Embedded nanopattern for selectively suppressed thermal conductivity and enhanced transparency in a transparent conducting oxide film

Transparent conductive oxide (TCO) thin films are cornerstones in many optoelectronic applications including displays, photovoltaics and touchscreens. In these devices, thin films with simultaneous high optical trans-parency and electrical conductivity are needed. Ideally, heat generated during normal device operation must ideally be compensated for to achieve optimum functionality. One possible way to address the thermal management problem is adding thermoelectric (TE) properties to TCO films. However, improving TE properties while maintaining optimal electrical conductivity and optical transparency is challenging: thermal and electrical transport properties are deeply intertwined. Here, we demonstrate an approach allowing for independent optimization of optical transparency, electrical conductivity and thermal conductivity. An embedded nanopattern structure is filled with indium tin oxide (ITO) and sandwiched between two ITO layers. The resulting triple-layered structure exhibits reduced thermal conductivity and excellent electrical conductivity. This is made possible by electron channels in the embedded ITO nanopattern that electrically connect top and bottom layers, while at the same time limiting phonon-mediated heat conduction. The filling fraction and thickness of the nanopattern are adjusted to improve optical transmission, achieving transparency higher than bare ITO film. The result is a transparent TCO triple layer film with simultaneous high TCO and thermoelectric figures of merit.


Introduction
Thermal fatigue is the main cause of failures and poor performances of electronic apparatus such as displays, smart windows, or photovoltaic devices [1][2][3][4].Consequently, heat management is very important for these devices.Strategies for heat management can rely on, e.g., passive (dissipation of heat) or active (Peltier cooling) approaches [1,[4][5][6].Regarding the latter, transparent conducting oxides (TCOs) with good thermoelectric (TE) properties and optical transparency would be highly desirable for heat management.In addition, TE energy harvesting devices have been highlighted for next generation renewable energy systems because they can be used to harness waste energy otherwise dissipated into the environment [7][8][9][10][11][12].For these reasons, thin film TE materials have been studied for decades [13][14][15][16].Research on thin film TE materials has hitherto mainly focused on telluride alloys due to their high performance.However, telluride alloys are toxic, expensive and exhibit low chemical stability [17][18][19][20].In contrast, oxide thin film TE materials based on TCOs such as ITO have high chemical stability, are non-toxic, and involve relatively low manufacturing costs [21][22][23][24][25].
Indium tin oxide (ITO) has been intensively studied in the past, mainly as a TCO material [26], but also regarding its TE properties [27,28].Despite the outstanding optical and electrical properties of ITO, it is difficult to obtain ITO-based films that exhibit adequate electronic, optical and thermal properties simultaneously.This problem can be summarised in the optimization of the thermoelectric figure of merit (ZT), defined as [29].
where σ, S, T, and κ tot are the electrical conductivity, Seebeck coefficient, absolute temperature, and total thermal conductivity, respectively.
On the other hand, σ is given by: where μ, n and e are the carrier mobility, carrier concentration and electron charge.S is given by the Mott formula [30], where k B , h, and m* are the Boltzmann constant, Planck constant, and carrier effective mass, respectively.As evident from Eqs. ( 3) and ( 4), ZT exhibits a trade-off relationship with respect to n, viz.σ ∝ n and S ∝ n -2/3 , yielding ZT ∝ n -1/3 , that is, ZT deteriorates (weakly) as n increases, i.e. if the electrical conductivity increases.Most studies on the applicability of TCOs as front electrodes in display and OLED devices focus on highly crystalline TCO thin films exhibiting optimal σ values due to a high n.Note that n cannot be too high.The plasma frequency must remain in the near-infrared region so that the optical transmittance in the visible is preserved [31][32][33].
Notwithstanding their excellent σ, this situation is not beneficial from a TE point of view.This is because, as discussed above, larger n values result in lower ZT.We have previously reported that an ITO thin film with an amorphous structure can exhibit fairly high σ without an accompanying increase of n and still have low κ l because of high carrier mobility due to non-directional In 5 s bonding with oxygen [27,28].However, amorphous thin films have relatively poor electrical conductivity compared to crystalline oxide TE materials, which is an inherent limitation of amorphous metal oxides [26,34,35].
Acknowledging the difficulties of achieving simultaneously good TE performance, good electrical conductivity and visible transparency, the challenge is to develop a transparent and conductive coating where the TE properties (represented by ZT) can be enhanced by decoupling the electronic and lattice contributions to the thermal conductivity.Here, we show that this can be achieved by employing a three-layered ITO/ nanopattern/ITO sandwich structure.The ITO layers were prepared by magnetron sputtering and the intermediate dot-or rod nanopatterned layer was fabricated using block copolymer (BCP) self-assembly.The nanopatterned layer, embedded between two ITO layers, effectively suppresses κ l without significantly deteriorating the electrical properties of the complete film structure, resulting in an improved ZT.At the same time, the optical transmittance is also improved by judicious interference matching using appropriate height and filling factor of the embedded nanopattern.
The transparent electrode performance can be assessed using Haacke's figure of merit (ϕ TC ) defined as [36]: where T is the average optical transmittance between 500 nm and 600 nm and R sheet is the sheet resistance.The three-layer ITO/nanopattern/ITO films reported here are shown to exhibit ϕ TC value superior to previous reports, while at the same time achieving high ZT values.Specifically, in this study, we show how nanostructures sandwiched between two amorphous ITO thin films can be used to decouple the thermal and electrical properties.High overall electrical conductivity is achieved due to high mobility and electrical connectivity of the thin, amorphous capping ITO films [37].At the same time, the optical transmittance in the visible region of the ITO/nanopattern/ITO coating is larger than that of the single ITO coating of equivalent thickness, which, as detailed below, is due to both the small ITO thicknesses and judicious choice of nanopatterning creating beneficial interference conditions.We suggest that our embedded-nanopattern approach could offer a new design principle for TE-based thermal management of optoelectronic devices employing transparent conducting films.

