We successfully developed an atomic layer deposition (ALD) method for making Ag noble nanoparticles on cheap, commercial filter paper consisting of three-dimensional porous glass fibers and investigated the evolution of Ag nanostructures with some key process parameters. By tuning Ag particle sizes and controlling the cycle numbers of ALD deposited Ag films, we were able to obtain high-density isolated Ag nanoparticles with average sizes in 3–9 nm without the formation of agglomerates and continuous Ag films. We proved the presence of strong localized surface plasmon resonance peaks near a target wavelength of 632 nm. We further proved the presence of surface enhanced Raman scattering (SERS) signals on the Ag coated filter paper substrates using pyridine as the test analyte. Our results demonstrate that ALD is a very promising technique for a rational design of SERS substrates and, thus, has great potential for the fabrication of large-area, low-cost SERS substrates for future commercial applications, as compared to other existing techniques.

Surface enhanced Raman scattering (SERS) has emerged as a promising spectroscopic tool for ultrasensitive trace detection of target molecules in the vicinity of nanostructured noble metal surfaces.1,2 Among many applications, it has proven to be a reliable, field deployable detection technique for assessing chemical threats, which include chemical warfare agents, energetic materials, illicit narcotics, and many other classes of compounds of interest to defense and law enforcement.3–6 Noble metal nanostructures (MNSs) of Au, Ag, or other noble metals are employed as a key part of SERS substrates to enhance the sensitivity of Raman spectroscopy because of their unique localized surface plasmonic resonance (LSPR) properties. Enhancement factors (EFs) as high as 106 (theoretically up to 1010–1011) have been achieved with such substrates by readily tuning the optical properties of MNS surfaces through engineering their shape, size, orientation, and architecture, even allowing for single-molecule SERS detection in certain specific cases. In order to achieve very high sensitivities with SERS over more than a single random location (i.e., isolated hot spots), it is necessary to take advantage of the localized electric field enhancement associated with MNS such as nanoparticles (NP), nanorods, nanowires, nanotubes, and core–shell nanostructures by tuning the gaps to create a high density of isolated “hot spots.” In addition, to optimize the EFs of SERS substrates over a large area, the ability to produce a substrate that is environmentally stable, capable of being recycled, conformal to the analysis surface of interest, and cost-effective is critical, particularly for the development a truly field deployable SERS detection system. Recently, three-dimensional (3D) SERS substrates with steric extension along the z axis dimension, such as multilayers, microporous substrates, and array SERS substrates, have attracted significant attention for their special construction features.7 3D SERS substrates have a larger overall surface area, more hotspot quantities, a higher utilization efficiency of laser, and greater flexibility in all dimensions to enhance SERS performance and simplify substrate pretreatments.

Since the discovery of SERS, many methods have been developed to fabricate SERS substrates, with most of them falling into three categories: metal NPs by colloidal chemistry, nanopatterned surfaces, physical and chemical vapor deposition (PVD/CVD), or a combination of these. Although large SERS enhancements have been achieved using these techniques, the fabrication of such substrates is limited in the probing amount and often involves complicated and sometimes irreproducible fabrication steps with high production cost. Furthermore, all the above techniques have limitations to coating hotspots directly inside 3D SERS substrates. So, on the one hand, there is high demand for superenhanced, ultrasensitive SERS substrates, and on the other hand, it is still challenging for SERS-active substrates to enter the realm of real, practical applications due to a lack of stability, reproducibility, and reusability for such substrates. Thus, larger-area, cost-effective, novel, and simpler fabrication techniques for SERS substrates are still in demand.

