Controlled assembly of artificial 2D materials based on the transfer of oxo-functionalized graphene

Functionalized 2D materials have unique properties, but are currently not used for the assembly of van der Waals heterostructures. Here, we present the controlled transfer of artificially synthesized, polar and highly transparent oxo-functionalized graphene, which can decouple graphene layers.

The fast-growing family of two-dimensional materials includes diverse members ranging in properties from insulators (e.g. h-BN and graphene oxide) 1 and semiconductors (MoS 2 , WS 2 , and PtSe 2 ) 2-4 to semimetals (graphene and TiS 2 ) 5,6 and superconductors (NbSe 2 ). 7 Recently, it became possible to assemble dissimilar 2D materials into heterostructures. Such heterostructures can exhibit properties not seen in individual building blocks such as exotic superconductivity, modied band structures, and superlattice effects. [8][9][10][11][12] To date, single layers of various 2D materials of the highest quality have been produced by mechanical exfoliation of their multi-layered crystals. In the case of graphene, the production can also be carried out by wet-chemical intercalation and oxidation of graphite. [13][14][15] The thus obtained graphite oxide can be gently delaminated via osmotic swelling yielding graphene oxide akes with a typical size of up to 100 mm. [16][17][18] Graphene oxide belongs to a class of materials with variable sizes of sp 2domains isolated by defects (e.g. functionalization and vacancy defects). 19 In particular, we here use our recently introduced oxo-functionalized graphene (oxo-G), which is a graphene oxide derivative with a very low defect concentration. 14,20,21 Its density of in-plane lattice defects as well as the surface chemistry are well controlled. 20,[22][23][24] Oxo-G possesses hydroxyl, epoxy and organosulfate groups with hydronium counter ions as surface functional groups, resulting in a highly polar and protic surface with a negative zeta-potential. 17,25 Aer reduction of oxo-G to a certain quality of graphene (red-oxo-G) only vacancy and hole defects remain. The presence of defects leads to the characteristic photoluminescence (PL) spectrum that can be used as a ngerprint of the material in a heterostructure. [26][27][28] In this eld, a graphene/ozone-treated graphene heterostructure was already prepared as a transparent conductive electrode for an organic light emitting diode. 29 However, no controlled transfer of chemically modied graphene was demonstrated (Fig. 1).
In recent years the chemical modication of 2D materials such as functionalized MoS 2 , 30 graphene, 31 phosphorene, 32,33 and antimonene 34,35 have been investigated by various working groups. Although akes of functionalized 2D materials were obtained, the controlled transfer of those akes is not yet considered for the assembly of heterostructures. In particular oxo-G akes available in large quantities were not yet transferred in a controlled way, although there are transfer techniques for exfoliated monolayers of e.g. graphene, MoS 2 , or other transition dichalcogenides. [36][37][38][39][40] Flakes of oxo-G were used to improve perovskite solar cells, 41 and organosulfate groups in oxo-G were exploited as templates for incorporating molecules, forming hybrid materials suitable for the fabrication of lowvoltage memory devices. 42 However, to establish the fabrication of functional heterosystems, which include oxo-G or oxo-G hybrid materials the controlled transfer of highly transparent and polar akes of graphene oxide and oxo-G, respectively, must be established.
Here, we systematically investigate the transfer of akes of oxo-G to make vertical heterostructures, including a structure of graphene/oxo-G/graphene (G/oxo-G/G).
First, we tested to make oxo-G/oxo-G and red-oxo-G/oxo-G structures by using the Langmuir-Blodgett technique. However, the approach is random and thus limited in terms of controlling the appropriate size, shape and quality of akes used for the assembly. Accordingly, the assembly of specic stacks of oxo-G derivatives using this method is highly unlikely. The details of this approach are summarized in the ESI (Fig. S1 †).
