Performance modulation of α-MnO2 nanowires by crystal facet engineering

Modulation of material physical and chemical properties through selective surface engineering is currently one of the most active research fields, aimed at optimizing functional performance for applications. The activity of exposed crystal planes determines the catalytic, sensory, photocatalytic, and electrochemical behavior of a material. In the research on nanomagnets, it opens up new perspectives in the fields of nanoelectronics, spintronics, and quantum computation. Herein, we demonstrate controllable magnetic modulation of α-MnO2 nanowires, which displayed surface ferromagnetism or antiferromagnetism, depending on the exposed plane. First-principles density functional theory calculations confirm that both Mn- and O-terminated α-MnO2 (1 1 0) surfaces exhibit ferromagnetic ordering. The investigation of surface-controlled magnetic particles will lead to significant progress in our fundamental understanding of functional aspects of magnetism on the nanoscale, facilitating rational design of nanomagnets. Moreover, we approved that the facet engineering pave the way on designing semiconductors possessing unique properties for novel energy applications, owing to that the bandgap and the electronic transport of the semiconductor can be tailored via exposed surface modulations.

T he morphology related properties of nanomaterials have attracted growing research interest for generating peculiar properties with great potential for practical innovative applications [1][2][3][4][5] . Crystal facet engineering is known to induce exotic physical and chemical performance in functional materials due to the distorted electronic structure and different exposed ions in the surface layers of inorganic crystals with different exposed planes [6][7][8][9] . Scientific and technological exploration has shown the profound influence of such surface layer in research on catalysis and photocatalysis. Xie et al. found that the (1 1 0) facet exposed Co 3 O 4 nanorods had the ability to catalyze CO oxidation at temperatures as low as 77K, because the (1 1 0) planes expose active Co 31 species at the surface and allow the CO oxidizes at Co 31 sites at such a low temperature 6 . In research on photocatalysis, anatase TiO 2 showed promise for energy and environmental applications if the active (0 0 1) planes were exposed on the surface 7 . Tian et al. synthesized platinum nanocrystals with an unusual tetrahexahedral shape, with the polyhedra enclosed by 24 high-index facets, such as (7 3 0) and (5 2 0) surfaces with high density of atomic steps and dangling bonds. These surfaces exhibit enhanced catalytic activity compared to equivalent conventional Pt surfaces towards electro-oxidation of small organic molecules such as ethanol and formic acid 8 . Thereafter, Zhang et al. investigated the catalytic reaction processes of triiodide reduction over {1 0 0}, {1 1 1} and {4 1 1} facets of Pt, indicating that the activity follows the order of Pt(1 1 1) . Pt(4 1 1) . Pt(1 0 0) using density functional theory 9 . The highest photovoltaic conversion efficiency of Pt(1 1 1) in dye-sensitized solar cells confirms the predictions of their theoretical study with the understanding of the mechanism of triiodide reduction at Pt surfaces 9 . The distorted electronic structure in the surface layer also induces exotic physical phenomena in the conductivity and magnetic coupling. The topological insulator is one such example of unique surface behavior 10 . The electronic band structure in the bulk of a non-interacting topological insulator resembles that of a normal insulator with the Fermi level falling in the gap of the conduction and valence bands. While the surface of a topological insulator shows special states falling within the bulk energy gap and allowing surface metallic conductive behavior. Lu et al. demonstrated the influence of facet effect on electrochemical performance of one-dimensional SnS nanobelts grown along the [0 2 0] direction and expose (1 0 0) facets 11 . The SnS nanobelts also showed unexpected strong photon absorption properties from the ultraviolet to the near-infrared region.
It is expected that facet engineering might be a possible way to modulate magnetism, because the magnetism is determined by the short-range interaction between the magnetic ions. The short-range interaction depends on the bond length, bond angle, and coordination environment of the magnetic ions. Facet engineering can tune these parameters through surface reconstruction to control the magnetic behavior. Indeed, Ohnishi et al demonstrated theoretically that in iron the magnetic moment increases from 0.73 m B /atom in the center layer to 2.98 m B /atom in the surface layer of an Fe (0 0 1) plane 12 . For the transition metal oxides, such as manganese oxides, the bond angle of Mn-O-Mn is 180u in the MnO 6 octahedral environment. The Mn ions are antiferromagnetically coupled through the O superexchange interaction. On the other hand, the bond angle of Mn-O-Mn is 90u in the MnO 4 tetrahedral environment. Mn-Mn can then show ferromagnetic coupling behavior through Heisenberg exchange coupling 13 . Our previous work indicates that a-MnO 2 nanowires exposed (2 1 0) planes on the side walls show extrinsic spin-glass performance with exchange-bias behaviour 14 .
