Dimensional Hexagonal Boron Nitride Nanosh
Supporting Information
Solution-Processed Dielectrics Based on Thickness-Sorted TwoDimensional Hexagonal Boron Nitride Nanosheets †
†
†
†
†
†
Jian Zhu , Joohoon Kang , Junmo Kang , Deep Jariwala , Joshua D. Wood , Jung-Woo T. Seo , Kan† †‡§ †‡§∥ Sheng Chen , Tobin J. Marks , and Mark C. Hersam* †Department of Materials Science and Engineering, ‡Graduate Program in Applied Physics, § Department of Chemistry, and ∥Department of Medicine, Northwestern University, Evanston, Illinois 60208, United States *Address correspondence to: [email protected]
Contents:
Section S1. Methods o Section S1.1. Dispersion of h-BN Flakes o Section S1.2. Size and thickness sorting of h-BN o Section S1.3. Van der Waals volume of F68 chain o Section S1.4. Measurement of h-BN buoyant density in each fraction o Section S1.5. Layer-by-layer (LBL) assembly of h-BN multilayers o Section S1.6. LBL assembly of h-BN layer on top of SiO2 and HfOx o Section S1.7. Fabrication of graphene field-effect transistors (GFETs) o Section S1.8. Areal capacitance measurement o Section S1.9. Characterization of GFETs o Section S1.10. Other characterization tools Section S2. Supporting Figures o Figure S1. Dispersion protocol for h-BN with F68 surfactants o Figure S2. Zeta potential of F68 and F68-stabilized h-BN o Figure S3. Using the h-BN solution supernatant for DGU separation o Figure S4. AFM images during DGU processing steps o Figure S5. h-BN thickness and size distributions o Figure S6. Raman spectra of bulk h-BN and sorted h-BN in the different fractions. o Figure S7. Thickness sorting of h-BN by DGU with sodium cholate (SC) as surfactants o Figure S8. Thermogravimetric analysis of polyethyleneimine (PEI) o Figure S9. Raman metrics for graphene on SiO2, HfOx, and h-BN o Figure S10. AFM images of h-BN on different dielectrics o Figure S11. Assembled h-BN film as an interfacial layer for other dielectrics used in GFETs o Figure S12. Model fitting of the transfer curve for graphene FETs with various
dielectrics S1
Section S1. Methods: 1. Dispersion of h-BN flakes 2 g of h-BN powder (Aldrich, ~1µm) was tip-sonicated in 50 ml of 2% w/v Pluronic F68 solution using a probe sonication system (Fisher Scientific Sonic Dismembrator Model 500) for 2 h at 50 W in an ice water bath. The resulting dispersion was centrifuged (Beckman Coulter Avanti® J-26 XPI) at 7,500 rpm (~10,000 g) for 10 min to remove the un-exfoliated flakes, and the supernatant was collected. 2. Size and thickness sorting of h-BN The collected supernatant was centrifuged (Beckman Coulter Optima™ L-80 XP with an SW28 rotor) at 20,000 rpm for 30 min to collect the large sized flakes, which were subsequently dispersed in 25 mL of 2% w/v Pluronic F68 by bath sonication for 1.5 h (End of Step 1). Step gradients were then prepared using an underlayer of 12 mL of 60% w/v iodixanol (density of 1.32 g/ml) and an overlayer of 25 mL of the large sized h-BN dispersions. The step gradients were then centrifuged at 32,000 rpm for 12 h at 22 °C. Subsequently, a piston gradient fractionator (Biocomp Instruments) was used to collect the fractions at a step of 1 mm. The four fractions (~2 mL) that were close to the concentrated white band of h-BN were combined and diluted by 1 mL of 2% w/v F68 for further separation (End of Step 2). Four layers in a centrifuge tube were stacked from bottom to top: 2 mL of 60% w/v iodixanol, 5 mL linear gradient of 20% to 60% w/v iodixanol, ~3 mL of h-BN dispersion from step 2, and 1 mL of 2% w/v F68, and they were then centrifuged (Beckman Coulter Optima™ L-80 XP with an SW41 rotor) at 41,000 rpm for 14 h at 22 °C. The piston gradient fractionator was then used to collect the fractions at a step of 0.5 mm (End of Step 3). 3. Van der Waals volume of F68 chain The van der Waals volume for F68 chain was calculated using Materials Studio (Accelrys, Inc.; San Diego, CA).1 The molecular structure of [PEO78PPO30PEO78] was built in the program.
