1 Introduction

Transparent conductive oxides (TCOs) can be widely used in passive and active electronic applications, such as flat-panel displays, photovoltaic devices, light-emitting diodes, smart windows, photodetectors, etc. [1,2,3,4]. However, the performance of p-type TCOs as an essential component used in the active devices is far behind that of n-type TCOs, which seriously limits the application of these active devices [2]. The discovery of the first p-type transparent semiconductor with application potential is CuAlO2 [5], after which increasing attention has been paid to CuI-based oxides, especially those with delafossite structure. CuMO2 with delafossite structure shows attractive p-type optoelectronic properties among various oxides [6,7,8,9]. Unfortunately, the figure of merit (FOM) of such materials is below 10 MΩ−1 [6,7,8,9,10,11,12]. Recently, correlated metals CaVO3 and SrVO3 and their p-type doped materials such as La2/3Sr1/3VO3 with high p-type conductivity of about 742.3–872.3 S·cm−1 have been reported. Unfortunately, their transmittance continues to deteriorate [13,14,15]. Therefore, improving the conductivity of p-type TCOs while maintaining their high transmittance is indeed a great challenge at present.

γ-Phase Cooper Iodide (γ-CuI), an inexpensive and non-toxic p-type semiconductor has attracted much attention. It is an intrinsic p-type semiconductor with low hole effective mass (0.30 m0) and high bulk hole mobility (> 40 cm2·V−1·s−1) [16, 17]. Its wide direct band gap (Eg = 3.1 eV) is advantageous for the fabrication of transparent semiconductors, while its high exciton binding energy of Ebx = 58 meV is also beneficial as used in optoelectronic devices [18, 19]. In addition, n-type CuI can also be realized by Zn doping [20], resulting in the realization of the p-CuI/n-CuI homo-junction, which further greatly expands the application of CuI in various optoelectronic devices. γ-CuI films have been synthesized through various physical and chemical methods, such as sputtering [21], pulsed laser deposition (PLD) [22], solution method [23], thermal evaporation [24], and iodination method [25]. However, the γ-CuI films deposited by different techniques generally have a high haze with coarse grains and a frosted glass-like appearance [16, 26,27,28]. This feature increases the light scattering and significantly reduces the transmittance of the film. At present, many efforts have been made to reduce the haze of CuI film and to improve the transmittance of the film. One of the most attractive means is the preparation of CuI films by the iodination method, i.e. using various precursors such as Cu [25], Cu3N [25], CuS [29] to react with iodide. In addition, Vidur Raj et al. have also incorporated a second phase such as TiO2 into CuI films to inhibit the growth of CuI grains and reduce the haze of the film [30]. However, the results are unsatisfactory: the transmittance of CuI films with a thickness below 100 nm prepared by the above methods is still well below 70%.

Consequently, novel strategy to prepare γ-CuI film with high visible transmittance and low haze is required. In the present work, CuI films were prepared by iodination of Cu3N precursor. The film’s haze as well as the surface morphology can be controlled by adjusting the contact area between iodine and the Cu3N precursor. The transmittance of the as-prepared γ-CuI films can reach 86%. The haze of the as-prepared γ-CuI films can be below 0.7%. Benefit to the high transmittance, the figure of merit (FOM) of the optimal CuI film above 230 MΩ−1can be realized.

2 Experimental section

2.1 Precursor Cu3N thin-film fabrication

2.5 cm × 2.5 cm glass substrates were cleaned with acetone in an ultrasonic bath and then dried with N2. The Cu3N thin films were then deposited on glass substrates from the copper target (purity of 99.99%) by reactive radio frequency (RF) magnetron sputtering. The sputtering power was maintained at 80 W. The background vacuum of the sputtering chamber was 5 × 10–5 Pa. Prior to deposition, the copper target was pre-sputtered with pure Ar for 10 min to remove the oxide layers from the target surface. The deposition pressure is 0.8 Pa with an argon flow of 40 sccm and a nitrogen flow of 20 sccm. The thickness of the Cu3N precursor is approximately 75 nm.

