Figure 1. h-BN crystal structure, thermodynamic stability, and synthesis methods. A) Crystallographic structure of cubic, wurtzite, hexagonal, and rhombohedral BN. Adapted with permission. [3] Copyright 2017, Wiley. B) Layered van der Waals structure of bulk h-BN and (C) planar view of an atomically thin sheet of sp 2 -bonded B and N atoms that comprise the h-BN crystal or boron nitride nanosheets (BNNS). Adapted with permission. [6] Copyright 2012, RSC. Synthetic bulk h-BN is generally found to stack in an AA′ configuration, however other stacking has also been observed in some cases. [5] . D) Thermodynamic P,T phase diagram for boron nitride. Adapted with permission. [7] Copyright 1999, ACS. E) Energy required for transformation between crystal phases. Cubic BN has the lowest energy, however significant energy required to transform one crystal phase of BN to another makes h-BN quite stable chemically and thermally. Adapted with permission. [8] Copyright 2003, APS. F) (Left) Example of single layered h-BN transferred to SiO 2 /Si wafer after growth via chemcial vapor deposition (CVD). Adapted with permission. [9] Copyright 2015, Springer Nature. F) (Right) Example of bulk h-BN crystal grown from Ni-Mo flux process. Adapted with permission. [10] Copyright 2007, AAAS. G) h-BN synthesis methods as a function of While the cubic BN form is the most thermodynamically favorable under ambient conditions, other meta-stable forms of BN can be stabilized under ambient conditions due to the relatively large transformation energies (experimental ~6.5-10.8 eV atom -1 , theoretical ~9.4 eV atom -1 for h-BN to c-BN conversion) associated with phase change (Figure 1D,E). [8] This review will focus on the h-BN phase and the synthesis thereof. Readers interested in other forms of BN are directed to several other excellent reviews. [16][17][18] As a ceramic material h-BN has been conventionally used for thermal management, heat shielding, lubrication, and as a filler material for structural composites. [16,19] The experimental isolation and characterization of graphene in 2004 sparked significant interest in isolating and studying monolayers of other layered van der Waals materials to exploit their unique properties. [20] In particular, h-BN has emerged as a material of interest for applications in nanoelectronics [13,[21][22][23][24][25][26][27][28][29][30][31][32][33] due to its wide bandgap [3] , high thermal conductivity, [34] ultra-flat and unreactive surface, and electron tunneling properties. [35] h-BN is of interest for photonic and optoelectronic applications [13,25,[36][37][38] owing to deep ultraviolet emissions, hyperbolic optical properties in the mid-IR, and room temperature single photon quantum emissions. [39] Since h-BN contains boron, an effective neutron absorber (especially 10 B), it has been investigated as a neutron detection material. [40][41][42][43][44][45][46] h-BN is also impermeable to all molecular species, allowing for ultra-thin oxidation protection, [47] and facilitates the development of atomically thin membranes for atomic and molecular separations. [35,[48][49][50][51][52][53][54][55][56][57] Despite significant research interest in h-BN, synthesis of high-quality h-BN remains challenging and the subject of ongoing work. Figure 1G and Table 1 provide a summary of the different synthesis methods, discusses crystal quality and process scalability, along with the advantages and disadvantages of each method, respectively. h-BN powders have been produced for more than a century and significant research has allowed for development of scalable cost-effective synthesis processes. [1,16,19] However, the synthesis of larger h-BN crystals of sizes greater than a few microns has yet to achieve parity with that of powder synthesis. [1,3,58] Large bulk h-BN crystals (millimeters) have been grown successfully, however synthesis methods have typically involved using high pressure and high temperature (HPHT) processes (>4 GPa and >1400 °C), [36,[59][60][61] or more recently from catalytic flux-based growth methods at atmospheric pressure but still requiring high processing temperatures (>1400 °C). [10,[62][63] Such flux based methods involve the dissolution of B and N into a metal solvent, then slowly cooling the flux, resulting in precipitation of B and N out of the metal solvent, forming h-BN crystals on the surface. [36,10,63] These bulk crystals have been mechanically exfoliated down to few layers and even atomically thin single layers using, for example, the scotch tape method, and transferred to appropriate target substrates for electronic devices, a process similar to that first developed to isolate graphene. [64,65] Although these methods allow for the highest h-BN quality, and have facilitated the majority of academic research projects, controlling crystal thickness and stacking configuration is nontrivial. Furthermore, the limited size of flakes produced by mechanical exfoliation of these h-BN crystals (generally micronscale, Figure 1F) as well as the high processing pressures and temperatures required for synthesis presents inherent challenges to achieving scalable, cost-effective, reliable, and re-producible synthesis for practical applications.

Deposition processes such as molecular beam epitaxy (MBE), chemical vapor deposition (CVD) and metal-organic chemical vapor deposition (MOCVD) allow for the synthesis of continuous mono and multilayer h-BN films. [3,[66][67][68] Unlike HPHT, these processes are typically performed under ambient or sub-ambient (vacuum) pressures and the areal coverage is, in-principle, only limited by the size of the substrate/catalyst (Figure 1F). However, the quality of h-BN crystals is slightly lower compared to HPHT, that is, polycrystalline thin films. [69] Nonetheless, these methods can yield large-area uniform h-BN films at ambient to low-pressures at growth temperatures ≈750-1100 °C (depending on substrate/catalyst). [70][71][72] Among the different deposition-based synthesis methods, CVD using catalytic polycrystalline metal foils has emerged as one of the most preferred route for scalable, cost-effective synthesis of large-area continuous h-BN films. However, the h-BN films synthesized via CVD typically need to be transferred from the catalyst surface to an appropriate substrate for fabricating functional devices. Here, we note roll-to-roll transfer processes developed for graphene could, in-principle, be adapted for h-BN transfer. [73][74][75][76] In catalytic CVD, the catalyst typically provides an alternative lower energy pathway for the h-BN synthesis reaction allowing for h-BN synthesis at temperatures and pressures lower than those that would be required for a pure BN phase change at thermodynamic equilibrium. Additionally, the catalyst shape and complexity directly impacts the shape and complexity of the growing h-BN nanostructure. For example, a flat, planer metallic catalyst substrate typically facilitates the growth of planer h-BN crystals/films, while nanoparticle catalysts template the formation of h-BN nanotubes (BNNTs) (Figure 2). [77,6] Developing a detailed fundamental understanding of h-BN growth mechanisms and its interaction with the catalyst during CVD is hence imperative to rational design of synthesis and processing routes. In situ characterization during growth can aid in understanding growth mechanisms as well as the catalyst-h-BN interaction during the growth process. However, limited compatibility of in situ characterization probes with commonly used pressures (pressure-gap) and temperature for CVD as well as the complexity of the catalyst (polycrystalline foils vs single crystals vs nanoparticles-known as the materials-gap) presents some challenges (Figure 2). [78][79][80][81][82][83] Recent and rapid advances in surface science to bridge the pressure-gap by enabling differential pumping and novel measurement configurations/devices have resulted in some of these techniques being compatible with h-BN CVD conditions. [82] Since h-BN is a synthetic crystal, synthesis approaches can be classified as primarily bottom-up for bulk as well as thin films of h-BN. Here, we focus on bottom-up h-BN growth on different substrates/catalysts with an emphasis on understanding the growth mechanisms to identify opportunities for advancing scalable high-quality synthesis for novel applications. For details on physics involving the complex h-BN substrate interactions, we refer the reader to a comprehensive review by Auwärter et al. [66] For details of other forms of BN synthesis including liquid-phase exfoliation of bulk h-BN crystals, chemical functionalization of h-BN, BNNTs, h-BN nanoribbons, and h-BN growth on 3D nanostructures, amorphous h-BN, [84][85][86] and conversion of graphene to h-BN [86][87][88] we refer the reader to relevant resources in the literature. [89,90]

Synthesis of h-BN films/layers has been explored using insulating/inert substrates as well as catalytic transition metal substrates, that is, polycrystalline metal foils, single crystals, pure metals, and alloys etc. A typical bottom-up h-BN growth process consists of a substrate/catalyst initially annealed and treated followed by exposure to precursors containing B and N compounds at elevated temperatures. A catalytic substrate facilitates the decomposition of precursors on the surface and subsequent nucleation and growth of h-BN.

Various precursors can be used to synthesis h-BN. For example, ammonia-borane (solid powder at room temperature) and borazine (liquid at room temperature) are often used since they naturally contain a stoichiometric ratio of N:B. However, on some systems (sapphire, for example) growers have found that tuning the ratio of N:B precursor (V/III) can improve film quality, likely because of differing precursor decomposition rates, intermediate species formed, sticking coefficients, and desorption/etching rates which could result in non-stochiometric availability of B:N on the growth surface despite stochiometric feed. [68] Also, for atomic-layer deposition (ALD) type strategies, independent N and B precursors are required. For nitrogen containing precursors, ammonia (NH 3 ) is the most common choice, however N 2 can also be used albeit the higher decomposition energy necessitates higher temperature or catalytic substrates. Diborane and boron trichloride are some common inorganic options, while trimethylboron (TMB) and triethylboron (TEB) are the most common organic options for boron-containing precursors. Unfortunately, most of these boron containing precursors are highly toxic to humans and/ or are extremely pyrophoric.

We review catalyst/substrate systems used for h-BN growth and summarize growth conditions and details in tables, while figures highlight salient observations, and discuss common emerging themes and state-of-the-art advances.

The availability of single crystalline sapphire wafers and their use for epitaxial growth of metal films as well as other III-V semiconductors (GaAs, GaN etc.) inspired early studies of h-BN growth on sapphire wafer (see Table 2 and Figure 3). Early experiments typically resulted in thick multi-layer h-BN films, that is, in 1986, Nakamura investigated h-BN growth via MOCVD, achieving a range of amorphous and turbostratic h-BN on the c-plane of sapphire by tuning ammonia (NH 3 ) and TEB precursor ratios between 0 and 200 (N:B) at 750-1200 °C. [68] Subsequent studies found that even higher ratios of N:B containing precursors than those previously studied by Nakamura et al. [68] resulted in higher crystalline quality of multilayer h-BN films (Figure 3A) [91][92][93][94][95][96] as observed via more pronounced/sharper XRD peaks corresponding to the (0002) h-BN crystal plane. Although, higher flux of N containing precursors resulted in smoother and more crystalline h-BN film growth on sapphire, maintaining a stoichiometric precursor balance at the reaction site was essential and too much NH 3 was found to limit growth due to saturation of the surface with nitrogen, limiting further adsorption of boron species. [96] Interestingly, alternating the flow of ammonia and TEB at high N:B ratios, allowed for effective growth of thick h-BN films. [41,43,44] Higher temperatures (>1200 °C) are required to grow high quality h-BN on sapphire due to the relatively inert nature/poor catalytic activity of the sapphire substrates. [24,[41][42][43][44]99] At growth temperatures >1400 °C, 2-6 layers of highly ordered stacked layers of h-BN is formed on the sapphire surface as observed under cross-sectional transmission electron microscopy (TEM) (see Figure 3B). [99] Low-energy electron diffraction (LEED) imaging reveals radially aligned diffraction spots corresponding to the h-BN and sapphire crystal structures indicating potential epitaxial relationship between h-BN and the substrate, that is, the zig-zag edge of h-BN is parallel to the (1120) plane of sapphire across the entire 2″ diameter wafer (see schematic in Figure 3B). [99] Interestingly, the h-BN film appears to be aligned out of plane indicating AA′ stacking (see cross-sectional TEM image in Figure 3B) [99] and the film appears free of wrinkles yet the out-of-plane distance between layers is observed to be ≈0.36 nm which is slightly larger than that for bulk h-BN (≈0.34 nm) in spite of the in-plane lattice constants of the armchair edge being similar to that of bulk h-BN (0.217 nm vs 0.218 nm, respectively). [99] The increased spacing has been hypothesized to originate from lateral strain induced by lattice-mismatch between the h-BN film and sapphire substrate that results in a slight corrugation in the lower h-BN layer (≈2.6% contraction reported) which may result in the upper h-BN layer being further separated. [99] Other studies reported significant wrinkles (Figure 3C) after growth of thicker h-BN films on sapphire, providing visual evidence that the growing h-BN film is subject to strain induced by the difference in thermal expansion coefficient and/or the lattice mismatch with the sapphire wafer. [92] The addition of buffer layers between the sapphire substrate and growing h-BN have been explored to alleviate the strain on the growing film. [93,95] For example, the addition of a thin (≈10-20 nm) buffer AlN or amorphous BN resulted in smoother h-BN film with a greater degree of uniformity (see Figure 3D). [95] Considering production scalability and application, h-BN growth on sapphire substrates presents the following advantages: i) sapphire is a widely available single crystalline substrate with well-established production methods already in place; ii) evidence of potential epitaxy between h-BN and sapphire indicate that large area single crystal h-BN growth may be possible and is only limited by the size of sapphire wafer; iii) gas phase thin film synthesis methods have been demonstrated to be effective at industrial scale for several other materials; and iv) the multi-layer h-BN films grown on sapphire can be delaminated, [45,102] allowing for integration into existing growth process lines that produce III/V semiconductors such as GaN, InN etc., as well as transfer of device stacks and potential re-use of the substrates. [23,24] While most h-BN growth studies on sapphire substrates have focused on MOCVD, processes such as pulsed layer deposition, have also been explored (Figure 3E), albeit the crystallinity of these films has yet to reach the same quality as those formed via MOCVD. [98,106] Further, the sapphire substrate can also be readily patterned, allowing for h-BN growth on 3D surface features (Figure 3F). [100] Finally, growth on sapphire is conducive to relatively thick films (micron scale) that are freestanding, allowing for flexible h-BN substrates for devices (Figure 3F). [40,45,102] The primary drawback to using X-ray photoelectron spectroscopy (XPS) advances have enabled nearambient pressure XPS allowing for probing of a wider pressure range. [35] Environmental transmission electron microscopy (ETEM) allows for in situ imaging during growth, but is limited to thin catalysts which can de-wet to form nanoparticles from which BNNTs can be grown. For h-BN formation on a solid surface, environmental scanning electron microscopy (ESEM) is effective at probing how surface morphology of h-BN changes during growth. Low-energy electron microscopy and diffraction (LEEM and LEED) allows for investigation of h-BN growth on single crystalline substrates over micron scale areas. Scanning tunneling microscopy (STM) remains largely limited to ultrahigh vacuum conditions and single crystalline substrates and probes small areas.

