Figure 2 shows changes in the morphology of the CdS films obtained from the formulation with ammonium chloride as a complexing agent and deposition times of 55 and 90 minutes. Both A-55 and A-90 films were composed of submicron spherical-shaped particles with few particles bigger than the average size. Average particle size was higher for A-90, indicating that a higher deposition time produces bigger particles. The A-55 and A-90 films had discontinuous surfaces with free spaces of about 600 nm and 400 nm, average particle sizes of 71.3 nm and 112 nm, and particle-size distribution functions with Full Width at Half Maximu (FWHM) of 0.50291 nm and 0.37171 nm, respectively. Both films showed nodular growth. It is worth noting that higher deposition times allowed the coalescence of nanoparticles, hence minimizing the number of free spaces and, apparently, film roughness (see Figures 2 c and 2d).

Figure 2

Figure 3 shows changes in the morphology of CdS films obtained from the formulation with sodium citrate as a complexing agent and deposition times of 55 and 90 minutes. Both B-55 and B-90 films had continuous surfaces, without cracks, made of splinter-shaped nanoparticles. The films showed a few submicron particles resulting from the coalescence of nanoparticles, achieving sizes of 500 nm for the B-90 film and 400 nm for the B-55 film (see Figures 3a and 3b). This translated into surfaces with heterogeneous morphology.

Figure 3

The micrographs of both films showed only a few black spots indicating zones with different chemical compositions. This suggests that films had a nearly homogeneous chemical composition.

B-55 and B-90 films showed splinter-shaped particles with width distribution functions centered at 31.7 nm and 37.5 nm, and FWHM of 0.603 nm and 0.5198 nm, respectively. At first, film roughness appeared to be in the nanometric range, like the width of the splinters. The average size of splinters increased with deposition time, indicating that a higher deposition time produces bigger splinters.

From the above, it appears that the B-90 film resulted in a better morphology for potential solar cell applications.

Figure 4 shows a SEM micrograph in LABE mode of the copper film deposited on the B-90-Troom sample, in order to examine its morphology for the purpose of copper doping. The film turned out to be continuous and without cracks, with a few particles of submicron dimensions (~500 nm). There are three things worth mentioning about the copper film: 1) The film appeared to have a uniform chemical composition across its surface; 2) The film's surface was composed of spherical-shaped nanoparticles; and 3) At first glance, film roughness appeared to be in the nanometric range.

Figure 4

Figure 5 shows the morphology of the B-90-T_a film as a function of T_a. Figures 5 a, 5c, 5e and 5g show micrographs of B-90-T_a samples for T_a = 140, 170, 200 and 230 °C, respectively. The micrographs reveal a surface composed of spherical-shaped particles and the presence of relatively big particles and potholes. The films annealed at 140 and 170 °C turned out to be morphologically similar (Figure 5d) but had different average particle size: 160 nm and 38 nm, respectively.

Figure 5

For T_a of 200 °C and higher, however, there was a change in morphology, with film surfaces composed of nanoclusters of spherical-shaped nanoparticles of 10 nm diameter, as shown in Figures 5 f and 5h. The average size of the clusters was of 65 nm and 56 nm for T_a = 200 and 230 °C, respectively. At first glance, film roughness appeared to be in the nanometric range. It is evident from the lower secondary electron images in Figures 5a, 5c, 5e and 5g that the films had a homogeneous chemical composition across their surface for all T_a values.

Figure 5b shows a micrograph of a cross section of the B-90-140 sample obtained by using low-angle backscattered electrons (LABE). A bilayer film of about 160 nm is observed, a result representative of all synthesized samples.

The upper film was 15 nm thick on average, but was thicker in certain zones. The inset in figure 5b shows an EDS analysis of the zone evidencing that Cu ions were incorporated into the film.

The crystalline structure of the A-55, A-90, B-55 and B-90 samples was obtained by XRD. Figure 6 shows that the crystalline phase of CdS films was cubic and did not vary with deposition time or the formulation, which is consistent with the results of Atefeh Jafari et al. [21]. The XRD pattern showed a wide band and three main peaks c entered at 2θ = 24.89 °, 26.51 ° and 43.99 °. The wide band was attributed to amorphous material. The peak centered at 2θ = 24.89 °, particularly the highest diffraction peak (111), was attributed to the presence of monoclinic CdSO₄-H₂O [22].

Figure 6

Indeed, this peak disappeared after thermal annealing, used to diffuse copper into the CdS film (see Figure 8). The peaks centered at 2θ = 26.51 ° and 43.99 ° were attributed to the (111) and (110) planes of the cubic crystalline phase of CdS. All samples had a lattice constant of 5.8196 Å and a preferred growth orientation along the (111) plane.

