Effect of an Interfacial Layer on Electron Tunneling through Atomically Thin Al2O3 Tunnel Barriers.

Subscriber access provided by LAURENTIAN UNIV

Article

The effect of an interfacial layer on electron tunneling through atomically-thin AlO tunnel barriers 2

3

Jamie Samantha Wilt, Ridwan Sakidja, Ryan Goul, and Judy Z. Wu ACS Appl. Mater. Interfaces, Just Accepted Manuscript • DOI: 10.1021/acsami.7b12170 • Publication Date (Web): 09 Oct 2017 Downloaded from http://pubs.acs.org on October 13, 2017

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.

ACS Applied Materials & Interfaces 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

ACS Applied Materials & Interfaces

The effect of an interfacial layer on electron tunneling through atomically-thin Al2O3 tunnel barriers Jamie Wilt1*, Ridwan Sakidja2, Ryan Goul1, and Judy Z. Wu1* 1

Department of Physics and Astronomy, University of Kansas, Lawrence, Kansas, 66045, USA Department of Physics, Astronomy and Materials Science, Missouri State University, Springfield, MO 65897, USA. Corresponding Authors Jamie Wilt, email: [email protected] Judy Wu, email: [email protected] 2

Keywords: Electron tunneling, interfacial layer, ultrathin film deposition, atomic layer deposition, Al2O3

ABSTRACT Electron tunneling through high-quality, atomically thin dielectric films can provide a critical enabling technology for future microelectronics, bringing enhanced quantum coherent transport, fast speed, small size, and high energy-efficiency. A fundamental challenge is in controlling the interface between the dielectric and device electrodes. An interfacial layer (IL) will contain defects and introduce defects in the dielectric film grown atop, preventing electron tunneling through the formation of shorts. In this work, we present the first systematic investigation of the IL in Al2O3 dielectric films of 1-6 Å’s in thickness on an Al electrode. We integrated several advanced approaches: molecular dynamics to simulate IL formation, in situ high vacuum sputtering-atomic layer deposition (ALD) to synthesize Al2O3 on Al films, and in situ ultrahigh vacuum scanning tunneling spectroscopy to probe the electron tunneling through the Al2O3. The IL had a profound effect on electron tunneling. We observed a reduced tunnel barrier height and soft-type dielectric breakdown which indicate that defects are present in both the IL and in the Al2O3. The IL forms primarily due to exposure of the Al to trace O2 and/or H2O during the pre-ALD heating step of fabrication. As the IL was systematically reduced, by controlling the pre-ALD sample heating, we observed an increase of the ALD Al2O3 barrier height from 0.9 eV to 1.5 eV along with a transition from soft to hard dielectric breakdown. This work represents a key step towards the 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

Page 2 of 23

realization of high-quality, atomically thin dielectrics with electron tunneling for the next generation of microelectronics.

Metal-insulator-metal tunnel junctions (MIMTJs), such as Josephson Junctions and Magnetic Tunnel Junctions are ubiquitous throughout the semiconductor industry.1,

2

For

optimal performance, the insulating dielectric layer must be high quality, uniform, and extremely thin as the tunneling current through the MIMTJ exponentially decays with barrier thickness.3, 4 In current technology, pushing beyond ultrathin (1-2 nm) down to atomic thicknesses (few Å) while maintaining a pinhole-free and defect-free dielectric is challenging. To push forward to the next-generation of MIMTJs, the metal-insulator (M-I) interface must be well controlled as it can significantly degrade the electron tunneling and hence the performance of MIMTJs.5, 6 In Josephson Junctions for example, oxygen vacancies within the dielectric tunnel barrier create two-level systems which are one of the major sources of decoherence in superconducting qubits.7-9 A present industry standard tunnel barrier for MIMTJ devices is thermal AlOx which is created through controlled oxygen diffusion into an Al wetting layer. When the AlOx thickness is pushed below roughly 4 Å, the oxygen diffusion process does not create a useful tunnel barrier due to the presence of defects, such as oxygen vacancies, and pinholes.10 As the thermal AlOx is grown thicker, pinholes become less prevalent, and the overall quality of the tunnel barrier increases as the defective interfacial layer (IL) is covered with AlOx.4, 10 Nevertheless, disorder and non-uniformity, primarily due to oxygen vacancies within the tunnel barrier and in the IL, limit the quality of thermal AlOx tunnel barriers which can be achieved.9 Atomic Layer Deposition (ALD) of Al2O3 provides an alternative to thermal oxidation for defect-free tunnel barrier growth with atomic-scale thickness control. ALD is a, typically low vacuum, chemical vapor deposition process which grows stoichiometric Al2O3 one atomic layer at a time through a series of precursor pulses, such as H2O and Trimethylaluminium (TMA). This cyclic series of precursor pulses allows for self-limited and conformal growth on the sample’s surface. Each ALD reaction cycle deposits a monolayer of Al2O3 which is 0.110.12 nm thick.11, 12 However, in most cases, a hydroxylated surface is required in order to enable TMA nucleation and ALD growth.13-15 While hydroxylation of a metal surface is possible through an incubation process by sacrificing the first tens of ALD cycles,14 the treated metal 2 ACS Paragon Plus Environment

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


surface typically contains a defective IL that is several nm in thickness.16-18 Such a thick IL is unacceptable for MIMTJs.

An IL can also develop prior to ALD during the sample

transportation between the physical vapor deposition (PVD) and ALD chamber, which is usually ex situ.19 However even in situ sample transportation, under low-vacuum, can create an IL which is greater than 0.5 nm in thickness.17, 20 In a prior exploratory work, we demonstrated that high quality ALD Al2O3 tunnel barriers can be grown on an Al wetting layer using an integrated ultrahigh vacuum (UHV) sputtering, and ALD system. In situ sample transport to a high vacuum (HV) ALD system, along with a well-controlled pre-ALD H2O pulse were the key breakthroughs achieved to overcome IL formation.21 However, the quality of the Al2O3 was found to be highly dependent on the sample heating conditions prior to the start of ALD. An IL may have formed even under HV. Therefore, to improve this technology to allow for high-quality ALD Al2O3 tunnel barriers which are suitable for MIMTJs, it is critical that the IL formation mechanisms are well understood. In this work, through the use of ab-initio molecular dynamics (AIMD) simulations along with in situ Scanning Tunneling Spectroscopy (STS), we probed the nucleation of the first ALD Al2O3 monolayer on the Al wetting layer while the pre-ALD heating process was systematically varied. This strategy is illustrated in Figure 1. The pre-ALD heating time required to reach the optimal temperature window of 150-190 °C was found to strongly affect IL formation through exposure to trace O2 or H2O which subsequently dissociates into adsorbed Oxygen ions (O ) on the Al surface. Thermal oxidation of the Al surface quickly follows to form an IL. Temperatures outside this window either impaired the Al surface hydroxylation from the pre-ALD H2O pulse or contributed to IL growth due to O formation on the Al surface.