S2
4. Measurement of h-BN buoyant density in each fraction A mechanical pipette was used to withdraw 20 μL of dispersion from each fraction in a vial. The mass difference was measured using an analytical balance (Mettler Toledo XS105). The density of the fraction was calculated as the measured mass divided by the volume (20 μL).
5. Layer-by-layer (LBL) assembly of h-BN multilayers The fraction f14 obtained in step 2 with a typical density of 1.14 g/mL was used for LBL assembly. A heavily doped p-type Si wafer (Montco Silicon Inc., , resistivity = 0.001–0.004 Ω cm) was first cleaned in oxygen plasma for 2 min, and rinsed with isopropanol and DI water. The treated substrate was immersed in 2 w/v % polyethyleneimine (PEI, average Mw ~25,000, Aldrich) for 2 min, and then dried by gentle N2 flow. The substrate was then immersed in f14 h-BN dispersion for 2 min, and similarly rinsed and dried. Those procedures complete one cycle of LBL assembly, and can be repeated to get a desired thickness. The obtained PEI/h-BN thin films were then annealed in air in a tube furnace for 30 min at 600 °C to remove the PEI. 6. LBL assembly of h-BN layer on top of SiO2 and HfOx A Si wafer with ~300 nm SiO2 on top (Silicon Quest International, , resistivity = 0.001– 0.005 Ω cm) was first cleaned with an O2 plasma, and then rinsed with isopropanol and DI water. The wafer was subsequently immersed in 2 w/v % (3-Aminopropyl)triethoxysilane (APTES, Aldrich) in DI water for 30 min, and then rinsed with DI water and dried. The substrate was then immersed into h-BN dispersion (f14) for another 30 min, and similarly rinsed and dried. Completing such a cycle can give a full coverage of h-BN on SiO2. The h-BN/SiO2 hybrid was then annealed at 600 °C in air. HfOx was deposited on a p+ type Si wafer (Montco Silicon Inc., , resistivity = 0.001–0.004 Ω cm) by atomic layer deposition (ALD) with a thickness of ~9 nm. The HfOx surface was cleaned by O2 plasma, isopropanol, and DI water, and subsequently immersed in 2 w/v % PEI for 30 min, and rinsed with
S3
DI water and dried. The substrate was then immersed into h-BN dispersion (f14) for another 30 min, and similarly rinsed and dried. Completing such a cycle can give a full coverage of h-BN on HfOx. The hBN/HfOx hybrid was then annealed at 600 °C in air. After annealing, the film structure was determined by ellipsometry to be: 4 nm h-BN/ 8.5 nm HfOx /2.2 nm SiO2 on Si. 7. Fabrication of graphene field effect transistors (GFETs) 120 µm × 120 µm, 2 nm Cr/50 nm Au electrodes on selected dielectrics were prepared by photolithography, metal evaporation, and lift-off. Monolayer graphene grown by chemical vapor deposition (CVD) on copper foil (Graphene Supermarket, Inc.) was then transferred onto those electrodes by a wellestablished PMMA-assisted transfer method. Specifically, the graphene/Cu foil was coated with PMMA (C4, MicroChem Corp) at 2,500 rpm for 2 min. The graphene on the backside of the foil was then etched by reactive ion etching (RIE) in O2 plasma. The foil was cut into required sizes, and then floated on 1.6 v/w% ammonium persulfate until the Cu was completely etched. The freestanding PMMA-coated graphene was then transferred into clean DI water with the assistance of a piece of polyethylene terephthalate foil; this was done several times for thorough rinsing. The floating PMMA/graphene was then gently transferred onto the substrates with patterned electrodes. The coated wafer was left in a vacuum oven overnight to completely dry off the water, and then annealed at 100 °C in ambient for 20 min, and at 190 °C in ambient for 30 min to improve the adhesion of graphene onto the dielectric. Finally, PMMA was dissolved in 40 °C acetone for 30 min, and rinsed with clean acetone and isopropanol. The GFETs were then made by photolithography and RIE to define channels with a length of 10 µm and width of 100 µm. Finally, the devices were cleaned twice in a resist stripper (Remover PG, MicroChem Inc.) at 60 °C for 30 min. 8. Areal capacitance measurement The areal capacitance and leakage current density of dielectrics were measured on the metal insulator semiconductor (MIS) capacitors before the graphene transfer step using a Keithley 4200-SCS and a Cascade Microtech Summit 12000 semi-automatic ambient probe station. The areal capacitance was