sapphire as a growth substrate is its inherent low catalytic activity leading to higher growth temperatures and perhaps lower quality crystals (Table 2).

h-BN growth has also been demonstrated on other inert substrates. For example, Lee et al. used ALD methods to grow few-nm thick h-BN films on SiO 2 at 600 °C using NH 3 and TEB as precursors. [107] h-BN has also been grown directly on SiO 2 substrates via low pressure chemical vapor deposition (LPCVD) using a solid ammonia-borane precursor, resulting in h-BN films up to a few tens of nm thick. [108][109][110] Similar LPCVD methods have been used to grow h-BN films directly on quartz. [108,109] Yang et al. used pulsed-mode metal-organic vapor phase epitaxy to grow an h-BN film of ≈20 nm directly on diamond (111) and (100) using NH 3 and TEB (V/III = 3000) at 1380 °C, [111] and found that the h-BN showed a highly ordered lattice with an epitaxial relationship to the underlying diamond (111) surface. [111] Finally Behura et al., demonstrate h-BN growth directly on Si 3 N 4 , Si, and SiO 2 using LPCVD from ammoniaborane at 1100 °C, and found that growth rate correlated with substrate electronegativity (Si 3 N 4 > SiO 2 > Si). [110] The authors suggest that the observed trend in growth rate is consistent with adsorption rates of precursor molecules calculated using molecular dynamic simulations. [110] Growth of h-BN directly on inert substrates is appealing since it allows for direct device integrations without the need for material transfer and the availability of single crystalline substrates potentially allows for epitaxial relations that could in principle enable single crystalline h-BN synthesis, however the resulting h-BN films suffer from poor crystallinity and generally demand high growth temperatures. [86] h-BN films grown on SiO 2 via LPCVD exhibited a larger full width at half maximum (FWHM) of the characteristic E 2g Raman peak as compared to h-BN grown on catalytic metal substrates (≈45 cm -1 vs ≈20 cm -1 ) and small crystalline domains (≈25 nm). [109] Finally, h-BN growth on 6H-SiC and Si (100) has also been demonstrated. Majety et al., grew h-BN on 6H-SiC using MOCVD from NH 3 and TEB precursors at 1300 °C after depositing ≈20 nm BN buffer layer at 800 °C. [112] B 1-x N x films have been grown on Si(100) at room temperature using N 2 + ion-beam assisted evaporation of boron [113] as well as hexagonal-like BN grown on Si(100) by pulsed layer deposition methods. [114] BN films have also been grown on Si substrates by magnetron sputtering with a boron target and N 2 environment. [115] However, the synthesis of continuous, highly crystalline films using these methods remains elusive.

Ni was among the first transition metal catalysts investigated as a catalyst for h-BN growth due to the very similar lattice parameters between the Ni (111) surface (a = 2.49 Å) and the basal plane of h-BN (a = 2.50 Å), allowing for near-epitaxial growth with minimal strain on the h-BN lattice. Under ultrahigh vacuum (UHV) conditions (≈1 × 10 -8 Pa), Nagashima et al., [116] reported growth of an h-BN film on Ni (111) via the decomposition of borazine at 800 °C. We note the significantly lower growth temperature on Ni compared to sapphire substrates (Table 2 vs Table 3), suggesting the reaction is mediated by increased catalytic activity of the Ni substrate. [116] While only 100 Langmuir borazine exposure (1 Langmuir = exposure at 10 -6 Torr for one second which leads approximately to one monolayer absorbed assuming a perfect sticking coefficient [117] ) was needed to form the first monolayer of h-BN, subsequent layer growth required about 300 times the exposure required for the first layer, further supporting the theory that contact with the Ni (111) surface plays an important catalytic function in h-BN growth. [116] Scanning tunneling microscopy (STM), X-ray photoelectron spectroscopy (XPS) and LEED have since confirmed epitaxial relationships between as-grown monolayer h-BN and the underlying Ni (111) substrate. [116, Studies have attempted to probe h-BN/Ni (111) interaction by intercalating ions at the h-BN/Ni (111) interface and measuring the corresponding XPS peak shift [66,127,150] and LEED intensity analysis have confirmed that the h-BN lattice forms a highly commensurate relationship with the Ni (111) substrate where the nitrogen atoms sit on top of Ni atoms and the B atoms reside over Ni vacancy sites (see schematic in Figure 4A). [132,156] Further, Gamou et al., [156] found that the monolayer h-BN lattice sits 2.04 Å from the topmost Ni layer which is significantly closer than the interlaying spacing of bulk h-BN (3.34 Å). [156] Prevost et al. performed scanning transmission electron microscopy (STEM) on multilayer h-BN films grown on polycrystalline Ni and observed that h-BN grown on Ni(111)-like grains appeared continuous and uniform, while h-BN grown on Ni(001)-like grains was observed to be polycrystalline with sub-micron sized grains. [158] Despite evidence that h-BN grown on Ni (111) forms a continuous and uniform h-BN crystal, [156] domain boundaries in h-BN have been observed in films on single-crystal Ni(111) substrates [132] as well as polycrystalline Ni. [148] It has been suggested that the domain boundaries observed in h-BN grown on single crystal Ni (111) are the result of two different pseudo-epitaxial relationships, one where the B atoms of h-BN occupy the space above Ni(111) fcc hollow sites, and the other where the B atoms occupy the space above hcp hollow sites (Figure 4A). [118,128,132] Finally, in situ LEEM during h-BN growth on Ni (111) confirm isothermal nucleation and growth [152] (Figure 4B), wherein B is more soluble in Ni (≈0.3 at%, 1085 °C) than N (0.004 at% at 1550 °C, [161] and likely lower at growth temperatures ≈1200 °C). [162] More recently, a LPCVD process on large single crystal Ni (111) substrates resulted in uniform few-layered growth (AA′A stacking) of almost entirely of single orientation, commensurate to the Ni (111) surface. [160] Notably the cooling rate did not affect the thickness of the grown h-BN film. Considering N is relatively insoluble in Ni, these observations suggest h-BN growth occurs via a surface-mediated isothermal process, rather than via precipitation. Interestingly, a thick (≈200 nm) Ni 23 B 6 alloy was observed on the Ni surface, just below the growing h-BN film, and the thickness of this alloy layer was dependent on cooling rate (slower cooling rate resulting in a thicker Ni 23 B 6 layer), strongly suggesting that this Ni-B alloy forms during cooling as B precipitates toward the surface (Figure 4D).

Despite uniform h-BN growth on single-crystal Ni (111), growth on other Ni crystal facets occur at widely different rates (Figure 4C). [147,149] Atomic force microscopy (AFM), Raman and TEM analysis of the synthesized h-BN films confirm that different crystalline facets in polycrystalline Ni films/foils exhibit significantly different h-BN growth rates (due to differences in precursor reactivity on each facet, catalytic activity toward h-BN formation and h-BN etching rates), resulting in films of inconsistent thickness. [103,147,149] Further, the choice of precursor may also influences h-BN growth, that is, Lee et al. [147] reported the fastest growth of h-BN on Ni(100) facets using ammoniaborane precursor at ≈800 °C, while Chou et al. [149] report the slowest growth on Ni(100) facets and the fastest growth on Ni(110) facets using ammonia and diborane at ≈1000 °C. Since these studies used different precursors at different growth temperatures, it is likely that the growth rate of h-BN on differing facets is limited by the catalytic decomposition of precursors and that decomposition efficiency may scale differently with temperature on differing facets resulting in varying B and N precursor flux at the h-BN reaction site. Regardless of which facet facilitates that fastest growth, h-BN films grown on polycrystalline Ni foils/films exhibit inhomogeneous www.advmat.de www.advancedsciencenews.com Investigation of crystallinity using XRD and X-ray rocking curve (XRC) measurements.

[145] Ni Atmospheric pressure 400-700 °C Borazine, anneal film at 1000 °C for 1 h after growth ~5-50 nm thick h-BN film. [146] Ni (111) Deposition pressure 10

-7 mbar 750 °C Borazine, 10 min Monolayer h-BN film. [147] Ni 10 Torr 800 °C Ammonia-borane, 120-125 °C 0-18 nm thick h-BN films exhibiting facet dependent thickness. [103] Polycrystalline Ni, Cu foils Low pressure. ≈135 mTorr 1025 °C Ammonia, diborane h-BN film of between 1 to ~100 layers on Ni. [148] Polycrystalline Pt, Ni, Cu foils <3 mTorr borazine exposure 950 °C Borazine Mono-to few-layer h-BN films. [149] Polycrystalline and single-crystalline Ni Base pressure ≈10 -6 Torr, 0.7 Torr during growth 1000 °C 10 sccm diborane, 30 sccm ammonia Multilayer h-BN films exhibiting facet dependent thickness. [150] Curved Ni(111) Growth pressure 2 × 10 -7 mbar 750 °C Borazine h-BN monolayer film formed and was used to investigate oxygen intercalation. [151] Ni(111) 30 mbar 1000 °C Ammonia, TEB: V/III ratio 23500, alternating B and N flow Multilayer h-BN formed on patterned Ni(111) on sapphire. [152] Ni(111) UHV 800 °C Borazine, 5 × 10 -8 Torr Epitaxial monolayer h-BN islands formed on clean Ni(111), while epitaxial and non-epitaxial h-BN islands are formed on Ni(111) containing near-surface B atoms. [47] Polycrystalline Ni foil 60 mTorr base pressure, 300 mTorr during growth 1000 °C Ammonia borane, 40-90 °C 2-10 layered h-BN films. [153] Polycrystalline Ni, Cu foils Atmospheric pressure 1000 °C Decaborane (70-72 °C), ammonia 2-6 layered h-BN films. Adv. Mater. 2023, 35, 2207374 15214095, 2023, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adma.202207374 by Vanderbilt University Medical, Wiley Online Library on [06/07/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License thickness due to combination of precursor reactivity, catalytic activity toward h-BN formation and different rates of etching of the formed h-BN on different crystal catalysts (Figure 4C). Despite these limitations, experimental results point toward uniform growth within each crystal facet in the polycrystalline Ni film/foil. [145] Further, the high catalytic activity of Ni and close lattice match of Ni( 111) and the h-BN crystal strongly facilitates growth and results in a close commensurate relationship with the substrate. While high quality h-BN can be formed on single crystal Ni, the difficulty in producing large-area single crystalline Ni (111) substrates as well as associated costs and highly facet dependent growth on polycrystalline Ni foils makes Ni a non-trivial substrate to scale h-BN growth.

