Observation of Highly Durable Silicone Resin for Encapsulating AlGaN-Based UVB Light-Emitting Diodes

In this paper, we report an AlN-based ceramic lead frame (LF) with encapsulating silicone between the surface of an AlGaN-based ultraviolet-B light-emitting diode (UVB-LED) chip and a quartz glass cover; the light output power (LOP) of this structure was 13.8% greater than that of the corresponding packaging structure without encapsulating silicone. Another packaging structure in which the silicone fully filled the cavity of the AlN-based ceramic LF included covering with quartz glass; in this case, the enhancement of the LOP was 11.7%. Reliability tests performed over a period of 3500 h at a forward current (If) of 100 mA revealed that the LOPs of these two silicone-containing packaging types decreased to 45.3 and 48.6%, respectively, of their initial values. The different degradation rates of these UVB-LEDs were not, however, correlated with the appearance of cracks in the encapsulating silicone during long-term operation. Excluding any possible mechanisms responsible for degradation within the UVB-LED chips, we suggest that the hermetic cover should be removed to avoid the appearance of cracks. Moreover, the main mechanism responsible for the slow degradation rates of LOPs in these proposed packaging structures involves the encapsulated silicone, after cracks have appeared, undergoing further deterioration by the UVB irradiation.


Introduction
III-Nitride-based deep-ultraviolet (DUV) light-emitting diodes (LEDs) have a wide range of applications. For example, the emission wavelength of AlGaN-based LEDs can be tuned by alloying GaN with AlN to cover almost the entire ultraviolet (UV) spectral range (200-365 nm), making them perfectly suited to applications in biological, environmental, industrial, and medical fields [1,2]. Furthermore, AlGaN-based ultraviolet-B LEDs (UVB-LEDs) have been attracting great attention since the implementation of the Minamata Convention on Mercury. There is much potential for such LEDs to be used in medical (e.g., cancer immunotherapy; treatment of psoriasis, vitiligo, and pemphigus vulgaris) and agricultural (e.g., plant growth under UVB lighting; production of phytochemicals in the green leaves of vegetables) applications [3][4][5][6][7]. For example, narrow-band (NB-UVB) emission therapy at 310 nm has been found to be more effective than broad-band (BB-UVB) therapy at 280-320 nm in the treatment of psoriasis and atopic dermatitis; furthermore, NB-UVB phototherapy with emission at 310 nm appears suitable for cancer therapy and skin cure therapy [8][9][10][11]. Several reports have disclosed that high external quantum efficiencies (EQEs) are possible for UVB-LEDs when combining a low-dislocation-density AlN template substrate, a well-optimized n-AlGaN buffer layer, an electron injection layer, a transparent p-AlGaN layer, and reflective electrodes [12][13][14][15]. At present, it is difficult to prepare highly conductive p-AlGaN contact layers with high Al contents (>50%); therefore, it has been necessary to use p-GaN as the contact layer to lower the series resistance of UVB-LEDs, even though its presence significantly decreases the light-extraction efficiency (LEE) as a result of strong absorption of UVB light. Moreover, several reports have suggested that silicone is unreliable as an encapsulating material against DUV light [16][17][18]. Therefore, to improve the reliability of UVB-LEDs, the conventional packaging design has involved flipping the LED chip bonded onto the AlN-based direct plating copper ceramic lead frames (AlN-DPC LFs) and using an attached quartz glass as the hermetic cover. This kind of packaging structure for UVB-LEDs results, however, in severely decreased light extraction and limited optical performance [19][20][21]. Furthermore, quartz glass is more expensive than silicone. Some DUV-transparent encapsulation materials have been proposed to ameliorate these drawbacks, including sol-gel methyl siloxane hybrid materials, fluorine resins, and silicone oil [22][23][24]. Therefore, in terms of simplifying the manufacture process, it would be profoundly useful to develop an inexpensive packaging structure using silicone as the transparent encapsulant that displays improved LEEs and acceptable reliability.
