Effect of Nd3+ incorporation on the microstructure and chemical structure of RF sputtered ZnO thin films

Materials Science and Engineering B 178 (2013) 609–616

Contents lists available at SciVerse ScienceDirect

Materials Science and Engineering B journal homepage: www.elsevier.com/locate/mseb

Effect of Nd3+ incorporation on the microstructure and chemical structure of RF sputtered ZnO thin films Gloria Gottardi a,∗ , Rajesh Pandiyan a,b,c , Victor Micheli a , Giancarlo Pepponi a , Salvatore Gennaro a , Ruben Bartali a , Nadhira Laidani a a

Fondazione Bruno Kessler, Center for Materials and Microsystems, Via Sommarive 18, 38123 Trento, Italy University of Trento, Physics Department, Via Sommarive 14, 38123 Trento, Italy c LGPPTS, ENSCP, Université Pierre et Marie Curie, 11 rue Pierre et Marie Curie, 75005 Paris, France b

a r t i c l e

i n f o

Article history: Received 16 July 2012 Received in revised form 19 November 2012 Accepted 9 December 2012 Available online 28 December 2012 Keywords: Rare earth doping ZnO thin films XRD Neodymium AES XPS

a b s t r a c t The present work aims at investigating the effects that different levels of Nd atoms incorporation can have on the microstructure and chemical structure of ZnO thin films. Undoped and Nd-doped ZnO films were deposited by RF co-sputtering from pure ZnO and metallic Nd targets in Ar plasma onto Si, quartz and glass substrates. The Nd concentration in the ZnO host matrix was varied in the range 0–26 at.% by varying the bias applied to the Nd target. A comprehensive characterization of the films properties was performed by X-ray photoelectron and Auger electron spectroscopies, X-ray fluorescence analysis, X-ray diffraction and scanning electron microscopy. At low Nd atomic concentration (Nd/Zn < 0.07) Nd atoms were successfully incorporated into the ZnO matrix, whose crystalline structure was preserved. A deterioration of the ZnO würtzite phase was observed on the contrary with increasing Nd content in the films together with the precipitation of a second phase, identified as Nd2 O3 . © 2012 Elsevier B.V. All rights reserved.

1. Introduction Doping is commonly used for altering phase structure, electronic structure, surface and bulk chemical and physical properties of a material depending on the pursued application. A crucial prerequisite for a successful doping is actually an accurate control over the dopant concentration in the host material. Moreover, since the physical and chemical properties of the host matrix have been proved to be influenced by the level of foreign elements incorporation, a full characterization of the material over an adequate range of dopant concentrations is recommended [1]. In this paper we will report the investigation of the effects of Nd atoms incorporation on the microstructure and chemical structure of RF sputtered ZnO thin films. Rare earth (RE) ions doped semiconductors have been recently attracting increasing attention due to their several promising applications in technological fields as photovoltaics, photocatalysis, optoelectronic devices and flat panel displays [1–3]. Wide band

∗ Corresponding author at: Fondazione Bruno Kessler – Plasma, Advanced Materials and Surface Engineering Research Unit, Center for Materials and Microsystems, Via Sommarive 18, 38123 Trento, Italy. Tel.: +39 0461 314475; fax: +39 0461 810851. E-mail address: [email protected] (G. Gottardi). 0921-5107/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.mseb.2012.12.013

gap oxides, such as TiO2 and ZnO, have been the focus of particular research interest since they can act as host candidates for doping with trivalent lanthanide ions, thus allowing a better exploitation of the solar spectrum through photon conversion [4]. ZnO above all, due to its wide direct band-gap of 3.37 eV, large exciton energy of 60 meV, chemical stability and low toxicity is gaining growing importance in both basic and applied research [2,5–7]. On the dopants side, trivalent neodymium (Nd3+ ) is an element of choice, due to its sharp and intense luminescence in the near infrared region, which falls within the range of intrinsic spectral absorption of silicon solar cells [8]. Nevertheless, detailed literature survey shows that few reports are there focusing on the incorporation of Nd atoms into ZnO films by RF sputtering [1,9]. Besides, due to the large mismatch in ionic radius and charge imbalance between Nd3+ and Zn2+ , the successful incorporation of neodymium ions into the ZnO lattice has always been rather challenging [2]. In the present work, a study was therefore conducted to describe the influence of incorporated Nd concentration on ZnO thin films properties and in particular to determine the Nd concentration limit in the host matrix beyond which dopant clustering may occur. The use of a multi-technique analytical approach allowed to characterize the films with respect to their structure and chemical composition, especially for low Nd concentrations.

