Fabrication and characterization of superconducting MgB₂ thin film on graphene Open Access

Since the first extraction through mechanical cleavage, graphene, a single atomic layer of carbon, has remained the subject of intense study owing to its exceptional properties and high promise in various practical applications.^1,2 By bringing graphene in contact with a superconductor, superconductivity may be induced in graphene via the proximity effect, i.e., the extension of superconducting correlations from the superconductor into graphene. Theory shows that, as graphene has a unique electronic structure with relativistic linear energy spectrum, the proximity effect in graphene may exhibit unusual properties, such as the presence of specular Andreev reflection in the vicinity of the Dirac point.³ Moreover, the carrier type and density in graphene can be tuned through the electric field effect, which provides an important way to modulate the property of the proximity effect, for instance, to modulate the supercurrent in the Josephson junction formed by graphene with two closely-spaced superconducting electrodes.^3,4 In addition, it has been proposed that, with the combination of the quantum Hall effect in graphene and superconductivity, graphene/superconductor junctions may host Majorana bound states when operating in the quantum Hall regime, offering a promising route to realize topological superconductivity.^5,6 These, in tandem with other features like the chemical stability of graphene at ambient environment, make the building of superconductor/graphene hybrid very attractive for both fundamental studies and superconducting device applications.

In experiments, the superconductor/graphene hybrid has been investigated by combining graphene with a variety of low-temperature (low-T_c) superconductors. For instance, gate-modulated bipolar supercurrent was first demonstrated in the Josephson junction made from the deposition of two Al superconducting electrodes on graphene.⁷ By incorporating two such Al/graphene/Al junctions in a superconducting loop, the modulation of the supercurrent by applied magnetic fields, i.e., the quantum interference effect, has also been illustrated.⁸ Other superconducting electrodes, including Ta,⁹ Nb,¹⁰ NbN,¹¹ etc., are used to make graphene junctions as well, in which various topics, such as the interplay between the quantum Hall effect and the Andreev reflection and the ballistic transport of the supercurrent, have been addressed. With tunneling spectroscopy, the induced superconductivity in graphene grown on Re thin film has been investigated.¹² At the interface of NbSe₂/graphene bilayer, interband specular Andreev reflection has been observed.¹³ In encapsulated graphene contacted by MoRe superconducting electrodes, supercurrent mediated by quantum Hall edge states has been demonstrated, marking a step in the quest for novel topological excitations.¹⁴ Very recently, superconductor/graphene hybrid has also been realized by placing graphene onto high-temperature (high-T_c) cuprate superconductors Pr_2−xCe_xCuO₄ and YBa₂Cu₃O₇, which permits the probe of induced unconventional superconductivity or Klein tunneling of Cooper pairs into graphene through conductance measurements.^15,16

In 2001, MgB₂ was discovered to be superconducting with a transition temperature T_c in bulk at 39 K, the highest among intermetallic compounds.¹⁷ Compared with low-T_c superconductors, by combining graphene with MgB₂ to create MgB₂/graphene hybrid would in principle allow for the investigation of the proximity effect in graphene to considerably higher temperatures, which would be quite beneficial for us to better explore fundamental issues regarding the interplay of superconductivity and relativistic quantum dynamics. From an application perspective, the realization of MgB₂/graphene hybrid would also be very appealing in developing related devices to function at elevated temperatures and hence to reduce the cryogenic cost and requirement. As a matter of fact, concerning MgB₂, a lot of efforts have been devoted to using it to make superconducting devices or circuits in order to take advantage of its high T_c.^18–21 In this respect, the MgB₂/graphene hybrid would offer a new route, for instance, to fabricate MgB₂/normal metal/MgB₂ Josephson junctions.²² Another notable feature of MgB₂, which sets it apart from many other superconductors including the high-T_c cuprate materials, is that it is a multiband superconductor.^23,24 Specifically, in MgB₂ there are two σ-bands and two π-bands crossing the Fermi level, resulting in four sheets of the Fermi surface and two groups of superconducting gaps, namely the σ- and π-gaps.^23–26 This multigap superconductivity has shown to lead to many properties of MgB₂ markedly different from that in conventional single-gap superconductors and, moreover, the presence of new physical phenomena that do not exist in single-gap superconductors.^18,27 For instance, novel vortex patterns not unattainable in single-gap superconductors, such as the vortex stripes or clusters and fractional vortices, have been experimentally or theoretically explored for MgB₂.^28,29 Actually, the research on multigap or multicomponent superconductivity has recently evolved into a fascinating field owing to the rich emergent quantum effects exhibited therein.³⁰ In light of this, the MgB₂/graphene hybrid may also provide a unique platform to explore possible new physics in the proximity effect in graphene resulting from the multigap superconductivity of MgB₂.

