# Fileset

[Agarwal_2025_J._Phys._Mater._8_045006.pdf](https://mdr.nims.go.jp/filesets/1b68b746-b2cb-4c33-9048-b5a54796a2ce/download)

## Creator

Hitesh Agarwal, Antoine Reserbat-Plantey, David Barcons Ruiz, Karuppasamy Pandian Soundarapandian, Geng Li, Vahagn Mkhitaryan, Johann Osmond, Helena Lozano, [Kenji Watanabe](https://orcid.org/0000-0003-3701-8119), [Takashi Taniguchi](https://orcid.org/0000-0002-1467-3105), Petr Stepanov, Frank H L Koppens, Roshan Krishna Kumar

## Rights

[Creative Commons BY Attribution 4.0 International](https://creativecommons.org/licenses/by/4.0/)

## Other metadata

[In situ engineering hexagonal boron nitride in van der Waals heterostructures with selective SF<sub>6</sub> etching](https://mdr.nims.go.jp/datasets/5f9fc2f1-5f92-41e1-a27f-6b6e3a338a38)

## Fulltext

In situ engineering hexagonal boron nitride in van der Waals heterostructures with selective SF6 etching     PAPER • OPEN ACCESSIn situ engineering hexagonal boron nitride in vander Waals heterostructures with selective SF6etchingTo cite this article: Hitesh Agarwal et al 2025 J. Phys. Mater. 8 045006 View the article online for updates and enhancements.You may also likeThe IFMIF-DONES Irradiation ModulesFrederik Arbeiter, Urszula Wicek, BeatrizBrañas et al.-Update on the status of the IFMIF-DONESTest SystemsSantiago Becerril, Jesus Castellanos,Pablo Ignacio Araya Carmona et al.-Remote maintenance in IFMIF-DONES:current status and future developmentprogramGioacchino Micciché, Fernando Arranz,Martin Mittwollen et al.-This content was downloaded from IP address 144.213.253.16 on 17/09/2025 at 02:35https://doi.org/10.1088/2515-7639/adfd15/article/10.1088/1741-4326/add172/article/10.1088/1741-4326/ade9dd/article/10.1088/1741-4326/ade9dd/article/10.1088/1741-4326/adcfb3/article/10.1088/1741-4326/adcfb3/article/10.1088/1741-4326/adcfb3https://pagead2.googlesyndication.com/pcs/click?xai=AKAOjssujAtxYF4nVdfe3nMPMpxeH7g1PgVfLr-WT3PEUyIx3-cO9XPffRGjcKZXfl3e1fw8X09u6YJLhhCck1sCzHHbmvzbgM-ktphmsXqxdQZfLUoWJf0b-o4GpbQz_M9P8gn6mlxx2G902rh9WqBHRWvzDv8dfp_oNWHV33ZmgfFPjeMG3PDAdf47SO62kMGveoF5oPc3vUr7rA99i03RmSkXPt46tKl8T4UUy-nx-94OdJ4c7bNoTmM8d-YfKSrXRu7t7xCwc1jsdIJEWzgUB8SuPpPEOvuOzTaQsHk6GS9btUzd_GmfKUNdxqAGRtbwpSZzZ0mXV7x-ghhiIlz5UjsJ6bK9reQbxeGVc-tTuVVuSAv1Pb2_&sig=Cg0ArKJSzDn0VgkhDNBX&fbs_aeid=%5Bgw_fbsaeid%5D&adurl=https://www.edinst.com/product/fls1000-photoluminescence-spectrometer/%3Futm_source%3Diopcover%26utm_medium%3DnanotechJ. Phys. Mater. 8 (2025) 045006 https://doi.org/10.1088/2515-7639/adfd15Journal of Physics: MaterialsOPEN ACCESSRECEIVED9 December 2024REVISED18 July 2025ACCEPTED FOR PUBLICATION19 August 2025PUBLISHED10 September 2025Original content fromthis work may be usedunder the terms of theCreative CommonsAttribution 4.0 licence.Any further distributionof this work mustmaintain attribution tothe author(s) and the titleof the work, journalcitation and DOI.PAPERIn situ engineering hexagonal boron nitride in van der Waalsheterostructures with selective SF6 etchingHitesh Agarwal1, Antoine Reserbat-Plantey1,2,∗, David Barcons Ruiz1,Karuppasamy Pandian Soundarapandian1, Geng Li1, Vahagn Mkhitaryan1, Johann Osmond1,Helena Lozano1, Kenji Watanabe3, Takashi Taniguchi4, Petr Stepanov1,5, Frank H L Koppens1,6,∗and Roshan Krishna Kumar1,7,∗1 ICFO—Institut de Ciències Fotòniques, The Barcelona Institute of Science and Technology, Av. Carl Friedrich Gauss 3, Castelldefels(Barcelona), 08860, Spain2 Université Côte d’Azur, CNRS, CRHEA, rue Bernard Grégory, 06560 Valbonne, France3 Research Center for Functional Materials, National Institute for Materials Science, Tsukuba 305-0044, Japan4 International Center for Materials Nanoarchitectonics, National Institute for Materials Science, Tsukuba 305-0044, Japan5 Department of Physics and Astronomy, University of Notre Dame, Notre Dame, IN 46556, United States of America6 ICREA—Institució Catalana de Recerca i Estudis Avancats, 08010 Barcelona, Spain7 Catalan Institute of Nanoscience and Nanotechnology (ICN2), Campus UAB, Bellaterra, 08193 Barcelona, Spain∗ Authors to whom any correspondence should be addressed.E-mail: antoine.reserbat-plantey@cnrs.fr, frank.koppens@icfo.eu and roshan.krishna@icn2.catKeywords: 2D Materials, graphene, nanofabrication, van der Waals heterostructures, nanopatterningSupplementary material for this article is available onlineAbstractVan der Waals heterostructures are at the forefront in materials heterostructure engineering,offering the ultimate control in layer selectivity and capability to combine virtually any material.Hexagonal-boron nitride, the most commonly used dielectric material, has proven indispensablein this field, allowing the encapsulation of active 2D materials preserving their exceptionalelectronic quality. However, not all device applications require full encapsulation but rather requireopen surfaces, or even selective patterning of hBN layers. Here, we report on a procedure toengineer top hBN layers within van der Waals heterostructures while preserving the underlyingactive 2D layers. Using a soft selective SF6 etching combined with a series of pre—and post-etchingtreatments, we demonstrate that pristine surfaces can be exposed with atomic scale flatness whilepreserving the active layers’ electronic quality. We benchmark our technique using graphene/hBNHall bar devices. Using Raman spectroscopy combined with quantum transport, we show highquality can be preserved in etched regions by demonstrating low temperature carriermobilities> 200,000 cm2V–1s−1, ballistic transport probed through magnetic focusing, andintrinsic room temperature phonon-limited mobilities. Atomic force microscopy brooming andO2 plasma cleaning are identified as key pre-etching steps for obtaining pristine open surfaceswhile preserving electronic quality. The technique provides a clean method for opening windowsinto mesoscopic van der Waals devices that can be used for local probe experiments, patterning tophBN in-situ, and exposing 2D layers to their environment for sensing applications.The synthesis of high-quality hexagonal-boron nitride [1] marked a turning point in two-dimensional (2D)materials research [2–4]. As an inert 2D crystal, it is an excellent dielectric material in 2D electronics [5],provides atomically flat surfaces with pristine interfaces [6], and protects active 2D layers from degradingatmospheric environments [7]. These properties enable engineering van der Waals heterostructures of thehighest quality with unique functionalities envisioned for next-generation semiconductor technologies [8].Additionally, they serve as powerful condensed matter simulators harboring physics spanning stronglycorrelated electron phenomena to topological physics [9]. In most van der Waals