Density of Nanometrically Thin Amorphous Films Varies by Thickness

May 17, 2017 - †Department of Materials Science and Engineering and ‡Russell Berrie Nanotechnology Institute, Technion − Israel Institute of Tec...
0 downloads 7 Views 3MB Size
Article pubs.acs.org/cm

Density of Nanometrically Thin Amorphous Films Varies by Thickness Yael Etinger-Geller,† Alex Katsman,† and Boaz Pokroy*,†,‡ †

Department of Materials Science and Engineering and ‡Russell Berrie Nanotechnology Institute, Technion − Israel Institute of Technology, 3200003 Haifa, Israel S Supporting Information *

ABSTRACT: Organisms in nature can alter the short-range order of an amorphous precursor phase, thereby controlling the resulting crystalline structure. This phenomenon inspired an investigation of the effect of modifying the short-range order within the amorphous phase of a selected material. Amorphous thin films of aluminum oxide deposited by atomic layer deposition method were found to vary structurally as a function of size. Thinner films, as predicted and also confirmed by atomistic simulations, exhibited more 4coordinated Al sites. These atomistic alterations were expected to change the amorphous thin film’s average density. The density indeed varied with the alumina layer thickness, and the measured effect was even stronger than predicted theoretically. This effect is explained in terms of the deposition process, where each newly deposited layer is a new surface layer that “remembers” its structure, which results in thin films of substantially lower density.



amorphous phases.9 Moreover, in many cases, a specific crystalline polymorph is required for a particular function,10 and indeed, many studies were conducted over the years to achieve controlled crystallization.11−14 We chose to utilize size effects to alter the atomistic structure in the amorphous state.15 Size effects have been extensively studied in crystalline materials and found to change various properties, both structural16,17 and functional,18,19 in nanosized materials. These effects, which arise from highly pronounced surface stress and energy, are not restricted to crystalline materials but can exist in any solid substance. Furthermore, amorphous materials, like crystalline ones, can exist in different solid structures,20 implying the possibility of switching between them. We chose atomic layer deposition (ALD) as our sample preparation technique, as this is an excellent method of producing high-quality, conformal, and pinhole-free thin films of different systems.21−25 In the quest to control the short-range order within an amorphous phase, our group recently performed a breakthrough study in which size effects were found to alter the short-range order in amorphous Al2O3 nanofilms.15 The results indicated that the near-surface layer of an amorphous Al2O3 nanofilm is characterized by a short-range order that is richer in 4-coordinated Al sites (Al4 sites)26 than that of the bulk structure. Thus, the thinner the amorphous film, the more its short-range order resembles that near the surface. These results are also supported by atomistic simulations.27 We therefore

INTRODUCTION Amorphous materials are important for a number of different applications in science and technology1 owing to their unique electronic, optical, and mechanical properties.2 Although such materials are in common use, scientists have only recently begun to explore their extraordinary structures.3 While crystalline materials are characterized by a periodic and predictable atomic arrangement, in amorphous materials, the order decays rapidly with the distance.3,4 It is nevertheless possible to describe the structure of amorphous materials in terms of short-range order (coordination number, nearest neighbors, bond length, bond angles), which is similar around atoms of the same kind, and has a typical bond distance of up to 2−3 atoms.3,5 Fine changes in the atomistic structure can lead to new, fascinating phenomena, most of which are not yet known. The inspiration for this study comes from nature, where amorphous phases characteristically have various advantages and play significant roles.6 One such characteristic is the ability to serve as a transient precursor phase for controlled mineralization into a specific crystalline structure, even if this is not the thermodynamically preferred one.7,8 This is achieved by controlling the short-range order in the amorphous precursor so that it resembles that of the desired crystalline polymorph.7 Such control is achieved in nature via different additives, such as polymeric molecules or magnesium ions, which become incorporated into the crystal and induce precipitation of a specific phase.6−8 Finding a way to emulate this manipulative technique synthetically would have a profound impact, as many technological applications utilize different characteristics of © 2017 American Chemical Society

