# Fileset

[s41467-020-16817-1.pdf](https://mdr.nims.go.jp/filesets/a978d29a-6fac-469a-bf18-86c0ec01e9a7/download)

## Creator

Zhujun Huang, Abdullah Alharbi, William Mayer, Edoardo Cuniberto, [Takashi Taniguchi](https://orcid.org/0000-0002-1467-3105), [Kenji Watanabe](https://orcid.org/0000-0003-3701-8119), Javad Shabani, Davood Shahrjerdi

## Rights

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

## Other metadata

[Versatile construction of van der Waals heterostructures using a dual-function polymeric film](https://mdr.nims.go.jp/datasets/5da4b4d4-e4f3-4a89-9117-e47273d7e694)

## Fulltext

Versatile construction of van der Waals heterostructures using a dual-function polymeric filmARTICLEVersatile construction of van der Waalsheterostructures using a dual-function polymericfilmZhujun Huang1,5, Abdullah Alharbi 1,2,5, William Mayer3, Edoardo Cuniberto1, Takashi Taniguchi4,Kenji Watanabe 4, Javad Shabani3 & Davood Shahrjerdi 1,3✉The proliferation of van der Waals (vdW) heterostructures formed by stacking layeredmaterials can accelerate scientific and technological advances. Here, we report a strategy forconstructing vdW heterostructures through the interface engineering of the exfoliationsubstrate using a sub-5 nm polymeric film. Our construction method has two main featuresthat distinguish it from existing techniques. First is the consistency of its exfoliation processin increasing the yield and in producing large (>10,000 μm2) monolayer graphene. Second isthe applicability of its layer transfer process to different layered materials without requiring aspecialized stamp—a feature useful for generalizing the assembly process. We demonstratevdW graphene devices with peak carrier mobility of 200,000 and 800,000 cm2 V−1 s−1 atroom temperature and 9 K, respectively. The simplicity of our construction method and itsversatility to different layered materials may open doors for automating the fabricationprocess of vdW heterostructures.https://doi.org/10.1038/s41467-020-16817-1 OPEN1 Electrical and Computer Engineering, New York University, Brooklyn, NY 11201, USA. 2 King Abdulaziz City for Science and Technology, Riyadh 11442, SaudiArabia. 3 Center for Quantum Phenomena, Physics Department, New York University, New York, NY 10003, USA. 4National Institute of Materials Science, 1-1 Namiki Tsukuba, Ibaraki 305-0044, Japan. 5These authors contributed equally: Zhujun Huang, Abdullah Alharbi. ✉email: davood@nyu.eduNATURE COMMUNICATIONS |         (2020) 11:3029 | https://doi.org/10.1038/s41467-020-16817-1 | www.nature.com/naturecommunications 11234567890():,;http://crossmark.crossref.org/dialog/?doi=10.1038/s41467-020-16817-1&domain=pdfhttp://crossmark.crossref.org/dialog/?doi=10.1038/s41467-020-16817-1&domain=pdfhttp://crossmark.crossref.org/dialog/?doi=10.1038/s41467-020-16817-1&domain=pdfhttp://crossmark.crossref.org/dialog/?doi=10.1038/s41467-020-16817-1&domain=pdfhttp://orcid.org/0000-0002-0846-2934http://orcid.org/0000-0002-0846-2934http://orcid.org/0000-0002-0846-2934http://orcid.org/0000-0002-0846-2934http://orcid.org/0000-0002-0846-2934http://orcid.org/0000-0003-3701-8119http://orcid.org/0000-0003-3701-8119http://orcid.org/0000-0003-3701-8119http://orcid.org/0000-0003-3701-8119http://orcid.org/0000-0003-3701-8119http://orcid.org/0000-0002-5955-1830http://orcid.org/0000-0002-5955-1830http://orcid.org/0000-0002-5955-1830http://orcid.org/0000-0002-5955-1830http://orcid.org/0000-0002-5955-1830mailto:davood@nyu.eduwww.nature.com/naturecommunicationswww.nature.com/naturecommunicationsVan der Waals (vdW) heterostructures are an excitingplayground for scientific discoveries among different dis-ciplines1–11. These structures are made by stacking dif-ferent layered materials that are held together by vdW forces. Thevast diversity of layered materials can result in a large number ofpossible combinations of layered material stacks, which could beemployed for future research in a wide range of fields, frommany-body physics to electrochemistry. However, efficientmethodologies for constructing diverse vdW heterostructures arestill lacking. What is needed is a versatile construction methodwith the ability to build arbitrary stacks of high-quality layeredmaterials using an identical fabrication process.Continuous progress on developing standard processes forassembling exfoliated flakes has made vdW heterostructures moreaccessible to the research community12–18. Past research hasdeveloped those standard processes around the direct exfoliationof layered materials on SiO2 substrates19. This method hasremained popular because of the ability to produce clean (i.e.,homogeneous and low-disorder) layered materials. Figure 1a, bshow the schematic representation of this conventional method ofdirect exfoliation on SiO2 and the following layer transfer of theexfoliated flake (e.g., graphene) onto a stamp. However, despitethis significant progress, efficient and high-yield fabrication ofdiverse heterostructures remains a difficult task.A major barrier to the efficient construction of vdW hetero-structures using the conventional method is the opposingrequirements of the exfoliation and layer transfer steps. In prin-ciple, promoting the adhesion between the substrate and thelayered material improves the exfoliation process20,21. However,the adhesion at the flake-substrate interface must be ideally weakduring the transfer process for two important reasons. First, itresults in high transfer yields by facilitating the release of theexfoliated flake. Second, it relaxes the requirement on the adhe-sion strength of other interfaces within the stamp (e.g., interface 2and 3 in Fig. 1b), since the adhesion in those interfaces mustexceed that of the flake-substrate interface.Because of these opposing requirements, the conventionalexfoliation method suffers from small flake size, a low number offlakes, and inconsistency of the exfoliation outcome among dif-ferent trials20,21. Hence, the exfoliation step is typically repeateduntil there are enough layered material flakes available for con-structing the heterostructure. Further, the opposing requirementsof the adhesion strength critically limit the choice of the transferstamp for detaching the layered material flake from SiO2. Thetear-and-stack method using a hexagonal boron nitride (hBN)flake (see Fig. 1b) is currently the most common method forbuilding hBN-encapsulated heterostructures12,16,22. However, tobuild arbitrary stacks of layered materials, each research lab hasdeveloped ad hoc solutions6,23–25. The above limitations make thefabrication of vdW heterostructures using existing methods non-standard, and tedious.An efficient and high-yield fabrication process could beachieved if the adhesion at the flake-substrate interface can beoptimized independently during the exfoliation and transfersteps. Here, we report a strategy that achieves this objective.Specifically, we modify the SiO2 substrate by coating it with a sub-5 nm poly(vinyl alcohol) (PVA) film (see Fig. 1c). We find thatthe adhesion strength of the PVA film can be tuned to promotethe exfoliation of layered materials. For the layer transfer process,we remove the PVA film on-demand, freeing the exfoliated flakesfrom the substrate (Fig. 1d). Our construction method is simple,high-yield, and generalizable to different layered materials.ResultsConsistent exfoliation and layer transfer via interface engi-neering. The exfoliation process begins by coating the SiO2substrate with a sub-5 nm PVA film (see Methods). A Scotch tapecontaining graphite crystals is then pressed onto the substrate,followed by a brief heat treatment at T ≈ 85 °C (which is close tothe glass transition of PVA). The tape is then pulled back slowly.We found that the heat treatment