Directed Atom-by-Atom Assembly of Dopants in Silicon - ACS Nano

7 days ago - Department of Physics and Astronomy, The University of Tennessee, Knoxville , Tennessee 37996 United States. ACS Nano , Article ASAP...
0 downloads 4 Views 1MB Size
Subscriber access provided by UNIVERSITY OF ADELAIDE LIBRARIES

Article

Directed Atom-by-Atom Assembly of Dopants in Silicon Bethany M Hudak, Jiaming Song, Hunter Sims, M. Claudia Troparevsky, Travis S. Humble, Sokrates T. Pantelides, Paul C Snijders, and Andrew R. Lupini ACS Nano, Just Accepted Manuscript • DOI: 10.1021/acsnano.8b02001 • Publication Date (Web): 11 May 2018 Downloaded from http://pubs.acs.org on May 12, 2018

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 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 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.

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 24 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 Nano

Directed Atom-by-Atom Assembly of Dopants in Silicon Bethany M. Hudak1,2,*, Jiaming Song1, Hunter Sims1,3, M. Claudia Troparevsky1, Travis S. Humble4, Sokrates T. Pantelides1,3, Paul C. Snijders1,5, and Andrew R. Lupini1,2 1. Materials Sciences and Technology Division, Oak Ridge National Laboratory, Oak Ridge, TN, 37831 USA. 2. The Institute for Functional Imaging of Materials, Oak Ridge National Laboratory, Oak Ridge, TN, 37831 USA. 3. Department of Physics and Astronomy, Vanderbilt University, Nashville, TN, 37235 USA. 4. Computational Sciences and Engineering Division, Oak Ridge National Laboratory, Oak Ridge, TN, 37831, USA. 5. Department of Physics and Astronomy, The University of Tennessee, Knoxville, TN, 37996 USA. Corresponding Author *[email protected]

ACS Paragon Plus Environment

1

ACS Nano 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 24

This manuscript has been authored by UT-Battelle, LLC under Contract No. DE-AC0500OR22725 with the U.S. Department of Energy. The United States Government retains and the publisher, by accepting the article for publication, acknowledges that the United States Government retains a non-exclusive, paid-up, irrevocable, world-wide license to publish or reproduce the published form of this manuscript, or allow others to do so, for United States Government purposes. The Department of Energy will provide public access to these results of federally sponsored research in accordance with the DOE Public Access Plan (http://energy.gov/downloads/doe-public-access-plan).

ACS Paragon Plus Environment

2

Page 3 of 24 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 Nano

Abstract The ability to controllably position single atoms inside materials is key for the ultimate fabrication of devices with functionalities governed by atomic-scale properties. Single bismuth dopant atoms in silicon provide an ideal case study in view of proposals for single-dopant quantum bits. However, bismuth is the least soluble pnictogen in silicon, meaning that the dopant atoms tend to migrate out of position during sample growth. Here, we demonstrate epitaxial growth of thin silicon films doped with bismuth. We use atomic-resolution aberration-corrected imaging to view the as-grown dopant distribution and then to controllably position single dopants inside the film. Atomic-scale quantum-mechanical calculations corroborate the experimental findings. These results indicate that the scanning transmission electron microscope (STEM) is of particular interest for assembling functional materials atom-by-atom because it offers both real-time monitoring and atom manipulation. We envision electron-beam manipulation of atoms inside materials as an achievable route to controllable assembly of structures of individual dopants. KEYWORDS. Atomic positioning; single-atom manipulation; scanning transmission electron microscopy (STEM); dopants; bismuth in silicon; quantum materials; quantum computing

ACS Paragon Plus Environment

3

ACS Nano 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 24

In Richard Feynman’s famous 1959 lecture, he proposed the idea of building nanoscale devices by directly arranging atoms ‘the way we want’ and pointed out the need for an improved electron microscope.1 These ideas were reflected in the recent suggestion to use an electron beam to directly manipulate materials at the atomic scale.2 However, such undertaking requires precise control over the size and position of the electron probe, which has only recently become possible through aberration correction. The current state of the art in manipulation at the atomic scale includes the positioning and imaging of single atoms,3 or even atomic scale surface lithography,4 using scanning tunneling microscopy (STM) and atomic force microscopy (AFM),5,6 performed upon two-dimensional surfaces. Recently, promising results for control of substitutional Si atoms found in two-dimensional single-layer graphene films have been demonstrated in a STEM.7–9 Ultimately one wishes to achieve controlled positioning of individual atoms, especially impurities, within a three-dimensional crystal, but no method for targeted manipulation of a specific atom within such a structure exists. Directly controlling the position of a single dopant atom offers opportunities for quantum engineering, such as quantum computing,10,11 optoelectronics,12 or catalysis.13,14 Bismuth was recently identified as a potentially superior candidate for donor-based quantum computing,15–20 and its large atomic number relative to Si makes Bi an ideal test subject for addressing difficulties related to single-atom manipulation in three-dimensional crystals. However, incorporation of Bi into an (isotopically purified) Si film is problematic 21–24 because Bi is almost insoluble in Si due to the large difference in covalent radius.25 Moreover, as evidenced by the efforts in the STM-based approach,4 progress in directed single atom manipulation is hindered by the limited experimental means to address single atoms within a solid. Even elementary questions such as “where exactly are the dopants?” often remain unanswered, and accurate

