WorksPhysics and Astronomy
First-principles dissociation pathways of BCl on the Si(100)-21 surface
Abstract
One of the most promising acceptor precursors for atomic-precision -doping of silicon is BCl. The chemical pathway, and the resulting kinetics, through which BCl adsorbs and dissociates on silicon, however, has only been partially explained. In this work, we use density functional theory to expand the dissociation reactions of BCl to include reactions that take place across multiple silicon dimer rows, and reactions which end in a bare B atom either at the surface, substituted for a surface silicon, or in a subsurface position. We further simulate resulting scanning tunneling microscopy images for each of these BCl dissociation fragments, demonstrating that they often display distinct features that may allow for relatively confident experimental identification. Finally, we input the full dissociation pathway for BCl into a kinetic Monte Carlo model, which simulates realistic reaction pathways as a function of environmental conditions such as pressure and temperature of dosing. We find that BCl is broadly dominant at low temperatures, while high temperatures and ample space on the silicon surface for dissociation encourage the formation of bridging BCl fragments and B substitutions on the surface. This work provides the chemical mechanisms for understanding atomic-precision doping of Si with B, enabling a number of relevant quantum applications such as bipolar nanoelectronics, acceptor-based qubits, and superconducting Si.
precision doping of Si with B, enabling a number of relevant quantum applications such as bipolar nanoelectronics, acceptor-based qubits, and superconducting Si.
Introduction
Atomic precision advanced manufacturing (APAM) techniques can be used to realize a wide variety of novel quantum and electronic devices in silicon. [1, 2, 3] In this technique, a hydrogen or halogen terminated silicon surface is exposed to a precision lithography tool (usually, but not always, [4, 5] scanning tunneling microscopy (STM)), which depassivates a region of interest in ultrahigh vacuum (UHV). This region is then exposed to a precursor gas, which selectively adsorbs onto the exposed silicon, but not the surrounding resist. The precursor gas will dissociate on the surface, ideally leading to incorporation of the dopant atom, often at carrier densities exceeding . [6]
Most of the work on APAM so far has focused on using a precursor gas of phosphine for donor doping of silicon, [7, 8, 9, 10] but there has been recent interest in expanding the range of applications available by introducing alternative donors such as As, [11, 12] as well as a variety of precursors with the goal of acceptor incorporation such as diborane (), [13, 14] aluminum trichloride (), [15] boron trichloride (), [16, 17] and organics such as trimethyl and triethyl aluminum. [18]
Of these, has shown particular promise as an APAM acceptor precursor, achieving an active carrier concentration of . [17] To make further progress on creating quantum devices with this precursor, it is necessary to develop a comprehensive understanding of how decomposes on the surface. Campbell et al. used density functional theory (DFT) to establish the initial decomposition pathway of on the silicon surface, [19] demonstrating that the molecule adsorbs without a reaction barrier, and quickly decomposes with low reaction barriers (). Based on this analysis, it was then predicted that doping under certain conditions could be used to achieve deterministic single acceptor doping with atomic precision lattice site. [20]
This analysis was limited, however, by the aim of single acceptor incorporation.
We only considered decomposition pathways taking place entirely along a single dimer row and ending with BCl fragments. This paper extends these reactions to take place across multiple silicon dimers and reach an isolated B, providing the theoretical basis for predicting decomposition pathways which end in multiple acceptors incorporated in the silicon, particularly -doping. This will be useful for predicting ideal APAM conditions for maximizing B incorporation, with useful applications such as improved coherence times in multi-atom qubits [21, 22] and superconducting Si regions.[23] Furthermore, comparing these dissociation pathways with experimentally observed dissociation on the silicon surface is difficult without generation of STM images matching configurations. Wyrick et al. have recently generated STM images for the full range of fragments on the Si(100)- surface and have achieved good agreement with experimental characterization.[24] This work will enable similar experimental analysis of fragments on Si.