610

G. Gottardi et al. / Materials Science and Engineering B 178 (2013) 609–616

Undoped- and neodymium-doped ZnO films were deposited by radio frequency (13.56 MHz) co-sputtering. High purity, commercially available ZnO (purity = 99.999%, diameter = 10 cm) target and metallic Nd target (purity = 99.99%, diameter = 5 cm) were sputtered simultaneously in a pure argon plasma. The reactor base pressure was kept at 4E−6 Pa, while the depositions were operated at 6.6 Pa working pressure by keeping the argon flow rate constant at 30 sccm. Separate power supplies (Advanced Energy RFX 600A for ZnO and Advanced Energy CESAR for Nd) were connected to each target providing independent power control. A constant selfbias voltage (VRF ) of −550 V was kept fixed on the ZnO cathode for all the deposition processes. On the contrary, to obtain different levels of Nd doping in the films, the power applied to the Nd target (WNd ) was varied from 0 to 7 W, corresponding to a self bias voltage (Vbias ) ranging between 0 and −67 V. The deposition time was 7 h for all the samples to obtain thicknesses in the range 0.85–1.89 ␮m. Organically degreased Si (1 0 0) wafers, quartz and glass pieces were used as substrates; they were placed on a rotating sampleholder, always maintained at the floating potential. The sample holder was water cooled to keep the substrates at room temperature during the depositions. All the samples were then treated by ex situ post deposition annealing, performed in a furnace for 6 h at ambient air and at the constant temperature of 600 ◦ C, reached with a heating ramp of 5 ◦ C/min.

diffractometer (TNX, Italy) where a Silicon drift detector (50 mm2 collimated active area, KETEK, Germany) has been mounted at 90◦ to the sample surface and 15 mm distance. XPS spectra were recorded with a KRATOS AXIS UltraDLD instrument equipped with an hemispherical analyzer and a monochromatic Al K␣ (1486.6 eV) X-ray source. The core lines (C1s, O1s, Zn2p, Nd4d) were acquired at 20 eV pass energy, which leads to an energy resolution of ∼0.4 eV. Compensation of the surface charging was performed by bombarding the surface with an electron flood gun during the analyses. After a Shirley-type background subtraction, the spectra were fitted using a non-linear least-squares fitting program adopting a Gaussian–Lorentzian peak shape. Since the deposited films were exposed to air, the C1s peak must include a significant amount of carbon due to ambient contamination. After deconvolution of the C1s core line therefore, the main peak at 285 eV corresponding to hydrocarbon contamination was used as internal reference to calibrate the spectra and correct their binding energy (BE) shift due to surface electrostatic charging, either under flood electron gun application or without. The phase structure of the films was investigated using a ˚ Bragg–Brentano X-ray diffractometer with Cu K␣ ( = 1.5406 A) radiation. In order to increase the sensitivity of the :2 geometry, the incident and diffracted beams were made quasi-parallel by means of a Göbel Mirror optics on the incident beam and a Soller slit along the detector side. XRD patterns were acquired using scan rate of 0.02◦ 2 step at 40 kV, 30 mA over a 2 angle range of 20◦ –80◦ . Crystallites size (D) was obtained according to the Scherrer’s equation