We show that the IL can be suppressed by controlling the heating time below 10-15 min, resulting in superior electron tunneling characteristics through atomically-thin ALD Al2O3 tunnel barriers.

3 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 4 of 23

Figure 1: An illustration of the strategy employed in this work to study and minimize IL formation. Al was magnetron sputtered and transferred, under high vacuum, to the ALD chamber where an initial H2O pulse hydroxylated the Al surface just prior to the ALD reactant pulses (TMA and H2O). In situ STS studied the electron tunneling properties of the insulator to observe the significance and effect of the IL on the ALD Al2O3. Then with insight gained from AIMD simulations, the pre-ALD heating conditions (temperature and time) were varied to optimize ALD conditions to minimize the IL.

RESULTS AND DISCUSSION Numerical Simulations on the Mechanisms of IL Formation In our UHV sputtering-ALD system, the sputtered Al wetting layer experiences a preALD exposure to vacuum during the in situ transport to the ALD chamber and the beginning of ALD. This is because sputter deposition and ALD operate in two different chambers, to prohibit cross contamination, and the samples had to bridge a temperature differential from ~12 °C to 100-350 °C for sputtering and ALD respectively.22

Bridging this temperature differential

inevitably took some time. Trace O2 or H2O originating from the ALD reactor during this heating step may form an IL on Al via the thermal oxidation process. To shed light on the microscopic mechanisms of this possible IL formation, a number of AIMD simulations were performed. Figure 2a-I shows the atomic trajectory of O on Al (111) at a temperature of 80 K

(-193°C). After 1.5 ps, trace O have distorted the topology of the Al surface lattice. The Al 4 ACS Paragon Plus Environment

Page 5 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


atoms are spontaneously “extracted” from their original surface positions, and a rough topology is created. A very similar phenomenon has been predicted by a previous theoretical study23 based on ground state Density Functional Theory calculations. Al extraction during the early stages of oxidation has been attributed to the strong propensity to form AlOx clusters on the Al surface, whereby the Al ions spontaneously move to the center of each cluster. Our AIMD simulations demonstrate that this process can occur at a temperature as low as 80 K (-193 °C). Higher temperatures accelerate this process.

Thus it is conceivable that any trace oxygen

impurities in the ALD chamber present during the pre-ALD sample heating, may initiate thermal oxidation to form an AlOx IL.

Figure 2: Snapshots from AIMD simulations are shown for (a) the proposed thermal AlOx IL formation mechanisms on Al. (I) thermal oxidation from exposure to trace oxygen. The temperature was 80 K (-193 °C) and the simulation was run for 1.5 ps. (II) thermal oxidation from exposure to H2O and it’s subsequent dissociation into O . The temperature was 600K (327 °C) and the simulation was run for 0.5 ps. (b) a simulated H2O pulse on the Al wetting layer. The temperature was (I) 300 K (27 °C), (II) 423 K (152 °C) and (III) 473 K (200 °C). The simulations was run for 1.5 ps, 3 ps, and 2 ps respectively. All AIMD simulations were run with 1 fs step sizes. 5 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 6 of 23

Aside from O2, a thermal AlOx IL may also form indirectly upon exposure to H2O, originating from either the initial H2O pulse or from exposure to trace H2O in the ALD reactor during the pre-ALD sample heating, which has a vacuum pressure on the order of 10-4 – 10-5 Torr. Figure 2a-II depicts the trajectory of adsorbed water molecules (H O ) on an Al (111) surface at 600 K after only 0.5 ps. An elevated sample temperature can dissociate H O into adsorbed hydroxyl molecules (OH ) which can then dissociate into O and H+.

The

O oxidize the Al layer similar to Figure 2a-I. This dissociation of H O into O is quite

temperature dependent, becoming more prevalent with higher temperatures.21 This sensitivity to temperature is demonstrated in Figure 2b which shows the simulated trajectories of H O on Al with a temperature of 300 K (27 °C), 423 K (152 °C) and 473 K (200 °C). When the sample temperature is too low, (e.g. at 300 K shown in Figure 2b-I), dissociation of H O into OH does not occur efficiently so some H O remain on the Al surface.

When the

temperature is just right (e.g. near 423 K (152 °C), shown in (Figure 2b-II), the majority of the H O dissociate into OH without much dissociation into O and H+. However when the

temperature is too high, (e.g. near and greater than 473 K (200 °C), shown in Figure 2b-III), the dissociation of H O into OH occurs with a subsequent dissociation of OH into O and remain behind to form a thermal AlOx H+. The H+ leave the surface to form H2 gas, and the O

IL, as is the case of Figure 2a-II. Further observations along with corresponding movies of the AIMD simulations are available in the Supporting Information.

Overall, these AIMD

simulations suggest that the sample temperature during the pre-ALD H2O pulse should be maintained within the range of 423-473 K (152-200 °C) to favor a hydroxylated Al surface with minimal dissociation of OH into O .

To understand how an IL may affect the ALD Al2O3 growth, we first must understand the ideal case of ALD growth on a hydroxylated Al surface without an IL. In Figure 3a, we evaluated the interaction of TMA with a hydroxylated Al (111) surface.

We set up this

simulation by placing a horizontally aligned TMA molecule on top of a pool of seven OH on

the Al (111) surface (Figure 3a-I). The first step of this TMA-OH- interaction is the adsorption of the TMA’s Al cation onto an OH . This adsorption partially elevates the hydroxylated Al

surface and slightly distorts the TMA molecule with a slight change in the bond angle between 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


the Al and the three methyl ligands. This interaction can be seen in Figure 3a-II, III. Evidently there is a strong preference for an Al-O bond. This finding is consistent with the results from a previous study on the interaction between TMA and a hydroxylated surface which found that the Al-O adsorption process is a highly energetically favorable exothermic reaction.24

Figure 3: Snapshots from AIMD simulations at 432 K (159 °C) up to 7 ps (1fs for each step) are shown for the (a) TMA adsorption onto a well-hydroxylated Al (111) surface. The initial setup is shown in (I), the TMA interaction after 1ps with an OH is shown in (II), and the subsequent + formation of an Al-O bond and H release from an OH is shown in (III). (b) subsequent reactions of the attached TMA molecule on the hydroxylated Al surface. The initial setup is shown in (I), the proton exchange between nearby OH and one of the CH3 groups of the TMA is shown in (II), and the final release of the CH4 molecule is shown in (III).