In this paper, we use DFT to further expand the possible range of decomposition reactions, including reactions that take place across multiple silicon dimer rows and reactions that lead to subsurface B. We simulate STM images corresponding to common fragments on the surface, which can be compared with experimental results for identification purposes. We then develop a kinetic Monte Carlo (KMC) simulation, which allows us to predict the most common fragments during decomposition as a function of dose pressure, time, and temperature, as well as the following anneal temperature. We predict surface coverages of B roughly in line with experimentally observed acceptor densities. We further predict that will be the dominant surface feature during typical -doping conditions. This is largely limited by the space to decompose, however, and in systems where there is more room to dissociate, bridging BCl is often observed, with the most common location for a single B to end as a substitution for a surface silicon. With this knowledge, atomic precision doping of silicon using can be fine tuned to provide optimal electronic properties for desired applications.
Methodology
Reaction Barrier Calculations
To compute reaction barriers between configurations, we use the Nudged Elastic Band (NEB) method, as implemented in QUANTUM-ESPRESSO,[25] and use norm-conserving pseudopotentials from the PseudoDojo repository[26] and the Perdew-Burke-Ernzerhof exchange correlation functional.[27] Kinetic energy cutoffs of 50 Ry and 200 Ry are used for the plane wave basis sets used to describe the Kohn-Sham orbitals and charge density, respectively, and a Monkhorst-Pack grid is used to sample the Brillioun zone.[28]
fragment calculations are performed on a supercell of a seven-layer thick Si(100)– slab with a 20 Å vacuum region, where a single unit cell has a size of 3.87 Å × 3.87 Å. Matching the experimentally measured Si structure, we model the Si surface with alternating buckled Si dimers. On the other end of the slab, the Si dangling bonds are passivated with Se atoms to prevent spurious surface effects. The bottom four layers of the slab are frozen and the geometry of the surface layers along with the adsorbate are relaxed until the interatomic forces are lower than 50 meV/Å.
Scanning Tunneling Microscopy Calculations
Simulated Scanning Tunneling Microscopy (STM) images of the configurations calculated were generated by QUANTUM-ESPRESSO’s postprocessing methodologies, which use the Tersoff-Hamann approximation.[29] We simulate constant current STM images, with a charge threshold of e and the voltages stated in our figures. It should be noted that these voltages are likely to somewhat underestimate real world energies,[30] and that this approximation does not take into account the effect of band bending on the substrate from the electric field introduced by the STM tip. In our calculations of reaction barriers for the dissociation of , the silicon dimers on our surface have an alternating tilt up and down– the lowest energy reconstruction. When we simulate STM images, however, we hold all the silicon dimers which are not directly connected to the relevant fragment flat. This mimics the rapid oscillation between tilt directions which leads to silicon dimers looking flat in experimentally obtained STM images. [31] In the main paper, we simulate STM images at normal resolution, but in the Supporting Information with all the simulated STM images, we also include images with an added gaussian blur to mimic realistic experimental noise. This approach mimics the STM simulation techniques used by Wyrick et al. in their analysis of PH fragments on the silicon surface. [24]
Kinetic Monte Carlo Simulations
Finally, we use a kinetic Monte Carlo (KMC) model[32, 33] as implemented in the KMCLib package[34] to determine the probability of incorporation. Our KMC model uses transition rates based on the Arrhenius equation , [35] where is transition rate, is the attempt frequency, is the reaction barrier found from our earlier DFT calculations, is the Boltzmann constant, and is the temperature. We set all attempt frequencies to as a reasonable order of magnitude estimate based on an analysis of attempt frequencies for the dissociation of phosphine on silicon, [36] which we assume to be analogous. We calculate the effusive flow rate of molecules landing on any particular silicon dimer as , where is the pressure of the incoming precursor gas, is the area of impingement, taken here as a single silicon dimer, is the mass of the precursor gas.