ACS Paragon Plus Environment

4

Page 5 of 24 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 Nano

placement and determination of dopant locations below the surface, therefore, remains a key challenge.26–28 Results and Discussion Here we demonstrate both the growth of heavily Bi-doped Si(111) and Si(100) thin films by solid phase epitaxy and the ability to controllably move and place the Bi dopants below the surface of a Si crystal at room temperature using STEM. This capability was indicated by two findings from our density functional theory (DFT) calculations: (1) the strain induced by the Bi dopant lowers the Si vacancy enthalpy of formation allowing for Si vacancies to be preferentially created adjacent to the large Bi atoms, and (2) there is no significant energy barrier for the Bi atom to hop into these vacancies once they have been formed. Using the electron beam, we exploit this phenomenon to direct and place the Bi dopants in specific columns within an oriented Si crystal. We explored the effect of the Sub-Ångstrom-sized electron beam on single dopants in aberration-corrected STEM at various accelerating voltages between 60 and 200 kV. By using an accelerating voltage of 160 kV, which is close to but above the knock-on damage threshold of silicon, we achieve control of dopant placement while reducing damage to the surrounding crystal. The method presented in this work is an indispensable step toward the fabrication of functional single-atom devices through the direct positioning of single dopants within a three-dimensional material. Figure 1 shows Z-contrast STEM images of a Bi-doped Si(111) layer grown using nonequilibrium solid-phase epitaxy (see Methods). Plan view images along the [111] direction (Figure 1a) show well dispersed Bi atoms distributed over the sample. A cross-section view (Figure 1b) ([110] direction) confirms the presence of a nominally 10 monolayer (ML) thick Bi-

ACS Paragon Plus Environment

5

ACS Nano 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 24

doped Si layer (1 ML = 0.31 nm). When imaged at high magnification in a 200 kV STEM (Figure 2a) we find that some dopants exhibit ‘streaks’ in their intensity, and others are only partially visible. This effect is not rooted in microscope instabilities, but arises because the dopant atoms move during the image scan 29. To confirm this inference, Figure 2b shows a selection of frames extracted from a fast sequence of larger images taken at 200 kV. Supporting Figures S1-S3 show a similar selection of frames, but for images acquired at 60 kV, 100 kV, and 160 kV. The apparent interaction of the beam with the dopants is enabled by beam-induced Si dynamics. For a 200 kV accelerating voltage, the maximum amount of energy that can be transferred to a Bi atom in a single elastic collision is less than 2.5 eV,30 whereas the diffusion barrier for Bi inside Si is about 4.1 eV 31 implying that, even at the highest beam energy used here, direct energy transfer to Bi is insufficient to cause the Bi mobility. However, because the Si atoms are lighter, significantly more energy (up to about 18 eV at 200 kV) can be transferred to Si in a single elastic collision, for which the bulk knock-on damage threshold is roughly 12.5 eV, corresponding to an accelerating voltage of about 140 kV.32 At accelerating voltages below the knock-on damage threshold of Si (100 kV and 60 kV) we see almost no dopant movement (Supporting Figures S2-S3). Using DFT calculations we found that the Si vacancy enthalpy of formation is significantly lower at sites adjacent to substitutional Bi atoms (1.5 eV versus 3.5 eV in pure Si) due to the strain induced by the dopant (Supporting Figure S4). Further, we used the nudged elastic band method to calculate the energy barrier for a Bi to hop into such a vacancy to be only about 40 meV, meaning that such a hop readily occurs thermally. When a single atomic column is irradiated by an electron beam with an energy above the knock-on damage threshold of the crystal, there is a high probability that an atom will be displaced from its lattice site. In the case that all atoms had the same bonding environment, they would have a similar probability of