Throughout the AIMD simulations, we observed intermolecular hydrogen bonds present ; as evident by the frequent re-alignments of the hydrogen atoms. The OH amongst the OH

intermolecular bond length is quite short, in the range of 1.4 Å to 2.0 Å. A hydrogen bond between two close OH , results in a horizontal OH alignment. This horizontal alignment can only be achieved with a high surface-packing density of OH on the Al (111) surface. As a consequence of this horizontal alignment, the O anion in the OH is exposed towards the Al



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 8 of 23

cation from the TMA. A vertically-aligned OH , on the other hand, has significant steric hindrance for TMA adsorption onto the OH . Thus horizontal alignment of the OH is critical

to enable an efficient TMA adsorption on the hydroxylated Al wetting layer. While it has been well-established that ALD reactions, particularly with respect to TMA are self-terminating, that is not the case for the hydroxylation process of the Al (111) surface. It is hence reasonable to assume that rougher Al surface topology, created by the ingress of oxygen from a thermal AlOx IL, would lead to a reduced OH surface density. The additional steric hindrance from the resulting vertically aligned OH will likely lead to a reduced TMA density in the first ALD

cycle which will, in turn, lead to defects in the ALD Al2O3 tunnel barrier. To look beyond the initial TMA adsorption, we ran further AIMD simulations that depict the release of CH4 from the TMA molecule following the reaction shown in equation (1). ||-OH + ||-O-Al-(CH3)3 ||-O + ||-O-Al(CH3)2 + CH4(g)

(1)

where ||-OH stands for hydroxylated Al (111) surface and ||-O-Al-(CH3)3 is the adsorbed TMA with the reaction yielding the adsorbed dimethylaluminium or ||-O-Al(CH3)2 and the methane gas CH4(g). AIMD snapshots for our simulation at 432K (159oC) of this reaction are shown in Figure 3b. A previous theoretical work24 has shown that the first CH4 dissociation process is characterized by an energy activation in the range of 0.35-0.9 eV. This wide range of energy is due to the varying degree of steric hindrance for the protonation onto the methyl ligands. The minimum energy path for this dissociation process suggests that a protonation reaction following reaction (1) results in the release of CH4 gas and a stable formation of dimethylaluminium. The dissociation mechanism is quite similar to that observed in an ALD reaction onto a Si(100) template.25,

26

Using Nudge Elastic Band analysis, we found that reaction (1) has an energy

barrier of about 0.5 eV which is quite comparable to that of α-Al2O3.24, 27 More details on this analysis are in the Supporting Information.

This analysis however, is predicated upon a

simplistic assumption that there are always going to be nearby hydrogen atoms on the hydroxylated Al surfaces to remove the CH3 ligand. We expect that the presence of an IL will make this reaction (1) more difficult due a distorted Al lattice and low OH density on the

surface.


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


The possible impact of the IL on the quality of the ALD Al2O3 tunnel barriers is schematically illustrated in Figure 4. As we have discussed in Figure 3, the defective structure of an IL may perturb the hydroxylated surface sufficiently to impair TMA nucleation and the overall ALD Al2O3 growth (Figure 4a). Gaps in the TMA coverage due to steric hindrance from an AlOx IL may lead to pinholes or localized locations of low quality alumina (Figure 4a-I). Eventually, once the initial alumina layer is established, the ALD process should proceed normally (Figure 4a-II). This means that in order to achieve a high-quality ALD Al2O3 tunnel barrier with atomic-scale thickness precision, as depicted Figure 4b, it is crucial that the IL is eliminated through precise control of the pre-ALD heating conditions. Without an IL, the Al2O3 density and resulting tunnel barrier quality should be constant with ALD cycle number from one ALD cycle (Figure 4b-I) to any number of ALD cycles Figure 4b-II.



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

Figure 4: An illustrative diagram of the hypothetical internal structure for (a) a ALD Al2O3 tunnel barrier with an IL which is (I) 1 ALD cycle and (II) 5 ALD cycles in thickness and (b) a ALD Al2O3 tunnel barrier without an IL which is (I) 1 ALD cycle and (II) 5 ALD cycles in thickness.

In Situ Scanning Tunneling Spectroscopy on the effect of IL development in the ALD Al2O3 tunnel barriers Based on the simulations in Figures 2 and 3, it is clear that the thermal budget of ALD must be controlled in order to minimize the Al surface’s exposure to O . This exposure is

especially critical to ALD Al2O3 growth during the first ALD cycle as any disruption to TMA nucleation will prevent full Al2O3 coverage on the Al (Figure 4a-I). To understand the effect of the heating parameters (temperature and time) on the tunnel barrier quality, we developed a dynamic pre-ALD heating strategy which allowed us to probe both parameters independently. Figure 5a shows the sample’s temperature as a function of exposure time in a preheated ALD reaction chamber. Three different blackbody heater powers for the ALD chamber were explored in this dynamic heating strategy: 156 W, 220 W, and 304 W.

As our ALD reactor is

cylindrically shaped,28 a little under half of this wattage was directed inwards towards the sample. Fits to the data of the form given in Eq.(2) found the sample’s steady-state temperature to be 187 °C, 220 °C, and 267 °C as time, t, goes to infinity and the time constant, τ, was 17.2 min, 15.0 min, and 12.6 min for the three powers respectively,

) = + − )1 −

)

(2)

where Ti is the sample temperature prior to heating and Tf is the steady-state temperature. As expected, increasing the blackbody heater power led to a significantly reduced τ and an increased Tf. This dynamic heating strategy has the advantage of bringing the sample’s temperature to the ALD suitable window quickly at the expense of a non-constant temperature during ALD. Depending on the heater power and time position on Figure 5a, the sample temperature can change at a rate as high as 10 °C/min. Thus this dynamic heating strategy is best suited for 10 ACS Paragon Plus Environment

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


growth of the ALD tunnel barriers of only a few Å thick so the growth can be completed within a few minutes.

