Epitaxial highly ordered Sb:SnO2 nanowires grown by the vapor liquid solid mechanism on m-, r- and a-Al2O3

Epitaxial, highly ordered Sb:SnO2 nanowires were grown by the vapor–liquid–solid mechanism on m-, r- and a-Al2O3 between 700 °C and 1000 °C using metallic Sn and Sb with a mass ratio of Sn/Sb = 0.15 ± 0.05 under a flow of Ar and O2 at 1 ± 0.5 mbar. We find that effective doping and ordering can only be achieved inside this narrow window of growth conditions. The Sb:SnO2 nanowires have the tetragonal rutile crystal structure and are inclined along two mutually perpendicular directions forming a rectangular mesh on m-Al2O3 while those on r-Al2O3 are oriented in one direction. The growth directions do not change by varying the growth temperature between 700 °C and 1000 °C but the carrier density decreased from 8 × 1019 cm−3 to 4 × 1017 cm−3 due to the re-evaporation and limited incorporation of Sb donor impurities in SnO2. The Sb:SnO2 nanowires on r-Al2O3 had an optical transmission of 80% above 800 nm and displayed very long photoluminescence lifetimes of 0.2 ms at 300 K. We show that selective area location growth of highly ordered Sb:SnO2 nanowires is possible by patterning the catalyst which is important for the realization of novel nanoscale devices such as nanowire solar cells.


Introduction
Metal oxide semiconductor nanowires (NWs) such as Sb:SnO 2 , 1 Sn:In 2 O 3 , 2 Al:ZnO 3 and In:ZnO 4 NWs have a high conductivity but they are also capable of light emission as shown for Sn:In 2 O 3 NWs by O' Dwyer et al. 5 and Gao et al. 6 Despite ongoing efforts into the growth and properties of such metal oxide (MO) NWs only a few have obtained epitaxial, ordered networks which is essential for the realization of novel nanoscale devices with improved performance like nanowire solar cells (NWSCs). More specically Wan et al. 7 has obtained ordered Sn :In 2 O 3 NWs by homo epitaxy on an Sn:In 2 O 3 buffer layer while Nguyen et al. 8 and Gao et al. 6 have obtained ordered Sn:In 2 O 3 NWs by hetero epitaxy on m-, a-and c-Al 2 O 3 . To the best of our knowledge all other MO NWs that have been obtained previously are not oriented in an epitaxial fashion along any particular direction. It is important then to obtain earth abundant MO NWs such as Sb:SnO 2 NWs as low cost alternatives to Sn:In 2 O 3 .
In the past we have grown SnO 2 NWs via the vapor liquid solid (VLS) mechanism at 800 C and 10 À1 mbar which had a carrier density of the order of 10 16 cm À3 and mobility of 70 cm 2 V À1 s À1 , as determined from THz conductivity spectroscopy. 9 High conductivity SnO 2 NWs have been obtained via the incorporation of Sb, 10 Mo 11 and F 12 in SnO 2 NWs, while recently, Ma et al. 13 showed theoretically that a semiconductor to semimetal transition is possible via the incorporation of Pb in SnO 2 . However, in most cases Sb has been used as an n-type donor impurity in SnO 2 NWs. [14][15][16][17][18][19][20][21][22][23][24] All of the Sb:SnO 2 NWs have been obtained by using metallic Sn and Sb in the past, but they were not ordered.
Nevertheless, it is necessary to point out that Mathur et al. 25 has obtained un-doped SnO 2 NWs via the VLS mechanism that were ordered on TiO 2 (001) which is isostructural with the tetragonal rutile crystal structure of SnO 2 . Similarly, Kim et al. 26 obtained un-doped, epitaxial, SnO 2 NWs on TiO 2 (101) while Leonardy et al. 27 investigated the structural properties of ordered, un-doped, SnO 2 NWs on m-and a-Al 2 O 3 at 700 C by using SnO under a ow of Ar at 10 mbar. Both Mathur et al. 25 and Leonardy et al. 27 obtained SnO 2 NWs which were oriented in two mutually perpendicular directions while the SnO 2 NWs of Kim et al. 26 were aligned in three directions. Others like Mazeina et al. 28 have grown vertical, un-doped SnO 2 NWs via the VLS mechanism on c-Al 2 O 3 at 900 C with limited ordering and a Nanostructured Materials and Devices Laboratory, School of Engineering, University of Cyprus, PO Box 20537, Nicosia, 1678, Cyprus. E-mail: zervos@ucy.ac.cy b Theoretical and Physical Chemistry Institute, National Hellenic Research Foundation, Vass. Constantinou 48, GR-11635 Athens, Greece uniformity. Lateral, but un-doped SnO 2 NWs have also been obtained by Kim et al. 29 and Choi et al. 30 while more recently Wang et al. 31 investigated lateral SnO 2 NWs that were aligned on the surface of m-Al 2 O 3 . All of these studies on epitaxial, ordered un-doped SnO 2 NWs 25-30 focused primarily on their growth and structural properties. It is imperative then to investigate the electrical and optical properties of similar Sb:SnO 2 NWs which is critical in evaluating their potential for subsequent use in devices such as NWSCs.