Each KMC calculation is repeated 100 times with different random seeds, and the sample mean of the results is reported. The code for these simulations has been uploaded to an open source repository. [37]
Results
Reaction Barriers
We begin by calculating novel reaction barriers for dissociation on the silicon surface. These reactions expand on our previous work [19] in three main directions. First, we consider reactions which take place across multiple silicon dimer rows. This provides another avenue for Cl to shed from the initial molecule and sometimes even for fragments to switch dimer rows as shown in Figure 1. Second, we discover a new configuration of BCl which has not been previously identified, a bridging BCl which is located entirely along one silicon dimer (see, e.g., Figure 1e). This is analogous to the previously seen location of PH molecules. [36]
Figure 1. Schematic diagrams of reaction barriers for BCl fragments dissociating on the Si surface. Panels a-d depict reactions which take place across multiple silicon dimer rows, while e-h represent single silicon dimer row reactions. Silicon atoms which have red underlines represent the lower end of the tilted silicon dimer. represents the reaction barrier of each reaction.
Finally, we extend our analysis beyond BCl fragments to look at decomposition into individual B atoms in Figure 2. The range of possible locations for a single B can be quite high as it migrates from residing solely on the surface to subsurface locations. We chose sites for B incorporation matching previous results for the initial incorporation of P into the Si surface. [36] While the locations we tested are not comprehensive, we looked at B locations both adsorbed on the surface often in similar configurations as previously shown fragments (as in Figure 2a and Figure 2b), one layer within the subsurface of the material (Figure 2c and Figure 2d), and B substituting for a silicon atom on the surface layer (Figure 2e and Figure 2f). Reaction barriers for reaching an isolated B are of the order eV, well within the range of typical reaction barriers seen in the dissociation steps up to BCl. We therefore expect that dissociation all the way to individual B atoms on the surface will be possible at reasonable anneal temperatures (e.g. 350 °C).
Figure 2. Continued schematic diagrams of reaction barriers for fragments dissociating on the Si surface. Silicon atoms which have red underlines represent the lower end of the tilted silicon dimer. The dashed purple lines connecting to the B atom in panels c and d indicate that the B is actually subsurface in these models. See Figure 3 for a ball and stick diagram of the B location.
Due to the combinatorial explosion of possible reactions once the boron reaches the subsurface, we have stopped our modeling of reaction barriers at this first level of boron atoms at or slightly below the surface. These systems all exhibit extremely strong ( eV) adsorption energies and high reverse reaction barriers (i.e. the barrier of moving the reaction in the opposite direction as shown within Figure 1- Figure 2). We therefore
assume within our models that once a molecule dissociates to the point of a single B atom, that B atom will remain on the surface throughout processing and for subsequent electrical measurements, will behave as an electrically active dopant.
Additional reaction barriers can also be found in the supporting information.
STM simulations of surface configurations
To aid in the experimental identification of the predicted configurations, we have performed STM simulations of all configurations. In Figure 3, we display the STM simulations for the most likely end points of decomposition, i.e. bridging BCl and B fragments, as well as subsurface B, and B substitutions into the Si lattice.
Figure 3. Scanning tunneling microscopy (STM) simulations of BCl and B fragments that serve as likely endpoints for BCl dissociation. In each row, we include the schematic view of what the BCl fragment looks like in the style of Figure 1-Figure 2, a ball and stick view of the fragment from the top and the bottom, and finally the simulated STM image at -2.0 V and +2.0 V.
Fortunately for later experimental verification, the majority of the simulated STM structures appear to have distinct features which could be used for identification. , shown in Figure 3a, has a bright barbell like pattern. To contrast, the bridging BCl fragment across multiple dimer rows, shown in Figure 3b, has barbell like patterns where a bright Cl atoms is immediately next to a darker depression (this pattern is particularly pronounced in the positive bias images). The STM patterns with the most similarity are the bridging BCl on a single Si dimer (Figure 3c) and the bridging B along a single Si dimer (Figure 3d) Both display a single bright spot corresponding to the elevated atoms.
In the bridging BCl case, the Cl atom is slightly offset from the silicon dimers, however, and the bright spot is thus centered between the silicon dimer rows. In contrast, the bridging B is entirely centered on the silicon dimer and the bright spot aligns neatly with the nearby dimers. This means that it should be possible to distinguish between , BCl and a bare B, even if they are occupying essentially the same position along the silicon surface. The subsurface B, in Figure 3e is the hardest to distinguish in our simulations, with the majority of the surface appearing to mimic a normal bare silicon surface. The subsurface B can be identified in negative bias images solely by a slightly less bright silicon dimer, whereas at positive bias, we begin to see a small gap between the relevant top silicon atoms. Finally, for a B substitution, shown in Figure 3f for a Si
atom on the surface, the B atom appears significantly less bright than the surrounding atoms, which is again clearer at positive bias. When switching from negative to positive bias, the observed trends are only exaggerated at lower and higher biases, as can be seen in the supporting information. These results emphasize the utility of taking multiple voltages for STM images when attempting to identify specific fragments on the surface.