Figure 5: dI/dV STS spectroscopy and the optimization of the heating conditions for the ALD Al2O3 process in order to minimize the formation of an IL. (a) The measured Sample temperature is shown as a function of exposure time in an ALD reaction chamber which has been preheated at the given wattages. The solid lines are fits to the data. (b) A Representative dI/dV spectra and corresponding IV spectra (insert) is shown for a 1 cycle ALD Al2O3 tunnel barrier. The barrier height, denoted by the position of the blue arrow, was determined by the intersection of two linear fits (shown in red) for the band gap and conduction band respectively. (c) The ALD Al2O3 Coverage on the Al surface is shown for a 1 cycle of ALD Al2O3 tunnel barrier as a function of the sample temperature during the start of ALD (solid circles). The corresponding barrier heights are shown with the open circles. The grayed out Area in (c) and (a) is a rough estimate for the ideal temperature window required to have high ALD surface coverage on the Al after only 1 11 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 12 of 23

ALD cycle. (d) The measured Barrier heights are shown for both 1 cycle and 5 cycle ALD samples as function of heating time. All samples were within the temperature window from (c).

The quality of the ALD Al2O3 tunnel barriers grown using this dynamic heating strategy were studied using in situ STS on 1 cycle ALD Al2O3 tunnel barriers. A representative dI/dV and I-V STS spectra is shown in Figure 5b for a 1 cycle ALD Al2O3 tunnel barrier that was heated for 15 min at a power of 304 W to a temperature of about 169 °C. STS dI/dV spectra probe the local density of states of the sample’s surface.29 Within the band gap of the insulator at a low bias voltage (tunneling electron energy), few states are available for tunneling so the tunnel current will be small and dI/dV flat. At the onset of the insulator conduction band (or valence band), states rapidly become available for tunneling, leading to a sharp rise in dI/dV which is roughly linear on the log dI/dV scale.30 The tunnel barrier height (Eb) is defined as this rapid onset and is reflective of the quality of the tunnel barrier. Defects within the insulator provide additional tunneling states which smear this conduction band onset toward the Fermi level, lowering Eb.9, 31 The Eb from Figure 5b was 1.58 eV which is typical for our ALD Al2O3 tunnel barriers with values in the range of 1.0-1.6 eV. This ALD Al2O3 Eb is significantly higher than the thermal AlOx tunnel barrier which has an Eb of 0.6-1.0 eV2, 4, 21 and is comparable to plasmaassisted Aluminum oxidation which has an Eb of around 1-3 eV.

2, 32, 33

What is remarkable

about this ALD Al2O3 process is that this high Eb value is maintained when the Al2O3 is only one atomic layer, or about 1-2 Å, in thickness. Competing Al2O3 growth methods such as sputtering, thermal, or plasma-assisted oxidation suffer from pinholes or oxidation of the underlying substrate when the alumina is grown near this thickness scale.10, 34, 35 The ALD Al2O3 coverage was then estimated as the percent of STS spectra which showed evidence of a tunnel barrier with an Eb > 1 eV and is shown in Figure 5c. Eb values less than 1 eV were considered to be thermal AlOx or pinholes. The average ALD Al2O3 Eb was constant with temperature with an average value of about 1.5 eV. Interestingly there is a maximum of ALD coverage with a value of about 93 % in Figure 5c for temperatures between 150 °C and 190 °C. This temperature range, noted with a grey box in Figure 5c and Figure 5a, roughly corresponds to the ideal temperature range for the hydroxylation of the Al wetting layer using the pre-ALD H2O pulse and matches the results from our AIMD simulations earlier in 12 ACS Paragon Plus Environment

Page 13 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


Figure 3b. In the lower temperature range of 60-150 °C, the ALD coverage drops down to ~78%. Although we did not fully exhaust the reaction sequences between TMA and H O on the Al surface, we postulate that the TMA did not nucleate successfully on areas of the Al surface which were covered by the water molecules. As a result of poor or no TMA nucleation, on an Al surface, it nucleates and Al2O3 did not grow. However when the TMA does find OH

the tunnel barrier is still of high quality (high Eb). In the higher temperature range of 190-225 °C, the ALD coverage reduces to ~85 %. One explanation for this drop in coverage may be that + some OH have dissociated into O and H . A slight thermal AlOx IL may result. As already

discussed in Figures 2 and 3, this thin IL will provide additional steric hindrance for TMA must not nucleation, accounting for the drop in coverage. However, this dissociation of OH

be too severe as Eb did not decrease in this temperature range. In Figure 5d, Eb was measured for both 1 cycle ALD Al2O3 and 5 cycle ALD Al2O3 tunnel barriers as a function of heating time with temperatures within the ideal window from Figure 5c. The three heating powers of 156 W, 220 W, and 304 W were used for heating times 75 min, 26 min, and 15 min respectively. The ALD coverage was constant with sample heating time with values greater than 90%. Due to its thicknesses relative to any IL, 1 cycle ALD Al2O3 tunnel barriers are affected more significantly by an IL than 5 cycle ALD Al2O3 tunnel barriers. Thus, assuming a defective IL with a low Eb, such as thermal AlOx, the value of Eb can be correlated with the presence of an IL. The 1 cycle ALD Al2O3 Eb values in Figure 5d drop at a roughly linear rate of about 10 meV/min, from a value of about 1.5 eV to 0.9 eV, over the course of one hour of extra heating (15 min to 75 min). This confirms that the IL in our ALD Al2O3 tunnel barriers grows thicker with increased pre-ALD exposure time through exposure to trace O2 or H2O at elevated temperatures. When these tunnel barriers are grown thicker, the contribution of the IL to tunneling reduces. Therefore at 5 cycles ALD (~0.6 nm) in thickness, the STS dI/dV spectra are primarily probing the Al2O3 density rather than the disorder at the Al-Al2O3 interface. As can be seen in Figure 5d, the 5 cycle ALD Al2O3 Eb did not decrease significantly between 15 min and 26 min heating in the ALD chamber with a rate of only -4.5 meV/min. However between 26 min and 75 min heating, this rate was about -7.8meV/min. Evidently, there must have been some IL that formed, during the long heating time of 75 min that was significant enough to impact the overall ALD Al2O3 growth and density similar to the schematic in Figure 4a-II. This trend of IL 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