Here we show that epitaxial, ordered Sb:SnO 2 NWs can be grown via the VLS mechanism on m-, r-and a-oriented Al 2 O 3 only in a narrow window of growth conditions. We describe their morphology, structural, electrical and optical properties, in detail and show that selective area location growth of ordered Sb:SnO 2 NWs is possible which in turn is attractive for the realization of NWSCs as a low cost alternative to Sn:In 2 O 3 NWs.

Epitaxial growth of Sb:SnO 2 NWs
The Sb:SnO 2 NWs were grown using a 1 00 hot wall, low pressure chemical vapour deposition (LPCVD) reactor, capable of reaching 1100 C, which was fed via a micro ow leak valve positioned on the upstream side, just aer the gas manifold which consists of four mass ow controllers. A chemically resistant, rotary pump that can reach 10 À4 mbar was connected downstream. For the growth of the Sb:SnO 2 NWs, metallic Sn and Sb (Aldrich, 100 Mesh, 99.9%) were weighed with an accuracy of AE1 mg. We used an excess of Sb, i.e. a mass ratio of Sn/Sb z 0.1, and the total mass of Sb and Sn was kept xed and equal to 100 mg or 0.1 g. Square samples of 10 mm Â 10 mm c-, m-, r-and a-Al 2 O 3 were cleaned sequentially in trichloroethylene, methanol, acetone, isopropanol, rinsed with de-ionised water, dried with nitrogen and then coated with z1 nm Au. The elemental Sb and Sn as well as the c-, m-, r-or a-Al 2 O 3 substrates were loaded in the same quartz boat which was positioned at the centre of the 1 00 LPCVD reactor. The latter was pumped down to 10 À4 mbar and purged with 1000 sccm of Ar for 10 min at 1 mbar. Subsequently the temperature was ramped up to 800 C at 30 C min À1 using the same ow of Ar. Upon reaching 800 C a ow of 10 sccm O 2 was added to the ow of Ar in order to grow the Sb:SnO 2 NWs over 10 min at 1 mbar, followed by cool down without O 2 . We have grown Sb:SnO 2 NWs on c-, m-, r-and a-Al 2 O 3 using these growth conditions and changed the growth temperature between 700 C to 1000 C.

Characterization of Sb:SnO 2 NWs
The morphology, crystal structure and composition of the Sb:SnO 2 NWs was determined by scanning electron microscopy (SEM), X-ray diffraction (XRD) and Energy Dispersive X-ray analysis (EDX). High resolution transmission electron microscopy (HRTEM) was carried out using a TECNAI F30 G 2 S-TWIN operated at 300 kV. The optical properties of the Sb:SnO 2 NWs were determined by steady state and transient absorptiontransmission spectroscopy. The steady state and time resolved photoluminescence (PL) were also measured between 10 K and 300 K, while the electrical properties, i.e. carrier density and resistivity, were measured by the Hall effect in the Van der Pauw geometry similar to Costa et al. 23,24 In particular the Sb:SnO 2 NWs were transferred from the m-, r-or a-Al 2 O 3 onto 10 mm Â 10 mm c-Al 2 O 3 by applying pressure. This results into a dry transfer of the ordered Sb:SnO 2 NWs onto the c-Al 2 O 3 and the formation of a planar interconnected network. We then deposited In contacts over the Sb:SnO 2 NWs by thermal evaporation using a shadow mask. The Sb:SnO 2 NWs on c-Al 2 O 3 was not heated up during the deposition, and the In contacts had diameters of z1 mm at the four corners of the 10 mm Â 10 mm c-Al 2 O 3 , but we did not anneal them. The Hall effect was measured using a GMW3470 Electromagnet at 0.3 Tesla. The magnetic eld was calibrated with a Hirst GM08 Gaussmeter. A Keithley 2635 A current source and Keithley 2182A nanovoltmeter, controlled by Lab View were used to provide a current and measure the voltages.

