Nirali A.
Bhatt
a,
Lizabeth
Quigley
a,
Shiyu
Zhou
a,
Anirutha
Gnanasabai
a,
Abhijeet
Choudhury
a,
Yizhi
Zhang
a,
Jianan
Shen
a,
Juanjuan
Lu
a,
Aleem
Siddiqui
c,
Raktim
Sarma
cd and
Haiyan
Wang
*ab
aSchool of Materials Engineering, Purdue University, West Lafayette, IN 47907, USA. E-mail: hwang00@purdue.edu
bSchool of Electrical and Computer Engineering, Purdue University, West Lafayette, IN 47907, USA
cSandia National Laboratories, Albuquerque, NM 87123, USA
dCenter for Integrated Nanotechnologies, Sandia National Laboratories, Albuquerque, NM 87123, USA
First published on 1st May 2025
ZnO thin films have attracted significant interest in the past decades owing to their unique wide band gap properties, piezoelectric properties, non-linearity and plasmonic properties. Recent efforts have been made in coupling ZnO with secondary phases to enhance its functionalities, such as Au–ZnO nanocomposite thin films with tunable optical and plasmonic properties. In this work, magnetic nanostructures of Ni are incorporated in ZnO thin films in a vertically aligned nanocomposite (VAN) form to couple magnetic and plasmonic response in a complex hybrid metamaterial system. Nickel (Ni) is of interest due to its ferromagnetic and plasmonic properties along with gold (Au) which is also plasmonic. Therefore, two approaches, namely, tuning of the deposition pressure and use of a ZnO–Au seeding layer have been attempted to achieve unique Ni nanostructures in addition to tuning of the microstructure. Together, both approaches demonstrate a range of microstructures such as core–shell, nanodisk, nanocup, and nanocube-like morphologies not previously attempted. Additionally, there is effective tuning of properties. Specifically, the seeding layer thickness causes hyperbolic behavior as well as redshift in the surface plasmon resonance (SPR) wavelength. The addition of the ZnO–Au seeding layer directly influences the optical properties. Plus, regardless of the different approaches, the films demonstrate magnetic anisotropy based on the composition and microstructure of the film which impacted the saturation magnetization and coercivity. This study demonstrates the potential of ZnO-based complex hybrid metamaterials with coupled electro-magneto-optical properties for integrated photonic devices.
VANs are a unique type of thin film nanostructure that is comprised of a pillar in matrix form. The structure itself can be tuned in terms of strain, functionality, and properties simply based on the materials that are chosen for the system. The large anisotropy that presents in this type of nanostructure aids in the strain tuning and results in the change in properties. VAN presents an advantage in the ability to choose two or more different materials with distinct properties and create a single film that couples those properties (i.e. optical, magnetic, and electric) and creates new emergent properties not present in either material, hence a metamaterial.5 Plus, VAN are self-assembled nucleation and growth which means that the fabrication is a relatively easy one-step growth method6 as long as the deposition conditions which control the kinetics, surface energy, and strain are optimized.7
Zinc oxide (ZnO) is a well-known oxide that has been used in metal-oxide hybrid metamaterials because it is a wide band gap semiconductor, is piezoelectric, and can be a transparent conductive oxide. Due to the benefits offered by VAN, ZnO has been incorporated into VAN growth and studied. Previous ZnO-based VAN have been studied. Specifically, there is the ZnO–Au system where the oxygen background pressure and laser frequency tuning8 were examined in addition to Au concentration variation.9 There were also many ZnO-alloyed films like the ZnO–Au–Cu alloy and ZnO–Au–Al alloy where a seeding layer approach was used to improve the growth of these films.10 The ZnO–Ag–Au alloy study looked at the impact of background pressure on morphology.11 For ZnO-Oxide studies, ZnO–LSMO (lanthanum strontium manganite, La0.7Sr0.3MnO3) is the main material system that has been explored thus far that shows VAN structure and it explored the impact of film thickness on magnetic properties and growth.12 Additionally, all the ZnO-metal material systems explored mainly at optical property tuning with the microstructure which presents a large gap in this material system. Since Au, Cu, Al, and Ag are not ferromagnetic, there is no understanding of the magneto-optical coupling in ZnO-based VAN films.
