Roll-to-roll sputtered and patterned Cu2−xO/Cu/Cu2−xO multilayer grid electrode for flexible smart windows

We fabricated cost-effective Cu2−xO/Cu/Cu2−xO multilayer grid electrodes using roll-to-roll (RTR) sputtering and patterning processes for use as transparent and flexible electrodes in flexible smart windows. To optimize the patterned Cu2−xO/Cu/Cu2−xO multilayer grid, the electrical and optical properties of the Cu2−xO/Cu/Cu2−xO multilayer grid electrodes were investigated as a function of grid width and pitch, which directly influence the filling factor of the grid. At the optimized grid width of 16 and pitch of 600 μm, the Cu2−xO/Cu/Cu2−xO multilayer grid had a sheet resistance of 7.17 Ohm per square and an optical transmittance of 87.6%. In addition, the mechanical properties of the optimized Cu2−xO/Cu/Cu2−xO multilayer grid electrode was compared to those of brittle ITO electrodes to demonstrate its outstanding flexibility. To show the potential of the Cu2−xO/Cu/Cu2−xO multilayer grid for smart windows, we fabricated a flexible and transparent thin film heater (TFH) and a flexible electrochromic (EC) device, which are key components of smart windows. The low saturation voltage of the Cu2−xO/Cu/Cu2−xO grid-based TFH and the fast on–off performance of the Cu2−xO/Cu/Cu2−xO grid-based EC device indicates that the RTR-processed Cu2−xO/Cu/Cu2−xO multilayer grid is promising as a low-cost and large-area flexible transparent electrode for high-performance smart windows.


Introduction
Multi-functional smart windows equipped with transparent displays, transparent heaters, electrochromic devices, and energy-harvesting devices have attracted signicant attention as next-generation exterior materials for buildings and automobiles. [1][2][3][4][5] Unlike conventional windows, which simply pass visible light into the building or automobile, smart windows provide several convenient and smart functions, such as information displays, energy harvesting, self-heating and cleaning, transmittance control, and indoor temperature and light control. In several components of smart windows, exible and transparent thin lm heaters (TFHs) can remove frost or ice by heating the window and exible electrochromic (EC) devices, which can adjust the indoor brightness by controlling the transmittance of the window. [5][6][7] The performance, stability, and fabrication cost of TFHs and EC devices are critically dependent on the electrical, optical, and mechanical properties of transparent and exible electrodes (TFEs). In addition, the fabrication cost of TFHs and EC devices for low-cost and largearea smart windows is closely related to the cost of the TFE materials and coating processes. Therefore, the development of highly transparent, conductive, exible, and cost-effective TFE materials and processing methods is imperative for massproducing TFHs and EC devices. A common transparent conducting electrode (TCE) material, Sn-doped In 2 O 3 (ITO) lms coated on polyethylene terephthalate (PET) substrates are typically employed as TFEs due to their low sheet resistance, high optical transmittance, well-known processing technology, and ease of use for large-area coatings. [8][9][10][11][12] However, sputtered ITO lms are critically limited as high-quality and cost-effective TFEs due to the relatively high sheet resistance and poor mechanical properties of ITO/PET lms as well as the high cost of indium. 13 Several TCE materials fabricated by vacuum-based or solution-based coating processes have been extensively reported as replacements for high-cost ITO lms. [14][15][16][17][18][19] Among these, sputtered oxide-metal-oxide (OMO) multilayer lms and printed metal (Ag, Cu) grid electrodes are considered to be promising replacements. However, sputtered OMO electrodes are still composed of high-cost indium-based oxide and Ag interlayers. In the case of the printed metal grid electrodes, the patterned grid is shiny due to the high reection on the surface. To solve both of these problems, it is necessary to develop a cost-effective OMO-based multilayer grid electrode that combines the merits of the OMO and the metal grid.
