Transparent and conducting boron doped ZnO thin films grown by aerosol assisted chemical vapor deposition

ZnO based transparent conducting oxides are important as they provide an alternative to the more expensive Sn : In2O3 that currently dominates the industry. Here, we investigate B-doped ZnO thin films grown via aerosol assisted chemical vapour deposition. B : ZnO films were produced from zinc acetate and triethylborane using either tetrahydrofuran or methanol (MeOH) as the solvent. The lowest resistivity of 5.1 × 10−3 Ω cm along with a visible light transmittance of ∼75–80% was achieved when using MeOH as the solvent. XRD analysis only detected the wurtzite phase of ZnO suggesting successful solid solution formation with B3+ substituting Zn2+ sites in the lattice. Refinement of the XRD patterns showed minimal distortion to the ZnO unit cell upon doping when MeOH was the solvent due to the immiscibility of the [BEt3] solution (1.0 M solution in hexane) in methanol that limited the amount of B going into the films, thus preventing excessive doping.


Introduction
Transparent conducting oxides (TCOs) are vital semiconductor materials widely used in many areas, such as screen displays, touchscreens, solar cells, LCD panels and OLEDs. [1][2][3][4] They have a wide band gap that allows for visible light transparency and relatively high carrier concentration (∼$×10 20 cm −3 ) that enables low electrical resistivity. 1,3,5 The high carrier concentration arises due to a combination of intrinsic impurities and extrinsic dopants. [6][7][8] Currently, tin doped indium oxide (ITO) and uorine doped tin oxides (FTO) are the most widely used TCO materials due to their high performance i.e., resistivities #5 × 10 −4 U cm and transparencies >80%. 2,3,[8][9][10][11] However, ITO is fast becoming nancially unviable due to the increasing cost of In whilst FTO suffers from intrinsic limitations that has meant further enhancement in optoelectronic performance is unattainable. 12 Research into new TCOs is needed to nd potential replacements for ITO and FTO. 5,[13][14][15][16][17] TCOs based on ZnO have the potential to become leading players due to the high abundance of Zn and low cost. 18,19 The band gap of ZnO is ∼3.37 eV leading to high transmittance. 20,21 Traditionally, ZnO is doped with Group 13 ions such as Al 3+ or Ga 3+ as their higher valence and acceptable ionic radii allow for an increase in carrier concentration without massive distortion of the ZnO lattice. 7,13,14 Typically resistivities as low as 5.6 × 10 −4 U cm 22 and 5.0 × 10 −3 U cm 7 have been achieved via both physical 22,23 and chemical vapor deposition 7 routes.
ZnO can also be doped substitutionally on Zn 2+ sites with boron in the +3 oxidation state to enhance carrier concentration and increase conductivity. B 3+ has a smaller ionic radius compared to Zn 2+ and is highly soluble in ZnO. Furthermore, the enthalpy of formation of B 2 O 3 (−13.18 eV) is higher than that of Al 2 O 3 (−17.37 eV) therefore suggesting unwanted secondary oxide phases that can negatively impact the conductivity are less likely to form when B is used as a dopant compared to Al. 24,25 For example, Lu et al. found for sputtered Al : ZnO lms at higher concentrations of dopant, possible cluster and precipitate formation within and on the boundaries of grains caused detrimental lm properties including a reduction in the conductivity, carrier concentration and mobility. 26 This may possibly be due to the formation of the thermodynamically favorable amorphous Al 2 O 3 phase. B : ZnO thin lms have been grown via magnetron sputtering, 27 metal organic chemical vapor deposition (MOCVD), 28 spray deposition 29 and sol-gel methods 30 yielding resistivities as low as 7.5 × 10 −3 U cm, 27 10 U/⧠, 28 4.5 × 10 −3 U cm, 29 and 2.2 × 10 2 U cm. 30 B : ZnO has been shown to be particularly good for increasing the efficiency of Cu(In, Ga)Se 2 (CIGS) photovoltaics. 31 In this paper, we use a specialized form of CVD called aerosol assisted (AA) CVD that allows the growth of transparent and conducting B : ZnO lms using non-volatile, commercially available and inexpensive precursors, namely zinc acetate hydrate and triethylborane. AACVD is unique in that the CVD precursors are dissolved in an appropriate solvent to form a solution that is transferred to the vapor phase in the form of aerosol droplets using a pizoelectronic device. The aerosol mist is then moved into the CVD growth chamber using a carrier gas. The AACVD method is advantageous as it enables device quality lms under scalable ambient pressure conditions. 6,8,11,[32][33][34][35] AACVD has been used to prepare many thin lms widely used in different areas such as for photovoltaics, sensors and photocatalysis. [36][37][38] Here, a series of B : ZnO thin lms from two different solvents (THF and methanol) have been prepared on glass substrates via AACVD and their material and optoelectronic characteristics tested. It was found that resistivities as low as 5.8 × 10 −3 U cm for THF as the solvent and 5.1 × 10 −3 U cm for MeOH as the solvent were obtainable with visible light transparency of ∼75-90% for all the thin lms.
All solutions were atomised using a piezoelectric device (Johnson Matthey liquifog ® ). The aerosol mist was delivered to the AACVD reaction chamber and passed over the heated substrate (oat glass with a SiO 2 barrier layer) using N 2 carrier gas at 1.0 L min −1 . 39 Depositions were carried out at 475°C and lasted until the precursor solution was fully used. Aer the depositions the substrates were cooled under a ow of N 2 . The glass substrates would not be removed unless that with the graphite block was cooled to below 50°C. The lms on the substrates were handled and stored in air.

