Epsilon-near-Zero Modes and Surface Plasmon Resonance in

Letter pubs.acs.org/journal/apchd5

Epsilon-near-Zero Modes and Surface Plasmon Resonance in Fluorine-Doped Cadmium Oxide Thin Films Evan L. Runnerstrom,† Kyle P. Kelley,† Edward Sachet,† Christopher T. Shelton,† and Jon-Paul Maria*,† †

Department of Materials Science and Engineering, North Carolina State University, Raleigh, North Carolina 27695, United States

Downloaded via UNIV OF CAMBRIDGE on October 10, 2018 at 06:47:33 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

S Supporting Information *

ABSTRACT: In this report we demonstrate fluorine-doped CdO as a model infrared plasmonic material by virtue of its tunable carrier density, high mobility, and intense extreme-subwavelength plasmon−polariton coupling. Carrier concentrations ranging from 1019 to 1020 cm−3, with electron mobility values as high as 473 cm2/V·s, are readily achieved in epitaxial CdO films over a thickness range spanning 50 to 500 nm. Carrier concentration is achieved by reactive sputtering in an Ar/O2 atmosphere with trace quantities of CF4. Infrared reflectometry measurements demonstrate the possibility of near-perfect plasmonic absorption through the entire mid-IR spectral range. A companion set of reflectivity simulations are used to predict, understand, and optimize the epsilon-near-zero plasmonic modes. In the context of other transparent conductors, CdO exhibits substantially higher electron mobility values and thus sharp and tunable absorption features. This highlights the utility of high-mobility transparent conducting oxides as a materials system for supporting strong, designed light−matter interactions. KEYWORDS: transparent conductive oxide, mid-infrared, epsilon-near-zero mode, surface plasmon polariton, reactive sputtering

T

exhibits the highest electronic mobility among TCOs and the best plasmonic performance in the IR.6,15−18 In 2015, we showed that dysprosium-doped CdO (Dy:CdO), grown by molecular beam epitaxy (MBE), can achieve mobilities reaching 500 cm2/V·s and tunable carrier concentrations between 1019 and 1021 cm−3.6 Despite the exceptional plasmonic performance of this material, MBEgrown Dy:CdO has two significant practical disadvantages: film growth by MBE is slow and difficult to scale commercially, and Dy is an expensive and reactive element. Nevertheless, given the promise of CdO as a plasmonic host, we are exploring alternative dopants and deposition methods to enhance the utility of this material. In a separate, contemporary report, we have shown that reactive high-power impulse magnetron sputtering (HiPIMS) is a viable method for depositing Ydoped CdO films with electronic properties on par with films deposited by MBE.19 These films skirt the aforementioned disadvantages, as Y is more prevalent and more stable than Dy, and HiPIMS combines the deposition rates and scalability of

hanks to sustained and growing interest in light−matter interactions at infrared (IR) energies,1 transparent conducting oxides (TCOs) are commanding considerable attention from the photonics and plasmonics communities. As a class of plasmonic materials, TCOs offer boutique optical properties and absorption by design throughout the near- and mid-IR. By virtue of accessible electron concentrations ranging from 1018 to 1021 cm−3, TCOs intrinsically interact with IR light and can be tuned to resonate with specific IR energies by doping.2−9 By contrast, traditional plasmonic materials, such as metals, have fixed electron concentrations and thus require sophisticated nanofabrication techniques to engineer IR light interactions.1,10 The ability to tune IR optical properties in TCOs enables advanced optical materials and devices operating at telecommunications and biologically relevant wavelengths.1,8 Additionally, some TCOs enjoy high electron mobility, particularly when defect chemistry is tightly controlled in high-quality crystals.6,11−13 Electron mobility has a strong impact on plasmonic performance, and high mobilities correspond to stronger light−matter interaction, more effective light concentration, and sharp plasmonic resonances with highquality factors.3,6,12,14 To date, rocksalt cadmium oxide (CdO) © 2017 American Chemical Society

Received: April 26, 2017 Published: July 18, 2017 1885

DOI: 10.1021/acsphotonics.7b00429 ACS Photonics 2017, 4, 1885−1892

ACS Photonics

Letter

Figure 1. Structural and morphological characterization of F:CdO thin films. (a) Typical 2θ−ω XRD scan of heteroepitaxial CdO, with a (001) growth habit, on r-plane sapphire. (b) Typical 2D reciprocal space map showing epitaxial relationship between CdO and r-plane sapphire. (c) Typical AFM scan of an as-grown F:CdO film with an rms roughness of 858 pm.