step noticeably promotes theexfoliation of graphene, resulting in large flakes, a high number offlakes per sample, and consistency among different exfoliationruns (discussed later). The illustration in Fig. 1c shows the PVA-assisted graphene exfoliation (PAGE) process.Identifying the number of layers within an exfoliated grapheneflake on a PVA-coated SiO2 can be done with well-establishedpractices19,26,27. We first identified candidate flakes throughoptical inspection. Figure 2a shows an optical image of a 10,000μm2 monolayer graphene produced using PAGE, illustrating theexcellent optical contrast between graphene and PVA. SubsequentRaman measurements on the candidate flakes are essential forobtaining the exact number of layers within each flake. Weobserved that the PVA film does not interfere with the Ramansignature of graphene (Fig. 2b). This property would allow the useof established Raman techniques (e.g., see refs. 27–29) for studyingthe structural properties of the exfoliated graphene flakes onPVA.Our simple interface engineering method is effective inproducing large monolayer flakes (see Supplementary Note 1and Supplementary Fig. 1). Figure 2e compares the dimensions ofthe 100 largest PAGE flakes against those produced using theconventional method. We obtained these flakes from 33 PAGEsamples and 77 conventional exfoliation samples. Each exfoliationtrial comprised one sample and the size of each sample was 1 × 1cm2. The comparison between these two methods shows theTapea�G-G�G-G�G-PVA�G-SiO2�G-hBNSiO2 SiO2SiO2H2OinjectionMXSiO2Exfoliation step Layer transfer step Exfoliation step Layer transfer stepGraphitehBNInterface 3 Polymeric stampsub-5 nm PVAT ≈ 85 °CInterface 2Interface 1Grapheneb c dFig. 1 Fabrication of vdW heterostructures. Schematic representations of a exfoliation and b layer transfer processing steps in the conventional fabricationmethods. In those methods, γG-SiO2 must exceed γG-G during exfoliation. Moreover, γG-hBN must exceed γG-SiO2 during layer transfer. γ denotes the adhesionenergy at the interface between two respective materials in contact with each other. These requirements are conflicting, thus complicating the process ofheterostructure fabrication. We introduce a simple interface engineering method to overcome this dilemma by using a nanoscale PVA release layer. c Inour process, PVA can be tuned to promote exfoliation. d Moreover, selective removal of PVA enables a high-yield layer transfer of the exfoliated flakeswithout requiring a specialized stamp, hence relaxing the choice of the stamp material (i.e., MX). The black arrows indicate the direction of the applieddisplacement force during the exfoliation and layer transfer steps.ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-16817-12 NATURE COMMUNICATIONS |         (2020) 11:3029 | https://doi.org/10.1038/s41467-020-16817-1 | www.nature.com/naturecommunicationswww.nature.com/naturecommunicationsability of our method to produce large monolayer grapheneflakes. The difference in the number of samples between the twomethods in this analysis is because PAGE consistently yields ahigher number of flakes per sample than the conventionalmethod. To study the yield and the consistency of the exfoliation,we analyzed the monolayer flake area distribution on samplesfrom 15 consecutive exfoliation trials, which we performed forboth methods. We accounted for all samples, including those withno monolayer flakes. The results in Fig. 2f indicate that,compared to the conventional method, our method is about 20times more likely to produce monolayer flakes that are biggerthan 1000 μm2 in each sample.Figure 1d shows the schematic illustration of our layer transfermethod. The process begins with landing the transfer stamp ontothe exfoliated flake. A drop of water is then applied locally nearone end of the flake. The stamp is withdrawn gradually at a slowspeed of 20–30 μm s−1 until the flake is fully released (Supple-mentary Movie 1). The slow speed ensures that the stampremains in full contact with the flake while allowing the water topermeate underneath the flake. Depending on the size of theflake, the release process takes only a few to ten seconds. Becauseof the local application of the water drop in our process, theexfoliated flakes in other regions of the substrate remain intactand usable for future fabrication experiments (SupplementaryFig. 4).Our layer transfer method relaxes the adhesion requirements ofthe stamp material. This is a major advance for realizing arbitrarystacks of layered materials. We demonstrated this capability,discussed next, by transferring exfoliated monolayer grapheneflakes onto two stamps that are known to have vastly differentadhesion strength to graphene.The first stamp consisted of a poly-propylene carbonate (PPC)coating on a polydimethylsiloxane (PDMS) layer, commonly usedin the stacking process of layered materials12,15. The adhesionstrength of this stamp is insufficient for detaching exfoliatedgraphene flakes produced by the conventional method. In ourlayer transfer method, however, this polymeric stamp cansuccessfully pick up large monolayer graphene flakes from theexfoliation substrate, as shown in Fig. 2c.We also produced a different stamp by attaching an hBN flakeonto the PPC/PDMS stamp12,15,16,18. Figure 2d shows the opticalimage of a large monolayer graphene flake, which was transferredfully from the substrate onto the stamp using our layer transfermethod. In this image, notice the graphene layer in regionsoutside the rectangular hBN flake, indicating that the layertransfer is independent of the adhesion to hBN. This outcome ofour layer transfer process contrasts with the tear-and-stackmethod using a PPC stamp12,15,16, where the shape of thetransferred flake conforms to that of the hBN flake. Fabrication ofa similar stack structure using the conventional method ispossible only by changing the polymer coating of the stamp (forexample, see ref. 14).Fabrication and material characterization of graphene het-erostructures. Next, we used monolayer graphene transferredonto the two different stamp types (i.e., PPC and hBN/PPC) forbuilding two groups of heterostructures: graphene-on-hBN (GB)and graphene encapsulated in hBN (BGB) stacks. To do so, we1500 2000 2500Intensity (a.u.)Raman shift (cm–1)AB0 5 10 15 20110100PAGEConv.CountMonolayer area (103 μm2)102102101101100103PAGEConv.2000X dimension (μm)Y dimension (μm)20,00010,0001000a bedfPVAGraphenecBAG2DPPC stampPPC stampGrapheneGraphenehBNFig. 2 PVA-assisted graphene exfoliation (PAGE). Our method allows the use of existing inspection techniques using a optical and b Raman microscopesfor identifying monolayer flakes. Scale bar in a is 50 μm. The Raman spectra in the b correspond to the regions in the optical image of the a marked with“A” and “B”. We showed the effectiveness of our layer transfer method by picking up monolayer graphene using c a PPC stamp, and d an hBN/PPC stamp.Scale bars are 20 μm. e PAGE reliably produced very large monolayer graphene flakes, which were not attainable by the conventional method in ourexperiments. The dashed lines are guides to the eye and represent the product of X and Y dimensions equal to 1000, 2000, 10,000, and 20,000 μm2. fThe histogram plot showing the distribution of the monolayer flake area for the samples obtained from 15 consecutive exfoliation trials. The data indicatethat PAGE is 20 times more likely than the conventional method (denoted as Conv.) to produce monolayer graphene with an area larger than 1000 μm2(notice the distribution beyond the yellow shading). The yellow shading denotes flakes with an area of <1000 μm2.NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-16817-1 ARTICLENATURE COMMUNICATIONS |         (2020) 11:3029 | https://doi.org/10.1038/s41467-020-16817-1 | www.nature.com/naturecommunications 3www.nature.com/naturecommunicationswww.nature.com/naturecommunicationsfirst assembled these stacks on the stamp itself by picking up thebottom hBN flake from SiO2. The full stacks were then laminatedat temperatures below 110 °C onto the support substrates. Weperformed the lamination following a similar procedure in ref. 15.Prior to our study, the construction of these two heterostructuresinvolved developing different strategies and processing steps (e.g.,see refs. 1,12,14,23,30). The versatility of our method overcomes thistechnical difficulty for fabricating diverse layered material stacks.Direct exfoliation of layered materials on clean SiO2 substratesmitigates the chance of material degradation due to contami-nants, enabling the fabrication of heterostructure devices withhigh carrier mobility12. In our method, the layered materialcomes into contact with PVA and water, raising a valid concernabout the degradation of graphene quality and its suitability formaking high-mobility devices. To understand the effect ofgraphene exposure to PVA and water, we studied the surfaceproperties of graphene using high-resolution atomic forcemicroscopy (AFM). Our AFM measurements indicate an increaseof both the mean phase and the average roughness of grapheneafter exposure to PVA and water (see Supplementary Note 11,Supplementary Figs. 11–20, and Supplementary Table 3). Whilenot possible to separate the effects of PVA and water, theseexperiments suggest an apparent negative impact on the graphenesurface properties. However, as we discuss next, this is not animpediment for producing graphene heterostructures with cleaninterfaces.Atomically clean interfaces are required for superior carriertransport in graphene heterostructures. Therefore, we examinedthe structural properties of our BGB stacks using AFM andRaman spectroscopy. In Fig. 3a, we show the optical image of aBGB stack after lamination. The data clearly shows that the as-fabricated BGB heterostructures made using our PVA-basedapproach suffer from microscopic blisters. While the exposure toPVA and water could amplify the likelihood of blister formation,we attribute this issue primarily to the sub optimal conditions ofour lamination process (see Supplementary Note 12 for details).Inspired by previous reports that the high-temperature processingcan mobilize blisters6,15,18, we performed annealing on fullyfabricated BGB stacks in ultra-high vacuum (UHV) at 400 °C (see“Methods”). Post-fabrication annealing at similar temperatures isroutinely performed for cleaning fully formed heterostructuresmade of flakes prepared by the conventional exfoliation (e.g., seerefs. 6,9,31–34). Interestingly, we observed that this annealing stepis highly effective in removing blisters. In Fig. 3b–d, we show theoptical and AFM images of the BGB stack of Fig. 3a after theUHV annealing step. The data illustrate that a large portion of theBGB stack is blister free, indicating the effectiveness of the UHVanneal in cleaning the graphene interfaces with the top andbottom hBN (t-hBN and b-hBN) flakes.Next, we used Raman spectroscopy to investigate further thematerial properties of graphene at three different stages of theBGB stack fabrication process. Figure 3e shows the typical Ramanspectrum of graphene after its exfoliation on PVA, after thelamination of the BGB stack, and after the UHV annealing step.The visible differences in the characteristics of the G and 2Dpeaks of these three spectra indicate detectable changes in thegraphene properties during the stack fabrication, which weelaborate below.The data in Fig. 3e show a noticeable blueshift of the 2D lineafter encapsulating graphene in hBN. This observation has beenattributed to the effect of dielectric screening by hBN on theelectronic structure of graphene35. We also observed a markedreduction in the full-width at half-maximum (FWHM) of the 2Dline (Γ2D) after the UHV annealing of the BGB stack. We attributethis observation to the removal of blisters. In particular, previousstudies reported that Γ2D is sensitive to structural deformation ofgraphene, for example, those caused by blisters in a BGB stack18,36.It has been observed that its value is generally >20 cm−1 inblistered regions of a BGB stack and is <20 cm−1 in blister-freeregions. Taking advantage of this knowledge, we observed that thespatially resolved measurement of Γ2D on the annealed stack,shown in Fig. 3i, provides precise information about the locationof blisters, consistent with the AFM images in Fig. 3c, d. The 2Dlines recorded in the blister-free region of the stack, marked with adashed box, have a Γ2D of 17 ± 0.45 cm−1. In contrast, the Γ2D inthe same region before annealing is above 22 cm−1, comparable tothose of exfoliated graphene on PVA (see Fig. 3g). The markeddecrease of Γ2D in this region of the sample after the UHV annealis an indication of the reduction in the nanometer-scale straininhomogeneities36.Analyzing the spatial Raman map in this region of the samplealso revealed the broadening of the G lines (i.e., ΓG) after theUHV annealing (Fig. 3h). This observation suggests the reducedcharge carrier doping of graphene after the UHV annealing36–38.Another important observation from this plot is the reducedvariations of ωG (i.e., ΔωG) after the UHV annealing. To examinethe origin of this observation, we plotted ω2D against ωG (Fig. 3f).The data in this plot provide critical information on howgraphene properties change at different stages of our BGBfabrication process. In particular, the data points of both thegraphene on PVA and the blister-free region of the annealed stackscatter tightly along a line with a slope of 2.2 (the solid anddashed black lines in the plot). This indicates that ΔωG at thesetwo stages is mostly due to the strain variations (Δε) and that thecontribution to ΔωG due to doping variations is negligible18,36,39.In contrast, the noticeable spread of the data points for the as-fabricated stack suggests significant variations of both doping andstrain within the sample before the UHV annealing.Another valuable information obtainable from this plot is theestimation of strain variations in graphene. In particular, localstrain fluctuations can critically limit carrier transport ingraphene devices40. Hence, we quantified the strain variationsin the blister-free region of the BGB stack, which we utilize fordevice fabrication. The estimated amplitude of strain usingRaman depends on the uniaxial or biaxial nature of the strain ingraphene41–43. However, the Raman spectrum of graphene at lowstrain gives no discernible signature for differentiating the natureof the strain41,43. In our calculations, though, we assumed auniaxial strain because it gives a worst-case estimate for thevariations of the strain amplitude in graphene (i.e., about threetimes larger than biaxial)41–43. In Fig. 3j, we show the spatiallyresolved distribution of the calculated strain amplitude in theblister-free region of the UHV-annealed BGB stack. The dataindicate the presence of tensile strain in graphene with an averageamplitude of ≈0.123%. This level of strain is low, and its effect onthe electronic band structure of graphene is negligible44. Moreimportantly, the small spread in ωG translates into Δε ≈ 0.067%.Moreover, the shift of data points towards lower values of ωGafter UHV annealing can be interpreted as a reduction in thecharge carrier doping of graphene37. This observation isconsistent with the broadening of the G lines after UHVannealing, discussed earlier in Fig. 3h.Our material characterization study using Raman spectroscopyand AFM imaging provides strong evidence that our technique iscapable of