In Situ Patterning of Ultrasharp Dopant Profiles in Silicon - ACS Nano

Feb 9, 2017 - (23) Although somewhat successful, the high kinetic energy (and therefore long mean-free-path) of the electron beam used leads to incomp...
1 downloads 8 Views 3MB Size
In Situ Patterning of Ultrasharp Dopant Profiles in Silicon Simon P. Cooil,*,†,‡ Federico Mazzola,†,∇ Hagen W. Klemm,§ Gina Peschel,§ Yuran R. Niu,∥ Alexei A. Zakharov,∥ Michelle Y. Simmons,⊥ Thomas Schmidt,§ D. Andrew Evans,‡ Jill A. Miwa,# and Justin W. Wells† †

Department of Physics, Norwegian University of Science and Technology (NTNU), N-7491 Trondheim, Norway Department of Physics, Aberystwyth University, SY23 3BZ Aberystwyth, United Kingdom § Fritz-Harber-Insitute Max-Planck Society, Faradayweg 4-6 14195 Berlin, Germany ∥ MAX IV Laboratory, Lund University, 221 00 Lund, Sweden ⊥ Centre of Excellence for Quantum Computation and Communication Technology, School of Physics, University of New South Wales, Sydney, New South Wales 2052, Australia # Department of Physics and Astronomy, Interdisciplinary Nanoscience Center (iNANO), University of Aarhus, 8000 Aarhus C, Denmark ∇ School of Physics and Astronomy (SUPA), University of St. Andrews, St. Andrews, Fife KY16 9SS, United Kingdom ‡

ABSTRACT: We develop a method for patterning a buried two-dimensional electron gas (2DEG) in silicon using low kinetic energy electron stimulated desorption (LEESD) of a monohydride resist mask. A buried 2DEG forms as a result of placing a dense and narrow profile of phosphorus dopants beneath the silicon surface; a so-called δ-layer. Such 2D dopant profiles have previously been studied theoretically, and by angle-resolved photoemission spectroscopy, and have been shown to host a 2DEG with properties desirable for atomic-scale devices and quantum computation applications. Here we outline a patterning method based on low kinetic energy electron beam lithography, combined with in situ characterization, and demonstrate the formation of patterned features with dopant concentrations sufficient to create localized 2DEG states. KEYWORDS: low-energy electron patterning, delta layers, silicon quantum confinement, 2DEG, LEEM, PEEM high electrical conductivity,9 long electron spin coherence and relaxation times,10−12 many-body interactions,13 and long spinrelaxation times,12 making the platform a promising candidate for application in areas such as quantum computing.14−16 To date, methods for patterning such δ-layers have relied on the electron-stimulated desorption (ESD) of a monohydride resist bound to the Si surface, referred to herewith as H:Si(001), which exposes Si dangling bonds for subsequent phosphine (PH3) adsorption.17 Scanning tunneling microscopy (STM) has made it possible to pattern with unprecedented precision.17−20 A major drawback of STM lithography is that it is relatively slow at patterning larger areas, often requiring many hours of patterning per μm2.21 It is features at this μm scale that

W

ith a continued effort to downscale electronic device dimensions, the number and difficulty of the technical challenges required are ever increasing.1 The metal oxide semiconductor field effect transistor (MOSFET) is expected to reach its limiting architecture size within the next decade, giving way to more modern electronic platforms such as nanowire and molecular electronics.2 Additionally, the challenges associated with creating contacts to these ever decreasing device features are also rising. An interesting and scalable platform that shows great promise in overcoming such challenges is silicon delta-layer (δ-layer) electronics, for example, leading to technological milestones such as a single atom transistor.3 The δ-layers consist of a dense and sharp change in the phosphorus dopant density beneath a silicon (001) surface which creates a metallic two-dimensional electron gas (2DEG).4−8 Such highly doped regions in Si have shown a variety of phenomena of fundamental interest such as © 2017 American Chemical Society

Received: November 7, 2016 Accepted: February 9, 2017 Published: February 9, 2017 1683