Atomic layer deposition (ALD) has made dramatic achievements in the past in various fields such as semiconductor development, catalysis, energy, and environmental applications.8 Due to its ALD's unique characteristics of super conformality, large-area uniformity, easy layer thickness and composition control with precision in atomic scale, low cost, and easy scale-up, it is an excellent technique for the bottom-up fabrication of nanoscaled materials and devices if designed rationally. The ALD method has been extensively investigated for the fabrication of SERS-active substrates such as Au or Ag NPs, ultrathin films, conformal coatings of 3D scaffolds, high-aspect-ratio NMNs, core–shell nanostructures, and tunable nanogaps and high-density hot spots on a nanostructured scaffold.9–12 For example, it has been shown by Cao et al.13 that large-scale SERS-active substrates with tunable nanogaps between Au NPs separated by an ultrathin Al2O3 layer by ALD generated highly enhanced multilayer Au/TiO2/Au SERS substrates. A similar work was done by the Cullum group.14 The arrays of Au NPs with high-density hot spots, introduced by ALD-assisted nanogaps, produced strong electromagnetic fields and exhibited strongly enhanced SERS signals. Similarly, it has been shown by Wang et al.15 that using an ALD Al2O3 thin layer, nanogaps between Ag NPs and Au nanogratings could be developed with hot spots of uniform and high intensity for enhanced SERS sensitivity. Furthermore, an ALD ultrathin coating with novel functional dielectric materials provides protection to NMNs against aggregation, oxidation, and surface contaminations.16 ALD Al2O3, TiO2, etc., have proven very effective as spacer/passivation layers for multilayer-enhanced SERS substrates. This is important from the point of view that an agglomeration of NMNs (immobilized on a substrate) from a colloidal sample leads to a damping of their optical properties. Therefore, ultrathin surface coatings of these MNS using adequate materials allow them to retain their desired/tailored functional properties. For example, Ma et al.17 developed TiO2-coated Ag nanorods as highly stable and sensitive SERS substrates with excellent recyclability due to the photocatalytic property of the TiO2 layer.

Our work focused on an innovative all-ALD approach, as shown in Fig. 1, for fabricating potentially superenhanced, ultrasensitive, SERS-active substrates using cheap commercial filter paper. Mesoporous filter membranes such as SiO2 glass fiber paper show fine pore and fiber sizes, a 3D network with a huge effective surface area, excellent temperature resistance, chemical stability, and physical strength, all of which provides an ideal SERS substrate matrix. Many such filter paper sheets are well developed, commercially available, mass-produced, and cheap. In addition, the 3D matrix of the filter paper could allow for both increased surface area and the ability to suppress background signals from potential interferents by taking advantage of the nature-filtering abilities of the underlying support and only allowing access of the analyte of interest to the interlaced SERS nanostructures. We demonstrated that this approach is a very promising one not only for the rational design of SERS substrates with good reproducibility, performance consistency, and reliability, as compared to other existing techniques such as colloidal chemistry, nanopatterning, and PVD/CVD, but also in terms of low cost and scalability for high-volume large-area manufacturing for future commercial SERS applications.

FIG. 1.

Illustration of the concept and design of an SERS substrate fabricated by ALD.

FIG. 1.

Illustration of the concept and design of an SERS substrate fabricated by ALD.

Close modal

Among only a few published ALD Ag processes, for example, either thermal or plasma enhanced,18–20 we have identified the following ALD process using PEt3-Ag(fod) as the Ag precursor and BH3(NHMe2) as the reducing agent.20 It has proved to be a thermal process with a self-limiting performance and very low impurity levels. A catalytic effect was observed that enhanced the growth rate of Ag films such as on a Ru metal surface or on an Ag surface. Isolated Ag NPs were observed, and average sizes were well controlled by cycle numbers. Films were readily obtained on Si with a growth rate per cycle (GPC) of ∼ 0.3Å/cycle in a temperature window ranging from 110 to 120 °C. The GPC increased with the number of cycles from 0.2 to ∼ 0.35Å/cycle after 1500 cycles, indicating that the growing Ag film catalyzes the reaction. The composition of the films was measured as 96.9 at. % Ag, 1.6 at. % oxygen, 0.8 at. % hydrogen, and 0.7 at. % carbon. More recently, Ag NPs were also deposited at 60 °C with a higher GPC at atmospheric pressure by a plasma-enhanced ALD using [(NHC)Ag(hmds)],19 which offers a potential low-cost Ag ALD process for future applications.