2.2. Films characterization

D = 0.9

2. Material and methods 2.1. Films deposition

The chemical structure and bulk and surface composition of the deposited samples were studied by Auger electron spectroscopy (AES), X-ray photoelectron spectroscopy (XPS), X-ray diffraction (XRD) and X-ray fluorescence (XRF). AES depth profiles were carried out using a Physical Electronics 4200 system, equipped with a variable resolution cylindrical mirror analyzer (CMA, energy resolution 0.6%) and a coaxial electron gun. Auger spectra were excited using a 5 keV electron beam that scanned an area of 400 ␮m2 , while film etching was performed by a 2 keV Ar+ beam. The C KLL (271 eV), O KLL (503 eV), Zn LMM (994 eV) and Nd MNN (91 eV) transitions were monitored for the film characterization while Si LMM (89 eV) for probing the signal coming from the silicon substrate [10]. The depth distribution of the elements was recorded as a function of the sputtered depth, assessed on the basis of the erosion time and the film thickness. The elements signal intensity was measured from the peak-to-peak heights in derivative spectra after correction with their specific sensitivity factors. Due to the AES detection limits being too high for the quantification of very low Nd content in the films, an estimate of the dopant concentration was in that case derived by X-ray fluorescence analysis. XRF spectra were acquired on a Phoenix theta–theta

 ˇcos 

(1)

where  is the X-ray wavelength, ˇ the full width at half maximum (FWHM) of the considered peak and  is the Bragg angle [11]. The c-axis length was on the other hand calculated by the equation c=

 sin 

(2)

where  is the X-ray wavelength and  is the Bragg diffraction angle in degree [1]. Sample morphology was checked with a scanning electron microscope (SEM). Secondary Electron images of the samples were acquired by using a Field Emission SEM (JEOL JSM 7401F) operated at 5 kV. Secondary electrons only were detected to maximize lateral resolution. The films thickness was measured with a KLA – Tencor P-6 stylus profilometer and reported in Table 1. 3. Results and discussion In the following sections we will describe the evolution of Nd doped ZnO films properties in dependence of the increasing neodymium content in the material. The co-sputtering process appears to be successful in doping the ZnO matrix at low Nd concentration, while much higher concentrations produce dramatic

Table 1 Plasma conditions, elemental analysis by AES and XRF, films thickness and grain size D of pure ZnO films and ZnO:Nd films deposited with increasing Vbias on the Nd target. Vbias on Nd target [V]

Nd at.% from AES

Zn at.% from AES

O at.% from AES

C at.% from AES

Nd/Zn atomic ratio from AES

0 −19 −32 −39 −55 −67

– <0.5 <0.5 3.3 21.8 26.3

50.9 51.2 49.6 48.4 19.5 9.1

47.1 47.3 49.0 46.8 57.2 62.9

0.9 0.7 0.6 1.1 1.1 1.2

0 – – 0.07 1.12 2.90

a

Not measurable.

Nd/Zn atomic ratio from XRF n.m.a 0.0013

O/Nd + Zn atomic ratio from AES

Film thickness [␮m]

D [nm]

0.92 0.92 0.98 0.91 1.39 1.78

0.85 0.91 1.29 1.10 1.45 1.89

32.1 29 38.5 37.8 24.6 13.6

G. Gottardi et al. / Materials Science and Engineering B 178 (2013) 609–616

3.2. Quantification of low neodymium content by XRF A crucial point of our study has been the quantification of Nd in samples co-sputtered with low power applied on the Nd target (Nd target Vbias < −39 V). As anticipated in the previous section, the amount of dopant atoms incorporated in these samples is in fact below the AES detection limit (<0.5 at.%, see Table 1). An estimate of

[a]

100 Zinc Oxygen Carbon Silicon Neodymium

80

Rel. at. conc. (%)

