n-Type conducting P doped ZnO thin films via chemical vapor deposition

Extrinsically doped ZnO thin films are of interest due to their high electrical conductivity and transparency to visible light. In this study, P doped ZnO thin films were grown on glass substrates via aerosol assisted chemical vapour deposition. The results show that P is a successful dopant for ZnO in the V+ oxidation state and is able to reduce resistivity to 6.0 × 10−3 Ω cm while maintaining visible light transmittance at ∼75%. The thins films were characterized by X-ray diffraction studies that showed only Bragg peaks for the wurtzite ZnO phase. Fitting of the diffraction data to a Le Bail model also showed a general expansion of the ZnO unit cell upon doping due to the substitution of Zn2+ ions with the larger P5+.


Introduction
Transparent conducting oxides (TCOs) are important materials that are widely used in optoelectronic devices such as solar cells, touchscreens, screen displays, LCD panels and OLEDs. [1][2][3][4] TCOs combine the seemingly orthogonal properties of high transmittance to the visible wavelengths (>80%) with low electrical resistivity (<10 À3 U cm). This is achieved due to TCO materials being wide band gap (>3.1 eV) semiconductors and carrier concentrations in the 10 20 cm À3 order or above due to intrinsic and/or extrinsic point defects. 1,[5][6][7][8] Currently, the most widely used TCO material is tin doped indium oxide (ITO), for example, 90% of the global display market is based on ITO transparent electrodes. 2,3,7,9,10 ITO achieved this dominance due to its superior properties including transparencies as high as 90% and resistivities as low as Â10 À5 U cm. However, the use of ITO is not sustainable owing to its high cost which is associated to supply issues of indium. 8,[11][12][13] As such, other semiconductor materials like SnO 2 and ZnO have been investigated as potential replacements.
ZnO is a mechanically, optically and electrically stable and inexpensive semiconductor with a direct optical band gap of 3.37 eV (via spectroscopic ellipsometry) 14 or 3.27 eV (by applying the Tauc method to UV-visible spectroscopy data). [15][16][17][18][19][20][21] The intrinsic n-type conductivity in ZnO is believed to arise from adventitious H (or C), not from defects such as O vacancies (deep donors) or Zn interstitials (unstable at room temperature) as previously thought. 22 In the nominally undoped form, ZnO is too electrically resistive for TCO applications and is therefore typically doped with trivalent donor dopants such as Al 3+ , Ga 3+ and In 3+ on Zn 2+ sites. 5,23 With these aforementioned dopants, shallow donor levels are formed that enable high electron densities (Â10 20 cm À3 ) and hence resistivities as low as 2 Â 10 À4 U cm. 1,15,16 Although ZnO shows resistance in forming shallow acceptor levels allowing for p-type conductivity, as evidenced from computational and experimental studies, [24][25][26] there have been numerous successful studies on p-type ZnO achieved from nitrogen (N O ), arsenic (As O ), phosphorus (P O ) and lithium (Li Zn ) but oen with limited stability and requiring suppression of compensating donor states. [24][25][26][27][28][29] Phosphorus is an interesting dopant candidate for ZnO as its multivalent nature existing in the IIIÀ, III+ and V+ states, technically allows it to be an n-type and/or p-type dopant via P 3+ / P 5+ on Zn 2+ sites or P 3À on O 2À sites, respectively. Along with ptype conductivity, previous studies have shown an enhancement in electron density and n-type conductivity originating through the formation of P Zn defects and P 3+ , P 5+ or P 3À related complexes. 24,[30][31][32][33][34] In this study, the effect of P doping on the material and optoelectronic properties of ZnO is investigated. The lms were grown using AACVD a scalable, highly tunable and industrially friendly route to determine the suitability of P as a dopant for real world TCO applications and to study whether p or n type conductivity is achieved.