magnetron sputtering while producing smooth and dense films comparable to those grown by MBE. Still lacking, however, are exhaustive studies of dopants for high-mobility plasmonic CdO. A number of aliovalent dopants that substitute Cd2+ on the cation sublattice have been studied, including Dy, Y, In, Sc, Sn, V, and Ti.6,15−17,20−23 Rarely investigated, however, are aliovalent dopants that substitute O2− on the anion sublattice. Examining the crystal ionic radius of six-coordinate O2− (126 pm) and considering anions that would act as donors at an oxygen site reveals F− (119 pm) as an appealing candidate in CdO.24 Fluorine-doped CdO (F:CdO) has been studied,25,26 but the electronic properties of F:CdO films are so far not comparable to Dy:CdO. Furthermore, with the notable exception of colloidal F:CdO nanocrystals,27,28 we know of no reports on the plasmonic properties of F:CdO thin films. By contrast, fluorine is a viable and very well studied dopant in tin oxide (SnO2).13,29−31 We aim here to determine if substitutional, aliovalent anion doping with fluorine is a strategy for producing plasmonic CdO films with tunable carrier concentration and high electron mobility. Fluorine is a promising dopant for a few reasons. First, given the F−/O2− ionic radius match, we anticipate only modest lattice strain associated with doping; this should minimize carrier scattering in the same manner as Dy3+ in CdO.6 Second, the CdO conduction and valence bands have primarily Cd 5s and O 2p/2s character, respectively.20,21 For F− substituted on an O2− site, we expect electronic structure perturbations primarily in the valence band, which would limit ionized impurity scattering in a manner analogous to modulation doping. This effect has been observed in SnO2 by computational studies.32 Third, from the coupled intrinsic and extrinsic defect reactions (written in Kröger−Vink notation) of F-doped CdO, x x Cd Cd + OOx ↔ Cd Cd + VO.. + 2e′ +

1 1 . F2,(g) ←→ ⎯ FO + e′ + O2,(g) CdO 2 2

1 O2,(g) 2

This would allow control over dopant concentration by changing the partial pressure of the dopant precursor gas, which simplifies the design and operation of the vacuum sputtering chamber, eliminating the need for multiple magnetrons or alloyed targets, and enabling quick and in-line changes to film properties without breaking vacuum. Herein, we present a study of the electronic, morphological, and plasmonic properties of F:CdO thin films deposited by HiPIMS. By using CF4 as a fluorine source, we access free carrier concentrations spanning an order of magnitude, from 1019 to 1020 cm−3, with electron mobilities as high as 473 cm2/ V·s. These properties, combined with control over film thickness, allow us to grow F:CdO films that sustain epsilonnear-zero (ENZ) modes in addition to surface plasmon resonance. We access ENZ modes spanning an energy range greater than 1800 cm−1 (2700 nm) across the mid-IR, with peak widths as small as 321 cm−1/350 nm and peak extinction values as high as 97%. These results rival Dy:CdO and Y:CdO in performance, despite the fact that the films are deposited by sputtering and using an alternative dopant. This underscores the performance and versatility of CdO as a plasmonic host.

■

RESULTS AND DISCUSSION We grow F:CdO thin films on r-plane (012) single-crystal sapphire substrates using reactive HiPIMS from a 99.9999% pure metallic Cd target. The sputtering environment is a mixture of argon, oxygen, and CF4, with a 6:4 ratio of Ar:O2 by pressure and a total sputtering pressure of 10 mTorr. The fluorine content of the sputtered F:CdO films is controlled by adjusting the background pressure of CF4 using a leak valve and ion gauge. X-ray diffraction (XRD) shows that F:CdO grows epitaxially on r-plane sapphire along the [001] direction, as symmetric 2θ−ω scans and reciprocal space maps (Figure 1a,b) show only substrate and CdO (002) peaks. Peak positions shift to slightly higher 2θ values at high dopant concentrations, suggesting a slight lattice contraction in our films, which could be due to the slightly smaller ionic radius of F− compared to O2−. Additionally, ω rocking curves about the CdO (002) peak and ϕ scans about the CdO (111) peak (Figure S1) confirm epitaxial growth with peak widths of 0.15° and 0.20° in ω and ϕ, respectively. These XRD data suggest that fluorine is a substitutional dopant on the oxygen sublattice and does not perturb epitaxial quality.

(1) (2)

LeChatlier’s principle predicts that F doping will drive the intrinsic defect equilibrium eq 1 toward the reactant side, suppressing the formation of oxygen vacancies. Note that we have already observed this behavior in Dy- and Y-doped CdO. Finally, F-doping presents the opportunity to use a gas phase precursor, such as fluorine (F2), sulfur hexafluoride (SF6), or tetrafluoromethane (CF4), in a reactive HiPIMS deposition. 1886


ACS Photonics

Letter

Figure 2. Electronic properties of F:CdO thin films. Electron concentration (filled circles), mobility (filled triangles), and conductivity (empty squares) are plotted as a function of CF4 background pressure in the deposition chamber. (a) The left panel displays the electronic properties of films as deposited, while the (b) right panel displays the electronic properties of the films after annealing under static O2 gas at 1 atm and 700 °C for 1 h. All films are 120 to 160 nm thick. The lines are meant as guides to the eye.