DOI: 10.1021/acsnano.6b07359 ACS Nano 2017, 11, 1683−1688

Article

www.acsnano.org

Article

ACS Nano are necessary to bridge the already demonstrated nanoscale features to the macroscopic world. A promising solution to multiscale patterning is electron beam lithography (EBL). EBL is now an industry standard technique for device processing and has demonstrated the ability to pattern features sizes ranging from a few nm to many μm.22 ESD of a monohydride resist for patterned δ-layer fabrication has previously been attempted with high-energy (≈ 25 keV) electrons using a scanning electron microscope (SEM).23 Although somewhat successful, the high kinetic energy (and therefore long mean-free-path) of the electron beam used leads to incomplete H desorption at the surface. Additionally, care is needed to avoid the formation of crystalline defects and the deposition of reacted hydrocarbons from the vacuum system as a result of using such an intense electron beam. The minimum feature size is also limited to ≈200 nm due to the electron interaction volume at the surface for high kinetic energy electrons.23 Here we demonstrate a method for patterning the hydrogen (H)-resist using low kinetic energy electrons (40−150 eV). The primary advantage afforded is that the electron interaction is surface localized, and therefore, removal of the H-resist is very effective. The mechanism of hydrogen removal in this energy range is thought to originate from a two-hole final state in the H:Si bonding orbitals as a result of Auger electron decay processes. An unscreened hole−hole repulsion occurs, thereby causing subsequent desorption of H.24−26 Low kinetic energy electron stimulated desorption (LEESD) is therefore a promising alternative method for patterning the H:Si mask. Furthermore, since patterning with EBL is scalable over many orders of magnitude, it is an ideal method for bridging the gap between atomically precise STM lithography and microscale feature patterning. A schematic representation of the processing steps performed is shown in Figure 1 (for details of the sample preparation see the Methods/Experimental section). Following the preparation of a clean Si(001) substrate, a monohydride resist layer was prepared (Figure 1a). Low-energy electron lithography (40−150 eV) was then performed to selectively remove hydrogen from the H:Si surface (Figure 1b). After patterning the H-resist, the sample was exposed to PH3 gas, terminating the Si dangling bonds created in the patterning step (Figure 1c). An incorporation anneal was performed to create a P:Si(001) region (Figure 1d) . Finally, a Si encapsulation layer ≲1 nm thick was thermally deposited (Figure 1e).

Figure 1. Schematic illustration detailing the processing steps used to pattern high dopant density regions on Si substrates. (a) Initial H:Si(001) surface; (b) electron stimulated H desorption; (c) the patterned H-resist mask; (d) PH3 deposition; (e) H desorption and incorporation of P into the Si surface; (f) encapsulation with a thin Si overlayer. For simplicity, only the topmost Si layers are drawn.

surface due to the sample being held at a more negative potential (start voltage (SV)) than that of the electron illumination source. Contrast in this image mode is therefore related to both topography and changes in surface workfunction.27 The surface is seen to be homogeneous, indicating a complete termination of the Si surface dangling bonds. This is further confirmed by the sharp (2 × 1) low-energy electron diffraction (LEED) pattern in the inset of Figure 2a. Although the LEED pattern looks similar to a (2 × 2) reconstruction, this is only due to the two possible rotations of the (2 × 1) reconstruction on a Si(001) surface.4,28 Within the electron illumination column of the instrument, a variety of circular apertures can be used which restricts the size of the illuminating electron beam. In the following examples, feature sizes of 6 and 70 μm diameter were patterned, however, a large range of scales is possible with this method. By exposing the surface to low-energy electrons, selected areas of the hydrogen resist were removed, as seen in the secondary electron image (SEPEEM) of Figure 2d. By imaging the emitted secondary electrons, an increase in the workfunction contrast can be achieved, as there is a strong dependence of the photoemission yield on the local chemical state (i.e., presence of adsorbates).29 Following desorption of hydrogen, a sudden change in potential occurs at the interface between the patterned and unpatterned regions. This results in a lensing effect in the microscope which causes the bright and dark fringes around the patterned features observed in the MEM image of Figure 2b.30 The effect also somewhat limits the lateral resolution of the instrument, making accurate measurements of the edge profile