In this work, we developed a grid-patterned OMO multilayer electrode with a Cu 2Àx O/Cu/Cu 2Àx O structure using a lab-scale roll-to-roll (RTR) sputtering system to replace high-cost ITO electrodes. To optimize the grid width and pitch of the Cu 2Àx O/ Cu/Cu 2Àx O multilayer grid, we investigated the electrical and optical properties of the Cu 2Àx O/Cu/Cu 2Àx O multilayer grid as a function of grid width and pitch. In addition, the mechanical properties of the Cu 2Àx O/Cu/Cu 2Àx O multilayer as a substitute for a typical sputtered ITO electrode were comprehensively investigated using lab-designed outer and inner bending tests. Furthermore, we used our patterned OMO multilayer grid electrode to fabricate exible TFHs and EC devices to demonstrate their feasibility for application in next-generation exible smart windows.  Fig. 1(b) shows the roll-to-roll patterning process for the Cu 2Àx O/Cu/Cu 2Àx O multilayer grid. The patterned Cu 2Àx O/Cu/ Cu 2Àx O grid had high optical transmittance due to the narrow grid width and wide pitch, as shown in Fig. 1(c), unlike the asdeposited Cu 2Àx O/Cu/Cu 2Àx O lms with dark brown color. The electrical and optical properties of the Cu 2Àx O/Cu/Cu 2Àx O multilayer grid electrode were investigated as a function of grid width and pitch using Hall measurements (HL5500PC, Accent Optical Technology) and a UV/visible spectrometer (UV 540, Unicam). In addition, the composition and binding energy of the reactive sputtered Cu 2Àx O lm in the multilayer grid electrode were analyzed by using X-ray photoelectron spectroscopy (XPS: ESCALAB250). The mechanical properties of the gridpatterned Cu 2Àx O/Cu/Cu 2Àx O multilayer were evaluated using a specially designed inner and outer bending system. In addition, a dynamic fatigue bending test was performed using a labdesigned cyclic bending system operating at 0.5 Hz for 10 000 cycles.

Fabrication and evaluation of thin lm heaters and electrochromic devices
To demonstrate the potential of the grid-patterned Cu 2Àx O/ Cu/Cu 2Àx O multilayer, we fabricated exible TFHs and EC devices on an optimized Cu 2Àx O/Cu/Cu 2Àx O multilayer grid electrode. Flexible TFHs with size of 25 Â 25 mm 2 with twoterminal side contacts was fabricated on the Cu 2Àx O/Cu/ Cu 2Àx O multilayer grid (Fig. S2 †). A 200 nm-thick Ag side contact electrode was sputtered onto the edge of the grid, and a DC voltage was supplied by a power supply (OPS 3010, ODA Technologies) to the grid-based TFHs through the Ag contact electrode at the lm edge. The temperature of the TFHs was measured using a thermocouple mounted on their surface and an infrared (IR) thermal imager (A35sc, FLIR). Fig. S3 † shows a picture of the temperature measurement system, including the thermocouple and IR thermal imager. A commercial ITO electrode with sheet resistance of 38.27 Ohm per square and optical transmittance of 87.52% was used as a reference. We also fabricated exible electrochromic devices on the optimized Cu 2Àx O/Cu/Cu 2Àx O multilayer grid electrode. Prior to the deposition of poly(3-hexylthiophene) (P3HT), a poly(3,4ethylenedioxythiophene):polystyrene sulfonate (PEDOT:PSS) lm (NanoWearable Co.) was spin-coated at 1500 rpm for 30 s and cured at 120 C for 5 min. A 2.2 wt% solution of P3HT (Aldrich) in chlorobenzene (Aldrich) was spin-coated on the PEDOT:PSS/grid substrates at 1500 rpm for 20 s and then dried on a hot plate at 60 C for 10 min. The thickness of the P3HT lms was around 60 nm, as measured using a fused ion beamscanning electron microscope (FIB-SEM) (Helios NanoLab 600i). Electrochromic tests were performed in a threeelectrode system with a propylene carbonate solution containing 0.5 M LiClO 4 . The working electrode was the P3HT lm on a grid electrode. Pt wire and Ag/Ag + wire were used as the counter and reference electrodes, respectively. The potential of the samples was controlled using a potentiostat/galvanostat (PGSTAT 302N, Autolab) and the optical properties of P3HT were measured using a UV/vis spectrometer (Cary 100, Agilent Technologies). The pulse potential tests were carried out by applying À0.2 V for coloring and 0.8 V for bleaching. Each coloring and bleaching time was set to 60 s.

