Advances in growth, doping, and devices and applications of zinc oxide

Zinc oxide (ZnO) has been a multifunctional material of research interest due to tunability in its characteristics and direct applications in various areas such as energy harvesting, electronics, optoelectronics, and biomedicine.^1–10 ZnO has a direct and wide bandgap and a high exciton binding energy. It is stable at high temperatures and power and is earth-abundant and environment-friendly in nature. This has raised interest in ZnO for photovoltaics, thermoelectrics, spintronics, piezoelectrics, electronics, sensing, optoelectronics, and biomedicine.^11–18 A high carrier mobility and breakdown voltage as compared to commonly used semiconductors such as silicon and gallium arsenide, makes it a potential material for future transistors and electronics. The magnetic properties of ZnO especially ferromagnetism at room temperature make it a candidate for spintronic, and quantum computing applications.^19–21 Stiffness, hardness, piezoelectric, and thermoelectric properties of ZnO have been applied in the cement and ceramics industry.²² Most of these interesting properties arise from a nonsymmetric structure of the ZnO unit cell. In addition to its role as a primary material of applications, ZnO with a wurtzite crystal structure could also be used as a substrate with other materials such as gallium nitride, with a matched lattice structure.^23–25 ZnO is abundant and nontoxic, which widens its applications in biomedicine. It is immune to high energy incidence and could be used for space applications and neutron detection.^17,26 ZnO could be grown as thin films, bulk, as well as nanostructures, and each structure could have its unique morphology.^4,27,28 The resulting properties of ZnO materials also depend on the growth methodologies.

However, the characteristics and defects structures of ZnO are neither well-understood nor comprehensively characterized. This has been hindering several interesting applications of ZnO-based materials. A better understanding of ZnO growth, defects, and properties is crucial for its utility in next generation applications. For example, bandgap tunability is a conducive property in photovoltaics, sensing, spintronics, photodetectors, and biomedicine. ZnO has shown bandgap tunability but the effects of doping, growth conditions, crystal structure, etc., have not been consistent on the bandgap. ZnO shows signs of spintronic, thermoelectric, nuclear, and biomedical properties. However, specific directionality in each of these areas is required for immediate advancements. In this article, impactful characteristics of ZnO-based materials are collated, these help in planning effective future research and development in ZnO and similar compound semiconductors.

This paper reviews advancements in growth techniques and emerging devices and applications of ZnO. Common and effective synthesis methods to grow high-quality ZnO are discussed. Native defects, doping, and bandgap engineering are discussed. ZnO devices and their current and future applications in thermoelectrics, photovoltaics, light-emitting diodes (LEDs), spintronics, neutron detection, and biomedicine are highlighted.

Optimum growth conditions and constituent elements' sources are required to grow ZnO with desired structure and properties depending on the applications. Growth techniques such as metal-organic chemical vapor deposition (MOCVD), molecular beam epitaxy (MBE), atomic layer deposition (ALD), pulsed laser deposition (PLD), hydrothermal method, sol–gel technique, and magnetron sputtering have been used to grow ZnO on substrates such as sapphire, silicon carbide, silicon, gallium arsenide, LiTaO₃, and glass.^{1–4,6,17,29–32} Sapphire is the most used substrate due to its lattice match, crystallinity, optical transparency, low cost, and stability at high temperatures and pressure. Growth conditions can influence the properties of ZnO grown and its applicability for applications.

MOCVD is an established technology to grow ZnO at high/low temperatures. It results in a well-controlled and uniform growth with a good crystal quality and growth rate over a large area. MOCVD could be employed to grow ZnO in the laboratory as well as on the commercial scale.^33–38 The process involves a controlled vaporization of a metal-organic zinc source, followed by a vapor-phase chemical reaction with the oxygen source under optimum growth conditions (temperature, pressure, substrate rotation). Figure 1(a) shows a flow flange of an MOCVD system chamber with points where zinc, oxygen, and carrier gas enter the chamber right above the susceptor that is shown in Fig. 1(b).

In MOCVD, optimum heat exchange is maintained between the flow flange and the susceptor to satisfy boundary layer conditions and effectively result in a chemical reaction and ZnO deposition. Growth temperature in a range of 300–600 °C, pressure from 30 to 50 Torr, O to Zn ratio of 100–2000, and susceptor disk rotation from 600 to 1200 rpm resulted in the growth of ZnO on sapphire substrates, as shown in Fig. 1(c).³³ Deposited films showed uniformity in thickness and optical properties. Figure 1(d) shows the photoluminescence properties of grown thin films, and the response is uniform through different space points over the sample.

Figure 2 shows the structural and optical properties of ZnO grown by MOCVD.³⁹ A (002) peak with a rocking curve full width half maximum of ∼200 arcsec and optical absorption band edge of 373 nm were obtained. ZnO shows Raman scattering at A₁ and E₁ longitudinal and transverse optical modes, E₂ low and high, and B phonon modes. Strong peaks from E₂ vibrations were observed (not shown here), indicating a good crystal quality.³⁹ 

Structures of Zn–O bonding were studied by reflective second harmonic generation (RSHG), Fig. 3.⁴⁰ The average polar strength of the Zn–O bond represents the quality of the ZnO (002) thin film. The mirror symmetry is caused by nonvanished polarity of twin boundaries due to the mismatch between the ZnO film and sapphire substrate. It is analyzed using s-polarized RSHG with s-polarized fundamental light irradiation. These results demonstrate that the Zn–O hetero-polar bonds on the smooth ZnO surface contribute to the SHG intensity.

MBE also results in a uniform growth of ZnO with high crystallinity; however, the growth rate could be lower than other techniques.^42,43 Magnetron sputtering and PLD are among other commonly used growth methods for ZnO.^36,44 In magnetron sputtering, an electric field is applied between the target and ions' source. Ions striking the target release atoms, which are deposited on a substrate. Magnetron sputtering usually results in defects, which makes it difficult to effectively control material properties. In the case of PLD, a laser beam is incident on the target instead of ions. Pure or doped/alloyed ZnO could be used as targets in this growth method.

