Solution phase synthesis of large-area ultra-thin two dimensional layered Bi2Se3: role of Cu-intercalation and substitution

Intercalation of high densities of guest species without affecting the host lattice is challenging. Here we report on a general solution-based synthesis route to intercalate high densities of zero-valent copper into layered Bi2Se3 nanosheets at room temperature. We develop a solution phase synthesis route to design large area single-crystalline two-dimensional ultrathin Bi2Se3 nanosheets with micron dimensions. Layered Bi2Se3 nanosheets possess rhombohedral crystal structure where the Bi and Se hexagonal planes remain in close stacked configuration forming quintuple layers along the c-direction. The coupling between two quintuple layers is predominantly van der Waals type, which allows intercalating smaller guest zero-valent copper within the layers of Bi2Se3 nanosheets. Such intercalation of guest species without affecting the lattices of Bi2Se3 is challenging considering the change in oxidation state of copper, which limits the intercalant concentration. Additionally, we show that the use of CuI-amine complex at high temperature reaction conditions yields CuI substituted CuI-Bi2Se3 nanosheets disrupting the host lattice of Bi2Se3 nanosheets. We have explored the role of intercalation and substitution on the electronic properties of pristine Bi2Se3 nanosheets. Development of new synthetic strategy for the synthesis of ultra-thin larger area 2D layered Bi2Se3 nanosheets and understanding the role of metal intercalation and substitution hold promises for fundamental understanding and energy related applications.

Despite of the intense research on the synthesis and application of metal chalcogenides NCs, more extensive research is prerequisite in order to gain control over size, shape, crystallographic phase, in addition to the underlying formation mechanism of NCs [22]. Recently, two-dimensional (2D) metal chalcogenide NCs showed a variety of surface active applications including electrochemical and topological insulators [23][24][25][26][27][28][29][30][31][32][33][34][35][36][37][38][39]. In 2D NCs, the charge carriers are confined along the thickness, however, are allowed to move along the lateral plane. Owing to the active basal planes and edges, 2D nanosheets can be modified by chemical functionalization by doping or intercalation to incorporate additional reactive sites. 2D nanosheets possess the ability to intercalate ions in between the adjacent layers and can provide intercalation pseudocapacitance [23][24][25]. Unlike rigid porous 3D materials, the nature of 2D nanosheets favours fast ionic transport through 2D channels. Due to their sub-nanometer thickness, 2D nanosheet have a high packing density that leads to a high volumetric capacitance, which is significant for manufacturing thin-film supercapacitors and microsupercapacitors [23]. Tuning the amount of intercalant concentration yields many surprising and unusual physical and chemical phenomena [24][25][26][27]. Although intercalation of alkali metals such as lithium can be high because of the small size, however, intercalation concentrations are mostly limited by the ionic nature of the intercalant [24]. The predominantly ionic nature of a guest species requires either a change in the host lattice oxidation states or the presence of vacancies to maintain charge neutrality, thus limiting the intercalant concentration. Intercalation of ionic guest species often initiates cation exchange reaction with host cation by substitution, which results in alloyed nanocrystals with different physical and chemical properties compared to the host NCs [7]. Additionally, formation of alloy may lead to the strain within the NCs due to size mismatch between the guest and host cations [7]. In contrary, zero-valent intercalants do not change the oxidation state of the host lattice, thus allowing the possibility of high concentration of intercalation [28][29][30][31][32]. Zero-valent intercalants manifest unusual physical effects such as superlattice structures and incommensurate charge density waves [30,[33][34][35][36][37]. The topological insulator characteristics of bismuth chalcogenides is well established from the recent theoretical and experimental studies [36][37][38][39]. The surface topological property is related to the shape of the bismuth chalcogenide NCs [37,38,40].
Depending on the choice of the cation precursor and reaction conditions, it is feasible to obtain either intercalated or substituted alloyed NCs. Recently, a technique has been developed using dilute concentration of metal in solution phase to intercalate high densities of zero-valent metals into layered materials [30]. For layered Bi 2 Se 3 crystal, it is possible to intercalate zero-valent Cu 0 in the van der Waals gap to obtain copper intercalated Bi 2 Se 3 NCs without disturbing the host lattice framework. On the other hand, foreign cation Cu + can undergoes cation exchange reaction which allows the substitution of the original host cation framework resulting in alloyed Bi 2 Se 3 NCs [41,42]. Notably, replacement of Bi III with Cu I will cause strain in the host crystal lattice considering the difference in the effective ionic radii of Bi III (1.03 Å) and Cu I (0.77 Å) in addition to the overall charge imbalance in the crystal. Additionally, the insertion of Cu I beyond in an excess limit may cause Cu 2-x Se NCs formation by complete cation exchange reaction [34,35].