In the supporting files for this work, we include STM simulation data and images for all the 26 dissociation fragments we have simulated.
Kinetic Monte Carlo Simulations of
Finally, we take the reaction barriers calculated from Section 3.1 and input them (along with the previously calculated reaction barriers for [19]) into a kinetic Monte Carlo (KMC) simulation. This method works by surveying the available reactions from an initial setup, enumerating all possibilities from the reaction list, and then choosing a reaction from the probability distribution determined by the reaction barriers (as well as a given temperature the simulation takes place at). This allows us to predict how will dissociate on the surface of silicon accounting for real stochasticity of kinetic incorporation. [38] Our aim with these simulations is to predict what will be the dominant fragments of on the silicon surface after a realistic dosing and incorporation schedule.
We initially simulate a system at low coverage, 0.03 L, using a dosing pressure of Torr for 60 s at room temperature, followed by an anneal at temperatures of , , or for 30 s. This allows us to examine what dissociation fragments we expect to see when each of the molecules is allowed to fully dissociate without necessarily running into steric hindrance from nearby fragments. In Figure 4, we show that bridging and Cl fragments tend to dominate on the surface at the end of the dosing run at nearly equal proportions. This can be attributed to the fact that the first Cl atom automatically detaches from the in the process. of adsorbing onto the silicon substrate,[19] thus creating an equal number of BCl and Cl molecules. Beyond this point, however, higher temperatures are needed to further dissociate the molecule. After annealing, the BCl molecules fully dissociate into individual B atoms, which are most likely to take the form of a substitution into the silicon lattice. By comparing between the dose and anneal plots, it becomes clear that essentially all of the BCl on the surface is being converted to B substitutions. This tendency toward B substitution can be attributed to the comparatively low barrier (1.06 eV) for a substitution, compared to all other single B reactions, which require overcoming barriers eV. Furthermore the 2 Cl atoms of the BCl fragment are shed and the percentage of Cl on the surface is approximately three times the amount of B. This matches the initial stoichiometric ratio of BCl and indicates that Cl desorption is not yet reached at these anneal temperatures within our model.
Figure 4. The percentage of the silicon surface taken up by different BCl fragments at the end of the (a) dose and (b) anneal, with a dependence on the anneal temperature. These results come from an initial dosing of 0.03 L, with a dosing pressure of Torr for 60 s at room temperature followed by a 30 s anneal at varying temperatures.
We then move on to simulating the formation of a -doped layer of B, using realistic monolayer coverages of BCl. We follow the settings of Dwyer et al. [17], using a dosing pressure of Torr with different times to create a 1.5 L coverage in Figure 5 and 3.6 L coverage in Figure 6.

Figure 5. The percentage of the silicon surface taken up by different BCl fragments at the end of the (a) dose and (b) anneal, with a dependence on the anneal temperature. These results come from an initial dosing of 1.5 L, with a dosing pressure of 1 × Torr for 150 s at room temperature followed by a 30 s anneal at varying temperatures.

Figure 6. The percentage of the silicon surface taken up by different fragments at the end of the (a) dose and (b) anneal, with a dependence on the anneal temperature. These results come from an initial dosing of 3.6 L, with a dosing pressure of Torr for 360 s at room temperature followed by a 30 s anneal at varying temperatures.