producing high-quality graphene heterostructures. Inparticular, our BGB stacks have desirable properties (e.g., low-residual doping and negligible strain variations), making themattractive for high-performance device applications.Transport studies of graphene heterostructure devices. Wefabricated gated-Hall bar devices from the GB and BGB stacks toARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-16817-14 NATURE COMMUNICATIONS |         (2020) 11:3029 | https://doi.org/10.1038/s41467-020-16817-1 | www.nature.com/naturecommunicationswww.nature.com/naturecommunicationsevaluate their electronic properties. To build these structures, weapplied our exfoliation and layer transfer methods only to gra-phene and used the conventional method for producing hBNflakes. This experimental design allows us to focus on the effect ofour process on graphene quality. Figure 4a, b show the schematicillustrations and example optical images of these gated-Hall bardevices. We used standard nanofabrication processes for buildingthese devices (see “Methods”).Electrical measurements on our GB and BGB devices indicatethe high material quality of monolayer graphene produced byPAGE. Figure 4c shows the typical four-point longitudinalresistivity (ρxx) of monolayer graphene in a GB device measuredas a function of the back-gate bias (Vg) (see “Methods”). Forevaluating the carrier mobility (μ), we assumed that a combina-tion of short- and long-range scattering mechanisms controls thecarrier transport in graphene45,46. As a result, the resistivity ofgraphene at charge carrier densities far away from the chargeneutrality point (CNP) follows ρxx= (enμL)−1+ ρs, where n is thegate-voltage induced carrier density, μL is the mobility due tolong-range Coulomb scattering, and ρs is a term due to short-range scattering. At sufficiently low carrier densities but far awayfrom CNP the contribution of ρs to the resistivity is generallynegligible, hence μ ≈ μL47. This approach gave μ ≈ 47,000 cm2 V−1 s−1 at 10 K for the GB device in Fig. 4c.While the estimated μ of the GB device is comparable with thepreviously reported counterparts23,48–50, it is significantly lower(by >20 times) than the state-of-the-art fully encapsulatedgraphene devices (for example, see refs. 12,18). We attribute thedegradation of the carrier mobility primarily to the graphenecontamination due to exposure to polymers and solvents. In thisdevice structure, graphene comes in contact with polymers andsolvents during both the GB stack construction and the devicefabrication. It is, hence, difficult to unambiguously identify theindependent contribution of the stack and device fabricationprocesses to the mobility degradation in this device structure.Unlike the GB structure, however, the top hBN layer in the BGB–0.15–0.10–0.051600 2600 2800Intensity (a.u.)On PVAAs-fabricatedAnnealedBGBBGB1LGG2D17.52625.51520251570 1580 1590266026802700�2D (cm–1)�G (cm–1)�G (cm–1)�2D (cm–1)�G (cm–1)�2D (cm–1)�2D  (cm–1)1LG on PVABGBannealedBGBas-fabricated2.22660 2680 27001020301580 159001020t-hBNb-hBNGraphene60 nm0 3 nm7 nma b c dfeg hijRaman shift (cm–1)� (%)Fig. 3 Evaluation of graphene material quality. Optical images of a BGB stack a after lamination (i.e., as-fabricated) and b after UHV annealing, indicatingthe effectiveness of the annealing step in removing blisters. High-resolution c AFM topography and d AFM topography error images confirm thecleanliness of the annealed BGB stack. Notice the remaining blisters in the top left corner of the AFM images. The black dashed box represents the region,which we used for comparing the Raman data before and after UHV annealing. e Typical Raman spectra of monolayer graphene (1LG) obtained at thedifferent stages of the BGB stack fabrication. The black dashed lines mark the position of G and 2D lines after UHV annealing. The numbers next to the 2Dlines give the FWHM. f The plot of ω2D against ωG. The dashed and solid black lines have a slope of 2.2. The scatter plots of (g) Γ2D against ω2D, and h ΓGagainst ωG. The symbols in these two plots correspond to the data in the f. i The spatial map of Γ2D of the annealed BGB stack, confirming the ability ofRaman to provide precise information about the location of blisters. The black dashed box in this plot is the same region as in the AFM image of the c. j Theestimated strain amplitude in the marked rectangular region. All scale bars are 5 μm.NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-16817-1 ARTICLENATURE COMMUNICATIONS |         (2020) 11:3029 | https://doi.org/10.1038/s41467-020-16817-1 | www.nature.com/naturecommunications 5www.nature.com/naturecommunicationswww.nature.com/naturecommunicationsstructure protects graphene from exposure to contaminantsduring the device fabrication process. Hence, the electricalcharacteristics of the BGB device provide a direct measure ofthe graphene cleanliness prepared using our PVA-basedtechnique.Figure 4d shows the ρxx of a BGB device at room temperatureand 9 K (the base temperature of our measurement system).The sheet resistivity of this device (BGB-1) at room tempera-ture is <40Ω at n ≈ 2 × 1012 cm−2 (Supplementary Note 3 andSupplementary Fig. 3), evidence of high graphene quality. Togain better insight, we then calculated the carrier-dependentmobility of the BGB-1 device at room temperature (Fig. 4e)using the Drude model of conductivity, σ= enμ. The carrierdensity was calculated by considering both the quantumcapacitance and the oxide capacitance. For calculations of theoxide capacitance, we used a dielectric constant of 3 for hBN(extracted from the magneto-transport measurements, seeSupplementary Note 9 and Supplementary Fig. 9) andmeasured the thickness of hBN using AFM. The data indicateroom-temperature mobility over 200,000 cm2 V−1 s−1 at lowcarrier densities. At larger carrier densities (n ≈ 3 × 1012 cm−2)the mobility is ≈62,000 cm2V−1 s−1, which is comparable to thetheoretical acoustic phonon-limited mobility calculated assum-ing a deformation-potential coupling constant (D) of 14 eV(Supplementary Note 4)51. The theoretical carrier-density-dependent mobility curve in this plot corresponds to atheoretically predicted ρe-ph of ≈32Ω52. We note that theroom-temperature carrier mobility of the BGB-1 deviceimproved after a thermal cycle in the measurement system(see Supplementary Note 7, Supplementary Fig. 7, andSupplementary Table 2). Thermal cycling has been previouslyobserved to improve the mobility of BGB devices by modulatingthe charge inhomogeneity33.We also measured μ of the BGB-1 device at 9 K. Using theDrude model of conductivity, we calculated a carrier mobility of≈800,000 cm2V−1 s−1 for this device, as shown in Fig. 4f. Thetransport characteristics of our BGB device are well within therange of the typical carrier mobility for the state-of-the-art fullyencapsulated graphene devices prepared using the polymer-freelayer assembly method (for example, see refs. 12,18). The electricalcharacteristics of the BGB devices confirm the ability of our PVA-based technique to produce high-quality graphene devices.We studied the charge carrier inhomogeneity (n*) as a measureof graphene quality. Interactions of graphene with its environ-ment (e.g., variations of chemical doping, charged impurities) cancause spatial charge inhomogeneity close to the Dirac point dueto electron-hole (e-h) puddle formation53,54. The charge carrierinhomogeneity can be estimated from the gate-induced carrierdensity below which the conductivity σ= (ρxx)−1 remainsunchanged with gating. From the data in the insets in Fig. 4c,d, we found n* to be ≈2 × 1010 and <5 × 109 cm−2 for the GB-1and BGB-1 devices. These values represent an upper-boundestimate of the charge inhomogeneity in our graphenesamples40,55. Further, the monolayer graphene in these devicesis nearly intrinsic (i.e., low doping), evident from the peakresistivity position at nearly zero gate-voltage (see ρxx plots inFig. 