Hence, the development of new synthetic strategy for the synthesis of 2D shape-controlled Bi 2 Se 3 nanosheets and understanding the role of metal intercalation or substitution are important for fundamental interest and technological applications. Here we demonstrate a rapid synthetic route for designing large area ultrathin singlecrystalline Bi 2 Se 3 nanosheets. The growth mechanism of Bi 2 Se 3 nanosheets involves oriented attachment mechanism where small Bi 2 Se 3 NCs undergo oriented attachment followed by epitaxial recrystallization. The layered Bi 2 Se 3 nanosheets consist of typical thickness of two quintuple layers (QLs) with a van der Waals gap within the layers. We observed that the overall shape of the nanosheets is sensitive to the synthesis temperature, nature of surfactants and annealing time. Short chain amine and acid surfactants and relatively low reaction temperature greatly enlarged the lateral dimension of the nanosheets. Furthermore, we are able to intercalate 30% of Cu 0 within the layers of Bi 2 Se 3 nanosheets without affecting the host lattice. On the other hand, use of Cu I salt substitutes bismuth from the host lattice thereby disrupting the host lattice.

Synthesis of Bi 2 Se 3 nanosheets
To synthesis of Bi 2 Se 3 nanosheets, first bismuth-oleate complex was prepared by dissolving 0.2 mmol of Bi(OAc) 3 in a mixture of 500 μl OLAc, 500 μl OLAm, 500 μl OctAc, 500 μl OctAm and 1 ml of 1-ODE in a three neck round bottom flask at 100°C. Formation of bismuth oleate complex results in a light yellow reaction colour. In a separate three neck round bottom flask, Se-complex was prepared by adding 0.3 mmol of Se powder in mixture of 2 ml OLAc, 2 ml OLAm, 500 μl OctAc, 500 μl OctAm and 1 ml of 1-ODE. The reaction mixture was heated to 180°C and annealed for 60 min at the same temperature till a dark orange colour appears. The formation of dark orange colour ensures the complete dissolution of Se powder and formation of Se-complex. Then, initially prepared bismuth complex solution was swiftly injected into the Se-complex solution at 180°C.
After annealing for 5 min at 180°C, the reaction flask was rapidly cooled down to room temperature by using water bath. As synthesized Bi 2 Se 3 nanosheets was precipitated out from the mixture using ethanol as antisolvent. Then, the solution was centrifuged at 5000 rpm to collect the precipitate and supernatant solution was discarded. Precipitate was washed with a mixture of chloroform and ethanol for three times. Finally, it was dispersed in toluene and stored for further use.
2.3. Synthesis of copper intercalated bismuth selenide nanosheets 2.3.1. Zero-valent copper (Cu 0 ) intercalation: Cu 0 intercalated Bi 2 Se 3 nanosheets was prepared by adding copper precursor solution into the Bi 2 Se 3 nanosheets dispersion in toluene under stirring condition. Tetrakis(acetonitrile)copper(I) hexafluorophosphate was used as a copper precursor. Copper precursor solution was prepared by dissolving 6 mg of tetrakis(acetonitrile)copper (I) hexafluorophosphate in 4 ml methanol. Bi 2 Se 3 nanosheets stock solution was prepared by dispersing the whole precipitate obtained from the synthesis of Bi 2 Se 3 nanosheets in 2 ml toluene. 100 μl of this Bi 2 Se 3 nanosheets dispersion was taken in a glass vial with 1 ml toluene and stirred well with a magnetic bar. 500 μl copper precursor solution in methanol was added to the Bi 2 Se 3 nanosheets dispersion and stirred for 30 min After 30 min of stirring, the reaction product was precipitated out using excess methanol and centrifugation at 3000 rpm. As obtained Cu 0 -Bi 2 Se 3 nanosheets was then dispersed in toluene for further use. Percentage of Cu 0 intercalation within the Bi 2 Se 3 nanosheets was controlled by adjusting the amount of copper precursor during the intercalation reaction. Cu 0 -Bi 2 Se 3 nanosheets was obtained via intercalation reaction using tetrakis (acetonitrile)copper(I) hexafluorophosphate with 5%, 10%, 30% and 60% compared to Bi 2 Se 3 (molar ratio of copper precursor with respect to Bi 2 Se 3 nanosheets). The reaction product obtained by using tetrakis (acetonitrile)copper(I) hexafluorophosphate with 5%, 10%, 30% and 60% are named as 5Cu-Bi 2 Se 3 , 10Cu-Bi 2 Se 3 , 30Cu-Bi 2 Se 3 and 60Cu-Bi 2 Se 3 .