Nanopattern embedding procedure
The detailed procedure for the synthesis of rod-nanopattern and embedding process is schematically represented in Fig. 1.First, a 70 nm ITO thin film is deposited onto a clean glass substrate by magnetron sputtering (Fig. 1b).Then, poly(4-vinylpyridine), P4VP is spin-coated on ITO and exposed to UV to for hardening the polymer structure (Fig. 1c).Polystyrene-block-dimethylsiloxane (PS-b-PDMS) is subsequently spincoated on the P4VP/ITO structure forming the block co-polymer, BCP (Fig. 1d, e).The PS-b-PDMS/P4VP/ITO is then exposed to toluene vapor to induce self-assembly of PS-b-PDMS (Fig. 1g) forming PDMS nanoarrays.Thereafter, inductively coupled plasma reactive ion etching (ICP-RIE), using and CH 4 and oxygen gases, is performed, which oxidizes the Si moieties in the self-assembled PDMS into SiO x , which serves as an etching mask (Fig. 1g), while PS is removed, forming the void space between the nanostructures (Fig. 1h).The protected P4VP layer below the SiO x etching mask form rods.In the case of the nanodot pattern, the P4VP process shown in Fig. 1c is step is omitted.The process is completed by the deposition of a third capping ITO thin film layer by magnetron sputtering that fills the void space between the rods or dots.As a result, triple layered "sandwich" films are obtained, either ITO/ nanodots/ITO, or ITO/nanorods/ITO (Fig. 1i).
In the ITO/nanopattern/ITO configuration, the nanopattern (either rods or dots) acts as a thermal conduction barrier yielding reduced κ l , and consequently low κ tot .At the same time, the nanopattern does not deteriorate the electrical properties and allows the formation of ITO electron channels connecting the two ITO layers.We note that the bottom and top ITO layers are amorphous due to their small thicknesses (See Fig. S1 a and b).