In this study, we developed a simple, inexpensive, and highly durable packaging structure, using reliable silicone as the transparent encapsulation material, to improve the LEEs of UVB-LEDs. We have investigated the effects on the LEEs of various encapsulation methods applied in AlN-DPC LFs, and compared them with a conventional packaging structure. Our results suggest that such reliable silicone has great potential to induce long-term stability when removing the hermetic cover. Figure 1 provides a schematic representation of the epitaxial structure of the AlGaNbased NB-UVB LED, which was obtained using low-pressure metal-organic chemical vapor deposition (LP-MOCVD). Trimethylaluminum (TMA), trimethylgallium (TMG), bis(cyclopentadienyl)magnesium (Cp 2 Mg), ammonia (NH 3 ), and silane (diluted with H 2 to a concentration of 30 ppm) were used as Al, Ga, Mg, N, and Si sources, respectively. The TMA, TMG, and Cp 2 Mg precursors, exhibited the typical purity grade of ≥99.9999%, were obtained from Nata Optoelectronic Material (China). Additionally, the purity grade of NH 3 was over 99.9999% through a purification process. First, a 2.2 µm-thick AlN buffer layer was grown on a 2 inch (0001)-oriented sapphire substrate. Next, a strain-relieving interlayer, consisting of 30 periods of an AlN/AlGaN superlattice with an equivalent Al composition of 0.72, was grown on the AlN buffer layer. A 1.5 µm-thick layer of undoped Al 0.6 Ga 0.4 N was then grown on the superlattice interlayer, followed by a 2 µm-thick Sidoped n-Al 0.5 Ga 0.5 N layer as the n-contact layer. The active region of multiple quantum wells (MQWs) included five pairs of 2.2 nm-thick Al 0.35 Ga 0.65 N quantum wells (QWs) and 10 nm-thick Al 0.45 Ga 0.55 N quantum barriers (QBs), followed by a twofold Mg-doped p-Al 0.55 Ga 0.45 N (10 nm)/Mg-doped p-Al 0.4 Ga 0.6 N (2 nm) structure as the electron blocking layer (EBL). Subsequently, a 50 nm-thick Mg-doped p-AlGaN layer, with the Al content grading from 30% to 0%, was grown on the EBL. Finally, a 20 nm-thick Mg-doped p-GaN layer was deposited, serving as the p-contact layer. After growing the epitaxial layers, the sample was annealed in situ under a N 2 ambient to activate the Mg dopants at 700 • C for 15 min. UVB-LED chips were fabricated using standard flip-chip processing technologies. All the metal sources exhibited the typical purity grade of ≥99.999%. Mesa structures were defined using an inductively coupled plasma (ICP) dry etching system to expose the n-Al 0.5 Ga 0.5 N layer surface down to an etching depth of 0.5 µm. Next, n-contacts of Ti/Al/Ti/Au (100/200/30/100 nm) were deposited through electron-beam (EB) evaporation and subjected to rapid thermal annealing (RTA) at 980 • C for 60 s. To form a transparent p-contact, a 50 nm-thick layer of indium tin oxide (ITO) was sputtered on top of the p-GaN layer, and then the sample was annealed at 600 • C for 10 min. The UVB-LED chips were then fabricated by using EB evaporation to deposit Ti/Pt/Au (50/30/100 nm) on both the n-contacts of the metals and the ITO film. Moreover, a 1-µm-thick passivation layer of SiO 2 was deposited. Finally, a 3 µm-thick layer of AuSn alloy was deposited to form the bonding pads. The processed wafer was then lapped and scribed into dimensions of 550 × 550 µm 2 .

Materials and Methods
The prepared UVB-LED chips were separately flip-bonded onto four AlN-DPC LFs through soldering; a quartz glass covering served as an optical lens for each packaged sample. Structural diagrams of the four package types (S1-S4) are displayed in Figure 2. The yellow part is the DPC film with an Au coating over the surface; the blue part is the LED chip; the green part is the encapsulated silicone; the pink part is the quartz glass; and the black part is the AlN ceramic material. S1 is a packaging type in which the encapsulated silicone is positioned above the surface of the UVB-LED chip, but with a gap preserved between the silicone and the quartz glass cover. S2 denotes a packaging type featuring the encapsulated silicone above the surface of the UVB-LED chip, but without a gap between the silicone and the quartz glass cover. Notably, the cavities in the AlN-DPC LFs of samples S1 and S2 did not have any filling silicone. S3 is a packaging type with the encapsulated silicone above the surface of the UVB-LED chip and a filling of silicone in the cavity of the AlN-DPC LF; in other words, sample S3 was prepared without a gap present between the encapsulated silicone and the quartz glass cover. Finally, S4 is a conventional packaging type with an attached quartz glass as the cover, prepared without encapsulating silicone; it was used as a reference sample. The silicone (JS-UV200-L) displaying high transmittance of wavelengths in the range from 280 to 320 nm, obtained from Jusheng Optoelectronics (Shenzhen, China) [25], was used as the encapsulant; the curing process took approximately 3 h at 150°C in each case. The dimensions of the AlN-DPC LF and quartz glass were 3.75 (L) × 3.75(W) × 1.4 (H) mm and 2.9 (D) × 0.3 (H) mm, respectively.
The LOP, forward voltage (V f ), and electroluminescence (EL) spectrum of each packaged sample were measured at a forward current (I f ) of 100 mA in a continuous wave (CW) HAAS-2000 LED photoelectric measurement system (Everfine) equipped with a 30 cm-diameter integrating sphere. Aging tests were performed over 3500 h to investigate the UVB-LED characteristics after lifetime aging; the packaged samples were mounted separately onto a printed circuit board (PCB) with series connection at a value of I f of 100 mA at room temperature (RT).

