Degradation Analysis of Silicone as Encapsulation and Molding Material in High Power LEDs

Abstract

Applications of LEDs have increased significantly, and increasing outdoor applications are observed. Some outdoor applications require high reliability as their failure can lead to hazardous consequences. Examples are their applications in automotive, street lamp lighting etc. To ensure the reliability of LEDs in outdoor applications, reliability test that include humidity on the LEDs must be done. However, it is found that accelerated life test of LEDs at high humidity level cannot be extrapolated to standard condition of lower humidity as the mechanism of degradation depends critically on humidity level. In fact, the degradation of LEDs in outdoor applications is mainly due to the degradation of their encapsulation and housing materials (or called packaging material as a whole) instead of the semiconductor chip itself. The decrease in lumen is the results of crack and discoloration of the LED packaging material. Detail understanding of the failure physics of the packaging material for LED under humidity is needed for extrapolation performed at accelerated stress condition so that LED luminary reliability can be predicted. This chapter reviews the different types of degradation physics of the packaging material using ab-initio simulation with excellent verification from experiments. The method of extrapolation is therefore derived from the physics-based model after detailed understanding of the degradation physics of LEDs. The model also provides strategy for industry to prolong the usage of LEDs in outdoor applications, either through materials or operating conditions selection.

1. Introduction

Light is one of the most essential natural elements that has helped humans since early ages, and our dependency on light is very high. With the development of artificial modes of electrically powered lighting and heating devices for indoor as well as outdoor applications, the demand of electrical power now more than ever. About one-fifth of the total power is used for lighting [1]. To reduce power usage while maintaining the standard of complexity in lifestyle as well as advancement of technologies, the search for reliable and cost-effective light sources is in demand more than ever.

With the discovery of high power light emitting diode (LED), LEDs have become the go-to sources for applications in lighting systems. Due to their prolonged lifetime, low maintenance, small sizes and energy efficient properties, LEDs have replaced most of the conventional sources of lighting such as incandescent and fluorescent lamps all over the world. The technology used in LEDs is solid-state lighting (SSL) which provides more energy saving than conventional fluorescent lamps. The comparative study between efficiency and power output of LEDs and other conventional sources of lighting is shown in Table 1 [1, 2].

Light sources	Illumination efficiency (1 m/W)	Lifetime (hours)
Incandescent	15–20	1000
Tungsten Halogen	12–35	2000–4000
Mercury Vapor	40–60	12,000
Lamps Compact Fluorescent	40–70	6000–12,000
Lamp Fluorescent Lamp	50–100	10,000–16,000
Induction Lamp	60–80	60,000–100,000
High Pressure Sodium	80–100	12,000–16,000
LED	80–160	50,000–100,000

Table 1.

Lighting efficiency and lifetime of some light sources [1, 2].

Apart from energy efficiency, LEDs also have a longer lifetime compared to fluorescent and incandescent lamps. These devices are compact and have higher mechanical resistance compared to conventional sources. The report by Guan et.al showed nanowire LEDs that were able to bent for −2.5 mm to +3.5 mm radii of curvature while the light intensities was consistent throughout the bending cycle tests [3]. If all the conventional light sources in the world were converted to the LEDs, energy consumption could be reduced by 230 typical 500-MW coal plants, and greenhouse gas emission can then be reduced by about 200 million tonnes [4]. Owing to these benefits, high-power LEDs are now not only used to replace incandescent and halogen lamps in houses, they are also used in outdoor lighting applications such as displays, automobiles, traffic signals, lamp posts, emergency medical services, marine applications, etc. [4, 5, 6, 7, 8, 9, 10, 11, 12]. Unfortunately, degradation in LEDs will not only exhibit reduced light output but also change in color which can affect some of their applications. For example, a dimmed LEDs in traffic light might cause road accident to happen. Sudden blackout due to the failure of LEDs in lamp posts can create danger to the pedestrians, and these have indeed resulted in lawsuits against the manufacturers [13, 14, 15, 16]. Thus, alongside with energy efficiency and conversion, the reliability of LEDs is of the utmost importance [17], especially our dependence on the lighting is so high.

The environmental stresses for LEDs in outdoor applications, such as humidity and temperature can lead to degradation of output light from LEDs. While reliability tests of LEDs can help to evaluate if they are suitable for various applications, a detailed understanding of the degradation physics of LEDs can provide the root causes of the unexpected degradation of LEDs so that early detection monitoring can be possible, and their lifetimes can also be enhanced for increased effectiveness in energy saving.