S4
obtained in the accumulation region. The 600 °C annealed HfOx and h-BN/HfOx had a capacitance of 734 nF/cm2 and 345 nF/cm2, respectively. The coated single layer of h-BN has an estimated dielectric constant of ~3. The SiO2 and h-BN/SiO2 samples have a typical capacitance of ~11 nF/cm2. 9. Characterization of GFETs The electrical performance of GFETs was measured in vacuum at a pressure less than 2 ×10-5 Torr with Keithley 2400 SourceMeters. The mobility can be extracted by correlating the device resistance (R) with gate voltage (Vg) using the following model:2,3 /
2
μ
and √ where unit,
is the contact resistance, L and W are the length and width of the channel, is the residual carrier concentration near the Dirac point (VDIRAC),
density,
is the areal capacitance of the dielectric,
is fundamental charge
is the gate field induced carrier
is the Planck’s constant divided by 2 , and
is the Fermi velocity of graphene (~106 m/s). μ or μ can be fit on the right and left side of the R vs. Vg curves with key fitting parameters of Rc, μ
μ , and
. The following is a list of typical fitting
parameters for different dielectrics: 1) Assembled h-BN dielectric, electron side: Rc = 76 Ω, µe = 6170 cm2 V-1s-1, n0 = 2.8 × 1011 cm-2; hole side: Rc = 76 Ω, µh = 7100 cm2 V-1s-1, n0 = 2.4 × 1011 cm-2. 2) SiO2, electron side: Rc = 35 Ω, µe = 4582 cm2 V-1s-1, n0 = 5.3 × 1011 cm-2; hole side: Rc = 20 Ω, µh = 4081 cm2 V-1s-1, n0 = 5.0 × 1011 cm-2. 3) h-BN/SiO2, electron side: Rc = 34 Ω, µe = 6543 cm2 V-1s-1, n0 = 4.2 × 1011 cm-2; hole side: Rc = 30 Ω, µh = 6496 cm2 V-1s-1, n0 = 3.8 × 1011 cm-2. 4) HfOx, electron side: Rc = 36 Ω, µe = 2352 cm2/Vs, n0 = 13.1 × 1011 cm-2; hole side: Rc = 22 Ω, µh = 2463 cm2 V-1s-1, n0 = 10.7 × 1011 cm-2. 5) hBN/HfOx, electron side: Rc = 27 Ω, µe = 3945 cm2 V-1s-1, n0 = 6.9 × 1011 cm-2; hole side: Rc = 29 Ω, µh = 5553 cm2 V-1s-1, n0 = 5.1 × 1011 cm-2.
S5
10. Other characterization tools Tapping Mode Atomic Force Microscopy (AFM) was conducted on an Asylum Cypher AFM or Bruker High Speed AFM. The AFM samples were prepared using the LBL assembly technique. X-ray photoelectron spectroscopy (XPS) was gathered using a Thermo Scientific ESCALAB 250 Xi. A charge compensating flood gun was employed for all samples. All core level scans were charge corrected to adventitious carbon at 284.8 eV. Transmission Electron Microcopy (TEM) analysis was carried out on a JEOL JEM-2100 TEM at an accelerating voltage of 200 kV. The cross-sectional TEM sample was prepared by focused ion beam (FIB) cutting using Helios NanoLab DualBeam. To protect the film surface from beam damage, a thin layer of osmium was coated on top before the FIB cutting. UV-vis optical absorbance spectroscopy was obtained with a Cary 5000 instrument. Thermogravimetric Analysis (TGA) used a Mettler Toledo TGA/SDTA851 system at a heating rate of 10 °C/min in air. Thicknesses were measured using a J.A. Woollam M2000U ellipsometer and fit with a Cauchy model. The Raman spectra of graphene were taken through a 100X objective on a Horiba XploRA ONETM Raman system with 532 nm laser and a 1200 gr/mm grating. The Raman spectra of h-BN were taken through a 100X objective with a 473 nm laser and a 1800 gr/mm grating.
S6
Section S2. Supporting Figures:
Figure S1. Dispersion protocol for h-BN with F68 surfactants. a) Typical SEM image of raw h-BN powders before processing. b) Molecular structure of copolymer surfactant F68. c) h-BN dispersed by F68 in water after a mild centrifugation of 5,000 rpm for 30 min. d) AFM image of h-BN with various sizes and thicknesses. e) Histogram of h-BN thicknesses.