formation is also evident in the Eb differences between 1 and 5 ALD Al2O3 cycles for these three heating times. When only heated for 15 min, the 1 cycle ALD Eb was about the same as the corresponding 5 cycle ALD Eb. Beyond 15 min heating, the 1 cycle ALD Al2O3 Eb was consistently lower than the corresponding 5-cycle ALD Al2O3 Eb. The constancy of Eb with Al2O3 thickness when the sample was only heated for 15min prior to ALD implies that the IL may be negligibly small for these samples. Shortening the heating time further may improve Eb slightly although it’s unlikely to increase Eb more than about 10%. This possible minor Eb improvement is likely insufficient to warrant the engineering efforts required to heat large wafers for MIMTJs through a temperature differential of about 150 °C in only a few minutes. In situ Atomic Force Microscopy (AFM) images in Figure S2 confirm that the IL does not generate any observable changes in surface roughness before and after 1 cycle of ALD Al2O3 using this heating setting. The AFM topography is also extremely conformal, confirming a very high Al2O3 coverage on the Al surface. The behavior of tunnel barriers under intense dielectric stress (>10 MV/cm) can provide additional insights as to the nature and significance of the IL. Dielectric breakdown (BD) may be initiated by the high local electric field generated by the STS tip as the bias voltage is ramped up and down many times during subsequent dI/dV spectra. Figure 6 shows representative BD behavior observed for three insulating tunnel barriers. That is, a thermally oxidized AlOx tunnel barrier which was about 0.3 nm in thickness (Figure 6a), a 1-cycle ALD Al2O3 tunnel barrier which was heated for 75 min (Figure 6b), and a 1-cycle ALD Al2O3 tunnel barrier which was heated for 15 min (Figure 6c). As we observed in Figure 6a, thermal AlOx breaks down under the STM tip in a gradual, soft manner as disorder increases within the tunnel barrier through defect migration. Eventually the STS spectrum becomes linear (metallic).9, 31, 36-39 Interestingly, this soft BD was also observed in the 1-cycle ALD Al2O3 tunnel barrier which was heated for 75 min. (with an AlOx IL). Although this spectra (Figure 6b) does not show complete soft BD, the zero voltage conductance, defined as the slope of the dI/dV spectra in the band gap regime, did increase with subsequent spectra similar to the case of thermal AlOx (Figure 6a). This soft BD behavior observed for 1 cycle ALD Al2O3 with 75 min heating indicates that defects are present within the tunnel barrier and/or at the M-I interface.


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


Figure 6: Representative STS dI/dV spectra are shown to illustrate the type of breakdown behavior observed. The types of dielectric breakdown behavior observed as the STM bias voltage is ramped up and down are shown for (a) the thermal AlOx, tunnel barrier (b) a 1-cycle ALD Al2O3 tunnel barrier which was headed for 75 min, and (c) a 1-cycle ALD Al2O3 tunnel barrier which was heated for 15 min. The insert shows the I-V curve for a hard breakdown event (bottom).

When there was not a significant IL present in the ALD Al2O3 tunnel barrier, hard BD was observed. As shown in Figure 6c, there was a massive, sudden increase in the tunneling current (Figure 6c insert). After this traumatic BD event, the insulator becomes metallic with linear STS spectra. This form of dielectric BD is typical for crystalline insulators (>10s of nm thick) in capacitors.40 In the ultrathin regime of 1-2 nm, hard BD is primarily observed by STS for epitaxial Al2O336. Rather than gradual defect migration within the barrier, hard BD represents the breaking of the Al2O3 bonds in the insulator.41 Therefore, the presence of hard BD in our 1cycle ALD Al2O3 tunnel barriers in Figure 6c confirms a low-defect atomically thin Al2O3 tunnel barrier which does not have a significant IL.



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 16 of 23

Finally, while it is difficult to calculate the magnitude of the breakdown field using STM (as the tip-sample distance is difficult to determine without crashing the tip), we can estimate the rough bounds for this breakdown field. Hard BD events typically occurred around 2V in our STS spectra for the 1-cycle ALD Al2O3 samples with 15 min heating. The thickness of this Al2O3 is expected to be 1.1-1.2Å. If we estimate the tip-sample distance as somewhere between 1Å and 1 nm, then the breakdown field is on the order of 16-100 MV/cm. This estimated breakdown field is rather high, considering that epitaxial Al2O3 has a value on the order of 11 MV/cm.36 However, this large breakdown field is consistent with MIMTJ measurements for ALD Al2O3 on GaAr42 and suggests that ALD Al2O3 tunnel barriers are significantly more robust than their thermal AlOx counterparts. This resilience to dielectric stress may in turn lead to higher MIMTJ yields per wafer, which is important to practical applications.43

CONCLUSIONS In conclusion, this work presents the first systematic investigation of the effect of the M-I interface on the electron tunneling behavior through atomically-thin insulating tunnel barriers by integrating several advanced approaches of AIMD to simulate IL formation, in situ sputteringALD to synthesize 1-6 Å thick Al2O3 on Al films, and in situ STS to probe ALD Al2O3 tunnel barrier height Eb and breakdown behaviors as the IL was systematically controlled. Several important insights have been obtained in this investigation. First, a pre-ALD Al oxidation can occur through exposure to trace O2 in the vacuum or though dissociation of H O into O on

the Al surface, resulting in a defective AlOx IL. Quantitatively, more serious IL formation occurs at elevated sample temperature and prolonged pre-ALD heating time in the ALD chamber. Secondly, any pre-ALD IL that does form will reduce the OH density on the Al surface which in turn reduces the number of horizontally aligned OH on the Al surface.

Consequently TMA nucleation is impaired, resulting in defective ALD Al2O3 growth. Thirdly, based on this understanding of the IL formation mechanisms, optimal pre-ALD processing conditions were developed to dynamically heat the sample to the optimal growth temperature window of 150 °C-190 °C. Temperatures below this range were found to lead to H O on the Al surface which impaired the TMA nucleation and temperatures above this range led to a rough


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


Al surface through OH dissociation into O and H+. In addition, reducing the dynamic

heating time down to about 15 min led to ALD Al2O3 growth on the Al surface with a negligible IL forming.

Finally, STS revealed the distinctive effect the IL had on electron tunneling

including a reduction in Eb and soft-type dielectric breakdown.

Both effects indicate the

presence of a defective IL and a defective ALD Al2O3 grown on top. As the IL formation is suppressed with optimal pre-ALD processing conditions, a thickness-independent Eb in the range of 1.42-1.56 eV and hard-type dielectric breakdown were achieved. This result illustrates that atomically-thin ALD Al2O3 tunnel barriers can have superior dielectric properties approaching that of crystalline Al2O3. .