h-BN growth on Cu also initially focused on single crystals, that is, Cu (111). For example, Preobrajenski et al., [133] in 2005 studied h-BN (a = 2.50 Å), growth on Cu (111) as a comparison against Ni (111) substrates. Specifically, the researchers aimed to elucidate the interaction between the h-BN and the substrate and probe whether the Ni 3d orbitals hybridize with the h-BN π orbitals. Cu (111) has a similar lattice constant (a = 2.56 Å) to Ni(111) (a = 2.49 Å) but very different 3d electron configurations. Using a combination of in situ near-edge X-ray absorption fine structure (NEXAFS) and XPS after growth, these studies showed that the band structure of the monolayer h-BN as-grown on Cu (111) was similar to bulk h-BN, while the band structure of monolayer h-BN as-grown on Ni(111) was significantly altered. The results suggested that the interaction of the as-grown monolayer h-BN with the Ni catalyst is perhaps stronger than monolayer h-BN grown on Cu(111) (although retaining some form of commensurate relationship). Notably, growth on Cu required an order of magnitude greater precursor flux for h-BN growth, indicating differences in catalytic activity between Cu (111) and Ni (111), wherein the more energetically favorable hybridization between Ni 3d states and the h-BN π-orbitals have been suggested to play a significant catalytic role in h-BN synthesis. [133] Despite the lower catalytic activity of Cu, the weaker catalyst/h-BN is possibly advantageous for h-BN synthesis on polycrystalline Cu since growth is observed to be less facet dependent than on polycrystalline Ni.

In 2010, soon after it was discovered that single layer graphene could be grown on polycrystalline copper, studies found h-BN could similarly be produced using a polycrystalline Cu catalyst (Figure 5A) without the heavy facet dependence that is observed on Ni. [20,71,147,[163][164][165][166][167][168] Further studies showed that atmospheric pressure growth on commercially available polycrystalline Cu foils resulted in few-layered h-BN films (see Table 4 and Figure 5), [71,[166][167][168][169] while low-pressure CVD predominantly resulted in monolayer h-BN growth. [163][164][165]169] These early studies speculated monolayer h-BN growth on Cu foils forms via a self-limiting mechanism analogous to graphene growth on Cu, despite the formation of few-layered h-BN films at atmospheric pressure. [71,[163][164][165][166][167][168]170] Insights into the growth mechanisms came from complementary in situ XPS and XRD observations made during h-BN CVD on polycrystalline Cu [78] (Figure 5B,C) despite the complications of Cu sublimation at low pressure and the high temperature conditions typically used for h-BN CVD. The in situ XPS results show the development of B1s and N1s peaks at 1000 °C upon exposure to borazine indicative of isothermal h-BN growth (Figure 5C). [78] In situ XRD evidenced an increase in lattice constant for Cu during exposure to borazine while controls with exposure to NH 3 at 1000 °C did not show a change in lattice constant indicating negligible N solubility and preferential B incorporation into the Cu bulk. [78] These observations were in agreement with solubility values of B in Cu ≈0.29 at% at 1000 °C, [195] while N exhibits almost no solubility in pure Cu. [196] These observations indicated that h-BN CVD on Cu proceeds isothermally via dissociation of the borazine precursor (with a stoichiometric balance of B and N) impinging on the Cu catalyst (Ji), followed by nucleation and growth of h-BN (J B' :J N' = 1) on the surface, while some B incorporates into the Cu bulk (J B ) and N desorbs (J N ) from the surface (Figure 5D). [78] Hence, the stoichiometric balance (B:N = 1) in the precursor is not preserved on the catalyst surface as h-BN forms and it is

Reference Substrate Pressure regime Substrate temperature Precursor Results [154] Polycrystalline Ni foil 6 mTorr 1100 °C Borazine Multilayer film. A higher growth rate was observed on Ni(110) or Ni(100) facets. [155] Ni(110) UHV 1000 K Borazine, 70 L Monolayer film. Two orientations observed, at 90° rotation with respect to each other. [156] Ni(111) UHV (≈1 × 10 -8 Pa base pressure) ≥ 600 °C Borazine, >100 L for 1 ML Epitaxial monolayer film analyzed using LEED. [157] Ni(111) 3 × 10 -2 Pa, H 2 environment 850-1150 °C Pure h-BN target, Ion beam (Ar) sputtering deposition Monolayer film and islands. [158] Polycrystalline Ni foils 0.3 mbar 1000 °C Borazine Multilayer single crystalline film on Ni(111)-like grains. ABC stacking with AB stacking faults observed. [159] Single crsytalline and polycrystalline Ni (carburized) UHV 880 °C Ammonia, B 2 O 3 Monolayer h-BN film. [160] Ni(111) foil LPCVD 1020-1320 °C Borazine, 0.025-0.1 sccm Epitaxial few-layer h-BN film, 2-5 layers.

Table 3. Continued. Adv. Mater. 2023, 35, 2207374 Adapted with permission. [132] Copyright 2003, Elsevier. B) LEEM time lapse of h-BN domain growing on Ni, indicating an isothermal growth process. Adapted with permission. [152] Copyright 2019, Wiley. C) (Left) Optical image and thickness of h-BN grown on polycrystalline Ni and transferred to SiO 2 showing highly facet dependent h-BN film thickness. Adapted with permission. [147] Copyright 2012, RSC. In a seprate study, Raman spectra and accompanying cross-sectional TEM images of h-BN grown on various Ni crystal facets is presented. The strongest E 2g signal is reported for the Ni(110) crystal face.

The facet dependent growth make polycrystalline Ni foil a non-trivial substrate for uniform scalable growth. Adapted with permission. [149] Copyright 2018, ACS. D) Growth mechanism of few-layered h-BN on Ni (111). It is suggested that absorbed B during the growth process results in a thick (≈200 nm) Ni 23 B 6 layer on the Ni(111) surface, underneath the grown h-BN film. AFM confirmed a 1.2 nm film step height, correlating to approximately three layers. Adapted with permission. [160] Copyright 2022, Springer Nature.

Adv. Mater. 2023, 35, 2207374 15214095, 2023, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adma.202207374 by Vanderbilt University Medical, Wiley Online Library on [06/07/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License © 2022 Wiley-VCH GmbH 2207374 (12 of 38) www.advmat.de www.advancedsciencenews.com Table 4. h-BN growth parameters on Cu. Reference Substrate Pressure regime Substrate temperature Precursor Results [133] Ni(111), Cu(111) UHV, Base pressure 1 × 10 -10 mbar 800 °C for Ni, 750 °C for Cu Borazine, 100L for Ni(111) and 2000L for Cu(111) Monolayer films with stronger chemisoption on Ni(111) and weaker chemisorption on Cu(111). [166] Polycrystalline Cu foil 1000 °C Ammonia-borane 120-130 °C 2-5 layer h-BN film [168] Polycrystalline Cu foil Atmospheric pressure 1000 °C Ammonia-borane, 110-130 °C 6-8 layer h-BN film [171] Cu foil ≈200 mTorr 900-1000 °C Ammonia-borane (45-120 °C), methane h-BNC hybrid multilayer film. [163] Cu foil 350 mTorr 1000 °C Ammonia-borane, 60-90 °C Monolayer film and islands. [167] Cu foil Atmospheric pressure 750 °C during growth, annealed after growth at 1000 °C Borazine 2 to ~20 nm thick film. [172] Cu foil ~30 mTorr 1050 °C Borazine, varying partial pressure 0-20 mTorr Self-limiting film to 10 nm thickness below ≈17 mTorr borazine; Film thickness reaching 91 nm at higher borazine partial pressures. [164] Cu foil ≈90 Pa 1000 °C Ammonia-borane, 120 °C (10-200 mg) 1-4 layer thick film [173] Cu foil Atmospheric pressure 1000-1030 °C Ammonia-borane, ≈60-70 °C, 4 mg compressed pellet Monolayer film templated by graphene edges. [165] Polycrystalline Cu foil 30-40 Pa 1000 °C Ammonia-borane, 90-100 °C Monolayer film. [148] Pt, Ni, Cu foils <3 mTorr 950 °C Borazine Mono-to few-layer films. [174] Cu foil 400 mTorr 1050 °C Ammonia-borane, 130 °C Monolayer film [78] Cu foil UHV ≈950-1000 °C Borazine, 1 × 10 -4 to 5 × 10 -3 mbar Monolayer films studied with systematic in situ XRD and XPS. [175] Cu foil Atmospheric pressure 1050 °C Ammonia-borane, ≈120 °C Monolayer film templated by graphene edges. [176] Cu foil Base pressure 2 × 10 -5 Pa, 3 × 10 -5 Pa during growth 600-1050 °C Pure h-BN source, ion beam sputtering deposition, 1 keV 0.3 mA × cm 2 Ar + source 0.7 -2.4 nm thick films. [71] Cu foil Atmospheric pressure 950-1050 °C Ammonia-borane, 50-80 °C, time varies Monolayer films. [177] Cu foil Atmospheric pressure 1050 °C Ammonia-borane, 5 mg, 60 °C Monolayer films and hexagon-shaped islands. [178] Cu foil Atmospheric pressure 965-1065 °C Ammonia-borane, 2-3 mg Monolayer films and islands whose shape is observed to be dependent on the distance of the Cu substrate from precursor source. [179] Cu (solid and molten Cu on W) Atmospheric pressure 1000 °C solid and 1100 °C liquid Ammonia-borane, 110 °C Mono-and few-layered films (1-10 layers). [9] Cu foil Low pressure 1000 °C Ammonia-borane, 55-120 °C Monolayer film. [180] Cu foil 30-40 Pa 1000 °C Ammonia-borane, 50, 70, 90, and 110 °C Monolayer film and islands whose shapes evolved during etching. [181,182] Resolidified Cu on W Atmospheric pressure 1075 °C Ammonia-borane, 85 °C Monolayer film, symmetrically aligned islands, and spiral shaped few-layered islands. [183] Polycrystalline Cu foil UHV, Base pressure < 10 -7 mbar 1000 °C Ammonia-borane, 15 mg Monolayer islands epitaxially aligned on Cu(100), Cu(110) and Cu(111) facets. [184] Cu foil Low pressure 1050 °C Ammonia-borane, 100 °C Monolayer film. [67] Cu foil Atmospheric pressure 950-1075 °C Ammonia-borane, 80-100 °C Monolayer film and islands. [185] Cu foil ≈0.1 Torr 1050 °C Ammonia-borane, 100 mg, ≈70 °C 1-3 layer islands and films. AA', AB, and 30° twisted stacking configurations observed. [186] Polycrystalline Cu foil Atmospheric pressure 1030 °C Ammonia-borane, 60-90 °C Monolayer film.