4). Measurements on additional GB and BGB devices resultedin similar values for μ and n* (see Supplementary Note 6,–2 0 2030048120 1 2 30123300 KPhonon limit modelBGB-1–40 –20 0 20 400123Vg (V)Vg (V)300 K9 K109 1010 1011 1012n∗–4 0 4024250 K10 K� xx (kΩ)� xx (S)� xx (kΩ)� (mS)1010 1011 101210 3–10 4–10 1–10 2–10 3–10 4–10–2n∗bab-hBNVGateVgSGrapheneSiO2IsourceDVg Isourceb-hBNSiO2t-hBNVGrapheneGB-1 BGB-1c defSDGATEb-hBN b-hBNSDBGB-19 K� (105 cm2 V–1s–1)n (1012 cm–2)n (1011 cm–2)n (cm–2) � (105 cm2 V–1s–1)� xx (S)n (cm–2) Fig. 4 Transport characteristics of graphene heterostructures. We fabricated gated-Hall bar devices from two types of graphene heterostructures: agraphene-on-hBN (GB) and b graphene encapsulated in hBN (BGB). The dashed green box in the optical images shows the device active region. The scalebars are 10 μm. The narrow FWHM of the four-point longitudinal resistivity (ρxx) of the c GB and d BGB devices suggests a low-charge inhomogeneity (n*)in graphene. The insets in c and d show the estimated n* of the GB and BGB devices. e The carrier-density-dependent mobility of the BGB-1 device at roomtemperature, giving mobility in excess of 200,000 cm2 V−1 s−1 at low carrier densities. For comparison, we provided the theoretically predicted acousticphonon-limited mobility45. f The corresponding low-temperature mobility of the BGB-1 device is 800,000 cm2 V−1 s−1 at low carrier densities.ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-16817-16 NATURE COMMUNICATIONS |         (2020) 11:3029 | https://doi.org/10.1038/s41467-020-16817-1 | www.nature.com/naturecommunicationswww.nature.com/naturecommunicationsSupplementary Figs. 5 and 6, and Supplementary Table 1). Theseobservations further confirm that the exfoliated grapheneprepared by PAGE is of high quality.Next, we performed magneto-transport measurements on theBGB-1 device at the base temperature of 9 K. The rationalebehind these measurements was to glean further informationabout the carrier transport properties of the BGB device. Inparticular, the transport scattering time, which determines carriermobility, is insensitive to both small- and large-angle scatter-ings56. For this reason, the measurement of mobility alone isinsufficient to provide a complete picture of carrier scattering ingraphene. In contrast, the elastic scattering time in graphene,which determines the quantum (i.e., Landau) level broadening, issensitive to small-angle scattering events. Therefore, the outcomeof these measurements provides additional information about thegraphene quality, which is inaccessible to transport mobilitymeasurements.In Fig. 5a, we show the Landau fan diagram of the longitudinalconductivity (σxx) for the BGB-1 device. To plot the intensity mapof the longitudinal conductivity, we calculated σxx from ρxx/(ρxx2+ ρxy2), where ρxy is the transverse resistivity. The Landau fandiagram shows well-resolved quantum Hall states (QHS),indicating the high quality of the two-dimensional electronicsystem in monolayer graphene.In a two-dimensional electronic system, the quantizationbegins at a sufficiently strong magnetic field, which yieldscyclotron energy larger than the Landau level broadening57.Quantifying the elastic scattering time in samples with chargeinhomogeneity is highly involved (e.g., see refs. 58,59). However,the onset of Shubnikov de Haas can be used for a qualitativeassessment of this scattering time by providing an estimate forDingle mobility at low carrier densities47. To do so, we examinedthe transition of Hall conductivity (σxy) to |ν |= 2 phase, asshown in Fig. 5b, because this filling factor appears first in theQHS sequence in graphene60. The data indicate that thetransition commences in fields around 40 mT, corresponding toan estimated Dingle mobility of at least 250,000 cm2V−1 s−1 at 9K. We also observed that the degeneracy lifting of the Landaulevels due to broken symmetry in our sample commences at low-magnetic fields. Figure 5c illustrates the start of σxy transition to ν= 0 phase at CNP occurring at below 1 T. Collectively, thesedesirable characteristics (common among BGB devices madeusing our technique, see Supplementary Note 10 and Supple-mentary Fig. 10) are evidence of the graphene cleanliness in theBGB device structure.DiscussionOur study establishes an efficient strategy for the high-yieldconstruction of diverse vdW heterostructures. A simple interfaceengineering scheme for modifying the exfoliation substrate with asub-5 nm PVA is the key to this advance. This scheme enabled usto optimize the exfoliation step independently of the layertransfer step and vice versa. This ability resulted in two majoroutcomes. First is a consistent exfoliation method for producinglarge monolayer flakes. The second outcome is a high-yield layertransfer of exfoliated flakes without requiring a specializedtransfer stamp. By using graphene as a model material, we furtherestablished the remarkable material and electronic properties ofthe resulting heterostructures.We attribute the remarkable improvements of the PAGEtechnique over the conventional exfoliation method to a strongeradhesion of graphene to PVA than to SiO2. This hypothesis isconfirmed by our observation that, when performing layertransfer at the glass transition temperature of PVA, a PVA-coatedPDMS stamp can reliably peel off monolayer graphene that hasbeen exfoliated directly on SiO2 (see Supplementary Note 8, andSupplementary Fig. 8).Our extensive material and electrical characterizations con-firmed the ability of our technique to produce heterostructureswith clean interfaces. However, our layer assembly experimentsrevealed two crucial insights. First, we found that a laminationtemperature of <110 °C is inadequate for achieving blister-freeheterostructures prepared using PAGE. One possible explanationis that the exposure to PVA and water negatively impacts thesurface properties of graphene. While we do not rule out thispossibility, we point out that heterostructures prepared inambient air using a polymer-free layer assembly could also sufferfrom a similar issue (e.g., see ref. 18 and Supplementary Fig. 21).These observations highlight the critical role of the laminationprocess on the interface quality of the resulting heterostructures.The second insight is that subsequent annealing of the as-fabricated stacks in a UHV environment was highly effective inremoving blisters. Indeed, our AFM and Raman measurementsconfirmed the cleanliness of the annealed heterostructures.However, one common observation about the high-temperatureannealing is the accumulation of blisters6. Based on these twoexperimental insights, an important future direction for buildingheterostructures in ambient air is to explore the effectiveness oflamination at elevated temperatures in removing blisters—forexample, using the process of Purdie et al. in ref. 18. The similaritybetween the glass transition temperatures of PMMA and PVA2820 mT30 mT40 mT60 mT200 mT250 mT100 mT700 mT800 mT1000 mT1500 mT2000 mT642040–2 2 610.001.000.100.011–6 –3 –1 0 –2 0 20 3 6Vg – VDirac (V) Vg – VDirac (V) Vg – VDirac (V)⎟ �xy⎟ (e2  h–1)�xx (e2 h–1)B (T)⎟ �xy⎟ (e2  h–1)  = –6  = –2  = 0a b cFig. 5 Magneto-transport measurements of graphene. a The fan diagram of the longitudinal conductivity (σxx) shows well-resolved quantum Hall states atthe base temperature of 9 K. The measurements were made on the BGB-1 device of Fig. 4. b The transition of Hall conductivity (σxy) to ν=−2 commencesat ≈40mT. c The lifting of the zero-energy Landau level degeneracy begins at below 1 T. These observations