A VEECO Savannah 200 ALD system was used for the synthesis of noble metal Ag NPs. A commercial booster was installed to increase the vapor pressure of the Ag precursor, and a flow-through coating fixture was also developed to enhance the diffusion of the Ag precursor inside the 3D porous filter paper. The first step was focused on establishing an Ag film process on flat substrates such as Si wafers with native SiO2 and fused silica glasses, which have a similar composition and surface conditions to those of the commercial filer paper consisting of silica glass fibers. Subsequently, we maximized the LSPR peak intensity and size distribution of Ag NPs as a function of cycle numbers. Finally, we tuned and optimized the ALD process in the so-called static mode to push Ag NPs inside the 3D porous structures of the filter paper in the depth direction to increase “hotspot” volume, which is the key for maximizing SERS signals and sensitivity and for providing accessibility to analytes. The growth temperature was held constant at 120 °C, and the typical one-cycle time was from 30 to 50 s.

A variety of cost-effective, commercially available 3D porous filter membranes/paper made of polymer or inorganic materials were tested and down-selected in this work, and these were PTFE (Teflon), Al2O3, AAO, and SiO2 glass fibers as SERS substrate bases. They feature with a 3D network and large surface area.

These polymer/inorganic filter membranes have been well developed for water purification and water desalination purposes and are commercially available at low cost. (trenchanttextiles.com; hawach.com, and sterlitech.com) There is a wide range of choices for pore size, filter diameter, and thickness that can be optimized for various applications (liquid sample analysis, air analysis, etc.). We systematically measured SERS background signals from these potential SERS substrates in different pore size ranges. For example, PTFE polymer substrates have a low SERS background signal but are not compatible with the ALD process due to a large thermal expansion coefficient when heated up to 100–200 °C. The remaining substrates like silica glass filter paper show not only a low SERS background signal but have also proven mechanically robust and compatible with the ALD process when heated up to 200 °C.

The measurement conditions of SERS spectral signals are an excitation wavelength of 514 nm and power at 250 mW with 2 s exposure, 100 accumulations. A backscatter excitation and collection geometry will be employed, with the incident laser being reflected off a holographic notch filter and focused by a microscope objective onto the sample. The same objective will be used to collect and collimate the backscattered SERS photons to ensure maximum overlap of the excitation and collection volumes of the system and, thus, sensitivity. The collimated SERS signal then passes through the same holographic notch filter for rejection of the Rayleigh scattered light prior to being focused on the slit of a 1/4 m spectrometer equipped with a 1200 groove/mm grating, a 600 groove/mm grating, and a 150 groove/mm grating, to provide flexibility to both high-resolution spectra (1200 groove/mm grating) and large spectral range spectra (150 groove/mm grating) to cover the entire fingerprint and CH stretching region. Detection of the signal is achieved using a liquid nitrogen cooled CCD detector to provide optimal signal-to-noise and minimal electronic background.

Characterization of the LSPR absorption bands of the Ag coated substrates was performed using an Ocean Optics handheld absorption spectrophotometer with a fiber optic reflectance probe. Ag NP morphology, thickness, and composition were measured by using a Hitachi S-4800 Cold-Field-Emission SEM equipped with an Oxford EDX System operated at 15 KV.

When creating a rational design of an SERS substrate, it is critical to predict, tune, and explain LSPR properties such as the dependence of wavelength and intensity on the size, gap, shape, volume fraction/surface coverage, and supporting medium of MNS in order to maximize the LSPR peak intensity, EFs, and sensitivity of SERS substrates. Because scattering from metal NPs can be complex, such as Mie scattering, Rayleigh scattering, and so on, different models are required in different situations depending on NP sizes relative to the excitation wavelength as well as to see whether such particles are embedded in supporting media.