MATERIALS AND METHODS Computational Simulations on the IL growth mechanisms and it’s effect on ALD growth were carried out using Density Functional Theory calculations using the Nudged-elastic band method44 and AIMD simulations under constant temperature and volume with 1 fs for each trajectory step as implemented in VASP45 and Quantum Espresso46 codes. Detailed procedures to model the Al surface reactions can be found in our previous works.47-49 To create the M-I structure for in situ STS, a bilayer of Nb (20 nm)/Al (7 nm) was DC magnetron sputtered onto a Si/Au (50 nm) substrate. Au was evaporated onto an undoped Si wafer with a native oxide to serve as a ground contact for the STS system (RHK technologies). In order to reduce oxidation following magnetron sputtering, the samples were transferred to the ALD chamber in a unique, home-built vacuum chamber which is capable of in situ sample transport under HV between the sputtering, ALD, and SPM chambers.28 After the in situ transport to the preheated ALD chamber, the samples were dynamically heated under HV in preparation for ALD. Following heating, 1-5 cycles of ALD Al2O3 was deposited through the standard H2O and TMA process13 with a 5 sccm N2 carrier gas to create the insulating tunnel barrier. The first reactant pulse was H2O in order to hydroxlate the Al surface to allow for ALD nucleation in the first ALD cycle. This first H2O pulse has been described in detail in our earlier work.21

To compare the breakdown characteristics of our ALD Al2O3 tunnel barriers to



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 18 of 23

traditional thermal AlOx tunnel barriers, a 0.3 nm thermal AlOx tunnel barrier was also fabricated using a 3.81 Torr-sec O2 exposure in the sputtering chamber.4 To examine the Al2O3 tunnel barriers in situ, the samples were transferred, under HV to the STS chamber for measurement immediately following tunnel barrier fabrication. The STS chamber had a pressure of about 2×10-10 Torr. A mechanically-cleaved Pt-Ir tip was used for all STS measurements at room temperature. Constant height IV and dI/dV spectroscopy were taken simultaneously using the lock-in amplifier method with a voltage modulation of 100 mV at 1 kHz or 30mV at 5kHz. To examine the dielectric breakdown of the Al2O3 tunnel barrier, The STS tip was held fixed at each scanned location and the bias was sequentially ramped up and down 20 times.

ASSOCIATED CONTENT Supporting Information The following files are available free of charge 1. Additional AIMD and Nudge-Elastic-Band simulations and analysis on the release of CH4 from TMA on the hydroxylated Al surface and on α-Al2O3. In situ AFM images were also taken of the Al surface before and after 1 cycle of ALD Al2O3. (file type: PDF) 2. Video of the trajectories from AIMD simulation at 300 K up to 1.5 ps showing that the minority of the initially adsorbed water ligands have turned into OH ligands with some parts remaining as H O . (file type: mp4) 3. Video of the trajectiries from AIMD simulation at 423K up to 3 ps showing the majority of the initially adsorbed water ligands have turned into OH ligands with some parts of the hydrogen atoms adsorbed onto the Al substrate or released as a hydrogen gas. (file type: mp4 4. Video of the trajectories from AIMD simulation at 473K up to 2 ps showing the removal of hydrogen from the water pool leaving adsorbed oxygen only on the surface. (file type: mp4) 5. Video of the TMA being readily adsorbed by hyroxolated Al surface (file type” mp4)

AUTHOR INFORMATION 18 ACS Paragon Plus Environment

Page 19 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


Corresponding Authors * [email protected] * [email protected] ORCID Jamie Wilt: 0000-0003-0913-2889 Judy Wu: 0000-0001-7040-4420 Author Contributions J.S.W. and J.Z.W. designed the experiment. J.S.W. prepared the samples for STS and performed the STS characterization and most of the analysis. R.G. helped with some of the STS analysis. R.S. did the simulations. All authors contributed to discussions of the results. J.S.W, R.S., and J.Z.W. led the effort in development of the manuscript. Funding Sources Support in part by NASA contract NNX13AD42A, ARO contract No. W911NF-16-10029, and NSF contracts Nos. NSF-DMR-1337737, and NSF-DMR-1508494. Notes The Authors declare no competing financial interest

ACKNOWLEGEMENT The Corresponding Authors acknowledge support in part by ARO contract No. W911NF-16-10029, and NSF contracts Nos. NSF-DMR-1337737, and NSF-DMR-1508494. We also acknowledge Jennifer Totleben for her assistance in the synthesis of STS substrates and Prof. Siyuan Han for many beneficial discussions.

REFERENCES (1) Barone, A.; Paterno, G., Physics and Applications of the Josephson Effect. Wiley Online Library: 1982; Vol. 1. (2) Zhu, J.-G.; Park, C. Magnetic Tunnel Junctions. Materials Today 2006, 9 (11), 36-45. (3) Dorneles, L. S.; Schaefer, D. M.; Carara, M.; Schelp, L. F. The Use of Simmons’ Equation to Quantify the Insulating Barrier Parameters in Al/Alox/Al Tunnel Junctions. Applied Physics Letters 2003, 82 (17), 2832-2834. (4) Kang, X.; Ying, L.; Wang, H.; Zhang, G.; Peng, W.; Kong, X.; Xie, X.; Wang, Z. Measurements of Tunneling Barrier Thicknesses for Nb/Al–Alox/Nb Tunnel Junctions. Physica C: Superconductivity 2014, 503 29-32.