indicate the high quality of the two-dimensional electronic system in our graphene heterostructure.NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-16817-1 ARTICLENATURE COMMUNICATIONS |         (2020) 11:3029 | https://doi.org/10.1038/s41467-020-16817-1 | www.nature.com/naturecommunications 7www.nature.com/naturecommunicationswww.nature.com/naturecommunicationssuggests that the removal of blisters by employing the laminationprocess of Purdie et al.18 should be feasible.Past research has shown that performing the layer assembly ina moisture-free environment (e.g., in a glovebox) is anothereffective solution for producing heterostructures with cleaninterfaces61,62. However, the use of water for releasing exfoliatedflakes in our current process limits its application for fabricatingheterostructures in a glovebox environment. This pitfall could beresolved by replacing water with an anhydrous solvent for dis-solving the PVA sacrificial layer (e.g., dimethyl sulfoxide63).Developing a glovebox-compatible process based on our techni-que constitutes an important future direction.Our methodology can be generalized for constructing hetero-structures from other layered materials besides graphene. Wedemonstrated the application of our exfoliation and layer transfermethods to MoS2, a material system that belongs to the largefamily of transition metal dichalcogenides. Applying our exfo-liation technique resulted in large (<1000 μm2) monolayer MoS2(Fig. 6a). This is an improvement over the conventional method,which is known to suffer from poor yield and small flakes (<300μm2)21. We used Raman spectroscopy for identifying the numberof layers within the candidate MoS2 flakes chosen through opticalinspection. Figure 6b shows a typical Raman spectrum of theMoS2 flake in Fig. 6a (taken at point “M”) against the spectrum ofthe PVA background (taken at point “P”), indicating a monolayerflake. The Raman data indicate the broadening of the A1g peak,the reduction of its intensity relative to the E2g peak, and thesmaller-than-expected spacing between these peaks (i.e., a theo-retical separation of 20 cm−1 for unstrained monolayer MoS264).These signatures of the Raman data suggest the presence ofsubstrate-induced strain and doping in the as-exfoliated flakeresiding on PVA64–68. Decoupling the quantitative effects ofdoping and strain requires additional systematic studies. Thespatial Raman map data (taken from the region marked with thesolid black box in Fig. 6a) indicate the uniformity of the structuralproperties across the flake (see Fig. 6c–f).We also demonstrated the utility of our layer transfer methodfor producing a stack of hBN/MoS2/hBN. In the resulting stack(Fig. 6g), the monolayer MoS2 is partially encapsulated in hBN; alarge region of MoS2 remains exposed. This configuration givesadditional degrees of freedom for building advanced devicestructures. For example, one can stack superconducting materialson the exposed regions of the semiconducting layered material tobuild Josephson junction transistors. Lastly, the strong intensityand the narrow FWHM of the photoluminescence (PL) spectrumof the stack in Fig. 6h indicates the high material quality of themonolayer MoS2.Our construction method simplifies the fabrication of vdWheterostructures. Compared to the conventional method, ourexfoliation method requires a significantly smaller number ofattempts for producing the same number of exfoliated flakes.Further, our method enables the long-term storage of the4 3I(A1g)/I(E2g)1 0.8516 20380 400 420b1.6 1.8 2.00200400600 ABIntensity (cts)Energy (eV)E2g A1gMoS2MoS2t-hBNb-hBNSiO2×1.4a c dcts0103ABPVAMPMPe f g h79Intensity (a.u.)Raman shift (cm–1)Δ� = 17 cm–1Δ� (cm–1)�(E2g) (cm–1) �(A1g) (cm–1)Fig. 6 Fabrication of heterostructures from other layered materials. a Applying PVA- assisted exfoliation technique to MoS2 consistently yields largemonolayer flakes. Scale bar is 20 μm. b Corresponding Raman spectra of “M” and “P” points in a. PVA does not have an overlapping Raman signature withMoS2, allowing the use of Raman spectroscopy for identifying the number of layers within the flake. c–f The spatial Raman maps of the flake in the aindicate the uniformity of its structural properties. The map was taken from the region inside the flake marked with the solid black box. In these plots, Δω isthe spacing between E2g and A1g peaks, I(A1g)/I(E2g) is the ratio of the peak intensity, Γ(E2g) and Γ(A1g) are the FWHM of the peaks. Scale bars are 5 μm. gThe optical image of a monolayer MoS2 encapsulated in hBN. The inset shows the spatial map of the PL peak intensity. Scale bars are 20 μm. h Therepresentative PL spectra of the monolayer MoS2 at the encapsulated region (point “A”) and the exposed region (point “B”). The red-shift of the PLspectrum suggests the presence of a small tensile strain within the encapsulated region. The FWHM of the “A” and “B” spectra are ≈46 and 54meV,indicating the high material quality of the exfoliated MoS2.ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-16817-18 NATURE COMMUNICATIONS |         (2020) 11:3029 | https://doi.org/10.1038/s41467-020-16817-1 | www.nature.com/naturecommunicationswww.nature.com/naturecommunicationsexfoliated flakes for future use. The reason is that our layertransfer process relies on the removal of PVA for releasing thelayered material. We fabricated a BGB device from a 3-week-oldPAGE sample that showed a mobility on-par with fresh samples(see Supplementary Note 5). This contrasts with our experiencewith the layer transfer of exfoliated flakes on SiO2, where wecould not mechanically detach monolayer flakes from SiO2 after afew days of storage. Lastly, given their consistent results andversatility to different layered materials, our exfoliation and thelayer transfer methods can be standardized, hence opening thedoor for building automated machines to perform those steps.The collective effects of the above features of our constructionmethod enable a high-yield production of vdW heterostructures.MethodsPreparation of PVA-coated substrates. In all experiments in this study, we useda dilute PVA solution with a concentration of 3% wt vol−1. We prepared thesolution by dissolving the PVA powder (MW= 9000, Sigma) in deionized water.Prior to the exfoliation experiments, SiO2/Si substrates (285 nm oxide) werecleaned with a Piranha solution, followed by the spin coating of the PVA solutionat a spin speed of 8000 rpm for 30 s. The PVA film was not baked after the spincoating and prior to the exfoliation step. We measured the thickness of the PVAfilm by fitting the ellipsometry data using the Cauchy model. The ellipsometry (J.A. Woollam) was done in the visible wavelength range and at three different angles.The Cauchy model consistently provided a refractive index of 1.46 ± 0.02, which isin good agreement with the known refractive index of PVA. The typical measuredPVA film thickness using this technique is about 3.2 nm.Graphene exfoliation. We prepared the bulk crystals of graphite on a scotch tape(3M, Cat. 105). After placing the tape on the PVA-coated substrate at roomtemperature, we performed a brief heat treatment at 85 °C for 10 s on a hotplate.We observed no apparent improvements in size or yield of the exfoliated flakeswith longer heat treatments. Before detaching the tape, the sample was removedfrom the hotplate. The tape was then peeled back slowly (at a speed of 2–3 mm s−1), starting from one end of the substrate. While we had no tight control over theexfoliation angle, we targeted for an angle below 30°.Preparation of the polymeric stamp and layer transfer. We followed the pro-cedure in ref. 16 to produce the PDMS stamps. These stamps were then modified byapplying a PPC coating, prepared following