Radio frequency (RF) magnetron sputtering has also been used to grow m-plane ZnO films grown on cleaned m-plane sapphire and p-type (100) Si substrates.⁴⁵ After deposition, the films were treated separately with air-annealing at 150, 175, and 200 °C for 1h, with five samples named Sa0, Sa1, Sa2, Si1, and Si2 with different thickness (190–380 nm) and annealing temperature. X-ray diffraction (XRD) θ−2θ scan patterns showed the diffraction peak of ZnO (10–10) at ∼31.8° and ZnO (11–20) at ∼56.6°, PL spectra at room temperature (RT) in Fig. 5(b) showed near band edge (NBE) emission at 3.27 eV, consisting of FX and the first order longitudinal optical (1-LO) phonon replica.

Spectroscopic ellipsometry (SE) measures the change in the polarization state psi (ψ) and delta (Δ) between the incidence and reflection of light on the samples. Thickness, surface roughness, optical constants, and bandgap of samples can be precisely extracted by utilizing theoretical models to fit the measured ellipsometric spectra (tan ψ and cos Δ). From the RT-SE measurement data and Tauc model simulation, the index of refraction n and (αhν)² versus (hν) of m-plane ZnO films was obtained, as shown in Fig. 6, also indicating the energy bandgaps for m-plane ZnO films.

High purity bulk ZnO can be grown using a pressurized melt-growth technique. This process involves melting of precursors followed by crystallization as the temperature is lowered.^20,46–48 ZnO powder is heated up to ∼1950 °C in a crucible, and the crucible is steadily moved away from the mix resulting in bulk crystallization. Figure 7 shows a schematic of a melt-growth setup, and x-ray diffraction results of ZnO grown by this method.⁴⁹ The composition of Mn and Co-doped ZnO was set by the stoichiometric amounts of source powders before starting the growth. Temperature changes and the pulling rate of bulk crystal control the resulting properties, defects, and grain boundaries of ZnO.

Sol–gel is a popular technique to grow ZnO and involves mixing constituent elements' precursors to form a clear solution. The solution is then spin-coated on a substrate or the substrate is dipped in the solution, at temperatures above ∼100 °C to deposit thin films.^50–52 ZnO could be grown as nanostructures to utilize the combined properties of ZnO and nanorelated characteristics. Figures 8(a)–8(f) show images of nanofibers grown using a sol–gel technique based electrospinning method.²⁷ The precursors' sol–gel solution was placed in a plastic syringe and spawned on an aluminum substrate. The samples were annealed at a temperature range of 80–700 °C to deposit ZnO nanofibers. The lengths, diameters, and compositions of the nanostructures could be controlled by postgrowth processing. It was observed that the ZnO nanofiber diameter decreased with high temperature annealing. As seen in Fig. 8(g), a crystalline structure was observed by annealing the grown nanofibers.

Hydrothermal and pulsed laser deposition are popular methods to grow ZnO.^53,54 Surface roughness, crystal orientations, and nonstoichiometry could be tuned with changes in the growth conditions. ZnO could also be grown using ALD that consists of sequential and self-limiting self-reactions at low temperatures (∼200 °C). ALD results in a conformal and uniform deposition of ZnO films, which are typically a few nanometers thick.^55,56 The growth rate is much lower (a few nm per hour) than other techniques; the reactor geometry, precursor injection schemes, and purge times could be potentially optimized to achieve higher growth rates.^55–59 

ZnO in its stable form has an asymmetric structure. Different site occupancies by dopants or cations in ZnO system can also result in different extents of nonstoichiometry.³⁰ Depending on the growth mechanisms, sites which are oxygen deficient, or zinc deficient, or which do not have Zn and O bonded, are likely. These cause defects and scattering in ZnO.^60–64 

Figure 9 shows the mobility and carrier density corresponding to various scattering mechanisms in MOCVD-grown ZnO on sapphire.⁶⁵ The trend could be affected by vacancy and interstitial defects in ZnO that act as compensation points for free carriers. Acoustic-phonon, polar optical-phonon, piezoelectric, and ionized impurity scattering mechanisms, or electron-plasmon interactions, could have weighted effects on the electrical properties of ZnO.⁶⁵ 

Defects in ZnO primarily arise due to vacancies, interstitials, and activation of unintended impurity sites. Zinc interstitials, oxygen vacancies, and zinc antisites are common donor defects, while oxygen interstitials, zinc vacancies, and oxygen antisites are acceptor defects.^66–68 Acceptors that are intrinsic in ZnO can have high formation energy while intrinsic donors can either have a high formation energy or form deep energy states. Oxygen vacancies typically have a lower formation energy compared to intrinsic acceptors and donors.⁶⁴ Acceptor-type defects have high energy of formation and result in instability atRT, while donor defects are more common and exist at RT in thermodynamically stable states. Common donor-type defects could be conducive to a high n-type carrier density of ZnO. However, these make p-doping of ZnO challenging.

Oxygen vacancy (V_o) draws the most research attention among all native defects because it is widely considered as the primary source of unintentional n-type conductivity and the nonstoichiometry of ZnO.^15,69–71 V_o has the lowest formation energy (from −0.9 to 3.9 eV) among donorlike defects. Zinc interstitial is another common defect. Zinc interstitial atom (Zn_i) could occupy a tetrahedral or octahedral site in the ZnO wurtzite lattice and is more stable with a lesser geometric strain at octahedral sites.^2,15,68 Zn_i is a shallow donor-type impurity and has a high formation energy in n-type ZnO as compared to p-type ZnO. Zn_i could be one of the compensating centers in p-type ZnO and could contribute to the intrinsic n-type nature. In a study of high-energy electron radiation in ZnO, it was observed that the Zn-sublattice interstitial defect is produced more on the Zn-face than the O-face.⁶¹ Another defect zinc vacancies (V_Zn) results in oxygen dangling bonds that are electron-deficient and form deep acceptorlike states. V_Zn are formed easily in n-type ZnO than p-type ZnO and are unlikely to induce an effective p-type conductivity.^60,61,68 Complexes formed by dopant group-V elements (such as As) substituting Zn sites and surrounded by zinc vacancies could possibly result in a net p-type conductivity. Accurate estimation and interrelation of defect levels with ZnO properties such as the bandgap could be complicated and is an area of progress.⁷⁰ 

The n-type nature of a defect complex V_o–Zn_i was experimentally investigated in MBE-grown ZnO.⁶² Oxygen vacancies were compensated by annealing in an oxygen environment. Incoming oxygen fill in the oxygen vacancies defects and deactivate Zn_i position, resulting in a reduced n-type carrier concentration. On the other hand, annealing in the Zn environment increased the carrier concentration as the Zn atoms could undergo hybridization with the inactive V_o to form n-type V_o–Zn_i defect. The results are tabulated in Table I and confirm that V_o–Zn_i acts as native donors in ZnO. The V_o–Zn_i defect complex also Affected properties such as mobility and resistivity. Most stable native defects in ZnO seem n-type, but it is important to achieve high-quality p-doping for devices and applications such as LEDs, solar cells, and transistors, where p–n junctions are preferred over metal-semiconductor junctions.