Electron channels and thermal conduction barrier
Fig. 2 shows schematic representations, as well as FE-TEM images and TEM-EDS mapping images corresponding to ITO/dot/ITO (panels a to e) and ITO/rod/ITO (panels f to j) multilayers.The dot pattern has a 20 nm-pitch and height of 10 nm and the capping ITO films are determined to be about 70 nm thick from FE-TEM (see Fig. 2b).Well-ordered dot patterns placed between the ITO layers are observed by crosssectional FE-TEM in Fig. 2b and c, with clear gaps connecting the top and bottom ITO layers.As shown in the element-specific TEM-EDS mappings (Fig. 2 d and e), which show the same cross-sectional region of the sample shown in Fig. 2c, In and Sn atoms fill these gaps, i.e., successful formation of electron channels is achieved.Analogously, in the case of ITO/rod/ITO multilayer structures, a well-ordered rod pattern between ITO layers is observed, Fig. 2g and h.In this case, the pitch and height are 20 nm and 35 nm, respectively, and the capping ITO films are again about 70 nm.Despite the deep gaps left by the rod pattern observed in Fig. 2h, In and Sn atoms fill the space between the rods, as shown by TEM-EDS mapping of the interface, Fig. 2i and j.
Fig. 2 confirms that dot-and rod-nanostructured layers are successfully embedded between the ITO layers, but also that the top and bottom ITO layers are inter-connected through metal-filled channels surrounding the nanostructures.The nanostructures (dot and rod) formed between ITO layers have the same size and pitch when grown onto Si substrates (see Fig. S2, panels a to c).
Photographs of a single, amorphous ITO coating with thickness about 100 nm, as well as multi-layered ITO(70 nm)/dot(10 nm)/ITO (70 nm) and ITO(70 nm)/rod(35 nm)/ITO(70 nm) structures are presented in Fig. 3a.Fig. 3b shows the transmittances corresponding to these coatings, whereas the inset shows the average transmittance in the wavelength interval between 500 and 600 nm.While the ITO/dot/ITO sample shows slightly higher transmittance compared to the single ITO sample, the ITO/rod/ITO structure displays an outstanding transmittance value (~93.7%) at 550 nm, exceeding that of bare glass.This is qualitatively confirmed by the photograph image in Fig. 3a, which displays a different visual appearance of the ITO/rod/ITO structure compared to the other structures.This high transmittance is caused by interference effects in the ITO/nanopattern/ITO structure, where the embedded nanopattern layer acts as an adjustable optical interference spacer layer, as confirmed by modelling of the optical properties (dashed curves in Fig. 3b).The ITO layer was modelled using three elements: a dielectric background ε ∞ ≈ 4.0, a Drude oscillator (determined by free charge carrier concentration n and electrical resistivity ρ) [38], and an OJL oscillator [39] for modelling the fundamental band gap absorption.The use of an OJL oscillator is justified due to the amorphous structure of ITO [39].In the OJL model, the band gap energy was set to 2.7 eV, i.e., at the onset of the fundamental absorption of ITO [40].The good agreement of our model with the experimental data suggests that the fundamental absorption of the multilayer structure is not affected by the insertion of the nanopattern, and is completely dominated by ITO.This is not surprising, since the fundamental absorption of P4VP is expected at 4.7 eV [41], located outside the rage of study.The refractive index of the nanostructures (dots and rods, composed of SiO X and P4VP) converged to a constant value n = 1.52 in the wavelength region of interest in the fitting process, in very good agreement with previous studies [42].The resulting embedded nanopatterned layer is considered   as a mix of ITO/dot or ITO/rod with an ITO volume fraction modelled by the filling factor, ff, using the Bruggeman approximation [43].The glass substrate is approximated by a constant refractive index n = 1.52.The fitting of the model to the experimental results was performed using the downhill simplex method implemented in the commercial software SCOUT [44].An excellent fitting to the experimental transmittance of the single ITO layer is achieved with Drude parameters corresponding to n = 4.20 × 10 20 cm -3 and ρ = 2.56 × 10 -4 Ωcm, and film thickness d = 91.0nm (Fig. 3b).The Drude parameters for ITO are in reasonably agreement with the parameters obtained by Hall measurements (Fig. 4a).For the ITO/dot/ITO structure, the Drude parameters obtained from model fitting were determined to be n = 3.6 × 10 20 cm -3 and ρ = 3.18 × 10 -4 Ωcm.The top and bottom ITO layers were found to be 42 and 46 nm, respectively.The embedded nanopattern layer was here modelled as a Bruggeman ITO/dot mix, where the thickness was 10 nm and the filling factor of the ITO component was 38%.Analogously for the ITO/rod/ITO arrangement, the best-fit parameters were found to be: n = 2.6 × 10 20 cm -3 and ρ = 4.6 × 10 -4 Ω cm, with top and bottom ITO thicknesses of 46 and 55 nm, respectively, and intermediate nanopatterned thickness 39 nm, and ITO filling factor of 20%.The obtained filling factors agree well with the estimate filling fractions obtained from SEM images in Figs.S2d and e using the open-software Gwyddion [45].From the projected SEM images the corresponding filling factor are predicted to be 37% for the nanodots and 24% for the nanorods, respectively.
The electrical properties, ρ, n and μ corresponding to single ITO, ITO/rod/ITO and ITO/dot/ITO multilayers obtained by Hall measurements are presented in Fig. 4a.The ITO/dot/ITO multilayer shows only marginally higher ρ compared to a single 100 nm thick amorphous ITO layer, while the ITO/rod/ITO multilayer shows about 4% higher resistivity.In the absence of electron channels, i.e., if the nanopattern provided complete electrical insulation between the upper and lower ITO layers, then the ITO/rod/ITO multilayer stack, of thicknesses 70 nm/35 nm/70 nm would be expected to exhibit a ρ value closer to the one measured in a about 50 nm-thick amorphous ITO film (Fig. S3).
However, ρ measured in the ITO/rod/ITO samples is very similar to a 100 nm-thick ITO layer, which again points to the formation of electron channels through the embedded nanopattern.We note that would be possible to improve the conductivity of the ITO/nanopattern/ITO multilayer even further by replacing the amorphous ITO films (or another type of TCO material) with corresponding crystalline films employing alternative routes for low temperature thin film TCO deposition.
The Seebeck coefficient, S, was calculated using Eq. ( 4) from the experimental n values presented in Fig. 4a.S and n are shown in Fig. 4b for the different sets of samples.The multilayer film is well-described as a degenerate semiconductor and follows the expected dependence, S ∝ n -2/3 , predicted by Eq. ( 4) (Fig. S4).In contrast, analysing the contribution to the conductivity, the ITO/dot/ITO and ITO/rod/ITO multilayers show low κ tot values compared to a single ITO, Fig. 4c.
Considering that the embedded nanopattern barely affects the overall electrical properties of the ITO/dot/ITO and ITO/rod/ITO layer stacks (Fig. 4a), and that κ tot can be separated in contributions from free conduction electrons κ e and lattice vibrations κ l (Eq.( 2)), we conjecture that the reduction observed in κ tot in the ITO/rod/ITO and ITO/dot/ITO multilayer films must be caused by a decrease of the κ l component.This is corroborated by time-domain thermo-reflectance measurements.Fig. 4c shows the measured κ tot, κ l and κ e for the different films.It is evident that κ tot is reduced in the multilayer films, and most for the ITO/ rod/ITO multilayer film (Fig. 4c).
An empirical relation between the total thermal conductivity, κ tot , and the height of the nanopattern (length of electron channels), L, can be deduced from the data reported in Fig. 4c, viz.
Here κ e, = 1.38 W m -1 K -1 is about the same for all types of ITO films, plain or nanopatterned sandwich films (Fig. 4c), and κ l,bulk, = 0.656 W m -1 K -1 is the measured value for the flat ITO film ("Single ITO" in Fig. 4c).The fitting parameter a is determined to be a = 0.030 nm from a linear least-square fit (Fig. S5).Eq. ( 6) explicitly shows that κ tot, decreases as a function of length, L, of the electron channel and reduces to Eq. ( 2) when L = 0.An interesting limiting case can be deduced from Eq. ( 6) when aL = 1.Then, κ tot, = κ e , and a critical length L = 1/a where the thermal conductivity equals the electronic contribution can be identified, where the nanostructure behaves as if lattice vibrations did not exist, i.e. like a perfect metal.In our case when L = 1/0.030≈ 33 nm, i.e. close to the nanorod length.We emphasize however that this is an empirical relation, which should be analyzed more thoroughly, but nevertheless can be used an approximate expression for designing embedded electron nanochannels in similar structures.