The LED lifetime varies somewhere between 50,000 and 70,000 hours [1, 2, 17]. The LED lifetime is measured by means of luminous flux (lumen) maintenance, which shows the fall of lumen of LEDs over time. The Alliance for Solid-State Illumination Systems and Technologies (ASSIST) has defined LED lifetime based on the time to 50% light output degradation (L50: for the display industry approach) or 70% (L70: for the lighting industry approach) light output degradation at room temperature [18, 19].

However, evaluation of lumen degradation at standard operating conditions for LED lifetime assessment is not feasible as the costs of the test and time are too high. To expedite the reliability test of LEDs, accelerated life tests are used to predict the LEDs lifetime and reliability while maintaining exact failure mechanism between operating and accelerated conditions so that extrapolation can be performed. The most common industrial standards used as qualification tests for LEDs are JEDEC and JEITA standards [20]. These standards include reflow soldering test, thermal shock test, temperature cycle test, moisture resistance cyclic test, high temperature storage test, temperature humidity storage test, low temperature storage test, vibration test, and electro-static discharge test.

The degradations of high-power LEDs are mostly due to three external stresses, namely temperature, humidity and electrical stress [21]. A large number of investigations are done to understand and examine their degradation physics under thermal and electrical stresses. On the contrary, the investigations on the effect of humidity related degradation mechanisms is quite low as reported by Singh & Tan [21]. Previously, it was believed that thermal and electrical stress are the most dominant factors in lumen degradation of LEDs. However, when the applications of LEDs are in outdoor or harsh environmental applications, it was found that instead of the LED chip, the materials of LED encapsulation and molding which is comprised of different silicone polymers degrade in presence of humidity that render major lumen degradation of LEDs in their applications. In this chapter, a detailed study of the dynamics of the degradation physics of silicone polymers in LED encapsulation and molding, and their effect on total lumen degradation of LEDs will be presented.

2. Identification of humidity related failure mechanisms in high-power LEDs

The failure mechanisms induced due to humidity or penetration of moisture into LED packages can be classified mainly into two categories: a) Degradation of the Chip which is further split into semiconductor dice and phosphor; b) degradation of the LED package which includes encapsulation of the LED chip or LED chip + phosphor, LED molding and the die attach with heat sink. A schematic diagram of an LED structure is shown in Figure 1.

Figure 1.

Schematic of moisture penetration into a typical LED structure. The red arrow represents the possible moisture diffusion paths into the LED package.

2.1 Degradation of the chip including phosphor layer

To characterize the chip level degradation in the case of white phosphor LEDs, the computation of the blue to Yellow ratio (BYR) of the output light, and ideality factor (n), Reverse saturation current (I_s) and series resistance between bond wires with chip and bond pads (R_s) are considered [22]. A method of analyzing the n, Is and R_s of the LEDs from the LED forward characteristics curve was developed by Tan et al. which was adopted for all such analysis mentioned hereafter [23]. The analysis of BYR is done in such a way that if the BYR decreases with time, it means the blue light of the LED chip has degraded indicating semiconductor dice degradation. On the other hand, if BYR increases, the yellow light conversion from phosphor decreases indicating the phosphor degradation. This phenomenon is further analyzed by observing the percentage change in the ideality factor (n) where we expect n to go out of the normal range of 2.0–7.0 for chip related degradation [23]. For Blue LEDs without phosphor layer, all the above-mentioned characterizations except BYR are done.

It has been seen that moisture penetration from ambient through the LED package has caused serious lumen degradations. This happened either due to the degradation of the chip or degradation of phosphor or both, and degradation of the silicone encapsulation and molding which will be discussed in the next section. An example of the degradation of the chip due to moisture penetration that reduces the lifetime of the LED chip where its BYR drops to half of its original value at 20% rapidly was reported in Ref.s [22, 24]. The intensity of the emitted blue light degrades over time. Upon further analysis of the cause of failure of the semiconductor dice, it was found that the moisture trapped in the die attach or in the electrical contact pad vaporized and the increase in vapor pressure resulted in the crack of the GaN based die [24, 25].

An example of degradation due to phosphor is the work by Tan et al. [24, 25, 26]. The BYR was found to increase for some samples which indicated the degradation of phosphor in LEDs. It was found that moisture trapping inside the LED package has caused the dissolution of phosphor which resulted in the overall lumen degradation.

2.2 Degradation of LED package

It is noteworthy that the humidity test reported until now has a metal reflector below the chip and the effect of moisture on the LED degradation can be identified as drastic [24, 25]. However, with the maturation of chip technology against humidity and the replacement of reflector molding material by the mixture of silicone polymers in high-power LEDs, question on its humidity reliability is re-visited.