As discussed in Sec II B, ZnO is inherently n-type, which makes it challenging to produce low resistive p-type ZnO with a high carrier density and mobility.^60,61,72,73 Although ZnO-based materials exhibit desirable properties and hold promises in various applications, p-type doping is still the bottleneck for device applications. Difficulties in p-type doping arise from low solubility of acceptor components, deep acceptor energy levels (high ionization energy), and strong self-compensating effects. An acceptor level can be treated as the energy difference between impurity energy level and valence band maximum, which is a measure of the ionization energy of holes from acceptorlike impurities. Except for deep acceptor levels, doped acceptors may also form donorlike defect centers and compensate holes.

Several elements from group-V (N, P, As, Sb), group-I (Li, Na, K), and group-IB (Ag, Cu) have been used to p-dope ZnO.^72–79 As per local density approximation, these dopants could form shallow acceptor states and result in a p-type behavior. Spin-polarized Kohn–Sham and hybrid functional theories showed that these corrected acceptor energy states could be deeper than predicted. Even if the theoretical analysis shows that compensating the inherent n-type behavior of ZnO could be difficult, there have been successful growths of p-type ZnO. However, the samples are usually not sufficiently stable or conductive.

Group-I elements such as K, Li, and Na have a small size and could diffuse in ZnO.^36,63 These dopants tend to occupy interstitial sites. Na, Ag, Li, As, and Sb could be doped to compensate the n-type nature and result in p-type ZnO. K-doped ZnO by a solution method had a hole density of 10¹⁴ cm⁻³ while Li-doped ZnO by sol–gel method showed 10¹⁹ cm⁻³ hole density. Na-doped ZnO showed a mobility range of 0–10⁹ cm²/V s. Li-doped ZnO by sol–gel method exhibited a resistivity of 2.6 Ω cm while Na-doped ZnO by sol–gel method had a resistivity of 1532 Ω cm. These dopants are discussed in detail elsewhere.^34,36,63

Annealing helps to move the dopants to substitutional sites but reduces the mobility and carrier concentration and increases the resistivity. Group-II and III elements such as Al, Mg, and Cd usually occupy substitutional sites in ZnO and result in a high carrier concentration, mobility, and conductivity.^{36,63,80–82} In the ZnO lattice, group-V elements can substitute oxygen sites and group-I elements can substitute zinc sites. Both scenarios could behave like acceptors. However, they either have high ionization energy or form compensation centers. When doping large group-V atoms into ZnO, the dopants could substitute Zn sites rather than oxygen sites. For instance, in nitrogen doping, a shallow acceptor N_o–V_Zn complex could be formed rather than N_o (N substituting O).⁸³ The formation of shallow acceptor complexes seems sensitive to growth conditions and could be challenging to control.

p-type, n-type, and semi-insulating nitrogen-doped ZnO were investigated using MOCVD growth technique.³⁴ As-grown nitrogen-doped ZnO samples by MOCVD showed p-type behavior at 0.2% and 1% NH₃ (nitrogen source) flow. ZnO:N with 4% NH₃ flow that was initially n-type showed a p-type behavior upon annealing in oxygen. All the p-type ZnO samples showed a high resistivity. A summary of the results is shown in Table II. Rapid thermal annealing at 800 °C for 15 min (not shown in Table II) made all the samples n-type. Nitrogen doping could result in defects that are n-type with varying formation energies and make it difficult to control the p-doping levels in ZnO. In a study of p-type ZnO grown by molecular beam epitaxy, N-doped ZnO showed hole concentrations (carrier density in the order of 9 × 10¹⁶ cm⁻³, mobility of 2 cm²/V s, and resistivity of 40 Ω cm).⁶⁰ 

Codoping, in which two or more elements are incorporated in ZnO, is also considered an effective method to enhance the p-type conductivity and structural stability.^84–94 These combinations include acceptor-donor codoping or dual-acceptor codoping. The purposes and advantages of codoping are (1) increasing concentration of acceptor dopants, (2) lowering ionization energy of acceptors via complex structures (3) Suppressing self-compensating centers and (4) stabilizing acceptors in substitutional sites.^55,95–98 Madelung energy could be used as a criterion of stability of the dopants. p-type doping with elements such as C, N, Li, or As increases the Madelung energy and could cause instabilities in the charge distributions, while n-doping with Al, Ga, In, B, or F decreases the Madelung energy.^95,98 When a donor-acceptor pair such as C–Ga or C–Al or N–Ga is codoped, the n-type dopant could maintain the Madelung energy and help toward maintaining the p-type nature and acceptor energy states that are introduced by the p-dopant. In the case of n-doping, formation of Zn–O bond is energetically favorable as compared to Zn–N bonds. Adding Ga or Al species which have more affinity than Zn toward N increase the solubility and incorporation of N in ZnO as a p-dopant.⁹⁷ P-type behavior has been experimentally achieved in ZnO by the codoping approach.^36,63 The N codoped ZnO by MOCVD had a hole density (h) up to 10¹⁸ cm⁻³, mobility (μ) of 1–10 cm²/V s, and a resistance (ρ) of 1–10 Ω cm. ZnO codoped with In–N by MOCVD had a h of 10¹⁸ cm⁻³, μ of 0.1–0.5 cm²/V s, and a ρ of 15–16 Ω cm. Various codoping combinations of Ag, Li, Mg, Be, B, Se, Te, S, Al, Ga, P, and S have been made with N or P to achieve a p-type behavior in ZnO. Techniques such as cluster-doping which includes engineering a locally stable chemical environment that includes dopant species can also lead to more effective p-type doping.

As mentioned in the Introduction, ZnO could be a substrate for materials with a lattice match.^23–25 Efforts are made for the growth of GaN/ZnO, InGaN/GaN/ZnO, and so on.⁹⁹ For example, InGaN thin films were grown on the ZnO substrate with a GaN buffer layer using MOCVD. High resolution XRD (HR-XRD) in Fig. 10 confirmed the single crystalline lattice structure for the nanometer scale InGaN/GaN/ZnO.