There are three different nanoparticle-optics models depending on particle size: the effective-medium region, the independent-scattering region, and the dependent-scattering region. Coatings that contain particles much smaller than the wavelength of light can be modeled by using effective-medium theories.21,22 In this case, we can conveniently use electrostatics to model the optical properties of the composite. The coating is modeled as a homogeneous “effective medium” with effective optical properties. The commonly used effective-medium theories such as the Maxwell–Garnett theory (MGT) and Bruggeman theory (BT) are based on different microstructure morphologies of composite materials (i.e., the MGT is suitable for a composite structure consisting of isolated metal particles surrounded by a dielectric matrix and BT is suitable for a composite structure consisting of a symmetrical interconnecting mixture of a metal phase and a dielectric phase). We might, therefore, expect the MGT to be applied for low Ag content, whereas BT should be applied for high Ag content. Sheng Ping's effective-medium theory (SPT), also called the symmetrical Maxwell–Garnett theory, is more suitable to describe the observed microstructures. SPT is based on two types of elementary microstructural units (i.e., B phase coated A phase spherical particles describing isolated metal particles A surrounded by dielectric matrix B, and A phase coated B phase spherical particles describing metal particles clustered around the matrix).

Figure 2(a) illustrates different microstructures and corresponding equations to determine effective dielectric constants. In our modeling, as shown in Fig. 2(b), only the MGT is used assuming that isolated Ag NPs are coated on a rough surface. For a polymer filter paper surface, ALD wetting layers such as Al2O3, TiO2, or ZnO must be deposited to change its surface to hydrophilic, which is essential for promoting further ALD dielectric and metal deposition. However, such wetting layers typically show a contiguous but granular surface. Ag NPs deposited on such surfaces can be treated as embedded in a dielectric matrix forming a nanocomposite film with an estimated dielectric constant of interparticle medium as a mixture of air and the wetting material and effective thickness determined by average NP sizes. For an inorganic filter paper surface like silica fibers, a higher refractive index dielectric layer such as Al2O3, TiO2, or ZnO could be also deposited to tune the LSPR peak positions for different SERS wavelength applications based on our modeling.

FIG. 2.

(a) Illustration of various effective-medium theories; (b) an illustrated model to determine the effective dielectric function of metallic NPs on/in fiber paper and modeling of reflectivity by using TFCal, an optical coating simulation tool (spectra.com).

FIG. 2.

(a) Illustration of various effective-medium theories; (b) an illustrated model to determine the effective dielectric function of metallic NPs on/in fiber paper and modeling of reflectivity by using TFCal, an optical coating simulation tool (spectra.com).

Close modal

By fine-tuning Ag nanostructures and controlling the cycle numbers of ALD deposited Ag films, we were able to obtain high-density isolated Ag NPs with average sizes of 3–9 nm depending on the cycle numbers without the formation of agglomerates and continuous Ag films. Figure 3 shows SEM micrographs of Ag NP morphology on Si with different cycles. Figures 4(b) and 4(c) shows x-ray dispersive spectrometry (EDX) from a 3000-cycle sample showing dominated Ag signals and low impurities besides Si substrate signals.

FIG. 3.

SEM micrographs of Ag NP morphology on Si with different cycles: (a) taken after 500 cycles; (b) taken after 1000 cycles; and (c) taken after 1500 cycles.

FIG. 3.

SEM micrographs of Ag NP morphology on Si with different cycles: (a) taken after 500 cycles; (b) taken after 1000 cycles; and (c) taken after 1500 cycles.

Close modal
FIG. 4.

(a) SEM SE image of a 3000-cycle Ag coated sample on Si and (b) and (c) an EDS spectrum on two different areas showing Ag rich NPs.

FIG. 4.

(a) SEM SE image of a 3000-cycle Ag coated sample on Si and (b) and (c) an EDS spectrum on two different areas showing Ag rich NPs.

