Promising ﬁ eld electron emission performance of vertically aligned one dimensional (1D) brookite ( b ) TiO 2 nanorods

We evidence ﬁ eld-electron emission (FE) studies on the large-area array of one-dimensional (1D) brookite ( b ) TiO 2 nanorods. The pure 1D b -TiO 2 nanorods of 10 nm width and 760 nm long were synthesized on Si substrate utilizing hot- ﬁ lament metal vapor deposition technique. X-ray di ﬀ raction (XRD) and transmission electron microscopy (TEM) analysis evidenced the b -TiO 2 nanorods to be composed of orthorhombic crystals in brookite ( b ) phase. X-ray photoemission spectroscopy (XPS) revealed the formation of pure stoichiometric ( i.e. 1 : 1.98) 1D TiO 2 nanorods. The values of turn-on ﬁ eld, required to draw current density of 10 m A cm (cid:1) 2 , was observed 3.9 V m m (cid:1) 1 for pristine 1D b -TiO 2 nanorods emitters, which were found signi ﬁ cantly lower than doped/undoped 1D TiO 2 nanostructures ( i.e. nanotubes, nanowires, nanorods) based ﬁ eld emitters. The enhanced FE behavior of the TiO 2 /Si emitter can be attributed to modulation of electronic properties due to the high aspect ratio of vertically aligned TiO 2 nanorods. Furthermore, the orthodox emission situation of pristine TiO 2 /Si emitters exhibit good emission stability and reveal their potentials as promising FE material.


Introductions
The one-dimensional (1D) nanostructure morphologies of transition metal oxides offer the advantages of a high aspect ratio, sharp tip features, good stability, and oxidation resistance, 1 which are prerequisites to enhance the eld emission (FE) properties and develop efficient functional devices. Among the various 1D metal-oxides, TiO 2 is one of the most fascinating functional materials. It has received considerable attention for applications in solar cell, 2 photocatalysis, 3 electrochromic displays, 4 supercapacitors, 5 and batteries, 6 etc., owing to its nontoxicity, abundance, and good chemical and thermal stability. The FE properties of TiO 2 are rather explored in spite of its low work function range (3.9 to 4.5 eV) than other popular metal-oxides, owing to its availability in limited morphological forms. 6 The geometrical similarity of TiO 2 nanotubes with carbon nanotubes inuenced many researchers to explore the FE characteristics of TiO 2 nanotubes 7-10 and nanotubes with sharp tips. 11 However, the reproducibility of the TiO 2 nanotubes remained challenging. Not only the diameter inuence the FE performance, but the density (i.e. separation between nanotubes) and clusters appearance of the TiO 2 nanotubes array appear to create eld screening effects. 7 Moreover, the uncontrolled and random dispersion of TiO 2 nanowires 12 and nanorods 13 affect their FE adversely. Highly dense rutile TiO 2 hierarchical nanorods network arranged in the form of dandelion ower like morphology suffer from signicant eld screening effect thereby exhibiting poorer FE behavior. 14 On the contrary, the inuence of oxygen vacancies on the electron affinity alter the electronic structure of TiO 2 and thereof FE properties. 15 Therefore, the doping of 1D TiO 2 nanostructures with N, 9 Fe, 16 and C, 17 etc. was seldom adopted to enhance the FE characteristics.
Theoretical analysis and experimental studies have proved the advantage of brookite (b) phase over anatase and rutile phase of TiO 2 . Mo et al. 18 have reported electronic and optical properties of three different polymorphs (rutile, anatase, and brookite) of TiO 2 . They used rst-principles orthogonalized linear combinations of atomic orbitals (OLCAO) method to calculate the electronic structure of TiO 2 and reported that the brookite phase has a much larger static dielectric constant of 7.89 compared to that of rutile (6.62) and anatase (6.04) phase. Furthermore, comparably better photoactivity is observed in brookite TiO 2 thin lms/nanoparticles than the anatase and rutile. 19,20 Shibata et al. 21 found brookite TiO 2 nanoparticles as excellent coating materials because of its better photoinduced hydrophilicity than anatase one. Koelsch et al. 22 reported better photovoltaic abilities of brookite TiO 2 nanoparticles over anatase phase in terms of bandgap and electrochemical characterization in water and acetonitrile. Furthermore, our detailed report on excellent electrochromic properties with 99% reversibility 4 and highly stable supercapacitive performance with long cycle lifetime (10 000 cycles) of brookite TiO 2 nanoneedles 5 corroborate signicant electronics advantages of the brookite phase over the anatase and rutile phase. However, the brookite consisting of distorted TiO 6 octahedra is less investigated. The synthesis of pure brookite phase is always challenging due to its metastable nature 23 and commonly accompanied with the anatase and/or rutile phases. 3,24,25 That is why numerous scientic reports are found on the utilization of rutile and anatase phases for variety of applications including eld emitters. [10][11][12]26,27 High temperature calcinations 25 and annealing 24 processes were unsuccessful to yield pure brookite phase. Nevertheless, b-phase is thermodynamically most stable in nanocrystalline form at dimensions between 11-35 nm. 3 Despite the facts above, to the best of our knowledge, there is no study report on FE performance of 1D b-TiO 2 nanorods. Therefore, for promising FE behavior, it is of scientic and technological importance to growing vertically aligned 1D b-TiO 2 nanorods.