this concentration on the surface that governs the nucleation and growth of h-BN. [78] Additionally, precipitation of the B from Cu after growth upon cooling is observed (inset in Figure 5C). These observations also indicate that once the monolayer h-BN is formed, the formation of successive h-BN layers require the transport of precursor through defects in the first layer to the Cu surface, since only B is dissolved in the Cu bulk, while N is not. [78] Maintaining a uniform supply of B:N = 1 to the Cu surface via randomly formed defects after the formation of monolayer of h-BN on Cu is non-trivial. [78] This challenge is generic to all other catalysts and epitomizes the challenges for uniform multi-layer h-BN synthesis on a catalytic surface and is highly relevant to the synthesis of other 2D materials with more than one constituent element. [78] Additionally, the in situ XPS also showed that as-grown h-BN is initially coupled to the Cu surface, [78] however, post-growth ambient exposure results in oxygen intercalation between the Cu and h-BN effectively de-coupling the h-BN from the Cu, similar to observations for graphene grown on Cu foils. [197] Such oxygen intercalation at the h-BN/Cu interface indicates relatively weak interactions between h-BN and Cu. Although, annealing in vacuum allows for deintercalation of the oxygen and restores coupling, heating in the presence of oxygen causes catalytic degradation of the h-BN. [78] Interestingly, the precursor flux, bulk B solubility in Cu and the interaction between h-BN and Cu impacts the shape of h-BN domains forming the monolayer film. For example, Kim et al., [163] reported the formation of triangular h-BN monolayer domains via CVD on Cu by subliming ammoniaborane at 60 °C with a shift toward more irregular shaped domains with an increase in sublimation temperature, that is, increased precursor delivery to the Cu surface. Meanwhile Wang et al., [180] reported triangular monolayer h-BN domains on Cu via CVD by subliming ammonia-borane at 50-110 °C and found the triangle edge changed from convex to concave with increasing sublimation temperature. Babenko et al., [194] provided insights on varying h-BN domain shapes with varying precursor sublimation temperature (Figure 5E) using mass spectrometry to observe the decomposition products of ammonia-borane precursors and related it to the supply of B and N to the Cu surface. The mass spectrometry analysis showed a mixture of varying amounts of ammonia, amino-borane, amino-diborane, borane, di-borane, and tri-borane with only minor compositions of borazine (B:N = 1) [194,198] upon subliming ammonia-borane below 130 °C (note most CVD processes sublime ammonia-borane ≈55-120 °C). [194] Further, the ratio of B to N in the gaseous products, and the total mass flow varied significantly at different decomposition temperatures and over time since heating of the precursor begins. [194] Further, the synthesis of hexagonal shaped monolayer h-BN domains on electropolished Cu foil have also been reported, [177] while others report suggest triangular h-BN domains on electropolished copper. [164,183] Stehle et al., [178] found the triangular h-BN domains transition to hexagonal shapes when the precursor ratio of B to N is increased (Figure 6A). While some of these variations can be attributed to differences in process conditions, the following generic theme emerges based on insights from observations reported by Stehle et al., [178] and Kidambi et al., [78] : h-BN domains with N terminated edges form as a result of a complex interplay between process kinetic and thermodynamics effects. [163] The N terminated edges are thermodynamically more stable (N terminated edges have been observed while etching holes in the h-BN lattice under electron beam [199] ), yet the rate of B and N addition can lead to kinetic control of the domain edge composition via: i) precursor supply (temperature and time duration of sublimation of precursor) and/or impingement flux and subsequent dissociation of the precursor; ii) mobility of the dissociated precursor on the Cu surface which typically increases with increasing temperature and attachment to the growing h-BN domain; iii) preferential solubility of B in the Cu bulk and desorption rate of N from the surface, both of which increase with temperature; and iv) etching by other gases such as H 2 or trace contaminants such as water and/or oxygen present in the reaction environment. On catalyst substrates that show high degree of interaction with the growing h-BN (i.e., Ni) thermodynamics is expected to play a dominant role, but on a less interactive substrates, such as Cu, the growth process is more susceptible to kinetic control. h-BN domains grown on Cu typically show little orientational alignment with the underlying Cu crystallographic facets leading to polycrystalline h-BN films with h-BN domain sizes depending on nucleation density

Reference Substrate Pressure regime Substrate temperature Precursor Results [187] Cu foil 0.5 and 760 Torr 1000 °C Ammonia-Borane, 10-200 mg, 50-130 °C 1.5-10.3 nm film. [188] Polycrystalline Cu foil Low-pressure 1050 °C Ammonia-borane, 2-3 mg, 75-85 °C Monolayer islands exhibiting facet-dependent orientation. [189] Cu(110) foil ≈200 Pa 1035 °C Ammonia-borane, 65 °C Epitaxial monolayer film. [190] Cu(111) UHV, Base pressure 4 × 10 -10 Torr 1000-1100 K Borazine, 0.5-4.5 × 10 -7 Torr partial pressure Dendritic monolayer triangular domains studied using in situ LEEM and LEED. [191] Cu foil 20 mTorr air and 10 sccm hydrogen 1030 °C Ammonia-borane, 80 °C ~3 nm thick film [192] Cu foil Low-pressure 1050 °C Borazine, diluted in N 2 , H 2 , Ar Monolayer film [193] Cu foil Atmospheric pressure 1020 °C Ammonia-borane, 8.5 mg, 80-100 °C ~3 nm thick film

Table 4. Continued. Left: Adapted with permission. [163] Copyright 2012, ACS. Right: Adapted with permission. [166] Copyright 2010, ACS. B) In situ XRD of Cu under various gaseous environments. A lattice expansion in Cu is observed under a borazine environment, but not under an ammonia environment, indicating that B is absorbed into the metal bulk, but N is not. Adapted with permission. [78] Copyright 2014, ACS. C) (Left) SEM image of Cu catalyst surface after various growth times. Small monolayer domains grown and merge to form a complete polycrystalline film. (Right) In situ XPS during h-BN growth on polycrystalline Cu where the development of B1s and N1s peaks during growth indicates isothermal growth of h-BN. Lack of h-BN growth during cooling is in agreement with the differing solubilities of B and N into Cu that make N unavailable to precipitate from the bulk on cooling. Adapted with permission. [78] Copyright, 2014, ACS. D) Proposed mechanism for h-BN growth based on in situ XRD and XPS. Adapted with permission. [78] Copyright 2014, ACS. E) Mass spectroscopy results of the decomposition products of ammonia-borane at 100 °C. Thermal decomposition of ammonia-borane leads to a variety of gaseous B and N containing species.

The ratios of such species vary with decomposition temperature and time from initial heating. Adapted with permission. [194] Copyright 2017, Springer Nature.

(see SEM of h-BN triangular domains with many different orientations within a single Cu grain in Figure 5C). Despite the polycrystalline nature of the grown films, the kinetic control coupled with the h-BN growth mechanisms on Cu (low N solubility in bulk Cu, Figure 5B) necessitating transport of the precursor through defects in the first layer to the Cu surface for dissociation to provide B:N = 1 for growth of additional layers enables uniform monolayer h-BN film growth across the different Cu crystal facets in polycrystalline Cu foils (Figure 5C). This is in sharp contrast to h-BN grown on Ni where the h-BN film thickness is highly dependent Ni crystal facet (see Figure 4C).

Although, h-BN is an insulator and the number of grain boundaries should in-principle not affect electronic applications, some applications such as corrosion barriers, [201,202] tunnel barriers, [3,32] and membranes [203] require minimal defects. Wang et al. [188] systematically mapped the orientation preferentiality of h-BN domains on various Cu crystal facets and found that while most Cu facets contained multiple h-BN alignments, some Cu facets exhibited h-BN oriented in a single direction (Figure 6B,C). Growth on single crystalline Cu (110) and Cu (111) exhibited high degree of h-BN orientational alignment (Figure 6D,E) and the merging of such aligned . Adapted with permission. [163] Copyright 2012, ACS. The differences in edge stabilities appear to influence the morphology of growing h-BN domains (separate study, bottom). Adapted with permission. [178] Copyright 2015, ACS. B) Effects of Cu crystal facets on h-BN growth. Cu substrate crystallography is observed to influence nucleation density, growth rate, and orientation preferentiality suggesting that different facets have different nucleation thresholds and different catalytic characteristics. Adapted with permission. [188] Copyright 2019, Wiley. C) EBSD map and accompanying SEM of h-BN islands on polycrystalline Cu show differences in h-BN growth. Adapted with permission. [9] Copyright 2015, Springer Nature. The Cu crystallography can be used to align h-BN domains. This is demonstrated on Cu(111) (D) and Cu(110) (E). D) Wafer scale Cu( 111) is synthesized by annealing a thin Cu film on c-plane sapphire, while E) (separate study) Cu (110) foil is synthesized using a Cu(110) seed crystal placed on a polycrystalline foil, followed by long annealing times. D) Adapted with permission. [200] Copyright 2020, Springer Nature. E) Adapted with permission. [189] Copyright 2019, Springer Nature.

domains could in-principle allow for single crystalline monolayer films. [164,181,[188][189] In particular, aligned monolayer h-BN domains (99% alignment with only 1% un-aligned) have been grown on the Cu(110) which merge into predominantly single crystalline h-BN film. [189] Notably, 10 × 10 cm 2 Cu (110) foil was prepared by seeding a large Cu(110) crystal on to a poly crystalline Cu foil and using the propensity of the larger crystal to grow by consuming smaller crystals (analogous to Oswald ripening in liquid phase crystal growth). [189,204] However, the process is time intensive and even the 1% of evenly distributed un-aligned h-BN domains could result in a polycrystalline film. [189] Another approach to grow wafer scale single crystalline h-BN monolayer leverages a thin film of Cu deposited on sapphire wafers and annealed to form Cu (111) substrates. (Figure 6D). [200] A ≈500 nm layer of Cu was sputter deposited onto a sapphire wafer and annealed under hydrogen to convert it into a single crystalline Cu (111) surface. [200] Single crystalline h-BN monolayer was grown from multiple aligned domains on the Cu (111) surface that are all oriented in the same direction, suggesting that some growth effects observed on a single crystal substrate under UHV conditions can potentially be translated to manufacturing compatible low-pressure CVD. [200] The choice of the 500 nm thick layer was attributed to maintaining adequate Cu film thickness despite the sublimation of Cu during CVD conditions (low pressure and high temperatures) as well as allowing commensurate Cu (111) formation with the sapphire. [200] Finally, Cu sublimation during low pressure CVD is detrimental to scalable manufacturing, [78,197] but can be mitigated by transitioning to higher pressures for deposition with dilute precursors and further research on these aspects is necessary. [197] Nonetheless, the ability to grow wafer scale single crystalline h-BN on Cu (111) over sapphire represents a major opportunity for scalable single crystalline h-BN synthesis.

Using Cu catalysts for h-BN growth is advantageous because single layer h-BN can be grown on a polycrystalline Cu foil. While the resulting h-BN film will be polycrystalline under such conditions, the wide availability of polycrystalline Cu foils makes it an inexpensive option. We note that roll-to-roll manufacturing techniques developed for growing and transferring graphene from Cu can theoretically be used to grow and transfer h-BN from Cu. [74][75][76] Finally, single crystalline Cu also aids the formation of monolayer h-BN films (by merging of many similarly oriented h-BN domains) with predominantly single crystalline orientation. [189,200]