Results and discussion
To investigate the stoichiometry and phase of the copper oxide fabricated by RTR sputtering, we carried out XPS analysis for the top Cu 2 O layer in the Cu 2Àx O/Cu/Cu 2Àx O multilayer grid. Fig. 2 shows the XPS core level spectra of the Cu 2p peaks and the O 1s peaks obtained from Cu 2Àx O/Cu/Cu 2Àx O grid multilayer lm. In general, reactive sputtered Cu could form two different oxides, such as cuprous oxide (Cu 2 O) and cupric oxide (CuO), depending on the oxygen ow ratio. 20,21 The binding energies of the Cu 2p 1/2 (951.43 eV) and 2p 3/2 (931.38 eV) were matched with general cuprous Cu 2 O phase. 22,23 The O 1s peak at 530.38 eV was also matched with cuprous Cu 2 O phase. 24 In addition, XPS analysis showed that the RTR sputtered Cu 2 O phase had a Cu-deciency (Cu/O ratio: 1.88-1.92). Therefore, we concluded that the multilayer grid was consisted of top and bottom Cu 2Àx O phase. In our previous work, we also conrmed the cuprous Cu 2 O phase of the reactive RTR sputtered copper oxide lms using XRD and TEM examinations. 23 Thickness of each layer in the Cu 2Àx O/Cu/Cu 2Àx O multilayer grid electrodes was determined by consideration of optical transmittance, conductivity and adhesion of the multilayer grid. Generally, the high reection from Cu metal interlayer resulted in glittering of the Cu grid, which prevent the use of Cu grid electrode for large area smart window. interlayer has a semiconducting property, the thick Cu metallic layer is necessary in the multilayer grid electrode for obtaining a low sheet resistance comparable to typical metal grid electrodes. Fig. 3 shows the sheet resistance and resistivity of the patterned Cu 2Àx O/Cu/Cu 2Àx O multilayer grid electrodes as a function of grid line width and pitch. In general, the electrical and optical properties of metal grid electrodes are critically dependent on the geometry of the grid, such as the grid width, grid height, and pitch. [25][26][27] Therefore, it is very important to optimize the grid width and pitch to obtain low sheet resistance and high transmittance. Fig. 3(a)-(d) show the sheet resistance and resistivity of the Cu 2Àx O/Cu/Cu 2Àx O multilayer grid electrodes at a specic grid width (10, 12, 14, and 16 mm) with increasing pitch length. Each inset shows the geometry of the constant grid width. It is clear that the sheet resistance and resistivity of the Cu 2Àx O/Cu/Cu 2Àx O multilayer grid electrode decreased with increasing grid width from 10 to 16 mm. However, the grid pitch did not affect the sheet resistance or resistivity; with increasing grid pitch length from 500 to 750 mm, the grid electrode showed similar sheet resistance and resistivity because the changes in grid pitch were small. At a grid width of 16 mm, the Cu 2Àx O/Cu/Cu 2Àx O multilayer grid showed the lowest sheet resistance of 7.69 AE 0.59 Ohm per square and resistivity of 3.46 AE 0.2 Â 10 À4 Ohm cm À1 . Due to the existence of the conductive Cu interlayer, the Cu 2Àx O/Cu/Cu 2Àx O multilayer grid exhibited better metallic conductivity than a typical ITO electrode. The calculated resistivity of the Cu interlayer was found to be 3.0 Â 10 À6 Ohm cm À1 , which is similar to that of bulk Cu (1.7 Â 10 À6 Ohm cm À1 ). 23 Therefore, the grid width of the Cu interlayer affected the electrical properties of the Cu 2Àx O/Cu/Cu 2Àx O multilayer grid because it provides the main current path. Fig. 4 shows the optical transmittance of the Cu 2Àx O/Cu/ Cu 2Àx O multilayer grid electrode as a function of grid width and pitch length. With increasing grid width and decreasing pitch length, the optical transmittance decreased, as shown in Fig. 4(a)-(d). At grid widths of 10 and 12 mm, the optical transmittance of the Cu 2Àx O/Cu/Cu 2Àx O multilayer grid electrode was almost constant regardless of grid pitch between 500 and 750 mm, as shown in Fig. 4(a) and (b). However, at grid widths of 14 and 16 mm, the optical transmittance of the multilayer grid was affected by pitch length. The optical transmittance of the electrode began to decrease at a grid pitch of 550 mm. Therefore, to obtain a high-performance multilayer grid electrode, the grid pitch should be larger than 550 mm. In the Cu 2Àx O/Cu/Cu 2Àx O multilayer grid electrodes, a high conductivity was determined by width of the metallic Cu layer because top and bottom Cu 2Àx O layers had semiconducting properties unlike typical OMO electrodes where the oxide layer is highly conductive oxide layer. In addition, the optical transmittance of the Cu 2Àx O/Cu/Cu 2Àx O grid electrode was dependent on the uncovered space in the grid structure because the Cu 2Àx O/Cu/ Cu 2Àx O multilayer had an optical transmittance of 0% (Fig. S1 †). Therefore, appropriate design of grid structure and geometry is very important to obtain high-quality multilayer grid electrodes.