In this work, we present 1D b-TiO 2 nanorods arrays as promising eld-emitters. The large area arrays of vertically aligned TiO 2 nanorods of brookite phase were synthesized using hot-lament metal vapor deposition (HF-MVD) technique, which is unique and simple technique to provide diverse nanostructures morphologies. [28][29][30][31][32][33] The structural morphology, size distribution, electronic structure, and chemical composition of as-synthesized large-area arrays b-TiO 2 nanorods was examined utilizing various techniques such as X-ray photoemission spectroscopy (XPS), eld-emission scanning electron microscopy (FESEM), transmission electron microscope (TEM), and X-ray diffractometer (XRD), etc. Interestingly, the vertically aligned and uniformly distributed 1D b-TiO 2 nanorods array exhibit low turn-on eld, good emission stability, and large eld enhancement factor.

Experimental
Large area arrays of TiO 2 nanorods were synthesized using hot-lament metal vapor deposition (HFMVD) technique. A clean titanium (Ti) wire (99.9% pure) of a diameter of 1 mm, passed through a pure graphite disc was xed on two supporting Cu electrodes in a vacuum chamber. Once the pressure of the vacuum chamber was pumped down to 1.3 Â 10 À2 mbar, the Ti wire was heated to $1300 C for 25 min to generate controlled hot titanium vapor. The hot titanium vapor encountered and reacted with the residual oxygen (or leaking air) to form a metaloxide vapor of TiO x (x # 2). TiO x vapors meet rather cold Si wafers (substrate) which had been placed on the graphite disc holder ($3 mm above the Ti wire) and condensed into onedimensional (1D) TiO 2 nanorods. Aer that, the surface morphology of the as-synthesized large-area arrays of b-TiO 2 nanorods was characterized using a eld emission scanning electron microscope (FESEM, JEOL JSM-6500F). The crystal structure of b-TiO 2 nanorods was conrmed with an X-ray diffractometer (Philips X'Pert PRO) with Cu-Ka radiations (l ¼ 1.541Å). Structural analysis of b-TiO 2 nanorods was carried out using a transmission electron microscope (TEM, JEOL, JEM-2100) with a selected-area electron diffractometer (SAED) at an accelerating voltage of 200 kV. The band-structure and chemical states present in as-synthesized b-TiO 2 nanorods was analyzed using X-ray photoelectron spectrometer (XPS, Thermo Scientic Inc. K-alpha) with a microfocus monochromated Al Ka X-ray. The individual FE studies of the TiO 2 nanorods were carried out in a vacuum chamber at a base pressure of $10 Â 10 À9 mbar. The semitransparent phosphor screen as an anode was maintained at an optimized distance from specimen/samples of b-TiO 2 nanorods. Moreover, to avoid the effect of contamination, preconditioning of the samples was carried out by applying a voltage of $3 kV for 30 min. The FE current (I) was measured with an electrometer (Keithley 6514) at dc voltage (V) applied using high voltage DC power supply (0-40 kV, Spellman, U.S.). The long-term stability of the FE current was recorded for the TiO 2 nanorods.