S7
Figure S2. Zeta potential of F68 and F68-stabilized h-BN (BN-F68).
Figure S3. Using the h-BN supernatant solution for DGU separation. a) Band separation in a centrifuge tube after step 3 using the supernatant from step 2 instead of the precipitate. AFM height images of b) fraction F14, c) fraction F19, and d) fraction F23. e) Thickness distribution. f) Size distribution.
S8
Figure S4. AFM images during DGU processing steps. AFM height images of a) f14 and b) f15 after step 2, and c) F40 and d) F44 after step 3.
Figure S5. h-BN thickness and size distribution. a) Thickness versus number of layers in h-BN measured by AFM. b) Size distribution of h-BN from various fractions.
S9
Figure S6. Raman spectra of bulk h-BN and sorted h-BN in the different fractions. The few layer h-BN samples show blue shifts of the characteristic E2g peak of up to 3 cm-1 in comparison to the bulk h-BN sample.4
Figure S7. Thickness sorting of h-BN by DGU with sodium cholate (SC) as the surfactant. a) Molecular structure of sodium cholate. b) Band separation in a centrifuge tube after step 2. c) Band separation in a centrifuge tube after step 3. d) AFM images of h-BN in different fractions. e) Media density profile after step 3. f) Buoyant density of h-BN with 1 to 3 layers fit by a mathematical model. g) Thickness distribution for fractions of f8, f17, and f29, respectively.
S10
Figure S8. Thermogravimetric analysis of polyethyleneimine (PEI).
Figure S9. Raman metrics for graphene on SiO2, HfOx, and h-BN. a) Raman spectra of transferred CVDgraphene on SiO2, HfOx, and h-BN dielectrics. b) ID/IG distribution for the transferred graphene. c) 2D peak intensity mapping of graphene across the GFET channel on h-BN. d) Correlation between the frequencies of the G and 2D Raman modes of graphene on SiO2 (black), HfOx (cyan), and h-BN (pink).5,6
S11
Figure S10. AFM images of h-BN on different dielectrics. AFM height image of a) h-BN assembled on SiO2 and b) h-BN assembled on HfOx.
Fig. S11. Assembled h-BN film as an interfacial layer for other dielectrics used in GFETs. (a, b, c, d) Transfer curve, hysteresis, Dirac point voltage, and mobility comparison for the SiO2 and h-BN/SiO2 dielectrics. (e, f, g, h) Transfer curve, hysteresis, Dirac point voltage, and mobility comparison for the HfOx and h-BN/HfOx dielectrics. Vd = 5 mV for the devices on SiO2, HfOx, and their hybrids with h-BN.
S12
Figure S12. Model fitting of the transfer curve for graphene FETs with various dielectrics. a) h-BN, b) SiO2 and h-BN/SiO2, and c) HfOx and h-BN/ HfOx.
References: (1) (2) (3) (4) (5) (6)
Arnold, M. S.; Suntivich, J.; Stupp, S. I.; Hersam, M. C. ACS Nano 2008, 2, 2291-2300. Kim, S.; Nah, J.; Jo, I.; Shahrjerdi, D.; Colombo, L.; Yao, Z.; Tutuc, E.; Banerjee, S. K. Appl. Phys. Lett. 2009, 94, 062107. Sangwan, V. K.; Jariwala, D.; Everaerts, K.; McMorrow, J. J.; He, J.; Grayson, M.; Lauhon, L. J.; Marks, T. J.; Hersam, M. C. Appl. Phys. Lett. 2014, 104, 083503. Gorbachev, R. V.; Riaz, I.; Nair, R. R.; Jalil, R.; Britnell, L.; Belle, B. D.; Hill, E. W.; Novoselov, K. S.; Watanabe, K.; Taniguchi, T.; Geim, A. K.; Blake, P. Small 2011, 7, 465-468. Ahn, G.; Kim, H. R.; Ko, T. Y.; Choi, K.; Watanabe, K.; Taniguchi, T.; Hong, B. H.; Ryu, S. ACS Nano 2013, 7, 1533-1541. Lee, J. E.; Ahn, G.; Shim, J.; Lee, Y. S.; Ryu, S. Nat. Commun. 2012, 3, 1024.