Results and discussion
In the past, we have shown that the reaction of Sn with O 2 at 800 C and 10 À1 mbar results into a high yield and uniform distribution of SnO 2 NWs on Si (001) or fused SiO 2 . The SnO 2 NWs have average diameters of z50 nm, lengths up to z100 mm and grow by the VLS mechanism whereby Sn enters the Au catalyst particles on the surface of Si (001) or fused SiO 2 and forms liquid Au:Sn particles. Upon saturation, solid SnO 2 forms beneath the liquid Au:Sn particles via the reaction with O 2 at the triple phase junction, as shown in Fig. 1(a), leading to one dimensional, bottom-up growth. 32 However, the SnO 2 NWs obtained on Si (001) or fused SiO 2 were not oriented or ordered along any direction, and had a carrier density of the order of 10 16 cm À3 with a mobility of 70 cm 2 V À1 s À1 , as determined from THz conductivity spectroscopy. 33 Hence doping is required to increase their conductivity. Recently, we showed that higher carrier densities of the order of 10 18 to 10 19 cm À3 may be readily obtained in Sb doped SnO 2 NWs grown on both Si (001) and fused SiO 2 . 34 We obtained a high carrier density of 4 Â 10 19 cm À3 in the Sb:SnO 2 NWs grown on c-Al 2 O 3 at 800 C and 1 mbar by adding metallic Sb to Sn but the Sb:SnO 2 NWs were not ordered or oriented in any particular direction, similar to those obtained previously on Si (001) and fused SiO 2 . 34 This is in contrast to the ndings of Mazeina et al., 28 who obtained vertical but un-doped SnO 2 NWs via the VLS mechanism on c-Al 2 O 3 at 900 C with a limited degree of ordering and uniformity. Similarly, we did not obtain ordered Sb:SnO 2 NWs on c-Al 2 O 3 by changing the growth temperature between 800 C and 1000 C. The epitaxial growth and ordering of Sb:SnO 2 NWs on c-Al 2 O 3 is not favorable in view of the fact that SnO 2 has a tetragonal crystal structure which will not match the hexagonal crystal structure of the underlying c-Al 2 O 3 . However, we observed the formation of ordered Sb:SnO 2 NWs on the sides of the c-Al 2 O 3 , as described in more detail in the ESI, † which is related to its specic crystallographic orientation. Therefore we carried out the growth of the Sb:SnO 2 NWs on m-, r-and a-Al 2 O 3 at 800 C and 1 mbar using the same growth conditions described above.   but their Sb:SnO 2 NWs were not aligned in any particular direction. The formation of the second segment can be prevented by reducing the growth time as will be described in more detail later. Similarly we nd that the Sb:SnO 2 NWs on a-Al 2 O 3 shown in Fig. 4(a-c) consist of two segments, similar to those obtained on r-Al 2 O 3 .
In all cases, we obtained a uniform distribution and ordering of Sb:SnO 2 NWs over the 10 mm Â 10 m-, r-and a-Al 2 O 3 in reproducible way, using Sn/Sb ¼ 0.1 at 800 C and 1 mbar as described above. Subsequently we varied the growth temperature between 700 C to 1000 C in order to nd if any changes occur in the growth directions and ordering. The Sb:SnO 2 NWs obtained at 900 C on m-Al 2 O 3 were very similar to those obtained at 800 C, but we nd that the Sb:SnO 2 NWs obtained at 1000 C are short, as shown in Fig. 2(c), due to the fact that the Sn and Sb are more or less completely transferred into the gas stream under the ow of Ar during the temperature ramp, before the onset of one dimensional growth. In fact, the depletion of Sb during the temperature ramp is more signicant than Sn, due to the fact that Sn has a melting point of 232 C and  Paper vapor pressure of 10 À5 mbar, while Sb has a higher melting point of 630 C, but a remarkably higher vapor pressure of 10 À1 mbar at 1000 C. This in turn implies an important doping limitation when trying to obtain high carrier densities and conductivity in Sb:SnO 2 NWs using metallic Sn and Sb. 34 Likewise, the Sb:SnO 2 NWs obtained on m-Al 2 O 3 at 700 C were short, similar to those obtained at 1000 C, due to the limited supply of Sn which has a lower vapor pressure at 700 C, but they remained orthogonally oriented to each other. No changes occurred in the morphology and growth directions of the Sb:SnO 2 NWs on r-and a-Al 2 O 3 by varying the temperature between 700 C and 1000 C. In contrast, the epitaxial growth and ordering of the Sb:SnO 2 NWs was critically dependent on the mass ratio of Sn/Sb. We obtained ordered Sb:SnO 2 NWs on m-, r-and a-Al 2 O 3 only for Sn/Sb ¼ 0.15 AE 0.05. One dimensional growth was completely suppressed when Sn/Sb < 0.1, and the Sb:SnO 2 NWs were not oriented in any particular direction for Sn/Sb > 0.2. Before elaborating further, it is important to mention that we did not obtain any Sb:SnO 2 NWs at all without adding O 2 to the ow of Ar during the growth step. In other words, the residual or background O 2 aer purging and the temperature ramp was not signicant, and the growth of the Sb:SnO 2 NWs occurred solely due to the oxygen supplied during the growth step.