Received: March 21, 2017 Revised: May 17, 2017 Published: May 17, 2017 4912

DOI: 10.1021/acs.chemmater.7b01139 Chem. Mater. 2017, 29, 4912−4919

Article

Chemistry of Materials

Figure 1. 27Al nuclei investigation using SS-NMR: (a) spectra of two abundant polymorphs of Al2O3; (b) spectra of thin Al2O3 films, deposited into mesoporous silica. Structure, Thickness, and Density Characterization. To study the structural aspects of the amorphous films, solid-state nuclear magnetic resonance33 (SS-NMR) measurements were performed, with a magnetic field of 9.4 [T] (Varian VNMRS, Varian, USA). These measurements were carried out in the Chemistry Department at Durham University, UK. By comparing the results to spectra of known Al2O3 polymorphs, it is possible to determine the structural species, which exist in the amorphous film. High-angle annular dark field scanning transmission electron microscopy (HAADF-STEM) technique was used to study the quality of the deposited films and to exclude other possible influences over the density. The thickness of the films was measured by direct spectroscopic ellipsometry (VASE, Woollam, USA) on the Si wafers. To examine the variations in density, we chose the X-ray reflectometry (XRR) technique, performed using the X-ray diffractometer (SmartLab, Rigaku, Japan). XRR is a surface-sensitive characterization method suitable for the study of thin films and allows the measurement of the thickness, roughness, and the density of a thin film.34 The electronic density is proportional to the square root of the critical angle,10 from which the mass density35 of a thin layer can be calculated. To directly evaluate the absolute values of densities of the deposited films, Rutherford backscattering spectroscopy31 (RBS) was used with a 2.024 [MeV] 4He+ ± 1 [KeV] beam (Pelletron accelerator, NEC, USA). These measurements were performed in the Institute for Nanotechnology and Advanced Materials, Bar Ilan University, Ramat Gan, Israel.

expect that the dependence of the short-range order on size in amorphous Al2O3 will yield, as in crystalline materials, different size-dependent physical properties. The density and densification of ALD films are intriguing and interesting topics, which have been widely studied both theoretically28,29 and experimentally.30,31 While the density was found to vary as a function of the deposition temperature,31 the relation between the density and the film thickness has not been studied up to now. In this study, we investigated the size dependence of density in amorphous Al2O3 nanofilms. We discuss the experimental results in the framework of the developed model and take into account surface reconstruction driven by a decrease in surface energy.



EXPERIMENTAL SECTION

Sample Preparation. Nanometrically thin films of Al2O3 were deposited by the ALD technique utilizing a plasma enhanced atomic layer deposition (PEALD) reactor (ALD R 200 Advanced tools, Picosun, Finland). Despite the plasma assistance capabilities, thermal ALD was utilized in this study. Deposition was performed on the surfaces of p-type Si (100) wafers (UniversityWafer Inc., USA), on low roughness, polished LiF (100) substrates (MTI Corporation, USA), and directly onto holey carbon grids (SPI supplies, USA). The selected working temperatures were 200 and 350 °C, as this process has a wide ALD window32 allowing conformal and precise deposition of amorphous alumina films, at various temperatures. Silicon wafers were rinsed with ethanol and dried under a N2 gas stream prior to the deposition processes. Polished LiF substrates and holey carbon grids were used “as received”. Trimethylaluminum (TMA) was used as the precursor and H2O as the oxidizer. The ALD system was operated under a continuous flow of N2 carrier gas (99.999% pure). A basic ALD cycle consisted of 0.1 s TMA pulse (room temperature), 6 s N2 purge, 0.1 s H2O pulse (room temperature), and 6 s N2 purge. One ALD cycle yields approximately one monolayer of substance,32 and by repetition of the basic cycle, the thickness of the film can be strictly controlled. The same conditions were used for the growth of all samples. An additional set of ALD Al2O3 films was obtained using TMA and O3 as the oxidizer. These conditions were used to grow Al2O3 thin films onto mesoporous silica, similar to those investigated by Bloch et al.15 These processes were done under “stop flow” conditions. In each process, an additional control small Si wafer was placed to monitor the thickness of the deposited layer.



RESULTS To study the size effect on the structure and density of the amorphous Al2O3, we used ALD to deposit very thin nanofilms of Al2O3 of varied thicknesses, having in mind that the size effect is expected to be more pronounced in the thinner films. Using this technique, we achieved good linear growth with excellent control over the thickness, and a growth rate of approximately 1 Å per cycle at both working temperatures, 200 and 350 °C. The short-range order in the amorphous Al2O3 nanofilms has been found to vary with size15 by using electron energy loss spectroscopy (EELS) in the transmission electron microscope (TEM). Nevertheless, the use of TEM might be problematic since an intense electron beam that goes through the sample might change its structure over time. Here, we chose to use another technique namely SS-NMR to characterize the shortrange order within the films. This method is not expected to 4913

DOI: 10.1021/acs.chemmater.7b01139 Chem. Mater. 2017, 29, 4912−4919

Article

Chemistry of Materials

Figure 2. (a) XRR measurements of amorphous Al2O3 samples of three different thicknesses; in circle, the critical angles’ range. (b) Shift in the critical angle with size; the arrows indicate the critical angle of each sample.

