Supporting Information Interface Anchored Effect on Improving Working Stability of Deep Ultraviolet Light-Emitting Diode Using Graphene Oxide-based Fluoropolymer Encapsulant Renli Liang 1,Jiangnan Dai 1 *, Linlin Xu 1, Yi Zhang 1, Ju He 1, Shuai Wang 1, Jingwen Chen 1, Yang peng 2, Lei Ye 3 *, Hao-chung Kuo 4 * and Changqing Chen 1 1 Wuhan National Laboratory for Optoelectronics, Huazhong University of Science and Technology, Luoyu Road 1037, Wuhan, 430074, China 2 School of Mechanical Science & Engineering, Huazhong University of Science and Technology, Luoyu Road 1037, Wuhan, 430074, China 3 School of Optical and Electronic Information, Huazhong University of Science and Technology, Luoyu Road 1037, Wuhan, 430074, China 4 Department of Photonics and Institute of Electro-Optical Engineering, National Chiao Tung University, 1001 Ta Hsueh Road, Hsinchu 300, Taiwan S-1
Corresponding email address: daijiangnan@mail.hust.edu.cn; leiye@hust.edu.cn; hckuo@faculty.nctu.edu.tw 1. The process of GO-based fluoropolymer via a solvent-exchange approach Figure S1. Photographs of 0.04 wt% GO-based fluoropolymer at each step of the solventexchange method. The dispersion of GO in ethanol (a) is centrifuged and sonicated to redispersed GO in fluoropolymer (b). And the insert is the image of 0.20 wt% GO-based fluoropolymer before centrifugation. Figure S2. UV transmittance of silicone resin and GO-based fluoropolymer with various GO contents. In the DUV region of 200-300nm, high transmittance values (>99%) can be achieved in S-2
neat fluoropolymer, and the transmittance of GO-based fluoropolymer just decreases by 2% as the content of GO increases to 0.10 wt% at 0.1 mm thickness. The insert represents the photographs of 0.04 wt% GO-based fluoropolymer before (a) and after (b) the solvent-exchange method. As Figure S1 displayed, stable GO-based fluoropolymer can be formed using a solventexchange method, which is an effective and simple approach. Firstly, prepared GO according to Hummer's method, then dispersed the GO powder in ethanol by using sonication for 1 hour. Next, the dispersed GO in ethanol was mixed with neat fluoropolymer. The weight ratios of GO powder in fluoropolymer and were 0.02, 0.04, 0.06, 0.08, 0.10 and 0.20. The mixture was then centrifuged for 1 hour at ten thousand RPM and natural dried to remove ethanol completely. Finally, the GObased fluoropolymer was stirred uniformly by planetary vacuum mixer. The fundamental role of an encapsulant is to ensure long-term working stability of the DUV-LED by protecting it, as well as allowing UV-light emission. Therefore, the high UV-light transmittance of the encapsulant plays an important role because it directly affects the package efficiency and the reliability of DUV-LED. To demonstrate the better UV-light transmittance of GO-based fluoropolymer composite than that of the conventional encapsulant [Silicone resin (MS- 1002, Dow Corning)], the measured transmittance results are shown in Figure S2. It is observed that the deep UV-light can be absorbed by the silicone and the transmittance decreases dramatically below 230 nm, but the deep UV-light transmittance of the fluoropolymer and the most GO-based fluoropolymer composites almost remains, except the composite with the GO content of 0.2%, suggesting that the GO-based fluoropolymer composites (their GO content is <0.2%) promise a potential application as the encapsulant. S-3
2. The process of surface treatment using APTS and fabrication of sandwich structure. Figure S3. Contact-mode AFM scan of graphene oxide. Height profile through the blue line was shown in the insert. According to the insert, the thickness of GO in this investigation is about 1 nm. Figure S4. The packaging process of sandwich structure. (a) The fluid A contained 95 wt% ethanol and 5 wt% deionized water; (b) The fluid B contained 1 wt% APTS was adopted to modify the surface of sapphire (DUV-LED chip) and quartz lens; (c) The encapsulant was dispensed on S-4
the surface of sapphire, then put a quartz lens on the top; (d) The encapsulant would fill the whole interface under the action of vacuum and capillarity. A baking process at 80 over 12 hours was necessary for a sufficient anchoring reaction. Figure S3 shows an AFM image of graphene oxide fabricated according to Hummer's method, and the thickness of GO in this investigation is about 1 nm. The packaging process of sandwich structure was detailedly described in Figure S4. The fluid B contained 1 wt% APTS was adopted to modify the surface of sapphire (DUV-LED chip) and quartz lens. After that, the encapsulant was dispensed on the surface of sapphire, then put a quartz lens on the top. The encapsulant would fill the whole interface under the action of vacuum and capillarity, as Figure S4(d) displayed. A baking process at 80 over 12 hours was necessary for a sufficient anchoring reaction. This chemical reaction may progress according to 2NH 2 (CH 2 ) 2 Si(O-) 3 +GO-COOH GOCONH(CH 2 ) 2 Si(O-) 3 +H 2 O (1) 3. The calculation of air barrier of DUV-LEDs using mapping method S-5