2.2 γ-CuI Thin Film Fabrication

γ-CuI thin films were fabricated via the chemical reaction between Cu3N precursor films and liquid, vapor or solid phase iodine:

$$2{\text{Cu}}_{3}\text{N}+{3\text{I}}_{2}\to 6\text{CuI}+{\text{N}}_{2}$$
(1)

The liquid, vapor, and solid iodination processes are schematically illustrated in Fig. 1a shows the liquid iodination process. First, 0.1 g iodine particles and 0.2 g KI powder were dissolved in 100 ml water as iodine aqueous solution. 5 ml iodine aqueous solution was transferred into a petri dish. Cu3N precursor was immersed in an iodine aqueous solution at room temperature. The iodination reaction is maintained for 5 min. Figure 1b shows the vapor iodination process. Iodine particles were distributed evenly at the bottom of the petri dish. Cu3N precursor films were buckled on the petri dish and exposed to iodine vapor. The petri dish was then kept at room temperature for 5 min. Figure 1c shows the solid iodination process. The Cu3N precursor films were placed face up in a petri dish. The Cu3N film was completely covered with iodine particles for 5 min. The diameter of CuI particles is about 2.5 mm.

Fig. 1
figure 1

Schematic illustrations of the (a) liquid iodination method (b) vapor iodination method and (c) solid iodination method

For the solid iodination method, the different arrangements of contact area and non-contact area were realized by controlling the amount of iodine particles (iodine particle density) on the surface of Cu3N precursor. The low haze sample was prepared by the Cu3N precursor completely covering with iodine particles, and the medium haze sample was prepared by placing 1/2 iodine particles of the low haze sample on the Cu3N precursor surface of the sample. High haze samples were prepared by evenly placing 1/2 of iodine particles on the surface of Cu3N precursor.

2.3 Characterization

The transmittance and reflectance of CuI thin films were measured using a PerkinElmer UV–visible-IR spectrophotometer in the wavelength range of 200 − 2000 nm. The phase structures of the films were analyzed through an X-ray diffractometer (XRD, Rigaku Ultima IV, Japan) with Cu anode (λ = 0.154 nm). The surface morphology and cross-sectional morphology of the films were observed using a field emission scanning electron microscope (FE-SEM; Nova Nano SEM 450, USA). The films’ surface roughness was detected by atomic force microscope (AFM; SPI3800N, SII). The resistivity (ρ), carrier concentration (nh), and carrier mobility (μ) of the films were determined by Hall effect measurements in the van der Pauw configuration (Keithley-4200 SCS, USA) at room temperature. The electrical resistivity of the films under bending was measured by the four-point probe method (RTS-8, China). Haze and CIELAB color values were measured by a spectral haze meter (HAM-300, China).

3 Results and discussion

3.1 Comparison of iodination methods

Figure 2 shows the XRD patterns of the Cu3N precursor and the as-prepared γ-CuI films prepared by liquid, vapor and solid iodination methods, respectively. The 220 nm thickness of CuI is consistent with the expected volume expansion of the precursor about three times. Irrespective of the synthesis method, the γ-CuI phase was formed in the films. Moreover, no other impurity phases such as Cu3N, α-CuI, and β-CuI can be observed. Considering the stability energy of the growth front, it is predicted that the growth rate along the (111) direction is the fastest, and thus, the deposited γ-CuI thin films are expected to exhibit < 111 > -orientation [31, 32]. The sharp and intense (111) peaks of all the as-prepared γ-CuI films indicate the high crystallinity of γ-CuI films in this study.

Fig. 2
figure 2

XRD patterns of (a) Cu3N precursor and b as-prepared γ-CuI films prepared by liquid iodination method (grey line) vapor iodination method (red line) and solid iodination method (blue line)

Top-view SEM images of γ-CuI thin films prepared by different iodination methods are shown in Fig. 3a-c. During the liquid and vapor iodination process, more iodine can be in contact with the precursor compared to solid iodination process. Reactions between the precursor and iodine occur simultaneously at each contact point, resulting in the formation of a CuI shell surrounding the Cu3N particles. The iodine then diffuses inwards through the CuI shell and reacts further with the precursor Cu3N inside. Due to the shape limitation of CuI shell, the volume expansion from Cu3N to CuI leads to the formation of individual larger Cul particles, as shown in SEM images Fig. 3a and b. In contrast, with fewer contact points between the precursor and iodine in solid iodination, the resulting fine particles can grow freely in different directions by diffusion of iodine into the precursor, especially without the shape limitation of the CuI shell layer as mentioned in liquid and vapor iodination methods. Thus, smaller CuI particles and denser morphology can be observed (Fig. 3c). AFM images of γ-CuI thin films prepared by different iodination methods are shown in Fig. 4a-c. The variation trend of the surface roughness is in agreement with SEM results.