performed slow high-resolution scans on samples with different thicknesses (scan speed −0.05°/min, scan step −5.2 × 10−3°). As seen in Figure 3, the critical angle depends significantly on the thickness of the amorphous film. The ALD deposition

cause any structural changes during the measurements. First, two known polymorphs of Al2O3 were studied: α-Al2O3, which exhibits 6-coordinated Al sites solely, and γ-Al2O3, in which there are both 4-coordinated and 6-coordinated Al sites (Figure 1a). Next, three samples of ALD deposited Al2O3 films into mesoporous silica with different thicknesses were studied via the 27Al-nuclei (Figure 1b). As can be seen in Figure 1 by comparing the spectra of the two polymorphs of alumina to the spectra of the ALD deposited alumina, thinner films exhibit a higher fraction of 4coordinated Al sites. This corroborates previous findings of the surface effect on the short-range order in amorphous Al2O3 nanometric films by this additional technique.15 To study the variation in properties of the amorphous films, samples of three different thicknesses were scanned by X-ray reflectivity (XRR) (Figure 2a). An XRR spectrum is characterized by its periodic pattern, from which the film’s thickness, density, and roughness can be extracted. The thickness of the film is proportional to the frequency of the intensity fluctuations,35 and the spectra obtained from the thicker samples are indeed seen to be characterized by higher fluctuation frequency. In addition, the extended spectra are indicative of the high quality and low roughness of the deposited films. Another parameter that can be extracted from the XRR spectra is the average film density. The density, ρ, is determined by measuring the critical angle θc, which is proportional to the square root of the density, according to the following expression:36 θc =

⎛ r λ2 ⎞ ⎜ e ⎟N0ρ ∑ xi(Zi + f i′ )/∑ xiMi ⎝ π ⎠ i i

Figure 3. Dependence of the critical angle on the film thickness; thin films deposited at 200 °C (black squares) and at 350 °C (red dots).

temperature has little influence on the critical angle (and on the density) of very thin films (below ∼20 nm), but the “saturation” of the critical angle with increasing thickness is achieved faster at a higher working temperature. The change in density of the films can be found from the critical angle change using eq 1 as follows: (ρ0 − ρ)

(1)

ρ0

where re is the classical radius of an electron, N0 is the Avogadro number, λ is the X-ray wavelength, Zi, Mi, xi, and f ′i are the atomic number, atomic weight, atomic ratio, and atomic scattering factor of the ith atom, respectively. The critical angle is defined as the angle at which, owing to refraction of the X-ray beams, the reflected intensity starts to decrease.35 It can be seen (Figure 2b) that the critical angle shows a shift toward higher angles for the thicker samples, indicative of their higher density. To examine the variation in critical angle, we

⎛ θ ⎞2 = 1 − ⎜⎜ c ⎟⎟ ⎝ θc,0 ⎠

(2)

where θc,0 is the critical angle for the thickest film, which also has the largest density, ρ0. Additional thermal treatments of the low-density films at various temperatures ranging from 300−750 °C had no measurable impact on the critical angle. The used ALD working temperatures did not influence the appearance of lowdensity very thin films, while the density “saturation” rate increased with increasing deposition temperature. These 4914

DOI: 10.1021/acs.chemmater.7b01139 Chem. Mater. 2017, 29, 4912−4919

Article

Chemistry of Materials

Figure 4. (a) Mass density of Al2O3 films, as dependent on the thickness of the films (achieved by RBS) and the Θc2 of Al2O3 films, as dependent on the thickness of the films (achieved by XRR). (b) Relation between the Θc2 and the mass density.

findings suggest that metastable low-density layers are formed in the amorphous alumina films of relatively larger thicknesses (15−20 nm). Those layers are probably a result of the combined effect of the surface layer reconstruction and the kinetics of the ALD process, as discussed further. For further investigation of the ALD amorphous films’ density, we utilized the RBS technique. RBS is a characterization method that can provide atomic concentration and areal density of the elements within a thin film, from which it is possible to calculate the mass density.37 Here, a low Z substrate was needed to get a good signal from the thin layers; hence, the thin films were deposited onto single crystalline LiF substrates. The new substrate did not reduce the quality of the ALD process, and smooth films were deposited with a growth rate of approximately 1 Å per deposition cycle, identical to that on the Si wafer. According to the obtained RBS results (see Figure 1S), the O/Al ratio changed from 1.79 ± 0.018 for 20 nm-thick films to 1.52 ± 0.02 for 42.5 nm-thick films, but these changes, according to calculations, may cause only ∼2% change in the density and by no means can it explain the effect observed. By modeling the achieved RBS spectra, we have obtained the areal densities and mass densities of the samples (Figure 4a). The critical angles for these films were also obtained by the XRR technique (Figure 4a) to compare the densities of the films deposited on Si and LiF substrates. As can be seen from Figure 4a, the density of the film indeed depends on the thickness and rises from ∼2.95gr/cm3 for the 20 nm-thick film to ∼3.22gr/cm3 for the 42.5 nm-thick film. These values coincide well with the known densities of amorphous Al2O3 films.31,38 The proportionality coefficient between the density and the square critical angle (evaluated from Figure 4b) was found to be 63.41 ± 0.44. By substituting this value, one can estimate the absolute density values for the films deposited on the Si wafer (Figure 5). Here, it can be seen again that the values coincide well with known density values for amorphous Al2O3. It is important to note that the thickness range examined in the RBS experiments was smaller than that in the XRR experiments, since the accuracy of the results is higher for thicker films, due to stronger experimental signal.