The most notable difference in these systems is that steric hindrance becomes a significant factor in the endpoint of BCl dissociation. At coverages L, the number of sites that a BCl molecule can dissociate into decreases rapidly. As a BCl adsorbs onto the silicon surface, it will automatically shed a Cl atom to the nearby silicon atom on the same dimer, forming BCl. Beyond this initial adsorption, however, it can be difficult to find room to shed the remaining Cl when most of the nearby sites are also occupied by similar BCl + Cl pairs. We do include Cl desorption in our model as an option, but with a reaction barrier of 4.85 eV, this does not become a significant factor until high temperature anneals. This means that each adsorption site inherently limits the number of sites that can be dissociated into by further BCl molecules. In Figure 5, we see that this initial adsorption process is dominant during the room temperature dosing, with the only significant BCl fragment found after dosing being BCl. This matches what we previously saw for 0.03 L dosing. When the anneal is applied, however, the BCl fragments have significantly less room to dissociate than in the 0.03 L case, due to the greater saturation of dosing sites from the higher Langmuir dose. After the anneal, the majority of fragments are still BCl; however, the percentage of the surface occupied by moves from to . The that does dissociate primarily moves into the form of two dimer bridging , at around 5% of the surface. This necessarily results in an increase in the percentage of on the surface from to , filling the remaining surface sites. As anneal temperatures increases, however, the number of single B increases slightly, also slightly increasing the proportion of . The dominant form of single B is still B substitutions into the silicon surface, but at 500 C, there is a non-negligible amount of subsurface B. Comparing this result to 0.03 ML coverage, we can infer that B substitution is the energetically easiest path, but as the surface sites saturate, moving to the subsurface becomes a viable option.
These same trends are amplified in the 3.6 L dosing. Here takes up a greater percentage of the surface, . By the end of the dose, the and sites take up essentially the entirety of the surface. This means, however, that there is essentially no room for the to dissociate, leading to the percentage of on the surface staying almost exactly the same after annealing. This indicates that if the goal is to maximize dissociation to B, it may be preferable to perform sub-Langmuir doses, as this provides more room for eventual dissociation. Another avenue, explored in previous work for phosphine, [38] may be to perform the dose at higher temperatures, which encourages dissociation reactions to take place during the dosing phase, before other precursor molecules have saturated the surface.
Notably, both the 1.5 L and 3.6 L doses create a total coverage of different fragments at ML at the end of the anneal. These numbers are roughly double the reported sheet densities of Dwyer et al.[17] of (0.26 ML). Given that many of these fragments are still in the form of , which has the relatively weakest adsorption energies of all fragments, it is somewhat likely that not all of them will electrically incorporate during the subsequent silicon growth. Furthermore, B is known to dimerize in silicon, which renders it electrically inactive.[39] These mechanisms may account for the diminished values seen for electrically active B in experiment compared to our results.
While we have focused on -layer formation for these kinetic simulations, the associated code[37] could also be used to generate the probability of incorporation for single or few acceptor systems, directly impactful for quantum devices.
Conclusions
In this work, we used DFT calculations to explore a significantly expanded dissociation pathway for on the Si(100)- surface. We find significant new configurations which had not been previously explored such as the single Si dimer bridging and also expand to dissociation to a lone B atom. This dissociation pathway remains achievable with modest reaction barriers, usually on the order of 1.5 eV. We furthermore explore what each of these fragments would look like on the surface under STM, finding distinct features for each. Finally, we use these reaction barriers to develop a kinetic model that allows us to predict how actually dissociates on the surface and what are the dominant end points. After room temperature dosing, fragments are by far the most common. After an anneal, if the has room to fully dissociate, such as the case of 0.03 L dosing, then the most likely outcome will be a B substitution into the silicon lattice. If the BCl becomes trapped by neighboring fragments, however, as in the 1.5 L and 3.6 L dosing cases, it is much less likely to fully dissociate after an anneal, with only a minority reaching substitutional boron. This is analogous to previous work with phosphine, [38], which has shown that dosing at lower pressures for longer times and/or higher temperatures can be used to increase the rate at which phosphine dissociates before being hampered by adsorption of PH fragments at nearby sites. This work allows us to fully understand the dissociation pathway of BCl on silicon and allows for advances in the creation of a number of unique quantum applications such as bipolar nanoelectronics, acceptor-based qubits, and superconducting Si.