Fig. 3
figure 3

SEM images of γ-CuI films prepared by a liquid method b vapor method and c solid method. Grain size distribution of γ-CuI films prepared by d liquid method e vapor method and f solid method

Fig. 4
figure 4

AFM images of γ-CuI films prepared by a liquid method b vapor method and c solid method

The Vis-IR spectra of CuI films prepared through liquid, vapor, and solid iodination methods are shown in Fig. 5a, where the solid line represents the transmittance and the dashed line represents the reflectance. According to the transmittance analysis shown in this figure, an obvious difference between CuI sample prepared by Cu3N solid iodination method and other two samples can be observed. The relatively high transmittance of the former sample with obvious Fabry–Perot interference fringes corresponds to flat surface morphology and low surface roughness. On the other hand, CuI films prepared by liquid and vapor iodination methods present degraded transmittance due to their rough surface, as shown in Fig. 3. Haze test is another approach to analyze the film’s surface quality which is relevant to light scattering in transparent films. The results show that the haze of CuI films prepared by Cu3N liquid and vapor iodination is between 30–50%, while the haze of CuI films prepared by Cu3N solid iodination can be controlled within 10%.

Fig. 5
figure 5

a The Vis-IR spectra and b the photographs of CuI films prepared by liquid, vapor, and solid iodination methods; c the transmittance as a function of the photon energy of CuI films prepared with different iodination methods. d the absorption coefficient as a function of the photon energy of CuI films prepared with different iodination methods. (The thickness of CuI films prepared from Cu3N is about 220 nm)

As for the reflection of the films prepared via different iodination methods, the light reflection from their surface includes specular reflection and diffuse reflection. For CuI film prepared by solid iodination method, the specular reflection from its smoother surface is much more obvious than other two samples. Meanwhile, there is more diffuse reflection occurs on the samples prepared through liquid and vapor iodination methods. As the spectrophotometer equipped here can only detect specular reflection, the reflection value of CuI film prepared by solid iodination method is relatively high.

The photographs for liquid, vapor and solid iodination methods are shown in Fig. 5b. The CuI film produced by liquid iodination has many large and spare grains and a rough surface. When a light beam irradiates the film with many cracks, the light is scattered and propagated in different directions [25]. As a result, a frosted glass-like appearance can be observed on CuI thin films fabricated by the liquid and vapor iodination methods, and the blurred view of the picture under these two samples are obtained. In contrast, CuI film produced by the solid iodination owns smaller grains and smoother surface, and better light transmission can be achieved. Thus, much clearer view of the picture under this sample is observed.

Furthermore, a pronounced excitonic absorption of CuI films appears at EZ1/Z2 = (3.05 ± 0.01) eV (Fig. 5d). Since γ-CuI has large exciton binding energy of about 0.062 eV [19, 22], the Eg of CuI of about 3.11 eV can be obtained by calculating the sum of EZ1/Z2 and the exciton binding energy of CuI. This result is consistent with theoretical calculations and other experimental reported [16, 19, 24]. Meanwhile, the transition due to spin–orbit coupling appears at EZ3 = (3.68 ± 0.01) eV.

The absorption coefficient was calculated by the Beer-Lambert relation

$$\alpha =\frac{1}{t}\mathit{ln}(\frac{1}{T})$$
(2)

where t is the thickness and T is the transmittance. Due to the poor surface quality of CuI films prepared by Cu3N liquid and vapor iodination, more significant light scattering and absorption occurs in these films, resulting in worse transmittance. According to the Beer-Lambert absorption formula, the higher the transmittance, the lower the absorption coefficient for the same film thickness (Fig. 5d).

In this study, the solid-state reaction is used to explain the obvious difference of CuI films prepared by iodization. Generally, the solid-state reaction is divided into two processes: the interfacial reaction and the diffusion of reactive species in the precursor. Jander’s equation is one of the most popular to examine solid-state reaction kinetics under isothermal conditions [33, 34]. Its expression is t

$${\left(\text{1} - \sqrt[3]{1-\beta}\right)}^{2}=\frac{2D{c}_{0}}{{r}^{2}}\cdot t=k^{'}\cdot t$$
(3)

where \(\beta\) is the conversion rate of reactants, D represents the diffusion coefficient, c0 represents the reactant concentration at the interface between the layer of reaction products and the reactant, t represents the time and r is the spherical reactant particles’ radius.