Figure 5. Calculated density values for amorphous Al2O3 films of various thicknesses, deposited on Si at two different temperatures, compared to the values achieved with RBS on LiF.

To rule out other influences on the films’ density, such as nanovoids or nanopores, the films were studied using the HAADF-STEM technique. This method is highly sensitive to zcontrast (Figure 6). As we showed, thinner films have lower density so if porosity indeed affects the density of these films, we would expect to see higher pores portion in the thinner ones. Nevertheless, it can be seen that the films are of high quality, uniform, conformal, and pinhole free.



DISCUSSION AND THEORETICAL MODELING The size dependence of the density of very thin films can be caused by a thermodynamic driving force provided by an excess in the total free energy owing to the presence of two interfaces. The energy of unrelaxed external surface of γ-alumina can be quite large (γunrel ≈ 10 J/m2), and it decreases substantially during relaxation (γrel ≈ 1 J/m2).39 This decrease is mainly achieved by formation of new bonds formed between undercoordinated surface atoms. It is accompanied by substantial reconstruction of the near-surface layer that resulted in significant lattice distortion, and even in its amorphization, as 4915

DOI: 10.1021/acs.chemmater.7b01139 Chem. Mater. 2017, 29, 4912−4919

Article

Chemistry of Materials

Figure 6. HAADF-STEM measurements of a 20 nm thick amorphous Al2O3 layer deposited on a holey carbon TEM grid. Left, low magnification; right, higher magnification.

was shown, in particular, by computer simulations.40,41 An additional energy stored in the distorted layer is a part of the surface energy gain. The internal energy increase can be realized through the decrease in the average interatomic interaction energy. The reduced interatomic bonds may correspond to the increased or decreased interatomic distances. As follows from the surface reconstruction analysis below, Al atoms passing to the oxygen layer, to bond with the nonbridging oxygen atoms, cause additional repulsive Al−Al forces, and this may result in increase of the equilibrium distance between the oxygen layers. To estimate maximum possible thermodynamic effect of the surface energy relaxation on the film density, let us assume that the all surface energy gain is consumed by redistributed internal atoms and all increase of the bulk potential energy is realized through increase of the Al−O interatomic distances. By considering the change in the average interatomic interaction energy during surface relaxation, the film’s density change can be approximated by the following expression (see Supporting Information Note I): ρ − ρ0 ρ0

2 4.67γsd111 3ΔE ≈ ∞ ≃ ∞ V O − AlN V O − AlN

reconstruction, cannot explain alone the experimentally observed decrease of the film density. The most superficial layer can be viewed as an outcome of the dynamic reconstruction of several outer atomic layers during continuous deposition of alumina by the ALD method. To illustrate such possible reconstruction, let us consider several near-surface parallel (111) atomic layers in an ideal γalumina structure, two of which, A and C, contain only oxygen atoms, and the layers of aluminum atoms, B and D, are split for three sublayers, Bi, Di, i = 1, 2, 3, B2, and D2 consisting of Al6 atoms at a distance d111/2 from the oxygen layers, and intermediate layers B1, B3, D1 and D3 containing Al4 atoms and located at a distance d111/4 from the nearest oxygen layer, excluding the outer (terminating) B1 layer, as shown schematically in Figure 7.

(3)

where V∞ O−Al is the O−Al interatomic potential in the bulk, ΔE = 6γs/5nOx s is proportional to the alumina surface energy γs, N is the number of close packed atomic oxygen monolayers with the atomic surface density nOx s , and d111 is the distance between (111) oxygen planes. 2 Assuming reasonable values V∞ O−Al = −8.3 eV, γs = 9.45 J/m , and d111 = 0.224 nm, yield ρ − ρ0 ρ0

≈−

1.7 N

(4)

Figure 7. Schematic structure of the near-surface atomic layers in alumina. (a) Unrelaxed ideal structure of γ−Al2O3; (b) Reconstructed structure.