RESULTS/DISCUSSION In order to perform the patterning and measure aspects of the various processing steps, an instrument that can illuminate and image the surface with both electrons and X-rays was required. A photoemission and low-energy electron microscope was therefore the ideal choice, as it allows for sample architecture, crystal structure, and chemical composition to be investigated in situ within an ultrahigh vacuum environment, while providing the low kinetic energy electron lithography capabilities. An important aspect of the current study is the preparation of a highly ordered H:Si(001) surface, sufficient to mask PH3 adsorption on the unpatterned regions. We adopt an already established method for H-terminating the Si surface.3,19,20,23 Figure 2a shows a mirror electron microscopy (MEM) image of the Si surface following H-termination. In MEM, the image is formed from electrons that are reflected in front of the sample 1684

DOI: 10.1021/acsnano.6b07359 ACS Nano 2017, 11, 1683−1688

Article

ACS Nano

using SEM patterning23 and also leads to insufficient dopant densities. In contrast to SEM patterning, we are able to achieve complete resist removal using comparable electron doses. As an example of such an electron dose dependence, we present two patterned features. For both areas, electrons with the same kinetic energy (in this example 40 eV) are used, but with a different order of magnitude in e-beam exposure time (5 and 60 min). The electron density per unit area was also kept constant by placing a circular aperture in the path of the incident electron beam to pattern the small area and using the full electron beam profile for the larger area (approximately circular, diameter = 70 μm2, the edges of the patterned features appear uneven due to profile of incident electron beam). A small circular region of 28 μm2 was illuminated for 5 min, while a larger circle of ≈3800 μm2 was illuminated for 1 h, as indicated with a star and square in Figure 2b, respectively. Both exposure times result in H-desorption, and hence both regions are clearly visible in the MEM image of Figure 2b, however, the P concentration after PH3 dosing and annealing differs greatly between the two patterned features. It is, therefore, apparent that the 5 min exposure results in an incomplete desorption of H and a much lower P concentration, while the 1 h exposure results in much more complete H desorption over 3800 μm2 (cf. 4.2 μm2 per hour with STM).21 Hence the P density is high, and the patterned area remains visible in the MEM and SEPEEM images of Figure 2c,d. The feature remains visible in Figure 2e even though it has now been encapsulated with a Si overlayer (thickness 0.6 ± 0.2 nm as calculated by the attenuation of the P 2p core level from the μXPS measurements shown in Figure 2a,c). The patterned feature appears darker because the photoemission secondary electron cutoff is at a higher kinetic energy than that of the unpatterned areas. The secondary electron cutoff shifts to higher kinetic energy by ≈140 meV, as shown in Figure 2f. The origin of the shifted onset is due to a workfunction decrease within the patterned feature, where the magnitude of the decrease is consistent with metallic levels of n-type doping in Si.31 Being able to clearly see the buried patterned layer is an important engineering aspect if contacts are to be made to the few μm2 contacts already demonstrated for STM patterned nanoscale devices.18,20 Streaks along the [110] and [1̅10] directions are seen in the LEED pattern of Figure 2d (which are not observed in the area outside of the patterned region, see for example the inset of Figure 2c). This indicates that incorporating the P dopants disturbs the surface periodicity. It is known that PH3 readily disassociates upon contact with the Si surface, giving up a H atom which quickly terminates a nearby Si surface dangling bond.32 If the newly H-terminated surface bond prefers a particular crystallographic direction, then streaks in that direction in the LEED pattern would arise. Likewise ejected Si during the incorporation anneal may also adopt similar sites.33 As the surface comprises two rotational (2 × 1) domains, streaks in both directions are observed. The resulting LEED pattern of the surface following encapsulation was a (1 × 1) structure, as shown by the inset LEED pattern in Figure 2e. This diffraction pattern was observed both within the patterned feature and outside, further indicating that the grown Si is identical across the whole measurable surface. The higher background intensity in this LEED pattern is most likely a result of residual H migrating to the surface during Si encapsulation, due to incomplete H desorption during the incorporation anneal at 350 °C.