This work demonstrates that the magnetism and electrochemical properties of a-MnO 2 nanowires can be modulated by exposing different crystal planes on the surface. We synthesized two batches of a-MnO 2 -based nanowires, one with exposed (1 1 0) planes on the side walls (defined as MnO 2 -110) and the other with exposed (2 1 0) planes on the side walls (defined as MnO 2 -210). It is interesting that the exposed surfaces of MnO 2 show significant influences on the magnetic and electrochemical properties of the materials. Magnetic measurements clearly demonstrate that MnO 2 -110 is ferromagnetic (FM) and MnO 2 -210 is mainly antiferromagnetic (AFM). Density functional theory (DFT) calculations confirm the different types of surface magnetism in these two samples. Collectively, we demonstrate two distinct sources contributing to the magnetism in the nanostructures: antiferromagnetic ordering in the core region and tuneable surface magnetism, which is mainly attributed to the surface Mn ions. It is also demonstrated that different exposed surfaces endow unique photocatalyst and lithium battery applications of a-MnO 2 nanowires. The energy-related applications of a-MnO 2 nanowires have been studied by taking advantage of the big size of the (2 3 2) tunnels along the c-axe as ion/molecule channels [15][16][17][18][19][20][21] , but the other possibility of their use as magnetic nanowires (MNWs) has been largely unexplored. Inspired by the intriguing structure of a-MnO 2 , naively, one may envisage that by selective cutting of the (2 3 2) tunnels along the planes with low Miller indices, such as (1 1 0) and (2 1 0) 22,23 , the tunnel structure of a-MnO 2 may be opened up, and consequently, different magnetic and electrochemical performances could be obtained. This reveals a possible route towards the selective modulation of the magnetic and chemical properties in nanostructured a-MnO 2 .

Results and Discussion
Phase, microstructure, and valence state. Both samples are high purity a-MnO 2 , as indicated by the X-ray diffraction (XRD) patterns shown in Figure 1 Microstructures of the MnO 2 -110 were observed by scanning electron microscopy (SEM) and by transmission electron microscopy (TEM), as shown in Figure 1(c). The morphology of the sample with (1 1 0) planes exposed consists of ultra-long nanowires with width of 30 nm and length of more than 10 mm, as shown in the inset of the SEM image. The HRTEM images and the selected area electron diffraction (SAED) pattern indicate that the a-MnO 2 nanowires have (1 1 0) planes exposed on the side walls, which is in agreement with the XRD refinement results. MnO 2 -210 has nanowires with a rectangular morphology, with width of ,20 nm and length of ,1 mm, based on SEM and TEM observations (Figure 1(d)). The HRTEM images and the SAED pattern indicate that the a-MnO 2 nanowires have (2 1 0) planes exposed on the side walls, which is in agreement with the XRD refinement results. The HRTEM images of the surface of a single nanowire reveal the highly crystalline nature of the a-MnO 2 nanowires. The different growth speeds of the different planes and the electronic structure are responsible for this structural variation. It should be noted that the tetragonal crystal structure (with space group I 4/m) of a-MnO 2 shows different preferred growth directions in different reaction environments.
The surface sensitive X-ray photoelectron spectroscopy (XPS) technique was employed to examine the valence state of Mn ions in a-MnO 2 nanowires. The survey scan indicates both MnO 2 -110 and MnO 2 -210 are high impurity samples, as shown in Figure S1. The resulting high resolution scans of Mn-2p 1/2 and Mn-2p 3/2 were fitted with four Gaussian-Lorentz peaks, p1-p4, respectively, as shown in Figure 1(e, f), where p1 and p2 are responsible for the observed 2p 1/2 peak of Mn 41 , and p3 and p4 for the 2p 3/2 peak. For Magnetic properties. a-MnO 2 has been reported as an antiferromagnetic substance with a Néel temperature (T N ) of ,24.5 K 24 . Both zero-field-cooled (ZFC) and field-cooled (FC) susceptibility were measured under a 100 Oe magnetic field, and the results are shown in Figure 2(a). The ZFC curve bifurcates from the FC one below ,13 K and shows a peak at ,13 K for both samples. The bifurcation indicates that the magnetic phase is making a transition from paramagnetism (PM) to a spin-glass-like state in the a-MnO 2 nanowires 25 , which is similar to the previously reported transition temperature in a-MnO 2 nanowires 26 . The spin-glass moments are easily polarized under low magnetic field, while the AFM susceptibilities are much lower. This means that that the characteristic of the AFM transition is almost buried in the ferromagnetic cluster ordering. Careful observation can also find the weak AFM transition feature in MnO 2 -110 between 20 and 30 K. The high temperature susceptibility data for a-MnO 2 are in good agreement with the Curie-Weiss law and therefore can be fitted to the equation  with that of MnO 2 in the high temperature region. Interestingly, the T N values in the two samples are similar. This can be understood that the coupling of the majority of the inside atoms in both samples is antiferromagnetic -resembling the case in bulk MnO 2 . Nevertheless, the distinct surface magnetic ground states of (2 1 0) and (1 1 0) show great influences on the Curie-Weiss Temperatures.