Consequently the suppression of one dimensional growth for Sn/Sb < 0.1 is attributed to the total depletion and transfer of Sn into the gas stream during the temperature ramp, before the onset of growth, as it forms a liquid alloy with Sb. Note also that we did not obtain any SbO 2 , Sb 2 O 3 or Sb 2 O 5 NWs by using only Sb.
On the other hand the Sb:SnO 2 NWs were not oriented in any particular direction for Sn/Sb > 0.2. From the above it is clear that an excess of Sb mixed with Sn provides exible control over the Sn supplied to the Au particles which must be carefully tuned to a minimum for epitaxial growth and ordering to occur. This is further corroborated by the fact that we did not obtain ordered Sb:SnO 2 NWs without mixing the Sb and Sn even when Sn/Sb z 0.15 AE 0.05.
In addition to the above it is important to point out that the epitaxial growth and ordering of the Sb:SnO 2 NWs also depends critically on the crystal quality of the m, r and a-Al 2 O 3 surfaces. It has been shown that m-Al 2 O 3 is thermodynamically unstable during high-temperature growth and nanostructured grooves composed of s-and r-facets develop along the [11 20] direction. 35 In contrast Wang et al. 31 claimed that both the a-and r-plane of Al 2 O 3 retain their surface structure at elevated temperatures. We annealed the m-, r-and a-Al 2 O 3 at 1000 C for 30 min under Ar and O 2 without any Sn, Sb or Au and observed a drastic reduction in the strength of the XRD peaks of the m-and a-Al 2 O 3 but the r-Al 2 O 3 appeared to maintain its surface crystallinity. Subsequently we deposited 1 nm of Au over the pre-annealed m-, r-and a-Al 2 O 3 and tried to grow epitaxial, ordered Sb:SnO 2 NWs at 800 C and 1 mbar using Sn/Sb ¼ 0.1. As expected, we obtained ordered Sb:SnO 2 only on r-Al 2 O 3 . It appears then, that the deposition of 1 nm Au over pristine m-, r-, and a-Al 2 O 3 at room temperature, prevents in some way the deterioration of the surface crystal structure and allows the epitaxial growth of ordered Sb:SnO 2 NWs at elevated temperatures between 700 C to 1000 C. In fact, the deposition of a 1 nm Au layer on the m-, r-, and a-Al 2 O 3 , which contain grooves or steps along specic crystallographic directions, leads to instabilities and ruptures of the Au at elevated temperatures as described by Hughes et al. 36 These ruptures occur at high curvature sites, i.e., peaks and ridges, which act as retracting edges leading to a net ux of atoms away from the high positive curvature regions. For sufficiently thin layers, this process exposes the texture or steps of the underlying substrate, and a self-assembly of the Au particles will occur along specic crystallographic orientations. 37 This in turn will instigate one dimensional epitaxial growth along specic lateral crystallographic directions via the VLS mechanism, as shown in Fig. 1(b). When Sn is added to the Au particles, it is expected to reduce the surface tension and contact angle q with the underlying m-, r-and a-Al 2 O 3 surface, as a consequence of the fact that Sn and Sb have surface tensions of z500 mN m À1 and 350 mN m À1 respectively, but the surface tension of Au is z1000 mN m À1 . A large contact angle implies that the contact area is small, and vice versa, a smaller contact angle implies a larger contact area. Consequently, an excess of Sn is expected to lead to the formation of Au-Sn particles having a small contact angle and larger contact area with the underlying m, r and a-Al 2 O 3 surface, in which case they might not be able to follow the variations in the surface topography which is necessary to obtain ordered Sb:SnO 2 NWs. The formation of ordered Sb:SnO 2 NWs is possible due to a reduction of the Sn by the excess of Sb during the temperature ramp, which in turn results into sufficiently small Au-Sn liquid particles that are able to follow steps or grooves on the surface during the growth step at elevated temperatures. Epitaxial growth, then, commences laterally, aer which a transition to inclined growth occurs leading to the formation of the ordered networks of Sb:SnO 2 NWs shown in Fig. 2, 3 and 4.