In this work, we propose to grow a new ZnO–Ni VAN system to close the gap by using Ni, a known ferromagnetic material, to understand how the magnetic properties couple with the optical properties. Additionally, this system demonstrates the growth of Ni as a secondary phase instead of the typical dopant material that is commonly seen in prior ZnO–Ni studies.13–16 Dopants are usually added to ZnO to tune its bandgap.17 A secondary phase could be advantageous because the properties of the matrix material are either maintained or enhanced due to the secondary phases by strain tuning and introduction of other functionalities. Ni can also grow as NiO, under oxygen background. NiO is antiferromagnetic in bulk form and has shown to be ferromagnetic at times in nanoparticle form.18 To achieve Ni–ZnO nanocomposite, two methods were used. The first method is based on the control of vacuum level. The second method employs a thin ZnO–Au seeding layer that was deposited before the ZnO–Ni film of interest. These methods are selected based on the impacts of both deposition pressure and seeding layers. Specifically, the vacuum level of thin film growth, usually in the range of 10−7 to 10−10 Torr, plays an important role in the mean free path of the adatoms and the impurity level of the film. A seed layer on the other hand, is a thin film layer grown underneath the main film of interest considering its ability to reduce strain, improve crystallinity19 and nucleation20 of the film grown above it, enhance properties,21 or aid in the growth of the film of interest. Fig. 1 shows the proposed films that will be discussed in this paper. Fig. 1a is the ZnO–Ni film grown in a high vacuum (2.1 × 10−6 Torr) which will be referred to as HVZN in this paper. Fig. 1b and c show the thin seeding layer and ZnO–Ni film, and a thick seeding layer and ZnO–Ni film, which will be referred to as 10ZA50ZN and 50ZA80ZN, respectively. Additionally, a low vacuum ZnO–Ni film grown at 7.4 × 10−6 Torr is shown in Fig. S1a† which will be referred to as LVZN. We aim to achieve tunable ZnO–Ni nanostructures with tailorable optical and magnetic properties that could find unique applications in photonics, spintronics and other device structures.
![]() | ||
Fig. 2 XRD θ–2θ scans for the (a) high vacuum sample, (b) thin seeding layer and thin film layer sample, and (c) thicker seeding layer and thin film layer sample. |
Fig. S1b† shows the XRD for the LVZN film grown at 7.4 × 10−6 Torr which shows epitaxial growth of ZnO (0002), NiO(111), and Ni (111) on c-Al2O3 (0006). This relationship works to explain the connection between the hexagonal wurtzite structure of the ZnO and the cubic Ni and NiO. Since Ni has a face-centered cubic crystal structure, the Ni rotates to be in the (111) to match its 3-fold symmetry with the 6-fold symmetry of hexagonal substrate and matrix.8,9 It shows clearly textured growth. Plus, there is a peak of both ZnO(101)/NiO(111) suggesting a strain coupling and co-growth of these two phases in the thin film which will be explained by the detailed microstructure analysis in the next section.
Fig. 2b shows the XRD for 10ZA50ZN while Fig. 2c shows the XRD for 50ZA80ZN. Clearly, both films show epitaxial growth of ZnO (0002), NiO(111), Au(111), and Ni(111) on the c-cut sapphire substrate. Fig. 2b shows that there is a more prominent Ni (111) peak as compared to the NiO(111) while in Fig. 2c there is the inverse with a more prominent NiO(111) peak as compared to Ni(111). Clearly, the ZnO–Au seeding layer aids in the growth of epitaxial ZnO, Au, and Ni onto the substrate.
The seeding layer thickness could influence the crystallographic orientation and phase distribution and thus is explored. The seeding layer thickness in the aforementioned samples is about 10 nm and 50 nm for the 10ZA50ZN and 50ZA80ZN samples, respectively. First, for the 10ZA50ZN sample, the seeding layer thickness is too thin that the Au, though Au(111), is not affecting the crystallographic nature of the ZnO–Ni growth on top. That is why the Ni(111) peak is comparable to what is seen in HVZN. There is a minor NiO(111) peak that could be explained by the small strain introduced by the thin seeding layer that causes Ni to grow as NiO(111) to accommodate the strain. Furthermore, for the thick seeding layer sample, 50ZA80ZN, there is a different phase distribution induced in the film. The 50 nm ZnO–Au layer results in the formation of well-defined Au particles as will be discussed in more detail with the microstructure next. In short, the aid of the Au (111) nanoparticles allows for nucleation of the NiO(111) which is the main phase of Ni that forms due to the ZnO co-growth and interface interactions.