Acknowledgement
We would like to thank Azadeh Farazeh, Robert Butera, Chris Allemang, Devika Mehta, Tom Sheridan, and Scott Schmucker for useful discussions surrounding the results of this work.
This work was supported by the LPS Qubit Collaboratory and by the Laboratory Directed Research and Development (LDRD) program at Sandia National Laboratories under project 229375. This work was performed, in part, at the Center for Integrated Nanotechnologies, an Office of Science User Facility operated for the U.S. Department of Energy (DOE) Office of Science. Sandia National Laboratories is a multi-mission laboratory managed and operated by National Technology & Engineering Solutions of Sandia, LLC (NTESS), a wholly owned subsidiary of Honeywell International Inc., for the U.S. Department of Energy’s National Nuclear Security Administration (DOE/NNSA) under contract DE-NA0003525. This written work is authored by an employee of NT-ESS. The employee, not NTESS, owns the right, title and interest in and to the written work and is responsible for its contents. Any subjective views or opinions that might be expressed in the written work do not necessarily represent the views of the U.S. Government. The publisher acknowledges that the U.S. Government retains a non-exclusive, paid-up, irrevocable, world-wide license to publish or reproduce the published form of this written work or allow others to do so, for U.S. Government purposes. The DOE will provide public access to results of federally sponsored research in accordance with the DOE Public Access Plan.
References
- [1]Ward, D. R.; Schmucker, S. W.; Anderson, E. M.; Bussmann, E.; Tracy, L.; Lu, T.-M.; Maurer, L. N.; Baczewski, A.; Campbell, D. M.; Marshall, M. T.; others Atomic precision advanced manufacturing for digital electronics. EDFA Technical Articles 2020, 22, 4–10.DOI
- [2]Bussmann, E.; Butera, R. E.; Owen, J. H.; Randall, J. N.; Rinaldi, S. M.; Baczewski, A. D.; Misra, S. Atomic-precision advanced manufacturing for Si quantum computing. MRS Bulletin 2021, 46, 607–615.DOI
- [3]Schofield, S. R.; Fisher, A. J.; Ginossar, E.; Lyding, J. W.; Silver, R. M.; Fei, F.; Namboodiri, P.; Wyrick, J.; Masteghin, M. G.; Cox, D. C.; others Roadmap on Atomic-scale Semiconductor Devices. Nano Futures 2025,DOI
- [4]Katzenmeyer, A. M.; Dmitrovic, S.; Baczewski, A. D.; Campbell, Q.; Bussmann, E.; Lu, T.-M.; Anderson, E. M.; Schmucker, S. W.; Ivie, J. A.; Campbell, D. M.; others Photothermal alternative to device fabrication using atomic precision advanced manufacturing techniques. Journal of Micro/Nanopatterning, Materials, and Metrology 2021, 20, 014901–014901.
- [5]Constantinou, P.; Stock, T. J.; Tseng, L.-T.; Kazazis, D.; Muntwiler, M.; Vaz, C. A.; Ekinci, Y.; Aeppli, G.; Curson, N. J.; Schofield, S. R. EUV-induced hydrogen desorption as a step towards large-scale silicon quantum device patterning. Nature Communications 2024, 15, 694.DOI
- [6]Keizer, J. G.; Koelling, S.; Koenraad, P. M.; Simmons, M. Y. Suppressing segregation in highly phosphorus doped silicon monolayers. ACS nano 2015, 9, 12537–12541.DOI
- [7]Schofield, S.; Curson, N.; Simmons, M.; Rueß, F.; Hallam, T.; Oberbeck, L.; Clark, R. Atomically precise placement of single dopants in Si. Physical review letters 2003, 91, 136104.
- [8]Rueß, F. J.; Pok, W.; Reusch, T. C.; Butcher, M. J.; Goh, K. E. J.; Oberbeck, L.; Scappucci, G.; Hamilton, A. R.; Simmons, M. Y. Realization of atomically controlled dopant devices in silicon. Small 2007, 3, 563–567.