The films were checked throughout their thickness by Auger depth profiling, in order to assess their composition uniformity and stoichiometry. The obtained elemental compositions, calculated as an average over the bulk of the films (surface and film/substrate interface regions excluded) and expressed in relative atomic percentages, are listed in Table 1 for all the deposited samples. In Fig. 1 we plotted three sample AES depth profiles, acquired respectively on the pure ZnO film (Nd target Vbias = 0, panel [a]), on the ZnO:Nd film with measured 3.3 at.% Nd content (Nd target Vbias = −39 V, panel [b]) and on the ZnO:Nd film with measured 26.3 at.% Nd content (Nd target Vbias = −67 V, panel [c]). All of them exhibit a rather sharp interface region toward the substrate and a perfectly homogeneous composition throughout their thickness. A negligible carbon contamination has been recorded in all the films, with the exception of the surface region where a certain amount of adventitious carbon is always present due to air exposition. As can be seen from data reported in Table 1 and Fig. 1 the Nd signal (obviously absent in profile [a]) becomes detectable by AES starting from samples deposited with a bias voltage on the Nd target larger than −39 V and it grows to be dominant with respect to Zn signal in profile [c], where a Nd content of 26.3 at.% has been measured. We can also observe that pure ZnO sample appears to be slightly sub-stoichiometric, with an O/Zn ratio equal to 0.92. Close to 0.9 is also the value of the O/(Nd + Zn) ratio obtained in all the films deposited with low power applied on the Nd target (Vbias ≤ −39 V) which are those where the Nd concentration is low (Nd ≤ 3.3 at.%). This is most probably due to the fact that at low doping levels neodymium atoms succeed to enter the host ZnO matrix by substituting into the Zn2+ sites. Further evidence of this deduction will come from XRD analyses as well. On the contrary, increasing the dopant concentration further than 21 at.% (Nd target Vbias ≥ −55 V) results in an increase of the O/(Nd + Zn) atomic ratio up to 1.78, well beyond the unit value expected for a zincite-like structure. This implies that high Nd doping level in the material lead to the formation of a different oxide phase involving neodymium, which can coexist and eventually become dominant with respect to the original host ZnO matrix. We suppose that a proper identification of such a phase is Nd2 O3 . The analysis of the line shape of the AES O KLL peak can help confirming our hypothesis. In Fig. 2 the AES O KLL peak acquired respectively on a ZnO standard [a], on a Nd2 O3 standard [b], on a pure ZnO film (Nd target Vbias = 0, [c]) and on a ZnO:Nd film with high Nd content (Nd = 26.3 at.%; Nd target Vbias = −67 V, [d]) have been plotted. It is evident that while the un-doped ZnO sample shows an OKLL peak perfectly resembling that of a ZnO standard (curve [c] and [a]), the highly doped one appears to have an AES oxygen line more similar (for both shape and position) to that of neodymium oxide rather than of ZnO (curve [d] and [b]). This means that we are not doping ZnO with Nd atoms any more, but that we are rather synthesizing a mixed phase where Nd2 O3 is prevailing over ZnO.

60

40

20

0 0

10

20

30

40

50

60

70

Erosion time (min)

80

[b]

100 Zinc Oxygen Carbon Silicon Neodymium

80 Rel. at. conc. (%)

3.1. Films composition from AES

the Nd concentration was derived therefore by another technique, X-ray fluorescence. Samples with sufficient Nd content, quantifiable by AES, were used to derive a calibration curve of the XRF data. In Fig. 3a we

60

40

20

0 0

10

20

30 40 50 Erosion time (min)

60

70

80 [c]

100 Zinc Oxygen Carbon Silicon Neodimium

80

Rel. at. conc. (%)

changes in the microstructure of the films and eventually the precipitation of two co-existing phases.

611

60

40

20

0 0

10

20

30

40

50

60

70

Erosion time (min) Fig. 1. AES depth profiles, acquired respectively on the pure ZnO film (Nd target Vbias = 0, panel [a]), on the ZnO:Nd film with measured 3.3 at.% Nd content (Nd target Vbias = −39 V, panel [b]) and on the ZnO:Nd film with measured 26.3 at.% Nd content (Nd target Vbias = −67 V, panel [c]).

612

G. Gottardi et al. / Materials Science and Engineering B 178 (2013) 609–616

OKLL acquired on a standard ZnO

OKLL acquired on a standard Nd2O3

[b]

dN(E)/dE

dN(E)/dE

[a]

480

490

500

510

520

530

480

540

490

510

520

530

540

Kinetic energy (eV)

Kinetic energy [eV]

OKLL acquired on a pure ZnO film (Vbias=0)

[d]

OKLL acquired on a ZnO:Nd film (Vbias=-67V)

dN(E)/dN

dN(E)/dE

[c]

500

480

490

500 510 520 Kinetic energy [eV]

530

540

480

490

500

510

520

530

540

Kinetic energy [eV]

Fig. 2. AES line shape of OKLL peak acquired on a ZnO standard [a], on a Nd2 O3 standard [b], on a pure ZnO film [c] and on a ZnO:Nd film deposited with Vbias = −67 V applied to the Nd target [d].

present two examples of XRF spectra, acquired respectively on a film with low Nd concentration (Nd target Vbias = −32 V), not detectable by AES, and on one with a Nd content of 3.3 at.% (Nd Vbias = −39 V) as measured by AES depth profiling. X-ray net intensities have been extracted by integration of the regions of interest and subtracting the background according to the trapezium rule, as described below, where specti is the spectrum (counts in channel (i) and ch1 and ch2 are the first and last channel of the region of interest for the X-ray line considered.