A diagram of CuI grain growth basing on Jander’s equation and intuitive results has been proposed, as shown in Fig. 6. Compared to the solid iodization method, the liquid and vapor iodination methods have many iodine molecules in contact with the Cu3N precursors, which makes the interface reaction area much larger. Thus, in liquid and vapor iodination situation, reactions between Cu3N precursor and iodine occur simultaneously at each contact point, resulting in the formation of a CuI shell surrounding the Cu3N particles. Meanwhile, iodine aqueous solution and iodine vapor have lower concentrations at the reaction interface than solid iodine particles, which makes all reactions slower. As a result, the iodine can diffuse inward through CuI shell and further react with the precursor Cu3N inside. Due to the shape restriction of CuI shell, the volume expansion from Cu3N to CuI leads to the formation of individual larger Cul particles, as shown in Fig. 6a. Then, coarser CuI grains and rough surface of the films are obtained (Fig. 3a and b). In contrast, fewer contact point between the precursor and iodine during solid iodination (Fig. 6b), resulting fine particles can grow freely in various directions through the diffusion of iodine into the precursor, especially without the shape restriction of CuI shell layer as mentioned in liquid and vapor iodination situations. Thus, smaller CuI particles and denser morphology can be observed (Fig. 3c).

Fig. 6
figure 6

Schematic illustrations of iodination process for (a) vapor, liquid, and b solid iodination methods

3.2 A new strategy

Based on the above analysis, it is inferred that the key factor in improving the transmittance of CuI films is to reduce the grain size to form a denser film structure with a smooth surface. A denser film structure with a smooth surface can reduce the light scattering of CuI films. The effect of solid iodization is the best among the three iodization methods. Thus, solid iodization process was further investigated.

When iodine particles come into contact with Cu3N precursor, they can be divided into contact areas and non-contact areas. The schematic diagram of Fig. 7a, b shows the preparation process of these two cases. In the contact reaction case, the contact area is prepared by completely covering the iodine powder. In the non-contact reaction case, the iodine particles are placed on the edge of the precursor for iodization, and the middle area is the non-contact iodization area.

Fig. 7
figure 7

a Schematic illustrations of contact iodination (left) process and non-contact process (right); b, c SEM images of solid iodination in the contact area and non-contact area d Comparative UV–vis-IR spectra for these two processes; e, f SEM images of solid iodination in the contact iodination and non-contact iodination

After iodination reaction, the γ-CuI films prepared by iodine particles contact iodination are smoother as shown in Fig. 7b and e. The root-mean-square (RMS) roughness of the scanned area of 10 × 10 μm2 was 27.1 nm. CuI films with a smooth surface present a high average transmittance of about 85% in visible light region. This can also be explained by the above analysis of solid-state reaction. Meanwhile, for the non-contact iodization condition, γ-CuI films have a frosted-glass-like appearance and a low average transmittance of about 73% (Fig. 7d) due to small uneven grains on the substrate in the no-contact region, which is consistent with a previous report [25].

Here, the iodination process can be divided into two steps:

  1. (1)

    the direct reaction between solid-state iodine and Cu3N precursor (denoted as S1), which is favorable to the nucleation of CuI grains.

  2. (2)

    mass transfer process of iodine to the surface of Cu3N precursor (denoted as S2), which is favorable to the growth of CuI grains. The rate of this mass transfer process is relatively slow, providing sufficient time for nucleus growth.

It is clear that S1 is the dominant process in the case of the contact reaction, whereas S2 is the dominant process in the case of the non-contact reaction. In the former case, the high density of contacts brought about by the iodine powder will result in a high density of nucleation sites. The rate of nucleation is much faster than that of crystal growth, resulting in fine grains and smooth surfaces. In contrast, in the latter case of the non-contact reaction process, the mass transfer rate is relatively slower than the reaction between iodine and Cu3N precursor, resulting in the CuI nucleus growth rate being greater than the nucleus formation rate. Thus, coarser grains are formed.

Improving the smoothness and grain density of the film can reduce light scattering and then increase the transmittance of the films and eliminate the frosted glass-like appearance of the film [25, 35]. According to the above finding results, we further prepared CuI samples with adjustable transmittance and haze by solid-phase iodination. Figure 8a shows the films prepared by solid iodination method with different contact areas of iodine particles and Cu3N precursor. Green color represents contact areas, gray color represents the non-contact area. In crowed case, when iodine particles directly contact the precursor, the initial nucleation rate increases, and the original crystal nuclei grow together to refine the grains and form a dense CuI film. As a result, the CuI film has a smoother surface, the sample scatters less light and the surface roughness of the film is lower. The transmittance and reflectance of three samples are shown in Fig. 8b. It is clear that CuI sample prepared with crowed condition presents best transmittance (> 85%). This is followed by the samples prepared with medium and sparse conditions. The variation of the film’s transmittance as a function of the film’s haze is shown in Fig. 8c. The films prepared with more non-contact areas possess high haze and low transmittance due to their rough surface. Therefore, the haze and transmittance of CuI films can be well controlled by adjusting the contact area between iodine particles and Cu3N precursor.