Thermoelectric and TCO figure of merit
The TE and TCO properties, ZT and ϕ TC were evaluated using Eqs.( 1) and ( 5), respectively [36].Fig. 5 shows ϕ TC and ZT at a temperature of 300 K. Fig. 5c shows a comparison of ϕ TC and ZT obtained in this work with previously reported data.
Fig. 5a shows experimental values of ZT for a single ITO layer and for ITO/dot/ITO and ITO/rod/ITO multilayers, based on experimental values of κ tot , n and μ.Additionally, Fig. 5a shows calculated ZT values as a function of n and μ assuming a κ tot value corresponding to a single ITO layer of the same thickness (red circles).It is evident that the multilayer samples exhibit improved ZT when compared to the equivalent ITO single layer.This improvement is larger in the case of ITO/rod/ITO than for ITO/dot/ITO multilayers.The increase of the ZT values is due to the decreases value of κ l , which decreases about 80% in the ITO/rod/ITO film respect to the single ITO film of same thickness (Fig. 4c), while the mobility only increases about 7% (Fig. 4a).
Fig. 5b shows a comparison of ZT and ϕ TC for single ITO, ITO/dot/ ITO and ITO/rod/ITO.Since the interference effects provided by the embedded nanopattern result in higher visible transmittance in the multilayers than in the equivalent ITO monolayer (Fig. 3b), the ITO/ dot/ITO and ITO/rod/ITO multilayers present not only superior ZT, but also superior ϕ TC .