In addition to their crystalline quality, F:CdO films are uniform and smooth, as characterized with atomic force microscopy (AFM, Figure 1c). As-grown, F:CdO films have a root-mean-square (rms) roughness on the order of 500− 1000 pm, which is beneficial for reducing electron scattering at the film surface and is particularly advantageous for maintaining high electron mobility and limiting plasmon damping at the low film thicknesses required for supporting ENZ modes. We characterized the electronic properties of F:CdO films using Hall effect measurements to determine carrier concentration, mobility, and conductivity as a function of fluorine content (Figure 2a). By varying the pressure of CF4 present during sputtering, we access carrier concentrations in the asgrown films ranging from 3 × 1019 cm−3 for intrinsic/ unintentionally doped (UID) CdO to a maximum of 1.6 × 1020 cm−3 for F:CdO. Increasing the CF4 pressure beyond this maximum results in rough and nonuniform films without an increase in carrier concentration, presumably as a result of resputtering or etching by reactive fluorine ion species. Within this composition range, carrier concentration increases monotonically with a logarithmic dependence on CF4 pressure, indicating that the presence of CF4 gas in a reactive sputtering environment leads to the generation of free carriers, presumably by F− incorporation. Coincident with carrier concentration increase, the electron mobility increases dramatically with the introduction of CF4, reaching a maximum of >400 cm2/V·s at 7.5 × 1019 e−/cm3 in the as-grown films. At higher CF4 pressures, the mobility decreases gradually to ∼350 cm2/V·s before dropping more precipitously at the highest electron concentration. The gradual drop is probably due to ionized impurity scattering of electrons from F−, and the stronger drop is likely from reactive F− etching or resputtering during deposition, adversely affecting film quality. This decrease in film quality is evident from AFM and XRD rocking curves of a F:CdO film deposited at the highest CF4 pressure (Figure S2). The high mobilities we achieve at elevated carrier concentrations let us rule out oxygen vacancies as the primary electron source in F:CdO. Oxygen vacancies are doubly charged defects and scatter electrons 4 times as strongly as singly ionized defects,2 so it would be impossible for the mobility and carrier concentration to increase simultaneously above UID levels if oxygen vacancies are the primary donor

defect. Rather, the mobility increase can instead be explained by invoking LeChatlier’s principle and the defect reactions 1 and 2, above, which show that fluorine incorporation enhances electron mobility by suppressing the formation of oxygen vacancies. On the basis of the carrier concentration and mobility trends and our XRD results, we assert that CF4 is serving as a source of F−, which is ultimately acting as a substitutional donor dopant on the oxygen sublattice. We also investigated how annealing in an oxygen-rich environment affects the electronic properties of F:CdO by heating the films to 700 °C in 1 atm of O2 for 1 h (Figure 2b). Because the concentration of oxygen vacancies in defect reaction 1 is proportional to p(O2)−1/6, annealing in pure oxygen could reduce oxygen vacancy concentrations by a factor as high as 7.5 (i.e., 4 mTorr O2 during deposition compared to 760 Torr during annealing), reducing ionized impurity scattering. Annealing may additionally heal other defects, such as dislocations and grain boundaries, to improve crystal quality and mobility. Indeed, we find that the electron mobility is significantly improved in every annealed sample, with a maximum mobility of 473 cm2/V·s at 5.4 × 1019 e−/cm3. This is comparable to the peak mobility we observed earlier in Dy:CdO.6 We also observe enhanced surface morphology by AFM in the annealed films, with rms roughness decreasing to 400−500 pm, and enhanced crystalline quality, as seen by a decrease in the rocking curve peak widths (Figures S3, S4). Finally, annealing decreases the carrier concentration in films with as-deposited carrier concentrations below 7.5 × 1019 cm−3, further indicating that oxygen defects are healed. Above 7.5 × 1019 cm−3, annealing increases carrier concentration slightly, which suggests that not all fluorine dopants are activated in the as-deposited films. Annealing may thus be providing the thermal energy necessary to ionize F− and inject free carriers into the conduction band and/or allowing fluorine ions to move from interstitial sites or displaced positions to equilibrium substitutional positions on the oxygen sublattice. Ultimately, we are able to reach conductivities of nearly 10 000 S/cm in annealed F:CdO films, a high value for TCOs.2 The electronic properties achievable here and the degree to which we can control them mean that F:CdO films are amenable to supporting strong, weakly damped plasmonic resonance in the infrared. The plasmonic properties of the films are dictated by their frequency-dependent dielectric function, 1887


ACS Photonics

Letter

Figure 3. IR-VASE measurements of very thin films of F:CdO in the Kretschmann configuration, with reflectivity plotted as the ratio of p-polarized to s-polarized light. (a) Simulated and experimentally measured reflectivity curves for a F:CdO film (sample f in Table 1), along with the simulated real and imaginary parts of the dielectric function. (b) Experimental reflectivity curves for several F:CdO films with varying carrier concentration. For ease of visual comparison, the data are normalized to the reflectivity minima and flat film reflectivity at 5000−6000 cm−1. The letters beneath each curve correspond to the sample ID in Table 1, which shows the electronic and optical properties of each sample. (c) Simulated mid-IR reflectivity map for sample f. (d) Experimental mid-IR reflectivity map for sample f. The dashed horizontal line cuts in the reflectivity maps correspond to the simulated and experimental curves shown in part a.