All of the Sb:SnO 2 NWs on m-, r-and a-Al 2 O 3 exhibited clear and well resolved peaks in the XRD, as shown in Fig. 2(d), 3(d) and 4(d) respectively, corresponding to the tetragonal rutile crystal structure of SnO 2 . For comparison, we have included the XRD of the m-, r-and a-Al 2 O 3 without the Sb:SnO 2 NWs. More specically, the Sb:SnO 2 NWs on m-, r-and a-Al 2 O 3 exhibit only one or two major peaks in their XRD, consistent with the fact that they grow along specic crystallographic directions. In contrast, we observed a multitude of major peaks from the Sb:SnO 2 NWs on c-Al 2 O 3 , due to the fact that they do not grow in an epitaxial fashion along specic directions, see ESI. † We did not observe any peaks related to oxides of Sb such as Sb 2 O 3 or Sb 2 O 5 which have melting points of 656 C and 380 C respectively. In addition, we do not observe any peaks suggesting the formation of Sb 2 O 4 , i.e. SbO 2 which is known to break down into Sb and O 2 at a higher temperature of 930 C. Nevertheless the Sb:SnO 2 NWs contain Sb donor impurities as shown by the EDX spectrum in Fig. 2(e) and as conrmed previously by Raman spectroscopy. 34 Small differences in the amount of metallic Sb, i.e. for Sn/Sb ¼ 0.15 AE 0.05, did not change the crystal structure or orientation of the Sb:SnO 2 NWs. We obtained exactly the same XRD spectra shown in Fig. 2, 3 and 4 for Sn/Sb ¼ 0.1 and Sn/Sb ¼ 0.15. However, it is important to point out that we did not detect any Sb in the Au particles on the ends of the Sb:SnO 2 NWs, as shown by the EDX spectrum in Fig. 3(e). This is consistent with the fact that we did not nd Sb in the Au aer trying to grow Sb 2 O 3 , Sb 2 O 5 or SbO 2 NWs, using just Sb, and leads us to suggest that the Sb donor impurities are incorporated into the SnO 2 NWs by surface diffusion, from their sides, as depicted in Fig. 1(c). Now, the Sb:SnO 2 NWs on m-Al 2 O 3 exhibited one dominant peak in the XRD, as shown in Fig. 2(d), corresponding to the (002), i.e. a multiple of (001), crystallographic planes of tetragonal rutile SnO 2 . The two dimensional lattice of (001) SnO 2 and the oxygen terminated surface of m-Al 2 O 3 are shown in Fig. 5 Fig. 2(a)-(c). This has also been conrmed by Wang et al. 31 who showed that un-doped SnO 2 NWs also grow laterally along two perpendicular directions and cross each other on m-Al 2 O 3 .
In contrast to the above we nd that the Sb:SnO 2 NWs on r-Al 2 O 3 exhibited two dominant peaks in the XRD, as shown in Fig. 3(d), corresponding to the (101) and (202) crystallographic planes of tetragonal rutile SnO 2 . The two dimensional lattice of (101) SnO 2 and the oxygen terminated surface of r-Al 2 O 3 are shown in Fig. 6(a) and (b) respectively. These have a larger lattice mismatch of 11%, and the in-plane epitaxial relationship is SnO 2 (101)kr-Al 2 O 3 . The growth of the Sb:SnO 2 NWs on r-Al 2 O 3 is very similar to the un-doped SnO 2 NWs of Kim et al. 38 which were also inclined at q ¼ 68 on r-Al 2 O 3 . Lateral SnO 2 NWs have also been obtained via the VLS mechanism by Kim et al. 29 on r-Al 2 O 3 using C and SnO 2 as opposed to Sn but they did not observe the transition of growth from lateral to vertical SnO 2 NWs.

Selective area location growth of Sb:SnO 2 NWs on r-Al 2 O 3
The VLS growth mechanism permits selective area location growth on m-, r-and a-Al 2 O 3 as we did not obtain any Sb:SnO 2 NWs without using Au. We obtained hexagonally ordered Sb:SnO 2 NWs on r-Al 2 O 3 by drop casting $10 ml of 9 mm diameter polystyrene spheres on r-Al 2 O 3 , followed by the deposition of a thin layer of $1 nm Au as shown in Fig. 7. Subsequently, the spheres were removed in isopropanol by ultrasonic vibration for 1 min, and the Sb:SnO 2 NWs were grown on the patterned Au on r-Al 2 O 3 at 800 C. The Sb:SnO 2 NWs on r-Al 2 O 3 do not consist of two segments due to the reduced growth time. One may clearly observe that the Sb:SnO 2 NWs grow on the r-Al 2 O 3 in a hexagonal pattern suggesting that one may also obtain different geometries in order to tailor the absorption-transmission spectrum in novel devices such as NWSCs.