Fig. 8
figure 8

a Schematic illustrations of three arrangements and their scattering, the green area represents contact areas, the gray area represents no contact area (b) Comparative UV–vis-IR spectra for three arrangements (c) Comparative visible transparent and haze for three arrangements, contact, no contact, and vapor iodination and d Comparative Lab color for three arrangements, the images (inside) are the color corresponding to the LAB color values. The films have a thickness of ~ 240 nm

Figure 8d shows the Lab color of those films. The negative value of the ‘a’ axis tends to be green, and the positive value tends to be red. The negative value of the ‘b’ axis tends to towards blue, and the positive value tends to towards yellow. When the distribution of smooth area is denser, the surface of CuI film is so smooth. The film has obvious Fabry Perot interference due to smooth surface as shown in Fig. 8b. The CIELab color values of a and b are less than 10 as shown in Fig. 8d. It indicates that the interference color does not affect its application as a transparent conductive film.

Different devices require different haze on the electrode. For example, touch-panel displays require transparent electrodes with low haze [36,37,38], but solar cells require transparent electrodes with high haze to improve light harvesting [39,40,41]. In this work, the adjusted haze of CuI film from 0.7% to 22% can be realized by this strategy to satisfy the needs of different electronic devices. This will greatly expand the application of CuI as transparent conductive films.

3.3 Optoelectric performance of p-type TCs

The overall optoelectric performance of the TC can be quantified by the Haacke figure of merit (FOM) [42], ΦTC = T10/Rs where T is the average transmittance in the visible region and Rs is the sheet resistance. A larger value of FOM indicates the better optoelectric performance of the TC. As shown in Table 1, the FOM of most novel p-type TCs is less than 20 MΩ−1. In contrast, the as-deposited γ-CuI film shows a FOM about 11 times higher, up to 233 MΩ−1, while providing a high Tvis above 86%. These results demonstrate the superior TC optoelectric performance of our γ-CuI films in all measures.

Table 1 Transmittance, resistivity, and the FOM of CuI films and other p-type TCs

Figure 9 shows the transmittance, conductivity, and FOM values of CuI films prepared by the new strategy and other methods. The films prepared by the new strategy have better performance than other CuI films. Compared to other preparation methods, this new strategy is simple and controllable. In addition, the haze can be adjusted from 0.7% to 22%. This expands the application range of CuI as a transparent conductive film.

Fig. 9
figure 9

Figures of merit. Graphical representation of electrical resistance, optical transmission, and FOM for as-prepared CuI films (red star) and other CuI films (PLD-CuI: Preparation of CuI films by pulsed laser deposition; Iodi-CuI: Preparation of CuI films by iodination; Sup-CuI: Preparation of CuI films by sputtering; TE-CuI: Preparation of CuI films by thermal evaporation; MA-CuI: Preparation of CuI films by mister atomizer) [22, 36, 47, 49,50,51,52]

4 Conclusion

In summary, γ-CuI films have been prepared via solid, vapour and liquid iodination of Cu3N precursors in this work. The SEM and AFM images show that the γ-CuI films prepared by the solid iodination method have smaller and denser grains, and lower surface roughness. These factors result in the high transmittance and good electrical properties. Furthermore, the contact and non-contact iodination between the Cu3N film and iodine particles in solid iodination have been investigated. The results show that solid iodination process can be divided into two steps. The first step is the reaction between the interface of iodine granules and Cu3N film. The second step is the diffusion of reactive species to the surface of the Cu3N layer. Different degrees of dominant steps will result in films exhibiting different surface roughness and haze, which in turn affects the optical properties of the film. Based on this phenomenon, a new strategy has been proposed to fabricate CuI films with high transmittance and low haze. The optimal FOM of the CuI film prepared by this new strategy is > 230 MΩ−1. Compared with other p-type transparent conductive materials and CuI films prepared by other methods, the optoelectric performance of CuI film prepared by this strategy is relatively excellent. Moreover, this strategy can also make the haze of the CuI film controllable. The results of this work can promote the application of CuI film in optoelectronic devices.