In this paper, we present the successful fabrication and characterization of superconducting MgB₂ thin film on graphene. The deposited MgB₂ thin film on graphene shows a continuous film surface with prevailing c-axis orientation, and importantly exhibits superconducting transition in width less than 3 K at a T_c of 36 K, close to the value in bulk MgB₂. It demonstrates the feasibility to obtain MgB₂/graphene hybrid structure with high T_c, which is critical to both investigating the proximity effect in graphene at relatively high temperatures and developing related superconducting devices such as the high operating-temperature Josephson junctions.

The process for fabricating MgB₂ thin film on graphene is illustrated in Fig. 1. We use the graphene on Cu foil prepared by chemical vapor deposition (CVD) method as the starting material to obtain monolayer graphene in large area to deposit MgB₂ thin film. Accordingly, the whole process can be divided into two major steps: the transfer of graphene from Cu foil to the target substrate and then the deposition of MgB₂ thin film on the transferred graphene. The step of graphene transfer follows the procedure described in our previous work.³¹ Briefly, as schematically shown in Fig. 1, the monolayer graphene on Cu foil (XF023, XFNANO Materials Co.) was first spin-coated with polyvinyl butyral (PVB, dissolved in alcohol with 5% mass concentration) in thickness of about 500 nm and placed at room temperature for about 30 min to cure. The 20 μm thick Cu foil was then etched away by an aqueous solution of iron nitrate (1 mol/L) over several hours, with the obtained PVB/graphene layer washed by deionized water and placed on the target substrate (5 × 5 mm² c-cut 6H-SiC substrate in this work). After heated at 60 °C for about 10 min to soften the PVB and with the PVB finally dissolved by alcohol, the transferred graphene on SiC substrate is ready for the next step of depositing MgB₂ thin film.

The deposition of MgB₂ thin film is performed using the hybrid physical-chemical vapor deposition (HPCVD) method,³² as described in detail previously.^33,34 During the deposition, the graphene/SiC substrate, placed on a resistive-heated susceptor, was surrounded by four pieces of Mg ingots in a chamber maintained at a high pressure (∼ 5 kPa) with a continuous flow of purified H₂ gas (300 sccm). When the temperature reached ∼ 680 °C, at which the Mg ingots melted generating high Mg vapor pressure around the substrate, the B₂H₆ gas (5% in H₂) flowed through the chamber (6 sccm), providing pure B source via its thermal decomposition to react with Mg vapor and hence initiating the MgB₂ film deposition. With a deposition time of 2 min, the thickness of the MgB₂ film is about 30 nm. To better evaluate the deposition of MgB₂ film on graphene, the deposition of MgB₂ film directly on SiC substrate without graphene on it has also been performed under the similar conditions.

To characterize the fabricated samples, several experimental techniques have been employed, including the Raman spectroscopy, X-ray diffraction (XRD), scanning electron microscopy (SEM), and the electrical resistance measurement. The Raman spectrum was obtained at room temperature with a Renishaw inVia Raman spectrometer at laser wave length of 532 nm. The XRD pattern was recorded by using a Rigaku D/max-RA X-ray diffractometer with Cu Kα radiation and a scan length of 20° − 80° (2θ) at 4°/min. The SEM image was taken with a FEI Quanta 200 microscope to investigate the morphology of the sample. The electrical resistance of the sample was measured in temperature range of 5-300 K in our homemade setup, which uses a Lakeshore calibrated DT-670 silicon diode thermometer and a Keithley 6221 current source coupled with a Keithley 2182A nanovoltmeter allowing for standard four-probe measurement.