The crystalline grain sizes were obtained by Debye-Scherrer formula [23]:

D   =   0.9 λ β cos ⁡ θ

Where: D = crystalline grain size, λ = CuKɑ radiation at 1.5426 Å, β = Full Width at Half Maximum of diffraction peak and θ = half of 2θ diffraction angle . The calculated D values were: 31.4 nm, 20.4 nm, 19.4 nm and 14.5 nm for the A-90, B-90, A-55 and B-55 films, respectively. This provides evidence of the greater effectiveness of sodium citrate (Formulation 3) than ammonium chloride (Formulation 2) as a complexing agent.

In Figure 7, the B-90 sample showed a film containing regularly spaced, asterisk-shaped black entities. These asterisks had a narrow size distribution function centered at 30 - 40 nm and were composed of several superposed nanocrystals smaller than 10 nm (Figure 7d). This is consistent with the average grain size obtained by XRD. These nanocrystals showed crystalline planes located far away from each other at 3.33 Å, corresponding to the (111) planes of cubic CdS (Inset in figure 7d). The electron diffraction pattern of the film (Inset in figure 7e) showed crystalline planes corresponding to the (111) planes of cubic CdS, with d(111) = 3.35 Å. This resulted in a lattice constant of 5.8024 Å, which is consistent with the lattice constant of 5.8196 Å obtained by XRD.

Figure 7

Figure 8 shows the XRD pattern of the B-90-T_a samples as a function of T_a, as well as the diffraction pattern of the B-90-T_room sample, that is, the sample that had a copper film deposited on its surface, but was not heat treated. The first sample showed peaks centered at 27.3 ° and 44.61 °, as well as a wideband. The peaks were attributed to the (111) and (220) planes of cubic CdS, and the wideband, to material that was still crystallizing and/or the amorphous copper film deposited by thermal evaporation [24].

Figure 8

After thermal annealing, the XRD pattern presented peaks at 26.76 °, 36.3 °, 43.3 °, 44.25 °, 50.6 ° and 52.4 °. The peaks centered at 26.76 °, 44.25 ° and 52.4 ° were attributed to the (111), (220) and (311) planes of cubic CdS [21]. The peaks centered at 43.3 ° and 50.6 ° belonged to cubic copper [25]. The peak centered at 36.3 ° as well as the low-intensity peaks in the region of 25 - 26 ° were attributed to traces of hexagonal CdS, in agreement with Kashiwaba et al. [26].

Although it might be thought that the 36.3 ° peak could correspond to the (111) planes of cubic Cu₂O [25], this could not be the case here because Cu₂O is formed simply by exposing copper to the oxygen in the air, and this 36.3 ° peak was not observed in the diffraction pattern of the B-90-T_room sample, which had a copper film exposed to the air for a long time. The average grain size for cubic CdS, hexagonal CdS and cubic Cu was 16 nm, 8 nm and 13 nm, respectively. The sample that was thermally annealed at T_a = 140 °C was the only one not showing CdS in its hexagonal phase.

All thin films showed CdS mainly in its cubic crystalline phase and a preferential growth along the (111) planes. Figure 8 shows a shift of cubic peaks towards lower 2θ values, indicating a uniform expansion of the CdS lattice as copper replaced Cd. There are two things worth mentioning in the XRD pattern of the B-90-T_a films as a function of T_a: the presence of cubic copper at T_a = 140 °C, which disappears as T_a increases; and the presence of hexagonal CdS, starting at T_a = 170 °C and becoming more evident as T_a increases.

Figure 9 shows the resistivity of copper-doped CdS film as a function of T_a. It can be observed that resistivity decreased drastically as T_a decreased, which seemed to be associated with a higher amount of crystalline copper and/or the predominance of cubic CdS over hexagonal CdS, in accordance with the results shown in the section Crystalline structure above. Further studies are required, however, to clarify the relationship among the above factors, and this is foreseen as future work. The lowest resistivity obtained was 5.822 χ 10^-5 Ω-cm for the B-90-140 sample, which showed the highest amount of cubic copper and no hexagonal CdS.

Figure 9

The absorbance spectra of the A-90, A-55, B-90 and B-55 samples are shown in Figure 10.

Figure 10

The energy band gap (Eg) was calculated using the Tauc relation [27], resulting in values of 2.27 eV, 2.32 eV, 2.43 eV and 2.52 eV for the A-90, A-55, B-90 and B-55 films, respectively (Figure 10b).