Therefore, the relative change in the average density is expected to be inversely proportional to the number of atomic monolayers, which is, of course, directly related to the thickness of the film. However, as follows from eq 4, substantial decrease in the density of about 20−25% can be achieved only for very thin layers with N ≤ 8, which correspond to thicknesses ≤2 nm. (The resulting decrease in density would be even smaller if only one external surface is reconstructed.) At the same time, the 20−25% decrease in the density was observed in the ALD films of 15−20 nm thickness (Figure 3). Therefore, a thermodynamic driving force, which may cause redistribution of potential energy between the interfaces and the bulk of the film during

Every oxygen atom in the close-packed layer (A or C) has three nearest-neighbor oxygen atoms in the second oxygen layer. Every Al atom in the octahedral position has three neighbor oxygen atoms in every oxygen layer, while an Al atom in the tetrahedral position has three neighbor oxygen atoms in the nearest (111) oxygen layer and one neighbor in the next one. Every Al4 atom thus has, on average, two oxygen neighbors in each layer. The distance between the oxygen layers corresponds to a zero total interaction force acting on every 4916

DOI: 10.1021/acs.chemmater.7b01139 Chem. Mater. 2017, 29, 4912−4919

Article

Chemistry of Materials

Figure 8. Possible change of (a) lattice parameter and (b) density as a function of Al atom fraction, z, which get incorporated into the oxygen layer A; z1 = zp1, z2 = zp2, p1 and p2 are the probabilities to find a nearest Al atom from the A layer in an octahedral or a tetrahedral position in the D2 and D1 layers, respectively.

oxygen layer in the direction perpendicular to it. In approximation of nearest neighbor pair interaction, this force can be expressed as follows: F1/nsOx =

1 ∂VAl − O x 2 ∂r

[3(n111 ⃗ n⃗ tet) + 1] + 3y rtet

(n111 ⃗ n⃗ oct) + (3/2)

∂VO − O ∂r

∂VAl − O ∂r

x

∂VAl − O ∂r

+ rtet

3y

∂VAl − O ∂r

roct

(5)

rO − O

roct

9 ∂VO − O 6 ∂r

3y

rtet

∂VAl − Al ∂r

∂VAl − O ∂r + d111/4

+ yzp1 roct

∂VAl − Al ∂r

9 ∂VO − O 6 ∂r

d111/2

=0 rO − O

(7)

where p1 and p2 are the probabilities to find a nearest Al atom from the A layer in an octahedral or a tetrahedral position in the D2 and D1 layers, respectively. With passing of aluminum atoms from the B to A layer, the equilibrium distance between the oxygen layers will increase due to the additional repulsive Al−Al forces. This result correlates with our above assumption about the elongation of the interatomic bonds during interface reconstruction. The results of the numerical solution of eq 7 for different values z1 = zp1 and z2 = zp2 are presented in Figure 8. As can be seen, the lattice parameter increases and the effective density decreases as Al atoms fraction passing to the outer oxygen layer increases. The relative changes in density may reach magnitudes of about 0.3 and larger for Z1 = 0.05−0.1 and Z2 = 0−0.05. The transfer of Al atoms and their embedding in the outer oxygen layer should trigger subsequent reconstruction of the A as well as of the B and C layers. Such reconstruction might result in destruction of the crystal structure of the layers, a process that can be viewed as amorphization of the near-surface region. Computer simulations39 with modified Born-MayerHuggins potentials have shown that the reconstruction of (111) surfaces of γ-alumina is extensive and that the near-surface regions become amorphous, or at least significantly distorted, at a depth of about 0.7 nm. Amorphization of the near-surface layer structure can be accompanied by a change in the average coordination number of Al atoms. A molecular dynamics study of amorphous alumina41 showed that the change in density from 3.0 to 3.3 gr/ cm3 results in a substantial change of short-range order. Moreover, experimental investigation of the size effect on the short-range order in a nanosized amorphous alumina15 confirmed that the fraction of tetrahedral Al sites is greater in thinner amorphous films. The pronounced effects on the presence of Al4 atoms were observed in films less than 20 nm

3(n111 ⃗ n⃗O − O) = 0

+

+

+ xzp2

where x and y are the fractions of Al4 and Al6 atoms, correspondingly (x = 0.3, y = 0.7 for ideal γ-alumina); n⃗tet and n⃗oct are unit vectors from tetrahedral and octahedral voids, respectively, to the nearest oxygen atom in the nearest oxygen layer; n⃗O−O is the unit vector in the direction between two nearest oxygen atoms in the neighbor layers; n⃗111 is the unit vector in the [111] direction; the coefficient (3/2) in the last term takes into account the ratio of oxygen and aluminum atoms. Using the values for the fcc γ-alumina (n⃗111n⃗tet) = 1/3, (n⃗111n⃗oct) = 1/√3, (n⃗111n⃗O−O) = 2/3 , rtet = a 3/4 , roct = a/2, rO−O = a/√2, and d111 = a/√3, one can rewrite eq 5: x