heterostructures, hBNtypically encapsulates active 2D layers. The bottom hBN protects layers from rough substrates [10], while thetop isolates it completely from its environment, pushing device quality to intrinsic limits [11]. However,© 2025 The Author(s). Published by IOP Publishing Ltdhttps://doi.org/10.1088/2515-7639/adfd15https://crossmark.crossref.org/dialog/?doi=10.1088/2515-7639/adfd15&domain=pdf&date_stamp=2025-9-10https://creativecommons.org/licenses/by/4.0/https://creativecommons.org/licenses/by/4.0/https://orcid.org/0000-0002-9418-7966https://orcid.org/0000-0002-9106-8750https://orcid.org/0000-0002-6271-2244https://orcid.org/0000-0002-9664-9095https://orcid.org/0000-0002-3138-8488https://orcid.org/0000-0001-6522-4695https://orcid.org/0000-0003-3701-8119https://orcid.org/0000-0002-1121-3146https://orcid.org/0000-0001-9764-6120https://orcid.org/0000-0003-0857-4466mailto:antoine.reserbat-plantey@cnrs.frmailto:frank.koppens@icfo.eumailto:roshan.krishna@icn2.cathttp://doi.org/10.1088/2515-7639/adfd15J. Phys. Mater. 8 (2025) 045006 H Agarwal et alFigure 1.Methods for in-situ engineering top hexagonal-boron nitride in van der Waals heterostructures. (a)–(h), schematics ofthe work flow presented in this work. The device schematic is sketched in a, showing a graphene (grey) encapsulated with hBN(green), contacted with Chromium, palladium, gold (Cr/Pd/Au) and fabricated on a silicon dioxide (SiO2) on silicon (Si)substrate which was used as an electrostatic gate. A series of lithography (b), O2 plasma ashing (c), (e) and dry etching (f) isperformed. At specific steps, AFM brooming is employed (d), (g) to remove contaminants and preserve electronic quality.(i) optical image of our device D1 after a window has been opened. Scale bar is 10 µm.certain applications and experiments require exposed surfaces. Hence, full encapsulation is not alwaysdesirable because it restricts access to active 2D layers.From an applications perspective, exposed surfaces offer advantages and novel functionalities, such astailoring light-matter interactions via patterned metasurfaces [12], leveraging 2D materials’ surfacesensitivity for biological and chemical sensing [13] and ensuring good electrical and magnetic contacts [14].From a fundamental standpoint, exposed surfaces allow direct access to the underlying electron system usinglocal probes. Some of the most powerful local spectroscopic probes, including scanning tunnelingmicroscopy [15–18] and Angle-resolved photoemission spectroscopy [19, 20], require exposed 2D layersachievable only through challenging heterostructure engineering. These techniques limit scanning areas andsuffer from polymer contamination which may degrade sample quality. In all these applications andexperiments, the bottom hBN remains crucial playing a major role in preserving electronic quality.To overcome these challenges, we introduce an in-situ hBN patterning method for van der Waalsheterostructures. We demonstrate that sulfur hexafluoride (SF6) can selectively etch [21] the top hBN inencapsulated heterostructures, opening windows into small regions of the device (figure 1) that serve asaccess points to the underlying 2D layers. Using graphene encapsulated with hexagonal-boron nitride, wetrack the quality of selectively etched regions through atomic force microscopy (AFM), Raman spectroscopyand quantum transport measurements, demonstrating that high electronic quality remains intact, limited byatmospheric conditions. The methodology is sketched in figures 1(a)–(h). Starting with fully encapsulatedhBN/graphene/hBN devices with electrical contacts (figure 1(a)), we pattern etching masks (figure 1(b))with PMMA. Following, a series of cleaning steps including O2 plasma etching and AFM brooming(figures 1(c)–(e) prepares the surface for SF6 etching of hBN (figure 1(f)).SF6 has proven highly effective in graphene nanofabrication due to its ability to selectively etch hBNwithout damaging graphene, enabling controlled exposure of clean graphene edges [22]. When used as anetchant for hBN, the energetic plasma breaks SF6 molecules, generating reactive fluorine radicals. Thesefluorine atoms react with the boron in the BN lattice forming volatile boron trifluoride (BF3), while thenitrogen is released as molecular nitrogen [21] (N). In a simplified overall reaction, one may write:2BN(s)+ 6F→ 2BF3 (g)+ N2 (g) where (s) and (g) refer to solid and gas phase respectively. In the plasmaprocess, fluorine atoms come from SF6, which is reduced to SF4 or other sulfur-fluoride species. The mainetch products are boron trifluoride (BF3), removed as a volatile gas, and nitrogen (N2) released into theatmosphere. Because of its excellent selectivity for hBN over graphene, it is commonly used to createlow-resistance electrical contacts [22]. Here, we use it instead to directly pattern the top encapsulating hBNlayers, selectively removing specific areas that expose the underlying graphene (figure 1) while preserving thebottom hBN substrate and maintaining graphene’s high electronic quality.In this work, we use a recipe inspired by previous works [21, 22]. However, we introduce additionalpre-etching steps and refine the etching procedure to optimize graphene’s electronic quality (seesupplementary section 1), making it suitable for various applications and experiments. While SF6 etchingstops at graphene, over-etching may still be detrimental, leading to fluorination or doping inhomogeneities.Calibrating etching time and recipes across different reactive ion etchers (RIE) is crucial to ensure selectivehBN removal without excessive graphene fluorination. Generally, a stable plasma should be maintained withthe highest chamber pressures but lowest radio frequency powers. Higher pressures reduce precision butenable slower, more controlled hBN removal, essential for preserving graphene’s electronic quality. We2J. Phys. Mater. 