In contrast, LVZN in Fig. S2† shows obvious vertical growth of Ni with the ZnO growing along with Ni with a thickness of ∼40 nm. This suggests a ZnO core and NiO shell structure that has been verified in Fig. S7† from a line scan. This microstructure supports the results shown in XRD. Due to the Ni growing where the ZnO grows, the more prominent NiO peak is justifiable. Additionally, the combined ZnO and NiO peak are supported by the core–shell structure. The lack of core–shell growth in the high vacuum film supports the greater intensity Ni(111) peak seen in the XRD. The core–shell seen in LVZN and the cube-like Ni growth in HVZN is due to those microstructures having the lowest energy in the system.24 It is surprising, however, that LVZN demonstrated a better epitaxial growth of ZnO(0002) while HVZN did not. This is counter-intuitive since typically high-quality epitaxy requires high vacuum growth.
The two samples with thin and thick seeding layer are shown in Fig. 3f–k for the thin seed layer sample 10ZA50ZN. For 10ZA50ZN, the Ni nucleates at the film–substrate interface as small particles in addition to a ‘nanodisk’ structure. The ZnO is present as the matrix material and the ZnO–Ni film has a thickness of 50 nm. The Au that is present due to the ZnO–Au seeding layer is only seen at the bottom of the film due to the very thin layer. The Au mostly looks like clusters that are not big enough to be defined as nanoparticles. It is also clear that the Ni grows where there is less ZnO shown. This explains the prominent Ni peak in the XRD as well as the very low intensity NiO.
For 50ZA80ZN (Fig. 3l–q), the Au grows as well-formed particles in Fig. 3p due to the thicker layer. Unlike the thinner seeding layer seen in Fig. 3j, the Au in 50ZA80ZN grows as particles that are slightly elongated vertically like a ‘pill’ shape. The seeding layer in this film has a thickness of 50 nm, five times that of the thin layer in Fig. 3j. This increased thickness is what leads to this better formation of Au particles. Additionally, there is a unique Ni ‘nanocup’ structure that only nucleates where an Au nanoparticle is present. Interestingly, through the EDS mapping of the Ni (Fig. 3n) and Zn (Fig. 3o), it is clear that the Ni growth is accompanied by ZnO. There is a clear interface in the “nanocup” where the Ni meets the ZnO. This could mean there is a possible core–shell structure.
Because of this interesting ‘nanocup’ structure, plan-view TEM/STEM images were taken and Fig. S8† shows the plan-view STEM images of 50ZA80ZN. Interestingly, from the HAADF STEM seen in Fig. S8a,† the film is made up of multiple domains which was not expected considering the highly textured growth seen in XRD. Ni (Fig. S8b†) is the major composition in the different domains that are not well connected. These Ni domains are filled by ZnO as seen in Fig. S8c.† The intensity of Au varies as typical plan-view samples have a variation in thickness. Regardless, there are Au particles. From the combined EDS map in Fig. S8e,† it is noted that at the edges of the grains it shows a darker blue suggesting that there is a greater concentration of Ni. This overall plan-view structure confirms the effects of Au seeding evidenced from the cross-section images seen in Fig. 3. The Ni nucleates only where there is a well-formed Au nanoparticle.
To better understand the change in composition of these films, Table S1† is included in the supplementary. The LVZN film (Table S1a†) has about 58%Zn and 42%Ni atomic fraction while for HVZN (Table S1b†) these change to about 6%Zn and 94%Ni. For the seeding layer films, the atomic fraction of Ni is more comparable, with ∼36% for 10ZA50ZN and ∼30% for 50ZA80ZN. Similarly, the Zn atomic fraction is ∼62% for 10ZA50ZN and ∼67% for 50ZA80ZN. The Au atomic fraction goes from 0.71% in 10ZA50ZN to 2.64% in 50ZA80ZN. It is clear that the pressure and seeding layer thickness varies the composition of the films.