Figure 2. Study of the patterning steps used to produce the buried δ-layer. The MEM image (SV = −0.4 eV) shown in panel (a) was recorded directly after hydrogen passivation of the surface, while the MEM image (SV = −0.4 eV) of panel (b) was recorded following the low-energy electron (SV = 40 eV) illumination of two areas on the surface. The measured (2 × 1) LEED pattern shown in the inset of (a) was representative of the surface inside and outside of the patterned features. The MEM (SV = 0.21 eV) and SEPEEM (SV = 0.5 eV) images shown in panels (c) and (d), respectively, were acquired following exposure of the patterned surface to PH3 and an incorporation anneal at ≈350 °C. The LEED patterns shown in the insets of these two panels are taken from outside and inside the P:Si regions, respectively. Panels (e−g) show measurements made following the deposition of a thin Si encapsulation layer. A SEPEEM image (SV = 0.5 eV) is shown in (e) along with the measured (1 × 1) LEED pattern, which was representative of the entire measurable surface. The secondary electron cutoff curves shown in (f) were measured inside and out of the patterned area using μXPS (hν = 200 eV). A line profile through the boundary of the patterned region is shown in (g). It is extracted across the red line shown in the SEPEEM image of panel (e). The curve fitting of the dI/dx peak along the line reveals a Gaussian width of approximately 380 nm.

at the boundary challenging, however, it allows us to estimate the patterning resolution to be ≤380 nm (see Figure 2g). Our investigations have revealed that the desorption of H using low-energy electrons depends on the total electron dose. Areas of the H-resist that have been patterned using low electron flux (or short illumination time) do not exhibit complete desorption of H. This becomes apparent when PH3 is introduced, since incomplete removal of the H-resist leads to a lower concentration of PH3 on the surface and, therefore, an insufficient dopant density to produce a metallic feature. A similar dependence on the electron dose has also been observed 1685

DOI: 10.1021/acsnano.6b07359 ACS Nano 2017, 11, 1683−1688

Article

ACS Nano

been shown to hinder device performance, as they do not act as active dopants. From careful analysis of the relative binding energy positions of the minority P 2p components, the large area sample appears to have slightly more PH, while the patterned sample has slightly more PH2. This difference is likely due to small variations in the annealing temperature (see Figure 5 in ref 32). The temperature variation arises as a result of the practical limitations of measuring the sample temperature in the various experimental setups. Careful optimization of the temperature has been shown to obtain complete incorporation of P into the surface.33,37 In our studies, a well-established recipe for P incorporation has been used.38 The recipe yields a maximum P density of 0.25 monolayers in the Si lattice. For the large area 2D δ-layer sample, a P coverage ≈0.25 monolayers was sufficient to provide a metallic layer.4 A very similar P coverage was found for the features patterned with our low kinetic energy electron desorption method, therefore, the patterned features here are expected to have similar electronic properties to the nonpatterned samples. Following encapsulation with a thin Si overlayer, it is interesting to question why the region of phosphorus dopants is visible in the SEPEEM image of Figure 2e. The LEED pattern is similar on and off of the patterned region (indicating similar crystal structure), and the dopants are now located beneath the surface. The answer to this question is that the sample workfunction is dramatically different, as exemplified by Figure 2f. The μXPS measurements (Figure 3) indicate that the doping density is the same as in a large area Si:P layer, which is known from previous studies to have a metallic electronic structure4 as well as metallic sheet resistance.5,9 This strong and metallic doping concentration is expected to result in a workfunction shift of approximately 100 meV (relative to an intrinsic sample)31 and is fully consistent with our measured workfunction shift of 140 meV on and off the patterned area.

In order to investigate the spatial distribution and chemical state of the incorporated P, small area (25 × 25 μm) photoelectron spectroscopy (μXPS) measurements were performed on a ≈ 490 μm2 feature, and the results are presented in Figure 3a,c,d. For comparison of the patterned photoemission results, XPS was also performed on large area δlayer sample (Figure 3b).