the procedure in ref. 15. To locallydissolve PVA during the layer transfer, a water drop was injected using a manualsyringe assembly, as shown in Supplementary Note 2, and Supplementary Fig. 2.The substrate temperature was 40 °C during this step. After transferring the flakeonto the stamp, the stamp was immersed in a deionized water bath at 40 °C forabout 1 h, while changing the DI water every 20 minutes. The stamp was then driedusing gentle N2 flow, then left in a desiccator for later use.Ultra-high-vacuum (UHV) annealing step. The UHV annealing for all stacks wasdone at 400 °C for 2 h. The base pressure of the system was about 1 × 10−10mbar.The ramp-up rate of the temperature was 5 °Cmin−1 up to 150 °C and was 10 °Cmin−1 after that.Fabrication of gated-Hall bar devices. Two-dimensional metal contacts wereformed on graphene for both GB and BGB structures using a combination ofelectron-beam lithography (EBL), e-beam metal evaporation (Cr 5 nm/Au 50 nm),and metal lift-off. The device structure was complete after defining the activeregion using a combination of EBL and reactive ion etching.Raman spectroscopy and strain calculation. Raman measurements were madeusing the Horiba Xplora micro-Raman system with a 532 nm laser. The G and 2Dlines were fitted with a single Lorentzian function. The uniaxial strain-sensitivity ofthe G mode (i.e., Δω/Δε= 23.5 cm−1/%) was used for calculating the residualstrain39.Electronic transport measurements. We used low-current, low-frequency lock-intechniques for measuring the longitudinal resistance (Rxx) and transverse Hallresistance (Rxy) of the graphene Hall bar devices. Low-temperature measurementswere made in a cryogen-free micro-manipulated Lakeshore probe station, CRX-VF.Low-temperature measurements below 9 K were made in a cyro-free super-conducting magnet system, TeslatronPT (see Supplementary Fig. 10).Data availabilityThe authors declare that the data supporting the findings of this manuscript are availablewithin the article and its Supplementary Information files. Extra data are available fromthe corresponding author upon reasonable request.Received: 1 October 2019; Accepted: 26 May 2020;References1. Amet, F. et al. Composite fermions and broken symmetries in graphene. Nat.Commun. 6, 5838 (2015).2. Yankowitz, M. et al. Emergence of superlattice Dirac points in graphene onhexagonal boron nitride. Nat. Phys. 8, 382 (2012).3. Dean, C. R. et al. Hofstadter’s butterfly and the fractal quantum Hall effect inmoiré superlattices. Nature 497, 598 (2013).4. Cao, Y. et al. Unconventional superconductivity in magic-angle graphenesuperlattices. Nature 556, 43 (2018).5. Cao, Y. et al. Correlated insulator behaviour at half-filling in magic-anglegraphene superlattices. Nature 556, 80 (2018).6. Larentis, S. et al. Band offset and negative compressibility in graphene-MoS2heterostructures. Nano Lett. 14, 2039–2045 (2014).7. Movva, H. C. et al. Tunable Γ- K valley populations in hole-doped trilayerWSe2. Phys. Rev. Lett. 120, 107703 (2018).8. Katoch, J. et al. Giant spin-splitting and gap renormalization driven bytrions in single- layer WS2/h-BN heterostructures. Nat. Phys. 14, 355–359(2018).9. Bediako, D. K. et al. Heterointerface effects in the electrointercalation of vander Waals heterostructures. Nature 558, 425–429 (2018).10. Velický, M. et al. Electron tunneling through boron nitride confirms Marcus-Hush theory predictions for ultramicroelectrodes. ACS Nano 14, 993–1002(2019).11. Shi, G. et al. Boron nitride-graphene nanocapacitor and the origins ofanomalous size-dependent increase of capacitance. Nano Lett. 14, 1739–1744(2014).12. Wang, L. et al. One-dimensional electrical contact to a two-dimensionalmaterial. Science 342, 614–617 (2013).13. Castellanos-Gomez, A. et al. Deterministic transfer of two-dimensionalmaterials by all- dry viscoelastic stamping. 2D Mater. 1, 011002 (2014).14. Zomer, P., Guimarães, M., Brant, J., Tombros, N. & Van Wees, B. Fast pick uptechnique for high quality heterostructures of bilayer graphene and hexagonalboron nitride. Appl. Phys. Lett. 105, 013101 (2014).15. Pizzocchero, F. et al. The hot pick-up technique for batch assembly of van derWaals heterostructures. Nat. Commun. 7, 11894 (2016).16. Kim, K. et al. van der Waals heterostructures with high accuracy rotationalalignment. Nano Lett. 16, 1989–1995 (2016).17. Kang, K. et al. Layer-by-layer assembly of two-dimensional materials intowafer-scale heterostructures. Nature 550, 229–233 (2017).18. Purdie, D. et al. Cleaning interfaces in layered materials heterostructures. Nat.Commun. 9, 5387 (2018).19. Novoselov, K. S. et al. Electric field effect in atomically thin carbon films.Science 306, 666–669 (2004).20. Huang, Y. et al. Reliable exfoliation of large-area high-quality flakes ofgraphene and other two-dimensional materials. ACS Nano 9, 10612–10620(2015).21. Desai, S. B. et al. Gold-mediated exfoliation of ultralarge optoelectronically-perfect monolayers. Adv. Mater. 28, 4053–4058 (2016).22. Cao, Y. et al. Superlattice-induced insulating states and valley-protected orbitsin twisted bilayer graphene. Phys. Rev. Lett. 117, 116804 (2016).23. Dean, C. R. et al. Boron nitride substrates for high-quality grapheneelectronics. Nat. Nanotechnol. 5, 722 (2010).24. Lee, G.-H. et al. Flexible and transparent MoS2 field-effect transistors onhexagonal boron nitride-graphene heterostructures. ACS Nano 7, 7931–7936(2013).25. Cui, X. et al. Multi-terminal transport measurements of MoS2 using a van derWaals heterostructure device platform. Nat. Nanotechnol. 10, 534 (2015).26. Ferrari, A. C. et al. Raman spectrum of graphene and graphene layers. Phys.Rev. Lett. 97, 187401 (2006).27. Ferrari, A. C. & Basko, D. M. Raman spectroscopy as a versatile tool forstudying the properties of graphene. Nat. Nanotechnol. 8, 235 (2013).28. Ni, Z. H. et al. Uniaxial strain on graphene: Raman spectroscopy study andband-gap opening. ACS Nano 2, 2301–2305 (2008).29. Malard, L., Pimenta, M., Dresselhaus, G. & Dresselhaus, M. Ramanspectroscopy in graphene. Phys. Rep. 473, 51–87 (2009).30. Banszerus, L. et al. Ultrahigh-mobility graphene devices from chemical vapordeposition on reusable copper. Sci. Adv. 1, e1500222 (2015).NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-16817-1 ARTICLENATURE COMMUNICATIONS |         (2020) 11:3029 | https://doi.org/10.1038/s41467-020-16817-1 | www.nature.com/naturecommunications 9www.nature.com/naturecommunicationswww.nature.com/naturecommunications31. Wang, L. et al. Evidence for a fractional fractal quantum Hall effect ingraphene superlattices. Science 350, 1231–1234 (2015).32. Yankowitz, M. et al. Intrinsic disorder in graphene on transition metaldichalcogenide heterostructures. Nano Lett. 15, 1925–1929 (2015).33. Komatsu, K. et al. Observation of the quantum valley Hall state in ballisticgraphene superlattices. Sci. Adv. 4, eaaq0194 (2018).34. Hung, T. Y., Camsari, K. Y., Zhang, S., Upadhyaya, P. & Chen, Z. Directobservation of valley-coupled topological current in MoS2. Sci. Adv. 5,eaau6478 (2019).35. Forster, F. et al. Dielectric screening of the Kohn anomaly of graphene onhexagonal boron nitride. Phys. Rev. B 88, 085419 (2013).36. Neumann, C. et al. Raman spectroscopy as probe of nanometre-scale strainvariations in graphene. Nat. Commun. 6, 1–7 (2015).37. Yan, J., Zhang, Y., Kim, P. & Pinczuk, A. Electric field effect tuning ofelectron-phonon coupling in graphene. Phys. Rev. Lett. 98, 166802 (2007).38. Casiraghi, C., Pisana, S., Novoselov, K., Geim, A. K. & Ferrari, A. Ramanfingerprint of charged impurities in graphene. Appl. Phys. Lett. 91, 233108(2007).39. Lee, J. E., Ahn, G., Shim, J., Lee, Y. S. & Ryu, S. Optical separation ofmechanical strain from charge doping in graphene. Nat. Commun. 