To determine the optimum grid width and pitch length of the Cu 2Àx O/Cu/Cu 2Àx O multilayer grid electrodes for exible smart windows, the gure of merit (FOM; T 10 /R sh ), as dened by Haacke, 28 was evaluated from the measured sheet resistance (R sh ) and optical transmittance (T) at a wavelength of 550 nm. As shown in Fig. 5(a) and (b), for very thin grid widths below 12 mm, the Cu 2Àx O/Cu/Cu 2Àx O multilayer grid electrode had a similar FOM value regardless of grid pitch length. The 12 mm grid showed slightly increased FOM values compared to the 10 mm grid due to the slightly decreased sheet resistance. However, at grid pitches of 14 and 16 mm, the FOM of the Cu 2Àx O/Cu/ Cu 2Àx O multilayer grid was affected by grid pitch, as shown in Fig. 5(c) and (d). Due to the low optical transmittance, the grid had a low FOM at grid pitch lengths of 500 and 550 mm. The Cu 2Àx O/Cu/Cu 2Àx O multilayer with a grid width of 16 mm had the highest FOM (37.34 Â 10 À3 Ohm À1 ) at a grid pitch length of 600 mm. The grid width and pitch length dependence can be explained by the lling factor (f F ) of the Cu 2Àx O/Cu/Cu 2Àx O multilayer grid. We calculated f F as follows. Ghosh et al. reported that a Ni grid electrode has low sheet resistance and high optical transmittance at a ll factor of 0.025. 27    To evaluate the mechanical exibility of the optimized electrode for exible TFHs and electrochromic devices, we measured the resistance change of the Cu 2Àx O/Cu/Cu 2Àx O multilayer grid electrode and a sputtered ITO electrode as the bending radius decreased during inner and outer bending of the substrate. Fig. 7 shows the inner/outer bending test results for both electrodes with decreasing bending radius from 25 to 2 mm. The change in resistance of the electrodes as a result of bending can be expressed as (R À R 0 )/R 0 , where R 0 is the initial measured resistance and R is the resistance measured during substrate bending. The Cu 2Àx O/Cu/Cu 2Àx O multilayer grid electrode showed constant resistance until a bending radius of 2 mm, which was the limit of our bending test machine. The grid-patterned electrode therefore had a very small critical bending radius of below 2 mm due to the outstanding mechanical exibility of the Cu interlayer. On the other hand, the sputtered ITO electrode broke at a bending radius of 5 mm in the outer bending test, while the resistance was constant for the inner bending test. As we previously reported, the resistance change is much lower during the inner bending test than during the outer bending test due to the overlapping of broken or laminated thin lms. 29 To verify the stability of the electrodes, a dynamic fatigue test was performed for the Cu 2Àx O/Cu/Cu 2Àx O multilayer grid electrode and a sputtered ITO reference electrode for 10 000 cycles, as shown in Fig. 8. Repeated outer bending was carried out at a xed outer bending radius of 5 mm, which is a fairly small radius considering large-area smart windows with large curvature. In the case of the Cu 2Àx O/Cu/Cu 2Àx O multilayer grid electrode, the resistance was constant even aer 10 000 cycles repeated bending. However, the resistance of the ITO single layer changed aer 100 cycles of repeated outer bending due to crack formation and the separation of cracked ITO lm. This separation led to an abrupt increase in measured resistance, as shown in Fig. 8(b). Based on the results of these tests, we conrmed that the Cu 2Àx O/Cu/Cu 2Àx O multilayer grid electrode is much more stable than an ITO single-layer electrode.