S13
Solution-Processed Dielectrics Based on Thickness-Sorted TwoDimensional Hexagonal Boron Nitride Nanosheets †
†
†
†
†
†
Jian Zhu , Joohoon Kang , Junmo Kang , Deep Jariwala , Joshua D. Wood , Jung-Woo T. Seo , Kan† †‡§ †‡§∥ Sheng Chen , Tobin J. Marks , and Mark C. Hersam* †Department of Materials Science and Engineering, ‡Graduate Program in Applied Physics, § Department of Chemistry, and ∥Department of Medicine, Northwestern University, Evanston, Illinois 60208, United States *Address correspondence to: [email protected]
Contents:
Section S1. Methods o Section S1.1. Dispersion of h-BN Flakes o Section S1.2. Size and thickness sorting of h-BN o Section S1.3. Van der Waals volume of F68 chain o Section S1.4. Measurement of h-BN buoyant density in each fraction o Section S1.5. Layer-by-layer (LBL) assembly of h-BN multilayers o Section S1.6. LBL assembly of h-BN layer on top of SiO2 and HfOx o Section S1.7. Fabrication of graphene field-effect transistors (GFETs) o Section S1.8. Areal capacitance measurement o Section S1.9. Characterization of GFETs o Section S1.10. Other characterization tools Section S2. Supporting Figures o Figure S1. Dispersion protocol for h-BN with F68 surfactants o Figure S2. Zeta potential of F68 and F68-stabilized h-BN o Figure S3. Using the h-BN solution supernatant for DGU separation o Figure S4. AFM images during DGU processing steps o Figure S5. h-BN thickness and size distributions o Figure S6. Raman spectra of bulk h-BN and sorted h-BN in the different fractions. o Figure S7. Thickness sorting of h-BN by DGU with sodium cholate (SC) as surfactants o Figure S8. Thermogravimetric analysis of polyethyleneimine (PEI) o Figure S9. Raman metrics for graphene on SiO2, HfOx, and h-BN o Figure S10. AFM images of h-BN on different dielectrics o Figure S11. Assembled h-BN film as an interfacial layer for other dielectrics used in GFETs o Figure S12. Model fitting of the transfer curve for graphene FETs with various
dielectrics S1
Section S1. Methods: 1. Dispersion of h-BN flakes 2 g of h-BN powder (Aldrich, ~1µm) was tip-sonicated in 50 ml of 2% w/v Pluronic F68 solution using a probe sonication system (Fisher Scientific Sonic Dismembrator Model 500) for 2 h at 50 W in an ice water bath. The resulting dispersion was centrifuged (Beckman Coulter Avanti® J-26 XPI) at 7,500 rpm (~10,000 g) for 10 min to remove the un-exfoliated flakes, and the supernatant was collected. 2. Size and thickness sorting of h-BN The collected supernatant was centrifuged (Beckman Coulter Optima™ L-80 XP with an SW28 rotor) at 20,000 rpm for 30 min to collect the large sized flakes, which were subsequently dispersed in 25 mL of 2% w/v Pluronic F68 by bath sonication for 1.5 h (End of Step 1). Step gradients were then prepared using an underlayer of 12 mL of 60% w/v iodixanol (density of 1.32 g/ml) and an overlayer of 25 mL of the large sized h-BN dispersions. The step gradients were then centrifuged at 32,000 rpm for 12 h at 22 °C. Subsequently, a piston gradient fractionator (Biocomp Instruments) was used to collect the fractions at a step of 1 mm. The four fractions (~2 mL) that were close to the concentrated white band of h-BN were combined and diluted by 1 mL of 2% w/v F68 for further separation (End of Step 2). Four layers in a centrifuge tube were stacked from bottom to top: 2 mL of 60% w/v iodixanol, 5 mL linear gradient of 20% to 60% w/v iodixanol, ~3 mL of h-BN dispersion from step 2, and 1 mL of 2% w/v F68, and they were then centrifuged (Beckman Coulter Optima™ L-80 XP with an SW41 rotor) at 41,000 rpm for 14 h at 22 °C. The piston gradient fractionator was then used to collect the fractions at a step of 0.5 mm (End of Step 3). 3. Van der Waals volume of F68 chain The van der Waals volume for F68 chain was calculated using Materials Studio (Accelrys, Inc.; San Diego, CA).1 The molecular structure of [PEO78PPO30PEO78] was built in the program.