This microstructure explains the higher intensity NiO peak in the XRD since the Ni is growing with the oxygen due to the ZnO core in the Ni ‘nanocup’. The thickness of the ZnO–Ni film is 80 nm. It is clearly noted that the growth of a thicker ZnO–Au seeding layer leads to well defined Au nanoparticles which aid in the nucleation of the Ni into the unique ‘nanocup’ structures.
Additionally, it is important to note that very little to no inter-diffusion was seen at the ZnO–Ni interface for the four samples discussed. Paldi et al. in their study of ZnO-based nanocomposites observed similar pillar in matrix morphology with minimal interdiffusion at the ZnO–metal interfaces.25 The LVZN sample shows slight inter-diffusion of the ZnO core into the Ni shell but that could be caused by the higher-pressure environment. Unlike copper (Cu) which inter-diffuses easily with oxides and can diffusion along phase boundaries.25,26 Ni does not diffuse as easily as that of Cu. One possible reason is related to relatively high melting temperature of Ni, Tm = 1455 °C, and relative inertness with ZnO and other oxides. Successfully Ni-based VAN growth has been reported in CeO2–Ni VAN27 and Ni–BZY VAN.28 Regardless, interdiffusion can cause issues in thin film performance and integrity and has been shown to be eliminated if there are secondary phases at the grain boundaries.29
Vacuum pressure plays an important role in the deposition of the film because it not only affects what adatom species arrive at the substrate surface but also during the laser–target interaction. The pressure during deposition determines the plume dynamics and therefore the kinetic energy and mean free path of the species that are ablated from the target.30 Plus, the level of vacuum will also affect the uniformity and purity of the films.30 The deposition rate also corresponds to the vacuum pressure.31 So, when the film is grown in high vacuum (low pressure), such as the case of HVZN, the plume is highly directional. Conversely, the plume in a high pressure (low vacuum) environment grows spherically due to more scattering with gas molecules.32 Therefore, it is reasonable to believe that the more directional plume for the HVZN sample resulted in the nucleation of ZnO matrix and Ni nanostructure as opposed to what was seen in LVZN.
Additionally, the measured reflection intensity spectra as a function of wavelength for various reflection angles is included in Fig. S3.† For all samples, the 30° angle has the lowest reflection intensity and it increases as the angle increases. From the reflection intensity, the relation to SPR phenomena can be understood since the p-polarized light is what excites the plasmons. This is why reflection intensity measured by ellipsometry is needed. Overall, since SPR results in a dip in reflection intensity since light energy is transported to the electrons in the thin film,33 a dip in the reflection intensity is expected. Therefore, the only sample that shows this dip is seen in Fig. S3c† for the 50ZA80ZN sample at ∼620 nm. This is due to the well-defined Au nanoparticles that are not present in any of the other films. Even the thin seeding layer film (10ZA50ZN) does not demonstrate this behavior because the ZnO–Au seeding layer is too thin for the p-polarized light to excite the plasmons in the sample. Ni does not possess as strong a plasmonic response as Au, that is why HVZN and LVZN do not show the large dip in reflection intensity.
The imaginary permittivity of all samples is shown in Fig. 4b and it describes the absorption losses. The 3 samples show relatively low losses and the out of plane imaginary permittivity is the highest for the thick seeding layer sample. This can be explained by SPR phenomena seen in Fig. 4c.
The real permittivity of LVZN seen in Fig. S4a† shows optical anisotropy and the film behaves overall as a dielectric, (i.e., real permittivity is positive in the measured wavelength range). It is observed that the film is anisotropic throughout all wavelengths, and this can be attributed to the highly anisotropic ZnO-core and Ni-shell structure. Though anisotropic, the film behaves like a dielectric in both the in-plane and out of plane directions because of the ZnO matrix and the coupling of the ZnO and NiO in the core–shell structure as explained by the XRD above. The imaginary part of the permittivity seen in Fig. S4b† is also low, thus having low losses for this sample as well as the other three. Plus, the out of plane imaginary permittivity becomes negative at 1429 nm which for metamaterials has been shown to be an inherent property resulting from the magnetoelectric coupling effect.34
The UV-Vis transmittance results are shown in Fig. 4c for all three samples. The property of interest was the surface plasmon resonance (SPR) which could show as a dip in the data where the frequency of light is absorbed if it matches that of the frequency of the metal electrons in the film. The SPR wavelength for HVZN is 506 nm. The SPR wavelength for 10ZA50ZN is 505 nm and it jumps to 545 nm with 50ZA80ZN. This large redshift could be attributed to the Au presence in the film. A previous study which used the same concentration ZnO–Au film reported the same SPR wavelength as seen in 50ZA80ZN.9 Additionally, since SPR is related to the absorption of light, the larger out of plane imaginary permittivity of 50ZA80ZN seen in Fig. 4b can be explained by this phenomenon and of course the out of plane anisotropy resulting from the metals. As mentioned above, the imaginary permittivity is related to the SPR and can lead to a broadening of the SPR peak35 which is seen for 50ZA80ZN and explains the out of plane imaginary permittivity for this sample.