Figure 3. Emitted photoelectron intensity of the P 2p core level at various processing steps as well as a comparable P 2p core level from a nonpatterned sample (hν = 150 eV). The data are shown (red markers) following the subtraction of a Shirley background function. A fit to the data is shown in black and comprises two pairs of spin orbit coupled Voigt functions representing P and PHx in blue and pink, respectively. Panels (a), (c), and (d) are acquired using μXPS within the same SPELEEM where the patterning was performed, while (b) is acquired using conventional XPS. (a, b) The P 2p intensity following P incorporation. (c) The P 2p core level collected following deposition of a thin Si overlayer. (d) P 2p core level acquired on the H:Si surface outside of the patterned region (following the incorporation anneal). The relative binding energy scale refers to the energy shift from of a oxygen-free Si 2p3/2 core level to facilitate comparison to refs 32, 34−36.

CONCLUSIONS In conclusion, areas of a monohydride resist bound to a Si(001) wafer can be removed effectively by utilizing LEESD, with a lateral resolution of few 100 nm. Electrons in the range 40−150 eV were used, however, 150 eV gave the highest efficiency in H desorption and hence the highest density of P dopants. This indicates that the Auger decay of electrons is a more effective mechanism for removal of hydrogen than valence to conduction band transitions. The primary advantage over previously developed methods is that the low kinetic energy electrons have an interaction volume that is confined to the surface. We demonstrate that the total electron dose required for complete removal of the monohydride resist is also an important aspect of the technique. We provide evidence that by utilizing this technique, the patterning of metallic features on the Si surface is possible, and by comparison of the core-level photoelectrons, the patterned regions are very similar to large area metallic δ-layers, which have a band structure highly desirable for the engineering of quantum devices. Therefore, using the method developed here, we believe it is now feasible to efficiently pattern feature sizes which overlap atomic-scale δlayer features in silicon, thus paving the way to multiscale patterning of quantum computer prototype devices and interconnects.

The large area XPS measurements presented in Figure 3b are from an experiment in which the δ-layer was shown to be metallic by a measurable δ-state in the angle resolved photoemission (ARPES) study presented in our earlier work.4 The total measured P 2p intensity is expected to consist of a spin−orbit coupled doublet with a separation of 0.7 eV, however in our measurements, we typically observe two such doublets. According to the XPS data of Lin et al.34−36 reinterpreted by Wilson et al.,32 the binding energy of P that has been fully incorporated into the Si surface would be 129.20 eV, while components at higher binding energies originate from remnant PHx species. The μXPS measurements confirm that P is confined to the patterned region alone, as shown in Figure 3a,d. For both the patterned and large area samples, the measured P 2p core levels are extremely similar (Figure 3a,b). In both cases, the doublets have their most intense peak at ≈129.10 eV binding energy (blue components), indicating that the majority of the P has been incorporated.32,35 For both examples, a small amount of PHx species is detected (pink components), and remnant PHx species have also been observed in other studies9 and have not 1686