Film synthesis
Depositions were carried out under a N 2 (BOC Ltd., 99.99% purity) ow. Zinc acetate dihydrate (Zn(OAc) 2 $2H 2 O), triethyl phosphate ([PO(OEt) 3 ] (99%) and methanol (99%)) were purchased from Merck. Films were grown on barrier coated ($50 nm SiO 2 ) oat glass (5 cm Â 15 cm Â 0.4 cm) which were cleaned using detergent, water and isopropanol then dried in a 70 C oven prior to deposition. Zn(OAc) 2 $2H 2 O (0.40 g, 1.82 mmol) in methanol (20 mL) was placed in a glass bubbler. [PO(OEt) 3 ] (x mol% based on Zn, x ¼ 0, 0.5, 1.0, 5.0, 7.0 and 10.0) was added in the same bubbler. The solution was atomised through a piezoelectric device (Johnson Matthey Liquifog®). The aerosol mist was delivered to the AACVD reaction chamber and passed over the heated substrate using N 2 carrier gas at 1.0 L min À1 . 35 Depositions were carried out at 500 C and lasted until the precursor solution was fully used. Aer the deposition the substrates were cooled under a ow of N 2 . The glass substrate was not removed until 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/step as step frequency. X-ray photoelectron spectroscopy (XPS) analysis was used to determine the surface elemental surroundings via a Thermo Scientic K-alpha photoelectron spectrometer using monochromatic Al Ka radiation. Higher resolution scans were recorded for the principal peaks of zinc (Zn 2p), phosphorus (P 2p), oxygen (O 1s) and carbon (C 1s) at a pass energy of 50 eV, and then the CasaXPS soware was used to analyse the data from the XPS. The binding energy of adventitious carbon was adjusted at 284.5 eV as calibration. The JEOL JSM-6301F Field Emission Scanning Electron Microscopy (SEM) with 1.5 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 optical properties were determined through a Shimadzu UV-2600 spectrometer scanning between 1100 and 300 nm. Hall effect measurements were used to calculate the electrical properties including bulk concentration (n), carrier mobility (m) and resistivity (r) via the van der Pauw method with a permanent magnet (0.58 T) and a constant current (1 mA).