cm−1,36 which is far from the spectral regions examined here. ENZ modes at comparable energies would correspond to free carrier concentrations of ≲5 × 1018 cm−3, about an order of magnitude lower than measured here. We expect F:CdO films to support an ENZ mode when the following relation is satisfied:

which, assuming that the films behave as Drude metals, is given by ε(ω) = ε∞ −

ωp2 ω 2 + iω Γ

(3)

where ε∞ is the high-frequency dielectric constant (4.9−5.5 for CdO),23 Γ is the damping rate, which is inversely proportional

1+

2

to mobility, and ωp = ne /m*ε0 is the plasma frequency, which is proportional to the square root of electron concentration. For CdO, the electron effective mass m* is 0.21.33 The carrier concentrations achievable here correspond to plasma frequencies in the near-IR, which will lead to surface plasmon and ENZ modes in the near- to mid-IR. This approach (i.e., using the Drude model to relate carrier concentration and mobility to the dielectric function) is widely and successfully used to model the mid- and near-IR optical and plasmonic properties of highly doped oxide semiconductors.3,5−7,12,14,34 Note that the ε∞ term generally captures the response of the charged lattice ions to incident light, as long as the energies are far away from the optical phonon modes of the lattice, avoiding the need to modify the dielectric function to include dispersive lattice vibrations as in ref 35. This is not necessary for CdO, which has optical phonon modes of energies around 300−450

ε1kz ,3 ε3kz ,1

⎛ ε2kz ,3 ε1kz ,2 ⎞ ⎟⎟ = j tan(kz ,2d)⎜⎜ + ε2kz ,1 ⎠ ⎝ ε3kz ,2

(4)

ω2

where d is film thickness, kz , i 2(ω) = εi c 2 − k 2 is the square of the longitudinal (i.e., perpendicular to the film surface in the z direction) wavenumber in layer i (1 = free space, 2 = F:CdO, 3 = sapphire), εi is the relative permittivity of the layer, and k|| is the transverse (i.e., parallel to the film surface) wavenumber, with Re(kz,i) + Im(kz,i) ≥ 0.37 Equation 4 results from solving Maxwell’s equations in the absence of external excitations for the thin film system considered here, and the (k||, ω) pair that satisfies the equation defines the ENZ mode of the system. The physical basis for ENZ modes in plasmonic materials is thoroughly discussed elsewhere,37−39 but for the purposes of this report, the ENZ mode can be considered to be the longrange surface plasmon in the limit of very thin films well below 1888


ACS Photonics

Letter

Table 1. Thickness, Electronic Properties, and Optical Properties of the F:CdO Films with ENZ Features As Shown in Figure 3 sample ID

thickness (nm)

electron concentration (1019 cm−3)

electron mobility (cm2/V·s)

absolute peak extinction value

a

161

3.1

232

0.74

b

124

4.4

352

0.80

c

93

6.6

384

0.91

d

76

9.1

384

0.90

e

47

11.3

317

0.76

f

74

14.6

248

0.97

g

57

15.7

193

0.79

extinction peak position 1800 cm−1; 5556 nm 2151 cm−1; 4649 nm 2610 cm−1; 3831 nm 2934 cm−1; 3408 nm 3175 cm−1; 3150 nm 3486 cm−1; 2869 nm 3668 cm−1; 2726 nm

extinction peak fwhm 396 cm−1; 1237 nm 383 cm−1; 834 nm 321 cm−1; 473 nm 351 cm−1; 409 nm 352 cm−1; 350 nm 483 cm−1; 399 nm 522 cm−1; 390 nm

Q 4.5 5.6 8.1 8.4 9.0 7.2 7.0

Figure 4. Simulated (a) and experimental (b) reflectivity maps of a 580 nm thick F:CdO film with 9 × 1019 e−/cm−3, 430 cm2/V·s, showing angledependent extinction characteristic of surface plasmon polariton resonance.

shown in Figure 3a, simulation provides a rather accurate prediction of the ENZ mode of a 74 nm thick F:CdO film with a carrier concentration of 1.5 × 1019 cm−3 and a mobility of 248 cm2/V·s, including a reflection minimum that represents coupling 97% p-polarized light into the film’s ENZ mode. (Hereafter, we will refer to this coupling simply as absorption.) Moreover, the simulated dielectric function (Figure 3a) confirms the ENZ nature of this mode, as the real part of the dielectric function crosses zero at approximately the same energy as the absorption peak. Using simulations as a guide, we are able to grow a range of F:CdO thin films with physical properties targeted to sustain ENZ modes spanning the entire mid-IR (Figure 3b). The electronic and optical properties of these films, characterized by Hall effect measurements and IR-VASE, respectively, are summarized in Table 1. As opposed to the initial film series, these ENZ layers are significantly thinner, which is necessary to achieve perfect absorption. As such, their surface area-tovolume ratios are larger and their mobilities are somewhat lower; this effect has been observed in CdO,22 among many other TCO and semiconductor systems. All of our ENZ films absorb over 74% of incident p-polarized light at their resonance frequency, with the best-optimized films absorbing 90% to 97% of p-polarized light. Thanks to the high mobility of our films, the absorption features are sharp, with peak widths as narrow as 321 cm−1 (40 meV), giving quality factors (the ratio of peak energy to peak width) as high as 9.