The well-defined Au particles that are present in the thicker seeding layer film can be attributed to this prominent jump in SPR. Plus, there is a broadening of the absorption edge that can also be attributed to the Au particles since they are very small in the thinner seeding layer but become elongated particles with more defined aspect ratios which has been shown to impact the SPR.36 The % T decreases when the seeding layer grows thicker meaning less light passes through the film due to the greater metal content from both the Ni and Au. Overall, it is apparent that the thicker ZnO–Au seeding layer drastically changes the permittivity and SPR properties of these ZnO–Ni thin films. The thicker seeding layer demonstrates hyperbolic behavior which is representative of metamaterials which have the ability to be used in many different applications such as photovoltaic devices and hyperlenses.37 Additionally, the SPR change demonstrates tunability simply based on the thickness of the ZnO–Au seeding layer. This tuning can be utilized in applications such as sensors. Au is an important material used for SPR. Future research could grow the seeding layer with different concentrations of Au or different thicknesses. Fig. S4c† shows the transmittance plot of Sample 4 with an absorption valley at 505 nm, just as that seen in Sample 2.
The bandgap was calculated using the transmittance data and applying the Tauc plot method which is seen in Fig. S5† for all samples. Both HVZN and 10ZA50ZN have comparable bandgaps of 3.49 eV and 3.46 eV (Fig. S5a and S5b†). 50ZA80ZN seen in Fig. S5c† has a bandgap of 3.58 eV while the bandgap of LVZN increases to 4.41 eV (Fig. S5d†). Clearly, varying nanostructures in these films leads to this change in bandgap. Since LVZN demonstrates the core–shell pillar structure, it differs from the other films because the interface between the ZnO core and NiO shell can impact the strain and therefore the electronic structure. This is a possible reason for the larger bandgap observed in LVZN than the other films.
Fig. S6a and S6b† demonstrate magnetic properties of the low vacuum film (LVZN). The film is magnetically anisotropic regardless of temperature due to the different loop shapes of the OP and IP measurements. Plus, the saturation magnetization for the OP is clearly greater than that of the IP with the coercivity being about the same regardless of temperature. This large anisotropy for both temperatures likely arises from the gaps in between the ZnO core and Ni shell structure as seen by the black areas in Fig. S2e.† Additionally, there is a clear shape anisotropy resulting from the OP growth of the Ni shell while the ZnO grows more IP. The aspect ratio of the Ni growth is contributing greatly to the OP anisotropy.
The magnetic properties of Ni is relatively weak as compared to Fe, because of its nearly full 3d orbital.39 Furthermore, magnetic domains are the cause of the ferromagnetic property observed in Ni.40 The thin films deposited under different vacuum pressure (i.e., HVZN and LVZN), however, are grown in nanometer length scales making their atomic magnetic moments and magnetic anisotropy behave differently than just bulk Ni. Additionally, exchange interactions which are the coupling of atomic or molecular magnetic moments between the different materials in thin films are also necessary to observe ferromagnetic behavior.40 With that being said, the difference in hysteresis between HVZN and LVZN is a direct result of the shape anisotropy in the individual films. Consequently, this impacts the coercivity. As the name suggests, shape anisotropy is dependent on the shape of the magnetic nanostructures and the axis on which they magnetize easily. Therefore, it is reasonable that the LVZN sample, which has a high aspect ratio core–shell structure with the Ni and NiO growing out of plane, to have large magnetic anisotropy between the in-plane and out-of-plane directions. HVZN on the other hand does not have a high aspect ratio Ni or NiO structure and therefore the magnetic anisotropy is less obvious. This is a reason why the coercivity is less for LVZN in the out-of-plane direction since the hard axis magnetizes easily due to dipole–dipole interactions. HVZN on the other hand, has larger coercivity both in-plane and out-of-plane because of the greater presence of Ni(111) than that in LVZN.