Close modal

Figure 5 shows measured reflectivity from the same Ag coatings as the above SEM micrographs with different cycle numbers. A broad LSPR peak between 400 and 500 nm is clearly present compared with an uncoated silica glass substrate, with the LSPR peak intensity and wavelength strongly dependent on the cycle number. For example, the 500-cycle sample shows a weak LSPR peak at 415 nm, while the 1500 cycle sample shows a much stronger and shifted LSPR peak at 480 nm. Our modeling result based on the effective-medium theory, as shown in Fig. 6, matches the measured reflectivity well.

FIG. 5.

Measured reflectivity of 500–1500 cycle Ag NPs coated on flat-frosted silica glass substrates.

FIG. 5.

Measured reflectivity of 500–1500 cycle Ag NPs coated on flat-frosted silica glass substrates.

Close modal
FIG. 6.

Shows the modeling results of Ag NPs with different rates of surface coverage (10%–70%) on the same flat glass substrate.

FIG. 6.

Shows the modeling results of Ag NPs with different rates of surface coverage (10%–70%) on the same flat glass substrate.

Close modal

We developed a flow-through glass fiber coating fixture, as shown in Fig. 7, that can force all precursors to pass through porous fibers to coat Ag NPs deeper and more uniformly across a large area and, thus, to maximize hotspot density and volume in the 3D direction, which is critical for increasing accessibility to biochemical analytes and SERS sensitivity. This may also help increase Ag precursor utilization and the surface reaction mostly inside the filter paper rather pass by. The current fixture can coat three filter paper sheets with 47 mm in diameter. However, each holder column can be expanded to accommodate multifilters easily in a Veeco S200 batch reactor in the future for volume production. This flow-through fixture has been proven to enhance deep and uniform coating inside the 3D porous glass fiber filter paper, and the fixture has also been proven to work well, as shown in Fig. 8, which clearly demonstrates a unique and superconformality of Ag NPs not only coated around individual glass fibers but also covering many layers of fibers along the depth direction inside 3D networks. EDS confirms that the NPs are Ag rich, and other EDS peaks are associated either with glass fiber substrates or with Ag precursors such as P, F, and B but with all contents <1%.

FIG. 7.

Flow-through coating fixture developed for enhancing Ag NP deposition deep inside the filter paper.

FIG. 7.

Flow-through coating fixture developed for enhancing Ag NP deposition deep inside the filter paper.

Close modal
FIG. 8.

SEM micrograph from an Ag coated glass fiber filter paper showing excellent conformality.

FIG. 8.

SEM micrograph from an Ag coated glass fiber filter paper showing excellent conformality.

Close modal

Figure 9 shows measured SERS signals from pyridine that were measurable and consistent over a glass fiber paper with 47mm in diameter, 2.7 μm in average pore size, and 600 μm in thickness coated with 1500 cycle Ag NPs. The two strong peaks highlighted in blue originate from the pyridine spontaneous Raman spectrum as a solid reference. The spontaneous Raman spectrum was taken from a pure pyridine solution with 1 W laser power, and the excitation laser wavelength was 514nm. SERS signals were measured from two different areas of spots 2 and 3 on the same filter paper sample to prove spatial uniformity. Note that a small offset of the reference peaks was made on purpose to avoid overlapping. When the pyridine solution passed the Ag coated glass fibers, the two SERS pyridine spectra showed shoulders labeled by arrows as pyridine bands. Our result clearly proves the presence of SERS signals and enhancement for the first time on ALD Ag coated commercial glass fiber filter paper.

FIG. 9.

SERS of pyridine passing through 1500 cycle Ag NP-coated glass fiber filters.

FIG. 9.

SERS of pyridine passing through 1500 cycle Ag NP-coated glass fiber filters.