the optical skin depth of the plasmonic material (ranging from 433 nm to 6 μm for our F:CdO across the mid-IR), with a resonant energy that is approximately equal to the zero point of the dielectric function, given by39 ωENZ ≈

ωp2 ε∞

− Γ2 (5)

The additional relevant characteristics of ENZ modes are that they are minimally dispersive, meaning that their energies are weakly dependent on incident angle (i.e., increasing k||), λp

they require film thicknesses d ≪ 5π , where λp is the wavelength of the plasma frequency, and the z-component of the electric field is nearly entirely confined within the film.37 This means that plasmonic films tailored to support ENZ modes will confine light more strongly and generate stronger electric fields. Such films can then be used in applications such as perfect light absorption over multiple incident angles.40,41 Using Mathematica code6 based on the Drude model and Fresnel’s reflection coefficients, we can simulate dielectric functions and ENZ modes in doped CdO to determine the thicknesses and carrier concentrations needed for near-perfect absorption. After growing the films, we measure their optical spectra using IR variable-angle spectroscopic ellipsometry (IRVASE) in the Kretschmann configuration, using a CaF2 prism to couple light into the film, and plot the reflectivity as the ratio of reflected p-polarized light to reflected s-polarized light. As 1889


ACS Photonics

Letter

the pump and by flowing Ar (20 sccm) and O2 (14.4 sccm) as process gas. Fluorine doping is achieved by introducing pure tetrafluoromethane (CF4) gas into the chamber using a vacuum leak valve and an ion gauge to set a background base pressure of CF4 prior to introducing the process gas. The HiPIMS plasma is controlled using a Starfire Industries IMPULSE pulsed power module, with a pulse width of 80 μs and a repetition rate of 800 Hz. The HiPIMS controller is driven by an Advanced Energy MDX 10k magnetron drive dc power supply. Deposition rate is controlled by the dc power supply current, generally fixed at 140 mA, and film thickness is controlled by deposition time. At this current, voltages ranging from 380 to 400 V are delivered to the HiPIMS controller, ultimately resulting in an average sputtering power of 30 W delivered to the Cd target as measured by the HiPIMS controller. All films are grown on epitaxial-polished r-plane sapphire substrates (Jiaozuo TreTrt Materials) affixed to a stainless steel sample holder using silver paint (Ted Pella). Prior to and during deposition, the sample holder and substrates are heated to 455 °C using a radiative sample heater in the deposition chamber. The substrate temperature is measured using a Raytek 1.6 μm MM series pyrometer. If desired, samples are annealed after deposition for 1 h at 700 °C in pure oxygen. Electronic properties of the films are measured using an Ecopia HMS-3000 Hall measurement system with a 0.51 T magnet. Film thickness is measured by X-ray reflectivity using a PANalytical Empyrean XRD in parallel beam geometry with a double-bounce hybrid monchromator and a parallel-plate collimator. Symmetric 2θ−ω scans, rocking curves, and ϕ scans are collected using the same geometry. Reciprocal space maps are collected using parallel beam geometry and a PANalytical PIXcel area detector. IR reflectivity data are measured on F:CdO films deposited on double-side polished r-plane sapphire in the Kretschmann configuration. A Woolam IR-VASE ellipsometer is used in conjunction with a home-built sample holder and a right-angle CaF2 prism (Thor Laboratories) to couple light into the plasmonic films from the back surface of the substrate. An index matching fluid (Cargille Series M, n = 1.720) is used between the prism and substrate for good light coupling.

We can also simulate and measure the angular dependence or dispersion of ENZ modes in our films. Figure 3c shows a simulated mid-IR reflectivity map of sample f (the same sample shown in Figure 3a) over multiple incident angles. The dark band of minimum reflected intensity at ∼3500 cm−1 represents strong coupling of incident light into the ENZ mode. As expected for an ENZ mode, the absorption band exhibits very little dispersion, and its energy does not change significantly with incident angle. Our experimental reflectivity map for sample f (Figure 3d), collected using IR-VASE over multiple incident angles, matches our simulation quite well. The absorption band in this sample is nondispersive, providing further evidence that we approach the limiting case (with decreasing thickness) of an ENZ mode in this extreme subwavelength layer. In addition to supporting ENZ modes, F:CdO films can also support surface plasmon polariton resonance (SPPR) provided the film thickness is an appreciable fraction of the skin depth, in this case ≳300−500 nm. We grew a 580 nm thick F:CdO film with 9 × 1019 e−/cm−3 and a mobility of 430 cm2/V·s so that we could observe mid-IR SPPR using IR-VASE. Figure 4 shows the simulated and experimentally measured IR reflectivity maps for this sample. The dark absorption band at 1800−2300 cm−1 is strongly dispersive and angle dependent, a clear signature of SPPR modes. As a result of the larger film thickness, we also observe a distinct interference fringe at higher energy in this film, which is consistent with simulation. We have shown that F:CdO is a high-performance plasmonic semiconductor. By controlling carrier concentration and thickness, we are able to manipulate the optical properties of F:CdO films to generate plasmonic ENZ and SPPR modes that are tunable across the mid-IR. Thanks to the high electronic mobilities achieved here, the absorption bands are intense and sharp with high quality factors and, particularly for ENZ modes, offer nearly perfect light extinction. While other TCOs, such as tin-doped indium oxide, can support SPPR and ENZ modes,9,41 the superior electronic properties of F:CdO lead to much stronger and narrower absorption features with finer control over the resonance energy. Gold-based metamaterials and devices offer another point of comparison,1,10,42−44 but they require extensive patterning to support spoof plasmons or perfect absorption in the IR, with additional complexity required for tunability, and the resulting optical features are broader and more strongly damped. Here, by contrast, we achieve sharp ENZ modes and near-perfect absorption using lithography-free, unpatterned, single sub-100 nm F:CdO films grown using straightforward reactive physical vapor deposition procedures. These results, especially in combination with our reported work on Dy- and Y-doped CdO, cement this material as a versatile plasmonic host. With the even greater ease of fabrication and manufacturing afforded by doping with fluorine from the gas phase, F:CdO thin films should be a viable building block for metamaterials, plasmonic sensors, and other advanced optical devices operating in the mid-infrared.