ACS Paragon Plus Environment

6

Page 7 of 24 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 Nano

being displaced, depending on the local irradiation (taking into account the 3D propagation of the beam). In general, surface atoms are most likely to be removed from the lattice because they have fewer bonds. In the case where the Si is strained by a neighboring Bi atom, reducing the vacancy enthalpy of formation nearby, these atoms are more likely to be displaced, or may even attract existing vacancies. We conclude that the electron beam can easily generate Si vacancies adjacent to substitutional Bi atoms and, through regulated beam motion, enable controlled movement of single Bi atoms to adjacent lattice sites. In Figure 3 we demonstrate the experimental ability to control the dopant movement and selectively place single Bi atoms within a Si crystal. The frames displayed in Figure 3 are extracted from the video shown in Supporting Movie 1. To direct the dopant movement, we used controlled positioning of the electron beam operated at 200 kV. First a high-angle annular dark field (HAADF) image is acquired (Figure 3a) to act as a survey image. With the scanning stopped, the electron beam is manually positioned near a column containing a Bi atom. A path is then traced by stepping the electron beam from column to column (Figure 3c-f). A survey image afterwards reveals that the Bi atom has followed the movement of the beam, ending up in the designated position (Figure 3b). In fact, two Bi atoms are moved into the column where the beam was last positioned: one that was targeted for directed placement and one that was located in an adjacent column before beam positioning. The intensity of this column after Bi positioning is higher than either of the Bi-containing columns previously, indicating that two Bi atoms are occupying the column. Remarkably, this directed movement is highly reproducible, as shown in Supporting Movies 1-3. Significantly, the Bi atom appears to be on the Si column and stable at the new position. Nudged elastic band DFT calculations (Supporting Figure S5) show that this

ACS Paragon Plus Environment

7

ACS Nano 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 24

behavior is not consistent with the motion of Bi atoms along the surface, in which case they are adsorbed in hollow sites (between columns) on the surface. We now use the electron beam to assemble atomic-scale structures. In Figure 4, we show that several Bi atoms in a single region can be individually repositioned into neighboring columns, demonstrating precise control of dopant placement within a two-dimensional projection of a three-dimensional crystal. In Figure 4, some untargeted Bi atoms randomly move due to repeated scans at 200 kV, consistent with our previous observations. However, this random movement does not negate the fact that we demonstrate controllable positioning of targeted Bi dopants. By lowering the beam energy closer to the knock-on damage threshold of Si, we reduce the vacancy formation probability per incident electron, allowing for better placement of single atoms. In Figure 5, we have reduced the accelerating voltage to 160 kV to draw a triangular dopant lattice composed of six regularly spaced Bi atoms in projection (Figure 5b). Supporting Figure S6 shows an estimate of the Bi dopant depth based on a 10 nm thick Si crystal. Supporting Movie 3 shows the process of manually assembling this more complex atomic structure. While there are still technical hurdles towards optimizing the accelerating voltage of the microscope and dopant concentration of the samples, these results demonstrate that the electron microscope can be used as a tool for building device architectures at the atomic (Figure 4) and nanoscale (Figure 5) inside a three-dimensional Si matrix. Although it is reasonable to assume that we are creating a chain of vacancies through the Si crystal, we never observe a Bi atom moving freely back along the movement path traced by the electron beam. This suggests one of two further steps is occurring: Either an interstitial Si, created through the knock-on process, fills in the vacancy left by the Bi atom; or the vacancy rapidly diffuses away from the dopant atom. Point-defect studies published in the literature

ACS Paragon Plus Environment

8

Page 9 of 24 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 Nano

suggest that either mechanism is plausible. However, the conventional wisdom regarding pointdefects in Si has been acquired primarily through methods that apply a more uniform amount of energy to a macroscopic Si wafer.33–35 Here, we use a focused electron beam to supply energy to a very small portion of the sample, approximately one atomic column at a time. The system will therefore lack the energy for long-range defect migration, either away from or towards the Bi dopant, thus the interstitial Si will be near enough to allow vacancy-interstitial recombination, filling in the vacancy left by the Bi and preventing backward diffusion.33 The accurate placement of donor atoms currently presents a major challenge for the synthesis of quantum devices.11 Here we demonstrate the ability to control the positioning of individual Bi atoms within a two-dimensional projection of a solid, providing a powerful method for fabricating atomically precise samples of doped silicon. Supporting Figure S7 shows aspositioned dopants after time without electron beam exposure. The lack of movement of the dopants when not exposed to the electron beam demonstrates the potential of this method to produce stable structures. Supporting Table S1 includes the success rate of dopant positioning. Based on 226 recorded attempts at dopant positioning, we achieve a success rate of about 75%, although the combination of imaging and motion available in the STEM allows multiple attempts to move dopants to the desired location. Further development is needed before such proof-of-concept results will find practical application. For example, silicon donor qubits are nanoscale devices that store quantum information in the spin state of the donor atom and use surface control electrodes to manipulate the spin state.36 Thus, the ability to reproducibly prepare accurately doped samples will prove important for the design and fabrication of the surface electrodes.37 More specifically, highfidelity operation of these devices relies on atomically precise donor placement in a plane 5-10