To apply the RTR-fabricated Cu 2Àx O/Cu/Cu 2Àx O multilayer grid electrode to a exible smart window, we rst fabricated exible thin lm heaters on the optimized electrode, as shown   in Fig. 9(a). The exible TFHs were fabricated with a size of 25 Â 25 mm 2 using a two-terminal Ag contact conguration (Fig. S2 †). A DC voltage was applied to the TFHs through the sputtered Ag metal contact electrodes at the lm edge, and the temperature proles were measured using a thermocouple placed on the surface and an IR thermometer (Fig. S3 †). A exible TFH was mounted on a specially designed sample jig to supply power and measure the temperature. Fig. 9(b) shows the temperature proles of the TFHs with Cu 2Àx O/Cu/Cu 2Àx O multilayer grid electrodes, plotted with respect to input voltage from 2 to 5.5 V. Generally, as the input voltage increased, the saturation temperature of the TFHs increased. Due to the low sheet resistance (7.18 Ohm per square) of the Cu 2Àx O/Cu/ Cu 2Àx O multilayer grid electrode, the exible TFHs on the electrode with a width of 16 mm and pitch length of 600 mm reached a saturation temperature of 100 C when a low DC input voltage of 5.5 V was applied. The higher saturation temperature of the exible TFHs with Cu 2Àx O/Cu/Cu 2Àx O multilayer grid electrodes at low input voltage implies that efficient transduction of electric energy through Joule heating occurred. Based on Joule's law, the saturation temperature of the exible TFHs can be expressed as follows.
The power (P) applied to the exible TFHs over a heating time (Dt) generates heat in the TFH, as illustrated in Fig. 9(a). In the above equation, V is the applied voltage, R is the device resistance, h conv is the heat transfer coefficient, A conv is the surface area, and T s and T i are the saturation and initial temperatures. Therefore, it is apparent that the saturation temperature of exible TFHs increases with increasing input voltage (V) and with decreasing resistance (R). Therefore, the low sheet resistance of the Cu 2Àx O/Cu/Cu 2Àx O multilayer grid electrode mainly results in the rapid attainment of a saturation temperature of 100 C even at low voltage, which is appropriate for removing frost or deicing a smart window. On the other hand, the exible TFH fabricated on an ITO electrode reached the saturation temperature of 100 C at a higher input voltage of 10.5 V due to the higher sheet resistance (38.27 Ohm per square) of the sputtered ITO lm. Although the ITO-based TFH did reach a temperature of 100 C, it cannot be applied to exed or specially shaped surfaces due to the brittleness of the sputtered ITO lms. This indicates that the roll-to-roll sputtered and patterned Cu 2Àx O/Cu/Cu 2Àx O multilayer grid electrode is a promising exible TCE for an effective transparent defroster in smart windows. It could be employed as a defogging/deicing window in automobiles, helmets, and smart windows due to its exibility and transparency. To investigate the durability of Cu 2Àx O/Cu/Cu 2Àx O multilayer grid in the exible TFHs, we performed repeated heating-cooling tests for 10 cycles. Fig. 10(a) shows the temperature proles of the Cu 2Àx O/Cu/ Cu 2Àx O multilayer grid-based TFHs for 10 repeated cycles. The Cu 2Àx O/Cu/Cu 2Àx O multilayer grid-based TFHs showed identical temperature proles, rapidly reaching a saturation temperature of 100 C when a DC voltage of 5.2 V was applied. Fig. 10(b) and (c) compared the sheet resistance and optical The electrochromic absorption of the exible P3HT/ PEDOT:PSS lm was monitored by UV-vis spectroscopy at different applied voltages. A series of spectroelectrochemical spectra for the sample are shown in Fig. 11(a). When a potential of from À0.2 V to 0.8 V was applied (oxidation reaction), the absorption peak at around 520 nm decreased and a new band formed around 700 nm. During this reaction, the color of the lm became lighter, changing from its original red color and eventually becoming transparent blue with the formation of bipolaronic P3HT (see inset of Fig. 11(a)). 