To address this question, high temperature-humidity tests at 85°C/85%RH was conducted on OSRAM LEDs according to IPC/JEDEC standard J-STD-020D (2007) [19, 20]. Such test was performed by Tan and Singh [26, 27, 28], but they included one more test parameter, namely the drive current of the LEDs, and such reliability tests were termed as moisture-electrical thermal (MET) tests.

The test was performed on two sets of white (with phosphor) and blue (without phosphor) LEDs where one set of each type was kept ON with 350 mA constant current and the other set was turned OFF.

A significant difference in degradation between the ON and OFF sets of each type of LEDs was found as seen in Figure 2. The white LEDs in the OFF condition did not experience any lumen degradation which is in agreement with reports from Tan et.al [22, 25, 26] where phosphor acts as a protective layer for the LED. However, in the ON condition, white LEDs experience 33% lumen degradation in a span of 144 hours of testing. On the other hand, a difference of 5% lumen degradation was observed between the ON and OFF blue LEDs and the difference increase to 18% after the tests was concluded for each case. It was also found that the BYR of the fresh and degraded LEDs exhibited insignificant changes which omits the possibility of chip degradation as mentioned in the previous section.

Figure 2.

Percentage lumen degradation vs. test time for blue and white LEDs tested under different operating conditions. (a) Blue LEDs under power-on condition; (b) blue LEDs under power-off condition; and (c) white LEDs under power-on condition. Red line (dotted line) denotes the initial degradation; green line (dash line) indicates the lumen recovery and the blue line (solid line) indicates the final degradation for the LEDs [26].

Upon closer examination of Figure 2, a sharp degradation of lumen is observed, and it is attributed to the moisture entrapment in the silicone encapsulation of the LEDs. After moisture penetrates through the LED encapsulation, two phenomena are most likely to happen viz. trapping of moisture in silicone encapsulation which causes output light scattering and reduced lumen output; and the diffusion of moisture to the LED chip including phosphor, thereby causing damage in the chip level and resulting in reduced lumen output as discussed in the earlier section. However, such rapid degradation might also have followed by a recovery if the above two damages were not happened. Furthermore, upon further investigations, it was concluded that in order to prevent such moisture penetration, the pore size of the silicone encapsulation must be lower than 50 nm. A detail description of the work can be found in Ref. [22].

3. Failure analysis of humidity reliability of LEDs at OFF state during MET test

Singh et al. identified three stages of degradation mechanisms during the MET test which has been characterized in Table 2. The amount of moisture adsorbed by the LED encapsulation and die attach has also been studied by Singh et.al as shown in Figure 3 which provide good insights in understanding the degradation mechanism of the LEDs. As the degradation mechanisms of LEDs due to moisture for LEDs at ON and OFF conditions are different as evident from the experimental finding, let us discuss the failure mechanism for the LEDs at OFF condition.

Stages	Blue LED ON	Blue LED OFF	White LED ON	White LED OFF
Initial Degradation	12% degradation	8.5% degradation	Nil	Nil
Recovery	7.5% degradation	1% degradation	Nil	Nil
Final Degradation	30% degradation	12% degradation	33% degradation	Nil

Table 2.

Stages of degradation in LEDs during 85/85 test at ON and OFF conditions [26].

Figure 3.

Amount of moisture penetration into LED package and die attach during testing under OFF condition. The results are computed using finite element analysis as described in the text [26].

It was found that at the beginning of the test, moisture adsorption in the LED encapsulation is much higher than the die attach area and saturates around 132 hours of testing as shown in Figure 3. This initial degradation stage is attributed to moisture adsorption in the encapsulant which causes a “cloudiness” or light scattering effect as mentioned in Ref. [25]. The moisture trapped in the silicone encapsulation causes discoloration of the encapsulant material attributing to the “cloudiness” effect which cause lumen degradation. However, for white LEDs the phosphor layer prevents moisture penetration into the die attach region as well as the heat generated from the phosphor during lumen measurements help to evaporate some moisture in the encapsulant. Hence, no significant lumen degradation was observed.

As the test proceeds over 132 hours in OFF conditions, moisture entered the die attach material that caused subsequent die attach delamination which in turn increased the thermal resistance of the die, renders degradation in the overall lumen output of the LEDs. Furthermore, the moisture in the die attach cannot be evaporated during testing [22, 25], and this renders significant lumen degradation. It was found that moisture penetration is 40% higher in blue LEDs as compared to their white counterparts and this excess moisture is housed in the die attach material, therefore when the blue LEDs were tested for 336 hours, die attach delamination has already occurred [26].