∂VAl − O ∂r

=0 rO − O

(6)

Solving eq 6 by substituting the interatomic potentials40 values into the equation, yields the lattice parameter a = 0.382 nm, which seems a rather good approximation of the real γalumina structure (aγ = 0.388 nm). During reconstruction of the surface, Al atoms may pass from the outer aluminum layers Bi to the oxygen layer A to bond with the nonbridging oxygen atoms, and will probably take the triangular void positions in the close-packed oxygen layer to form a new relaxed outer layer A’, as it has been shown by computer simulation.39 This process is dictated by a reduction in total energy of the near-surface structure and therefore is thermodynamically driven. The change in the interlayer distance depends on the fraction of Al atoms (z) passing to the oxygen layer A′ layer. The interlayer equilibrium distance is a function of z and can be estimated by using the following equation: 4917

DOI: 10.1021/acs.chemmater.7b01139 Chem. Mater. 2017, 29, 4912−4919

Chemistry of Materials



thick. However, the intriguing question remains: how can the reconstruction of a very narrow near-surface layer (only ∼0.7− 1.0 nm thick) result in a low-density layer of 15−20 nm? The main feature of the ALD process is the following: each newly deposited atomic layer in turn becomes an outer surface layer. Accordingly, several formerly outer atomic layers will now constitute a near-surface layer with an amorphous structure characterized by a higher fraction of Al4 sites and by lower density. This layer, although by now an inner layer, “remembers” its surface structure for a comparatively long time. The formation of relatively thick and stable low-density layers (15−20 nm) may lead to the occurrence of a new metastable amorphous phase of low density with an enlarged fraction of Al4 sites. Reconstruction and densification of this layer into a regular amorphous phase might be restrained for various reasons, in particular owing to a high activation-energy barrier to further reconstruction. An increase in density where the thickness is more than ∼15 nm signifies such phase transformation in the depth of the film. The dependence of the density “saturation” rate on the ALD working temperature exemplifies a kinetic factor that determines the friable/dense amorphous/amorphous phase transformation. The densities of Al2O3 films as low as (2.5 − 3.0) gr/cm3 have been observed in the low temperature ALD alumina films at growth temperatures (33 − 177)°C by Groner et al.,31 where it has been found that the lower deposition temperature the lower film density. This observation confirms the kinetic influence on the density decrease. Therefore, combination of thermodynamically driven nearsurface-layer reconstruction and kinetics of the alumina film growth during ALD process results in formation of comparatively thick low density amorphous alumina films with enlarged fractions of Al4 atoms.

Article

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.chemmater.7b01139. RBS graphs of the studied samples: experimental and simulation spectra; note on change in density, as driven by decrease of external surface energy during nearsurface-layer reconstruction (PDF)



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. ORCID

Boaz Pokroy: 0000-0003-0480-7250 Author Contributions

The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS The research leading to these results has received funding from the European Research Council under the European Union’s Seventh Framework Program (FP/2013-2018)/ERC Grant Agreement No. 336077. We are grateful to Dr. Oleg Kreinin for helping with preparing the samples and operating the ALD system. We are also grateful to Dr. Yaron Kauffmann for helping with TEM experiments and Dr. Olga Girshevitz from the Bar Ilan University for performing the RBS experiments.





ABBREVIATIONS ALD, atomic layer deposition; SS-NMR, solid-state nuclear magnetic resonance; STEM, scanning transmission electron microscopy; XRR, X-ray reflectometry; RBS, Rutherford backscattering spectroscopy

CONCLUSIONS The near-surface-layer reconstruction driven by the decrease in surface energy, combined with the ALD kinetic process in which every newly deposited atomic layer appearsalbeit transientlyas the most superficial layer, results in the formation of a metastable, low-density, amorphous alumina structure. Thus, the density of ALD-produced thin amorphous alumina films remains significantly lower than that of bulk amorphous alumina with thicknesses of up to 40−50 nm and enlarged fractions of Al4 atoms. This effect does not arise from structural imperfections, such as nanovoids or nanopores. The average short-range order parameter, being a function of the film density, conforms to the thickness of the film. The lower density thus corresponds to a smaller coordination number of aluminum atoms. We believe that the reported effect is significant for both science and industry since different physical properties, such as dielectric properties42 and refractive index,30 are dependent on the density. ALD is a method, which is widely employed in various technological applications, ranging from the semiconductor industry, to optical coatings and even to wearresistant materials.43 Our findings prove that the density is strongly dependent on size; thus, we expect that other physical properties would also be size-dependent. The ability to tune one property or another by size, according to a specific requirement, can open new possibilities for materials selections and applications. This will be the focus of our future study.