Results and discussion
The surface morphology in FESEM images ( Fig. 1) reveals the formation of uniformly distributed and vertically aligned TiO 2 nanorods on Si substrate. Uniformly distributed TiO 2 nanorods over a large area ( Fig. 1(a)) signies variation in their diameters (inset of Fig. 1(a)), which were conned to the very limited range and all are smaller than 20 nm. The array contains $2550 nanorods per square micrometer. The high magnication tilted view ( Fig. 1(b)) of TiO 2 nanorods recorded at a tilt angle of 35 to the surface normal direction shows vertically standing and well separated nanorods with their clearly visible textural boundaries. The high magnication FESEM image in Fig. 1(c) shows a side view of the TiO 2 nanorods array with an average length of $760 nm. This indicates that no distinct thin layer of SiO 2 or TiO 2 nanoparticles has formed before the growth of the TiO 2 nanorods even aer a high temperature of the Ti source (1300 C). A statistical histogram of the diameter distribution ( Fig. 1(d)) illustrates that the diameter of the TiO 2 nanorods falls in the range of 5 to 20 nm. Most of the TiO 2 nanorods have a diameter of about $10 nm. This distribution of varying diameters can be tted to a log-normal distribution function as, XRD pattern is used as a ngerprint to identify the crystalline structure of the TiO 2 nanorods. The room temperature XRD pattern of as-synthesized TiO 2 nanorods arrays ( Fig. 2(a) The remaining diffraction peaks (indicated by asterisks) represents the Si substrate in cubic crystalline form of the space group P2 1 3 with lattice contacts a ¼ b ¼ c ¼ 7.16 nm, and a ¼ b ¼ g ¼ 90 (JCPDS -850621). Further, the crystalline structure of b-TiO 2 nanorods was conrmed by selected area electron diffraction (SAED) pattern ( Fig. 2(b)) of a b-TiO 2 nanorod extracted from large area array. The SAED pattern was collected from the portion of the b-TiO 2 nanorod shown in the inset of Fig. 2(b), which evidenced four-fold symmetric diffraction spots indexed to the [101] zone axis. The indexing of the reections in the SAED pattern is well consistent with the XRD analysis discussed above. This clearly evidenced that the b-TiO 2 nanorods were indeed composed of orthorhombic crystals in brookite (b) phase. Most of the phase transformation results suggest that thermodynamic phase stability for the three polymorphs of TiO 2 is rutile > brookite > anatase. Kominami et al. 25 reported that anatase is thermodynamically stable at crystalline dimensions below 11 nm, brookite is most stable for dimensions between 11-35 nm, and rutile is stable above the 35 nm. Therefore, decisive synthesis parameters (i.e. temperature, time and pressure) of the HFMVD technique were optimized to achieve the diameter of TiO 2 nanorods below 35 nm. Therefore, dimensions of nanorods (i.e. <20 nm) observed in this study ruled out the possibility of formation of rutile phase. The synthesis process involving additional reactants such as water and acid/base medium assists the faster nucleation process than the growth of the morphology (e.g. 1D or 0D) favor for the formation of anatase phase. 8,9,11,17,34,35 Furthermore, faster nucleation process than the growth leads to the lower size morphologies. Synthesis of TiO 2 nanorods is carried out by heating Ti-metal lament at a higher temperature in the absence of additional reactants. Consequently, the formation of b-TiO 2 nanorods is observed, and other phases such as anatase and rutile have been precluded.
XPS studies were carried out for the quantitative analysis of the electronic structure and chemical properties of b-TiO 2 nanorods. To precisely determine the features of the double peaks of Ti (2p 3/2 ) and Ti (2p 1/2 ), the Ti (2p) XPS spectra was decomposed via Voigt curve tting within the Shirley background ( Fig. 3(a)). The perfect t for two peaks located at binding energies (BE) of 458.98 and 464.60 eV, respectively, corresponds to the Ti (2p 3/2 ) and Ti (2p 1/2 ) core levels of Ti 4+ cations and not of Ti 3+ . 36,37 The energy separation of 5.62 eV between Ti (2p 3/2 ) and Ti (2p 1/2 ) peaks and their area ratio of 2.49 reects a strong bonding between Ti and O atoms. The full width at half maximum (FWHM) of the Ti (2p 3/2 ) and Ti (2p 1/2 ) peak are 1.32 and 2.22, respectively, indicative of the high resolution of the Ti (2p) XPS spectrum in comparison with previous studies. 