h-BN growth on Pt was initially reported in the early 1990s under UHV conditions on Pt(111) single crystals. [205][206][207][208] STM imaging of h-BN grown on Pt(111) revealed a highly regular moiré pattern indicating that the h-BN grown exhibited an orientational preference with respect to the Pt(111) surface (Figure 7A), despite the large difference in lattice constants between Pt(111) and h-BN (a = 0.277 nm vs a = 0.250 nm, respectively. ≈11% difference). [205][206][207][208] Particularly, two different configurations of h-BN on Pt (111) are seen: i) where the B and N alternate on Pt(111) top and bridge sites; and ii) where N sits on top sites and B sits on fcc or hcp hollow sites (Figure 7A). [208] To further investigate the catalyst/h-BN relationship, Nagashima et al., [131,135] used angle resolved photoemission spectroscopy (ARPES) to study the electronic structure of the h-BN on Pt (111) and found that the electronic structure was un-affected by the Pt substrate, indicating that the h-BN film was similar to an independent crystal physisorbed to the Pt surface. STM and XPS also confirmed the h-BN grown on Pt (111) is atomically flat and chemically detached from Pt (111) surface indicating a weak interaction between h-BN and Pt (111). [129,214,215] In situ low energy electron microscopy (LEEM) observations confirmed isothermal growth of h-BN on Pt (111) and alignment of the growing h-BN domains, even though there is relatively little chemical interaction between the grown h-BN film and the underlying substrate (Figure 7C). [214,216,210] Hemmi et al., also showed a strong dependence of h-BN epitaxy on Pt (111) with growth temperature during wafer-scale synthesis. [217] It is also found that h-BN grown on Pt(110) may cause Pt atoms to shift underneath the h-BN layer to accommodate strain, suggesting that despite the weak Pt/h-BN interaction suggested from XPS and ARPES, the strain at the h-BN/ Pt interface can cause changes at the atomic scale on the Pt surface (Figure 7B). [209] In 2013 two groups demonstrated h-BN synthesis on polycrystalline Pt foils via low pressure CVD as opposed to UHV conditions (see Table 5). Kim et al., [218] observed entirely monolayer h-BN growth while Gao et al., [211] observed monolayer, bilayer, and few layer h-BN growth depending on the temperature at which they sublimed the ammonia-borane precursor. For example, when the precursor was sublimed at 70 °C, monolayer domains were observed that coalesced into a single monolayer. Longer exposure times albeit at 75 °C, bilayer domains were observed (Figure 7D). [211] The observed h-BN domains were triangular with smoothed and rounded edges that the authors suggest has to do with the large lattice mismatch (≈11%) between h-BN and platinum, [211] although rounded corners on h-BN domains grown on polycrystalline Cu foils have been shown to occur as the ratio of B:N increases on the catalyst surface. [178] Interestingly, these studies demonstrated electrochemical delamination of the h-BN from Pt foil and subsequent re-use of the Pt foil for h-BN growth. In principle these methods potentially allow reuse of the expensive Pt foil indefinitely for h-BN growth, representing significant advantages for scalable h-BN synthesis. [218,211] h-BN films on Pt catalyst foils have been observed to have fewer wrinkles as compared to h-BN films grown on Ni and Cu foils which has been attributed to the relatively weak substrate interaction between h-BN film and Pt, and the relatively smaller difference in thermal expansion coefficients compared to Cu and Ni. [213] The density of wrinkles in the h-BN film was dependent on the cooling rate after growth, wherein slower cooling rates resulted in fewer wrinkles. [213] Here, the loosening of h-BN/Pt interface interactions during gradual cooling is suggested to allow the Pt to contract and h-BN to expand with lower strain at the interface (Figure 7F). [213] Significant insights into Pt-catalyzed h-BN growth came from studies by Wang et al. [212] Using in situ XPS they observed the evolution of B1s and N1s core level signatures during h-BN growth at temperature, confirming isothermal growth of h-BN on Pt foils. Further, they reported on a facile method to transform the polycrystalline Pt foil into a single Pt(111) crystal via annealing while exposing it to borazine (Figure 8). The exact mechanism of this transformation remains unclear, however it is suggested that the accelerated Pt recrystallization is a result of B absorption into Pt grain boundaries and subse-quent grain boundary unpinning via deoxidizing or removal of pre-existing solutes. [212] Subsequently, monolayer h-BN films were grown using borazine exposure, resulting in nucleation of h-BN domains that predominantly aligned themselves with the underlying Pt substrate. However, the h-BN domains were not (111). Adapted with permission. [208] Copyright 2005, ACS. B) On Pt (110) the Moiré pattern is also manifested in a 1D pattern perpendicular to steps on the 110 surface. STM and DFT simulated model of h-BN on Pt(110) surface. Adapted with permission. [209] Copyright 2019, ACS. C) In situ LEEM of h-BN growth on Pt. Adapted with permission. [210] Copyright 2008, Elsevier. D) AFM maps revealing monolayer island and bilayer islands with a higher precursor flux. Reproduced with permission. [211] Copyright 2013, ACS. E) Schematic of growth and dry transfer of h-BN off of Pt surface, demonstrating feasibility of reusing expensive Pt catalyst. Adapted with permission. [212] Copyright 2019, ACS. F) Proposed model for wrinkle reduction when the growth substrate is cooled slowly. Adapted with permission. [213] Copyright 2014, ACS.

www.advmat.de www.advancedsciencenews.com Figure 8. h-BN growth on Pt. Schematic of CVD growth process. Precursor is flowed into the chamber first, supplying boron which quickly facilitates recrystallization of the Pt surface, and then precursor is flowed a second time to grow h-BN. Associated SEM images of h-BN growth using the process described above. Adapted with permission. [212] Copyright 2019, ACS.

UHV, Base pressure: ≈1 × 10 -8 Pa 700-800 °C Borazine Monolayer film. Electronic properties studied using XPS, ARUPS and ARSEES. [135] Pt (111), Pd (111), Ni

UHV, Base pressure:

8 Pa 700-800 °C Borazine Monolayer film. Electronic properties studied using XPS, ARUPS, and EELS. [208] Pt(111) UHV, Base pressure: 10 -6 mbar 1000K B-trichloroborazine (ClBNH) 3 Monolayer film studied with LEED. Two different configurations are observed. [211] Pt foil Atmospheric pressure 1000 °C Ammonia-borane, 70-80 °C Monolayer, bilayer and few-layer films. [218] Pt foil 0.1 Torr 1100 °C Ammonia-borane, 130 °C Monolayer film. [148] Pt, Ni, Cu foils <3 mTorr 950 °C Borazine Mono-to few-layered film. [213] Pt foil 100 mTorr, 500 mTorr 1100 °C Borazine Monolayer film formed at 0.1 Torr while thicker multilayer films formed at 0.5 Torr on Pt(111) facets and thinner multilayer films on Pt(001) facets. [212] Pt foil UHV, Base pressure: <2 × 10 -6 mbar 1200 °C Borazine, 1 × 10 -5 mbar during seeding, and 2.5 × 10 -6 mbar during growth Monolayer film.

[209] Pt(110) UHV, Base pressure ≈5 × 10 -11 mbar 1020 and 1120K Borazine Epitaxial monolayer film. [219] Pt(110) UHV, base pressure ≈3 × 10 -10 mbar 1000 K Ammonia-borane, 40 °C, partial pressure of 10 -6 mbar for 30 min required for full coverage Epitaxial monolayer film. all aligned as observed under UHV conditions on single-crystalline Pt. This can possibly be attributed to the larger impingement flux under LPCVD conditions resulting in process kinetic dominating (compared to thermodynamics) and driving nucleation with poor alignment. Notably, when borazine exposure is removed at growth temperature, any h-BN grown quickly disappears via etching. Regardless, a relatively high-quality monolayer of h-BN is formed as the domains coalesce. Despite some success in growing h-BN on Pt, [220] the high price of Pt is a significant hurdle in scalable production. Electrochemical delamination of the h-BN from Pt foil and subsequent reuse of the Pt foil for h-BN growth presents an exciting solution to this problem. In principle, such methods potentially allow for reuse of the expensive Pt foil indefinitely for h-BN growth. [218,211] Here, Wang et al. [212] also demonstrated a facile method to transfer the h-BN film from the Pt surface by leveraging the weak h-BN/Pt interactions to remove the 2D material without electrochemical evolution of H 2 gas at the Pt/h-BN interface or etching the Pt. Specifically, oxygen intercalation between the as-grown h-BN film via ≈5 h of ambient exposure was followed by casting polyvinyl acetate (PVA) and peeling of the PVA/h-BN stack from the Pt surface. The h-BN can then be stamped onto a desired substrate or used to pick up another 2D material for building vertical heterostructures and the Pt catalyst reused (Figure 7E). Hence, the ability to grow high-quality, uniform monolayer h-BN on Pt foils and its facile removal to allow for reuse of the Pt presents an opportunity for scalable, reliable, and reproducible synthesis of h-BN monolayers.

h-BN growth on iron was first reported in 1979, [221] wherein a thick BN film of in-homogeneous crystallinity was grown on carbon-doped (0.3 at% C) iron (see Table 6). Here we focus on h-BN grown on pure Fe and discuss alloys further below.

In 2012, Vinogradov et al., demonstrated that h-BN grown on Fe (110) surface exhibited a periodic, wave-like corrugated structure (Figure 9A and Table 6). [222] The wave like structure with a periodicity of ≈2.6 nm and an out of plane amplitude of ≈0.8 Å was attributed to minimization of interfacial energy between the h-BN and the Fe (110), that is, relaxation of in-plane strain due to lattice mismatch in the [111] and [111] Fe crystal directions. [222] Subsequent research of growth on iron has focused on polycrystalline Fe foils and films.

Fe exhibits significant solubility of boron and nitrogen within the bulk, with profound implications for h-BN growth. It also undergoes a significant phase change from an α-phase (BCC) to a γ-phase (FCC) at ≈911 °C, [228] adding to the complicated dynamics of how the transformation will affect facet dependent relations and newly grown h-BN layers upon cooling. [79] There are some differences in reported B and N solubility values in Fe at growth temperatures, for example, Babenko et al., [227] report solubility of B ≈ 0.11 at% [229] and N ≈ 0.10 at% in γ-Fe at 950 °C, [230,231] Caneva et al. [79] report nitrogen uptake ≈0.6 at% based on in situ XRD when annealing Fe in NH 3 at ≈900 °C (XRD confirmed γ-phase Fe), and Kim et al., [223] note that maximum N solubility in Fe at 1000 °C reaches ≈8 at%. Despite such differences in reported literature values of solubility, both B and N are readily absorbed into the Fe bulk metal (contrary to, for example, Cu and Ni where N solubility is lower than B) and this adds significant complexity to h-BN growth, that is, the bulk diffusivity and precipitation of B and N must also be considered for h-BN growth. For example, B and N precipitation from the bulk can facilitate h-BN formation underneath existing h-BN layers, often during cooling after CVD resulting in multilayer h-BN formation. Multilayer h-BN domains on Fe typically exhibit inverted pyramid-type structures wherein surface growth likely forms topmost layer in the pyramid (farthest away from the Fe substrate) and bulk diffusion/precipitation of B and N enables the formation of the other layers below (Figure 9C). [79,223] As a further indication of bulk absobtion/diffusion playing a role in h-BN growth on Fe, the rate of cooling after CVD was found to significantly influence the morphology of multilayer h-BN grown on polycrystalline Fe foil, particularly at atmospheric pressure. [79,223] For example, rapid cooling resulted in secondary nuclei (small triangles in web-like patterns) [223] creating an inhomogeneous h-BN film. Slow cooling on the other hand (≈5 °C min -1 ), resulted in a relatively smooth multilayer h-BN film [223] , that is, the multilayer h-BN film has relatively uniform thickness over regions of ≈20-60 µm (Figure 9B). [223] Interestingly, Bayer et al., [225] showed that the h-BN domains on Fe in some cases do not merge by bonding covalently in the 2D plane with a domain boundary as seen is the case with other catalyst systems (Cu, Pt, etc.). Instead, the growing h-BN domains on Fe overlap each other as seen from the moiré pattern formed via overlapping h-BN layers of different crystalline orientation (Figure 9F). Further, they suggested the width of the overlapping region is correlated to the amount of dissolved B and N into the Fe bulk indicating the role of precipitation upon cooling. [225] One approach to mitigate bulk effects and form predominantly monolayer h-BN on Fe is via pre-filling the Fe bulk with N, limiting further B solubility, and thereby mitigating the effects of precipitation on h-BN growth and confining the growth dynamics to the Fe surface. [79] These studies observed primarily multilayer pyramidal discontinuous h-BN domains when the Fe catalyst foil/film was pre-cleaned via annealing in H 2 and in situ XRD, indicated B dissolution into the Fe catalyst bulk via the presence of various iron borides and phase transition for Fe catalyst from γ-Fe to α-Fe during borazine exposure at temperature (Figure 9D). However, when the Fe was annealed in NH 3 prior to borazine exposure, they observed predominantly monolayer h-BN domains that merged to result in a continuous monolayer h-BN film. Corresponding in situ observations showed a lattice expansion of γ-Fe indicating N dissolution and no further lattice expansion, phase transformation, or borides present during borazine exposure (Figure 9E). However, if too much nitrogen is dissolved into the Fe bulk, Fe-nitrides are formed and are detrimental to h-BN growth. [79] Babenko et al., developed a slightly different method to control bulk solubility of B and N in Fe. [227] Having observed that multilayer growth occurred primarily along rolling striations in the Fe foil, they showed via energy dispersive X-ray spectroscopy (EDS) that these striations often contained localized carbon impurities. To remove carbon from the surface, they oxidized the Fe foil to effectively reduce the concentration of C at the Fe surface and limit preferential nucleation along the striations in the metal. Next, the Fe was annealed in an NH 3 environment, and briefly exposed to C 2 H 2 before CVD via exposure to borazine. Tuning the amount of C 2 H 2 exposure allowed monolayer h-BN growth via carburization of the Fe foil which decreased the bulk co-solubility of N, and combined with a short NH 3 exposure, the co-solubility of B was minimal. Effectively, they find that a combination of oxidation, carburization, and NH 3 anneal can control the bulk solubility B and N thereby mitigating bulk reservoir effects from the Fe catalyst and form large domains of monolayer h-BN. In contrast to the prior studies [79] they observed that pre-annealing in ammonia was not sufficient to limit multilayer growth. However, we note that differences in growth times, film and foil system, and precursor flow likely could influence the direct comparability of these studies. Regardless, fine control of bulk absorption parameters appears to be essential in growing monolayer h-BN on Fe and Babenko et al., [227] report up to 1.1 mm h-BN domains which represent one of the largest single crystal monolayer h-BN domains on Fe (Figure 9G).