S2
4. Measurement of h-BN buoyant density in each fraction A mechanical pipette was used to withdraw 20 μL of dispersion from each fraction in a vial. The mass difference was measured using an analytical balance (Mettler Toledo XS105). The density of the fraction was calculated as the measured mass divided by the volume (20 μL).
5. Layer-by-layer (LBL) assembly of h-BN multilayers The fraction f14 obtained in step 2 with a typical density of 1.14 g/mL was used for LBL assembly. A heavily doped p-type Si wafer (Montco Silicon Inc., , resistivity = 0.001–0.004 Ω cm) was first cleaned in oxygen plasma for 2 min, and rinsed with isopropanol and DI water. The treated substrate was immersed in 2 w/v % polyethyleneimine (PEI, average Mw ~25,000, Aldrich) for 2 min, and then dried by gentle N2 flow. The substrate was then immersed in f14 h-BN dispersion for 2 min, and similarly rinsed and dried. Those procedures complete one cycle of LBL assembly, and can be repeated to get a desired thickness. The obtained PEI/h-BN thin films were then annealed in air in a tube furnace for 30 min at 600 °C to remove the PEI. 6. LBL assembly of h-BN layer on top of SiO2 and HfOx A Si wafer with ~300 nm SiO2 on top (Silicon Quest International, , resistivity = 0.001– 0.005 Ω cm) was first cleaned with an O2 plasma, and then rinsed with isopropanol and DI water. The wafer was subsequently immersed in 2 w/v % (3-Aminopropyl)triethoxysilane (APTES, Aldrich) in DI water for 30 min, and then rinsed with DI water and dried. The substrate was then immersed into h-BN dispersion (f14) for another 30 min, and similarly rinsed and dried. Completing such a cycle can give a full coverage of h-BN on SiO2. The h-BN/SiO2 hybrid was then annealed at 600 °C in air. HfOx was deposited on a p+ type Si wafer (Montco Silicon Inc., , resistivity = 0.001–0.004 Ω cm) by atomic layer deposition (ALD) with a thickness of ~9 nm. The HfOx surface was cleaned by O2 plasma, isopropanol, and DI water, and subsequently immersed in 2 w/v % PEI for 30 min, and rinsed with
S3
DI water and dried. The substrate was then immersed into h-BN dispersion (f14) for another 30 min, and similarly rinsed and dried. Completing such a cycle can give a full coverage of h-BN on HfOx. The hBN/HfOx hybrid was then annealed at 600 °C in air. After annealing, the film structure was determined by ellipsometry to be: 4 nm h-BN/ 8.5 nm HfOx /2.2 nm SiO2 on Si. 7. Fabrication of graphene field effect transistors (GFETs) 120 µm × 120 µm, 2 nm Cr/50 nm Au electrodes on selected dielectrics were prepared by photolithography, metal evaporation, and lift-off. Monolayer graphene grown by chemical vapor deposition (CVD) on copper foil (Graphene Supermarket, Inc.) was then transferred onto those electrodes by a wellestablished PMMA-assisted transfer method. Specifically, the graphene/Cu foil was coated with PMMA (C4, MicroChem Corp) at 2,500 rpm for 2 min. The graphene on the backside of the foil was then etched by reactive ion etching (RIE) in O2 plasma. The foil was cut into required sizes, and then floated on 1.6 v/w% ammonium persulfate until the Cu was completely etched. The freestanding PMMA-coated graphene was then transferred into clean DI water with the assistance of a piece of polyethylene terephthalate foil; this was done several times for thorough rinsing. The floating PMMA/graphene was then gently transferred onto the substrates with patterned electrodes. The coated wafer was left in a vacuum oven overnight to completely dry off the water, and then annealed at 100 °C in ambient for 20 min, and at 190 °C in ambient for 30 min to improve the adhesion of graphene onto the dielectric. Finally, PMMA was dissolved in 40 °C acetone for 30 min, and rinsed with clean acetone and isopropanol. The GFETs were then made by photolithography and RIE to define channels with a length of 10 µm and width of 100 µm. Finally, the devices were cleaned twice in a resist stripper (Remover PG, MicroChem Inc.) at 60 °C for 30 min. 8. Areal capacitance measurement The areal capacitance and leakage current density of dielectrics were measured on the metal insulator semiconductor (MIS) capacitors before the graphene transfer step using a Keithley 4200-SCS and a Cascade Microtech Summit 12000 semi-automatic ambient probe station. The areal capacitance was