Conclusions
Transparent and conductive ITO/dot/ITO and ITO/rod/ITO multilayers were obtained by embedding dot-or rod-nanostructured layers, prepared by block copolymer self-assembly methods, between two ITO sputtered films.The embedded nanostructures reduce the thermal conductivity by reducing the lattice contribution to the overall thermal conductivity, while preserving good electrical properties thanks to the formation of narrow electron channels between the top and bottom ITO layers.In addition, the refractive index of the nanostructures (n = 1.52), filling factor of ITO in the nanopattern, and thickness of the rodstructure can be adjusted to achieve beneficial interference effects that increases the optical transmittance of the multilayer stack in the visible region.The combined properties can be optimized independently with large degree of freedom without imparting adversely on the overall thermoelectric and TCO figure of merits.In particular, we have here shown that ITO/rod/ITO multilayer films result in ZT and ϕ TC values of 0.033 and 17.5 mΩ -1 , respectively, yielding superior combined TCO and thermoelectric properties compared to state-of-the-art.The embedded nanopatterning concept presented in this work demonstrates a TCO film structure that allows for independent optimization of optical transparency, electron and thermal conductivity of TCO coatings, which can be used to significantly increase their thermoelectric properties reaching practically applicable values.The length of the rods, pitch and diameter, and also in-filling material (here ITO), can be optimized for transport and optical properties, while still selectively suppressing thermal conduction.A straightforward implementation of the present work would be to change one of the sandwiching ITO thin films to a p-type semiconducting to realize a working TE device.We envisage that the proposed concept can be used for heat management of transparent displays and solar energy systems.