Net counts =

ch2 

specti −

i=ch1

(spectch1 + spectch2 ) · (ch2 − ch1) 2

(3)

In Fig. 3b, Nd (La + Lb)/Zn Ka counts against AES Nd/Zn atomic ratios have been plotted. Both intensity ratios as acquired and corrected for absorption effects in the layer are displayed. The absorption correction factor has been calculated integrating the absorption factor in the Sherman equation [12] over the layer thickness. Absorption correction factor =

1 − exp(−(/)s,E0 s (T/ sin (/)s,E0 (s / sin

0 ) − (/)s,Eline s (T/ sin

0 ) + (/)s,Eline (s / sin

1 ))

1)

(4)

where (/)s is the mass absorption coefficient of the sample, E0 is the energy of the primary beam, Eline the energy of the X-ray line considered, 0 is the angle of incidence of the primary radiation, 1 is the angle of detection, s is the density of the sample and T is the thickness of the layer. For the calculation of the mass absorption coefficients (/)s the sample composition determined by AES and tabulated cross sections [13] were used. Zn and Nd atomic ratios were converted into the relative concentrations of ZnO and Nd2 O3 . The data was fitted with a straight line, forcing the intercept through the origin. The Nd concentration for the sample deposited with Nd Vbias = −32 V was then simply obtained interpolating the fitted line (see Table 1). Analysis of sample deposited with Nd Vbias = −19 V did not provide on the contrary a measurable XRF signal. 3.3. Chemical structure from XPS XPS analyses were performed on all the samples and regarded the core levels Zn2p, O1s, Nd4d, C1s and survey regions. In Fig. 4 we plotted in particular the survey and the O1s acquired on a pure ZnO sample (panels [a] and [c] respectively) and the survey and O1s acquired on the ZnO:Nd film with measured 26.3 at.% Nd content (Nd target Vbias = −67 V, panels [b] and [d]). The O1s of two samples with intermediate Nd concentrations are displayed as well (3.3 at.% in panel [e] and 21.8 at.% in panel [f]). From the surveys in panel

G. Gottardi et al. / Materials Science and Engineering B 178 (2013) 609–616

Fig. 3. [a] XRF spectra of samples deposited with −32 V and −39 V bias on the Nd target; [b] calibration of XRF data by means of AES results. Raw experimental XRF intensities and absorption corrected ones are reported. The straight-line fit of the corrected values is also shown.

[a] and [b] it is immediately evident the remarkable variation in the composition of the two extreme samples of our experimental range: the first showing a pure zinc oxide surface, the second a surface characterized by the presence of strong neodymium signal in addition to those of zinc and oxygen. The XPS core lines of Zn 2p3/2, O1s and Nd4d were studied in detail for all the samples. However, in the same way of AES analyses, XPS could reveal and quantify Nd signal only in samples deposited with a bias voltage on the Nd target larger than −39 V. This fact has limited the possibility of performing an accurate description of the surface chemical structure of Nd doped ZnO films to those samples having a Nd content higher than 3 at.%. Besides, due to the overlapping between the Nd3d and the OKLL peaks, the Nd4d level was acquired, as an alternative, in detail. It was found that the introduction of neodymium in the films has no significant influence on the XPS spectra in the Zn2p level. After charging correction, the Zn 2p3/2 peak was always fitted with two components (data not shown here): the main one, located at 1021.6 eV, corresponding to zinc bounded to oxygen in the zinc oxide lattice and a weaker shoulder, shifted by 1.1 eV toward high binding energies with respect to the first component, attributed to zinc hydroxide, situated at 1022.7 eV [14–17]. A more meaningful description of the effects exerted by different levels of dopant incorporation on the chemical structure of the films can be derived instead from the deconvolution of the O1s peak, as shown in Fig. 4 (panels c–f). Independently on the Nd content, the