S4
obtained in the accumulation region. The 600 °C annealed HfOx and h-BN/HfOx had a capacitance of 734 nF/cm2 and 345 nF/cm2, respectively. The coated single layer of h-BN has an estimated dielectric constant of ~3. The SiO2 and h-BN/SiO2 samples have a typical capacitance of ~11 nF/cm2. 9. Characterization of GFETs The electrical performance of GFETs was measured in vacuum at a pressure less than 2 ×10-5 Torr with Keithley 2400 SourceMeters. The mobility can be extracted by correlating the device resistance (R) with gate voltage (Vg) using the following model:2,3 /
2
μ
and √ where unit,
is the contact resistance, L and W are the length and width of the channel, is the residual carrier concentration near the Dirac point (VDIRAC),
density,
is the areal capacitance of the dielectric,
is fundamental charge
is the gate field induced carrier
is the Planck’s constant divided by 2 , and
is the Fermi velocity of graphene (~106 m/s). μ or μ can be fit on the right and left side of the R vs. Vg curves with key fitting parameters of Rc, μ
μ , and
. The following is a list of typical fitting
parameters for different dielectrics: 1) Assembled h-BN dielectric, electron side: Rc = 76 Ω, µe = 6170 cm2 V-1s-1, n0 = 2.8 × 1011 cm-2; hole side: Rc = 76 Ω, µh = 7100 cm2 V-1s-1, n0 = 2.4 × 1011 cm-2. 2) SiO2, electron side: Rc = 35 Ω, µe = 4582 cm2 V-1s-1, n0 = 5.3 × 1011 cm-2; hole side: Rc = 20 Ω, µh = 4081 cm2 V-1s-1, n0 = 5.0 × 1011 cm-2. 3) h-BN/SiO2, electron side: Rc = 34 Ω, µe = 6543 cm2 V-1s-1, n0 = 4.2 × 1011 cm-2; hole side: Rc = 30 Ω, µh = 6496 cm2 V-1s-1, n0 = 3.8 × 1011 cm-2. 4) HfOx, electron side: Rc = 36 Ω, µe = 2352 cm2/Vs, n0 = 13.1 × 1011 cm-2; hole side: Rc = 22 Ω, µh = 2463 cm2 V-1s-1, n0 = 10.7 × 1011 cm-2. 5) hBN/HfOx, electron side: Rc = 27 Ω, µe = 3945 cm2 V-1s-1, n0 = 6.9 × 1011 cm-2; hole side: Rc = 29 Ω, µh = 5553 cm2 V-1s-1, n0 = 5.1 × 1011 cm-2.
S5
10. Other characterization tools Tapping Mode Atomic Force Microscopy (AFM) was conducted on an Asylum Cypher AFM or Bruker High Speed AFM. The AFM samples were prepared using the LBL assembly technique. X-ray photoelectron spectroscopy (XPS) was gathered using a Thermo Scientific ESCALAB 250 Xi. A charge compensating flood gun was employed for all samples. All core level scans were charge corrected to adventitious carbon at 284.8 eV. Transmission Electron Microcopy (TEM) analysis was carried out on a JEOL JEM-2100 TEM at an accelerating voltage of 200 kV. The cross-sectional TEM sample was prepared by focused ion beam (FIB) cutting using Helios NanoLab DualBeam. To protect the film surface from beam damage, a thin layer of osmium was coated on top before the FIB cutting. UV-vis optical absorbance spectroscopy was obtained with a Cary 5000 instrument. Thermogravimetric Analysis (TGA) used a Mettler Toledo TGA/SDTA851 system at a heating rate of 10 °C/min in air. Thicknesses were measured using a J.A. Woollam M2000U ellipsometer and fit with a Cauchy model. The Raman spectra of graphene were taken through a 100X objective on a Horiba XploRA ONETM Raman system with 532 nm laser and a 1200 gr/mm grating. The Raman spectra of h-BN were taken through a 100X objective with a 473 nm laser and a 1800 gr/mm grating.
S6
Section S2. Supporting Figures:
Figure S1. Dispersion protocol for h-BN with F68 surfactants. a) Typical SEM image of raw h-BN powders before processing. b) Molecular structure of copolymer surfactant F68. c) h-BN dispersed by F68 in water after a mild centrifugation of 5,000 rpm for 30 min. d) AFM image of h-BN with various sizes and thicknesses. e) Histogram of h-BN thicknesses.