DOI: 10.1021/acsnano.6b07359 ACS Nano 2017, 11, 1683−1688

Article

ACS Nano

(8) Mazzola, F.; Edmonds, M. T.; Høydalsvik, K.; Carter, D. J.; Marks, N. A.; Cowie, B. C. C.; Thomsen, L.; Miwa, J.; Simmons, M. Y.; Wells, J. W. Determining the Electronic Confinement of a Subsurface Metallic State. ACS Nano 2014, 8, 10223−10228. (9) Polley, C. M.; Clarke, W. R.; Miwa, J. A.; Scappucci, G.; Wells, J. W.; Jaeger, D. L.; Bischof, M. R.; Reidy, R. F.; Gorman, B. P.; Simmons, M. Exploring the Limits of N-Type Ultra-Shallow Junction Formation. ACS Nano 2013, 7, 5499−5505. (10) Büch, H.; Mahapatra, S.; Rahman, R.; Morello, A.; Simmons, M. Y. Spin readout and addressability of phosphorus-donor clusters in silicon. Nat. Commun. 2013, 4, 3017. (11) Yang, C. H.; Rossi, A.; Ruskov, R.; Lai, N. S.; Mohiyaddin, F. A.; Lee, S.; Tahan, C.; Klimeck, G.; Morello, A.; Dzurak, A. S. Spin-Valley Lifetimes in A Silicon Quantum Dot With Tunable Valley Splitting. Nat. Commun. 2013, 4, 3069. (12) Morello, A.; Pla, J. J.; Zwanenburg, F. A.; Chan, K. W.; Tan, K. Y.; Huebl, H.; Mottonen, M.; Nugroho, C. D.; Yang, C.; van Donkelaar, J. A.; Alves, A. D. C.; Jamieson, D. N.; Escott, C. C.; Hollenberg, L. C. L.; Clark, R. G.; Dzurak, A. S. Single-Shot Readout of an Electron Spin in Silicon. Nature 2010, 467, 687−691. (13) Mazzola, F.; Polley, C. M.; Miwa, J. A.; Simmons, M. Y.; Wells, J. W. Disentangling Phonon and Impurity Interactions in δ-Doped Si(001). Appl. Phys. Lett. 2014, 104, 173108. (14) Hollenberg, L. C. L.; Greentree, A. D.; Fowler, A. G.; Wellard, C. J. Two-Dimensional Architectures For Donor-Based Quantum Computing. Phys. Rev. B: Condens. Matter Mater. Phys. 2006, 74, 045311. (15) Kane, B. E. A Silicon-Based Nuclear Spin Quantum Computer. Nature 1998, 393, 133−137. (16) Hill, C. D.; Peretz, E.; Hile, S. J.; House, M. G.; Fuechsle, M.; Rogge, S.; Simmons, M. Y.; Hollenberg, L. C. L. A Surface Code Quantum Computer in Silicon. Sci. Adv. 2015, 1, e1500707. (17) Hallam, T.; Rueß, F. J.; Curson, N. J.; Goh, K. E. J.; Oberbeck, L.; Simmons, M. Y.; Clark, R. G. Effective Removal of Hydrogen Resists Used to Pattern Devices in Silicon Using Scanning Tunneling Microscopy. Appl. Phys. Lett. 2005, 86, 143116. (18) Simmons, M. Y.; Ruess, F. J.; Goh, K. E. J.; Hallam, T.; Schofield, S. R.; Oberbeck, L.; Curson, N. J.; Hamilton, A. R.; Butcher, M. J.; Clark, R. G.; Reusch, T. C. G. Scanning Probe Microscopy for Silicon Device Fabrication. Mol. Simul. 2005, 31, 505−515. (19) Ruess, F. J.; Oberbeck, L.; Simmons, M. Y.; Goh, K. E. J.; Hamilton, A. R.; Hallam, T.; Schofield, S. R.; Curson, N. J.; Clark, R. G. Toward Atomic-Scale Device Fabrication in Silicon Using Scanning Probe Microscopy. Nano Lett. 2004, 4, 1969−1973. (20) Rueß, F. J.; Oberbeck, L.; Goh, K. E. J.; Butcher, M. J.; Gauja, E.; Hamilton, A. R.; Simmons, M. Y. The Use of Etched Registration Markers To Make Four-Terminal Electrical Contacts to STMPatterned Nanostructures. Nanotechnology 2005, 16, 2446. (21) Hallam, T. The Use and Removal of a Hydrogen Resist on the Si (001) Surface for P-in-Si Device Fabrication. Thesis Ph.D., University of New South Wales, 2006. (22) Manfrinato, V. R.; Zhang, L.; Su, D.; Duan, H.; Hobbs, R. G.; Stach, E. A.; Berggren, K. K. Resolution Limits of Electron-Beam Lithography Toward the Atomic Scale. Nano Lett. 2013, 13, 1555− 1558. (23) Hallam, T.; Butcher, M. J.; Goh, K. E. J.; Ruess, F. J.; Simmons, M. Y. Use of a Scanning Electron Microscope to Pattern Large Areas of a Hydrogen Resist for Electrical Contacts. J. Appl. Phys. 2007, 102, 034308. (24) Fuse, T.; Fujino, T.; Ryu, J.-T.; Katayama, M.; Oura, K. Electron-Stimulated Desorption of Hydrogen From H/Si(001)-1 × 1 Surface Studied by Time-of-Flight Elastic Recoil Detection Analysis. Surf. Sci. 1999, 420, 81−86. (25) Albert, M. M.; Tolk, N. H. Absolute Total Cross Sections for Electron-Stimulated Desorption of Hydrogen and Deuterium From Silicon(111) Measured by Second Harmonic Generation. Phys. Rev. B: Condens. Matter Mater. Phys. 2000, 63, 035308. (26) Knotek, M. L.; Feibelman, P. J. Ion Desorption by Core-Hole Auger Decay. Phys. Rev. Lett. 1978, 40, 964−967.