Mono-valent copper (Cu I ) substitution
Cu I substituted Bi 2 Se 3 nanosheets was synthesized by injecting the copper precursor solution directly into the reaction solution of crude Bi 2 Se 3 nanosheets at high temperature. Here crude Bi 2 Se 3 nanosheets obtained after the reaction was used without purification. Copper precursor solution was prepared by dissolving 50 mg of CuCl 2 .2H 2 O in a mixture of 1 ml OLAm, 500 μl OctAm and 1 ml 1-ODE at 150°C to form the Cu-ammine complex. This Cu-ammine complex solution was injected into the crude Bi 2 Se 3 nanosheets at 180°C and annealed for 5 min. Aliquots are taken at different time interval to monitor the progress of the reaction.

Characterization
Powder XRD measurements were carried out using Bruker D8 powder diffractometer with Cu Kα (λ=1.54 Å) as the incident radiation. Thin film of the Bi 2 Se 3 nanosheets was used for the XRD measurements. Thin film was prepared by drop casting the Bi 2 Se 3 nanosheets on glass substrate. Transmission electron microscopy (TEM) was performed using UHR-FEG-TEM, JEOL; JEM 2100 F model operating at 200 kV. Size of Bi 2 Se 3 NCs was analysed using Digital Micrograph (DM3) software from JEOL. Bi 2 Se 3 NCs size was estimated by drawing line profile on the TEM image of NCs. Size distribution histogram plot was obtained by measuring the size of more than 100 Bi 2 Se 3 nanoparticles. Lattice spacings of Bi 2 Se 3 nanosheets were calculated from the HRTEM images of Bi 2 Se 3 nanosheets by using the line profile tool of DM3 software. Selected area electron diffraction (SAED) and energy dispersive x-ray spectroscopy (EDX) were also measured with the same electron microscope. Raman spectra were recorded using a J-Y HORIBA T64000 triple Raman spectrophotometer. UV-vis-NIR absorption spectra were measured using Varian Carry 5000 UV-vis-NIR spectrophotometer. Atomic force microscopy (AFM) imaging of the Bi 2 Se 3 nanosheets on a mica substrate was performed with an Asylum Research MFP-3D AFM in tapping mode using AC160TS silicon probes with nominal tip radii <10 nm. X-ray photoelectron spectroscopy (XPS) was carried out using Omicron x-ray photoelectron spectrometer with an Al Kα x-ray source.

Results and discussion
3.1. Preparation of Bi 2 Se 3 nanosheets Layered Bi 2 Se 3 possess rhombohedral crystal structure with five atoms in one-unit cell [space group D 3d 5 (R-3m)] (figure 1(a)) [43]. In rhombohedral Bi 2 Se 3 crystal structure, the Bi and Se hexagonal planes remain in close stacked configuration with a repetition of every five atomic layers as Se(Bi)-Bi-Se(Bi)-Bi-Se(Bi) ( figure 1(a)). The five-atom layers, known as quintuple layers (QLs), are arranged along the c-direction [34,35]. The coupling between two atomic layers within one QL is strong, however, is much weaker in between two QLs, where the interaction is predominantly van der Waals type. Hence, crystalline Bi 2 Se 3 can be exfoliated as 2D nanosheets similar to the graphene from bulk graphite. Additionally, guest atoms can be accommodated within the van der Waals gap between two QLs [35]. The TEM image of as-synthesized Bi 2 Se 3 nanosheets shows micrometer long dimension and a narrow thickness ( figure 1(b)). High-resolution TEM (HRTEM) image of nanosheets reveals clear lattice fringes implying single-crystalline nature of the nanosheets ( figure 1(c)).