Rutherford backscattering spectrometry (RBS) experiment and simulation were used to determine the layer thickness and composition for the MOCVD-grown InGaN/GaN/ZnO samples. Figure 11(a) shows the results of random RBS spectra recorded on three InGaN/GaN samples prepared on the ZnO substrate (S1, S3, and S4) with growth temperatures at 656, 700, and 720 °C, respectively. Figure 11(b) shows the random RBS spectra for the other two samples (S2 and S5) of InGaN with growth temperatures at 680 and 736 °C, respectively. The thickness and composition of five samples grown from 656 to 736 °C were acquired from the simulation on the RBS experimental spectra. The thicknesses of the GaN buffers are around 28–57 nm, while the thicknesses of InGaN layers were in the range of 52–91 nm. The contents of In at InGaN layers are 0.65, 0.52, 0.49, 0.37, and 0.21, for samples with growth temperatures of 656, 680, 700, 720, and 736 °C, respectively. The RBS simulation results are consistent with the HR-XRD data in Fig. 10. Both show that the growth temperature has a significant impact on the incorporation of indium and that the In content is higher when the growth temperature is higher. This correlation is associated with the higher incorporation of indium from the increase of temperature, i.e., more indium is incorporated at higher temperatures.⁹⁹ 

Besides stable n-type and p-type doping, applications of ZnO also depend on the feasibility of bandgap engineering.^56,100–108 Most device designs rely on heterostructures to confine carrier or optical properties, for example, LEDs, laser diodes, solar cells, and high electron mobility transistors (HEMTs). These heterostructures are formed by layers of different materials or compositions; the bandgap of each layer and the valence–conduction band offsets between the layers are important. Bandgap engineering of ZnO enables applications across various bandgap requirements. Interfacial defects and mismatches could be minimized if multiple layers of ZnO with varying bandgaps are stacked for applications such as solar cells, transistors, etc.

A popular approach to achieve bandgap engineering in ZnO is to alloy it with MgO or CdO.^{56,100–106,109} However, MgO and CdO have rock-salt structures at ground states and cause phase separations in ZnO with high concentrations of MgO and CdO. Despite phase separations, these alloys can still provide a relatively wide range of bandgap from 2.3 to 4.0 eV. MgZnO films have been grown on a variety of substrates, including c-sapphire, GaN/c-sapphire, and ZnO using various techniques, such as PLD, MOCVD, and MBE. MgZnO remains wurtzite for Mg concentration up to 33% with a bandgap up to 4 eV. X-ray diffraction measurements indicated that with an increase of Mg concentration, lattice parameter a increases and c decreases.^102,110 An increase in exciton binding energy has also been reported for MgZnO alloys.⁶⁹ 

While adding Mg leads to an increase in bandgap, adding Cd leads to a decrease in the bandgap. Cadmium-doped ZnO grown on quartz substrates by PLD showed a redshift in the band edge with the Cd content at 10% Cd as shown in Fig. 12.¹⁰⁰ With an increase in the temperature from 25 to 600 °C, a bandgap reduction was observed due to increased lattice expansion and atomic vibrations. A redshift of 0.3 eV was observed in MOCVD-grown CdZnO on c-sapphire.¹⁰⁶ However, the samples exhibited large inhomogeneities and regions with varying lateral Cd concentrations. Since both Mg and Cd alloying change lattice parameters and induce strain in ZnO-based materials, it is essential to understand how strain affects band structures of ZnMgO and ZnCdO. This helps to design devices that rely on bandgap engineering.

First principal calculations were used to study the absolute valence and conduction band offsets at MgO/ZnO and CdO/ZnO interfaces.^111–113 MgZnO and CdZnO remain wurtzite with moderate Mg or Cd concentration. Instead of calculating offsets for specific alloy compositions, those are determined by interpolating the parent compound and considering uniaxial or biaxial strains, hydrostatic strains, and related bonding/antibonding and kinetic energy states introduced by the alloying material. Absolute deformation potential is smaller at ZnO/CdO as compared to the ZnO/MgO interface. Conduction band offsets are larger than valence band offsets. The valence band offset is significantly larger in ZnO/MgO as compared to the ZnO/CdO interface. Zn and Cd both contain d-state electrons, while Mg does not which results in the larger valence band shift in ZnO/MgO.^111–114 Even with high CdO concentrations, CdO/ZnO alloys typically have weak confinement of holes. This agrees with experimental results that the bandgap shift is relatively small in these compounds. Optical transmission (OT) spectra showed that the optical bandgap energy increased with x(Mg), Fig. 13.¹¹⁵ The UV-PL spectra under the 325 nm excitation exhibited that PL peaks shifted toward higher energy side.

The bandgap of ZnO can also be tuned by doping with transition-metal elements.^{19,20,35,100,116–118} MOCVD-grown ZnNiO thin films on sapphire substrates showed a reduction in the band edge with nickel content under varying growth conditions [set A (400 °C/100 Torr) and set B (450 °C/30 Torr)], as shown in Fig. 14.³⁵ Under varying growth scenarios, the bandgap variation trend was similar, but the bandgap reduction rate changed with pressure and temperature. Sites occupied by the dopant and possible defects activated under different growth conditions could result in varying bandgap reduction rates.

ZnO alloyed with Cr, another type of transition-metal element, can form ZnCr₂O₄ (ZCO).¹¹⁹ A series of ZnCr₂O₄ films were grown epitaxially on c-sapphire by PLD with varied oxygen pressure. As the deposition oxygen pressure was increased, the atomic ratio of Zn/Cr in the ZCO films increased linearly from 0.316 to 0.585. Also, the bandgap of the ZCO films increased from 3.61 to 3.87 eV with the oxygen pressure, Fig. 15. This widening of bandgap was attributed to the enhancement of cations-anions interaction and reduction of oxygen vacancies with increasing deposition oxygen pressure.¹¹⁹ 

Optical transmission spectra (TS) data and SE data denote bandgap values obtained from the TS and SE measurements, respectively. A linear relationship of the bandgap versus the deposition oxygen pressure for ZCO thin films was obtained from fits to the TS and SE measurement data.