For the controlled assembly of specic heterostructures, akes of oxo-G were rst deposited by the LB technique on Si/ 300 nm SiO 2 wafers. The size and quality of akes are subsequently probed by optical microscopy, atomic-force microscopy (AFM) or Raman spectroscopy. Flakes of interest are selected, as shown in Fig. 2A and B. For the transfer, a polydimethylsiloxane (PDMS) stamp is used with a sacricial polymer layer consisting of poly(methylmethacrylate-co-n-butylmethacrylate) (short term: Elvacite 2550), as reported in previous work. 36 A schematic representation of the transfer slide is shown in Fig. 3B. The stamp is brought into direct contact with the desired ake, controlled by optical microscopy. The pick-up of oxo-G akes is facilitated with additional capillary force resulting from applying water vapor as described in the experimental section. 37 The release is achieved by heating the substrate to 120 C melting the Elvacite polymer (setup is described in the ESI and Fig. S2 †). Aerwards, the substrate is cooled to 90 C to ensure the integrity of the ake while removing the transfer slide. The successful transfer of the oxo-G ake is shown in Fig. 2C on top of the oxo-G ake of Fig. 2A.
While performing these transfer processes several obstacles emerged. For example, Raman mapping associated with laserinduced heating, conducted on the desired oxo-G ake before transfer, may inhibit the pick-up.
We propose that water molecules underneath the ake may determine the adhesion to the SiO 2 surface. This assumption is supported by the observation that lms of oxo-G akes deposited by the LB-technique on the SiO 2 substrate do sometimes not adhere to the SiO 2 substrate, which is very likely due to a lm of water between the substrate and oxo-G ake. 37,43,44 As described by Zomer et al. akes of graphene and h-BN can be transferred using a polymer coated adhesive tape on a PDMS stamp placed on a glass slide. 36 The oxo-G akes and red-oxo-G akes are barely visible on Elvacite using optical microscopy (see Fig. S3A †). When looking through the media (glass, PDMS, tape, Elvacite) of the transfer slide, the akes are not visible during the transfer process. Locating akes is however possible using impurities on the Elvacite polymer, such as few-layers or dust, which act as reference points. Nevertheless, these dust particles are more oen not gathered along with the desired ake, losing the reference points and thereby also the ake. For this reason, we used standard electron-beam lithography to create alignment marks using 3 nm Cr/20 nm Au on the glass slide substrate (Fig. 3B, see Fig. S3B †). Furthermore, we surprisingly recognized that a PDMS thickness of less than 0.5 mm increases the visibility of oxo-G akes and thus facilitates the location and positioning of akes.  This journal is © The Royal Society of Chemistry 2020 As depicted in Fig. 2C, the transfer process damaged the transferred ake. The height prole shows both the successful overlap of two akes and emerged wrinkles as well as cracks. Moreover, multiple folding is observed with a height of 10 nm. In order to determine the cause of additional wrinkles and cracks, the surface roughness of the exposed Elvacite lm was evaluated by AFM before and aer pick-up of akes. Fig. 3A shows that the Elvacite surface is composed of hills and valleys roughly 15 nm peak-to-peak, and akes may be placed in valleys with the tendency to form wrinkles (Fig. 3C) or on hills leading to cracks (Fig. 3D). Regarding the roughness, we observe that a roughness of 2.0 nm guaranties an extensive pick-up of akes while a roughness of 3.8 nm is not sufficient to pick up akes reliably. The low roughness of Elvacite can be achieved (see Fig. S4 †) by drop-casting a solution of Elvacite followed by inclining the substrate and letting the solution run down. In contrast a standard spin-coating process leads to an insufficiently high roughness (see ESI †).
With the developed method we prepared articial stacks of 2D materials, in particular oxo-G on oxo-G and red-oxo-G on oxo-G and also mechanically exfoliated graphene on oxo-G. The results of the transfer approach are presented in Fig. 4, and the different heterostructures are characterized by Raman microscopy. In Fig. 4B the Raman map of the oxo-G on oxo-G transfer (Fig. 4A) is shown. The overlap is accompanied by the increase of the intensity of the D peak, which is also demonstrated in the AFM image of Fig. 4C and by the average spectra of the bottom oxo-G (red line), the top one (blue line) and the overlap (green line). Herein the intensities (integrated area) of the peaks in the single spectra contribute in rst approximation additively to the spectrum of the overlap (see Fig. S5 †).