8 (2025) 045006 H Agarwal et alFigure 2. Atomic force microscopy measurements of etched regions. (a) an AFM image of device 1 in the regions whereD1_window1 and D1_window2 have been opened (see white labels). The gold regions with the largest height (t) correspond tothe electrical contacts of the device. The inset displays an optical image of the device. The scale bar is 2 µm. (b) Zoomed spatialmap of the height profile (t) in D1_window 2. The area is indicated by the white dashed box in panel a, (c), a 1× 1 µm2 area ofD2_window3 after annealing.perform systematic calibration using reference regions on hBN layers and use optical microscopy to track thehBN removal during sequential etching steps until complete. The calibrated recipe can then be applied to theactive region. We characterize our devices in the etched regions before and after exposure of graphene tobenchmark its electronic quality and the importance of the steps employed in b–g.To develop the selective etching recipe, we studied graphene-based/hBN heterostructures. This includedtwo monolayer graphene devices encapsulated with hBN (D1/2) and a small angle twisted bilayer graphenedevice (D3). All devices were fabricated using standard methods (see methods). The process involvedheterostructure assembly, followed by nanofabrication of mesoscopic devices with Hall bar geometries(figure 1(i)). Low-temperature (10 K) quantum transport measurements characterized the devices, revealingexcellent electronic quality, including size-limited mobilities [23, 24] (see discussions below). All deviceswere studied before and after etching using AFM. However, Raman spectroscopy and quantum transportmeasurements post-etching were performed exclusively on monolayer graphene/hBN heterostructuresbecause of its well-known electronic properties which enabled proper calibration.Following low-temperature device characterization, we applied our selective etching method. Optimizedrecipes, discussed in methods and supplementary section 1, were tested on all devices. Here, we presentresults on hBN/Graphene/hBN devices, where three windows were opened on two devices-D1 (figure 2(a))and D2 (figure 2(c))-demonstrating the importance of pre- and post-etching steps. The first is a small squareon the device’s left side (figure 2(a)). For this window (D1_window1), we follow steps figures 1(b)–(f),directly etching the device after patterning the PMMAmask. The second window (D1_window2) extends theentire device width. In this case, additional pre-etching steps were used, where the top hBN surface was firstcleaned using AFM brooming [25–27] to remove polymer residues from the targeted region (figures 1(d)and (e)), followed by O2 plasma cleaning. The third window in D2 (D2_window3) followed the sameprocedure, but was additionally annealed under ultra-high vacuum post-etching.All three windows were characterized using AFM. Figure 2(a) plots a topographic map of D1 afteretching windows 1 and 2. The two regions exhibit notable differences. D1_window1 shows significant surfaceroughness resembling contaminants accumulating on the surface of graphene. In contrast, the D1_window2appears much smoother with a root square mean roughness (Rq) below 1 nm. Figure 2(b) further illustratesthis by plotting the height profile (t) as a function of x,y spatial coordinates for a 1× 1 µm2 areas. Weattribute these differences to the crucial importance of AFM brooming. Without brooming, any surfacecontamination on the top hBN falls onto the graphene. With brooming the top hBN can be cleaned enablingselective etching that exposes the graphene with a pristine surface. The residual surface roughness tells usthat some contamination still remained, possibly due to hydrocarbon adsorbates or PMMA residues.Following the same procedure in a third device (D3_window4), we achieved atomically flat areas(Rq = 0.2 nm) over 1× 1 µm2 areas (see supplementary section 2). With additional post-etching annealingin D2_window3, even smoother surfaces were achieved. This included atomically flat areas over 2× 2 µm2,meeting the requirements of sensitive scanning probe experiments (figure 2(c)).While AFM measurements suggest clean exposed surfaces can be achieved, the chemical procedure mayfluorinate graphene or degrade its crystalline quality, which AFM alone cannot easily detect. Therefore, weperformed micro-Raman spectroscopy to characterize the exposed graphene’s quality further, compare itwith the fully encapsulated graphene, and evaluate the impact of etching. We focus on the E2g Raman activemodes, leading to the G band [28] at around 1582 cm–1, and for hBN [29] at 1362 cm−1. Notably, we3J. Phys. Mater. 8 (2025) 045006 H Agarwal et alFigure 3. Investigation of window opening and surface brooming effects on graphene via micro-Raman spectroscopy (a):Hyperspectral Raman map of the sample displaying the amplitude of the hBN E2g peak normalized by the amplitude of thegraphene G peak, which mitigates interference effects. The zones highlighted in pink including the left square area, correspond toregions where the top hBN layer has been removed (etched windows) (b): Spatially averaged Raman spectra of the E2g modes forhBN and the graphene G band. The spatial averages correspond to the pink (etched windows, D1_window2) and blue (fullyencapsulated heterostructure) regions indicated in panel a (c) and (d): Correlation plots of the positions of the graphene G and2D bands before (c) and after (d) AFM brooming. The pink and blue data points correspond to the etched windows and fullyencapsulated areas shown in panel a, respectively. The black dashed lines represent theoretical predictions of (ωG, ω2D) forgraphene under constant doping and uniform strain [35] (e) and (f): Similar correlation plots as in panels c and d, but showingthe relationships between the linewidths of the G and 2D peaks. The dashed lines, with a slope of 2.2, correspond to the expectedtrend for strain-induced broadening of the Raman peaks [10], illustrating the increased broadening due to strain variationswithin the laser spot area.observed the 2D band in our exfoliated graphene monolayer sample, with intensity more than 5x times largerthan the G band and with a Lorentzian profile, an indicator of weakly doped monolayer graphene [30](supplementary information 3). At no point did we detect the emergence of the D or D’ bands fromgraphene [31], even after opening the hBN window (figure 3(b)). This absence of the defect bands suggestsgood preservation of the crystalline structure after etching and therefore negligible fluorination.We first performed hyperspectral mapping of the sample (figure 3(a)), wherein the laser spot(500 nm diameter) was scanned over the device, and a spectrum was recorded at each point. We identifiedthe etched zones—D1_window1/2—by plotting the hBN E2g phonon peak area (AhBN) centred at1362 cm−1 (figure 3(b)) defined as AhBN = π/2 (IhBNΓhBN), where IhBN and ΓhBN refer to the peak intensityand peak width respectively. Removing hBN material alters the optical planar cavity formed by theheterostructures on the SiO2 dielectric and the back silicon mirror [32]. The interferences, governed by thelocal optical gain dependent on different layers’ thicknesses and refractive indices, affect the pump laser (at532 nm) and the Raman scattered light [33]. Consequently, the net interference pattern can be complex [34].To better observe the effect of removing the hBN top layer, we normalized the hBN peak area by the G bandarea, assuming it remains constant during etching. This assumption is supported by the absence of the Dband post etching, suggesting good preservation of sp2 carbon bonds and confirming graphene’s crystallineintegrity [34]. We report a∼ 35% reduction in the hBN area signal in the window zone (figure 3(b)).Interestingly, a 55% decrease was expected, given the top hBN thickness of 17.5 nm and the bottom of 14 nm(measured via AFM). This discrepancy may result from normalizing by the G band area, which does not fullyeliminate the Raman interference effect, as there is still a small shift in the scattered light wavelength betweenthe two peaks.To further analyse the quality of graphene after the etching, we plotted the correlation between the G and2D band positions [35] (figure 3(c)), isolating clusters of points corresponding to the D1_window2 area4J. Phys. Mater. 