The 5 K hysteresis loops of MnO 2 -110 after ZFC or FC under 10 kOe magnetic field from 350 down to 5 K are presented in Figure 2(b), with measurements between 670 kOe. Both the ZFC and the FC loops deviate from antiferromagnetism under magnetic field, showing high remnant magnetism and a strong coercive field, as shown in the upper left inset of Figure 2(b). This provides evidence of a mixed state of a component from the antiferromagnetic core of MnO 2 combined with stable net surface spins. The high remnant magnetism indicates the great amount of net magnetic spin on the surface, and the strong coercive field indicates anisotropic magnetic coupling. The hysteresis loops are not saturated under 670 kOe due to the contribution of the antiferromagnetic core as well as the spinglass component, which is a common phenomenon in the case of nanocrystalline compounds, alloys, and oxide materials 27,28 . The open loop, as shown in the lower right inset of Figure 2 Figure 2(c). Obvious exchange bias behaviour was observed in the FC loop, but is absent from the ZFC loop. H max , the maximum applied magnetic field, is crucial for investigating the exchange bias effect, because small H max may lead to the displacement of the magnetic hysteresis loop, even for FM and glassy magnetic substances. This is attributed to the irreversible magnetization processes known as minor loop effects 29 . When H max is small, the FC hysteresis loops are always shifted towards the negative field and positive magnetization. The M EB value is very small (less than 0.018 emu/g) in the M(H) loop. Thus, the exchange bias effect is indeed present in the a-MnO 2 Comparing the magnetization behaviour of MnO 2 -110 and MnO 2 -210, it is found that the former sample shows much stronger and more stable net magnetic coupling, as shown in Figure 2(d), but does not display strong exchange bias behaviour. The difference is attributed to their surface structures: the (1 1 0) plane contains chains of MnO 6 octahedra on the smooth matrix of (2 3 2) tunnels, while the (2 1 0) plane forms a step-type surface with chains of MnO 6 octahedra. The higher density of MnO 6 octahedral chains in MnO 2 -110 is responsible for its strong magnetization. On the other hand, the weak coupling between the core AFM spins and the surface spins cannot generate an intensive exchange bias when the magnetic field is reversed. In MnO 2 -210, the core AFM spins couple intensively to the weak surface spins during the magnetic field reversal, which is the origin of the exchange bias behavior.
Owing to their inherent shape anisotropy and the ability to incorporate different components, magnetic nanowires (MNWs) offer unique magnetic properties distinct from those of bulks, thin films, and particles [30][31][32][33] . A key property of MNWs lies in the strong coupling of magnetic properties with the nanowire orientations 34 . For practical applications, it is desirable to synthesize nanowires with tuneable magnetic ordering, as they can offer greater flexibility in the design and optimization of nanodevices. Bulk a-MnO 2 has a Hollandite-type structure (tetragonal; space group I4/m; a 5 9.777 Å and c 5 2.855 Å ) 35 . This tunnel-structured oxide is characterized by double chains of edge-sharing MnO 6 octahedra, which are linked at corners to form one-dimensional (1D) (2 3 2) and (1 3 1) tunnels that extend in a direction parallel to the c-axis of the tetragonal unit cell (Figure 3(a)). According to Néel's model 36 , neighboring pairs of octahedrally coordinated Mn 41 (d 3 , S 5 3/2) ions are aligned antiparallel to each other (Figure 3(b)), leading to an antiferromagnetic ground state (see Figure 3(c) and Supplementary Figure S2).
To confirm the influence of the surface on the magnetism, density functional theory (DFT) calculations were performed to distinguish the magnetism of bulk a-MnO 2 from those of exposed (1 1 0) and (2 1 0) surfaces. Bulk a-MnO 2 possesses an antiferromagnetic ordering between the corner-sharing MnO 6 octahedra and a ferromagnetic ordering between the edge-sharing MnO 6 octahedra, as shown in Figure 3  (1 1 0)-Mn, the exchange energy in the outermost surface layer is 533 meV/2Mn. The observed suboptimal weak ferromagnetism is likely to be due to the rough surfaces. Atomically smooth surface preparation is notoriously challenging -especially for chemical growth methods such as we employed in this study. Nevertheless, such intrinsic strong surface ferromagnetism may pave the way to interesting practical applications. To this end, further progress will require novel manufacturing techniques that allow control over nanowires so as to achieve atomic smoothness.