Most of the reported methods on the synthesis of Bi 2 Se 3 NCs relied on the solvothermal method, where tuning of size, shape and composition appears to be challenging [43]. We have synthesized Bi 2 Se 3 nanosheets using colloidal route by rapid injection of bismuth-complex precursor to selenium-complex precursor in presence of high boiling solvents and capping ligands at 180°C. Bismuth acetate was dissolved in pure oleic acid (OLAc) by heating the mixture above 150°C [44]. At this high temperature, there remains a possibility of formation of metallic Bi NCs or bismuth oxide which can be identified by visualizing the clear reaction solution into a black turbid medium [44,45]. This unwanted reaction product may affect the reaction kinetics and the morphology of the final product. Hence, we have prepared bismuth complex precursor at a relatively low temperature (100°C) with the aid of mixture of short chain octanoic acid (OctAc) and octylamine (OctAm) ligands along with OLAc. Reactivity of Se towards different metal salts depends on the nature of Se precursor solution used in the reaction [46]. Different Se precursors have been used so far to synthesize a variety of metal selenide NCs [47].
The use of Se-thiol precursor for the synthesis of bismuth chalcogenides induces immediate formation of metal sulphide NCs below 200°C [48]. Oleylamine (OLAm) and OLAc induced Se dissolution requires much higher reaction temperature above 300°C [49,50]. High temperature dissolution of Se also leads to the formation of small Se NCs [46]. OLAm induced heating of Se powder does not lead to the complete dissolution of Se even at high temperature [51]. In contrary these literatures, our synthesis route is novel since we have used a combination of four ligands (OLAm, OLAc, OctAm, OctAc) along with non-coordinating solvent octadecene to dissolve Se at relatively low temperature (180°C). Injection of Bi-complex into Se-complex leads to immediate TEM image of the aliquot taken after 10 s of the Se precursor injection shows the formation of Bi 2 Se 3 NCs ( figure 2(a)). Size distribution histogram shows that as obtained Bi 2 Se 3 NCs have an average size of ∼2.5 nm ( figure 2(a), inset). TEM image of the aliquots collected at 1 min and 2 min of annealing reveal the coexistence of both Bi 2 Se 3 NCs and nanosheets (figures 2(b), (c)). Size distribution histogram reveals that the Bi 2 Se 3 NCs have an average size of ∼2.9 nm for 1 min anneal sample ( figure 2(b), inset). However, the TEM image of the aliquot collected at 5 min of annealing shows only Bi 2 Se 3 nansheets with large planar area ( figure 2(d)). These observations reveal a fast formation of the single-crystalline Bi 2 Se 3 nanosheets with large planar area, which occurs via 2D oriented attachment of small Bi 2 Se 3 NCs followed by epitaxial recrystallization process [52].
X-ray diffraction (XRD) pattern of the Bi 2 Se 3 nanosheets shows major reflection peaks corresponding to the (006), (101), (015) and (110) planes of the rhombohedral crystal structure of bulk Bi 2 Se 3 ( figure 3(a)). XRD pattern of the Bi 2 Se 3 nanosheets obtianed at different annealing time at 180°C reveals that phase pure Bi 2 Se 3 NCs is formed from the begining of the reaction ( Figure S2, supporting information). We have estimated the thickness of the as-synthesized Bi 2 Se 3 nanosheets from the AFM topography image captured using tapping mode ( figure 3(b)).  The AFM height profile shows a thickness ∼5 nm of nanosheets (inset, figure 3(b)) including the surface attached ligands on the both the top and bottom planes of the nanosheets [10]. Considering the geometric length of the surface attached ligands (OLAc, OLAm, OctAc, OctAm), the actual thickness of inorganic core of the nanosheet can be estimated to be ∼2 nm which corresponds to 2 QLs thickness [10,43].
Selective insertion of copper (Cu 0 or Cu I ) into the Bi 2 Se 3 nanosheets can be achieved by treating the asprepared Bi 2 Se 3 nanosheets with a suitable copper precursor in an appropriate reaction condition. This can be achieved by treating Bi 2 Se 3 nanosheets by Cu I salt at room temperature or by injecting hot Cu I -amine complex into the reaction medium at high temperature. We have adopted both the reaction routes for selective insertion of copper (Cu 0 or Cu I ) into the Bi 2 Se 3 nanosheets. We observed that the use of Cu I -salt at room temperature leads to the formation of Cu 0 intercalated Bi 2 Se 3 nanosheets. Whereas, the use of Cu I -amine complex results in the formation of Cu I substituted Bi 2 Se 3 nanosheets at high temperature reaction. We have systematically studied the Cu 0 intercalation and Cu I substitution into the Bi 2 Se 3 nanosheets using reaction processes both at room and high temperatures.