Nanopattern formation process
A PS-b-PDMS BCP (Polymer Source Inc., Canada) and P4VP homopolymer (Polymer Source Inc., Canada) were used to prepare the embedded nanopattern structures.P4VP (60 kg mol -1 ) and PS-b-PDMS (56 kg mol -1 ) were dissolved in dimethylformamide (P4VP 3 wt%) and toluene solvent (BCP 1 wt%), respectively.The preparation steps for preparing the nanopattern embedded ITO structure are schematically illustrated in Fig. 1 (panels a to i).The glass substrate was cleaned using acetone, ethanol, and finally isopropyl alcohol in an ultrasonic bath for 15 min, respectively (Fig. 1a).The ITO thin films were then sputtered on the cleaned glass substrate to a thickness of 700 µm (Fig. 1b).P4VP was spin-coated on the ITO thin film, and then the sample was exposed to UV (254 nm) for 3 h to crosslink the polymer layer.Subsequently, the BCP was spin-coated on the P4VP layer, resulting in a multilayer BCP/P4VP/ ITO/glass (Fig. 1e).BCP/P4VP/ITO/glass was annealed by toluene solvent vapour at a temperature of 65 • C. The BCP/P4VP/ITO/glass was then dry etched using ICP-RIE system with CF 4 plasma (25 s) and O 2 plasma (45 s) employing 50 W plasma formation power and 10 W substrate power, respectively.Dry etching causes the self-assembled Si ions in PDMS to be oxidized to SiO X which serve as an etching mask during the dry-etching process (Fig. 1g).Finally, ITO was sputtered on the BCP/ P4VP/ITO/glass structure, resulting in the following configuration: ITO/nanopattern/ITO/glass. For the ITO/dot/ITO sample, the P4VP process (Fig. 1c) is omitted.

Thin film characterization
Electrical characterization of the different samples was carried out in a Hall effect measurements system (HMS2000, ECOPIA).
The optical transmittance of the different samples was measured in a Perkin-Elmer Lambda 900 spectrophotometer equipped with an integrating sphere.Cross sectional images and elemental composition mapping of the different samples were obtained by field-emission transmission electron microscopy (FE-TEM) in a JEOL JEM-2100 F microscope operated at 200 kV.Additionally, micrographs of the nanopatterns were obtained by field-emission scanning electron microscopy (FE-SEM) in a ZEISS LEO1550.
Thermal conductivity values were determined using time-domain thermo-reflectance (TDTR) method.A femto-second pulsed beam working at 80 MHz was used as pump and probe beam in the TDTR measurement.The beam was split into two oppositely polarized beams by a polarized beam splitter.One of the split beams was modulated at 9.8 MHz and used as a pump beam, and the other beam, at the original frequency, was used as a probe beam.Aluminium with thickness of 85 nm was coated on the sample and a Ti:sapphire laser was used for TDTR measurements [57].The contribution by the free electrons κ e was calculated using the carrier density value, n, from Hall measurements and applying the Wiedemann-Franz law: κ e = L 0 T/ρ, where L 0 is the Lorenz number.Finally, κ l values were obtained from κ l = κ tot − κ e .
Jang-Hee Yoon is a director of Busan center at Korea Basic Science Institute.He received his Ph.D. degree in chemistry.He is an expert in the field of surface analysis.Yoon's current scientific interests include electro-analysis of heavy metals, electrolysis of hazardous materials, solar cells of conducting polymer and functional thin film coatings.
Yunju Choi is currently working as a senior researcher at the Korea Basic Science Institute (KBSI).She operates a fieldemission transmission electron microscope and is conducting related research.Her research activities focus on carbon nanotubes, graphene, other two-dimensional materials, energy storage materials, photocatalytic semiconducting materials, and bulk carbon materials.
Pungkeun Song is professor in Department of Materials Science and Engineering, Pusan National University, South Korea.He received his Ph.D. degree from University of Tokyo in 1999, in the field of Applied Chemistry.He is expertise experience and knowledge on transparent conductive oxide fabricated by magnetron sputtering system, and also has abundant knowledge on sputtering target.His research filed includes design of plasma process and physical properties of thin films, nondegradable organic compounds treatment using insoluble electrodes.