Bandgap variations were also observed in ZnO bulk materials. Bulk ZnO doped with Mn, Fe, and Co and grown using a pressured melt-growth process exhibited a reduction in the bandgap as shown in Fig. 16.^19,20

Co-doped ZnO has additional intermediate peaks around 600 nm possibly due to d–d transitions of Co. The transmission spectra agreed with the bulk materials' colors upon visual inspection. As per x-ray diffraction study, doping resulted in an increase in the lattice size and a reduction in the crystal quality. The cooling process in the melt-growth technique could result in a nonuniform distribution of dopants in the material. Another set of dopants Ga, Er, Fe, Co, Mn, Mg–Li, and Ho were doped in ZnO by a modified melt-growth method. Bandgap reduction was seen in Co-, Fe-, and Mn-doped bulk ZnO. Transmission % was lesser in Co-, Ga-, and Mg–Li-doped ZnO, as compared to the other dopants, likely due to differences in defects or impurity or spin states introduced by doping. ZnO exhibits a controllable tunability in the bandgap with doping and growth conditions while maintaining the material quality.

Zinc oxide has great potential for next generation applications and devices considering its unique properties as reviewed in Secs. II and III. Semiconductor bandgap can determine the current-voltage characteristics not only by external voltage bias but also optical illumination, heat, and mechanical energy. However, only a fraction of the optical and heat energy is efficiently absorbed to generate electricity. A material that could convert dissipated or naturally available heat energy into electricity is of interest. Multijunction solar cells with varying bandgaps of a consistent base material such as ZnO could be integrated in a tandem device to potentially result in an enhanced power conversion efficiency.^2,5,120,121 Considering the electronics aspect, there has been a technical and economic growth as also predicted by Moore's law; electronic components' sizes have been reducing with an increase in their speed, circuit densities, and device performance.^1,122–125 However, lithography technology has been reaching an atomic scale and assembling more and more silicon devices or circuits in a dense area has been causing undesirable heat dissipation, and an overall poor performance. ZnO is a candidate material to address these challenges as it exhibits a combination of interesting electrical, thermoelectric, photovoltaic, magnetic, and emission properties. This section discusses the applicability of ZnO and its properties in various areas of energy harvesting, electronics, and sensing such as thermoelectrics, photovoltaics, light /laser emitters, spintronics, neuromorphic computing, neutron detection, applications in civil infrastructure, and biomedicine.

Spintronics is among the technologies that could pave the way toward faster scalable electronics and future computing needs related to quantum computing, big data, artificial intelligence, and neuromorphic computing.^{1,122–124,126–130} Understanding and control over quantum properties, which are closely related with the spin and magnetic properties of a material is essential for such applications. ZnO could be doped with transition metals or rare earth elements in dilute amounts to add magnetic properties.¹³¹ Ferromagnetism is a preferable property in these materials for integrated memory-logic applications. While such properties are difficult to achieve at RT in commonly used semiconductors such as Si and GaAs, ZnO doped with transition metal could have a Curie temperature above RT and could be an interesting material for spintronic applications.^49,124,132

ZnO in bulk form doped with Mn, Co, and Fe was grown using a melt-growth technique for the study of magnetic properties.^47,48 Mn-doped ZnO showed antiferromagnetic behavior, while Co-doped ZnO showed signs of ferromagnetism at RT. However, the origin of this magnetic behavior is not clear and is most likely due to Mn and Co clusters. 3d orbitals of the dopants could overlap with impurity bands in ZnO. Doping ZnO with Mn and Co could also help to tune its dielectric constant from ∼8.5 to over 25.¹³³ In another study, Fe was doped in bulk ZnO and ferromagnetic hysteresis curve was observed at RT.⁴⁷ Ni-doped ZnO thin films grown by MOCVD also exhibited RT ferromagnetism. Rare earth element Gd-doped ZnO thin films showed ferromagnetism at RT, but ZnGdO bulk material showed a diamagnetic response. Figure 17 shows magnetization hysteresis plots of ZnO:Fe bulk crystal grown by melt-growth technique, and MOCVD-grown ZnO:Gd thin films. The ferromagnetic behavior could be due to interactions of the dopant carriers with the host material carriers, or defect/impurity states, or magnetic clusters, or due to dopant-induced defects. ZnO doped with transition metal or rare earth element has clearly shown signs of RT ferromagnetism and a potential for spintronic applications.

Well-aligned Zn_0.94Cr_0.06O nanorod arrays were synthesized by the RF magnetron sputtering deposition at varying substrate temperatures.¹³⁴ The Zn_0.94Cr_0.06O nanorod arrays were aligned perpendicular to the Si substrate. The Zn_0.94Cr_0.06O nanorod arrays exhibited stable RT ferromagnetic ordering. The saturated magnetization was 1.16 mB per Cr ion at 650 °C and it decreased with substrate temperature. PL and O K-edge x-ray absorption near edge structure (XANES) analyses indicated the existence of numerous Zn vacancies. No secondary phases in the sample were found within the XANES and HRTEM detection limits. The ferromagnetism could originate from the V_Zn-mediated bound magnetic polaron model.¹³⁴ Characterization results are shown in Fig. 18. V_Zn could be a tuning parameter for ferromagnetism in ZnCrO.

ZnO doped with Ni and Co grown with colloids-based synthesis showed ferromagnetism above RT due to the increase in the domain volumes and lattice defects.¹⁸ Transition metal (Co, Ni)-doped ZnO grown by a colloidal procedure that involves hydrolysis and condensation in dimethyl sulfoxide, showed magnetic circular dichroism and the Zeeman effect above RT.¹²² Undoped ZnO quantum dots and nanostructures indicating the Rashba effect as per theoretical calculations also have the potential for room temperature spintronics.¹²⁵ 

Neuromorphic computing involves high-speed processing of enormous data with low power consumption and integrated memory and processing architectures, as compared to the currently used von Neumann architecture. It is inspired by brainlike parallel and event-driven processings.^135–138 Quantum information involves control and manipulation of quantum states and could be an enabling factor for high-speed computing.^123,139 Memristors, quantum diodes, transistors, atomic switches, and nano-oscillators are examples of potential high-speed devices for neuromorphic and quantum computing.