Figure 2.
Structural diagrams of the packaging structures (samples S1-S4). The yellow part is the DPC film with an Au coating over the surface; the blue part is the LED chip; the green part is the encapsulated silicone; the pink part is the quartz glass; and the black part is the AlN ceramic material. Table 1 lists the values of V f , the EL peak wavelength, the EL full width at half maximum (FWHM), and the LOP at a value of I f of 100 mA, all measured before and after filling the silicone into the samples S1-S3 as the encapsulant and attaching a quartz glass as the cover. Notably, S4 was a reference sample prepared without encapsulating silicone, but featuring a covering quartz glass. All the EL spectra featured a near-constant peak wavelength of approximately 311 nm and an EL FWHM of approximately 9.5 nm. Table 1. Variations in the values of V f , the EL peak wavelength, the EL FWHM, and the LOP of samples S1-S4 measured at 100 mA, before and after encapsulating with silicone and covering with quartz glass. Furthermore, the average variations in the values of V f were all less than 1%. In contrast, significant enhancements in the LOP occurred for samples S2 and S3 after encapsulation: 7 and 5% (tolerances: ±1%), respectively. We attribute this behavior to lower degrees of total internal reflection (TIR) after decreasing the refractive index (n) gap between the UVB-LED chip (sapphire; n = 1.82) and air (n = 1) by encapsulating silicone within the gap [26]. Moreover, compared with the LOP for sample S4, the LOPs for samples S1-S3 after encapsulation were enhanced by 3.2, 13.8, and 11.7%, respectively, due directly to the influence of the encapsulating silicone. Thus, the LEEs of the UVB-LEDs were improved after filling silicone inside the AlN-DPC LFs of the proposed packaging structures. Figure 3 displays the relative LOPs of the packaging samples over time; the values were normalized to the initial value. After operation with an aging time of 200 h, the relative LOPs of samples S1-S4 decreased to 70, 67.5, 71.4, and 70.9% of their initial values, respectively. Although the rates of degradation of samples S1, S3, and S4 were not obviously different, that of sample S2 was slightly faster during the 200 h of aging. Furthermore, the rates of degradation of samples S2 and S3 (−19.6 and −17.5%, respectively) were faster than those of samples S1 (−12.3%) and S4 (−9.7%) during operation between 200 and 700 h. Interestingly, cracks were not present in the encapsulated silicone of samples S1-S3, observed using a microscope, prior to an aging time of 700 h. Cracks did appear, however, in the encapsulated silicone of samples S1-S3 after operation with an aging time of 1500 h. Furthermore, in comparison with reference sample S4, cracks were evident in photographs of samples S1-S3 after aging for 3500 h (Figure 4). Thus, the different rates of degradation of the UVB-LEDs, when silicone was introduced as the encapsulating material, were not obviously correlated with the phenomena of cracks appearing in the silicone during long-term operation. Nevertheless, we will need to further investigate whether the mechanism behind the degradation of LOPs during aging could be attributed only to the cracking phenomenon or to the decrease in transparency of the encapsulated silicone [16,18,20]. Moreover, as the aging time increased from 700 to 1500 h, the relative LOPs of samples S1-S4 were 50.3, 45.3, 48.6, and 56.6% of their initial values, respectively. In this case, however, the rates of degradation of samples S1-S3 (−18.1, −16.6, and −17.5%, respectively) were significantly faster than that of sample S4 (−11.6%). Furthermore, the rate of degradation of sample S1 was faster than those of samples S2 and S3. In contrast, the degradation of samples S2 (−13.9%) and S3 (−17.1%) was accelerated in comparison with that of samples S1 (−5.6%) and S4 (−3.4%) during operation from 1500 to 3500 h. At the end of aging (after 3500 h), the absolute LOPs of samples S1-S4 were 2.94, 2.67, 2.87, and 3.41 mW, respectively. Thus, the relative LOPs of samples S1-S4 had decreased to 47.5, 39, 40.3, and 54.7% of their initial values, respectively.  In addition, in the absence of attached quartz glass, we observed no significant cracks in the encapsulated silicone filling the cavity of the AlN-DPC-LF over an aging time of 2000 h. Similar results during 1000 h of operation at 40 mA have been discussed previously [27]. Thus, excluding any mechanisms responsible for degradation in the UVB-LED chips, it might be possible that the silicone present between the UVB-LED chip and the attached quartz glass had undergone serious deterioration as a result of thermal stress and concomitant decay by UVB irradiation in samples S1-S3. The induced thermal stresses in samples S2 and S3 were more obvious than that in sample S1 due to the silicone touching the attached quartz glass. Thus, the degradation of the silicone in samples S2 and S3 was faster than that in sample S1, despite a lack of cracks prior to the aging time reaching 700 h. During the aging period from 700 to 1500 h, we suspect that the rate of degradation of sample S1 was faster than that of samples S2 and S3, because the appearance of cracks in the silicone of sample S1 was obviously