613

O1s peak appears dominated by the component at 530.2 eV due to oxygen atoms involved in the würtzite structure of the ZnO host matrix. A second component, placed at ∼531.3 eV, was found to exist in between the dominant Zn O and the higher BE components (O1s C O/O C O and O1s OH) due to carbonated species and to adsorbed water originating from surface contamination. It has been assigned to zinc hydroxide species (Zn OH). It is well known, in fact, that in many transition metals hydroxylation occurs, giving rise to a peak (due to metal OH bond formation) whose energy is in accordance with ours [16–18]. Finally, an additional component (∼529 eV) arises at the lower BE side of the O1s peak as soon as the Nd signal becomes detectable by XPS, that is to say for Nd concentrations higher than 3 at.% (Fig. 4, panels d–f). This component is completely separated from the main one due to oxygen involved in the ZnO würtzite structure (530.2 eV) and can be assigned to oxygen atoms bound to Nd [8,19,20]. It is worth to be noted at this point that Nd4d peak BE position, around 121.4 eV, is in good agreement with an oxidation state of Nd 3+ [8,19,20], suggesting that with increasing doping level Nd atoms could be present in the form of its oxide (Nd2 O3 ). This result agrees with what already proposed on the base of AES analyses: when the Nd concentration increases above the limit of 3 at.% two different phases are formed in the material (ZnO and Nd2 O3 ) which tend to segregate from each other. To conclude the description of the XPS spectra in Fig. 4, it is interesting to note that the intensity of the peak due to Nd O bounds is lower than that of the main component attributed to O bound to Zn in the ZnO lattice for the sample with the highest Nd concentration (panel [d]). This is in contrast with what expected on the base of the quantification results (Table 1), from which we know that Nd content is almost three times the Zn one in this sample. The discrepancy can be explained in view of the oxygen excess revealed by AES depth profiles, which can reasonably be attributed to the adsorption of water on the samples surfaces. The component placed at ∼533 eV, normally reported to pertain to adsorbed molecular H2 O [17,21,22] confirms this supposition. It is just the existence of adsorbed water which can promote the formation of Nd(OH)3 by the reaction Nd2 O3 + 3 H2 O → 2 Nd(OH)3 in the oxide film [19]. Oxygen atoms involved in neodymium hydroxide give a signal which overlaps with the Zn O component, whose intensity results this way enhanced.

3.4. Film structure from XRD and SEM analysis In Fig. 5 we plotted the diffraction patterns of all the ZnO:Nd samples deposited with increasing the self-bias voltage on the Nd target. The data refer to annealed samples. As deposited films were analyzed as well (patterns not shown) revealing their crystalline nature. Post deposition annealing was performed to improve the crystal quality of their structure. All the films deposited with Vbias ≤ −39 V crystallized in the characteristic hexagonal würtzite type phase of ZnO and appear to be highly textured, presenting mainly one dominant preferred orientation along the [0 0 2] axis (2 = 34.4◦ , in accordance with the ICDD 36-1451 card). At a normal view, no other phases could be identified. A magnified scale of Y-axis is used here in order to reveal lower intensity peaks. A peak pertaining to the (0 0 4) reflection can this way be recognized, as well as two small peaks placed at around 31◦ and 37.1◦ likely due to the presence of Zn(OH)2 (ICDD 20-1435 card) impurities, as revealed also by XPS analyses. When increasing power to the Nd target up to −39 V, two other small peaks corresponding to the (1 0 1) and (1 1 2) directions of the ZnO würtzite structure become visible, but no reflections attributable to Nd containing phases can be observed. This means that, for Nd doping level <3 at.%, ZnO:Nd films do not contain any secondary phases

614

G. Gottardi et al. / Materials Science and Engineering B 178 (2013) 609–616

Fig. 4. XPS spectra: [a] survey and [c] O1s of pure ZnO film; [b] survey and [d] O1s of ZnO:Nd film with measured 26.3 at.% Nd content (Nd target Vbias = −67 V); [e] O1s of ZnO:Nd film with 3.3 at.% Nd content (Nd target Vbias = −39 V); [f] O1s of ZnO:Nd film with 21.8 at.% Nd content (Nd target Vbias = −55 V).