Room temperature single-photon emission with lifetimes of 1–4 ns in ZnO nanoparticles was reported; such single systems could help to better understand intrinsic defects, effects of local environment, and quantum states in ZnO.¹⁴⁰ ∼1.8 eV emissions from the conduction band to the zinc vacancy trap were observed. Point defects such as zinc and oxygen vacancies, interstitials, and antisites result in emissions that could also be influenced by ambient fluctuations. Signs of neuromorphic synapses in the form of loops in current-voltage characteristics were observed in In₂O₃–ZnO on a glass substrate coated with fluorine-doped tin oxide (FTO), as a result of charge carrier trapping and release.¹⁴¹ A transient and biodegradable W/MgO/ZnO/Mo device was developed to function as a memristor with W and Mo as electrodes, and ZnO and MgO as resistance switch layers.¹⁴² The device exhibited analog memristor effects, and synaptic long-term depression and potentiation along with spike time-dependent plasticity. Short and long-term synaptic plasticity was also seen in a ZnO-based ionic-electronic hybrid thin film transistor on polyamide substrates.^143,144 An Al/ZnO/FTO memristor showed a combination of analog and digital resistive switching; the analog effect is due to Schottky conduction, and the digital effect is due to the space charge region.¹⁴⁵ Endurance cycles in the order of 10⁴ with 10⁴ s retention period were measured. Ga-doping in ZnO thin films up to 0.5% showed enhanced resistive switching (as compared to bare ZnO) and could be optimized for desired synapsis endurance and retention with a semirectifying current-voltage relation. Ga in dilute amounts could increase the O²⁻ ions proportion, electrical carrier concentration, and synaptic weights.^146,147 Multistate memory levels could be potentially achieved by using the gate effect and combining ZnO with gate insulators such as tantalum oxide in ZnO-based thin film transistor.¹⁴⁸ 

Neutron detection and scintillating materials are important in nuclear fusion and fission reactions, sensing high energy radiations in space and materials, medical imaging, nondestructive testing, geological purposes, and energy characterization.^17,26,53,149 There has been an increasing interest in new materials for neutron detection, especially after the recognition of a scarcity of materials to build ³He detectors.^150,151 Scintillators absorb high energy radiation and emit radiation that are lesser energetic and could be processed by optical detectors. Neutrons do not carry charge or ionization capabilities, which makes their detection complicated. ZnO has been explored for scintillation applications due to its large cross section to interact with neutrons particles and a high exciton binding energy.

MOCVD-grown ZnO has been reported to undergo carrier recombination during scintillation and has a rise and decay time of 30 ps and 0.65 ns, respectively.²⁶ A neutron detection response of bare ZnO, and ZnO with poly radiator for ⁶⁰Co gamma and PuBe neutrons spectrum, is shown in Fig. 19.

Luminescence in undoped ZnO is usually due to its intrinsic bandgap and self-absorption. Doping results in the addition of donor and acceptor states, and the luminescence could be tuned toward lower energies with a reduced self-absorption. Alpha and triton particles propagating through ZnO upon incidence recombine to form electron-hole pairs and result in scintillation. Gamma-ray and electron recoiling interaction is in the millimeter scale, while alpha interaction is in the micrometer range. Hence, strong emissions from neutron-alpha matter interactions could be effectively detected as shown in Fig. 20. High-quality ⁶Li- and ¹⁰B-doped ZnO thermal neutron scintillator detector grown using MOCVD and studied under gamma-rays and ²²⁶Ra incidence identified high sensitivity, and high neutron-to-gamma-ray discrimination, and seem to experimentally follow an ideal detection response; this is currently under investigation.

Cu-doped ZnO grown by a liquid phase epitaxy method on ZnO substrates was also explored for neutron detection.¹⁵² A scintillation light yield that is 140% times more than the commonly used BiGeO scintillator was achieved. Cu-doped ZnO films had a decay time of 21 500 nm as compared to 2300 ns for undoped ZnO. Photoluminescence spectra as well as radioluminescence responses when irradiated with ²⁴¹Am 5.5 MeV showed emission at 450–650 nm. These states could be the results of recombination of oxygen vacancy donor states with deep acceptor-type energy states of Cu²⁺.

A direct bandgap, high carrier diffusivity, carrier concentration/mobility, along with a nontoxic nature, abundancy, and a crystalline structure, enables ZnO to be applicable in various parts of solar cells.^{5,34,35,38,153–165} Bandgap engineering and photon absorption are factors that make it a good active layer. ZnO could also be a substrate, passivation layer, carrier blocking layer, or transparent oxide layer. Having multifunctionality in a single material base could reduce interface-related defects in layered and multijunction solar cell structures.

Aluminum-doped ZnO and indium-doped ZnO are the most common transparent conductive oxides with a high carrier absorption.^{36,121,154–161} Boron-doped ZnO has shown a low absorption, low carrier concentration, but a high mobility. Boron-doped ZnO are self-textured and have good light-trapping applications. ZnO could be used in heterojunction solar cell assemblies such as ZnO/CdSe, ZnO/CdS, ZnO/Cu₂O, ZnO/PbSe, ZnCuAlO, ZnO/ZnS, and ZnO/ZnTe. ZnO also has applications in dye-sensitized solar cells that are based on the excitation of dyes and redox based reactions in an electrolyte.^166–168 Roles of ZnO such as electron transport layer, and hole block layer, are directly applicable as carrier selective materials in perovskite solar cells.^169–175 ZnO has a conduction band at 4.4 eV and a valence band at 7.6 eV, which could effectively excite electrons from the lowest unoccupied molecular orbital and block holes from the highest occupied molecular orbital level in a perovskite solar cell.

ZnO-based Schottky junction solar cells were built using MOCVD-grown ZnO as the active layer.¹²¹ Schottky junction solar cells avoid p-doping related challenges and have a simple device structure and lower cost. Figure 21 shows the photovoltaic responses and has insets of the ZnO/Ag Schottky junction structure with Ti/Au ohmic contacts. The current-voltage characterization exhibits diodelike rectifying behavior. Open circuit voltage and short circuit current increase with illumination intensities of solux lamp, confirming the photovoltaic performance.

A heterojunction solar cell structure consisting of ZnO and silicon active layers as shown in Fig. 22 was theoretically studied.¹²⁰ Input parameters of ZnO for the simulation were experimentally acquired using MOCVD-grown ZnO characterization. The refractive index of ZnO matches with anti-reflection requirement for silicon solar cells, so ZnO acts as an anti-reflection layer. A reference external quantum efficiency measured for 0.5 μm thick ZnO as shown in Fig. 22(b) was used in the simulations. Figure 22(c) shows the simulation results. Short circuit current and the open circuit voltage decrease with ZnO thickness due to a reduction in the quantity of photons reaching the space charge region. Optimized ZnO thickness and optical parameters could be used in the development of multijunction solar cells.