8 (2025) 045006 H Agarwal et alFigure 4. Quantum transport measurements in etched region (a), mobility (µ) as a function of carrier doping (n) measured at10 K in the same region of the device before (blue circles) and after (pink circles) etching D1_window1 and D2_window2. Thegreen solid line traces the device width limited mobility. The top right inset sketches the measurement geometry. (b), Magneticfocusing resistance RTMF plotted as a function of carrier doping (n) and magnetic field (B) at 10 K. Top left inset plots a line cutfor carrier density n=− 3.1× 1012 cm−2. Top right inset sketches the measurement geometry. (c) Bend resistance RB plotted asa function of carrier doping n at 10 K for two geometries V1 and V2. Top right inset sketches the measurement geometries. (d),µ(n) measured around D1_window2 (orange) and in the fully encapsulated region (purple) at 300 K.(pink) and the fully encapsulated zone (blue). The two rectangular zones are shown in figure 3(a). Althoughboth clusters are close, we observe a systematic shift typically linked to doping differences between regions.This shift aligns with a local change in the Fermi level due to contaminant absorption directly on graphene orfluorination [35]. Following the framework analysis of previous works [10, 35], which shows doping around0.1× 1012 cm–2, consistent with our low-temperature quantum transport measurements (see supplementarysection 5). In contrast, encapsulated graphene remains protected from the environment. In both cases, dataclusters elongate along the strain axis, indicating the strain distribution is not perfectly uniform across thescanned area [10]. Interestingly, elongation is more pronounced in the etched window, suggesting thatremoving the top encapsulant caused a local strain redistribution within the monolayer.We then performed an additional step called AFM brooming, where an AFM tip in scanning contactmode cleans the exposed surface, acting as a nanoscale broom. We repeated the same Raman analysis afterbrooming (figure 3(d)) and observed that the data clusters almost perfectly overlap, indicating efficientremoval of contaminants and a nearly identical Fermi level in both etched and encapsulated areas. FurtherRaman correlation analysis, comparing G and 2D linewidths (figures 3(e) and (f)), provides insight intostrain distribution at the nanoscale [10]. Here, we observe a clear difference in strain distribution betweenthe fully encapsulated pristine heterostructure and the etched window. The higher center position of the datacluster for the etched window suggests more inhomogeneous strain distribution at the nanoscale, varyingover a smaller length scale than in the fully encapsulated case. At first glance, this may seem unexpected sinceencapsulation between hBN layers initially established graphene’s strain distribution. However, removing thetop hBN layer creates mechanical asymmetry and relaxes boundary conditions that stabilize straindistribution [36, 37]. As a result, graphene undergoes local strain redistribution when the top encapsulant isremoved, following strain transfer mechanisms similar to those in thin-film structures [38]. Notably, straindistribution does not change significantly after brooming, but the two clusters in figure 3(f) move closer dueto reduced and more uniform doping.Next, we characterized exposed regions using low-temperature quantum transport, comparing its qualitywith previous measurements before etching and in fully encapsulated regions. The blue curve in figure 4(a)plots the mobility extracted from the Drude conductivity measured at 10 K before etching. It showsmobilities larger than 400,000 cm2V–1s−1 over the entire doping range. In pristine grapheneheterostructures, the channel is so clean that carriers do not scatter until reaching device edges, resulting in amobility limited by the device width [23, 24]. The green solid line in figure 4(a) plots the width-limitedmobility corresponding to our devices (w = 8 µm). For hole doping (n< 0), mobilities reach this limitdemonstrating excellent electronic quality. For electron doping mobility is slightly less, potentially due toimpurities in the hBN substrate. Because of the excellent electronic quality, our devices exhibited clearsignatures of ballistic transport, including magnetic focusing [39–41] and negative bend resistance [7].The AFM and Raman data from D1_window2 are particularly promising, suggesting high-qualitygraphene can be preserved even after removing top hBN provided AFM brooming [26, 27] is employed. Tobenchmark electronic quality, we performed low-temperature quantum magneto-transport measurements,comparing device characteristics before and after opening the windows. The pink data in figure 4(a) plots themobility of our device measured at 10 K in the same region, using the same contact pairs, before the window5J. Phys. Mater. 8 (2025) 045006 H Agarwal et alwas opened (inset of figure 1(a)). While the mobility decreases for all the doping, it still remains relativelyhigh> 200,000 cm2 V–1s−1, competitive with the state-of-the-art reports of graphene on hBN withoutencapsulation [6, 42]. Even with pristine etching that avoids graphene fluorination, some degradation may beexpected because the graphene is now exposed to atmospheric conditions and may be sensitive to absorbates.To further assess device quality, we performed magnetic focusing experiments [39–41] (inset offigure 4(b)). These measurements serve as an excellent characterization tool, tracking ballistic trajectories ofsemi-classical quasi-particles exhibiting Lorenz-like motion. The measurement geometry is sketched in theinset of figure 4(b). In magnetic focusing experiments, ballistic carriers injected at side contacts are curved byfinite magnetic fields and collected at adjacent electrodes. For specific values of magnetic field, when thesemi-classical cyclotron radius is commensurate with contact spacing, resistance peaks appear. We choose ageometry that tracks