Although the ZnO–Au seeding layer is not composed of materials known to have prominent magnetic properties, the growth of the seeding layer before the ZnO–Ni film could add to the anisotropy and change the magnetic properties of the films by affecting the ZnO–Ni microstructure. Fig. 5c and d show the magnetic properties data for the film with the thinner ZnO–Au seeding layer (10ZA50ZN) at both 10 K and 300 K respectively. The film is magnetically anisotropic and has low coercivity and higher saturation magnetization for both temperatures. Fig. 5e and f show the magnetic data for the film with the thicker ZnO–Au seeding layer (50ZA80ZN) at 10 K and 300 K respectively. This film has low coercivity and lower saturation magnetization when compared to the thinner seeding layer film.
In order to understand why 10ZA50ZN and 50ZA80ZN show these properties, it is important to understand the microstructure. Bulk ZnO and Au are reported to be diamagnetic41,42 while bulk Ni is ferromagnetic, and bulk NiO is antiferromagnetic.43 However, the Au, Ni, and NiO present in these films grow with unique nanostructures which can impact the magnetic properties. The thinner seeding layer film has shown to grow more Ni (111) than NiO (111) through XRD measurements aiding to the improved saturation magnetization. 50ZA80ZN shows the lowest saturation magnetization due to the Au and NiO in the sample which would lower the amount of magnetization that can be achieved in the sample. Saturation magnetization can be influenced by microstructure, atomic structure, and magnetic interactions44 which explains why each sample behaves the way it does magnetically.
Furthermore, the composition of the films, as discussed previously and seen in Table S1† plays a role in the magnetic properties. As the Ni composition changes, so does the magnetic moments of the films. This is due to the plume kinetics in different vacuum environments.
It is clear that all 3 film samples are soft magnetic thin films due to their extremely thin hysteresis loops characterized by high saturation magnetization and low coercivity.45 This could be a result of the contribution of more Au and NiO.
Interestingly, the microstructure seen in HVZN more closely resembled the Ni growth in the combinatorial growth study of LSMO-NiO grown on STO substrates.46 One of the main differences is the use of two separate targets, a NiO target and an LSMO target. In this study, a composite target of ZnO–Ni was used which deposits Ni and ZnO simultaneously on the substrate. A combinatorial growth process in an oxygen background with annealing was used with the LSMO–NiO which allowed for the cubelike growth of the Ni in the LSMO matrix. Plus, this combinatorial approach would be an interesting method to recreate with the ZnO–Ni to see if the same Ni microstructure results.
Overall, this work gives a better understanding of the ZnO–Ni system by investigating it from different growth background pressures and seeding layer. With the change in vacuum condition and the growth of a seeding layer to aid in the growth of the ZnO–Ni film, there were new insights into the growth and properties of the films. First, it was shown that a ZnO–Au seeding layer can be utilized to achieve epitaxial growth of ZnO–Ni. Not only did this improve film growth but incorporated plasmonic properties which aid in making the nanocomposites metamaterials. Second, it was seen that the vacuum level impacts the microstructure drastically. The higher vacuum ZnO–Ni film grew more cube-like and not epitaxially while the lower vacuum ZnO–Ni grew as core–shell and epitaxially. Plus, the core–shell growth was achieved through a single step growth and a composite target. Lastly, there was tuning of the optical properties because of the thickness variation of the seeding layer. Clear hyperbolic behavior resulted as the seeding layer thickness increased because it improves the ZnO–Ni microstructure. Future directions for this work include the need to understand why lower vacuum results in epitaxial growth of ZnO while higher vacuum does not. Perhaps, trying a wider range of vacuum pressure on ZnO–Ni growth could provide a better understanding of how the microstructure could vary and in turn, the growth quality. Clearly, the ZnO, Ni, and Au provide a unique mix of properties such as plasmonic and ferromagnetic, which make these metamaterial thin films good candidates for optoelectronic devices, sensors, and data storage applications.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d5na00207a |
This journal is © The Royal Society of Chemistry 2025 |