ZnO was also used as a passivation layer in a metal/GaAs Schottky solar cell.¹⁶⁵ Passivation layers can reduce surface recombination, interface traps, and dangling bonds, resulting in a better photovoltaic performance. Two types of passivation layers were deposited on GaAs-based Schottky solar cells, one with a 20 nm Al₂O₃ layer and the other with an 18/2 nm Al₂O₃/ZnO layer. Al₂O₃/ZnO passivation resulted in a better photovoltaic response (Fig. 23). Al₂O₃ introduces negative fixed charges that increase the electric field to separate photo-generated electron-hole pairs, while ZnO further increases the fixed negative charges and reduces As-related oxide layers at the surface. This enhances the electric field and reduces carrier recombination at the device surface and the reverse saturation current.

A high free exciton binding energy of 60 meV, tunable bandgap, and high carrier concentration, make ZnO a narrowband emitter, and a potential add-on to GaN as a typical semiconductor material for light-emitting diode (LED).^{16,64,174–177}

In ZnO LED built from MOCVD-grown p–n junction, a turn-on voltage of 3.3 V and reverse breakdown voltage of 10 V was reported.¹⁶ An electroluminescence peak at 384 nm was seen as shown in Fig. 24 due to a recombination of the nitrogen luminescent centers (p-type dopant) and shallow donors in n-type ZnO. ZnO could either be used as a homojunction LED or could even be interfaced with other p-type material, for example, p-GaN/n-ZnO LED.¹⁷⁶ Some other configurations include ZnO on Si, ZnO on SiC, nanowire and quantum dots LEDs wherein the nanostructure-related properties would append to the inherent ZnO characteristics that are conducive for LEDs' operations.

In another study, ZnO/GaN-based LEDs with improved asymmetric double heterostructure of Ta₂O₅/ZnO/HfO₂ were fabricated. Three types of devices: LED 1 with structure of n-ZnO/i-ZnO/p-GaN, LED 2 with n-ZnO/i-ZnO/HfO2/p-GaN, and LED 3 with n-ZnO/Ta2O5/i-ZnO/HfO2/p-GaN were considered.¹⁷⁸ Energy bands of the LEDs are shown in Fig. 25. I–V characteristics are shown in Fig. 26.

Electroluminescence (EL) of the three LEDs is shown in Fig. 27. The EL performance was enhanced by the HfO₂ electron blocking layer and further improved by the Ta₂O₅ hole-blocking layer. The origins of the emission indicate that the Ta₂O₅/ZnO/HfO₂ asymmetric structure more effectively confined carriers in the active i-ZnO layer and suppress unintended radiation from GaN. This device exhibits good stability in long-time running. The asymmetric double heterostructure could be helpful for the development of the future ZnO-based LEDs.

There has been a recent interest in thermoelectric materials, considering the potential of effectively harnessing heat energy that is naturally available and also usually wasted as a byproduct.^{2,17,21,71,121,179–182} Thermoelectric materials are characterized by a thermoelectric figure of merit defined as ZT = S²σ/(k_e+ k_l), where T is the temperature, S is the Seebeck coefficient, σ is the electrical conductivity, k_e is the carrier thermal conductivity, and k_l is the lattice thermal conductivity; S²σ is the power factor. A high power factor and a large Seebeck coefficient are conducive for energy generation, and sensing applications. Materials with a low thermal conductivity and high electrical conductivity are of interest, in order to maintain a temperature difference and yet have a carrier flow through the material. σ and k_e are inter-dependent as per the Wiedemann–Fran relation, which makes it difficult to achieve a figure of merit greater than unity. Bi₂Te₃ and Sb₂Te₃ are commonly explored thermoelectric materials; however, they are expensive, toxic, and not sufficiently stable at high temperatures.

ZnO could be an effective alternative to conventional thermoelectric materials due to its nontoxicity, abundancy, and high temperature stability. Co-doping with Al and Ga could enhance the thermoelectric properties of ZnO.^121,180 Bulk ZnO has been reported to have a Seebeck coefficient up to 478 μV/K, with a power factor of 0.75 × 10⁻⁴ W/m K².^121,180 ZnO doped with Al and Ga have been reported to have a ZT up to 0.45 (and power factor up to 15 W/m K²) at 1000 K, however, the electrical conductivity was low.^121,180 Figure 28 shows Seebeck coefficients and power factors of ZnO materials. High Seebeck coefficients are seen in ZnO and ZnAlO thin films. Bulk ZnAlGaO and ZnAlO have a high power factor. Al or Ga dopant could substitute Zn in ZnO and result in an enhanced electrical conductivity, yet maintaining a low thermal conductivity. Figure of merit and the Seebeck coefficient of ZnO:Al are improved by codoping with Sm, Fe, and Ni, however, a low electron mobility was observed.¹⁷ 

ZnO nanoparticles could be embedded in cement to utilize the heat energy and temperature differences in concrete structures.^21,166 ZnO and Al-doped ZnO were mixed in cement with a varying ZnO content in one set of cement and varying ZnO:Al in another.²² As seen in Fig. 29, the Seebeck coefficient increases, and thermal conductivity decreases with ZnO or ZnO:Al incorporation in the cement paste. ZnO could reduce hydration reactions and result in an improved thermoelectric performance.

Zinc oxide nanoparticles could be used as additives or sensing agents in construction materials, resulting in a reduced damage potential and increased service life and safety of infrastructure.