ballistic properties through the etched region. An example of the resonances can be seenclearly in the inset of figure 4(b) showing an oscillatory structure appearing for one sign of the magneticfield. Figure 4(b) plots the magnetic focusing resistance RTMF as a function of magnetic field (B) and carrierdensity (n). Notably, strong magnetic focusing resonances can be observed for electron and hole doping. Forhole doping, higher-order resonances corresponding to multiple reflections from device edges are visible upto p= 5. The observation of magnetic focusing demonstrates the excellent quality of our devices even afterremoval of the top hBN. To our knowledge, this is the first demonstration of ballistic transport overmicron-length scales in single-sided graphene encapsulation. To further characterize the ballistic properties,we performed bend resistance measurements. The geometries are sketched in the inset of figure 4(c). For V1we observe a strong negative bend resistance appears with doping indicating carriers propagate ballisticallyacross the device over 8 µm. In V2 the negative response was strongly suppressed, likely due to trajectoriespassing through the etched region where the mobility is lower (figure 4(a)). Nonetheless, ballistic carriersremain detectable (figure 4(c) inset). The strong negative bend resistance in V1 confirms that unetchedregions retain high electronic quality after SF6 treatment. Similar observations were made near D1_window1.Although mobility degradation was more severe due to surface contamination, high-order magnetic focusingfeatures remained visible, corresponding to trajectories avoiding the window region (see supplementarysection 4). This further demonstrates that high electronic quality is preserved in unetched regions.Figure 4(a) shows that carrier mobility does not degrade significantly at 10 K and remains high across thedoping range. For device applications, assessing quality at room temperature is also crucial. Figure 4(d) plotsmobility measured in the etched region (D1_window2) and pristine fully encapsulated regions at 300 K. Atthese temperatures, mobility appears largely unaffected by opening a window in the hBN, remaining nearlyidentical for hole doping. We attribute this behavior to dominant phonon contributions, which limitgraphene’s intrinsic mobility. In other words, at room temperature, etched regions approach the intrinsicmobility limits of graphene encapsulated with hBN [11] with competitive values> 50,000 cm2V–1s–1.1. DiscussionOur measurements show that clean in-situ patterning of top hBN in van der Waals heterostructures can bemade without significantly degrading the electronic quality of underlying 2D layers. AFM measurementsconfirm that pristine surfaces can be opened if AFM brooming is performed. Our experiments showed thatthis step is essential in obtaining clean surfaces after etching with< 1 nm surface roughness (figures 2(a) and(b)). However, some variability in achievable surface roughness on the sub nanometer scale remains to beunderstood. The cleanest windows were observed in D2_window3 (figure 2(c)) and D3_window4(supplementary section 2), showing regions of atomically flat surfaces (< 200 pm). This enhanced qualitycompared to D1_window2 may be due to several factors. First, molecular adsorbates may contribute to thesurface roughness [43]. Second, the AFM brooming conditions may require further optimization, dependingon the number of pre-etching steps involving PMMA deposition. In D1, polymer masks were deposited andwashed twice for opening D1_window1 and then D1_window2. In contrast, D2_window3 and D3_window4underwent one etching procedure, uniquely defined on different devices. Furthermore, the PMMAmask wasnot removed after etching in D3_window4. In summary, D2/D3 generally faced less exposure to PMMAcontamination than D1. This step may be important for achieving the atomically pristine surfaces, aswashing in solvents can lead to nanoscale contaminants adsorbing onto the window region. Nonetheless,residual PMMA can still be removed through high-temperature annealing. This was evident in D2_window3which showed a vast improvement in surface roughness post-annealing (see supplementary section 7).Further work is needed to optimize processing steps for reproducing pristine atomic surfaces.In some applications, selective etching may be needed before depositing contacts, such as when 2Dcontacts to exposed graphene are required. In this case, Raman spectroscopy can characterize graphenequality due to its sensitivity to fluorination, serving as an initial screening of the active 2D layer beforecontact deposition. Strong fluorination is typically detected through the appearance of D’ peaks [44, 45].6J. Phys. Mater. 8 (2025) 045006 H Agarwal et alHowever, previous studies required long exposure times to resolve Raman signatures. Our etching recipes usesignificantly shorter times, lasting minutes rather than hours. Thus, it is unsurprising that excess fluorinationis not observed. However, low-temperature quantum transport measurements show slight qualitydegradation. When plotting resistivity as a function of carrier doping (see supplementary section 5), wenotice peak splitting at the Dirac point, suggesting doping inhomogeneities in the sample. This may resultfrom slight fluorination undetectable by Raman signals or surface contaminants. Notably, brooming thedevice surface after etching seemingly restored pristine quality (figure 3(d)/(f)), indicating degradation likelyoriginates from surface contamination. While there may be some minute/undetectable damage to graphenedue to electron irradiation, we believe it is much less compared to what is normally induced via dry etchingor chemical processes, and hence not relevant in the context of our work.Finally, we note that over-etching can degrade graphene´s electronic quality. This was evident inD2_window3, where we intentionally over-etched for 2 min. Raman and quantum transport measurementson this sample showed strong doping inhomogeneities post-etching (see supplementary section 6). Thisexperiment highlights the need for careful calibration of etching procedures and demonstrates thatroom-temperature Raman spectroscopy effectively probes surface quality in exposed graphene, even withoutthe presence of D’ bands. However, the surface can be further treated through high-temperature annealing(600◦ C in vacuum). This is evident in Raman spectroscopy measurements performed on D3 before andafter annealing (see supplementary section 6), showing that doping and strain profiles changepost-annealing. Additionally, annealing at moderate temperatures (250◦ C in an argon-hydrogen mixturewith 10% H2) may help reverse fluorination effects [44].2. ConclusionOur