Although large monolayer h-BN domains can be grown on Fe, oxidation of the Fe underneath the h-BN domains has been observed. [226] The oxides typically nucleate at regions underneath defect in h-BN or regions of Fe not covered by h-BN and spread outward at the h-BN/Fe interface wherein a nonself-limiting wet oxidation of Fe propagates in the presence of oxygen and ambient humidity. [226] This presents a challenge for using h-BN as a passivating coating on Fe albeit thermal annealing of the h-BN/Fe structure can reduce the oxide at the interface. [226] The solubility of B and N in the Fe bulk emerges as the central challenge in h-BN synthesis via CVD on Fe. Despite these challenges, Fe is a readily available metal and the large h-BN domains that have been grown, indicating that Fe may be ideal for applications that require large area single crystal domains. However, the overlapping nature of domains of h-BN on Fe may render them unsuitable for applications that require a continuous covalently bonded h-BN monolayer film.

h-BN grown on catalytic substrates with significantly different lattice constants (i.e., rhodium (Rh), ruthenium (Rh), and iridium (Ir)) show distinct corrugated morphology (see Figure 10 and Table 7). [232][233][234][235][236][237][238][239][240][241][242] The corrugation is thought to originate from the large lattice mismatch resulting in strain wherein the substrate causes the h-BN to ruffle into a meshlike pattern (Figure 10A). The corrugated h-BN layer typically consists of regions closely interacting with the substrate, and regions more loosely adhered, that is, a balance between the strong interaction of the h-BN to the substrate and the strain caused by the lattice mismatch. XPS of h-BN on catalyst systems reveal higher energy catalyst/h-BN interactions and more regions of close interaction for various rare earth metals (degree of interaction: Pt(111) < Ir(111) < Rh(111) < Ru(0001)). [214] Stania et al., showed the higher interaction of h-BN to Rh (111) persists even when growing h-BN on PtRh (111), resulting in reduced Pt enrichment under the h-BN and these observations were also supported by DFT calculations. [243] Despite the lattice mismatch, a strong crystallographic orientation preference is observed for h-BN domains with minimal rotational misalignment, for example, h-BN domains on a Ru(0001) surface (Figure 10A), [246] and periodic mesh structure of the h-BN is observed consistently across the 0001) (top middle) with periodicity of a few nanometers caused by a corrugation of the h-BN layer to minimize strain caused by the lattice mismatch between substrates. Adapted with permission. [237] Copyright 2007, ACS. (Separate study, bottom left) Modeled electrostatic potential around the Rh(111)/h-BN interface caused by the corrugated structure. Adapted with permission. [244] Copyright 2011, ACS. (separate study, Right) XPS of h-BN as-grown on various noble metals. Binding energy of nitrogen core levels indicating a shift in the nitrogen peak position of h-BN on Pt (111), Ir (111), Rh (111), and Ru(0001). The shift arises from increasing interaction with the metal surface. Adapted with permission. [214] Copyright 2007, Elsevier. B) Model of h-BN nanomesh structure on Rh(111) (top) and after hydrogen intercalation (bottom). The intercalated hydrogen separates the h-BN from the Rh lattice and allows the h-BN to relax into a uniform morphology. Adapted with permission. [242] Copyright 2010, Wiley. C) h-BN grown on Rh (111) under UHV conditions before (top) and after (bottom) Ar ion treatment and annealing. The post treatment etches and decouples the h-BN areas more closely bound to the Rh (111), peeling, via Ar intercalation, Adv. Mater. 2023, 35, 2207374 15214095, 2023, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adma.202207374 by Vanderbilt University Medical, Wiley Online Library on [06/07/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License © 2022 Wiley-VCH GmbH 2207374 (23 of 38) www.advmat.de www.advancedsciencenews.com Table 7. h-BN growth parameters on noble metals. Reference Substrate Pressure regime Substrate temperature Precursor Results [238] Rh(111) UHV 1070 K Borazine, 3 × 10 -7 mbar for 40 L Two-layer nanomesh film [239] Rh(111) UHV 820 °C Borazine, 5 × 10 -7 mbar for 675 L Monolayer nanomesh film [242] Rh(111) UHV 1070 K Borazine Nanomesh film. Intercalation of atomic hydrogen alleviated corrugation of monolayer [245] Rh(111) UHV 1050 K Borazine Monolayer nanomesh film. Ion bombardment and annealing generated 2 nm vacancies [248] Rh(111) UHV 1070 K Borazine at 3 × 10 -7 mbar partial pressure Nanomesh film studied using STM, LEED and surface XRD. [235] Rh(111) UHV 300-1200 K Borazine, 1.2-2.5 × 10 -9 mbar partial pressure. Nanomesh film formation studied using in situ STM. [240] Ir(111) UHV, base pressure 2 × 10 -10 mbar 170-1150 K Borazine Epitaxial monolayer formed at temperatures >1000 K. [249] Ir(111) UHV 1070 K Borazine, steady exposure at 1070 K, or exposure at RT followed by annealing at 1270 K Monolayer film [241] Ir(111) UHV (1 × 10 -10 mbar base pressure) 1000-1300 K Borazine, 1 × 10 -6 mbar partial pressure Monolayer film. Nucleation and growth occured under CVD conditions, whereas RT borazine exposure followed by annealing resulted in a network formation without nucleation. [250,251] Re(0001) UHV ~650 °C followed by annealing at ~730 °C for some experiments Ammonia-borane, ~150 °C Monolayer h-BN film and h-BN/graphene lateral heterostructure. Grain boundary defects studied using STM [252] Re(0001) UHV 115-1500 K Borazine, ~1-40 L Adsorption, decomposition, and desorption of borazine is studied. [246] Ru(0001) UHV 780 °C Borazine, 7 × 10 -9 1.3 × 10 -8 Torr partial pressure Monolayer film growth studied using in situ LEEM. [234] Ru(0001) UHV (2 × 10 -10 mbar base pressure) 700-1000 K Borazine, 5 × 10 -8 mbar partial pressure Monolayer nanomesh film and islands. [237] Ru(0001) UHV 1100 K Borazine, ~60 L Monolayer nanomesh film [253] Ru(0001) UHV 1030 K Borazine, ~40 L Monolayer nanomesh film. Electronic properties are studied [247] Ru(0001) UHV base pressure 2 × 10 -10 Torr, total pressure 10 -2 Torr RT sputtering and annealing, or growth at 850 °C Boron sputtered by RF magnetron (10 W) in Ar/ N 2 gas mixture. Alternating sputter and anneal used for thick films 1-10 layer film. [254] Ru(0001) UHV 1000 °C Borazine, 1 × 10 -6 Torr partial pressure Monolayer film. [255] Ru(0001) UHV 1030 K Borazine, 6 × 10 -7 mbar partial pressure Epitaxial monolayer film. [256] Ru(0001) UHV 630-700 °C Ammonia-borane, 100 °C Monolayer film growth and subsequent oxygen intercalation studied using in situ LEEM and PEEM.

the more loosely attached region of h-BN from the surface with "nanovoids" ≈2 nm from where more closely bound h-BN remains attached to the Rh (111). Adapted with permission. [245] Copyright 2018, ACS. D) In situ LEEM images of h-BN monolayer growing on Ru(0001) surface revealing the merging of several oblong shaped domains. Adapted with permission. [246] Copyright 2011, ACS. E) Magnetron sputtering setup consisting of a boron target in an Ar and N 2 environment. Uniform multilayer films can be grown by altering deposition and high temperature annealing. Adapted with permission. [247] Copyright 2013, ACS.

substrate surface. Hydrogen intercalation between the h-BN and the substrate has been found to flatten the corrugated h-BN structure, for example, for h-BN on Rh(111), [242] indicating that the corrugation is a consequence of strain from the close interaction and that the hydrogen intercalation alleviates some of the strain (Figure 10B). For more details on band structures and experimental results on the interactions between h-BN and noble metal substrates, we refer the reader to other focused reviews. [66] Interestingly, exposing the corrugated h-BN on Rh (111) to low energy Ar ions followed by annealing at ≈900 K allows for facile etching of nanoscale defects ≈2 nm in regions that were more closely bound to the substrate (Figure 10C) via the "can-opener" effect. The regions more closely adhered to the catalyst surface stay adhered and peel from the rest of the loosely adhered film as it further separates from the surface leading to periodic defects corresponding to the regions that stayed attached to the catalyst surface. [245,257] Such structures have then been used to demonstrate porous membranes for molecular separations. [245] In situ LEEM during h-BN growth on Ru(0001) under UHV conditions (Table 7) shows that h-BN growth proceeds isothermally via nucleation events that grow in time and merge to form a continuous monolayer film (Figure 10D). [246] Further, it has been observed that h-BN growth at UHV conditions on Ru(0001) does not proceed beyond a complete monolayer suggesting the role of strong interaction with the substrate and limited precursor access to the catalytic substrate surface that limits growth of subsequent layers. [246] Sutter et al., [247] proposed a novel process to form uniform multi-layer h-BN on Ru(0001) using a magnetron sputtering system composed of a boron target and nitrogen/Ar feed gas (Figure 10E). [247] Initially, they formed monolayer h-BN via CVD on Ru (0001) with borazine as a precursor. Interestingly, magnetron sputtering of B in a N 2 /Ar atmosphere at 10 -2 Torr at high temperature (≈850 °C) resulted in bilayer h-BN growth on Ru (0001). [247] Next, in a breakthrough discovery they found that alternating between sputtering at room temperature and high temperature annealing at ≈850 °C allowed for uniform multilayer h-BN films on Ru (0001). [247] In each cycle, a thin amorphous BN layer is deposited onto the h-BN surface and then the high-temperature annealing converts the amorphous film to, on average, ≈1 h-BN layer. This transformational advance demonstrated the potential for excellent layer thickness control in few-layered h-BN films. Further, they found that the stacking configuration of the h-BN film can be influenced by the underlying substrate crystal, [247], for example, AA′ stacking was observed for h-BN films when grown on Ru(0001), but ABC stacking, also known as rhombohedral boron nitride (r-BN), was observed for growth on monolayer graphene covering the Ru(0001) surface.

The growth of h-BN films with uniform thickness and controlled stacking configuration represents a transformative advance for their use in electronic applications. [247] Although, this method has been shown to be highly successful on Ru(0001), sputter deposition followed by film annealing approaches to grow multilayer h-BN on more cost-effective catalysts such as Cu or Ni, [176,258] remains work in progress.