Preparation of Cu 0 -Bi 2 Se 3 nanosheets
For intercalation zero-valent metal into the layered materials various metal salts have been used which undergoes disproportionation reaction to yield zero-valent metal [34]. However, disproportionation reaction of metal salt depends on the stability of oxidation states of the corresponding metal ion, which depends on the types of metal. Mono-valent copper tends to disproportionate in solution following to the equation The resulting Cu 2+ remains in the solvent as complex, thus releasing Cu 0 in dilute solution [41]. For a typical disproportionation reaction, the NCs, copper salt and solvent are taken into a reaction medium and heated just below reflux temperature. However, heating the reaction close to boiling point temperature of the solvent often damages the morphology of the NCs [42]. We have adopted modified method for disproportionation reaction for intercalation of copper atom into the Bi 2 Se 3 nanosheets in comparison to the previous reports [30,31,34,53]. We performed the reaction at room temperature using two partially immiscible solvents, toluene and methanol, which is advantageous in controlling the rate of intercalation of Cu 0 into the Bi 2 Se 3 nanosheets. We have used air-stable Cu I salt, tetrakis(acetonitrile)copper(I) hexafluorophosphate in toluene and methanol, which undergoes disproportionation redox reaction generating zero-valent copper. As discussed earlier, in the rhombohedral crystal lattice of Bi 2 Se 3 nanosheets, the hexagonal planes of Bi and Se are closely stacked along the c-axis with every five repeat atomic layers in a sequence Se(1)-Bi-Se(2)-Bi-Se(1) forming QLs. The van der Waals gap between two neighbouring Se(1) layers is ∼2.6 Å, which is larger than the radius of a Cu 0 (1.28 Å) ion [54]. This makes Bi 2 Se 3 nanosheets ideal host for the intercalation of zero-valent copper. The amount of zerovalent copper intercalation was controlled by tuning the duration of intercalation and the concentration of copper precursor salt relative to Bi 2 Se 3 nanosheets (see experimental section) [30,31,34,53]. Figure 4(a) shows the TEM images of Cu 0 -Bi 2 Se 3 nanosheets synthesized using 10% tetrakis(acetonitrile)copper(I) hexafluorophosphate (molar percentage). Pristine Bi 2 Se 3 nanosheets in toluene were treated with methanolic solution of tetrakis(acetonitrile)copper(I) hexafluorophosphate for zero-valent Cu 0 intercalation. The TEM image reveals that the morphology of nanosheets remain the same with the pristine Bi 2 Se 3 nanosheets upon zero-valent copper intercalation ( figure 4(a)).
Notably, the use of higher concentration of tetrakis(acetonitrile)copper(I) hexafluorophosphate results in the distortion of the morphology of the pristine nanosheets ( figure S3, supporting information). HRTEM image ( figure 4(b)) of Cu 0 -Bi 2 Se 3 nanosheets shows lattice spacings of 0.35±0.02 nm similar to pristine Bi 2 Se 3 nanosheets, which corresponds to the (101) planes of bulk rhombohedral Bi 2 Se 3 (JCPDS # 12-0732). SAED pattern ( figure 4(c)) of Cu 0 -Bi 2 Se 3 nanosheets shows diffraction spots similar to the pristine Bi 2 Se 3 nanosheets. These results imply that the inherent crystal structure of the Bi 2 Se 3 nanosheets remains unaltered with Cu 0 intercalation. EDX analysis ( figure 4(d)) shows the atomic percentage of Cu:Bi:Se≈12:38:50 supporting intercalation of copper into the Bi 2 Se 3 nanosheets. We have extracted the chemical oxidation states of Cu, Bi and Se of the Cu 0 -Bi 2 Se 3 nanosheets using XPS ( figure 5). Comparison of the XPS spectra of pristine Bi 2 Se 3 nanosheets with Cu 0 -Bi 2 Se 3 nanosheets reveals the effect of copper intercalation into the Bi 2 Se 3 nanosheets. High-resolution XPS spectrum of pristine Bi 2 Se 3 for Bi 4 f shows peaks at 157.3 and 162.6 eV with a splitting of 5.3 eV, which correspond to the Bi 4 f 7/2 and Bi 4f 5/2 respectively ( figure 5(a)).