Figure 2 shows the Raman spectra taken at various stages of the fabrication process. To check the transfer of graphene on SiC substrate, the Raman spectrum of the sample was measured before the deposition of MgB₂ thin film, as shown in Figs. 2(a) and 2(d), with the latter obtained at a higher laser intensity focusing the spectrum range of 1400-3200 cm⁻¹. It is shown in Fig. 2(a) that, at 100-2000 cm⁻¹, the spectrum features many peaks, with two strongest centering at 788 cm⁻¹ and 968 cm⁻¹. By comparing with previous experimental studies on Raman spectra of SiC,^35–37 it is found that all these peaks can be ascribed to the SiC substrate as their positions agree very well with that established for 6H-SiC, with the two strongest assigned to the E₂ and A₁ phonon modes respectively. At 2000-3200 cm⁻¹, the spectrum shows a single peak centering at 2692 cm⁻¹, the well-known 2D peak for graphene.³⁸ Hence, the spectrum shown in Figs. 2(a) and 2(d) confirms the successful transfer of graphene on SiC substrate. Note that, another peak typical for graphene, the G peak at ∼ 1580 cm⁻¹, may have been overwhelmed in the spectrum by the peaks at similar positions from the SiC substrate, as can be perceived in the magnified view of Fig. 2(d).

After the deposition of MgB₂ thin film on graphene/SiC substrate, Fig. 2(b) shows that a broad peak centering at ∼ 600 cm⁻¹ appears in the spectrum. From previous studies,^39–43 it is known that this broad peak is typical for MgB₂, assigned to the E_2g phonon mode. Owing to the cover of MgB₂ film, Figs. 2(b) and 2(e) also show that some of the SiC peaks and the 2D peak for graphene have not been detected in the spectrum. Hence, to make a cross-check, the Raman spectrum has been taken after finally removing the deposited MgB₂ thin film via conventional acid etching, as shown in Figs. 2(c) and 2(f). It is seen that, the broad peak around 600 cm⁻¹ disappears as expected, some weak peaks for SiC reappear, and importantly the 2D peak of graphene shows up again. This demonstrates unambiguously that the deposition of MgB₂ thin film on graphene has been achieved. It also indicates that the transferred graphene has a good contact with the underlying SiC substrate.

In Fig. 3, the XRD pattern of the MgB₂ thin film on graphene/SiC substrate has been compared with that of the MgB₂ film deposited directly on SiC substrate and that of the SiC substrate alone. It is shown that the SiC substrate exhibits only the (006) and (0012) peaks within the scan length of 2θ, consistent with its c-cut property. For the MgB₂ film deposited directly on SiC, it is shown that, apart from the substrate peaks, only MgB₂ (001) and (002) peaks are present, suggesting the c axis of the film oriented normal to the substrate surface. This is in agreement with previous reports of the epitaxial growth of MgB₂ on SiC,^32,44,45 primarily owing to the very close lattice match between the hexagonal 6H-SiC (a = 3.081 Å) and the hexagonal MgB₂ (a = 3.086 Å).^17,44 The c-axis lattice constant of the film deduced from the position of the MgB₂ peaks is found to be slightly smaller than the bulk value of 3.524 Å,¹⁷ which may indicate an in-plane tensile strain in the film, as reported previously for MgB₂ films deposited by HPCVD on SiC substrate.⁴³ For the MgB₂ thin film deposited on graphene/SiC substrate, Fig. 3 shows that, similar to the film deposited directly on SiC, only MgB₂ (001) and (002) peaks are observed in addition to the substrate peaks, which indicates that, on graphene, the growth of MgB₂ film with the c axis perpendicular to the film surface has also been realized, although the lattice mismatch between the hexagonal graphene (a = 2.460 Å)^3,46 and MgB₂ is considerably larger than that between 6H-SiC and MgB₂. From this lattice mismatch, one might expect compressive strain in the film. It is interesting, however, to note that the position of the MgB₂ peaks is nearly the same as that for the film deposited directly on SiC, which seems to suggest an in-plane tensile strain as well in the MgB₂ film on graphene/SiC substrate. According to previous study,⁴³ besides the lattice mismatch between the film and the substrate, other factors, such as the mode of the film growth and the thermal expansion mismatch, would also result in strain in MgB₂ films. Hence, to obtain precise knowledge of the strain in MgB₂ film on graphene, more detailed investigations are needed. In Fig. 3, one may also note that for the MgB₂ film on graphene/SiC substrate there are two peaks appeared near the position of SiC (006) and the peak of SiC (0012) appears at a slightly higher value of 2θ in comparison with the other two SiC substrates. This is likely due to some slight variation in lattice parameters among different batches of the 6H-SiC substrates which are n-type doped with certain amount of nitrogen.