Electrical properties of Sb:SnO 2 NWs on m-, r-and a-Al 2 O 3
The carrier density of the Sb:SnO 2 NWs grown on m-, r-and a-Al 2 O 3 at 800 C and 1 mbar with Sn/Sb ¼ 0.1 was measured by the Hall effect. We obtained a carrier density of 8 Â 10 19 cm À3 Paper which is signicantly larger than that measured previously in un-doped SnO 2 NWs, that was of the order of 10 16 cm À3 . 33 The conductivity of the Sb:SnO 2 NWs was found to be z3 Â 10 2 (U cm) À1 giving a mobility of 20 cm 2 V À1 s À1 which is lower compared to 70 cm 2 V À1 s À1 in un-doped SnO 2 NWs that was previously measured by THz conductivity spectroscopy. 33 We did not observe a signicant variation in the carrier density of the Sb:SnO 2 NWs on m-, r-and a-Al 2 O 3 , or due to slight variations of Sn/Sb ¼ 0.15 AE 0.05. We measured the Hall effect of the Sb:SnO 2 NWs obtained with Sn/Sb ¼ 0.15 and found a carrier density of 8.3 Â 10 19 cm À3 which is very close to that obtained with Sn/Sb ¼ 0.1. However the carrier density in the Sb:SnO 2 NWs obtained at 900 C was smaller i.e. 6 Â 10 18 cm À3 while those obtained at 1000 C had an even smaller carrier density of 4 Â 10 17 cm À3 and conductivity of z3 (U cm) À1 . This is attributed to the transfer of the Sb into the gas stream during the temperature ramp, and before the onset of one dimensional growth, due to the high vapor pressure of Sb, which in turn limits the supply and incorporation of Sb impurities into the SnO 2 NWs during the growth step. This trend is consistent with  the ndings of Klamchuen et al., 22 who found that the doping level in SnO 2 NWs grown at 650 C was twice that obtained at 750 C, attributed to a suppression of impurity re-evaporation. They also observed that the doping level did not increase further upon reducing the growth temperature from 650 C to 550 C, which was explained by a suppression of the diffusion length of impurity ad-atoms on the surface of the SnO 2 NWs, and the presence of a temperature activated energy barrier which is necessary for the incorporation of the Sb impurities into the host lattice of SnO 2 .
We suggest that the incorporation of Sb impurities into the SnO 2 NWs occurs through the sides of the SnO 2 NWs by thermal diffusion, or, via the triple phase junction, as shown in Fig. 1(c), similar to the mechanism proposed by Klamchuen et al. 22 This is also corroborated by the fact that we did not detect any Sb in the Au particles aer attempting to grow SbO 2 , Sb 2 O 3 or Sb 2 O 5 NWs on r-Al 2 O 3 , by using only Sb, and also by the fact that we did not detect any Sb in the Au:Sn particles on the ends of the Sb:SnO 2 NWs, as shown in Fig. 3(e). It is also consistent with the ndings of McGinley et al., 39 who showed that the surface of SnO 2 nanoparticles is terminated by an oxygen rich layer, but when doped n-type with 9% or 17% Sb, the impurity atoms are concentrated near the surface of the SnO 2 nanoparticles with an oxidation state of ve.
In order to obtain a more detailed understanding of the electronic properties of the Sb:SnO 2 NWs we carried out electronic structure calculations from rst principles using the CASTEP plain wave DFT code 40,41 and the Heyd-Scuseria-Ernzerhof (HSE) 42 exchange correlate-correlation energy functional. Aer relaxation of the cell, we obtained lattice constants of a ¼ b ¼ 4.82Å and c ¼ 3.23Å which are in close agreement to reported values. 42 DOS calculations were performed for the case of (a) the perfect SnO 2 cell and (b) the Sb-doped SnO 2 cell, shown in Fig. 8(a) and (b) respectively. The maximum of the valence band (VB) is set at zero energy level. It is evident from the partial density of states shown in Fig. 8(a) and (b) that the VB is dominated by O and the CB by Sn. 60 The band gap was found to be 3.6 eV for the perfect structure but a slight reduction of the band gap down to 3.5 eV was observed with doping. This semiempirical method yields a much more accurate band gap than calculations performed based on the Perdew-Burke-Ernzerhof (PBE) GGA functional which are not shown here. In addition we nd that the incorporation of Sb into the SnO 2 crystal does not produce any deep levels within the band gap which are in general detrimental to the operation of optoelectronic devices such as solar cells, light emitting diodes etc. In other words the Sb impurities are incorporated into the SnO 2 lattice as substitutional donors and our calculations show that the Fermi level resides 0.46 eV above the CB edge for 12.5 at% Sb (h4 Â 10 21 cm À3 ).