The position of the both Bi 4f 7/2 and Bi 4f 5/2 peaks are shifted slightly (0.2 eV) to the higher binding energies of 157.5 and 162.8 eV respectively for Cu 0 -Bi 2 Se 3 nanosheets in comparison to the pristine Bi 2 Se 3 ( figure 5(a)). XPS spectrum of Se 3d of Bi 2 Se 3 nanosheets is broad and asymmetric in nature, which is deconvoluted into two peaks ( Figure S4, supporting information). One peak appears at binding energy of 53.1 eV corresponding to the Se 3d 5/2 and the other peak appears at binding energy of 54 eV related to Se 3d 3/2 ( figure 5(b)). The XPS match well with the reported Bi 2 Se 3 nanostructures [34,35]. Cu 0 -Bi 2 Se 3 nanosheets show similar XPS spectrum of Se 3d orbital, however, is shifted about 0.7 eV towards higher binding energy compared to pristine Bi 2 Se 3 nanosheets ( figure 5(b)). The XPS results show that both the Se 3d and Bi 4 f peaks are shifted to higher binding energies upon Cu 0 intercalation suggesting a modified charge density distribution in Cu 0 -Bi 2 Se 3 nanosheets [53,55]. XPS spectrum of Cu 2p orbital of Cu 0 -Bi 2 Se 3 nanosheets shows doublet peaks for 2p 3/2 and 2p 1/2 at binding energies of 932.2 eV and 952.2 eV respectively with a separation of 20.0 eV ( figure 5(c)). The absence of any characteristic satellite peak in between the 2p 3/2 and 2p 1/2 peaks suggests the nonexistence of mono-valent or di-valent states (Cu + or Cu 2+ ) in Cu 0 -Bi 2 Se 3 nanosheets [30]. The XPS analyses confirm the successful intercalation of Cu 0 within the Bi 2 Se 3 nanosheets [34,35].
Furthermore, we have carried out powder XRD measurements Cu 0 -Bi 2 Se 3 nanosheets synthesized with the different amount of Cu 0 and compared with the pristine Bi 2 Se 3 nanosheets ( figure 6(a)). XRD peaks for Cu 0 -Bi 2 Se 3 nanosheets and pristine Bi 2 Se 3 nanosheets appear in the same position for Cu 0 intercalation till 30% suggesting that the host lattice of Bi 2 Se 3 nanosheets remains unaltered till 30% of Cu 0 intercalation. Lower amount of Cu 0 can intercalate in the van der Waals gap between the QLs (figure S5, supporting information). However, a shift towards higher angles is observed for Cu 0 -Bi 2 Se 3 nanosheets with 60% of Cu 0 -intercalation ( figure 6(a)). This observation indicates a change in the lattice spacings for high densities of Cu 0 intercalation. As  mentioned earlier, the use of larger amount of Cu 0 intercalation into Bi 2 Se 3 nanosheets also changes the shape of the nanosheets (figure S3, supporting information). The intercalation of Cu 0 in larger volume causes the lattice contraction, which is in-line observed shift of the XRD peaks towards higher angles. The distortion of lattice may be attributed to a wide range of factors including the volume of the intercalants and the electrostatic interactions between the host Bi 2 Se 3 nanosheets and guest Cu 0 intercalants [34]. UV-vis-NIR absorption spectrum of Bi 2 Se 3 nanosheets shows a broad absorption with a peak at 700 nm ( figure S6, supporting information). UV-vis-NIR absorption spectra of Cu 0 -Bi 2 Se 3 nanosheets show the same absorption peak till 30% Cu 0 intercalation ( figure  S6, supporting information). However, a blue shift to 130 nm is observed for the Cu 0 -Bi 2 Se 3 nanosheets with 60% of Cu 0 intercalation. The observed blue shift of the absorption peak may be attributed to the observed contraction of lattice spacing of the pristine Bi 2 Se 3 nanosheets with larger amount of Cu 0 intercalation.
The Raman spectrum of bulk Bi 2 Se 3 exhibits prominent peaks at 72 cm -1 , 131 cm -1 , and 174 cm -1 , which are assigned to A 1g 1 , E g 2 , and A 1g 2 modes respectively [56,57]. Bi 2 Se 3 nanosheets also show two A 1g modes at ∼69.38 and ∼172.87 cm −1 and a E g mode at ∼129.95 cm −1 ( figure 6(b)). We found that all the peaks are shifted to lower wavenumber by ∼2 cm -1 ( figure 6(b)) for Bi 2 Se 3 nanosheets in comparison to the bulk Bi 2 Se 3 [34]. The A 1g modes correspond to the out-of-plane vibrations parallel to the c-axis and the E g phonon mode corresponds to in-plane bond vibrations perpendicular to the c-axis (figure 6(c)) [34,56,57]. Hence, A 1g mode is more sensitive to thickness since it reflects the out-of-plane vibrations of the Se and Bi atoms, [56] and the interlayer van der Waals interactions influence the effective restoring forces acting on these atoms. The observed shift can be attributed to the lower degree of the in-plane and out-of-plane vibrations in the nanosheets compared to the bulk Bi 2 Se 3 [57]. Cu 0 -Bi 2 Se 3 nanosheets exhibits similar Raman active phonon modes (A 1g 1 , A 1g 2 and E g 2 ) like Bi 2 Se 3 nanosheets ( figure 6(b)). This observation indicates that intercalated Cu 0 forms non bonding interaction with the host Bi 2 Se 3 nanosheets since any bonding would have affected the phonon modes. Additionally, Cu 0 intercalation does not show any new peak in the Raman spectrum owing to the infinitely polarisable nature [34]. This observation also suggests that the intercalated copper remains in zero-valent character. This observations confirm that intercalated Cu 0 remains within the QLs of Cu 0 -Bi 2 Se 3 nanosheets rather than substituting Bi cations from the Bi 2 Se 3 lattice.