30,31 During the reverse scan from 0.8 V to À0.2 V, the reduction reaction occurred and the P3HT lms recovered their original red color. In the initial stages of the electrochemical reaction in this voltage range, the electrochromic coloration was stable without any degradation of the polymer lms. However, the P3HT lm coated directly on a grid substrate did not show any color change during the electrochemical reactions when a potential from À0.2 V to 0.7 V was applied, as shown in Fig. S4(a). † The abrupt decay in the absorption spectra at 0.8 V might be attributed to partial dissolution of the P3HT lm due to a high electric eld locally concentrated on the grid. These ndings indicate that coating the Cu 2Àx O/Cu/Cu 2Àx O multilayer grid electrode with a conductive PEDOT:PSS layer homogenized the electric eld in the lm, and that the P3HT lm was found to enable electrochemically stable optical modulation at the given potential. 32 Although PEDOT is well known to be a cathodically coloring  material with strong blue absorption, 33 the PEDOT:PSS layer was optically inactive during electrochemical reactions between À0.2 and 0.8 V (Fig. S4(b) †), which implies that the electrochromic absorption of the P3HT/PEDOT:PSS lm entirely originated from the P3HT layer. Fig. 11(b) presents the transmittance data for a P3HT lm during cyclic potential switching between À0.2 V and 0.8 V. The differences in transmittance (DT) were 48.3% and 48.8% measured at 520 nm and 700 nm, respectively. When the oxidation potential was extended to 1.2 V (as shown in Fig. 11(c)), DT at 520 nm gradually increased to 68.5%, while DT at 700 nm was almost the same or slightly decreased. This means that the main absorption peak of the P3HT lm located at 520 nm could be modulated by extending the oxidation potential. In particular, the bleached P3HT lm under 1.2 V of oxidation potential was highly transparent, with 91.4% of transmittance (Fig. S5 †). The response times during the bleaching (s b ) and coloring (s c ) processes were estimated as the time to reach 90% of the total transmittance difference. The s b and s c measured at 520 nm were 49.5 s and 30.4 s, respectively. Compared to EC devices with silver grid substrates, 34 these values are quite reasonable with a high DT, but could be improved by further optimization of grid geometry or of the conductive PEDOT:PSS layer.

Conclusions
In summary, we developed a Cu 2Àx O/Cu/Cu 2Àx O multilayer grid electrode for TFHs and electrochromic devices using roll-to-roll sputtering and patterning processes at room temperature. The optimized Cu 2Àx O/Cu/Cu 2Àx O multilayer grid electrodes exhibited a low sheet resistance of 7.17 Ohm per square and high optical transmittance of 87.6%. In addition, we found that the Cu 2Àx O/Cu/Cu 2Àx O multilayer grid had outstanding exibility due to the high exibility of the Cu interlayer. The mechanical stability of the Cu 2Àx O/Cu/Cu 2Àx O multilayer grid was compared with that of a commercial ITO electrode using inner/outer bending and fatigue tests. Due to the lower sheet resistance of the electrode, the exible TFH with the transparent Cu 2Àx O/Cu/Cu 2Àx O multilayer grid required a lower input voltage (5.5 V) to reach a saturation temperature of 100 C than the ITO-based TFHs. Furthermore, a P3HT lm coated on a Cu 2Àx O/Cu/Cu 2Àx O substrate exhibited efficient coloring/ bleaching performance. The temperature proles and EC properties indicate that the Cu 2Àx O/Cu/Cu 2Àx O multilayer grid electrode is a promising TCE for exible smart windows.

Conflicts of interest
The authors declare no competing nancial interests.