Film characterisation
The X-ray diffraction (XRD) analysis scanning from 10 to 65°( 2q) used a modied Bruker-Axs D8 diffractometer with parallel beam optics and a PSD LynxEye silicon strip detector. The scans used a monochromated Cu Ka source operated at 40 kV and its emission current was 30 mA with 0.5°as incident beam angle and 0.05°at 1 s per step as step frequency. The JEOL JSM-6301F Field Emission Scanning Electron Microscopy (SEM) with 5 keV as accelerating voltage was used to investigate the surface morphologies of the thin lms. To avoid charging, all the samples were coated with gold before the analysis. The X-ray photoelectron spectroscopy (XPS) analysis was used to determine the surface elemental surroundings by a Thermo Scientic K-alpha photoelectron spectrometer using monochromatic Alk a radiation. Higher resolution scans were recorded for the principal peaks of zinc (Zn 2p), boron (B 2s), oxygen (O 1s) and carbon (C 1s) at a pass energy of 50 eV, and then the CasaXPS soware was used to deal with the data from the XPS analysis. The binding energy of adventitious carbon was adjusted at 284.5 eV as calibration. The Filmetrics F20 thin-lm analyzer was used to measure the thickness of thin lms optically using reectance spectroscopy. The optical properties were determined through a PerkinElmer Fourier transform Lambda 950 spectrometer scanning between 2500 nm and 300 nm. Hall effect measurements were used to determine the of the lms resistivity (r) via the van der Pauw method with a permanent magnet (0.58 T) and one constant current (1 mA, 1 mA).  3 ] solution in hexane were added when MeOH was the solvent. All depositions were carried out at a substrate temperature of 475°C and N 2 ow rate of 1.0 L min −1 to allow for optimal substrate coverage, ZnO crystallinity and lm thickness. All the B : ZnO thin lms were well adhered to the substrate and passed the Scotch tape test. 8

X-ray diffraction
The X-ray diffraction patterns of the undoped and B-doped ZnO lms from THF and MeOH are illustrated in Fig. 1 Texture coefficients were calculated for the ZnO lms to determine the extent of preferential orientation of the crystallographic planes (ESI †). 40 For the nominally undoped and B : ZnO thin lms from THF and MeOH (Fig. 1) preferred orientation was observed in the (002) plane, which is expected as this is the lowest surface energy plane (see ESI Fig. S1 and 2 †) and therefore most likely to dominate. 41 Tables 1 and 2 show the unit cell parameters for the undoped and B-doped ZnO lms from THF and MeOH solutions as determined from Le Bail renement of the powder XRD data. Interestingly, a general decrease in the ZnO unit cell volume from 47.50(2)Å 3 for the 0 mol% to 45.86(22)Å 3 at 15 mol% for the lms gown using THF was observed. This is due to the smaller B 3+ (0.23Å) ions substituting for the larger Zn 2+ (0.60Å) ions resulting in a unit cell contraction. 13,42 With MeOH as the solvent, where solubility of Zn(OAc) 2 $H 2 O is high and the [BEt 3 ] solution is immiscible, no change in the ZnO unit cell volume was observed upon doping. This is likely due to the very low concentrations of B actually doping into the lm therefore minimizing the distortion caused to the ZnO lattice.
X-ray photoelectron spectroscopy X-ray photoemission spectroscopy (XPS) was carried out to determine the surface composition and oxidation state of the B : ZnO lms, as shown in Fig. 2. For all lms, the Zn 2p 3/2 peaks were centered at ∼1020.6 eV which correspond to Zn 2+ (Fig. 2a and c). 43 For lms grown using THF, the B 1s peaks (when detected) were centered at ∼191.6 eV, which corresponds to B in the expected 3+ oxidation state. 18 For the lms grown using MeOH as the solvent, the signal to noise ratio for the B 1s was low compared to the THF samples, again providing more evidence for the low concentration of B in these ZnO lms, as also suggested by the XRD data (Table 2). In fact, no B was detected even when 100 mol% of [BEt 3 ] was used in the MeOH solution. For both the ZnO based lms deposited using THF and MeOH as solvents, the nominally undoped ZnO lm consisted of randomly oriented grains in varying sizes, similar to what has previously been seen for CVD grown ZnO. 13 As B was introduced into the lms, minimal impact on the morphology was observed for the lms grown from THF solutions, however when MeOH was used as the solvent the presence of the B dopant caused a more noticeable change in the surface morphology. Previous reports have described the inuence that MeOH can have on the microstructure of thin lms deposited via AACVD, and in general aerosols from different solvents can inuence the microscopic surface morphology besides their normal transportation effect. 44,45 Therefore, the variations observed in the morphology of the lms in this study is consistent with literature.