Close modal

We preferably used Ag NPs to fabricate the SERS substrates not only because Ag substrates outperform comparable Au substrates by about two orders of magnitude in enhancement, but also because Ag precursors have significant cost advantage compared with Au precursors. However, the surfaces of Ag NPs are well known to be not as stable as Au NPs chemically, and, thus, the enhancement could decay significantly within hours of fabrication because of oxidation/sulfurization/tarnishing and a rapid degradation of the LSPR peak intensity. However, one of the unique advantages of an ALD process for a rational design of SERS substrates is that it is possible to apply subnanometer super thin passivation layers such as Al2O3 and TiO2 to prevent oxidation/tarnishing of the Ag NP surface, thus preventing enhancement decay.23 For example, using a few cycles of Al2O3, Ag nanorods wrapped with ultrathin Al2O3 layers exhibited excellent SERS sensitivity and outstanding SERS stability17 with a significantly extended shelf-life time of up to 50 days but still showed no sign of degradation in Raman intensity, even though SERS sensitivity became weakened by 30%–50% after the Al2O3 coating, which was dependent on cycle numbers (1–5). A most recent study24 revealed that Ag dendrites coated by a ten-cycle ZnO exhibited improved SERS sensitivity compared with pristine Ag dendrites. COMSOL theoretical simulations also proved that ultrathin ZnO coating can promote plasmonic coupling in Ag nanogaps. More importantly, this SERS substrate also shows excellent thermal stability at an elevated temperature of 200 °C. Therefore, both the SERS sensitivity and the thermal stability of Ag dendrites can be enhanced via ALD surface protection.

Note that most previous works on these SERS substrates were limited to ALD conformal and contiguous coatings of dielectrics on MNS as spacer layers and surface passivation layers for assisting and improving SERS substrate performance. Our results demonstrate that the ALD technique can also be used to directly generate MNS or metallic NPs as hotspots with strong and wavelength-tunable LSPR signals. Combined with its well-known superconformality in coating 3D network substrates like commercial filter paper, the ALD method could dramatically increase the volume coverage of metallic NPs and maximize the EFs and sensitivity of SERS signals, which are usually limited in traditional PVD/CVD coating techniques. In future work, we will further optimize our ALD process including an Ag precursor delivery kit and improve the flow-through coating fixtures in order to further maximize hotspot density and coverage inside the 3D filter paper. It is expected that SERS EFs of at least 105–106 over a large area will be achieved, with stable enhancements as high as 1012 also possible.

In addition, this method may significantly reduce the complexity and, thus, cost by fabricating SERS substrates in a single-batch ALD coating process in which all dielectric wetting layers, metallic NP deposition layers, and spacer/passivation layers can be conducted in the same coater, thus minimizing metallic NP surface degradation and cross contamination.

We have demonstrated the formation of Ag NPs, the presence of signature LSPR peaks, and the presence of the SERS enhancement effect on cheap commercial 3D porous filter paper all coated by ALD. This work opens up great potential for the possibility that large areas and large quantities of SERS-active substrates can be fabricated by a single ALD process in a cost-effective way.

The work was funded with government support under Contract No. W911SR22C0025 awarded by the U.S. Department of Defense. Special thanks go to Professor Brian Cullum in the Department of Chemistry and Biochemistry at the University of Maryland Baltimore County for granting permission to use his facility and for engaging in discussions and providing suggestions for the work.

The authors have no conflicts to disclose.

Feng Niu: Conceptualization (lead); Data curation (equal); Formal analysis (equal); Funding acquisition (equal); Investigation (equal); Methodology (equal); Supervision (equal); Writing – original draft (equal); Writing – review & editing (equal). Yimin Hu: Funding acquisition (equal); Investigation (supporting); Methodology (supporting); Resources (lead); Supervision (supporting). Stephen LeKarz: Data curation (equal). Wei Lu: Funding acquisition (supporting); Resources (lead); Writing – review & editing (supporting).