In addition, we calculated the conduction band (CB) potential prole, and one dimensional electron gas (1DEG) charge distribution, along the radial direction, via the self-consistent solution of the Poisson-Schrödinger (SCPS) equations, in the effective mass approximation, as described in detail elsewhere. 43 The SCPS calculations were carried out by taking into account the effective mass and dielectric constant of SnO 2 , i.e. m * e ¼ 0:3 (ref. 44 and 45) and 3 r ¼ 13.5 (ref. 46 and 47) respectively. In order to obtain a carrier density of 8 Â 10 19 cm À3 as determined from the Hall effect we have taken the Fermi level to be energetically located z0.3 eV above the CB edge, at the surface of the Sb:SnO 2 NWs. This is consistent with the electronic structure calculations of Mishra et al., 48 who showed that the Sb impurities in SnO 2 form a band, which has an energetic overlap with the conduction band and a nearly free electron structure, i.e. behaves like a metallic band. It is also consistent with Li et al., 49 who calculated the electronic structure of Sb:SnO 2 for 6.25% Sb, and found that the Fermi level moves into the CB upon the incorporation of Sb, and displays metallic character, but also with Farahani et al., 50 who showed that the Fermi level resides z0.33 eV above the CB edge at the surface of Sb:SnO 2 epitaxial layers grown by molecular beam epitaxy on r- The CB potential prole with respect to the Fermi level, i.e. E C À E F and 1DEG charge distribution, versus distance along the radial direction of the Sb:SnO 2 NWs, is shown in Fig. 8(d), where we have taken the donor impurities to be conned between r ¼ 20 and 40 nm. The CB edge potential prole is near at band in the vicinity of the core and the 1DEG charge distribution has a maximum at the core and a local maximum at the surface. It should be noted that we obtain similar band proles and charge distributions by taking a uniform distribution of Sb impurities throughout the Sb:SnO 2 NWs. However the 1DEG in the vicinity of the core is expected to have a higher mobility when the density of donor impurities is larger at the surface than the core, as is the case in Fig. 9(d). In other words, the incorporation of Sb impurities into the SnO 2 NWs via surface diffusion is not a drawback in the end. We estimate that the mobility in the Sb:SnO 2 NWs with a maximum carrier density of the order of 10 20 cm À3 is a few tens of cm 2 V À1 s À1 (ref. 23 and 24) so the resistivity of the Sb:SnO 2 NWs is of the order of 10 À3 U cm which in turn is attractive for the deposition of metal contacts with low resistance and the fabrication of high performance devices.

Optical properties of Sb:SnO 2 NWs on m-, r-and a-Al 2 O 3
In the past the transmission of light through Sb:SnO 2 NWs 14-24 has not been investigated as they were not ordered or oriented along any particular direction which in turn leads to a considerable suppression of transmission and transparency. In contrast a higher transmission of light is expected to occur through ordered MO NWs. We measured the steady state transmission through the Sb:SnO 2 NWs that were grown at 800 C on r-Al 2 O 3 as shown in Fig. 9(d). The Sb:SnO 2 NWs have an optical transmission of 80% above 800 nm but they absorb light in the visible between 400 nm to 800 nm. We do not observe any interference effects since the spacing between the Sb:SnO 2 NWs is considerably smaller than the wavelength of light. In addition we nd that the Sb:SnO 2 NWs on r-Al 2 O 3 obtained at 900 C and 1000 C have a higher transparency compared to those obtained at 800 C due to their shorter lengths. In order to understand the origin of the absorption we have measured the transient absorption through the Sb:SnO 2 NWs on an fs time scale as shown in Fig. 9(a)-(c). In all cases, we observe a strong peak in the time evolution of the differential absorption between 500 nm and 600 nm attributed to the surface plasmon resonance (SPR) of the Au-Sn particles, corresponding to sizes between 10 nm and 100 nm (ref. 51) and the existence of surface states lying energetically in the energy gap of SnO 2 . 52 One may observe a slight blue shi of the peak from 900 C to 1000 C in Fig. 9(b) and (c), consistent with that observed in the steady state transmission, that might be related to the smaller size of the Au-Sn particles due to the depletion of Sn that occurs during the temperature ramp and/or the elimination of mid gap states at elevated temperatures. It is important to point out, that the transparency of the Sb:SnO 2 NWs on r-  Al 2 O 3 may be increased further by selective area location growth as shown in Fig. 7 which leads to a higher transmission through the voids between the Sb:SnO 2 NW.