3, 1–8(2012).40. Couto, N. J. et al. Random strain fluctuations as dominant disorder source forhigh-quality on-substrate graphene devices. Phys. Rev. X 4, 041019 (2014).41. Mohiuddin, T. et al. Uniaxial strain in graphene by Raman spectroscopy: Gpeak splitting, Grüneisen parameters, and sample orientation. Phys. Rev. B 79,205433 (2009).42. Zabel, J. et al. Raman spectroscopy of graphene and bilayer under biaxialstrain: bubbles and balloons. Nano Lett. 12, 617–621 (2012).43. Yoon, D., Son, Y.-W. & Cheong, H. Strain-dependent splitting of the double-resonance Raman scattering band in graphene. Phys. Rev. Lett. 106, 155502(2011).44. Guinea, F., Katsnelson, M. & Geim, A. Energy gaps and a zero-field quantumHall effect in graphene by strain engineering. Nat. Phys. 6, 30–33 (2010).45. Hwang, E., Adam, S. & Sarma, S. D. Carrier transport in two-dimensionalgraphene layers. Phys. Rev. Lett. 98, 186806 (2007).46. Morozov, S. et al. Giant intrinsic carrier mobilities in graphene and its bilayer.Phys. Rev. Lett. 100, 016602 (2008).47. Mayorov, A. S. et al. How close can one approach the Dirac point in grapheneexperimentally? Nano Lett. 12, 4629–4634 (2012).48. Du, X., Skachko, I., Barker, A. & Andrei, E. Y. Approaching ballistic transportin suspended graphene. Nat. Nanotechnol. 3, 491 (2008).49. Mayorov, A. S. et al. Micrometer-scale ballistic transport in encapsulatedgraphene at room temperature. Nano Lett. 11, 2396–2399 (2011).50. Bolotin, K. I. et al. Ultrahigh electron mobility in suspended graphene. SolidState Commun. 146, 351–355 (2008).51. Hwang, E. & Sarma, S. D. Acoustic phonon scattering limited carrier mobilityin two- dimensional extrinsic graphene. Phys. Rev. B 77, 115449 (2008).52. Park, C.-H. et al. Electron-phonon interactions and the intrinsic electricalresistivity of graphene. Nano Lett. 14, 1113–1119 (2014).53. Martin, J. et al. Observation of electron-hole puddles in graphene using ascanning single-electron transistor. Nat. Phys. 4, 144 (2008).54. Xue, J. et al. Scanning tunneling microscopy and spectroscopy of ultra-flatgraphene on hexagonal boron nitride. Nat. Mater. 10, 282 (2011).55. Tan, Y.-W. et al. Measurement of scattering rate and minimum conductivityin graphene. Phys. Rev. Lett. 99, 246803 (2007).56. Hwang, E. & Sarma, S. D. Single-particle relaxation time versus transportscattering time in a two-dimensional graphene layer. Phys. Rev. B 77, 195412(2008).57. Monteverde, M. et al. Transport and elastic scattering times as probes of thenature of impurity scattering in single-layer and bilayer graphene. Phys. Rev.Lett. 104, 126801 (2010).58. Syed, S., Manfra, M., Wang, Y., Molnar, R. & Stormer, H. Electron scatteringin AlGaN/GaN structures. Appl. Phys. Lett. 84, 1507–1509 (2004).59. Hong, X., Zou, K. & Zhu, J. Quantum scattering time and its implications onscattering sources in graphene. Phys. Rev. B 80, 241415 (2009).60. Zhang, Y. et al. Landau-level splitting in graphene in high magnetic fields.Phys. Rev. Lett. 96, 136806 (2006).61. Masubuchi, S. et al. Autonomous robotic searching and assembly of two-dimensional crystals to build van der Waals superlattices. Nat. Commun. 9,1–12 (2018).62. Telford, E. J. et al. Via method for lithography free contact and preservation of2D materials. Nano Lett. 18, 1416–1420 (2018).63. Tacx, J., Schoffeleers, H., Brands, A. & Teuwen, L. Dissolution behavior andsolution properties of polyvinylalcohol as determined by viscometry and lightscattering in DMSO, ethyleneglycol and water. Polymer 41, 947–957 (2000).64. Doratotaj, D., Simpson, J. R. & Yan, J.-A. Probing the uniaxial strains in MoS2using polarized Raman spectroscopy: a first-principles study. Phys. Rev. B 93,075401 (2016).65. Li, H. et al. From bulk to monolayer MoS2: evolution of Raman scattering.Adv. Funct. Mater. 22, 1385–1390 (2012).66. Rice, C. et al. Raman-scattering measurements and first-principles calculationsof strain-induced phonon shifts in monolayer MoS2. Phys. Rev. B 87, 081307(2013).67. Chakraborty, B. et al. Symmetry-dependent phonon renormalization inmonolayer MoS2 transistor. Phys. Rev. B 85, 161403 (2012).68. Alharbi, A. & Shahrjerdi, D. Analyzing the effect of high-k dielectric-mediateddoping on contact resistance in top-gated monolayer MoS2 transistors. IEEETrans. Electron Devices 65, 4084–4092 (2018).AcknowledgementsWe acknowledge partial financial support from NSF (grants EECS-1638598, CMMI-1728051, DMR-1420073) and the instrumentation grants from NSF (MRI-1531664) andGordon and Betty Moore Foundation (GBMF 4838). K.W. and T.T. acknowledge sup-port from the Elemental Strategy Initiative conducted by the MEXT, Japan and theCREST (JPMJCR15F3), JST. This work was performed in part at the ASRC NanoFab-rication Facility of CUNY in New York. D.S. acknowledges Prof. J. Uichanco of theUniversity of Michigan Ann Arbor for helpful discussions.Author contributionsZ.H., A.A., and D.S. conceived and designed the experiments. Z.H., A.A., E.C., and D.S.performed the experiments. W.M. and J.S. contributed to the magneto-transport mea-surements. K.W. and T.T. prepared the hBN material. Z.H., A.A., W.M., E.C., J.S., and D.S. contributed to the data analysis. The manuscript was written with input from allauthors.Competing interestsThe authors declare no competing interests.Additional informationSupplementary information is available for this paper at https://doi.org/10.1038/s41467-020-16817-1.Correspondence and requests for materials should be addressed to D.S.Peer review information Nature Communications thanks Achint Jain and the other,anonymous, reviewer(s) for their contribution to the peer review of this work.Reprints and permission information is available at http://www.nature.com/reprintsPublisher’s note Springer Nature remains neutral with regard to jurisdictional claims inpublished maps and institutional affiliations.Open Access This article is licensed under a Creative CommonsAttribution 4.0 International License, which permits use, sharing,adaptation, distribution and reproduction in any medium or format, as long as you giveappropriate credit to the original author(s) and the source, provide a link to the CreativeCommons license, and indicate if changes were made. The images or other third partymaterial in this article are included in the article’s Creative Commons license, unlessindicated otherwise in a credit line to the material. If material is not included in thearticle’s Creative Commons license and your intended use is not permitted by statutoryregulation or exceeds the permitted use, you will need to obtain permission directly fromthe copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.© The Author(s) 2020ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-16817-110 NATURE COMMUNICATIONS |         (2020) 11:3029 | https://doi.org/10.1038/s41467-020-16817-1 | www.nature.com/naturecommunicationshttps://doi.org/10.1038/s41467-020-16817-1https://doi.org/10.1038/s41467-020-16817-1http://www.nature.com/reprintshttp://creativecommons.org/licenses/by/4.0/http://creativecommons.org/licenses/by/4.0/www.nature.com/naturecommunications Versatile construction of van der Waals heterostructures using a dual-function polymeric film Results Consistent exfoliation and layer transfer via interface engineering Fabrication and material characterization of graphene heterostructures Transport studies of graphene heterostructure devices Discussion Methods Preparation of PVA-coated substrates Graphene exfoliation Preparation of the polymeric stamp and layer transfer Ultra-high-vacuum (UHV) annealing step Fabrication of gated-Hall bar devices Raman spectroscopy and strain calculation Electronic transport measurements Data availability References Acknowledgements Author contributions Competing interests Additional information