Results and discussion
Nominally undoped and P doped ZnO thin lms on glass substrates were prepared from the AACVD reaction of zinc  No additional peaks for secondary oxide phases such as Zn 3 P 2 or P 2 O 5 were visible. Calculations performed on the XRD data showed that P had little inuence on the preferred orientation, with the (102) plane having the strongest preference followed by (002) and a lack of preference for the (100). This is most likely due to a combination of substrate inuence and the fact that the (002) is the lowest energy surface in the wurtzite crystal structure, therefore disproportionately favored for growth. 37 The XRD data was also used to determine the ZnO lattice parameters prior to and aer P doping, the results of which are shown in Table 1, and suggest a general contraction in the hexagonal wurtzite unit cell upon P incorporation due to the smaller P 3+ (0.44-0.58 A) or P 5+ (0.31-0.34 A) occupying Zn 2+ (0.74 A) sites in the lattice. If P was in the IIIÀ state (1.8-2.1 A) on O 2À (1.3-1.4 A) sites, then an expansion in the ZnO unit cell would have been observed. 38 X-ray photoelectron spectroscopy (XPS) was used to determine the surface oxidation state and composition of the thin lms (Fig. 1b and c). The Zn 2p data were t with a doublet of peaks separated by 23.1 eV. The Zn 2p 3/2 peaks for the doped and undoped lms were centered at $1022.1 eV, corresponding to Zn 2+ . 39 The P 2p experimental data were also t with a doublet of peaks at a separation of 0.87 eV. The P 2p 3/2 transitions were at $132.9 eV, corresponding to literature reports for P 5+ . [40][41][42] No indication of P 3+ or P 3À was found on the surface of the thin lms. The peaks at $139 eV that were seen in the P 2p scan range belongs to Zn 3s transitions. 41 XPS depth proling carried out on the 14.3 at% doped lm (see ESI Fig. S1 †) showed the P to be surface segregated as opposed to evenly distributing across the depth of the lms. Surface segregation of dopants in ZnO is common and has been seen previously in literature with anionic and cationic dopants. 43,44 The surface morphology of the nominally undoped and series of P doped ZnO thin lms were investigated via scanning electron microscopy (SEM) (Fig. 2). The morphology was similar to what has previously been seen for ZnO lms grown from [Zn(OAc) 2 $2H 2 O] in methanol and consisted of compact dome/ platelet like features. The size of the features were $200 nm at up to 3.4 at% doping but increased to 300-400 nm at higher  (Fig. 3a). The nominally undoped ZnO lm was too resistive to measure via the Hall instrument but crude measurements of resistance using a two-point probe showed values in the MU region. Upon doping of ZnO with P an enhancement in the lm conductivity was instantly observed to measurable levels with a negative Hall coefficient, indicative of  n-type conductivity. The 2.7 at% doped lm showed resistivity of 1.6 Â 10 À2 U cm, carrier concentration of 5.3 Â 10 19 cm À3 and electron mobility of 7.4 cm 2 V À1 s À1 . At 3.4 at% P, resistivity was further reduced to 1.0 Â 10 À2 U cm and at 6.5 at% P, the lowest resistivity of 6.0 Â 10 À3 U cm was achieved owing to an increase in the electron concentration of 8.9 Â 10 19 cm À3 and 1.6 Â 10 20 cm À3 , respectively. The increase in the electron concentration was observed due to the successful replacement of Zn 2+ with P 5+ , thereby releasing up to three electrons for conduction for every Zn 2+ replaced. Further increase in P to 8.6 and 14.3 at% resulted in an increase in resistivity to 1.0 Â 10 À2 and 1.1 Â 10 À2 U cm due to both a decrease in carrier concentration (to 9.1 Â 10 19 and 1.4 Â 10 20 cm À3 , respectively) and carrier mobility (6.6 and 4.3 cm 2 V À1 s À1 ). The decrease in carrier concentration is attributed to self-compensating mechanisms, such as O interstitials or Zn vacancies that increase with increasing P levels in ZnO, and/or the formation of electrically inactive secondary phases such as Zn 3 P 2 or P 2 O 5 that were undetected by XRD and XPS. [45][46][47] The reduction in carrier mobility with increasing levels of P is due to increased ionized impurity scattering and increased secondary electrically inactive phase formation. There are numerous examples of the AACVD growth of cation doped ZnO using [Zn(OAc) 2 $2H 2 O] for TCO applications in the literature ( Table 2). The electrical results from our study of P doped ZnO show that P is indeed a practically suitable dopant yielding enhanced n-type conductivity to levels acceptable for TCO application. In addition we compare our study with other P doped ZnO thin lms from different precursors and synthesis methods ( Table 3). The electrical properties were in a similar range ($10 À3 U cm) and the synthesis method in this study, AACVD, is considered as a more convenient method to prepare thin lms with simple process and relatively low cost, 6,7,11 which offers more advantages for TCO application. Ultraviolet-visible (UV-vis) spectra between 300 and 1100 nm for the undoped and P doped thin lms on glass substrates are shown in the Fig. 3b. All lms were transparent across the visible wavelengths (400-700 nm) with an average transmittance of $75%. The band gaps were determined by applying the Tauc formula to the UV/vis data and shown in Fig. 3c. The nominally undoped ZnO lm had a band gap of 3.3 eV, which is close to previous literature Tauc plot results of 3.28 eV for undoped ZnO. Upon the introduction of P at 2.7, 3.4 and 6.5 at%, the ZnO band gap remains at 3.3 eV however at higher dopant concentrations of 8.6 and 14.3 at% the band gap widens to 3.5 eV.

Conclusion
A series of undoped and P doped ZnO thin lms with different concentrations of P were grown on glass substrates via AACVD. XPS showed P to be only in the V+ oxidation state on the surface. Evidence for successful doping and solid solution formation was provided by XRD analysis where only reections for ZnO wurtzite phase were observed. Hall effect measurements showed that upon doping the electron density increased due to effective replacement of Zn 2+ in the lattice with P 5+ . The lowest resistivity of 6.0 Â 10 À3 U cm was obtained at 6.5 at% P concentration and comparable to results obtained for ZnO:Al but better than ZnO:In from [Zn(OAc) 2 $2H 2 O] through AACVD. Optical measurements showed 75% transmittance to visible light and absorption in the NIR region for the lms with high electron densities. The results of this study show that P doping of ZnO under a simple, saleable and industrially relevant technique such as AACVD with inexpensive and easy to handle precursors is able to produce sufficiently n-type conductive and stable thin lms.

Conflicts of interest
The authors declare no conict of interest.