METHODS/EXPERIMENTAL Flat and oxygen free n-type Si(001) substrates (resistivity 1−10 Ω cm−1, doped with phosphorus) were characterized by XPS, LEEM, and LEED following in-vacuum annealing cycles up to 1250 °C. The measurements revealed a (2 × 1) reconstruction in the LEED measurements and a lack of oxygen in μXPS. Hydrogen termination was performed by exposing the substrate to a constant flux of atomic hydrogen (pressure = 5 × 10−7 mbar) at a sample temperature of ≈240 °C for 5 min, following ref 11. This dose of hydrogen is likely to be far in excess of that required for termination of the Si surface dangling bonds. The sample was then cooled within the hydrogen flux to ≈200 °C for a further 2 min before finally terminating the flow of hydrogen and cooling the sample at a rate of 1 °C s−1 to room temperature, in order to minimize the possibility of H etching the Si surface. This method was found to give the most complete Htermination maintaining the (2 × 1) reconstruction of the surface. PH3 was introduced to the system at a pressure of 5 × 10−9 mbar. The incorporation anneal was performed at approximately 350 °C. The temperature was measured by a thermocouple positioned close to the sample and IR pyrometry, and an uncertainty of ±20 °C is estimated. Encapsulation of Si was performed by thermally depositing Si in situ from a calibrated evaporation cell on to the sample at room temperature at vacuum pressures ≤5 × 10−10 mbar and deposition rates of ≈8 Å min−1. The silicon overlayer was then given a postdeposition anneal to 400 °C in order to obtain the LEED pattern shown in Figure 2e. Measurements were carried out at the UE49PGM-SMART end-station at Helmholtz-Zentrum Berlin, Germany, and I311 MAXLab Lund, Sweden.

AUTHOR INFORMATION Corresponding Author

*E-mail: [email protected]. ORCID

Simon P. Cooil: 0000-0002-0856-6020 Notes

The authors declare no competing financial interest.

ACKNOWLEDGMENTS We thank HZB and MAX IV for the allocation of synchrotron radiation beam-time. J.A.M. acknowledges support from the Danish Council for Independent Research, Natural Sciences under the Sapere Aude program (grant no. DFF-6108-00409). REFERENCES (1) Peercy, P. S. The Drive to Miniaturization. Nature 2000, 406, 1023−1026. (2) Vogel, E. Technology and Metrology of New Electronic Materials and Devices. Nat. Nanotechnol. 2007, 2, 25−32. (3) Fuechsle, M.; Miwa, J. A.; Mahapatra, S.; Ryu, H.; Lee, S.; Warschkow, O.; Hollenberg, L. C. L.; Klimeck, G.; Simmons, M. Y. A Single-Atom Transistor. Nat. Nanotechnol. 2012, 7, 242−246. (4) Miwa, J. A.; Hofmann, P.; Simmons, M. Y.; Wells, J. W. Direct Measurement of the Band Structure of a Buried Two-Dimensional Electron Gas. Phys. Rev. Lett. 2013, 110, 136801. (5) Polley, C. M.; Clarke, W. R.; Miwa, J. A.; Simmons, M. Y.; Wells, J. W. Microscopic Four-Point-Probe Resistivity Measurements of Shallow, High Density Doping Layers in Silicon. Appl. Phys. Lett. 2012, 101, 262105. (6) Carter, D. J.; Marks, N. A.; Warschkow, O.; McKenzie, D. R. Phosphorus δ-Doped Silicon: Mixed-Atom Pseudopotentials and Dopant Disorder Effects. Nanotechnology 2011, 22, 065701. (7) Goh, K. E. J.; Oberbeck, L.; Simmons, M. Y.; Hamilton, A. R.; Butcher, M. J. Influence of Doping Density on Electronic Transport in Degenerate Si:P Delta-Doped Layers. Phys. Rev. B: Condens. Matter Mater. Phys. 2006, 73, 035401. 1687