In addition to the catalytic substrates described above, monolayer h-BN growth has also been reported on highly magnetic metals such as cobalt (Co). [259] Similar to Ni (111), the growth rate of the h-BN film on Co films was significantly reduced after the first monolayer. Further MBE methods have also been explored for h-BN on Co. [260] Semiconductors such as germanium (Ge) has also been demonstrated to facilitate growth of h-BN. [261] Interestingly, on the Ge (110), triangular h-BN domains were observed to only grow oriented at 0° or 180° with respect to the [0 0 1] crystal direction of the Ge(110) facet. Meanwhile on the Ge(100), the triangular h-BN domains oriented along 0°, 90°, 180°, and 270° with respect to the [1 1 0] crystal direction with equal probability. [261] h-BN has also been grown on Ag (111). [262] However, no orientation preference is observed between the h-BN domains and the Ag (111) substrate, in contrast to several other single crystalline substrates. Based on ab initio calculations, the binding interaction of h-BN and Ag ( 111) is suggested to be too weak to result in the enforcement of any particular h-BN crystal orientation that effectively minimizes interfacial energy. [262] Finally, thick films (several µm) of h-BN have also been grown on Si(110) using LPCVD methods. [263] Quasi-hexagonal Mo(110) has also been used as a catalyst substrate for h-BN synthesis, which resulted in one-dimensional h-BN moirés. [264]

Doping of metallic catalysts films has been shown to effectively control the nucleation density of h-BN domains. [224,265] Controlling the density of nucleation leads to larger h-BN domains and h-BN films with fewer grain boundaries. The nucleation density of h-BN domains on Fe [224] (Figure 11A) and Cu [265] (Figure 11B) has been shown to be controlled by Si doping, that is, the nucleation density decreases with increasing Si doping. The Si impurities serve to locally reduce the nucleation threshold around the impurity and thereby allow for the nucleation of an h-BN crystal at a lower overall density than the pure Fe metal catalyst. [224] In addition to doping, alloying of the catalyst substrates has also been explored a method to control nucleation density and growth morphology of monolayer h-BN. [266][267][268][270][271] Cu-Ni alloys in particular have received considerable interest as catalysts for h-BN growth (Figure 11D). [270,271,267] By alloying Ni into Cu in ratios below ≈30%, Lu et al., [267] and Yang et al., [271] demonstrated that the nucleation density of h-BN could be effectively suppressed by increasing the percentage of Ni in the alloy for low pressure and atmospheric pressure CVD processes, respectively. Additionally, Lu et al., [267] use density functional theory (DFT) simulations to demonstrate that the energy required to decompose H 2 BNH 2 on the alloy surface is reduced by adding more Ni to the Ni/Cu alloy ratio, suggesting that adding more Ni to the alloy increases its catalytic activity. However, it remains unclear why the h-BN domain sizes increases with increasing Ni/Cu ratio up to 20 at% Ni, A) The effects of doping and Fe substrate with Si. Increasing doping of Si reduces the nucleation density up to a point allowing for the growth of large area domains. Adapted with permission. [224] Copyright 2015, ACS. B) SEM images of h-BN domains on Cu with increasing Si doping, phase diagram of Cu with Si doping demonstrating a need to operate at lower temperatures and nucleation density as a function of Si doping. Coverage ratio of h-BN domains obtained via precipitation during cooling as a function of distance from the larger isothermal domains. Both Si doping of Fe and Cu result in a decreased nucleation density. Si defects locally reduce the nucleation threshold and facilitate nucleation before the bare/pristine catalyst surface, resulting in larger domains. Adapted with permission. [265] Copyright 2019, Wiley. C) Multilayer h-BN grown from a Ni-Fe alloy. By tuning the alloy composition of Ni to Fe, the solubility of B and N can come closer to a stoichiometric ratio of 1:1 in the bulk catalyst, allowing for stoichiometric precipitation of B and N out of the bulk to create uniform multilayers. The Ni/Fe(100) facet facilitates multilayer growth as compared to the Ni/Fe(111) facet due to decreased atom packing on this surface. Adapted with permission. [266] Copyright 2018, ACS. D) Effects of alloying Ni and Cu on h-BN domain size during growth. Larger domain sizes are achieved at 10-20 at% Ni. Adapted with permission. [267] Copyright 2015, Springer Nature. E) Mechanism of h-BN growth on Fe 2 B substrate using nitrogen gas and the substrate itself as the reactants. SEM and TEM images after which the domain size sharply decreases. [267] Further, Yang et al., [271] observed a change in shape toward more rounded h-BN domains with increasing Ni composition. While the exact mechanisms behind h-BN growth on Cu/Ni alloys is still an area of active research, the community has taken advantage of the larger h-BN domains and the possibility of growing graphene/h-BN heterostructres. [270] Ni-Fe alloys have been shown to effectively allow for multilayer h-BN synthesis (Figure 11C). [266] Although, pure Ni films/foils readily grow multilayer h-BN films, the strong facet dependence leads to inhomogeneous multi-layer h-BN film growth on Ni foils/ films that is detrimental to applications with stringent quality requirements. Pure Fe also exhibits phase transition during growth along with the inherent complexities of different B and N solubility that restrict its effectiveness in producing large area uniform multilayer films. Specifically, Ni-Fe alloy, deposited on a spinel (MgAl 2 O 4 ) either (100) or (111) single crystal (Figure 11C), maintains the same phase (fcc) throughout the entire range of operating temperatures during h-BN growth and facilitates a near equal flux of precipitating N and B to grow uniform multi-layer h-BN films. Alloying Fe and Ni on the spinel surface also results in significantly larger metal domains than that of either pure metal, helping reduce variability from facet dependence. Interestingly, fcc Ni-Fe deposited on a spinel (111) surface formed a Ni-Fe (111) surface but showed little h-BN growth, however the same alloy composition when deposited on spinel(100) resulted in a Ni-Fe(100) surface and showed uniform multilayer h-BN growth. These differences are attributed to the Ni-Fe(100) facilitating diffusion of B and N in the bulk and are hence more conducive to uniform h-BN film formation than the Ni-Fe (111). [266] Another alloy that has been used to grow multi-layer h-BN films is Fe 2 B. [268,272,273] In this process, N supplied through NH 3 gas and B from the bulk alloy react at elevated temperatures to result in the growth of thick h-BN multi-layer films that form on top of the Fe 2 B (Figure 11E). Films grown in such manner, however, suffer from relatively poor crystalline quality. [268] In addition to monolayer and multi-layer h-BN growth, growth of bulk h-BN crystals is also typically performed using an alloyed metal flux. [36,10,[59][60][61][62][63]269,[274][275][276][277][278][279][280][281][282] Growth from a metal flux occurs via precipitation whereby the flux (usually a liquid metal with solubilized B and N atoms) is slowly cooled, often over many hours, slowly reducing the solubility limits of the dissolved B and N atoms, forcing them to precipitate to the surface and form h-BN crystal. The slow growth and good mobility of precursor within the liquid flux results in high quality bulk h-BN crystals with domain sizes often ≈500-1000 µm and thickness ranging from 10s to 100s of microns depending on the growth conditions. [36,10,[59][60][61][62][63]269,[274][275][276][277][278][279][280][281][282] This method has been popular in producing bulk h-BN crystals at atmospheric pressures. While fluxes of a single metal (Ni, Mg, Ba) can effectively produce h-BN crystals, [36,60,61,274,278,280] growth from an alloyed metal flux allows for tuning B and N solubility closer to stoichiometric balance B:N = 1 desired for h-BN formation via precipitation upon cooling. For example, Ni-Cr alloy has been investigated for h-BN growth because while Ni has high B solubility, the addition of Cr is necessary to increase the solubility of N such that, at atmospheric flux conditions, sufficient B and N precipitate to the flux surface in stoichiometric balance (Figure 11F). [62,276,277,281] Careful alloying can enable close to equal solubility of B and N, for example, Ni-Mo and Fe-Cr flux alloys have been demonstrated to successfully grow bulk h-BN crystals.

Here, we emphasize that flux methods have typically been the primary route to synthesize research-grade bulk h-BN crystals since the method consistently produces high quality crystals compared to state-of-the-art CVD methods. Although progress has been made in adapting the flux composition to be compatible at atmospheric pressures, the long growth time (often 36 + hours), high melt temperatures (often >1300 °C) and fine control over cooling rates needed as well as the lack of precise control over the number of layers or stacking order, presents some challenges (Table 8). [36,10,[59][60][61][62][63]269,[274][275][276][277][278][279][280][281][282]

h-BN growth on liquid metal has recently emerged as a potential route to synthesize large-area, monolayer h-BN that is nearly single crystalline with few to no grain boundaries. [72,179,286,287] A typical synthesis process consists of melting the h-BN growth catalyst over a support which facilitates the wetting of the catalyst liquid followed by the introduction of B and N precursors that react on the surface of the liquid metal to form a h-BN film (Figure 12A).

Growth on liquid metal eliminated issues of preferential/ differential h-BN growth on different crystalline facets of a solid catalyst film that results in inhomogeneity. Further features such as catalyst domain boundaries and rolling striations are eliminated upon melting. For example, Han et al., [287] compared growth on solid Ag to that on liquid Ag and found the h-BN films grown on the liquid surface to be more continuous and of uniform thickness with fewer defects such as cracks and pinholes than h-BN grown on the solid Ag (Figure 12A). [287] These observations were attributed to the absence of surface defects and impurities which are often present on the highly oxidation prone Ag surface. [287] Similarly, Khan et al., [179] compared the growth of h-BN on Cu foil with liquid Cu and found that while h-BN domains preferentially form at Cu domain boundaries on the foil, on liquid Cu a uniform h-BN film was grown (Figure 12B). [179] A liquid catalyst surface also allows for the floating h-BN domains to be mobile and self organize as they grow and merge. Using ammonia-borane precursor Geng et al., report on a highly ordered, h-BN films on liquid Cu. [286] The h-BN domains appear to merge with hexagonal symmetry with each monolayer domain featuring a star-shaped multilayer structure at the center (Figure 12C). The highly ordered structure suggests that the h-BN domains undergo some sort of self-organization to assemble and organize themselves and suggests that the growing domains are mobile on of the thick h-BN that grows as a result of the reaction of Fe 2 B and nitrogen. The h-BN begins to replace the top Fe 2 B surface. Adapted with permission. [268] Copyright 2019, RSC. F) Flux-type h-BN growth from metal alloys has enabled atmospheric pressure growth at high temperatures (≈1400 °C). Bulk h-BN crystals are grown via precipitation of B and N from a liquid metal flux as the system is slowly cooled. Large bulk crystals are produced whose size and quality are improved by the alloying of metals in the flux to more closely achieve a stoichiometric balance of B and N precipitating from the flux. Adapted with permission. [269] Copyright 2018, ACS.

the liquid metal surface. The observed hexagonally packed ordering suggests minimization of free space on the liquid surface. However, the multilayer structure on top or underneath of each of the monolayer h-BN domains remains unexplained. Further evidence of mobile, self-organizing monolayer h-BN domains on a liquid surface came from Lee et al., [72] (Figure 12D). They observed h-BN nucleation and growth forming circular domains (rather than triangular) on the liquid Au. The circular domain shape suggests that, on the Au liquid surface, neither boron nor nitrogen terminated edges are significantly preferred, [72] Further, as the individual domains grow on the liquid surface, they appear to be mobile on the surface and order themselves into a packed structure similar to that during Geng et al., growth on liquid Cu. [72,286] Examining the completely merged h-BN film, Lee et al., found that the merged h-BN crystal had a single orientation, [72] indicating that the individual h-BN domains rotate such that they seamlessly merge with their neighbors to create a single crystal h-BN layer. This arrangement was termed "self-collimation" and attributed to electrostatic interactions between the locally polar h-BN domain edges. Such a process presents potential for the synthesis of large-area single crystal monolayer h-BN without using an expensive single crystalline solid catalyst substrate. Major challenges of growing on a liquid metal surface will likely be the high rate of sublimation of liquid catalysts Cu, Au, and Ag, the cost of precious metal catalysts, and wrinkles formed in the h-BN upon cooling the liquid metal with the h-BN film. In the latter, how the catalyst recrystallization affects the as-synthesized h-BN film remains yet to be fully studied (Table 9).