retarded relative to that in samples S2 and S3. During operation from 1500 to 3500 h, the rates of degradation of all four of the samples decreased simultaneously. In other words, the influence of the cracked encapsulated silicone on the rate of degradation became a minor factor. An alternative mechanism might be that the silicone was slightly destroyed by the UVB irradiation. As displayed in Figure 3, however, the LOPs decreased noticeably between the onset of operation and 200 h, but the degradation slowed thereafter during operation between 200 and 3500 h. Such time-dependence of the LOPs of AlGaN-based LEDs has been reported previously, but without considering the effect of an encapsulant [28][29][30]. The mechanism behind this faster degradation rate is usually associated with diffusion of the metal contact or dopant atoms in the AlGaN-based LED chip. Interestingly, while the LOPs decreased upon increasing the aging time in this study, no significant changes occurred in the peak wavelengths or FWHMs after 3500 h of aging, within an accuracy of ±1 nm, in the EL spectra (Figure 5a-d). This behavior suggests that the properties of the QWs did not change over time. Nevertheless, the effect of aging on sample S4 in Figure 3, without any encapsulating silicone, followed neither an exponential nor a square root timedependence. Figure 6 presents the time-dependence of the forward voltages, measured at a value of I f of 100 mA, for all of the samples during the aging tests. After 48 h aging, the forward voltages of samples S1-S4 decreased by 4.2, 2.7, 4.7, and 4.4%, respectively, from their initial values. Thereafter, the forward voltages remained more stable during operation between 48 and 3500 h, whether silicone was encapsulated or not. To separate the influence of the encapsulated silicone, two possible mechanisms have been suggested as being responsible for the degradation processes occurring in UVB-LED chips [30][31][32][33]. A possible scenario is that the number of non-radiative centers increased at the onset of operation due to an increased junction temperature, causing more rapid decay at the early stage; thereafter, a slow proliferation rate was gradually achieved, after saturating the number of non-radiative centers, through the other mechanism in response to thermal stress inside the UVB-LED chip. In other words, from our investigation of the degradation of LOPs during long-term operation, we suggest that, when encapsulating silicone in the cavities of the LFs when packaging UVB-LEDs, the hermetic cover of the proposed package structure should be removed to avoid cracks forming in the encapsulated silicone. Moreover, excluding any influences of degradation processes in the UVB-LED chips, we suggest that the main mechanism responsible for the different rates of degradation of the LOPs in these proposed UVB-LED packaging structures involves the encapsulated silicone undergoing slight destruction in response to thermal stress and UVB irradiation without cracks appearing in the encapsulant. If cracks did appear in the encapsulated silicone, we attribute the slow rate of degradation to the silicone being further deteriorated by the UVB irradiation during the aging period.

Conclusions
We have found that an AlN-DPC LF featuring encapsulation with durable silicone between the surface of an AlGaN-based UVB-LED chip and a quartz glass cover can enhance the LEE relative to that of the conventional packaging lacking the encapsulating silicone. The maximum enhancement in the LOP was 13.8% for the packaging type S1, in which the encapsulated silicone was positioned above the surface of the UVB-LED chip, but without a gap between the silicone and the quartz glass cover. In another packaging type (S2) in which the encapsulated silicone fully filled the cavity of the AlN-DPC LF that included an attached quartz glass as cover, the improvement in the LOP was 11.7%. The enhancement in the LOP was only 3.2% in the packaging type S3, in which a gap existed between the surface of the UVB-LED chip and the cover. Reliability tests performed over a period of 3500 h at a value of I f of 100 mA revealed that the LOPs of these three proposed packaging types decreased to 45.3, 48.6, and 50.3%, of their initial values, respectively; cracks had appeared in the encapsulated silicone after operation over an aging time of 1500 h and were evident in photographs taken after aging for 3500 h. Nevertheless, the different rates of degradation were not obviously correlated with the phenomena of cracks appearing in the encapsulated silicone during such long-term operation. From an investigation of the degradation of the LOPs, we suggest that, if silicone is used to encapsulate the cavity of the LF when packaging UVB-LEDs, the hermetic cover should be removed from the proposed package structure to avoid cracks occurring in the encapsulated silicone. Moreover, excluding any influences of degradation processes in UVB-LED chips, we suggest that the main mechanism responsible for the slow rates of degradation of the LOPs in these proposed UVB-LED packaging structures during longer periods of operation, after cracks had appeared in the encapsulated silicone, involves the silicone being further deteriorated by the UVB irradiation.