DOI: 10.1021/acsnano.6b07359 ACS Nano 2017, 11, 1683−1688

Article

ACS Nano (27) Schmidt, T.; Heun, S.; Slezak, J.; Diaz, J.; Prince, K. C.; Lilienkamp, G.; Bauer, E. SPELEEM: Combining LEEM and Spectroscopic Imaging. Surf. Rev. Lett. 1998, 05, 1287−1296. (28) Miwa, J. A.; Warschkow, O.; Carter, D. J.; Marks, N. A.; Mazzola, F.; Simmons, M. Y.; Wells, J. W. Valley Splitting in a Silicon Quantum Device Platform. Nano Lett. 2014, 14, 1515−1519. (29) Bauer, E.; Mundschau, M.; Swiech, W.; Telieps, W. Surface Studies by Low-Energy Electron Microscopy (LEEM) and Conventional UV Photoemission Electron Microscopy (PEEM). Ultramicroscopy 1989, 31, 49−57. (30) Jesson, D. E.; Tang, W. X. Surface Electron Microscopy of Ga Droplet Dynamics on GaAs (001). Microscopy: Science Technology, Applications and Education, Microscopy Book Series No. 4; Méndez-Vilas, A., Díaz, J.; Formatex: Badajoz, Spain, 2010; Vol 3, p 11. (31) Allen, F. G.; Gobeli, G. W. Work Function, Photoelectric Threshold, and Surface States of Atomically Clean Silicon. Phys. Rev. 1962, 127, 150−158. (32) Wilson, H. F.; Warschkow, O.; Marks, N. A.; Curson, N. J.; Schofield, S. R.; Reusch, T. C. G.; Radny, M. W.; Smith, P. V.; McKenzie, D. R.; Simmons, M. Y. Thermal Dissociation and Desorption of PH3 on Si(001): A Reinterpretation of Spectroscopic Data. Phys. Rev. B: Condens. Matter Mater. Phys. 2006, 74, 195310. (33) McKibbin, S. R.; Clarke, W. R.; Fuhrer, A.; Simmons, M. Y. Optimizing Dopant Activation in Si:P Double Delta-Layers. J. Cryst. Growth 2010, 312, 3247−3250. (34) Tsai, H.-W.; Lin, D.-S. Comparison of Thermal Reactions of phosphine on Ge(100) and Si(100) by High-Resolution Core-Level Photoemission. Surf. Sci. 2001, 482-485, 654−658. (35) Lin, D. S.; Ku, T. S.; Sheu, T. J. Thermal Reactions of Phosphine with Si(100): a Combined Photoemission and Scanning-TunnelingMicroscopy Study. Surf. Sci. 1999, 424, 7−18. (36) Lin, D.-S.; Ku, T.-S.; Chen, R.-P. Interaction of Phosphine with Si(100) From Core-Level Photoemission and Real-Time Scanning Tunneling Microscopy. Phys. Rev. B: Condens. Matter Mater. Phys. 2000, 61, 2799−2805. (37) Keizer, J. G.; McKibbin, S. R.; Simmons, M. Y. The Impact of Dopant Segregation on the Maximum Carrier Density in Si:P Multilayers. ACS Nano 2015, 9, 7080−7084. (38) McKibbin, S. R.; Polley, C. M.; Scappucci, G.; Keizer, J. G.; Simmons, M. Y. Low Resistivity, Super-Saturation Phosphorus-inSilicon Monolayer Doping. Appl. Phys. Lett. 2014, 104, 123502.

1688

DOI: 10.1021/acsnano.6b07359 ACS Nano 2017, 11, 1683−1688