S7
Figure S2. Zeta potential of F68 and F68-stabilized h-BN (BN-F68).
Figure S3. Using the h-BN supernatant solution for DGU separation. a) Band separation in a centrifuge tube after step 3 using the supernatant from step 2 instead of the precipitate. AFM height images of b) fraction F14, c) fraction F19, and d) fraction F23. e) Thickness distribution. f) Size distribution.
S8
Figure S4. AFM images during DGU processing steps. AFM height images of a) f14 and b) f15 after step 2, and c) F40 and d) F44 after step 3.
Figure S5. h-BN thickness and size distribution. a) Thickness versus number of layers in h-BN measured by AFM. b) Size distribution of h-BN from various fractions.
S9
Figure S6. Raman spectra of bulk h-BN and sorted h-BN in the different fractions. The few layer h-BN samples show blue shifts of the characteristic E2g peak of up to 3 cm-1 in comparison to the bulk h-BN sample.4
Figure S7. Thickness sorting of h-BN by DGU with sodium cholate (SC) as the surfactant. a) Molecular structure of sodium cholate. b) Band separation in a centrifuge tube after step 2. c) Band separation in a centrifuge tube after step 3. d) AFM images of h-BN in different fractions. e) Media density profile after step 3. f) Buoyant density of h-BN with 1 to 3 layers fit by a mathematical model. g) Thickness distribution for fractions of f8, f17, and f29, respectively.
S10
Figure S8. Thermogravimetric analysis of polyethyleneimine (PEI).
Figure S9. Raman metrics for graphene on SiO2, HfOx, and h-BN. a) Raman spectra of transferred CVDgraphene on SiO2, HfOx, and h-BN dielectrics. b) ID/IG distribution for the transferred graphene. c) 2D peak intensity mapping of graphene across the GFET channel on h-BN. d) Correlation between the frequencies of the G and 2D Raman modes of graphene on SiO2 (black), HfOx (cyan), and h-BN (pink).5,6
S11
Figure S10. AFM images of h-BN on different dielectrics. AFM height image of a) h-BN assembled on SiO2 and b) h-BN assembled on HfOx.
Fig. S11. Assembled h-BN film as an interfacial layer for other dielectrics used in GFETs. (a, b, c, d) Transfer curve, hysteresis, Dirac point voltage, and mobility comparison for the SiO2 and h-BN/SiO2 dielectrics. (e, f, g, h) Transfer curve, hysteresis, Dirac point voltage, and mobility comparison for the HfOx and h-BN/HfOx dielectrics. Vd = 5 mV for the devices on SiO2, HfOx, and their hybrids with h-BN.
S12
Figure S12. Model fitting of the transfer curve for graphene FETs with various dielectrics. a) h-BN, b) SiO2 and h-BN/SiO2, and c) HfOx and h-BN/ HfOx.
References: (1) (2) (3) (4) (5) (6)
Arnold, M. S.; Suntivich, J.; Stupp, S. I.; Hersam, M. C. ACS Nano 2008, 2, 2291-2300. Kim, S.; Nah, J.; Jo, I.; Shahrjerdi, D.; Colombo, L.; Yao, Z.; Tutuc, E.; Banerjee, S. K. Appl. Phys. Lett. 2009, 94, 062107. Sangwan, V. K.; Jariwala, D.; Everaerts, K.; McMorrow, J. J.; He, J.; Grayson, M.; Lauhon, L. J.; Marks, T. J.; Hersam, M. C. Appl. Phys. Lett. 2014, 104, 083503. Gorbachev, R. V.; Riaz, I.; Nair, R. R.; Jalil, R.; Britnell, L.; Belle, B. D.; Hill, E. W.; Novoselov, K. S.; Watanabe, K.; Taniguchi, T.; Geim, A. K.; Blake, P. Small 2011, 7, 465-468. Ahn, G.; Kim, H. R.; Ko, T. Y.; Choi, K.; Watanabe, K.; Taniguchi, T.; Hong, B. H.; Ryu, S. ACS Nano 2013, 7, 1533-1541. Lee, J. E.; Ahn, G.; Shim, J.; Lee, Y. S.; Ryu, S. Nat. Commun. 2012, 3, 1024.
S13