and that neodymium atoms are successfully incorporated into the hexagonal ZnO lattice. At the same time, what can be observed is a slight shift in the position of the (0 0 2) peak toward lower angle side with increasing dopant level in the films. This shift allowed to calculate the corresponding variation in the c-axis length [1], which is plotted toward the bias applied on the Nd target in Fig. 6. The increasing trend of this parameter for samples with low dopant content indicates that ZnO lattice expands along the c-axis when Nd atoms enter the matrix and substitute into the Zn2+ sites. Such an expansion can be qualitatively understood considering the size of the ions involved ˚ is much in the doping process: the ionic radius of Nd3+ (0.995 A) ˚ which creates lattice strain and larger than that of Zn2+ (0.74 A), consequently increases the lattice parameter [7,9,23].

On the other hand, for films deposited with Vbias on the Nd target more negative than −39 V, a strong reduction of the (0 0 2) peak intensity occurs, together with an abrupt decrease in the average crystallite size D (see Table 1). This indicates that the ZnO lattice becomes deformed owing to the presence of high amounts of foreign atoms: high level of Nd incorporation in the films restrains ZnO crystallites growth leading eventually to a visible deterioration of the ZnO würtzite structure [7,9,23]. Beyond 21.8 at.% Nd content, characteristic XRD peaks attributable to possible phases of Nd2 O3 and Nd(OH)3 (ICDD card nos. 21-0579, 28-0671, 06-0601) become visible (Fig. 5, upper panels). This means that Nd atoms segregated from ZnO forming a co-existing nano-crystalline phase in the film [19,24,25].

G. Gottardi et al. / Materials Science and Engineering B 178 (2013) 609–616

*/#

*

615

Vbias= -67V

* */#

**

*

Vbias= -39V (112)

(101)

Intensity [a.u.]

Vbias= -55V

Vbias= -32V

Vbias= -19V

(004)

(002)

Vbias= 0

°

° 30

40

50 60 2θ [deg]

70

80

Fig. 5. Diffraction patterns of ZnO:Nd films deposited with increasing bias voltage on the Nd target. The ZnO main crystallographic planes are identified in the plots; the Y scale was magnified to reveal the lowest intensity peaks. (◦ ), (*) and (#) indicate possible phases of Zn(OH)2 , Nd2 O3 and Nd(OH)3 respectively.

The scanning electron micrographs presented in Fig. 7 closely reflect the above described structural changes. Pure ZnO films show a rather rough surface with large crystallites (Fig. 7a) whereas at high concentration of Nd in the film, the material appears to

Fig. 7. SEM micrographs of a pure ZnO film [a] and of a ZnO:Nd film deposited with Vbias = −67 V applied to the Nd target, resulting in a Nd content in the film equal to 26.3 at.%.

be nano-crystalline, with grains of visibly reduced dimensions (Fig. 7b). 4. Conclusions

Fig. 6. Lattice c parameter variation for ZnO:Nd samples deposited with increasing bias voltage on the Nd target. The dashed line is to guide the eye.

In the present work we studied the effect of neodymium doping on the microstructure and chemical structure of ZnO thin films prepared by RF co-sputtering. The Nd concentration was varied in the range 0–26.3 at.% in the ZnO matrix. A detailed characterization of the pure and doped ZnO films, performed by AES, XPS, XRF and XRD techniques, allowed a complete description of the material evolution in function of the Nd doping level. In fact, marked structural modifications turned up together with variations in the films composition. In particular, at low Nd atomic concentration (Nd ≤ 3 at.%) Nd atoms appeared to be successfully incorporated into the ZnO matrix, whose crystalline structure was maintained. A deterioration of the würtzite phase was observed on the contrary with increasing Nd doping level above 3 at.%, which inhibits growth of ZnO crystallites. The study allowed in other words to determine the Nd concentration limit in the host matrix beyond which dopant clustering may occur. Nd atoms segregation consisted in the precipitation of a different Nd-containing phase, which we could identify as Nd2 O3 .