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 20 of 23

(5) Diešková, M.; Konôpka, M.; Bokes, P. Atomic and Electronic Structure of Ultra-Thin Al/Alo X/Al Interfaces. Surface Science 2007, 601 (18), 4134-4137. (6) Gangineni, R.; Bellouard, C.; Duluard, A.; Negulescu, B.; Baraduc, C.; Gaudin, G.; Tiusan, C. Interfacial Electronic Transport Phenomena in Single Crystalline Fe-Mgo-Fe Thin Barrier Junctions. Applied Physics Letters 2014, 104 (18), 182402. (7) McDermott, R. Materials Origins of Decoherence in Superconducting Qubits. Applied Superconductivity, IEEE Transactions on 2009, 19 (1), 2-13. (8) Martinis, J. M.; Cooper, K.; McDermott, R.; Steffen, M.; Ansmann, M.; Osborn, K.; Cicak, K.; Oh, S.; Pappas, D.; Simmonds, R. Decoherence in Josephson Qubits from Dielectric Loss. Physical Review Letters 2005, 95 (21), 210503. (9) Tan, E.; Mather, P.; Perrella, A.; Read, J.; Buhrman, R. Oxygen Stoichiometry and Instability in Aluminum Oxide Tunnel Barrier Layers. Physical Review B 2005, 71 (16), 161401. (10) Kleinsasser, A. W.; Miller, R. E.; Mallison, W. H. Dependence of Critical Current Density on Oxygen Exposure in Nb-Alo/Sub X/-Nb Tunnel Junctions. Applied Superconductivity, IEEE Transactions on 1995, 5 (1), 26-30. (11) Rigetti, C.; Gambetta, J. M.; Poletto, S.; Plourde, B.; Chow, J. M.; Córcoles, A.; Smolin, J. A.; Merkel, S. T.; Rozen, J.; Keefe, G. A. Superconducting Qubit in a Waveguide Cavity with a Coherence Time Approaching 0.1 Ms. Physical Review B 2012, 86 (10), 100506. (12) Miikkulainen, V.; Leskelä, M.; Ritala, M.; Puurunen, R. L. Crystallinity of Inorganic Films Grown by Atomic Layer Deposition: Overview and General Trends. Journal of Applied Physics 2013, 113 (2), 021301. (13) George, S. M. Atomic Layer Deposition: An Overview. Chemical Reviews 2009, 110 (1), 111-131. (14) Elliott, S. D. Atomic-Scale Simulation of Ald Chemistry. Semiconductor Science and Technology 2012, 27 (7), 074008. (15) Martin, M.-B.; Dlubak, B.; Weatherup, R. S.; Yang, H.; Deranlot, C.; Bouzehouane, K.; Petroff, F.; Anane, A.; Hofmann, S.; Robertson, J. Sub-Nanometer Atomic Layer Deposition for Spintronics in Magnetic Tunnel Junctions Based on Graphene Spin-Filtering Membranes. ACS nano 2014, 8 (8), 78907895. (16) Groner, M.; Elam, J.; Fabreguette, F.; George, S. M. Electrical Characterization of Thin Al 2 O 3 Films Grown by Atomic Layer Deposition on Silicon and Various Metal Substrates. Thin Solid Films 2002, 413 (1), 186-197. (17) Elliot, A. J.; Malek, G.; Wille, L.; Lu, R.; Han, S.; Wu, J. Z.; Talvacchio, J.; Lewis, R. M. Probing the Nucleation of in Atomic Layer Deposition on Aluminum for Ultrathin Tunneling Barriers in Josephson Junctions. Applied Superconductivity, IEEE Transactions on 2013, 23 (3), 1101405-1101405. (18) Kukli, K.; Ritala, M.; Pilvi, T.; Aaltonen, T.; Aarik, J.; Lautala, M.; Leskelä, M. Atomic Layer Deposition Rate, Phase Composition and Performance of Hfo2 Films on Noble Metal and Alkoxylated Silicon Substrates. Materials Science and Engineering B 2005, 118 112-116. (19) Moyerman, S. M.; Feng, G.; Krayer, L.; Stebor, N.; Keating, B. G. Atomic Layer Deposition of Tunnel Barriers for Superconducting Tunnel Junctions. Journal of Low Temperature Physics 2014, 176 (3-4), 237242. (20) Lu, R.; Elliot, A. J.; Wille, L.; Mao, B.; Han, S.; Wu, J. Z.; Talvacchio, J.; Schulze, H. M.; Lewis, R. M.; Ewing, D. J. Fabrication of Josephson Junctions Using in Situ Magnetron Sputtering and Atomic Layer Deposition. Applied Superconductivity, IEEE Transactions on 2013, 23 (3), 1100705-1100705. (21) Wilt, J.; Gong, Y.; Gong, M.; Su, F.; Xu, H.; Sakidja, R.; Elliot, A.; Lu, R.; Zhao, S.; Han, S. Atomically Thin Al 2 O 3 Films for Tunnel Junctions. Physical Review Applied 2017, 7 (6), 064022. (22) Gieraltowska, S.; Wachnicki, L.; Witkowski, B. S.; Mroczynski, R.; Dluzewski, P.; Godlewski, M. Characterization of Dielectric Layers Grown at Low Temperature by Atomic Layer Deposition. Thin Solid Films 2015, 577 97-102. 20 ACS Paragon Plus Environment

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


(23) Lanthony, C.; Ducéré, J. M.; Rouhani, M. D.; Hemeryck, A.; Estève, A.; Rossi, C. On the Early Stage of Aluminum Oxidation: An Extraction Mechanism Via Oxygen Cooperation. The Journal of Chemical Physics 2012, 137 (9), 094707. (24) Weckman, T.; Laasonen, K. First Principles Study of the Atomic Layer Deposition of Alumina by Tma– H 2 O-Process. Physical Chemistry Chemical Physics 2015, 17 (26), 17322-17334. (25) Halls, M. D.; Raghavachari, K. Atomic Layer Deposition Growth Reactions of Al2o3 on Si(100)-2×1. The Journal of Physical Chemistry B 2004, 108 (13), 4058-4062. (26) Dillon, A. C.; Ott, A. W.; Way, J. D.; George, S. M. Surface Chemistry of Al 2o 3 Deposition Using Al(Ch 3) 3 and H 2o in a Binary Reaction Sequence. Surface Science 1995, 322 230-242. (27) Sheppard, D.; Henkelman, G. Paths to Which the Nudged Elastic Band Converges. Journal of computational chemistry 2011, 32 (8), 1769-1771. (28) Elliot, A. J.; Malek, G. A.; Lu, R.; Han, S.; Yu, H.; Zhao, S.; Wu, J. Z. Integrating Atomic Layer Deposition and Ultra-High Vacuum Physical Vapor Deposition for in Situ Fabrication of Tunnel Junctions. Review of Scientific Instruments 2014, 85 (7), 073904. (29) Zandvliet, H. J.; van Houselt, A. Scanning Tunneling Spectroscopy. Annual Review of Analytical Chemistry 2009, 2 37-55. (30) Feenstra, R. M.; Lee, J.; Kang, M.; Meyer, G.; Rieder, K. Band Gap of the Ge (111) C (2× 8) Surface by Scanning Tunneling Spectroscopy. Physical Review B 2006, 73 (3), 035310. (31) Mather, P.; Perrella, A.; Tan, E.; Read, J.; Buhrman, R. Tunneling Spectroscopy Studies of Treated Aluminum Oxide Tunnel Barrier Layers. Applied Physics Letters 2005, 86 (24), 242504-242504. (32) Ando, Y.; Hayashi, M.; Iura, S.; Yaoita, K.; Yu, C.; Kubota, H.; Miyazaki, T. Growth Mechanism of Thin Insulating Layer in Ferromagnetic Tunnel Junctions Prepared Using Various Oxidation Methods. Journal of Physics D: Applied Physics 2002, 35 (19), 2415. (33) Kurnosikov, O.; de Nooij, F.; LeClair, P.; Kohlhepp, J.; Koopmans, B.; Swagten, H.; de Jonge, W. StmInduced Reversible Switching of Local Conductivity in Thin Al 2 O 3 Films. Physical Review B 2001, 64 (15), 153407. (34) Barner, J.; Ruggiero, S. Tunneling in Artificial Al 2 O 3 Tunnel Barriers and Al 2 O 3-Metal Multilayers. Physical Review B 1989, 39 (4), 2060. (35) Åkerman, J. J.; Slaughter, J.; Dave, R. W.; Schuller, I. K. Tunneling Criteria for Magnetic-InsulatorMagnetic Structures. Applied Physics Letters 2001, 79 (19), 3104-3106. (36) Magtoto, N.; Niu, C.; Ekstrom, B.; Addepalli, S.; Kelber, J. Dielectric Breakdown of Ultrathin Aluminum Oxide Films Induced by Scanning Tunneling Microscopy. Applied Physics Letters 2000, 77 (14), 2228-2230. (37) Rippard, W.; Perrella, A.; Albert, F.; Buhrman, R. Ultrathin Aluminum Oxide Tunnel Barriers. Physical review letters 2002, 88 (4), 046805. (38) Perrella, A.; Rippard, W.; Mather, P.; Plisch, M.; Buhrman, R. Scanning Tunneling Spectroscopy and Ballistic Electron Emission Microscopy Studies of Aluminum-Oxide Surfaces. Physical Review B 2002, 65 (20), 201403. (39) Mather, P. Electronic Structure of Oxide Tunnel Barriers and Gaas – Ferromagnet Interfaces. Ph.D. Dissertation, Cornell University, 2006. (40) Teverovsky, A. In Degradation of Leakage Currents and Reliability Prediction for Tantalum Capacitors, Reliability and Maintainability Symposium (RAMS), 2016 Annual, IEEE: 2016; pp 1-7. (41) Oliver, B.; Tuttle, G.; He, Q.; Tang, X.; Nowak, J. Two Breakdown Mechanisms in Ultrathin Alumina Barrier Magnetic Tunnel Junctions. Journal of applied physics 2004, 95 (3), 1315-1322. (42) Lin, H.; Ye, P.; Wilk, G. Leakage Current and Breakdown Electric-Field Studies on Ultrathin AtomicLayer-Deposited Al 2 O 3 on Gaas. Applied physics letters 2005, 87 (18), 182904. (43) Tolpygo, S. K.; Amparo, D. Electrical Stress Effect on Josephson Tunneling through Ultrathin Alox Barrier in Nb/Al/Alox/Nb Junctions. Journal of Applied Physics 2008, 104 (6), 063904. 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

