Katagiri, M.; Stucky, G. D. The Journal of Physical Chemistry B 1997, 101, 1125–1129. (27) Fadley, C. S. Progress in Surface Science 1984, 16, 275–388. (28) Powell, C. J.; Jablonski, A. NIST Electron Effective-Attenuation-Length Database; National Institute of Standards and Technology: Gaithersburg, MD, 2011; Version 1.3. 21 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

(29) Avramov, I.; Höche, T.; Henderson, G. S. Journal of Non-Crystalline Solids 2008, 354, 4681 – 4684. (30) Politi, Y.; Arad, T.; Klein, E.; Weiner, S.; Addadi, L. Science 2004, 306, 1161–1164. (31) Nakamura, R.; Ishimaru, M.; Yasuda, H.; Nakajima, H. Journal of Applied Physics 2013, 113, 064312. (32) Jakschik, S.; Schroeder, U.; Hecht, T.; Gutsche, M.; Seidl, H.; Bartha, J. W. Thin Solid Films 2003, 425, 216–220. (33) Adiga, S. P.; Zapol, P.; Curtiss, L. A. Phys. Rev. B 2006, 74, 064204.


Page 22 of 23

Page 23 of 23

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Graphical TOC Entry

In this research we show that the short range order and the crystallization behavior of amorphous alumina is affected by size. Moreover, the evolution of short range order during crystallization is also size-dependent.


Size Effect on the Short Range Order and the Crystallization of

Recommend Documents