Preparation of Cu I -Bi 2 Se 3 nanosheets
Cu I substituted Bi 2 Se 3 nanosheets (Cu I -Bi 2 Se 3 nanosheets) were synthesized by injecting hot Cu(I)-oleylamine precursor solution into the pre-synthesized Bi 2 Se 3 nanosheets at 180°C. Cation or anion exchange route of the pre-synthesized NCs is advantageous since the shape of the resultant NCs can be retained [58,59]. We have adopted the cation exchange method to form Cu I -Bi 2 Se 3 nanosheets by using oleylamine complex of copper which controls the rate of exchange reaction.
TEM image shows that the morphology of pristine Bi 2 Se 3 nanosheets remains the same after Cu I substitution ( figure 7(a)). HRTEM image shows lattice spacings of 0.40±0.02 nm which corresponds to the (101) planes of rhombohedral Bi 2 Se 3 crystal structure (JCPDS # 12-0732, figure 7(b)). This reveals that lattice spacing of (101) plane increase after Cu I substitution in comparison to 0.35±0.02 nm for pristine Bi 2 Se 3 nanosheets. SAED pattern also shows diffraction spots corresponding to the (101) planes of the rhombohedral Bi 2 Se 3 ( figure 7(c)). EDX analysis reveals the atomic ratio of 15:29:56 for Cu:Bi:Se ( Figure S7, supporting information). UV-vis-NIR absorption spectra show a change in the absorption band of Cu I -Bi 2 Se 3 nanosheets compared to the pristine Bi 2 Se 3 nanosheets ( figure 8(a)). Absorption spectra of longer annealed samples reveal that the broad absorption maximum of Bi 2 Se 3 nanosheets gradually disappears upon insertion of copper into the Bi 2 Se 3 nanosheets. Notably, the absorption spectrum of Cu I -Bi 2 Se 3 nanosheets appears different from the absorption spectra of Cu 0 -Bi 2 Se 3 or pristine Bi 2 Se 3 nanosheets (figure S8, supporting information). The substitution host atoms by Cu I influenced Bi 2 Se 3 nanosheets lattice structure thereby changing band gap. Furthermore, we have carried out powder XRD measurements of Cu I -Bi 2 Se 3 nanosheets and compared with the pristine Bi 2 Se 3 nanosheets ( figure 8(b)). XRD peaks for Cu I -Bi 2 Se 3 nanosheets shows new reflections in addition to the peaks of the pristine Bi 2 Se 3 nanosheets suggesting that the host lattice of Bi 2 Se 3 nanosheets is altered upon Cu I substitution. Diffraction peaks shifted to the lower angle for Cu I -Bi 2 Se 3 nanosheets compared to pristine Bi 2 Se 3 nanosheets. This also indicates that the lattice spacing expands upon Cu I substitution and is consistent with the HRTEM observations ( figure 7(b)). Additionally, completely new peaks appear for longer annealing time, which corresponds to the reflection planes of Cu 2-x Se (JCPDS # 03-065-1656) [60]. This observation confirms that Cu I completely substitutes Bi III of Bi 2 Se 3 nanosheets at longer annealing time to yield a phase transition to Cu 2-x Se NCs.