Fig. 1 .
Fig. 1.Schematic procedure (steps (a) -(i)) to synthesize nanopattern embedded ITO structure to fabricate TE-TCO films.The pattern size and pitch are 20 nm.The height is 5 nm and 35 nm for dot and rod structure, respectively.To fabricate nanodot pattern, step (c) is excluded.

Fig. 2 .
Fig. 2. Schematic images (a; f), FE-TEM (b, c; g, h) and EDS mapping (d, e; i, j) images of nanopattern embedded ITO structures comprised of dots: (a)-(e), and rods: (f)-(j).Cross-sections of the nanopattern between ITO layer are shown with higher magnification.The electron channel formed between the upper and bottom ITO layers are clearly observed in the element specific EDS mappings.

Fig. 3 .
Fig. 3. (a) Photographs of single ITO, dot ITO and rod ITO multilayer films, exhibiting distinctly different colors.(b) Transmittance spectra of single layer ITO film (ITO/glass), and ITO multilayer films (ITO/dot/ITO/glass and ITO/rod/ ITO/glass).The rod ITO shows the highest transmittance, even higher than bare glass, due to optical interference in the embedded nanostructure.The dashed curves show results from effective medium calculations.Inset in (b) shows the average transmittance between 500 and 600 nm for the different films.

Fig. 4 .
Fig. 4. (a) The resistivity (carrier density and mobility) slightly decreases (increases) as a function of increasing nanopattern height going from single ITO, dot ITO to rod ITO.(b) Seebeck coefficient increases inversely to carrier density.(c) Measured total thermal conductivity, electron conductivity, and lattice thermal conductivity as a function of nanopattern height.Electron thermal conductivity shows similar values, while the lattice thermal conductivity is dramatically suppressed for the rod ITO nanopattern.

Fig. 5 .
Fig. 5. (a) ZT determined from the measured total thermal conductivity ( ), and the predicted ZT values using the lattice thermal conductivity for single ITO for all samples, i.e. in absence of nanopatterning ( ; dashed line shows the least-square linear fit).The deviation from linear fit is due to the suppressed lattice thermal conductivity in the nanopattern sandwich-structures.(b) Thermoelectric (ZT) and transparent electrode (FoM, T 10 /R sheet ) figure of merit for the bare ITO and nanopatterned TE films.The samples have similar electrical properties, but show significant increase of ZT and FoM for the nanopatterned TE films.(c) Transparent conducting oxide FoM values plotted as a function of ZT values using recently reported data on TE performance for transparent electrode materials (references indicated in brackets adjacent to data points).The polymer embedded ITO shows excellent properties for both ZT and FoM.Red circles ( ) denote n-type semiconductors, and blue triangles ( ) show p-type.
Österlund is Professor in Solid State Physics at Uppsala University.He obtained a PhD in physics from Chalmers University of Technology 1997 with specialization in surface science.He performed a postdoc at Aarhus University in Flemming Besenbacher's group between 1997 and 2000, after which he joined the Centre for Catalysis at Chalmers in 2000.Österlund's group performs fundamental research on photon and electron stimulated surface reactions, materials optics, nanopatterning, and applied research in photocatalysis, solid state gas sensors, self-cleaning and antimicrobial surfaces with applications in building and display technology, air and water treatment, and medical technology.