The above indication of the film growth from XRD pattern has also been checked by the SEM images, as shown in Fig. 4. For the MgB₂ film deposited directly on SiC, it is shown in Figs. 4(c) and 4(d) that the film is dense and has a smooth surface, with grains, in average size of about 500-700 nm, intimately connected with each other. In Fig. 4(c), a few small hexagonal-shaped grains are also seen on the very top of the film surface, demonstrating further the c-axis orientation of the film as indicated from the XRD pattern. By comparison, Fig. 4(b) shows that, for the MgB₂ film deposited on graphene/SiC substrate, the average size of the grains is smaller, typically below 200 nm. The grains are also well-connected, but the film does not look as dense as the film deposited directly on SiC. The majority of the grains are shown to align along the surface of the film, which, consistent with the above XRD result as well, corresponds to a prevailing c-axis orientation of the film. At the same time, there are also grains not aligned along the surface but leaning out of the surface of the film, shown as bright spots in Fig. 4(b). In a larger scale of Fig. 4(a), it is seen that these grains, i.e., bright spots, are dispersed over the film surface. The above results suggest that, while the present deposited MgB₂ film on graphene/SiC shows a decent quality in both the continuity of the film surface and on the whole the c-axis orientation, its epitaxial property is inferior to that of the film deposited directly on SiC substrate. This could be partly attributed to the considerably larger lattice mismatch between MgB₂ and graphene, as pointed out earlier. On the other hand, the transferred graphene may not lie completely flat on top of the SiC substrate, which could also result in the less perfect epitaxial growth of MgB₂ film on graphene/SiC than the film on SiC substrate directly.

In Fig. 5, the electrical resistance has been plotted as a function of temperature for both MgB₂ films grown directly on SiC and on graphene/SiC substrate, with the region around the superconducting transition magnified in Fig. 5(b). For the thin film on SiC, it is shown in Fig. 5(a) that, in the normal state prior to the superconducting transition, the resistance decreases with decreasing temperature, i.e., exhibits a metallic behavior, as typically observed for MgB₂ films prepared by HPCVD on SiC substrate.^44,45,47 From Fig. 5(b), one can see that the T_c of the film, defined by the midpoint of the superconducting transition, is 39.2 K, and the width of the superconducting transition ΔT_c, defined by the difference between the onset temperature T_c,onset and the zero-resistance temperature T_c,0 of the transition, i.e., ΔT_c = T_c,onset − T_c,0, is about 1 K. These demonstrate a reasonably narrow superconducting transition of the film with a value of T_c comparable to that in bulk samples.¹⁷ Here, one may notice that, on SiC substrate, MgB₂ thin films with even higher T_c and narrower ΔT_c could be grown with the HPCVD method, as illustrated in previous works,^43,44 implying that there is still room for us to further optimize the currently-employed deposition parameters.

For the MgB₂ film deposited on graphene/SiC substrate, it is interesting to see in Fig. 5(a) that the resistance initially increases as temperature decreases from 300 K down to about 120 K, showing a semiconducting behavior. Below 120 K, the resistance drops with reducing temperature, similar to the film deposited directly on SiC. We note that, in previous works, such type of semiconducting behavior at elevated temperatures has ever been reported for MgB₂ films prepared by other experimental techniques on traditional substrates. For instance, in MgB₂ films prepared on Si or SiC substrate by using electron-beam evaporation with post-annealing,^48,49 the resistance has been observed to show similar semiconducting behavior with temperature decreasing from room temperature down to T_c or down to some value above T_c. It was suggested that this phenomenon may reflect Mg deficiency in the films as a result of the Mg loss during the post-annealing process in vacuum.^48,49 In our eyes, this picture may not be applicable in the present case because in HPCVD method there is no post-annealing process involved and during the growth of the film the high Mg vapor pressure near the substrate would usually ensure the stoichiometry of the film.¹⁹ For the present observation of semiconducting behavior, we suspect that it may relate to, as shown in Figs. 4(a) and 4(b) and discussed earlier, the slanted grains dispersed in the film, which introduce additional grain boundaries in the film and may hold back to some extent the continuous flow of electrical current between grains aligned along the film surface. For the present MgB₂ film on graphene, it is worth noting that the underlying graphene may also carry a fraction of electrical current, which could further complicate the resistance behavior of the film. More detailed examination of the above issues is the subject of further study.