- [9]Fuechsle, M.; Miwa, J. A.; Mahapatra, S.; Ryu, H.; Lee, S.; Warschkow, O.; Hollenberg, L. C.; Klimeck, G.; Simmons, M. Y. A single-atom transistor. Nature nanotechnology 2012, 7, 242–246.
- [10]Wang, X.; Khatami, E.; Fei, F.; Wyrick, J.; Namboodiri, P.; Kashid, R.; Rigosi, A. F.; Bryant, G.; Silver, R. Experimental realization of an extended Fermi-Hubbard model using a 2D lattice of dopant-based quantum dots. Nature Communications 2022, 13, 6824.DOI
- [11]Stock, T. J.; Warschkow, O.; Constantinou, P. C.; Li, J.; Fearn, S.; Crane, E.; Hofmann, E. V.; Kolker, A.; McKenzie, D. R.; Schofield, S. R.; others Atomic-scale patterning of arsenic in silicon by scanning tunneling microscopy. ACS nano 2020, 14, 3316–3327.
- [12]Stock, T. J.; Warschkow, O.; Constantinou, P. C.; Bowler, D. R.; Schofield, S. R.; Curson, N. J. Single-Atom Control of Arsenic Incorporation in Silicon for High-Yield Artificial Lattice Fabrication. Advanced Materials 2024, 2312282.
- [13]Škereň, T.; Köster, S. A.; Douhard, B.; Fleischmann, C.; Fuhrer, A. Bipolar device fabrication using a scanning tunnelling microscope. Nature Electronics 2020, 3, 524–530.
- [14]Campbell, Q.; Ivie, J. A.; Bussmann, E.; Schmucker, S. W.; Baczewski, A. D.; Misra, S. A model for atomic precision p-type doping with diborane on Si (100)-2× 1. The Journal of Physical Chemistry C 2021, 125, 481–488.
- [15]Radue, M. S.; Baek, S.; Farzaneh, A.; Dwyer, K.; Campbell, Q.; Baczewski, A. D.; Bussmann, E.; Wang, G. T.; Mo, Y.; Misra, S.; others AlCl3-dosed Si (100)-2× 1: Adsorbates, chlorinated Al chains, and incorporated Al. The Journal of Physical Chemistry C 2021, 125, 11336–11347.DOI
- [16]Silva-Quinones, D.; He, C.; Butera, R. E.; Wang, G. T.; Teplyakov, A. V. Reaction of BCl3 with H- and Cl-terminated Si (1 0 0) as a pathway for selective, monolayer doping through wet chemistry. Applied surface science 2020, 533, 146907.DOI
- [17]Dwyer, K. J.; Baek, S.; Farzaneh, A.; Dreyer, M.; Williams, J. R.; Butera, R. E. B-doped δ-layers and nanowires from area-selective deposition of BCl3 on Si (100). ACS Applied Materials & Interfaces 2021, 13, 41275–41286.DOI
- [18]Owen, J.; Campbell, Q.; Santini, R.; Ivie, J.; Baczewski, A.; Schmucker, S.; Bussmann, E.; Misra, S.; Randall, J. Al-alkyls as acceptor dopant precursors for atomic-scale devices. Journal of Physics: Condensed Matter 2021, 33, 464001.DOI
- [19]Campbell, Q.; Dwyer, K. J.; Baek, S.; Baczewski, A. D.; Butera, R. E.; Misra, S. Reaction pathways of BCl_3 for acceptor delta-doping of silicon. arXiv preprint arXiv:2201.11682 2022,arxiv.org/abs/2201.11682
- [20]Campbell, Q.; Baczewski, A. D.; Butera, R.; Misra, S. Hole in one: Pathways to deterministic single-acceptor incorporation in Si (100)-2× 1. AVS Quantum Science 2022, 4.DOI
- [21]Hsueh, Y.-L.; Büch, H.; Tan, Y.; Wang, Y.; Hollenberg, L. C.; Klimeck, G.; Simmons, M. Y.; Rahman, R. Spin-lattice relaxation times of single donors and donor clusters in silicon. Physical review letters 2014, 113, 246406.