Finally it is worthwhile considering the PL obtained from the Sb:SnO 2 NWs, grown on r-Al 2 O 3 at 800 C, as shown in Fig. 10(a), in which case we observe emission at l ¼ 600 nm (h2.1 eV) and 300 K. Bulk SnO 2 has a direct energy band gap of 3.7 eV but the evenparity symmetry of the conduction-band minimum and valence-band maximum states prohibits bandedge radiative transitions which has limited the use of SnO 2 for the fabrication of light emitting diodes. The PL at 2.1 eV is attributed to radiative recombination between deep donor and acceptor like states residing energetically in the energy band gap of SnO 2 that are related to oxygen vacancies. We observe a suppression of the maximum at l ¼ 600 nm and the emergence of emission at l ¼ 470 nm (h2.6 eV) by decreasing the temperature from 300 K to 10 K due to radiative recombination via shallower levels as proposed by Luo et al. 53 However an interesting aspect of the PL emission at 2.1 eV and 2.6 eV is that it has a lifetime of s z 0.2 ms as shown by the time resolved PL in Fig. 10(b). This is considerably higher than the lifetimes extracted from SnO 2 rods and particles which are of the order of 100 ns (ref. 54) and comparable to Eu doped SnO 2 nanocrystals. 55 Hence, in principle, the Sb:SnO 2 NWs described here may be processed into devices capable of light emission but also NWSCs. 56,57 To the best of our knowledge no one has previously used ordered networks of Sn:In 2 O 3 or Sb:SnO 2 NWs to make NWSCs despite the fact that Battaglia et al. 58 has showed that periodic photonic nanostructures outperform their random counterparts in trapping light in solar cells. It is desirable then to use these highly conductive, ordered networks, of Sb:SnO 2 NWs in order to improve the performance of all-solid state NWSCs. In addition the Sb:SnO 2 NWs can be used in perovskite solar cells as an electron transport layer (ETL). According to Jiang et al. 59 a traditional ETL such as TiO 2 , is not very efficient for charge extraction at the interface, especially in planar structures. In addition, the devices using TiO 2 suffer from serious degradation under ultraviolet illumination. SnO 2 shows a better band alignment with the perovskite absorption layer and higher electron mobility, which is helpful for electron extraction. Consequently the specic ordered networks of Sb:SnO 2 NWs described here may serve as a scaffold on top of which one may deposit a perovskite absorber layer and a hole transport layer in order to create a p-n junction solar cell.

Conclusions
We have grown epitaxial, ordered Sb:SnO 2 NWs via the VLS mechanism on m-, r-and a-Al 2 O 3 between 700 C and 1000 C using metallic Sn containing an excess of Sb i.e. for Sn/Sb ¼ 0.15 AE 0.05 under a ow of Ar and O 2 at 1.5 AE 0.5 mbar. One dimensional growth was suppressed for Sn/Sb < 0.1 while the Sb:SnO 2 NWs were not oriented along any particular direction for Sn/Sb > 0.2. Consequently highly conductive and directional Sb:SnO 2 NWs may only be obtained in a narrow window of growth conditions. All of the Sb:SnO 2 NWs have the tetragonal rutile crystal structure and square sections. The Sb:SnO 2 NWs are oriented along two mutually perpendicular directions forming a rectangular mesh on m-Al 2 O 3 with a maximum lattice mismatch of 0.1%. In contrast the Sb:SnO 2 NWs on r-Al 2 O 3 are all oriented in one direction but have a larger lattice mismatch of 10%. The morphology and growth directions of the Sb:SnO 2 NWs on m-, r-and a-Al 2 O 3 did not change by varying the growth temperature between 700 C and 1000 C but the carrier density changed from 8 Â 10 19 cm À3 to 4 Â 10 17 cm À3 due to the reevaporation and limited incorporation of Sb donor impurities into the SnO 2 NWs with increasing temperature. All of the Sb:SnO 2 NWs had a high transmission of 80% above 800 nm and absorbed light between 400 nm to 800 nm primarily due to the SPR of the Au particles. The transmission may be improved signicantly by selective area location growth which we have shown that is possible on r-Al 2 O 3 by patterning the catalyst. In addition the Sb:SnO 2 NWs on m-, r-and a-Al 2 O 3 are capable of light emission with remarkably long lifetimes of 0.2 ms and are attractive for the realization of NWSCs.

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