REFERENCES

(1) Inoue, A.; Hashimoto, K. Amorphous and Nanocrystalline Materials: Preparation, Properties, and Applications; Springer Science & Business Media, 2013; Vol. 3. (2) Stachurski, Z. H. On structure and properties of amorphous materials. Materials 2011, 4, 1564−1598. (3) Drabold, D. Topics in the theory of amorphous materials. Eur. Phys. J. B 2009, 68, 1−21. (4) Zallen, R. The Physics of Amorphous Solids; Wiley, 1998. (5) Hufnagel, T. C. Amorphous materials: Finding order in disorder. Nat. Mater. 2004, 3, 666−667. (6) Gower, L. B. Biomimetic model systems for investigating the amorphous precursor pathway and its role in biomineralization. Chem. Rev. 2008, 108, 4551−4627. (7) Addadi, L.; Raz, S.; Weiner, S. Taking advantage of disorder: amorphous calcium carbonate and its roles in biomineralization. Adv. Mater. 2003, 15, 959−970. (8) Aizenberg, J.; Addadi, L.; Weiner, S.; Lambert, G. Stabilization of amorphous calcium carbonate by specialized macromolecules in biological and synthetic precipitates. Adv. Mater. 1996, 8, 222−226. (9) Mort, J. Applications of amorphous materials. Phys. Technol. 1980, 11, 134. (10) Zolotoyabko, E. Basic Concepts of Crystallography: An Outcome from Crystal Symmetry; Vch Verlagsgesellschaft Verlag GmbH & Company KGaA, 2011. 4918

DOI: 10.1021/acs.chemmater.7b01139 Chem. Mater. 2017, 29, 4912−4919

Article

Chemistry of Materials

(32) Dingemans, G.; Van Helvoirt, C.; Van de Sanden, M.; Kessels, W. Plasma-Assisted Atomic Layer Deposition of Low Temperature SiO2. ECS Trans. 2011, 35, 191−204. (33) Puurunen, R. L.; Root, A.; Haukka, S.; Iiskola, E. I.; Lindblad, M.; Krause, A. O. I. IR and NMR study of the chemisorption of ammonia on trimethylaluminum-modified silica. J. Phys. Chem. B 2000, 104, 6599−6609. (34) Yasaka, M. X-ray thin film measurement techniques. Rigaku J. 2010, 26, 2. (35) Stoev, K.; Sakurai, K. Recent theoretical models in grazing incidence X-ray reflectometry. Rigaku J. 1997, 14, 22−37. (36) Kojima, I.; Li, B. Structural characterization of thin films by Xray reflectivity. Rigaku J. 1999, 16, 31. (37) Chu, W.-K. Backscattering Spectrometry; Elsevier, 2012. (38) Lizárraga, R.; Holmström, E.; Parker, S. C.; Arrouvel, C. Structural characterization of amorphous alumina and its polymorphs from first-principles XPS and NMR calculations. Phys. Rev. B: Condens. Matter Mater. Phys. 2011, 83, 094201. (39) Blonski, S.; Garofalini, S. Molecular dynamics simulations of αalumina and γ-alumina surfaces. Surf. Sci. 1993, 295, 263−274. (40) Vashishta, P.; Kalia, R. K.; Nakano, A.; Rino, J. P. Interaction potentials for alumina and molecular dynamics simulations of amorphous and liquid alumina. J. Appl. Phys. 2008, 103, 083504. (41) Gutiérrez, G.; Johansson, B. Molecular dynamics study of structural properties of amorphous Al 2 O 3. Phys. Rev. B: Condens. Matter Mater. Phys. 2002, 65, 104202. (42) Nelson, S. O. Density dependence of the dielectric properties of particulate materials. Trans. ASAE 1983, 26, 1823−1829. (43) Putkonen, M. I. ALD applications beyond outside IC technology-existing and emerging possibilities. ECS Trans. 2009, 25, 143−155.