Figure S5. Reconstructed and mapped filled and air barrier of fluoropolymer encapsulant. Considering for a feasible calculation, mapped meshing was used to rebuild the filled area and air barrier. The images were zoomed to the standard scale. The interface was divided into an 15 15 mesh. The size of the mesh element in this study was 85 µm, which allowed for the accurate identification of most of the insufficiently filled areas in the bonding structures. If the air barrier in each mesh element occupied more than half of the element, the entire element was recognized as air barrier, the marked with, as shown in Figure S5. It was observed that there are many air layers or gaps around the DUV-LED with neat encapsulant. S-6
4. The powder X-ray diffraction (XRD) results of GO and modified GO (m-go) Figure S6. XRD patterns of graphene, GO and m-go. The major peaks of unmodified GO powder are observed at 2θ angles of 10.2 and 40.2, but of m-go modified by ATPS are obtained of 6.4 and 21.4. Powder X-ray diffraction (XRD) was performed on a Bruker D8 Advance DAVINCI working in Bragg Brentano (θ/2θ) geometry. Diffractograms were recorded under Cu Kα radiation, with a step size of 0.013, and using variable divergent slits. The XRD patterns of graphene, GO and m- GO were presented in Figure S6. For the graphene sample, the (0 0 2) peak appeared at 26.50 indicating an interlayer spacing of 0.35 nm. GO had a large interlayer distance (~0.86 nm, 2θ = 10.2 ) due to the formation of hydroxyl, epoxy, and carboxyl groups which indicated that the graphene had been successfully exfoliated during the chemical modification process. The XRD spectrum of m-go showed two obvious peaks. The one peak at 2θ = 6.4 (~1.38 nm), was expected from the grafted APTS on the GO. The other widish peak at 2θ = 21.4 was expressed by S-7
the (0 0 2) peak of graphene. Comparing the peak of m-go with the peak at 2h = 26.50 of graphene, the interlayer spacing of m-go was still larger than that of the graphene for the introduction of APTS. 5. The XPS results of GO and GO (m-go) Figure S7. The XPS spectra of (a) GO and (b) m-go modified by ATPS, and high resolution region scan of (c) C1s and (d) Si2p of m-go. The insert of (a) shows high resolution region scan of C1s of unmodified GO. Table S1. XPS survey data for the most concentrated elements before and after modification. Samples Peak designation Band (ev) At% conc. before C 286.96 67.37 O 532.71 32.63 after C 284.81 49.65 S-8
O 523.56 21.8 N 400.8 6.05 Si2p 102.3 16.4 Si2s 153.8 6.46 In-depth chemical information on the surface can be gained via XPS. The results of the XPS analyses are shown in Figure S7. Table S1 provides details. In fact, nitrogen, oxygen, silicon and carbon are associated with APTS. During the configuration process, condensation reactions between the NH 2 groups in ATPS and COOH groups in GO are expected and are accelerated by the hydrolysis. And the contents of N and Si were obviously increased after the reaction, as shown in the Table S1. C1s XPS spectrum of the m-go modified by APTS is shown in Figure S7(c), the C=O (286.7 ev) and C-OH (285.9 ev) mode peaks are less prominent, indicating that cross-link network of APTS is formed on the surface of m-go. In addition, the high resolution region scan of Si2p is shown in the insert of Figure S7(d), and the formation of Si O C and Si O Si structures in m-go is observed. The spectrum of Si2p has been deconvoluted into two Gaussian peaks centered at 102.3 and 103.2 ev, respectively. The full width of half maximums (FWHMs) of the peaks at 102.3 and 103.2 ev are 1.37 and 1.15 ev. The Gaussian peak at 102.3 ev accounts for Si O C bridged bond formed between APTS and hydroxyl on the surface of m-go, and the peak at 103.2 ev should be Si O Si bonds between APTS molecules, indicating that cross-link network of APTS is also formed on the surface of m-go. The results indicated that a complex chemical crosslink could be formed between the APTS and GO-based fluoropolymer. S-9
6. The thermal improvement of anchoring structure Figure S8. Difference structure functions of 0.00 wt%, 0.04 wt% and 0.10 wt% GO-embedded DUV-LEDs. Although the most of the heat of junction is passed down to the heat sink, some heat spread to the external environment through the encapsulation. Therefore, the GO-based fluoropolymer encapsulant had better thermal performance than traditional structure benefit from the GO as filler. The thermal resistances of 0.00 wt%, 0.04wt% and 0.10 wt% DUV-LEDs were measured using T3ster, and the structure functions were shown in Figure S8. Accordingly, compared with traditional structure, the total thermal resistance of DUV-LED with 0.10 wt% GO-based fluoropolymer is reduced by 0.7 K/W owing the thermal improvement of encapsulant, to indicate the fluoropolymer encapsulant with GO embedded can improve the thermal dissipation around S-10
chip to benefit the performance of DUV-LEDs. 7. The influence of humidity and junction temperature on the voltage of DUV-LEDs. Figure S9. The working voltage of DUV-LEDs with different encapsulants at the current of 300 ma. The measurements of working voltage corresponded to the forgoing discussion. It was observed that a sharp drop in working voltage by 0.5 V of neat and 0.04 wt% GO-based fluoropolymer DUV-LEDs was obtained at 150 hours and 450 hours, respectively, which reasonably led to a light failure. It was because that the structure of epitaxial layer was destroyed in the interaction of humidity and heat, and electronics tended to be leaky instead of optical conversion. However, the DUV-LED with 0.10 wt% GO-based fluoropolymer had a relatively S-11
stable working voltage profit from protection of anchoring effect, and was believed to be promising in DUV-LED applications. S-12