ACS Paragon Plus Environment

9

ACS Nano 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 24

nm below the silicon oxide surface of the device. Incorporating STEM-guided donor positioning into the synthesis of Bi-doped silicon substrates would enable methods for patterning donor configurations within samples. High-resolution STEM imaging, as presented here, requires the use of electron-transparent samples with a thickness of 10-20 nm. However, similar techniques can potentially be applied to cases where it is harder to image the dopants. Our approach need not depend on direct imaging but may instead use pre-defined programmable or adaptive sweeps of the electron beam to drive dopants to specified points. Another route to incorporating this technique into a functional device, while avoiding the damaging (S)TEM sample preparation steps, would be through the use of doped thin-film Si membranes fabricated from silicon-oninsulator (SOI) wafers.38 These single-crystal Si films etched from SOI can be transferred to a foreign host material for fabricating thin-film transistors.39,40 Conclusions We have presented the ability to incorporate significant concentrations of Bi in Si films, and, using aberration-corrected STEM and insights from density functional theory calculations, we have introduced a technique to position dopant atoms within a three-dimensional crystal, with control of the movement in two dimensions. In the future, tilting the sample to different zone axes should allow three-dimensional location and positioning through control in different projections. However, the challenges inherent to creating a functional single-atom qubit in this way are still abundant and overcoming such obstacles will require collaboration across multiple disciplines. Here we demonstrated that STEM can be used to exploit vacancy-mediated motion to directly position substitutional Bi dopants in Si, below the surface of the crystal. This ability represents one step toward realizing a single-atom qubit device built from direct atom positioning. Sub-Ångstrom resolution affords us precise control over the electron beam location,

ACS Paragon Plus Environment

10

Page 11 of 24 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 Nano

and samples are monitored in near real-time. This process could also be extended to other systems in which vacancies can be directly created in the host material. More precise control over the beam scanning and faster feedback will allow automation of this process. The directed motion of single-atom dopants within a three-dimensional system represents a promising advance in atom-by-atom fabrication, enabling the tuning of material properties through controlled dopant positioning. Methods and Experimental Solid-phase epitaxial growth of Bi-doped Si films Bi-doped Si films were grown both on Si(111) and Si(100) substrates. The Si(111) and Si(100) surfaces were prepared with a (7x7) and (2x1) reconstruction, respectively, using standard ultrahigh vacuum degassing and flashing procedures established for these surfaces. The surface quality was examined by STM and low energy electron diffraction (LEED). On the Si(111)-7x7 surface, approximately 0.5 monolayer (ML) of Bi atoms (i.e. ~3.9 10^14 Bi atoms/cm2) was subsequently deposited with the Si substrate held at 500 °C, using a standard effusion cell held at 470 °C. This Bi dose is higher than would be desirable for qubit applications, but was chosen in order to have a sufficient doping concentration for good visibility and easy dopant structure assembly. Next, a nominally ~10 ML Si capping layer was grown by co-depositing Si and Bi with the substrate held at room temperature. To recrystallize the amorphous capping layer, the sample was annealed at approximately 575 °C for 15 s until LEED showed the surface structure had recrystallized. The resulting surface showed a (√3×√3) diffraction pattern evidencing that some of the Bi atoms had segregated to the surface and

ACS Paragon Plus Environment

11

ACS Nano 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 24

formed a (√3×√3)R30° reconstruction, and the images in Figure 1 confirm that some remain in the film. To reduce the segregation of Bi inside the silicon films on the Si(100) surface, Bi nanolines were grown on Si (100) at 570 °C.41 Subsequently, a thin crystalline Si film with a thickness of about 2.3 nm was formed on this Bi nanoline surface by recrystallizing the room-temperature deposited amorphous Si at 434 °C for 5 s. This destroys the Bi nanolines and gives rise to a locally high doping level of Bi atoms inside the thin film. To protect the Bi doped crystalline Si film, a thick amorphous Si layer was subsequently deposited at room temperature. Sample Preparation Cross-section STEM sample preparation consisted of taking the as-grown sample, cutting it in half to produce a sandwich, which is then sliced into sections and glued to a support grid. The samples are mechanically polished to approximately 10 microns, then ion-milled to electron transparency. We are deeply indebted to J. Meyers for this work. Immediately before imaging, the samples were baked at a nominal 140 ˚C in vacuum overnight to reduce surface contamination. STEM Aberration-corrected STEM was performed in a Nion UltraSTEM 200 operated at 200 kV, and at 160 kV; and in a Nion UltraSTEM 100 operated at 100 kV and at 60 kV. Note that the electron beam energy in a (S)TEM is typically in the range of 30-300 kV, while binding energies are of the order of a few eV. However, conservation of momentum implies that only a very small fraction of the kinetic energy can be transferred from the fast electron to a nucleus in a single

ACS Paragon Plus Environment

12

Page 13 of 24 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 Nano

elastic collision. Electronic interactions or multiple collisions can also occur. However, the low beam current (