For better understanding of the diffusion of moisture in LED degradation, finite element method (FEM) simulation was performed. The three most important parameters considered for such modeling are the moisture diffusivity D which is a function of temperature, moisture saturation concentration C_sat and moisture absorption and desorption times. The moisture diffusion in water permeable body is given in Eq. (1) as (Table 3) [30].

Property	Thermal	Moisture
Primary Variable	Temperature, T	Wetness, W
Density	ρ (kg/m^3)	1
Conductivity	k (W/m*^0C)	D. Csat (kg/s*m)
Specific heat	c (J/kg*^0C)	Csat (kg/m^3)

Table 3.

Correspondence table for thermal/moisture analogy [29].

$$\frac{\partial C}{\partial t}=D(\frac{{\partial }^{2}C}{\partial x^2}+\frac{{\partial }^{2}C}{\partial y^2}+\frac{{\partial }^{2}C}{\partial z^2}$$E1

A term W, called normalized concentration or wetness was introduced so that thermal modeling can be applied for moisture modeling as moisture concentration may not be continuous across an interface of two materials, unlike temperature. W is given in Eq. (2) as follows [29]:

$$W=C/C_{sat}$$E2

where C is the moisture concentration (kg/m ³) in a material.

It was found that during the absorption cycles, the moisture concentration was highest near the die-attach region and during the desorption cycle, the moisture concentration was high in the silicone than ambient. FEM results also indicated that moisture can reach the die attach under 24 hours of test [31, 32].

Tan et.al [28] used FEM to understand the experimental phenomena of initial rapid light output degradation followed by a recovery. The “recovery” is the result of moisture absorbed by die attach material that suck the moisture from the silicone. Thus the “recovery” of lumen degradation is actually associated with the degradation in the internal structure of the LED package which is not reversible.

4. Failure analysis of LEDs at ON state during MET test

From Table 2, it is obvious that the LEDs under power-on condition degrade at a faster rate than the one under power-off condition as expected. Since the LEDs at ON state generate heat and are able to reach a temperature of 135°C, the silicone in the encapsulation and molding start to expand due to excess heat. As the material compositions for encapsulation and molding materials (even though both of them are silicone) are different, their difference in thermal co-efficient of expansion led to interfacial void, cracks and delamination as found by Singh et.al [26]. These delaminated interfaces produce gap for moisture to penetrate preferentially into the LED packaging, causing lumen degradation for both white and Blue LEDs.

The Blue LEDs however were found to undergo a recovery phase of light output after 24 hours of the test and the lumen started degrading again after 48 hours. This phenomenon was studied extensively using FEM simulations and experimental verification through confocal scanning acoustic microscopy (c-SAM) [28]. It was found that in the first 24 hours the moisture trapped at the LED encapsulation caused its discoloration and cloudiness which led to a sharp lumen drop of 12%. However, as the LEDs were turned ON, moisture seeping into the die attach occurred thereby exhibiting a recovery stage during the LED test. After 48 hours, delamination in the encapsulation-molding started to occur due to the heat generated by the LED chip. It led to a higher flux of moisture inside the LED package causing constant lumen degradation hereafter.

In case of White LEDs, the phenomenon of reliability recovery was not found as in the case of blue LEDs, and a total degradation of 33% occurred continuously till 144 hours. To further understand the root cause of this degradation, optical microscope imaging was used to observe changes in LEDs after every 24 hours of testing. The optical images revealed heavy discoloration of the LEDs which was suspected to be in the silicone encapsulation and molding part of the LEDs as shown in Figure 4 [33].

Figure 4.

Optical micrograph of silicone discoloration in (a) white LEDs tested for 144 hours, (b) blue LEDs tested for 356 hours respectively when compared to (c) fresh LEDs. (d) a closer look of the top view of the degraded white LED. Encapsulant and molding parts are also detached from one another for detail examination. Optical micrographs at 500 X magnification of (e) encapsulant and (f) molding part of degraded white LEDs tested under 85°C/85% RH and ON condition for 144 hours [33]. (reproduced with permission from [33]).

The discoloration in the molding part and encapsulation of the white LEDs were found to be more severe as compared to blue LEDs due possibly to the excess heat generated by the phosphor layer during the light conversion process [33]. Destructive analysis of the LED encapsulation and molding parts revealed that cracks were observed in both components. However, the cause of such cracks in both encapsulation and molding parts of the LED package were different as reported by Singh et.al [33, 34].