In Fig. 5(b), it is shown that the MgB₂ film on graphene/SiC substrate has a T_c of 36.3 K and a transition width ΔT_c of 2.7 K. Compared with the film deposited directly on SiC substrate, the T_c is about 3 K lower and the ΔT_c is about 2 K wider. This suggests a weak suppression in T_c in the MgB₂ film on graphene/SiC, which might arise from, as indicated above by both the SEM images and the normal-state resistance behavior, the increased level of disorder in the film. For MgB₂ thin films, as discussed in detail in our previous study,⁴⁷ the higher level of disorder may result in an enhancement of the interband scattering (between σ and π bands), which has a pair-breaking effect and would lead to T_c suppression. This could also account for the slight widening of the superconducting transition. On the other hand, it is noted that the proximity effect between the MgB₂ thin film and the graphene beneath it may also render the T_c of the film lower than that of the MgB₂ bulk sample or thin film deposited directly on SiC substrate.⁴⁷ While the above phenomenon of weak T_c suppression is worth further investigating, the present result demonstrates that it is experimentally achievable to grow MgB₂ thin film on graphene with T_c retaining close to the bulk value of 39 K, which is of importance in illustrating the good possibility to explore superconductivity induced at relatively high temperatures in graphene by connecting it with MgB₂ thin film.

Finally, let us point out that, as we have employed the procedure of depositing MgB₂ thin film on graphene to yield the MgB₂/graphene hybrid, to explore the superconducting proximity effect of the sample, subsequent micro- or nano-patterning of the sample would be needed in obtaining the desired structure. Figure 2 indicates that it is applicable to utilize traditional etching method to remove the upper MgB₂ thin film and expose the graphene underneath. Hence, by further incorporating the standard lithography technique,^22,34 the fabrication of MgB₂/graphene micro-structures or junctions could be envisioned. Alternatively, one may first use the lithography technique and certain masks to define the deposition area on graphene, as widely employed in the literatures to fabricate graphene Josephson junctions by using low-T_c superconductors,^7–11 and then deposit MgB₂ thin film to realize the targeted structure. Currently we are investigating these two approaches on the basis of the present work.

One may also note that, compared with the monolayer graphene, the MgB₂ film is relatively thick, in the order of 100 MgB₂ monolayers, in the presently fabricated MgB₂/graphene bilayer films. In recent years, superconducting ultrathin films with thickness down to a few monolayers have drawn considerable interests from the community; one reason, among others, is that they provide another exciting route to observe multiband superconductivity.^30,50 In atomically thin MgB₂, in particular, theoretical work showed that the surface states can make a major contribution to the superconducting gap spectrum, which, distinct from the bulk σ- and π-gaps, can be further tuned by the number of monolayers and the strain in the film.⁵¹ On few-monolayer MgB₂ film prepared by using the molecular beam epitaxy (MBE),^52,53 the predicted surface states and the opening of the additional gap have been experimentally demonstrated.⁵⁴ How these new features influence the proximity effect in MgB₂/graphene hybrid with atomically thin MgB₂ would be very appealing to explore. In previous works we ever performed the HPCVD growth of MgB₂ film (not on graphene) in thickness down to about 6 nm, i.e., about 20 MgB₂ monolayers.^33,47 It is interesting to see whether this method with refinement of the growth parameters, or alternative method like the MBE,^52–54 is capable to produce atomically thin MgB₂ film on graphene in the near future to allow for the exploration on such new topics.

In summary, we have experimentally explored the fabrication of MgB₂/graphene bilayer films. By transferring CVD-grown monolayer graphene from Cu foil to the 6H-SiC substrate, MgB₂ thin film has been deposited on graphene/SiC with the use of the HPCVD technique. The deposited film shows predominant c-axis orientation with well-connected grains, and displays superconducting transition within 3 K at mid-point T_c of 36 K, illustrating the effectiveness of the present method in obtaining MgB₂/graphene hybrid with decent properties. It is worth pointing out that, in the present method, the bottom substrate is not necessarily SiC, as graphene can be similarly transferred to other desired substrate materials; also, the layer number of graphene could be readily tuned by repeating the transfer process; and moreover, the fabrication of further complicated structures like the stacking of alternating graphene and MgB₂ films should be possible. These indicate the potential of the present method to fabricate MgB₂/graphene hybrid in more different configurations. With further refinement of the fabrication process and the integration of suitable patterning techniques, such MgB₂/graphene hybrid could serve as an appealing platform for both studying the interplay of superconductivity and relativistic quantum dynamics and developing related devices such as the Josephson junctions to work at comparatively high temperatures.

This work was supported in part by the Ministry of Science and Technology of China under 973 Program 2011CBA00106, and by the National Natural Science Foundation of China under Grant 11074008.