The full exploitation of new applications of h-BN has been limited by the availability of high-quality h-BN films. Before the early 2000's, h-BN was only available in powder form and used as a filler material for composites due it its excellent thermal conductivity, chemical resistance, and temperature stability. [16,19,289,290] Additionally, h-BN powder (similar to graphite powder) was used as a lubricant additive, desired for its layered structure and lack of any dangling bonds on the surface, creating an incredibly low friction material. [291,292] Once h-BN films started to become more available with greater crystal size dimensions via the development of flux-based and depositionbased growth techniques, a wider array of applications have been made possible. Figure 13 summarizes some of the most . Adapted with permission. [287] Copyright 2019, ACS. B) SEM of h-BN grown on solid (left) and liquid (right) Cu under high precursor flux. While growth predominantly is observed along Cu domain edges, the sample grown on liquid is complete and uniform due to the lack of surface features on the liquid surface. Adapted with permission. [179] Copyright 2015, Springer Nature. C) SEM (left) and AFM (right) of highly ordered h-BN structure grown on liquid Cu at significantly lower precursor flux. The h-BN domains appear to merge with hexagonal symmetry with each monolayer domain featuring a star-shaped multilayer structure at the center of each h-BN domain. Adapted with permission. [286] Copyright 2018, Wiley. D) Proposed mechanism for self-collimated h-BN domains on a liquid Au. h-BN domains rotate due to electrostatic interactions to then merge seamlessly. SEM images of h-BN film on Au at various time points. The h-BN forms circles which then align, and the gaps fill in to yield a large single crystalline sheet of h-BN. Adapted with permission. [72] Copyright 2018, AAAS.

www.advmat.de www.advancedsciencenews.com Synthesis process development timeline and applications. h-BN was first synthesized in powder form that has been often used in composite materials as a ceramic filler substance to increase strength, thermal stability, and thermal conductivity. Powder h-BN has also been used as a lubricant material similar to graphite powder. The development of flux-based growth in the early 2000's enabled development of bulk crystal applications, however flux-based growth is a difficult synthesis method to scale to large crystal sizes and different stacking configurations. Chemical vapor deposition methods for h-BN synthesis of multilayer and monolayer h-BN emerged around the 2010's and was more suitable for large area h-BN synthesis. Currently, various strategies are being developed to costeffectively produce single-crystal h-BN using vapor deposition methods. The development of scalable processes to produce large area single crystal h-BN will enable high quality devices and applications that require ultrathin h-BN. Such processes developed to produce large-area single crystal h-BN at scale will then need to be expanded on to enable multilayer deposition, ideally with control over stacking configuration. Scalable deposition of single crystal h-BN of thickness ranging from monolayer to hundreds of microns thick represents the highest complexity of h-BN synthesis development and will allow for the realization of several new applications.

promising and emerging applications for h-BN and links the applications to the h-BN quality required.

Once bulk h-BN was produced via flux-based growth in the early 2000's, the indirect bandgap of h-BN was experimentally determined to be ≈5.955 eV, making it one of the few materials that emits light in the deep ultraviolet range. [30,36,37,41,10,97] h-BN's wide indirect bandgap and ultra-flat, unreactive surface make h-BN an ideal substrate for nanoscale and 2D electronic devices. For example, graphene on SiO 2 exhibits electron mobility of ≈10 000 cm 2 V -1 s -1 , [20] but graphene on the surface of h-BN crystals have significantly higher mobility, even reaching µ ≈60 000-500 000 cm 2 V -1 s -1 (when graphene is either supported on or sandwiched between two h-BN crystals). [30,293] Additionally, the exceptionally high in-plane thermal conductivity of h-BN allows for efficient dissipation of heat away from the devices and is a promising candidate for thermal management. [294] The combination of h-BN's large bandgap, inert chemical structures, and high thermal conductivity make h-BN an ideal substrate for future electronics. [13,21,[26][27][28][29][30]36,77,[295][296][297][298][299][300][301][302] Not only is h-BN of interest to the photonics community due its bandgap, but h-BN is also a naturally a hyperbolic material (the permittivity tensors along different crystallographic directions are opposite in sign) in the mid-IR, where these optical modes spectrally overlap with the molecular fingerprint region of the electromagnetic spectrum. [13,303,304] The ability to support hyperbolic phonon-polaritons (coupling of infrared photons with optical phonons [39,305] ) make it especially interesting for use in advanced optical signaling devices. [303,306] Additionally, these volume-confined hyperbolic phonon polaritons enable label-free, subdiffractional hyper-lens imaging [307][308][309] of other materials and has enabled broader research into understanding polaritonic behavior of other nanomaterials. [307,[310][311][312][313][314][315][316] Notably, significant quantum emission of polarized and ultrabright single-photons has been observed from atomic scale defects in h-BN at room temperature, whereas most quantum emitters have only been observed at cryogenic temperatures. The possibility of atomic vacancies in h-BN acting as quantum sources for room temperature quantum computing devices remains exciting. [317] For a more detailed review on photonic applications of h-BN, we refer the reader to Caldwell et al. [39] Bulk growth techniques using B 10 isotope enriched reactants have enabled the production of highly efficient neutron detecting materials. Most conventional neutron detecting devices require a gaseous component with 3 He that is rare and short in supply, meanwhile h-BN present a promising solidstate alternative for portable detectors for national security as well as preventing illicit transport of materials at ports of entry. [40][41][42][43][44][45][46]101,318,319] CVD based growth of few, and monolayer h-BN has enabled a host of applications that require one or only a few atomic layers. Monolayer and few-layered h-BN were first demonstrated as tunnel barriers in tunneling transistors and spintronic devices from exfoliated crystals. CVD grown monolayers have enabled larger devices and easier assembly of such devices. h-BN is an excellent electron tunneling material due to its thin and insulating nature and the characteristics of such device can be tuned by adjusting the number of layers. [13,22,27,35,296,299,320] Additionally, monolayer h-BN grown directly on ferromagnetic Fe has been used as an effective system for a magnetic tunnel junction using h-BN as the spin tunnel junction barrier. [32] Growing h-BN directly onto a device component (Fe is this case) ensures a clean, polymer free interface as well as a corrosion barrier for the magnetic metals used for spin injection. Such direct use of h-BN on its own growth substrate is perhaps exemplified using h-BN as an oxidation barrier coating. Being ceramic and impermeable to oxygen, h-BN has drawn interest as an ultrathin, high-temperature coating for oxidation protection. [47,202,[321][322][323] By growing a thin h-BN film on a target metal, a clean coating/substrate interface is ensured. It is worth noting, however, that atomically-thin h-BN will only be effective if there is a strong h-BN/metal interaction and the underlying substrate does not oxidize via a wet-oxidation mechanism as point-sources of oxygen inevitability are available through occasional defects on the h-BN surface. [226] h-BN has also been demonstrated to be a useful buffer material for easy removal of other grown Van der Waals materials, especially III/V semiconductors. [324,325] The lattice transparency of monolayer h-BN allows for newly grown semiconductors to be templated by an underlying substrate, but the inert h-BN separates the new growth from the template and allows for a facile removal of the newly grown film. [23,24] Transferred off the growth substrate, CVD-grown h-BN has been demonstrated as an exciting material in the fields of separations. [35,203] h-BN presents opportunities in the field of proton exchange membranes as well as isotope separations because the pristine h-BN lattice is impermeable to all gases including hydrogen gas and hydrated ions but remains highly permeable to protons, making it useful for integration into fuel cells and other electrochemical systems. [35,326,327] Passage of hydrogen isotopes through pores in the electron cloud of h-BN in Nafionh-BN-Nafion devices allows for energy efficient isotope separations due to differences in the vibrational zero-point energy of the isotope that is transiently bound to the Nafion before being incident on the h-BN. [35] This results in in slower transport of deuteron and tritium compared to protons, making h-BN of interest for nuclear decontamination efforts. [327][328][329][330][331] h-BN also presents new opportunities for size-selective molecular separations. Leveraging developments made into nanoporous graphene membranes, h-BN offers additional advantages of higher chemical and thermal stability allowing for operation under harsher conditions owing to h-BN ceramic nature. Notably, the nitrogen terminated edges are more stable in h-BN and pores/defects in h-BN tend to be triangular in shape. [56] DNA translocation through pores in h-BN, [49,57,332,333] ionic transport in the aqueous phase, [245,334,335] and molecular separations in the gas phase [48,[50][51][52][53][54][55]336] have been demonstrated.

With most applications of h-BN, the ideal material is single crystalline h-BN with controllable film thickness, since, atomic defects at grain boundaries will impact the tunneling properties and proton transport characteristics in tunneling transistors and proton-selective membranes, respectively. Developing scalable methods of producing high quality large-area h-BN single crystals is therefore imperative to advance applications. Currently, state-of-the-art synthesis processes are being developed to produce large area single crystals of monolayer h-BN. [72,189,200] In the future, such methods will need to be expanded to allow for deposition of additional layers with control over stacking order.

h-BN exhibits an array of unique properties ranging: physical, electronic, quantum, optics, amongst others. Due to these unique properties, h-BN has attracted significant research interest and we summarize the most promising and emerging applications for h-BN. While device application and characterization research has flourished, with seemingly new applications and features discovered on a regular basis, it will be essential to develop rational growth strategies to produce high-quality h-BN at scale to bring this material to market. Here, we reviewed research into the synthesis of mono-and multi-layer h-BN films on metallic, insulating, alloyed, single-crystalline, polycrystalline, and liquid catalysts/substrates. Each growth substrate offers advantages and disadvantages that come into play in terms of film quality, ease of growth, domain size, film thickness, transferability, and cost.

Despite tremendous progress in synthesis, some challenges remain. The first of which is being able to move away from current flux-based synthesis methods. This is an inevitable transition that will need to happen to produce enough uniform highquality h-BN films at scale. This will, however, be a difficult transition as flux methods still produce vastly higher quality crystals compared to CVD methods. CVD can effectively produce high-quality monolayer or few-layer films, however h-BN of micron-scale thickness via CVD remains challenging.

The second major challenge of scalable growth of h-BN using CVD is the sacrificial growth catalyst and transfer process. In the most basic sense, for any growth process to be scalable, the price of the material produced must be of significantly higher value than the materials and energy used to produce it. Growing significant areas of monolayer h-BN on a catalytic metal and then proceeding to etch away the catalyst metal to isolate the h-BN is simply not sustainable at scale. The cost per area of h-BN would have to far exceed that of the catalytic substrate and for metals such as nickel, iron, or copper, etching remains unviable at scale. Perhaps an effective methodology that recycles the dissolved catalyst could partially alleviate this cost, however such recrystallization from solution would be energy intensive. Solutions to this issue will involve: i) continued development of effective ways to transfer grown h-BN to the target application substrate without impairing the catalyst (e.g., catalyst recycling [218,212] ); ii) development of growth procedures to grow directly on typically used electronic application substrate; or iii) development of application substrates that facilitate h-BN growth. All of these will involve significant collaboration between crystal growers and device makers to effectively connect high quality h-BN with end use.

Even if a growth method is optimized to effectively produce large area h-BN at scale, the transfer methods that are employed to move the h-BN to the desired substrate need to be seriously considered for their effect on the ultimate device properties. For example, a PMMA based transfer of h-BN typically results in polymer residue on h-BN surface which could negatively influence applications with stringent quality requirements, for example, tunnel barriers. Here, h-BN's high thermal stability (compared to graphene) enables high temperature annealing of the newly transferred h-BN to bake off a significant amount of the residue, however, there inevitably will be some polymer left behind. This is not to say, however, that polymer residue cannot be tolerated in all applications. For example, if the h-BN is being used to encapsulate an air sensitive device, some polymer residue at the air h-BN interface may not affect the passivation the h-BN provides to the device. Significant thought is therefore needed when considering not only the substrate on which h-BN is being grown, but also on the post-processing steps that are required to then use the h-BN effectively in a device.

Finally, the potential for quantum computing, atomically thin ceramic membranes, proton sieves, isotope purification, and atomically thin tunneling junctions continue to feed interest into the development of scalable h-BN growth practices. Notably, the insights gained from h-BN growth will be relevant to the vast majority of other 2D materials that typically contain more than one constituent element and thereby, enable the next generation of transformative applications.