■

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acsphotonics.7b00429. Additional XRD characterization of F:CdO thin films (ϕand ω-scans), AFM and XRD comparison of F:CdO films deposited at intermediate and high CF4 pressures, AFM and XRD comparison of F:CdO films before and after annealing (PDF)

■

■

METHODS Heteroepitaxial F:CdO films are prepared using reactive HiPIMS from a circular 2 in. metallic Cd target made inhouse from pure Cd (99.9999%, Osaka Asahi Metal). The target is affixed to a MeiVac MAK 2 in. magnetron sputter source in a high-vacuum sputtering system with a turbomolecular pump (base pressure 5 × 10−8 Torr). The sputtering pressure is adjusted to 10 mTorr using a gate valve in front of

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. ORCID

Evan L. Runnerstrom: 0000-0002-9054-708X Notes

The authors declare no competing financial interest. 1890


ACS Photonics

■

Letter

and Directed Self-Assembly in a New Semiconductor Building Block, Indium-Doped Cadmium Oxide (ICO). Nano Lett. 2013, 13, 2857− 2863. (19) Kelley, K. P.; Sachet, E.; Shelton, C. T.; Maria, J.-P. High Mobility Yttrium Doped Cadmium Oxide Thin Films. APL Mater. DOI: 10.1063/1.4993799. In press. (20) Jin, S.; Yang, Y.; Medvedeva, J. E.; Ireland, J. R.; Metz, A. W.; Ni, J.; Kannewurf, C. R.; Freeman, A. J.; Marks, T. J. Dopant Ion Size and Electronic Structure Effects on Transparent Conducting Oxides. ScDoped CdO Thin Films Grown by MOCVD. J. Am. Chem. Soc. 2004, 126, 13787−13793. (21) Yang, Y.; Jin, S.; Medvedeva, J. E.; Ireland, J. R.; Metz, A. W.; Ni, J.; Hersam, M. C.; Freeman, A. J.; Marks, T. J. CdO as the Archetypical Transparent Conducting Oxide. Systematics of Dopant Ionic Radius and Electronic Structure Effects on Charge Transport and Band Structure. J. Am. Chem. Soc. 2005, 127, 8796−8804. (22) Yu, K. M.; Detert, D. M.; Chen, G.; Zhu, W.; Liu, C.; Grankowska, S.; Hsu, L.; Dubon, O. D.; Walukiewicz, W. Defects and Properties of Cadmium Oxide Based Transparent Conductors. J. Appl. Phys. 2016, 119, 181501. (23) Liu, C. P.; Foo, Y.; Kamruzzaman, M.; Ho, C. Y.; Zapien, J. A.; Zhu, W.; Li, Y. J.; Walukiewicz, W.; Yu, K. M. Effects of Free Carriers on the Optical Properties of Doped CdO for Full-Spectrum Photovoltaics. Phys. Rev. Appl. 2016, 6, 064018. (24) Shannon, R. D. Revised Effective Ionic Radii and Systematic Studies of Interatomic Distances in Halides and Chalcogenides. Acta Crystallogr., Sect. A: Cryst. Phys., Diffr., Theor. Gen. Crystallogr. 1976, 32, 751−767. (25) Santos Cruz, J.; Torres Delgado, G.; Castanedo Pérez, R.; Zúñiga Romero, C. I.; Zelaya Angel, O. Optical and Electrical Characterization of Fluorine Doped Cadmium Oxide Thin Films Prepared by the Sol−Gel Method. Thin Solid Films 2007, 515, 5381− 5385. (26) Deokate, R. J.; Pawar, S. M.; Moholkar, A. V.; Sawant, V. S.; Pawar, C. A.; Bhosale, C. H.; Rajpure, K. Y. Spray Deposition of Highly Transparent Fluorine Doped Cadmium Oxide Thin Films. Appl. Surf. Sci. 2008, 254, 2187−2195. (27) Ye, X.; Fei, J.; Diroll, B. T.; Paik, T.; Murray, C. B. Expanding the Spectral Tunability of Plasmonic Resonances in Doped MetalOxide Nanocrystals Through Cooperative Cation-Anion Codoping. J. Am. Chem. Soc. 2014, 136, 11680−11686. (28) Diroll, B. T.; Guo, P.; Chang, R. P. H.; Schaller, R. D. Large Transient Optical Modulation of Epsilon-Near-Zero Colloidal Nanocrystals. ACS Nano 2016, 10, 10099−10105. (29) Rakhshani, A. E.; Makdisi, Y.; Ramazaniyan, H. A. Electronic and Optical Properties of Fluorine-Doped Tin Oxide Films. J. Appl. Phys. 1998, 83, 1049−1057. (30) Stjerna, B.; Olsson, E.; Granqvist, C. G. Optical and Electrical Properties of Radio Frequency Sputtered Tin Oxide Films Doped with Oxygen Vacancies, F, Sb, or Mo. J. Appl. Phys. 1994, 76, 3797−3817. (31) Mientus, R.; Ellmer, K. Structural, Electrical and Optical Properties of SnO2‑X:F-Layers Deposited by DC-Reactive MagnetronSputtering From a Metallic Target in Ar−O2/CF4 Mixtures. Surf. Coat. Technol. 1998, 98, 1267−1271. (32) Li, C.; Li, J.; Li, S.-S.; Xia, J.-B.; Wei, S.-H. Selection Rule of Preferred Doping Site for N-Type Oxides. Appl. Phys. Lett. 2012, 100, 262109. (33) Burbano, M.; Scanlon, D. O.; Watson, G. W. Sources of Conductivity and Doping Limits in CdO From Hybrid Density Functional Theory. J. Am. Chem. Soc. 2011, 133, 15065−15072. (34) Agrawal, A.; Johns, R. W.; Milliron, D. J. Control of Localized Surface Plasmon Resonances in Metal Oxide Nanocrystals. Annu. Rev. Mater. Res. 2017, 47, 1−31. (35) Cada, M.; Blazek, D.; Pistora, J.; Postava, K.; Siroky, P. Theoretical and Experimental Study of Plasmonic Effects in Heavily Doped Gallium Arsenide and Indium Phosphide. Opt. Mater. Express 2015, 5, 340−352. (36) Cuscó, R.; Ibáñez, J.; Domenech-Amador, N.; Artús, L.; ZúñigaPérez, J.; Muñoz-Sanjosé, V. Raman Scattering of Cadmium Oxide