experiments outline a new technique for engineering the top encapsulating hBN in van der Waalsheterostructures. We demonstrate the technique can be used to open windows into graphene encapsulatedwith hexagonal-boron nitride, while preserving the electronic quality of the 2D surface. Aside from hBN,preliminary experiments on selectively etching transition-metal dichalcogenides (see supplementary section8), show that high quality of underlying graphene can also be preserved, highlighting the broaderapplicability to graphene encapsulated with 2D semiconductors [46]. The technique has strong prospects inscanning probe microscopy, enabling the possibility to access fully encapsulated quantum transport devicesto bridge the gap in understanding between global and local measurements. With further optimization, theprocedure offers a more controllable method for obtaining pristine surfaces compared to the stack-and-flipmethod currently employed [20]. Cryogenic techniques may further enhance controllability and resolutionin advanced nano-patterned structures [47]. It may also be used in scattering near-field optical microscopy(SNOM) experiments, enhancing the light-matter coupling between the tip and the underlying substrate[48]. From an application perspective, the capability to pattern the top hBN offers exciting opportunities.Aside from chemical sensing, it offers unique directions in 2D meta-materials [49] including engineering ofpolaritonic launches, excitonic landscapes, advanced [50] and artificial superlattice structures [51, 52].3. Methods3.1. Device fabricationThe samples are fabricated using typical methods in heterostructure assembly. Typically, a thin hBN flake(∼10–15 nm) is picked using hot-pick up technique [22, 53] using a polypropylene carbonate (PC) film on apolydimethylsiloxane (PDMS) stamp at 90◦ C. This hBN flake is then later used to pick up the graphenemonolayers, mechanically exfoliated on Si++/SiO2 (285 nm) from highly oriented pyrolytic graphite, andpre-characterized using optical microscopy, and Raman spectroscopy [28]. In D3, twisted graphene wasassembled between hBN layers and an additional graphite layer was picked up after the bottom hBN. Finally,the stack is used to pick up a last layer of hBN and later dropped on a pre-patterned marker chip ofSi++/SiO2 (285 nm) at 180◦ C, squeezing out the bubbles, and impurities as previously reported. The stack isthen shaped into a Hall bar geometry using SF6 plasma, O2 plasma to etch top hBN, and graphenerespectively, and further metalized using 3/15/30 nm of Cr/Pd/Au. The thickness of hBN flakes is identifiedby their color shading through optical contrast in a microscope, and consequently via AFM when assembledin the heterostructure. The heights can be distinguished from steps in the height profiles induced by exposedgraphene edges.3.2. SF6 etching ratesThe etching was performed in an OXFORD or SAMCO RIE (see supplementary information 1 for detailedrecipe). For D1_window1, D1_window2, and D2_Window3, etching were done in OXFORD RIE with hBN7J. Phys. Mater. 8 (2025) 045006 H Agarwal et aletch rate of∼ 7 nm/min. For D3_window4, etching was done in SAMCO RIE with hBN etch rateof∼ 5 nm/min. Before etching, the RIE chamber was pre-conditioned for 15 min with the same SF6 recipe.3.3. Raman spectroscopyRaman spectroscopy measurements were performed at room temperature using a 532 nm laser with opticalpower of 0.5 mW, focused on a 500 nm spot.3.4. Quantum transportLow-temperature quantum transport measurements were performed on an Advanced Research Systems 4 KCryostat. Measurements were performed using standard lock-in techniques (SR 860) in constant currentmode sourcing a small AC current (<100 nA) while measuring the four-probe voltage. For magnetic fieldmeasurements, a 1 T electromagnet was used (GMW associates).Data availability statementAll data supporting the findings of this study are publicly available under the Creative Commons Attribution4.0 International License (CC-BY 4.0) via Zenodo at: https://doi.org/10.5281/zenodo.15869927 (reference[54]). Additional data including those from the Supplementary Information are available from thecorresponding authors upon request.AcknowledgmentsWe would like to thank Carmen Rubio-Verdú (ICFO), Tymofiy Khodkov (ICFO), Lene Gammelgaard(DTU), Juan Sierra, Patricia Aguilar, Sergio Valenzuela (ICN2) and Qian Yang (University of Manchester) forimportant insights and useful discussions. We further thank Matteo Ceccanti for making the illustrationpresented in figure 1. H.A. acknowledges funding from the European Union’s Horizon 2020 research andinnovation programme under Marie Skłodowska-Curie grant Agreement No. 665884. A.R.-P. acknowledgesfunding from ANR JCJC NEAR-2D and Welcome Package Idex (UniCA) as well as AAP Tremplin Complex2023 ‘2DNEUROTWIST’ (ANR-15-IDEX-01). R.K.K. acknowledges funding by MCIN/AEI/10.13039/501100011033 and by the ‘European Union NextGenerationEU/PRTR’ PCI2021-122020-2A within theFLAG-ERA grant [PhotoTBG], by ICFO, RWTH Aachen and ETHZ/Department of Physics.R. K. K. also acknowledges support from the Ramon y Cajal Grant RYC2022-036118-I funded byMICIU/AEI/10.13039/501100011033 and by ‘ESF+’. F.H.L.K. acknowledges support from the Gordon andBetty Moore Foundation through Grant GBMF12212, and the Government of Spain (QTWISTRD0768/2022, PID2022-141081NB-I00; Severo Ochoa CEX2019-000910-S, and CEX2024- 001490-S[MCIN/ AEI/10.13039/501100011033). This work was also supported by the European UnionNextGenerationEU/PRTR (PRTR-C17.I1) and EXQIRAL 101131579, Fundació Cellex, Fundació Mir-Puig,Generalitat de Catalunya (CERCA, Department of Digital Policies and Territory, AGAUR, 2021 SGR014431656), and BBVA Foundation 2022 Fundamentals Program (An Electronic Quantum Simulator).Views and opinions expressed are those of the author(s) only and do not necessarily reflect those of theEuropean Union Research Executive Agency. Neither the European Union nor the granting authority can beheld responsible for them. This material is based upon work supported by the Air Force Office of ScientificResearch under Award Number FA8655-23-1-7047. Any opinions, findings, conclusions, orrecommendations expressed in this material are those of the author(s) and do not necessarily reflect theviews of the United States Air Force.ORCID iDsHitesh Agarwal 0000-0002-9418-7966Antoine Reserbat-Plantey 0000-0002-9106-8750David Barcons Ruiz 0000-0002-6271-2244Karuppasamy Pandian Soundarapandian 0000-0002-9664-9095Vahagn Mkhitaryan 0000-0002-3138-8488Johann Osmond 0000-0001-6522-4695Kenji Watanabe 0000-0003-3701-8119Petr Stepanov 0000-0002-1121-3146Frank H L Koppens 0000-0001-9764-6120Roshan Krishna Kumar 0000-0003-0857-44668https://creativecommons.org/licenses/by/4.0/https://doi.org/10.5281/zenodo.15869927https://orcid.org/0000-0002-9418-7966https://orcid.org/0000-0002-9418-7966https://orcid.org/0000-0002-9106-8750https://orcid.org/0000-0002-9106-8750https://orcid.org/0000-0002-6271-2244https://orcid.org/0000-0002-6271-2244https://orcid.org/0000-0002-9664-9095https://orcid.org/0000-0002-9664-9095https://orcid.org/0000-0002-3138-8488https://orcid.org/0000-0002-3138-8488https://orcid.org/0000-0001-6522-4695https://orcid.org/0000-0001-6522-4695https://orcid.org/0000-0003-3701-8119https://orcid.org/0000-0003-3701-8119https://orcid.org/0000-0002-1121-3146https://orcid.org/0000-0002-1121-3146https://orcid.org/0000-0001-9764-6120https://orcid.org/0000-0001-9764-6120https://orcid.org/0000-0003-0857-4466https://orcid.org/0000-0003-0857-4466J. Phys. Mater. 