It appears that Eg was larger for lower deposition times, regardless of whether sodium citrate or ammonium chloride was used as a complexing agent. This is in agreement with the grain sizes obtained by XRD for the same samples. Regardless of deposition time, E_g values were higher for the sodium citrate formulation than for the ammonium chloride formulation. The fact that E_g became bigger with decreasing grain size can be explained in terms of quantum confinement effect that becomes stronger in nanostructured films [28].

Figure 11 shows the absorbance spectrum of copper-doped CdS film as a function of T_a. It can be seen that all samples had an absorption band centered at 490 nm, which was attributed to the surface plasmon of copper nanoparticles [29] and is consistent with the presence of copper peaks shown in XRD results.

Figure 11

The plasmon absorption was most intense at T_a = 140 °C, followed by T_a = 170 °C, in agreement with the relative intensities of the XRD peaks associated with cubic Cu (Figure 8). The position of the plasmon band in the absorbance spectrum is consistent with the 13-nm grain size obtained from the XRD results above.

The Eg of the B-90-T_a films was calculated as a function of T_a using the Tauc relation (Figure 11b). Eg resulted very similar for all T_a values: 2.93 eV at 140 °C; 2.82 eV at 170 °C; 3.04 eV at 200 °C; and 2.90 eV at 230 °C. The average Eg was 2.92 eV, which is consistent with a CdS crystal lattice that expands when Cu is incorporated into it, but that still preserves its cubic phase [30].

Figure 12 shows the photoluminescence spectra of CdS and Cu-doped CdS films. At first, it seems that the spectra of CdS films obtained from the sodium citrate formulation are similar to those obtained from the ammonium chloride formulation, since both present two wide emission bands in the ranges of 500 - 600 nm and 600 - 800 nm. However, a closer examination reveals that the band in the 500 - 600 nm region is asymmetric for the films prepared from ammonium chloride, but not for those prepared from sodium citrate. The band in the 600 - 800 nm region is also narrower when the ammonium chloride formulation is used. After deconvolution of the spectra, the films obtained from the ammonium sodium citrate formulation presented bands centered at 524 nm, 685 nm and 730 nm. The band centered within the energy chloride formulation presented emission bands centered at 543 nm, 580 nm and 670 nm, while those obtained from the range of 2.28 - 2.37 eV (524 - 543 nm) was present for both formulations, corresponding to a band-to-band electronic transition, and could be attributed to the bandgap energy of each corresponding film [31,32]. This is in agreement with the absorbance measurements presented above (Figure 10). The PL emission band centered at 580 nm was attributed to electronic transition from donor energy states to valence band. Donor energy states moved into the energy gap due to the presence of interstitial cadmium (Cd_i) [33,34]. The emission band centered within the 1.81 - 1.85 eV energy range (670 - 685 nm) was present for both formulations and was attributed to the presence of cadmium vacancies (V_Cd), particularly with the electronic transition from the conduction band (CB) to V_Cd [34,35]. This transition could be assumed to take place in a range of 1.81 - 1.85 eV below the conduction band, implying that the V_Cd states are located in a range of 0.47 - 0.71 eV above the valence band, which is reasonably consistent with earlier reports [36,37]. The broad emission peak centered at 730 nm was attributed to the presence of sulfur vacancies (V_s) [31,38]. The presence of the 730 nm band in the films synthesized with sodium citrate, and its absence in those obtained from ammonium chloride, indicate that sodium citrate, but not ammonium chloride, promotes the formation of Vs. On the other hand, the presence of the 580 nm band in the films synthesized with ammonium chloride and its absence in those obtained from sodium citrate indicate that ammonium chloride, but not sodium citrate, promotes the formation of Cd_i.

Figure 12

The relative intensity values of PL emission bands, corresponding to a band-to-band electronic transition and electronic transitions assisted by defects, show that using the ammonium chloride formulation resulted in CdS films with better optical quality (or a minor number of defects).

Figure 12b shows the PL emission spectra of the B-90-T_a films as a function of T_a. The spectra showed a blue shift of the peak associated with a band-to-band electronic transition, which is consistent with the absorbance measurements presented above, showing an increase in Eg when CdS films were doped with Cu (Figure 11). Two points are also worth mentioning about the emission band associated with crystalline defects: 1) It only shows the V_s-related emission and the V_Cd-related emission disappears, which is consistent with the incorporation of Cu ions into Cd sites; and 2) Its relative intensity decreases, meaning that film optical quality increases with Cu doping. Only the film obtained by using T_a = 200 °C showed the presence of VCd, although more studies are needed to clarify this finding since it was not the purpose of this study.