ACKNOWLEDGMENTS We gratefully acknowledge support for this work by NSF grant CHE-1507947 and by Army Research Office grants W911NF16-1-0406 and W911NF-16-1-0037. We also thank the Efimenko and Genzer groups (NCSU, CBE) for providing us access to the IR-VASE.

■

REFERENCES

(1) Stanley, R. Plasmonics in the Mid-Infrared. Nat. Photonics 2012, 6, 409−411. (2) Ellmer, K. Past Achievements and Future Challenges in the Development of Optically Transparent Electrodes. Nat. Photonics 2012, 6, 809−817. (3) Lounis, S. D.; Runnerstrom, E. L.; Llordés, A.; Milliron, D. J. Defect Chemistry and Plasmon Physics of Colloidal Metal Oxide Nanocrystals. J. Phys. Chem. Lett. 2014, 5, 1564−1574. (4) Boltasseva, A.; Atwater, H. A. Low-Loss Plasmonic Metamaterials. Science 2011, 331, 290−291. (5) Naik, G. V.; Shalaev, V. M.; Boltasseva, A. Alternative Plasmonic Materials: Beyond Gold and Silver. Adv. Mater. 2013, 25, 3264−3294. (6) Sachet, E.; Shelton, C. T.; Harris, J. S.; Gaddy, B. E.; Irving, D. L.; Curtarolo, S.; Donovan, B. F.; Hopkins, P. E.; Sharma, P. A.; Sharma, A. L.; et al. Dysprosium-Doped Cadmium Oxide as a Gateway Material for Mid-Infrared Plasmonics. Nat. Mater. 2015, 14, 414−420. (7) Sachet, E.; Losego, M. D.; Guske, J.; Franzen, S.; Maria, J.-P. MidInfrared Surface Plasmon Resonance in Zinc Oxide Semiconductor Thin Films. Appl. Phys. Lett. 2013, 102, 051111. (8) Noginov, M. A.; Gu, L.; Livenere, J.; Zhu, G.; Pradhan, A. K.; Mundle, R.; Bahoura, M.; Barnakov, Y. A.; Podolskiy, V. A. Transparent Conductive Oxides: Plasmonic Materials for Telecom Wavelengths. Appl. Phys. Lett. 2011, 99, 021101. (9) Losego, M. D.; Efremenko, A. Y.; Rhodes, C. L.; Cerruti, M. G.; Franzen, S.; Maria, J.-P. Conductive Oxide Thin Films: Model Systems for Understanding and Controlling Surface Plasmon Resonance. J. Appl. Phys. 2009, 106, 024903. (10) Pendry, J. B.; Martín-Moreno, L.; Garcia-Vidal, F. J. Mimicking Surface Plasmons with Structured Surfaces. Science 2004, 305, 847− 848. (11) Bhachu, D. S.; Scanlon, D. O.; Sankar, G.; Veal, T. D.; Egdell, R. G.; Cibin, G.; Dent, A. J.; Knapp, C. E.; Carmalt, C. J.; Parkin, I. P. Origin of High Mobility in Molybdenum-Doped Indium Oxide. Chem. Mater. 2015, 27, 2788−2796. (12) Runnerstrom, E. L.; Bergerud, A.; Agrawal, A.; Johns, R. W.; Dahlman, C. J.; Singh, A.; Selbach, S. M.; Milliron, D. J. Defect Engineering in Plasmonic Metal Oxide Nanocrystals. Nano Lett. 2016, 16, 3390−3398. (13) Calnan, S.; Tiwari, A. N. High Mobility Transparent Conducting Oxides for Thin Film Solar Cells. Thin Solid Films 2010, 518, 1839−1849. (14) Lounis, S. D.; Runnerstrom, E. L.; Bergerud, A.; Nordlund, D.; Milliron, D. J. Influence of Dopant Distribution on