8 (2025) 045006 H Agarwal et alReferences[1] Kubota Y, Watanabe K, Tsuda O and Taniguchi T 2007 Deep ultraviolet light-emitting hexagonal boron nitride synthesized atatmospheric pressure Science 317 932–4[2] Zhang K, Feng Y, Wang F, Yang Z and Wang J 2017 Two dimensional hexagonal boron nitride (2D-hBN): synthesis, properties andapplications J. Mater. Chem. C 5 11992–2022[3] Molaei M J, Younas M and Rezakazemi M 2021 A comprehensive review on recent advances in two-dimensional (2D) hexagonalboron nitride ACS Appl. Electron. Mater. 3 5165–87[4] Moon S, Kim J, Park J, Im S, Kim J, Hwang I and Kim J K 2023 Hexagonal boron nitride for next-generation photonics andelectronics Adv. Mater. 35 2204161[5] Young A F, Dean C R, Meric I, Sorgenfrei S, Ren H, Watanabe K, Taniguchi T, Hone J, Shepard K L and Kim P 2012 Electroniccompressibility of layer-polarized bilayer graphene Phys. Rev. B 85 235458[6] Dean C R et al 2010 Boron nitride substrates for high-quality graphene electronics Nat. Nanotechnol. 5 722–6[7] Mayorov A S et al 2011 Micrometer-scale ballistic transport in encapsulated graphene at room temperature Nano Lett. 11 2396–9[8] Liu Y, Weiss N O, Duan X, Cheng H-C, Huang Y and Duan X 2016 Van der Waals heterostructures and devices Nat. Rev. Mater.1 16042[9] Andrei E Y, Efetov D K, Jarillo-Herrero P, MacDonald A H, Mak K F, Senthil T, Tutuc E, Yazdani A and Young A F 2021 Themarvels of moiré materials Nat. Rev. Mater. 6 201–6[10] Neumann C et al 2015 Raman spectroscopy as probe of nanometre-scale strain variations in graphene Nat. Commun. 6 8429[11] Wang L et al 2013 One-dimensional electrical contact to a two-dimensional material. science (1979) Science 342 614–7[12] Koepfli S M et al 2024 Controlling photothermoelectric directional photocurrents in graphene with over 400GHz bandwidth Nat.Commun. 15 7351[13] Liu Y, Dong X and Chen P 2012 Biological and chemical sensors based on graphene materials Chem. Soc. Rev. 41 2283–307[14] Guarochico-Moreira V H et al 2022 Tunable spin injection in high-quality graphene with one-dimensional contacts Nano Lett.22 935–41[15] Wong D, Nuckolls K P, Oh M, Lian B, Xie Y, Jeon S, Watanabe K, Taniguchi T, Bernevig B A and Yazdani A 2020 Cascade ofelectronic transitions in magic-angle twisted bilayer graphene Nature 582 198–202[16] Choi Y et al 2019 Electronic correlations in twisted bilayer graphene near the magic angle Nat. Phys. 15 1174–80[17] Xue J, Sanchez-Yamagishi J, Bulmash D, Jacquod P, Deshpande A, Watanabe K, Taniguchi T, Jarillo-Herrero P and LeRoy B J 2011Scanning tunnelling microscopy and spectroscopy of ultra-flat graphene on hexagonal boron nitride Nat. Mater. 10 282–5[18] Rubio-Verdú C et al 2022 Moiré nematic phase in twisted double bilayer graphene Nat. Phys. 18 196–202[19] Utama M I B et al 2021 Visualization of the flat electronic band in twisted bilayer graphene near the magic angle twist Nat. Phys.17 184–8[20] Lisi S et al 2021 Observation of flat bands in twisted bilayer graphene Nat. Phys. 17 189–93[21] Grenadier S, Li J, Lin J and Jiang H 2013 Dry etching techniques for active devices based on hexagonal boron nitride epilayersJ. Vac. Sci. Technol. A. 31 061517[22] Pizzocchero F, Gammelgaard L, Jessen B S, Caridad J M, Wang L, Hone J, Bøggild P and Booth T J 2016 The hot pick-up techniquefor batch assembly of van der Waals heterostructures Nat. Commun. 7 11894[23] Kumaravadivel P et al 2019 Strong magnetophonon oscillations in extra-large graphene Nat. Commun. 10 3334[24] Wang W et al 2023 Clean assembly of van der Waals heterostructures using silicon nitride membranes Nat. Electron. 6 981–90[25] Lindvall N, Kalabukhov A and Yurgens A 2012 Cleaning graphene using atomic force microscope J. Appl. Phys. 111 064904[26] Rosenberger M R, Chuang H-J, McCreary K M, Hanbicki A T, Sivaram S V and Jonker B T 2018 Nano-“squeegee” for the creationof clean 2D material interfaces ACS Appl. Mater. Interfaces 10 10379–87[27] Goossens A M et al 2012 Mechanical cleaning of graphene Appl. Phys. Lett. 100 073110[28] Ferrari A C et al 2006 Raman spectrum of graphene and graphene layers Phys. Rev. Lett. 97 187401[29] Gorbachev R V et al 2011 Hunting for monolayer boron nitride: optical and Raman signatures Small 7 465–8[30] Basko D M, Piscanec S and Ferrari A C 2009 Electron-electron interactions and doping dependence of the two-phonon Ramanintensity in graphene Phys. Rev. B 80 165413[31] Eckmann A, Felten A, Mishchenko A, Britnell L, Krupke R, Novoselov K S and Casiraghi C 2012 Probing the nature of defects ingraphene by raman spectroscopy Nano Lett. 12 3925–30[32] Blake P, Hill E W, Castro Neto A H, Novoselov K S, Jiang D, Yang R, Booth T J and Geim A K 2007 Making graphene visible Appl.Phys. Lett. 91 063124[33] Yoon D, Moon H, Son Y-W, Choi J S, Park B H, Cha Y H, Kim Y D and Cheong H 2009 Interference effect on Raman spectrum ofgraphene on SiO2/Si Phys. Rev. B 80 125422[34] Reserbat-Plantey A, Klyatskaya S, Reita V, Marty L, Arcizet O, Ruben M, Bendiab N and Bouchiat V 2013 Time- andspace-modulated Raman signals in graphene-based optical cavities J. Opt. 15 114010[35] Lee J E, Ahn G, Shim J, Lee Y S and Ryu S 2012 Optical separation of mechanical strain from charge doping in graphene Nat.Commun. 3 1024[36] Vincent T, Panchal V, Booth T, Power S R, Jauho A-P, Antonov V and Kazakova O 2019 Probing the nanoscale origin of strain anddoping in graphene-hBN heterostructures 2D Mater. 6 015022[37] Peña T, Chowdhury S A, Azizimanesh A, Sewaket A, Askari H and Wu S M 2021 Strain engineering 2D MoS2 with thin film stresscapping layers 2D Mater. 