UV-visible-near infrared spectroscopy
The optical property of the lms, namely transmittance, has been determined using ultraviolet-visible-near infrared spectroscopy (UV-vis-NIR) (Fig. 4). All the B : ZnO thin lms regardless of solvent used for the AACVD, showed transmittance between ∼75-90% in the visible rangemaking them suitable for TCO application. The lm thickness, as determined via reectance UV-vis spectroscopy using a Filmetrics instrument, increased with increasing amount of [BEt 3 ] used in the precursor solution (See ESI Tables S1 and S2 †). This is attributed to interactions between the zinc acetate and the [BEt 3 ], either in solution in the bubbler or in the gas phase in the CVD chamber and the formation of intermediate products that decomposed more efficiently to give B : ZnO lms.  In the near infrared area, a decrease in transmittance was observed with increasing B concentration for both solvent systems though this was more pronounced for the MeOH samples. This is associated with the increase in free carrier concentration caused by B 3+ substitution that leads to an increase in the plasmon resonance frequency from the NIR towards the visible. 6,13,46 Hall effect measurements The resistivities of the B : ZnO lms were measured via the Hall effect measurement while the parameter lm thickness was calculated through the reectance spectroscopy, as given in Tables S1 and S2 in ESI. † The nominally undoped lm deposited using MeOH was too resistive to obtain any values but crude measurements via a two-point probe showed the resistance to be in the MU order. For the THF solvent system, the nominally undoped lm was measurable but still high at 2.12 × 10 −1 U cm. The differences observed may be due to intrinsic vacancies/dopants such as oxygen vacancies, zinc interstitials or even adventitious hydrogen. 47 For both systems, an increase in the B concentration caused a decrease in resistivity is likely due to an increase in the carrier concentration (as also observed in the UV-vis-NIR spectra as a decrease in the NIR transmittance). The lowest resistivity of 5.8 × 10 −3 U cm for THF as the solvent and 5.1 × 10 −3 U cm for the MeOH based lms were achieved using 7.5 and 300 mol% of [BEt 3 ] in the AACVD solution respectively. According to the signicant difference in miscibility for the B source ([BEt 3 ] solution in hexane) in THF (high) and MeOH (low), the initial B concentrations of 7.5 mol% with THF as solvent and 300 mol% with MeOH as solvent were adopted in order to achieve similar bulk B concentrations (at%) aer depositions, close to the B ZnO thin lms also have been investigated as TCOs materials from some other synthetic routes, such as radio frequency magnetron sputtering 18 and chemical spray pyrolysis 29 and their lowest resistivities were 5.65 × 10 −3 U cm and 4.5 × 10 −3 U cm, respectively, which are similar to the lowest resistivities in this study although here the scalable and inexpensive synthesis method of AACVD was used (Fig. 5).

Conclusion
B-doped ZnO lms were grown using Zn(OAc) 2 $2H 2 O and dopant quantities of [BEt 3 ]. Two solvents were studied, THF and MeOH, due to the differing solubilities/miscibility of the precursors and to investigate the effect of the change in solvent system on the lm properties. XRD alluded to the successful solid solution formation involving substitution of B 3+ on Zn 2+ sites in the ZnO lattice. Furthermore, XPS studies showed that B was indeed in the 3+ oxidation state thus donating one electron for conduction for every Zn 2+ substituted. An increase in carrier concentration resulted in reduced transmittance in the NIR region of the UV-vis-NIR spectra for the doped samples. This was more pronounced in the MeOH samples compared to THF therefore suggesting that the formers carrier concentration was higher. The lowest resistance of 5.1 × 10 −3 U cm was achieved for the 300 mol% [BEt 3 ] using MeOH as the solvent.

Conflicts of interest
The authors have no conicts of interest to declare. Fig. 4 The optical data for the undoped and B doped ZnO thin films on glass substrates using THF (a) and MeOH (b) as solvents prepared via AACVD showing the UV/vis spectra. Fig. 5 The resistivities of the undoped and B doped ZnO films using THF and MeOH as solvents grown through AACVD derived from Hall measurements.