Injection of Cu I -oleylamine precursor to host Bi 2 Se 3 nanosheets initiates partial substitution of Bi III at the beginning of the reaction, which in turn gradually replaces Bi III at longer annealing time. We have compared the XPS spectra of Cu I -Bi 2 Se 3 nanosheets with the Bi 2 Se 3 nanosheets. The XPS spectrum of Cu I -Bi 2 Se 3 nanosheets for Bi 4 f orbital shows that the Bi 4f 7/2 and Bi 4f 5/2 peaks are shifted towards higher binding energy compared to pristine Bi 2 Se 3 nanosheets ( figure 8(c)). The XPS spectrum of Cu I -Bi 2 Se 3 nanosheets for Se 3d orbital shows a broad nature and is shifted towards higher energy of 1.1 eV in comparison to pristine Bi 2 Se 3 nanosheets ( figure 8(d)). XPS analysis of Cu 2p orbital of Cu I -Bi 2 Se 3 nanosheets shows doublet peaks of Cu 2p 3/2 and Cu 2p 1/2 at binding energies of 932 eV and 952 eV respectively ( figure 8(e)). Interestingly, a low intensity satellite peak marked by blue asterisk appears in between the Cu 2p 3/2 and Cu 2p 1/2 peaks at a binding energy of 944 eV ( figure 8(e)). The satellite peak was absent in case of Cu 0 -Bi 2 Se 3 nanosheets or pristine Bi 2 Se 3 nanosheets (figures 5(c) and S9c, supporting information).
The appearance of the low intensity satellite peak confirms the presence of Cu I oxidation state within the Cu I -Bi 2 Se 3 nanosheets [34,35]. These observations indicate that Cu I interacts with the host Bi 2 Se 3 nanosheets by replacing bismuth ion to form Cu I -Bi 2 Se 3 nanosheets. The Raman spectrum of Cu I -Bi 2 Se 3 nanosheets exhibits prominent peaks at 70 cm -1 , 129.5 cm -1 , and 175 cm -1 corresponding to A 1g 1 , E g 2 , and A 1g 2 modes respectively ( figure 9). The out-of-plane phonon mode A 1g 1 and in-plane mode E g 2 show almost similar peak positions compared to the pristine Bi 2 Se 3 nanosheets ( figure 9(a)). However, the out-of-plane phonon mode A 1g 2 shows a shift to longer wavenumber compared to A 1g 2 mode of pure Bi 2 Se 3 nanosheets ( figure 9(a)). The bonding interaction between the mono-valent copper and host Bi 2 Se 3 nanosheets clearly affects the out-of-plane phonon modes. This observation also indicates towards the cation substitution of Bi 2 Se 3 nanosheets by Cu I , which is in sheer contrast with the Cu 0 -Bi 2 Se 3 nanosheets with Cu 0 intercalation. Raman spectra with different annealing time show a gradual shift of the A 1g 2 peak of Cu I -Bi 2 Se 3 nanosheets to longer wavenumber in comparison to the Bi 2 Se 3 nanosheets ( figure 9(b)). Raman spectra with different annealing time also confirm Bi III substitution by Cu I into the Bi 2 Se 3 nanosheets. Prominent peaks of pristine Bi 2 Se 3 tends to disappear for Cu I -Bi 2 Se 3 nanosheets at 5 min

Conclusions
In conclusion, large area and ultrathin 2D Bi 2 Se 3 nanosheets are synthesized using colloidal synthesis method. Formation and growth mechanism of Bi 2 Se 3 nanosheets reveals oriented attachment of small Bi 2 Se 3 NCs followed by epitaxial recrystallization to form single-crystalline nanosheets. We have systematically studied the Cu 0 intercalation and Cu I substitution into the Bi 2 Se 3 nanosheets by choosing appropriate reaction conditions. Selective intercalation of copper Cu 0 into the van der Waals gap between two QLs of Bi 2 Se 3 nanosheets was achieved by treating the as-prepared Bi 2 Se 3 nanosheets with Tetrakis(acetonitrile)copper(I) hexafluorophosphate as copper precursor at room temperature. The Cu 0 intercalated Bi 2 Se 3 nanosheets show similar structural and optical properties like pristine Bi 2 Se 3 nanosheets. However, intercalated Cu 0 in between the adjacent layers of Bi 2 Se 3 nanosheets can be beneficial for supercapacitor applications and fast ionic transport through the 2D channels. On the other hand, the use of Cu I -amine complex results in the formation Cu I substituted Bi 2 Se 3 nanosheets at high temperature reaction conditions. Cu I substituted Bi 2 Se 3 nanosheets was obtained at the shorter reaction time while a complete cation exchange reaction leads to the formation of Cu 2-x Se phase. The development of new synthetic strategy for the synthesis of 2D shape-controlled Bi 2 Se 3 nanosheets and understanding the role of metal ion intercalation or substitution are important for fundamental understanding and energy related applications.