the Plasmonic Properties of Indium Tin Oxide Nanocrystals. J. Am. Chem. Soc. 2014, 136, 7110−7116. (15) Coutts, T. J.; Young, D. L.; Li, X.; Mulligan, W. P.; Wu, X. Search for Improved Transparent Conducting Oxides: a Fundamental Investigation of CdO, Cd2SnO4, And Zn2SnO4. J. Vac. Sci. Technol., A 2000, 18, 2646. (16) Yan, M.; Lane, M.; Kannewurf, C. R.; Chang, R. P. H. Highly Conductive Epitaxial CdO Thin Films Prepared by Pulsed Laser Deposition. Appl. Phys. Lett. 2001, 78, 2342. (17) Wang, A.; Babcock, J. R.; Edleman, N. L.; Metz, A. W.; Lane, M. A.; Asahi, R.; Dravid, V. P.; Kannewurf, C. R.; Freeman, A. J.; Marks, T. J. Indium-Cadmium-Oxide Films Having Exceptional Electrical Conductivity and Optical Transparency: Clues for Optimizing Transparent Conductors. Proc. Natl. Acad. Sci. U. S. A. 2001, 98, 7113−7116. (18) Gordon, T. R.; Paik, T.; Klein, D. R.; Naik, G. V.; Caglayan, H.; Boltasseva, A.; Murray, C. B. Shape-Dependent Plasmonic Response 1891


ACS Photonics

Letter

Epilayers Grown by Metal-Organic Vapor Phase Epitaxy. J. Appl. Phys. 2010, 107, 063519. (37) Campione, S.; Brener, I.; Marquier, F. Theory of Epsilon-NearZero Modes in Ultrathin Films. Phys. Rev. B: Condens. Matter Mater. Phys. 2015, 91, 121408. (38) Campione, S.; Liu, S.; Benz, A.; Klem, J. F.; Sinclair, M. B.; Brener, I. Epsilon-Near-Zero Modes for Tailored Light-Matter Interaction. Phys. Rev. Appl. 2015, 4, 044011. (39) Campione, S.; Wendt, J. R.; Keeler, G. A.; Luk, T. S. NearInfrared Strong Coupling Between Metamaterials and Epsilon-NearZero Modes in Degenerately Doped Semiconductor Nanolayers. ACS Photonics 2016, 3, 293−297. (40) Luk, T. S.; Campione, S.; Kim, I.; Feng, S.; Jun, Y. C.; Liu, S.; Wright, J. B.; Brener, I.; Catrysse, P. B.; Fan, S.; et al. Directional Perfect Absorption Using Deep Subwavelength Low-Permittivity Films. Phys. Rev. B: Condens. Matter Mater. Phys. 2014, 90, 085411. (41) Yoon, J.; Zhou, M.; Badsha, M. A.; Kim, T. Y.; Jun, Y. C.; Hwangbo, C. K. Broadband Epsilon-Near-Zero Perfect Absorption in the Near-Infrared. Sci. Rep. 2015, 5, 12788. (42) Cooper, C. T.; Rodriguez, M.; Blair, S.; Shumaker-Parry, J. S. Mid-Infrared Localized Plasmons Through Structural Control of Gold and Silver Nanocrescents. J. Phys. Chem. C 2015, 119, 11826−11832. (43) Liu, N.; Mesch, M.; Weiss, T.; Hentschel, M.; Giessen, H. Infrared Perfect Absorber and Its Application as Plasmonic Sensor. Nano Lett. 2010, 10, 2342−2348. (44) Liu, X.; Starr, T.; Starr, A. F.; Padilla, W. J. Infrared Spatial and Frequency Selective Metamaterial with Near-Unity Absorbance. Phys. Rev. Lett. 2010, 104, 207403.

1892


Epsilon-near-Zero Modes and Surface Plasmon Resonance in

Recommend Documents