(11) Whitesides, G. M.; Aizenberg, J.; Black, A. J. Control of crystal nucleation by patterned self-assembled monolayers. Nature 1999, 398, 495−498. (12) Aizenberg, J. Crystallization in patterns: a bio-inspired approach. Adv. Mater. 2004, 16, 1295−1302. (13) Lin, J.; Cates, E.; Bianconi, P. A. A Synthetic Analog of the Biomineralization Process: Controlled Crystallization of an Inorganic Phase by a Polymer Matrix. J. Am. Chem. Soc. 1994, 116, 4738−4745. (14) Wang, T.; Cölfen, H.; Antonietti, M. Nonclassical crystallization: mesocrystals and morphology change of CaCO3 crystals in the presence of a polyelectrolyte additive. J. Am. Chem. Soc. 2005, 127, 3246−3247. (15) Bloch, L.; Kauffmann, Y.; Pokroy, B. Size Effect on the Short Range Order and the Crystallization of Nanosized Amorphous Alumina. Cryst. Growth Des. 2014, 14, 3983−3989. (16) Solliard, C.; Flueli, M. Surface stress and size effect on the lattice parameter in small particles of gold and platinum. Surf. Sci. 1985, 156, 487−494. (17) Tavakoli, A. H.; Maram, P. S.; Widgeon, S. J.; Rufner, J.; van Benthem, K.; Ushakov, S.; Sen, S.; Navrotsky, A. Amorphous alumina nanoparticles: Structure, surface energy, and thermodynamic phase stability. J. Phys. Chem. C 2013, 117, 17123−17130. (18) Shyjumon, I.; Gopinadhan, M.; Ivanova, O.; Quaas, M.; Wulff, H.; Helm, C.; Hippler, R. Structural deformation, melting point and lattice parameter studies of size selected silver clusters. Eur. Phys. J. D 2006, 37, 409−415. (19) Roduner, E. Size matters: why nanomaterials are different. Chem. Soc. Rev. 2006, 35, 583−592. (20) Cartwright, J. H.; Checa, A. G.; Gale, J. D.; Gebauer, D.; SainzDíaz, C. I. Calcium carbonate polyamorphism and its role in biomineralization: how many amorphous calcium carbonates are there? Angew. Angew. Chem., Int. Ed. 2012, 51, 11960−11970. (21) Leskelä, M.; Ritala, M. Atomic layer deposition (ALD): from precursors to thin film structures. Thin Solid Films 2002, 409, 138− 146. (22) George, S. M. Atomic layer deposition: an overview. Chem. Rev. 2010, 110, 111−131. (23) Johnson, R. W.; Hultqvist, A.; Bent, S. F. A brief review of atomic layer deposition: from fundamentals to applications. Mater. Today 2014, 17, 236−246. (24) Miikkulainen, V.; Leskelä, M.; Ritala, M.; Puurunen, R. L. Crystallinity of inorganic films grown by atomic layer deposition: Overview and general trends. J. Appl. Phys. 2013, 113, 021301. (25) Leskelä, M.; Ritala, M. Atomic layer deposition chemistry: recent developments and future challenges. Angew. Chem., Int. Ed. 2003, 42, 5548−5554. (26) Kimoto, K.; Matsui, Y.; Nabatame, T.; Yasuda, T.; Mizoguchi, T.; Tanaka, I.; Toriumi, A. Coordination and interface analysis of atomic-layer-deposition Al2O3 on Si (001) using energy-loss nearedge structures. Appl. Phys. Lett. 2003, 83, 4306. (27) Adiga, S.; Zapol, P.; Curtiss, L. Atomistic simulations of amorphous alumina surfaces. Phys. Rev. B: Condens. Matter Mater. Phys. 2006, 74, 064204. (28) Elliott, S.; Scarel, G.; Wiemer, C.; Fanciulli, M.; Pavia, G. Ozone-based atomic layer deposition of alumina from TMA: Growth, morphology, and reaction mechanism. Chem. Mater. 2006, 18, 3764− 3773. (29) Mastail, C.; Lanthony, C.; Olivier, S.; Ducéré, J.-M.; Landa, G.; Estève, A.; Djafari Rouhani, M.; Richard, N.; Dkhissi, A. Introducing densification mechanisms into the modelling of HfO 2 atomic layer deposition. Thin Solid Films 2012, 520, 4559−4563. (30) Cimalla, V.; Baeumler, M.; Kirste, L.; Prescher, M.; Christian, B.; Passow, T.; Benkhelifa, F.; Bernhardt, F.; Eichapfel, G.; Himmerlich, M.; et al. Densification of thin aluminum oxide films by thermal treatments. Mater. Sci. Appl. 2014, 5, 628. (31) Groner, M.; Fabreguette, F.; Elam, J.; George, S. Lowtemperature Al2O3 atomic layer deposition. Chem. Mater. 2004, 16, 639−645. 4919

DOI: 10.1021/acs.chemmater.7b01139 Chem. Mater. 2017, 29, 4912−4919