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

1.
J.
Langer
,
D. J.
Aberasturi
,
J.
Aizpurua
,
R. A.
Alvare-Puebla
, and
L. M.
Liz-Marzán
,
ACS Nano
14
,
2817
(
2020
).
2.
P. L.
Stiles
,
J. A.
Dieringer
,
N. C.
Shah
, and
R. P.
Van Duyne
,
Annu. Rev. Anal. Chem.
1
,
60
(
2008
).
3.
B.
Cullum
,
J.
Mobley
,
Z.
Chi
,
D.
Stokes
,
G.
Miller
, and
T.
Vo-Dinh
,
Rev. Sci. Instrum.
71
,
1602
(
2000
).
4.
I. R.
Lewis
,
N. W.
Daniel
, Jr.
,
N. C.
Chaffin
,
P. R.
Griffiths
, and
M. W.
Tungol
,
Spectrochim. Acta A
51
,
1985
(
1995
).
5.
N. T.
Kawai
and
J. A.
Janni
,
Spectroscopy
15
,
36
(
2000
).
6.
D. S.
Moore
and
R. J.
Scharff
,
Anal. Bioanal. Chem.
393
,
1571
(
2009
).
7.
W. Y.
Hu
,
L.
Xia
,
Y. F.
Hu
, and
G.
Li
,
Microchem. J.
172A
,
106908
(
2022
).
9.
J.
Prakash
,
H. C.
Swart
,
G.
Zhang
, and
S.
Suna
,
J. Mater. Chem. C
7
,
1447
(
2019
).
10.
A. V.
Whitney
,
J. W.
Elam
,
P. C.
Stair
, and
R. P.
Van Duyne
,
J. Phys. Chem. C
111
,
16827
(
2007
).
11.
M.
Liu
,
L.
Sun
,
C. W.
Cheng
,
H. L.
Hu
,
Z. X.
Shen
, and
H. J.
Fan
,
Nanoscale
3
,
3627
(
2011
).
12.
W. J.
Xu
,
Z. Y.
Chen
,
N.
Chen
,
H.
Zhang
,
S. P.
Liu
,
X. M.
Hu
,
J. X.
Wen
, and
T. Y.
Wang
,
Sensors
17
,
467
(
2017
).
13.
Y. Q.
Cao
,
K.
Qin
,
L.
Zhu
,
X.
Qian
,
X. J.
Zhang
,
D.
Wu
, and
A. D.
Li
,
Sci. Rep.
7
,
5161
(
2017
).
14.
P.
Strobbia
,
E. R.
Languirand
, and
B.
Cullum
,
Opt. Eng.
54
,
100902
(
2015
).
15.
Y.
Wang
et al,
Microelectron. Eng.
172
,
1
(
2017
).
16.
P.
Strobbia
,
A. J.
Henegar
,
T.
Gougousi
, and
B. M.
Cullum
,
Appl. Spectrosc.
70
,
1375
(
2016
).
17.
L. W.
Ma
,
Y.
Huang
,
M. J.
Hou
,
Z.
Xie
, and
Z. J.
Zhang
,
Sci. Rep.
5
,
15442
(
2015
).
18.
S.
Wack
,
P. L.
Popa
,
N.
Adjeroud
,
J.
Guillot
,
B.
Rita Pistillo
, and
R.
Leturcq
,
J. Phys. Chem. C
123
,
27196
(
2019
).
19.
T.
Hasselmann
et al,
Adv. Mater. Technol.
8
,
2200796
(
2023
).
20.
M.
Mäkelä
,
T.
Hatanpää
,
K.
Mizohata
,
K.
Meinander
,
J.
Niinistö
,
J.
Räisänen
,
M.
Ritala
, and
M.
Leskelä
,
Chem. Mater.
29
,
2040
(
2017
).
21.
C. F.
Bohren
and
D. R.
Huffman
,
Absorption and Scattering of Light by Small Particles
(
Wiley-Interscience Publication
,
New York
,
1983
).
22.
F.
Niu
, “
Functional nanocomposite thin films by co-sputtering
,” in
Functional Thin Films and Functional Materials
, edited by
D. L.
Shi
(
Springer-Verlag
,
Berlin
,
2001
).
23.
J. F.
John
,
S.
Mahurin
,
S.
Daib
, and
M. J.
Sepaniaka
,
J. Raman Spectrosc.
41
,
4
(
2010
).
24.
X. X.
Wang
et al,
Appl. Surf. Sci.
611
,
155626
(
2023
).