8 045001[38] Schadler L S and Noyan I C 1991 Experimental determination of the strain transfer across a flexible intermediate layer in thin filmstructures as a function of flexible layer thicknessMRS Online Proc. Library 239 151–6[39] Berdyugin A I et al 2024 Minibands in twisted bilayer graphene probed by magnetic focusing Sci. Adv. 6 eaay7838[40] Lee M et al 2016 Ballistic miniband conduction in a graphene superlattice Science 353 1526–9[41] Taychatanapat T, Watanabe K, Taniguchi T and Jarillo-Herrero P 2013 Electrically tunable transverse magnetic focusing ingraphene Nat. Phys. 9 225–9[42] Amet F, Bestwick A J, Williams J R, Balicas L, Watanabe K, Taniguchi T and Goldhaber-Gordon D 2015 Composite fermions andbroken symmetries in graphene Nat. Commun. 6 5838[43] Temiryazev A, Frolov A and Temiryazeva M 2019 Atomic-force microscopy study of self-assembled atmospheric contamination ongraphene and graphite surfaces Carbon Capture Sci. Technol. 143 30–37[44] Nair R R et al 2010 Fluorographene: a two-dimensional counterpart of Teflon Small 6 2877–849https://doi.org/10.1126/science.1144216https://doi.org/10.1126/science.1144216https://doi.org/10.1039/C7TC04300Ghttps://doi.org/10.1039/C7TC04300Ghttps://doi.org/10.1021/acsaelm.1c00720https://doi.org/10.1021/acsaelm.1c00720https://doi.org/10.1002/adma.202204161https://doi.org/10.1002/adma.202204161https://doi.org/10.1103/PhysRevB.85.235458https://doi.org/10.1103/PhysRevB.85.235458https://doi.org/10.1038/nnano.2010.172https://doi.org/10.1038/nnano.2010.172https://doi.org/10.1021/nl200758bhttps://doi.org/10.1021/nl200758bhttps://doi.org/10.1038/natrevmats.2016.42https://doi.org/10.1038/natrevmats.2016.42https://doi.org/10.1038/s41578-021-00284-1https://doi.org/10.1038/s41578-021-00284-1https://doi.org/10.1038/ncomms9429https://doi.org/10.1038/ncomms9429https://doi.org/10.1126/science.1244358https://doi.org/10.1126/science.1244358https://doi.org/10.1038/s41467-024-51599-whttps://doi.org/10.1038/s41467-024-51599-whttps://doi.org/10.1039/C1CS15270Jhttps://doi.org/10.1039/C1CS15270Jhttps://doi.org/10.1021/acs.nanolett.1c03625https://doi.org/10.1021/acs.nanolett.1c03625https://doi.org/10.1038/s41586-020-2339-0https://doi.org/10.1038/s41586-020-2339-0https://doi.org/10.1038/s41567-019-0606-5https://doi.org/10.1038/s41567-019-0606-5https://doi.org/10.1038/nmat2968https://doi.org/10.1038/nmat2968https://doi.org/10.1038/s41567-021-01438-2https://doi.org/10.1038/s41567-021-01438-2https://doi.org/10.1038/s41567-020-0974-xhttps://doi.org/10.1038/s41567-020-0974-xhttps://doi.org/10.1038/s41567-020-01041-xhttps://doi.org/10.1038/s41567-020-01041-xhttps://doi.org/10.1116/1.4826363https://doi.org/10.1116/1.4826363https://doi.org/10.1038/ncomms11894https://doi.org/10.1038/ncomms11894https://doi.org/10.1038/s41467-019-11379-3https://doi.org/10.1038/s41467-019-11379-3https://doi.org/10.1038/s41928-023-01075-yhttps://doi.org/10.1038/s41928-023-01075-yhttps://doi.org/10.1063/1.3695451https://doi.org/10.1063/1.3695451https://doi.org/10.1021/acsami.8b01224https://doi.org/10.1021/acsami.8b01224https://doi.org/10.1063/1.3685504https://doi.org/10.1063/1.3685504https://doi.org/10.1103/PhysRevLett.97.187401https://doi.org/10.1103/PhysRevLett.97.187401https://doi.org/10.1002/smll.201001628https://doi.org/10.1002/smll.201001628https://doi.org/10.1103/PhysRevB.80.165413https://doi.org/10.1103/PhysRevB.80.165413https://doi.org/10.1021/nl300901ahttps://doi.org/10.1021/nl300901ahttps://doi.org/10.1063/1.2768624https://doi.org/10.1063/1.2768624https://doi.org/10.1103/PhysRevB.80.125422https://doi.org/10.1103/PhysRevB.80.125422https://doi.org/10.1088/2040-8978/15/11/114010https://doi.org/10.1088/2040-8978/15/11/114010https://doi.org/10.1038/ncomms2022https://doi.org/10.1038/ncomms2022https://doi.org/10.1088/2053-1583/aaf1dchttps://doi.org/10.1088/2053-1583/aaf1dchttps://doi.org/10.1088/2053-1583/ac08f2https://doi.org/10.1088/2053-1583/ac08f2https://doi.org/10.1126/sciadv.aay7838https://doi.org/10.1126/sciadv.aay7838https://doi.org/10.1126/science.aaf1095https://doi.org/10.1126/science.aaf1095https://doi.org/10.1038/nphys2549https://doi.org/10.1038/nphys2549https://doi.org/10.1038/ncomms6838https://doi.org/10.1038/ncomms6838https://doi.org/10.1016/j.carbon.2018.10.094https://doi.org/10.1016/j.carbon.2018.10.094https://doi.org/10.1002/smll.201001555https://doi.org/10.1002/smll.201001555J. Phys. Mater. 8 (2025) 045006 H Agarwal et al[45] Withers F, Russo S, Dubois M and Craciun M F 2011 Tuning the electronic transport properties of graphene throughfunctionalisation with fluorine Nanoscale Res. Lett. 6 526[46] Soundarapandian K P, De Fazio D, Bernal-Texca F, Hoffmann R, Ceccanti M, De Bonis S L, Tongay S and Koppens F H L 2023Hysteresis-free high mobility graphene encapsulated in tungsten disulfide Appl. Phys. Lett. 123 063103[47] Xu J et al 2024 Deep-reactive ion etching of silicon nanowire arrays at cryogenic temperatures Appl. Phys. Rev. 11 021411[48] Hesp N C H et al 2021 Observation of interband collective excitations in twisted bilayer graphene Nat. Phys. 17 1162–8[49] Danielsen D R, Lyksborg-Andersen A, Nielsen K E S, Jessen B S, Booth T J, DoanM-H, Zhou Y, Bøggild P and Gammelgaard L 2021Super-resolution nanolithography of two-dimensional materials by anisotropic etching ACS Appl. Mater. Interfaces 13 41886–94[50] Kapfer M et al 2023 Programming twist angle and strain profiles in 2D materials Science 381 677–81[51] Forsythe C, Zhou X, Watanabe K, Taniguchi T, Pasupathy A, Moon P, Koshino M, Kim P and Dean C R 2018 Band structureengineering of 2D materials using patterned dielectric superlattices Nat. Nanotechnol. 13 566–71[52] Barcons Ruiz D et al 2022 Engineering high quality graphene superlattices via ion milled ultra-thin etching masks Nat. Commun.13 6926[53] Purdie D G, Pugno N M, Taniguchi T, Watanabe K, Ferrari A C and Lombardo A 2018 Cleaning interfaces in layered materialsheterostructures Nat. Commun. 9 5387[54] Agarwal H et al 2025 Data for main text figures - In situ engineering hexagonal boron nitride in van der Waals heterostructureswith selective SF6 etching Zenodo (https://doi.org/10.5281/zenodo.15869927)10https://doi.org/10.1186/1556-276X-6-526https://doi.org/10.1186/1556-276X-6-526https://doi.org/10.1063/5.0151273https://doi.org/10.1063/5.0151273https://doi.org/10.1063/5.0166284https://doi.org/10.1063/5.0166284https://doi.org/10.1038/s41567-021-01327-8https://doi.org/10.1038/s41567-021-01327-8https://doi.org/10.1021/acsami.1c09923https://doi.org/10.1021/acsami.1c09923https://doi.org/10.1126/science.ade9995https://doi.org/10.1126/science.ade9995https://doi.org/10.1038/s41565-018-0138-7https://doi.org/10.1038/s41565-018-0138-7https://doi.org/10.1038/s41467-022-34734-3https://doi.org/10.1038/s41467-022-34734-3https://doi.org/10.1038/s41467-018-07558-3https://doi.org/10.1038/s41467-018-07558-3https://doi.org/10.5281/zenodo.15869927 In situ engineering hexagonal boron nitride in van der Waals heterostructures with selective SF6 etching 1. Discussion 2. Conclusion 3. Methods 3.1. Device fabrication 3.2. SF6 etching rates 3.3. Raman spectroscopy 3.4. Quantum transport References