- [22]Watson, T. F.; Weber, B.; Hsueh, Y.-L.; Hollenberg, L. C.; Rahman, R.; Simmons, M. Y. Atomically engineered electron spin lifetimes of 30 s in silicon. Science advances 2017, 3, e1602811.DOI
- [23]Shim, Y.-P.; Tahan, C. Bottom-up superconducting and Josephson junction devices inside a group-IV semiconductor. Nature communications 2014, 5, 4225.DOI
- [24]Wyrick, J.; Wang, X.; Namboodiri, P.; Kashid, R. V.; Fei, F.; Fox, J.; Silver, R. Enhanced atomic precision fabrication by adsorption of phosphine into engineered dangling bonds on H–Si using STM and DFT. ACS nano 2022, 16, 19114–19123.
- [25]Giannozzi, P. et al. QUANTUM ESPRESSO: a modular and open-source software project for quantum simulations of materials. Journal of Physics: Condensed Matter 2009, 21, 395502.DOI
- [26]van Setten, M. J.; Giantomassi, M.; Bousquet, E.; Verstraete, M. J.; Hamann, D. R.; Gonze, X.; Rignanese, G. M. The PSEUDODOJO: Training and grading a 85 element optimized norm-conserving pseudopotential table. Computer Physics Communications 2018, 226, 39–54.DOI
- [27]Perdew, J. P.; Burke, K.; Ernzerhof, M. Generalized Gradient Approximation Made Simple. Physical Review Letters 1996, 77, 3865–3868.
- [28]Monkhorst, H. J.; Pack, J. D. Special points for Brillouin-zone integrations. Physical review B 1976, 13, 5188.
- [29]Tersoff, J.; Hamann, D. R. Theory of the scanning tunneling microscope. Physical Review B 1985, 31, 805.
- [30]Stowasser, R.; Hoffmann, R. What do the Kohn-Sham orbitals and eigenvalues mean? Journal of the american chemical society 1999, 121, 3414–3420.
- [31]Engelund, M.; Godlewski, S.; Kolmer, M.; Zuzak, R.; Such, B.; Frederiksen, T.; Szymonski, M.; Sánchez-Portal, D. The butterfly–a well-defined constant-current topography pattern on Si (001): H and Ge (001): H resulting from current-induced defect fluctuations. Physical Chemistry Chemical Physics 2016, 18, 19309–19317.DOI
- [32]Bortz, A.; Kalos, M.; Lebowitz, J. A New Algorithm for Monte Carlo Simulation of Ising Spin Systems. Journal of Computational Physics 1975, 17, 10–18.
- [33]Gillespie, D. T. A general method for numerically simulating the stochastic time evolution of coupled chemical reactions. Journal of Computational Physics 1976, 22, 403–434.
- [34]Leetmaa, M.; Skorodumova, N. V. KMCLib: A general framework for lattice kinetic Monte Carlo (KMC) simulations. Computer Physics Communications 2014, 185, 2340–2349.DOI
- [35]Arrhenius, S. Über die Reaktionsgeschwindigkeit bei der Inversion von Rohrzucker durch Säuren. Zeitschrift für Physikalische Chemie 1889, 4, 226–248.
- [36]Warschkow, O.; Curson, N.; Schofield, S.; Marks, N.; Wilson, H.; Radny, M.; Smith, P.; Reusch, T.; McKenzie, D.; Simmons, M. Y. Reaction paths of phosphine dissociation on silicon (001). The Journal of Chemical Physics 2016, 144, 014705.DOI
- [37]Apam-KMC: A package of python scripts for running Kinetic Monte Carlo on precursor molecules on Si(100)-2x1 surface. 2025; https://github.com/quantumquinn/apam-kmc.
- [38]Ivie, J. A.; Campbell, Q.; Koepke, J. C.; Brickson, M. I.; Schultz, P. A.; Muller, R. P.; Mounce, A. M.; Ward, D. R.; Carroll, M. S.; Bussmann, E.; others Impact of incorporation kinetics on device fabrication with atomic precision. Physical Review Applied 2021, 16, 054037.DOI
- [39]Tarnow, E. Theory of two boron neutral pair defects in silicon. Journal of Physics: Condensed Matter 1992, 4, 5405.DOI