616

G. Gottardi et al. / Materials Science and Engineering B 178 (2013) 609–616

Acknowledgements This work was supported by Fondazione Caritro in the frame of the project OXI-SOLAR. References [1] M. Subramanian, P. Thakur, S. Gautam, K.H. Chae, M. Tanemura, T. Hihara, S. Vijayalakshmi, T. Soga, S.S. Kim, K. Asokan, R. Jayavel, Journal of Physics D: Applied Physics 42 (2009) 105410–105415. [2] Y. Liu, W. Luo, R. Li, H. Zhu, X. Chen, Optics Express 17 (12) (2009) 9748–9753. [3] Y. Liu, W. Luo, R. Li, X. Chen, Journal of Nanoscience and Nanotechnology 10 (3) (2010) 1871–1876. [4] T. Trupke, M. Green, P. Wurfel, Journal of Applied Physics 92 (2002) 1668. [5] S.-Y. Seo, M.-J. Kim, J.H. Shin, Applied Physics Letters 83 (2003) 2778. [6] L.-C. Chao, J.W. Huang, C.-W. Chang, Physica B 404 (2009) 1301–1304. [7] B. Roy, S. Chakrabarty, O. Mondal, M. Pal, A. Dutta, Materials Characterization 70 (2012) 1–7. [8] M. Yuan, J. Zhang, S. Yan, G. Luo, Q. Xu, X. Wang, C. Li, Journal of Alloys and Compounds 509 (2011) 6227–6235. [9] Z.Y. Huang, P. Luo, M. Chen, S.R. Pan, D.H. Chen, Materials Letters 65 (2011) 2345–2347. [10] L.E. Davis, N.C. MacDonald, P.W. Palmberg, G.E. Riach, R.E. Weber, Handbook of Auger Electron Spectroscopy: A Reference Book of Standard Data for Identification and Interpretation of Auger Electron Spectroscopy Data, 2nd ed., Physical Electronics Division, Perkin Elmer Corporation, 1978.

[11] H.P. Klug, L.E. Alexander, X-ray Diffraction Procedures for Polycrystalline and Amorphous Materials, 2nd ed., John Wiley & Sons, New York, 1974. [12] J. Sherman, Spectrochimica Acta 7 (1955) 283–306. [13] H. Ebel, R. Svagera, M.F. Ebel, A. Shaltout, J.H. Hubbell, X-Ray Spectrometry 32 (2003) 442–451. [14] H.T. Cao, Z.L. Pei, J. Gong, C. Sun, R.F. Huang, L.S. Wen, Journal of Solid State Chemistry 177 (2004) 1480–1487. [15] N. Asakuma, T. Fukui, M. Toki, K. Awazu, H. Imai, Thin Solid Films 445 (2003) 284–287. [16] S. Ben Amor, M. Jacquet, P. Fioux, M. Nardin, Applied Surface Science 255 (2009) 5052–5061. [17] U. Meier, C. Pettenkofer, Applied Surface Science 252 (2005) 1139–1146. [18] N. Laidani, P. Cheyssac, J. Perriere, R. Bartali, G. Gottardi, I. Luciu, V. Micheli, Journal of Physics D: Applied Physics 43 (2010) 485402–485413. [19] X.M. Wang, X.Q. Zeng, G.S. Wu, S.S. Yao, L.B. Li, Applied Surface Science 253 (2007) 9017–9023. [20] Y. Uwamino, Y. Ishizuka, H. Yamatera, Journal of Electron Spectroscopy and Related Phenomena 34 (1984) 69. [21] W.P. Yang, D. Costa, P. Marcus, Journal of the Electrochemical Society 141 (1994) 2669. [22] H.P. Marques, A.R. Canario, A.M.C. Moutinho, O.M.N.D. Teodoro, Applied Surface Science 255 (2009) 7389. [23] R. Saleh, S.P. Prakoso, A. Fishli, Journal of Magnetism and Magnetic Materials 324 (2012) 665–670. [24] R. Yu, K. Yu, W. Wei, X. Xu, X. Qiu, S. Liu, W. Huang, G. Tang, H. Ford, B. Peng, Advanced Materials 19 (2007) 838–842. [25] L. Zhao, Z. Wang, D. Han, D. Tao, G. Guo, Materials Research Bulletin 44 (2009) 984–988.