ZnO nanoparticles in optimum amounts could react with concrete lime and its high surface energy could result in faster hydration. Nanoparticles could improve concrete reinforcement, reduce steel bars corrosion, and fill in pores.^183–185 Zinc oxide nanoparticles up to 0.5% in self-compacting concrete prepared using Ordinary Portland Cement (ASTM C150 standard) increased the early-age flexural strength from 2.0 MPa with 0% ZnO to 4.0 MPa at 0.5% ZnO in 7 days,¹⁸³ along with a decrease in settling time. The strength did not show significant changes after 7 days and was 2.0 MPa at 0% ZnO and 2.3 MPa at 0.5% ZnO after 28 days. Also, split tensile strength showed an increase from 4.7 to 7.0 MPa after 7 days with the addition of 0.5% ZnO nanoparticles. ZnO nanowires grown on high strength polyacrylonitrile based carbon fabric exhibited a 20% increase in strength, 7% increase in Young's modulus, and 88% increase in interlaminar shear strength as compared to bare carbon fabric.¹⁸⁶ Adding ZnO nanoparticles to asphalt mixes showed better adhesion of aggregate with asphalt binder, especially in wet conditions.¹⁸⁷ ZnO-modified asphalt could have improved and tunable antiaging (especially anti-UV aging), high-temperature stability, ductility, and viscosity-recovery characteristics as compared to bare asphalt.¹⁸⁸ ZnO additives in construction materials such as concrete reduce air pollutants by controlling release of carbon dioxide and increasing oxygen levels due to their photocatalytic properties.^189–191 

ZnO nanoparticles could be applied as piezoelectric transducers, strain sensors, and energy harvesting materials in structural health monitoring and smart infrastructure.^192,193 Structural health monitoring is crucial to identify cracks, deterioration, corrosion, evaluate infrastructure's health condition and reduce chances of catastrophic failures. ZnO-based transistors have exhibited a pressure sensing sensitivity of up to 4 nA per kPa with a latency of less than 1 ms.¹⁹² Embedding ZnO nanoparticles in poly(vinyl fluoride) (PVDF) matrix increased its piezoelectricity systematically with the ZnO content from 0% to 20% and exhibited polarization-to-electric field hysteresis with remnant polarizations up to 0.0078 C m⁻².¹⁹⁴ Hammer impact tests showed that the ratio of corresponding maximum voltages of ZnO/PVDF to commercial PVDF could increase up to 1.6 with the 10%–20% ZnO content. Effects of adding ZnO nanoparticles were further validated by dynamic strain sensing which showed an increase in sensitivity from 1.93 at 0% ZnO to 3.24 at ∼10% ZnO. Ferroelectric Li-doped ZnO nanowires added to polymer composites have a potential for dynamic motion sensing as the resistance and piezoelectric effect are sensitive to mechanical movements and temperatures.¹⁹⁵ 

ZnO has applications in flexible electronics considering its tunable bandgap, exceptional electrical characteristics, physical flexibility, transparency, biocompatibility, and low temperature synthesis.^196–202 ZnO (grown by sputtering) based thin film transistors (TFTs) exhibited a field-effect mobility up to 50 cm² V⁻¹ s⁻¹, current on/off ratio in the order of 10⁶, threshold voltage of ∼0 V, and subthreshold slope of 3 V/decade.²⁰³ ZnO-based TFTs could have a threshold voltage up to 5.4 V.¹⁹⁶ These characteristics could be enhanced by alloying ZnO with indium.²⁰⁴ Field-effect properties of ZnO were also improved through the formation of heterostructures such as ZnO/ZnMgO with a two-dimensional electron gas at the interface, resulting in a mobility up to 9.1 cm² V⁻¹ s⁻¹, current on/off ratio up to 10⁸, turn-on voltage of −2.75 V, and a subthreshold slope of 3330.78 V/decade.^17,205 Nonlinear current-voltage characteristics (turn-on voltage, breakdown voltage, and current-voltage slope) of ZnO ceramics are also utilized in varistors.^206–208 

In addition to the above-mentioned applications, ZnO could be used in gas sensing, pharmaceuticals, cosmetics, and textile industries.^6,209–211 ZnO nanostructures are permeable, block UV-radiation, and can be used as textile coating materials. Antibacterial and disinfecting properties of ZnO make it a component of medicines, creams, medication to heal wounds, dental pastes, and dietary supplements.^6,211–213 ZnO nanoparticles could be incorporated in food packaging material to inhibit the growth of microorganisms on the food surface.²⁰⁹ ZnO nanoparticles are also explored for antitumor activities, antidiabetic activities, therapies for lung cancer, gastric cancer, hepatocarcinoma, cervical cancer, ovarian cancer, breast cancer, colon cancer, leukemia, functionalizing nano-element and carriers to target cancer cells, and biomedical imaging.^{209,214–218}

ZnO is a material of ongoing research with developing growth techniques and interesting characteristics. ZnO could be grown in bulk, thin films, and nanostructure forms with good crystal quality. An understanding of the growth mechanisms helps to control ZnO quality and resulting properties. As native defects in ZnO are n-type, p-type doping has been difficult and needs to be investigated more so that device-quality p–n junctions could be built on a large scale. In spite of the current challenges, prospects for controlled p-doping have been experimentally observed, and ZnO has shown promising characteristics. ZnO solar cells and light-emitting diodes have been developed. The large Seebeck coefficient, electron mobility, and carrier concentration have been observed. Interesting spintronic, neuromorphic computing, neutron detection, civil infrastructure-related, and biomedical capabilities have been reported. While there are challenges in ZnO yet to be completely resolved, it could be an enabling material for faster progress in several areas of optoelectronics, energy harvesting, sensing, and biomedicine and a potential multifunctional material for the next generation of devices and applications.

The authors would like to acknowledge the National Science Foundation CAREER Project No. CMMI 1560834 and DARPA Nascent Light-Matter Interactions Programs for financial support to conduct this research.

The authors have no conflicts to disclose.

Vishal Saravade: Conceptualization (equal); Data curation (equal); Formal analysis (equal); Investigation (equal); Methodology (equal); Project administration (equal); Writing – original draft (lead); Writing – review & editing (lead). Zhe Chuan Feng: Data curation (equal); Formal analysis (equal); Investigation (equal); Visualization (equal); Writing – original draft (lead); Writing – review & editing (lead). Manika Tun Nafisa: Software (equal); Validation (equal); Visualization (lead). Chuanle Zhou: Formal analysis (equal); Methodology (equal); Supervision (equal); Writing – original draft (equal); Writing – review & editing (equal). Na Lu: Conceptualization (equal); Funding acquisition (equal); Project administration (equal); Supervision (equal); Writing – original draft (equal); Writing – review & editing (equal). Benjamin Klein: Funding acquisition (equal); Project administration (equal); Resources (equal); Supervision (equal); Validation (equal). Ian Ferguson: Conceptualization (equal); Funding acquisition (equal); Investigation (equal); Project administration (lead); Resources (equal); Validation (equal); Writing – original draft (equal); Writing – review & editing (equal).

The data that support the findings of this study are available from the corresponding author upon reasonable request.