Tuneable electronic bandgap energy. MnO 2 has been demonstrated to be a highly efficient photocatalyst 37 , either alone or in MnO 2 /TiO 2 heterogeneous photocatalysts 38,39 . Figure 4(a) shows the ultravioletvisible (UV-vis) absorption spectrum of the a-MnO 2 nanowires. Broad absorption bands ranging between 300 and 600 nm with peak positions of ,400 nm for MnO 2 -210 and ,450 nm for MnO 2 -110 are observed. The d2d transitions of Mn ions in the a-MnO 2 nanowires is responsible for the absorption in the visible light range. The Mn 3d energy level splits into lower (t 2g ) and higher (e g ) energy levels in the ligand field of MnO 6 octahedra, and the energy difference between the e g and t 2g states is responsible for the optical bandgap energy 40 . The bandgap energy E g for the a-MnO 2 nanowires was estimated using the Kubelka-Munk function to plot the product of the square root of the absorption coefficient and the photon energy against the incident photon energy (hv) 41 . A straight line in a photon energy range close to the absorption threshold can be fitted, as shown in the inset of Figure 4(a). a-MnO 2 nanowires have an indirect electronic transition near the bandgap 41,42 . The bandgap energy for the a-MnO 2 nanowires can be derived as 0.98 eV for the sample with the exposed (1 1 0) planes, while it is 0.84 eV for the sample with exposed (2 1 0) planes, as derived from the intercept of the linear portion with the abscissa. Remarkable differences in the optical properties of nanostructured MnO 2 materials were previously observed. For example, Pereira et al. found that the absorption of MnO 2 colloid at longer wavelengths strongly decreases as the MnO 2 particles become smaller 43  Electrochemical properties. Lithium storage properties of the a-MnO 2 nanowires were investigated using the galvanostatic chargedischarge method. The capacity difference between the two samples is obvious, as shown in Figure 5(a). The origin of the performance variation in a-MnO 2 nanowires can be attributed to the different intercalation/absorption behavior of lithium ions as they interact with exposed (1 1 0) and (2 1 0) surfaces, as illustrated in Figure 5(b). The capacity of the batteries depends on the intercalation Li 1 ions in a-MnO 2 lattice under the charge/discharge voltage. One of the determinant factors is the penetration ability of Li 1 ions in the electrolyte through the close-packed plane of MnO 6 into the (2 3 2) tunnels. The more the (2 3 2) tunnels exposed to electrolyte, the more chance for the Li 1 ions intercalate into the a-MnO 2 lattice. Judging from the theoretical period structure of outmost layers of exposed with (1 1 0) and (2 1 0) plane as demonstrated in Figure S5, the (2 3 2) tunnels have more chance to accept Li 1 ions to build up the capacity of the material. The MnO 6 as blockers on the (1 1 0) and (2 1 0) surfaces were highlighted in Figure S5 to demonstrate the intercalation chance for Li 1 into the (2 3 2) tunnels. It can be roughly estimated that direct exposure rate of the (2 3 2) tunnels to electrolyte is 2/3 when the (1 1 0) plane was the exposed facet. The rate increased to 4/5 for (2 1 0) plane as the exposed facet. This may be one of the reasons that the capacity of MnO 2 -210 is double to that of MnO 2 -110.
In summary, the evidences of facet of nano particles suggest the facet control of nano materials for practical application is essential for the exploration of high performance. The magnetic property dependence on exposed crystal plane of a-MnO 2 nanowires reveals that the variation of the size and morphology dependence of  46 . KMnO 4 (Aldrich, 99.0%) and NH 4 F (Aldrich, 99.99%) were used to form a-MnO 2 under neutral hydrothermal conditions. In a typical procedure, KMnO 4 (0.001 mol) and NH 4 F (0.001 mol) were dissolved under magnetic stirring in 40 mL doubly deionized water to form a clear solution. The solution was transferred into a 50 mL autoclave with a Teflon liner. The autoclave was sealed and maintained at 150uC for 24 h, and then cooled to room temperature naturally. The suspension was then alternately centrifuged with doubly deionized water and ethanol several times, and the resulting brown precipitate was dried in an oven at 80uC for 10 h.  21 . In a typical synthesis, MnCO 3 (Aldrich, 99.9%), (NH 4 ) 2 S 2 O 8 (Aldrich, .98%), HNO 3 (.90%), and H 2 SO 4 (Aldrich, 95-98%) were used as received without further purification. MnCO 3 (0.02 mol) was dispersed in deionized water (200 mL), and HNO 3 (0.04 mol) was then added to make a transparent solution. Then, (NH 4 ) 2 S 2 O 8 (0.02 mol) was added, and the solution was diluted to 300 mL. After the addition was completely dissolved, concentrated H 2 SO 4 (20 mL) was added, and the solution was diluted to 400 mL and stirred for 30 min. The hydrothermal treatment was performed in a Teflon-lined autoclave, with heating at 140uC for 1 hour in a microwave device. After the reaction was completed, the solution was cooled to room temperature, and the resulting suspension was centrifuged in order to separate the precipitate from the supernatant liquid. The precipitate was washed and centrifuged two times and then dried at 80uC overnight.