37,38 Similarly, O (1s) XPS spectra of nanorods was decomposed via Voigt curve tting within the Shirley background ( Fig. 3(b)). The results demonstrate the perfect ts to two peaks located at 530.16 and 531.53 eV, with FWHM's of 1.42 and 2.22 eV, respectively. The lower BE peak at 530.14 eV corresponds to the O (1s) core level of the O 2À anions in b-TiO 2 nanorods. However, higher BE at 531.53 eV is ascribed to surface contamination, such as carbon oxides or hydroxides. [39][40][41] The O (1s) peak observed at BE of 530.16 eV, is associated with the Ti-O chemical bonding (O Ti-O 1s ). 36 The atomic    42 This conrms again that, nanorods array is formed of pure stoichiometric TiO 2 . FE measurements of pristine b-TiO 2 nanorods (h 1D b-TiO 2 / Si) were performed in a planar diode conguration. The macroscopic area of the emitting device was $0.30 cm 2 , and values of the anode-cathode separation used were 500, 1000, 1500 and 2000 mm. The variation in the macroscopic electron emission current density (J) as a function of applied electric eld (E) shown in Fig. 4(a). In this work, the applied eld E is dened by E ¼ V/d sep , where V is the voltage applied between electrodes separated by a distance d sep . This eld E is not uniform in our apparatus but is a form of the average eld between the electrodes. The emission current increased rapidly with the gradual increase in applied voltage. Interestingly, the larger emission current density of $470 mA cm À2 was drawn at an applied eld of 6 V mm À1 for the anode-cathode separation of 2000 mm. When the separation increases from 500 to 2000 mm, the turn-on (E on ) eld required to extract emission current density of 10 mA cm À2 , is decreased steadily from 5.6 to 3.9 V mm À1 . These values of E on are much lower than that reported for anodically fabricated TiO 2 nanotip arrays of anatase and rutile phases (i.e. 8.6 to 10.8 V mm À1 ), 11 free-standing TiO 2 nanotube arrays with ridged structures (i.e. 34 V mm À1 ), 2 thermally evaporated single-crystalline TiO 2 nanowires (5.7 V mm À1 ), pristine TiO 2 nanotube arrays (18.86 V mm À1 ), 43 carbon-doped TiO 2 nanotube array (5.0 V mm À1 ), 44 rutile TiO 2 nanorods annealed at high temperatures (3.96-7.46 V mm À1 ), 13 TiO 2 nanotubes/ nanowires of differentiated heights (7.5/6.7 V mm À1 ), 35 TiO 2 nanoowers composed of nanoneedles (4.76 V mm À1 ), 45 and 3D microsphere (4.07 V mm À1 ) of rutile TiO 2 . 46 The E on of b-TiO 2 nanorods is analogous to hydrothermally synthesized quasimicrosphere carrying urchin-like TiO 2 nanostructures of rutile phase (3.81 V mm À1 ). 46 Furthermore, 1D b-TiO 2 nanorods achieved lower E on than the TiO 2 nanotubes anodized in ethylene glycol electrolyte with 2 and 15 vol% H 2 O content (8.8 and 9.1 V mm À1 , respectively) which was reduced with a further increase in the vol% of H 2 O content, 47 but the specic reasons in support of this reduction are unknown yet. Moreover, the E on of b-TiO 2 nanorods is more promising than that obtained from hydrothermally synthesized ZnO nanotubes (7.0 V mm À1 ) at current density of 0.1 mA cm À2 . 48 Relatively lower E on reported for ZnO nanocombs (3.6 V mm À1 ) and Ni-doped ZnO nanowires (3.21 V mm À1 ) is achieved at current density of 1 mA cm À2 and 0.1 mA cm À2 , respectively. 49,50 Furthermore, b-TiO 2 nanorods produced larger emission current density than electrodeposited ZnO nanosheets (50.1 mA cm À2 ) at an applied eld of 6.4 V mm À1 , respectively. 51 The threshold eld (E thr ) correspond to the current density of 100 mA cm À2 was 7, 5.8, 5 and 4.8 V mm À1 , for the separations of 500, 1000, 1500, and 2000 mm, respectively. This decrease in threshold eld with increasing anode-cathode separation is somewhat surprising. For a well dened array of silicon nanotip emitters, similar behavior, was observed by Cheng et al., 52 and claimed that, for a very short anode-cathode distance, where the potential barrier seek by an electron tunneling through a vacuum gap has increased dramatically; and an electron either require more energy or a large applied eld to tunnel through the potential barrier compared with electron emission at large anode-cathode distance. Therefore, except for the electrostatic screening effect depending on the emitter's density and sharpness, the vacuum gap is also important when considering the effect of potential barrier on the transport of electrons.