Page 22 of 23

(44) Henkelman, G.; Jónsson, H. Improved Tangent Estimate in the Nudged Elastic Band Method for Finding Minimum Energy Paths and Saddle Points. The Journal of Chemical Physics 2000, 113 (22), 99789985. (45) Hafner, J. Ab-Initio Simulations of Materials Using Vasp: Density-Functional Theory and Beyond. Journal of Computational Chemistry 2008, 29 (13), 2044-2078. (46) Paolo, G.; Stefano, B.; Nicola, B.; Matteo, C.; Roberto, C.; Carlo, C.; Davide, C.; Guido, L. C.; Matteo, C.; Ismaila, D.; Andrea Dal, C.; Stefano de, G.; Stefano, F.; Guido, F.; Ralph, G.; Uwe, G.; Christos, G.; Anton, K.; Michele, L.; Layla, M.-S.; Nicola, M.; Francesco, M.; Riccardo, M.; Stefano, P.; Alfredo, P.; Lorenzo, P.; Carlo, S.; Sandro, S.; Gabriele, S.; Ari, P. S.; Alexander, S.; Paolo, U.; Renata, M. W. Quantum Espresso: A Modular and Open-Source Software Project for Quantum Simulations of Materials. Journal of Physics: Condensed Matter 2009, 21 (39), 395502. (47) Li, N.; Sakidja, R.; Ching, W.-Y. Ab Initio Study on the Adsorption Mechanism of Oxygen on Cr2alc (0 0 0 1) Surface. Applied Surface Science 2014, 315 45-54. (48) Dharmawardhana, C. C.; Sakidja, R.; Aryal, S.; Ching, W. Y. In Search of Zero Thermal Expansion Anisotropy in Mo5si3 by Strategic Alloying. Journal of Alloys and Compounds 2015, 620 427-433. (49) Liu, Q.; Gong, Y.; Wilt, J. S.; Sakidja, R.; Wu, J. Synchronous Growth of Ab-Stacked Bilayer Graphene on Cu by Simply Controlling Hydrogen Pressure in Cvd Process. Carbon 2015, 93 199-206.


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


An illustration of the strategy employed in this work to study and minimize IL formation. Al was magnetron sputtered and transferred, under high vacuum, to the ALD chamber where an initial H2O pulse hydroxylated the Al surface just prior to the ALD reactant pulses (TMA and H2O). In situ STS studied the electron tunneling properties of the insulator to observe the significance and effect of the IL on the ALD Al2O3. Then with insight gained from AIMD simulations, the pre-ALD heating conditions (temperature and time) were varied to optimize ALD conditions to minimize the IL. 635x282mm (72 x 72 DPI)

ACS Paragon Plus Environment

Doping graphene with an atomically thin two dimensional molecular layer.

Al2O3 gate stacks using an Al2O3 barrier layer.

ITO structures through insertion of HfO2:Al2O3 (HAO) dielectric thin layer.

Interfacial nanobubbles on atomically flat substrates with different hydrophobicities.

Atomically thin nonreciprocal optical isolation.

Surface determination through atomically resolved secondary-electron imaging.

Plasmonics in atomically thin materials.

Bias dependence of tunneling magnetoresistance in magnetic tunnel junctions with asymmetric barriers.

Al2O3 thin films with high laser-induced damage thresholds.

Voltage-controlled quantum light from an atomically thin semiconductor.

TiO2 nanolaminate thin film encapsulation for organic thin film transistors via plasma-enhanced atomic layer deposition.

Mechanisms of photoconductivity in atomically thin MoS2.

Ultimate permeation across atomically thin porous graphene.

Probing the role of an atomically thin SiNx interlayer on the structure of ultrathin carbon films.

Protecting the properties of monolayer MoS₂ on silicon based substrates with an atomically thin buffer.

Tunable plasmons in atomically thin gold nanodisks.

Nonlinear phononics using atomically thin membranes.

Dehalogenation and coupling of a polycyclic hydrocarbon on an atomically thin insulator.

Atomically thin quantum light-emitting diodes.

Direct On-Chip Optical Plasmon Detection with an Atomically Thin Semiconductor.

Long-range electron tunneling.

An approach for an advanced anode interfacial layer with electron-blocking ability to achieve high-efficiency organic photovoltaics.

Al2O3 on Black Phosphorus by Atomic Layer Deposition: An in Situ Interface Study.

Photoswitchable organic field-effect transistors and memory elements comprising an interfacial photochromic layer.