Synthesis of
Physical characterization. Both samples were microstructurally characterized by Xray diffraction (XRD: GBCMMA, Cu K a , l 5 0.154056 nm) in conjunction with Rietveld refinement (Rietica), UV-Visible spectrophotometer (Cary 5000 UV-Vis-NIR, Agilent), X-ray photoelectron spectroscopy (XPS: EscaLab 220-IXL, Al K a ), field emission gun scanning electron microscopy (FEG-SEM: JSM-6700F), and transmission electron microscopy (TEM: JEOL-2010) with high resolution TEM (HRTEM), operating at 200 kV. Selected area electron diffraction (SAED) patterns were also collected for crystal structure analysis. Magnetic properties were measured using a commercial vibrating sample magnetometer (VSM) model magnetic properties measurement system (MPMS: Quantum Design, 14 T) in applied magnetic fields up to 70 kOe. The nanoparticles were filled into a polypropylene powder holder, which is an injection moulded plastic part as powder container during the VSM measurement process. The polypropylene powder holder was mounted into a brass trough, which is made from cartridge brass tubing with a cobalt-hardened gold plating finish. Both polypropylene powder holder and brass trough were made by Quantum Design as commercial VSM sample holders with very low magnetic moments, which are much lower than the moments of a-MnO 2 samples.
Electrochemical characterisation. Electrochemical characterisation of MnO 2 -110 and MnO 2 -210 was conducted in 2032-type coin cells. The working electrodes were prepared by mixing 80 wt% a-MnO 2 nanowires and 10 wt% carbon black, along with 10 wt% polyvinylidene difluoride (PVdF), in the presence of N-methyl pyrrolidinone (NMP), and this slurry was pasted on aluminium foil and then heat-treated at 80uC under vacuum overnight. CR2032 coin type cells were employed in the battery testing, with lithium foil serving as counter electrode and a porous Celgard polypropylene membrane as separator. The electrolyte consisted of a solution of 1 M LiPF 6 dissolved in a mixture of the solvents ethylene carbonate and dimethyl carbonate in a volume ratio of 151. Galvanostatic charge-discharge measurements were performed over the potential range from 2 V to 4.5 V (vs. Li/Li 1 ) at a constant current density of 20 mA/ g on a Land CT2001A battery tester.
First principles simulation. All calculations were performed using spin-polarised DFT with the generalized gradient approximation 47 (GGA) for the exchangecorrelation functional, as implemented in the all-electron DMol 3 code 48,49 . Earlier study shows that GGA functional predict the correct ground magnetic states for a range of Manganese-oxides 50 . The wave functions are expanded in terms of a double numerical quality localized basis set with a real-space cut-off of 10 bohr. For bulk calculations, the Brillouin zone (BZ) integration was performed using Monkhorst-Pack grids of 12 3 12 3 36 was used. The calculated antiferromagnetic a-MnO 2 lattice constants are a 5 b 5 9.731 Å and c 5 2.854 Å , which compare well with the experimental ones. For surface supercells, a 30-40 Å vacuum region is used between adjacent slabs. All surfaces are fully relaxed, while keeping the innermost three centre layers fixed at the bulk values. The Brillouin zone (BZ) integration was performed using Monkhorst-Pack grids of 8 3 8 3 1, with 18 k points in the irreducible part of the BZ for all the surfaces. The convergence criteria for the forces on the atoms are less than 0.01 eV/Å , and for the total energy 0.05 meV. with different exposed crystal planes. The (2 1 0) exposed sample shows much higher lithium battery performance. (b), Schematic diagram of lithium ions (yellow balls) and their intercalation/absorption performance as they interact with exposed (1 1 0) and (2 1 0) surfaces in electrolyte.