A modied Fowler-Nordheim (F-N) equation, used to express the emission current density and electric eld relationship of semiconducting nanostructures is as follows, 53 where, J is the device average FE current density, a f is a macroscopic pre-exponential correction factor, a and b are constants (a ¼ 1.54 Â 10 À6 A eV V À2 , b ¼ 6.83089 Â 10 3 eV À3/2 V mm À1 ), F is the work function of the emitter (i.e. 4.3 eV for TiO 2 ), E is the applied average electric eld, b is the local electric eld enhancement factor, and n F is a particular value of the principal Schottky-Nordheim barrier function n (correction factor). The emitter surface for the 1D nanostructures deposited over a larger area is treated to be rough. Therefore, the applied and local electric eld differs from each other at the emission sites, and their ratio is known as the eld enhancement factor (b). The eqn (2) is further analyzed by plotting a graph of ln{J/E 2 } versus (1/E), known as Fowler-Nordheim (F-N) plot. In an approach compatible with eqn (2), the eld enhancement factor (b) is determined from the value of F and the slope S of this FN plot, by using the equation.
where, s is an appropriate value of the slope correction factor for the Schottky-Nordheim barrier. A typical value for s is 0.95, but for simplicity we use the approximation, s ¼ 1 here. Fig. 4(b) shows the F-N plot for b-TiO 2 /Si emitter at various separations between anode and cathode. In most cases, the FN plots exhibit two discrete sections. For low eld values (righthand side) the plot has positive slope; for high eld values (le-hand side) the plot has a negative slope. Fowler-Nordheimtype equations predict that FN plots have a negative slope, so it seems highly probable that the measured low-eld current is not due to cold eld electron emission. Although, similar behavior was also accounted for MoO 2 nanostars composed of nanorods, 54 the exact cause of this low-eld current is not known at present, but one possibility is that it may be some form of leakage current. The increase in the separation between cathode and anode commenced for an increase in eld enhancement factor. The values of b in high eld region are estimated to be 953, 1287, 1271, and 1476 for the cathodeanode separations of 500, 1000, 1500 and 2000 mm, respectively. These values of b for b-TiO 2 /Si emitter are higher than that reported for amorphous and nanoparticle decorated anatase ridged TiO 2 nanotubes, 10 N-and Fe-doped anatase TiO 2 nanotubes, 9,16 rutile TiO 2 nanorods 14 and ZnO nanotubes. 48 To conrm the viability of the eld emission measurements, particularly eld enhancement factor (b) of b-TiO 2 /Si emitters, orthodoxy text was performed using spreadsheet provided by Forbes in the ref. 55. The scaled-barrier-eld (f) values obtained for all cathode-anode separations in b-TiO 2 /Si emitters are shown in Table 1. The emission situation was found orthodox in all cathode-anode separations for both the lower (f low ) and highest (f high ) scaled-barrier-eld values. However, the f high value of 0.55 obtained for the separation of 1000 mm reveals the apparently reasonable emission situation. The individual dispersion, identical smaller diameter ($10 nm), near-perfect vertical alignment, and thoroughly uniform separation between b-TiO 2 nanorods enhances the effective eld at its top, which turns out to give enhancement of FE with better values of b and low E on for b-TiO 2 /Si. For most viable use of eld emitter in a variety of applications, stable FE current is one of the prerequisites. Fig. 4(c) shows the FE stability of b-TiO 2 /Si emitters. The inset of Fig. 4(c) shows the FE image of the b-TiO 2 . The emission current (I)-time (t) behavior recorded at preset current value of 1 mA found no obvious degradation in current within 180 min. The b-TiO 2 /Si emitters exhibit good stability with slight current uctuations of AE15% for average current values. This could be attributed to the excellent thermal and chemical stability of TiO 2 .

Conclusions
In conclusion, the large area arrays of vertically aligned 1D b-TiO 2 nanorods were synthesized on Si substrate utilizing simple and unique HFMVD technique. The formation of pure stoichiometric b-TiO 2 (i.e. Ti : O ¼ 1 : 1.98) nanorods was conrmed by XRD, TEM and XPS analysis. The E on (at 10 mA cm À2 ) of 3.9 V mm À1 was observed for pristine 1D b-TiO 2 nanorods. The individual dispersion, identical smaller diameter, near-perfect vertically alignment, and thoroughly uniform separation between the b-TiO 2 nanorods direct injected electrons toward emission sites and prominently contributed to the low turn-on eld, better FE characteristics and good emission stability. The b-TiO 2 nanorods offer strong potential for applications in FE based vacuum micro-nanoelectronic devices such Table 1 The scaled-barrier-field (f) values for b-TiO 2 /Si emitters obtained using spreadsheet provided by 'Forbes'  as FE displays and an intense 'point' electron sources. Moreover, the FE behavior of the pristine TiO 2 /Si emitters may be enhanced via rational design of hetero-architectures.