Next, transferring of red-oxo-G was conducted. While a partial overlap of oxo-G on oxo-G and on red-oxo-G respectively is possible with LB, it is not for red-oxo-G on oxo-G because overlapping oxo-G akes would be reduced by reducing agents. The transfer was performed in the same way as described earlier for oxo-G. The result is shown in the AFM image of Fig. 4D. The height prole of the AFM measurement indicates the overlapping area with a height of approximately 3 nm compared to red-oxo-G with about 1 nm height and oxo-G with a height of about 2 nm. Furthermore, the Raman map of the intensity of the D band is shown in Fig. 4E, in which the more intense, red coloured scale corresponds to the overlap.
The average spectra of the overlapping red-oxo-G/oxo-G region resemble the individual spectra of the red-oxo-G ake and the oxo-G ake. In addition, the transfer of mechanically exfoliated graphene onto oxo-G (oxo-G/G) was conducted to prove that other 2D materials, here in particular graphene can be transferred onto the akes of oxo-G. In Fig. 4G the AFM image as well as the height prole shows the overlapping area of akes with some glue residue from the tape on the graphene. The corresponding Raman map in Fig. 4H depicts the intensity of the 2D peak, used to visualize the different materials stacked on top of each other.
The average spectra in Fig. 4I, generated from single spectra obtained by Raman mapping, exhibit a sharp 2D peak for graphene and virtually no defect-induced D peak, while oxo-G is highly functionalized. Comparing the 2D peaks, it is found that the overlap is the sum of the graphene and the oxo-G intensity (see Fig. S5C †).
Finally, we demonstrate that oxo-G deposited between two layers of CVD graphene (G/oxo-G/G) leads to decoupling. The structure was then heated to 100 C so that impurities could diffuse out more easily and the structure could relax. The 2D peak measured (Fig. 5E) is the sum of individual 2D peaks, instead of a broadened 2D peak expected for bilayer graphene. 45 However, we propose that a random orientation of the graphene layers prevents perfect stacking as observed for bilayer graphene. Also impurities on the CVD graphene layer, as seen in Fig. 5C and in the AFM image in Fig. S6, † may inuence properties. However, Jeon et al. determined the critical distance for decoupling graphene layers to be 1.66 nm maximum depending on their orientation. 45 It is clearly seen in the height prole of the AFM image (Fig. S6 †) that oxo-G spaces the graphene layers according to its height of 2 nm. Thermal processing (tp) of that structure leads to the disproportionation of oxo-G into a ruptured carbon framework and CO 2 . 22 As depicted in Fig. S7 † the Raman 2D peak redshis by 11 cm À1 from 2684 cm À1 for G/ oxo-G/G to 2695 cm À1 for G/tp-oxo-G/G indicating the change of doping from p to n for the trilayer structure. 46

Conclusion
We introduced the controlled transfer of highly transparent oxo-G and demonstrated the specic assembly of oxo-G/oxo-G, oxo-G/red-oxo-G, oxo-G/G and also G/oxo-G/G, which identies oxo-G as an interfering layer. We solved inherent problems of transferring oxo-functionalized graphene derivatives, which are barely visible and may crack or fold. We propose that a layer of water between the surface and oxo-G derivatives allows successful transfer. Moreover, by preparing a marker labelled glass substrate for the transfer stamp the precise orientation of transferred akes becomes possible. The method is compatible also for transferring graphene and very likely other 2D materials, such as h-BN, TMDs or functionalized 2D materials in This journal is © The Royal Society of Chemistry 2020 general. In future research the developed technique can be used to incorporate functionalized 2D materials into heterostructures to fabricate optoelectronic devices. In addition, the inuence of defects in graphene on the properties of heterostructures can be studied. In general, with the presented method, novel 2D heterostructures with unexplored properties become accessible.

Conflicts of interest
There are no conicts to declare.