4.1 Analysis of encapsulant degradations

The surface of the fresh LEDs was found to be smooth, while after degradation, cracks started to appear and flakes or tube like structures were observed on the encapsulation of the blue LEDs. The depth of cracks on the encapsulant was quite minimal where average depth was around 3 μm for blue LEDs and that for white degraded LED encapsulant, it is almost negligible and not able to measure using optical microscope [33].

Scanning Electron Microscopy (SEM) was employed to further investigate the surface of the blue and white degraded LED’s encapsulant and compared to a fresh LED encapsulant. SEM images also found cracks on the degraded encapsulants which was consistent with optical imaging results. The flake like structures were more evident on the blue degraded LEDs surface as compared to the white degraded LEDs. Both SEM and optical micrographs shows that the crack density and crack depth are higher in blue LED’s encapsulant than the white LED [33].

However, if the total rise in temperature of both blue and white LEDs are compared when LEDs are powered on at the same drive current, it can be easily found that the total temperature of the white LEDs are more. The higher temperature of white LEDs is attributed to mainly two reasons; the junction temperature of the LED junction in the die and the heat generated when phosphor layer is activated for conversion of blue to yellow light. Therefore, it is expected that, in comparison to the percentage lumen degradation of both blue and white LEDs, and taking into account the heat generation in white LEDs, the crack in the encapsulation of the white LEDs should be more than blue LEDs. However, this was not the case, and as reported above, the encapsulants of blue LEDs experienced severe cracks than the white LEDs.

Energy dispersive system (EDS) was employed to examine the crack surface of the degraded and fresh encapsulants to provide a better understanding of the nature of crack on the encapsulants. EDS results showed that the percentage increase in oxygen is higher in the case of the blue degraded LED encapsulant as compared to that of the white LEDs. A higher percentage in oxygen element could indicate that the surface of the material has undergone oxidation which can be through a series of chemical reactions. To narrow down the type of chemical reaction which has occurred in the encapsulant surface, FTIR spectroscopic analysis was employed. FTIR analysis indicated that the degraded encapsulant of both the LEDs contained –OH peaks which indicates degradation by hydrolysis. However, the OH peaks concentration was higher in case of blue LEDs than its white counterpart. This means that moisture accumulation in white LED encapsulants was lower compared to that in blue LEDs, and thus less severe hydrolysis of the white LEDs than blue. The moisture concentration of the white LEDs was lower due to excess heat generated by the phosphor layer that drives away the moisture from the surface of the LED encapsulant [33]. The degradation mechanism related to cracks in encapsulation was thus found to be hydrolysis of the silicone polymers of the LED encapsulant [33, 34].

4.2 Analysis of molding part degradations

The degradation mechanisms of the molding part of the LEDs have not been extensively investigated, to the best knowledge of the authors, and we now present detailed investigation of the white LED molding degradation in this section. It was found that alongside with the yellowing of the molding part, cracks also started to develop as the test progressed from 45 to 150 hours. Optical micrograph investigation of the crack depths revealed that the average crack depth was found at 105 hours of testing corresponding to 25–27 μm. When the test progressed towards 150 hours, depths began to decrease by an average of 10 μm. This indicates that different types of chemical changes in the silicone molding must have taken place to observe such changes [34].

This hypothesis was later confirmed by the reports from FTIR analysis where it revealed two sets of exothermic and endothermic reactions in the silicone molding part. It was found that in the initial phases of the test (0–45 hours), the moisture penetration towards the molding part of the LEDs caused the hydrolysis of the silicone polymers which is an endothermic reaction [34, 35, 36]. The endothermicity of hydrolysis was further confirmed by the drop in temperature in the LED packages as reported by Singh et al. [34]. The endothermic nature of the hydrolysis reaction was further confirmed by Shabir et.al using density functional theory (DFT) calculations where it was found that 460 kJ/mol of energy was absorbed during hydrolysis of the silicone polymers under study [35]. Hydrolysis reaction was immediately followed by condensation of the silicone polymers which is an exothermic reaction where a raise in temperature was also observed by Singh et.al and others [34, 37]. Shabir et al also verified that condensation is exothermic as DFT calculations show 545 kJ/mol of energy release during condensation reaction [35].

As seen from Figure 5, it is evident that in the initial 48–72 hours of test, discoloration of the molding part was prominent, and the development of cracks was negligible. DFT calculations on the hydrolysis and condensation of silicone polymers found that condensation reaction is highly spontaneous as compared to hydrolysis during this time. Also, due to the exothermic nature of condensation reaction, the energy released from condensation is continuously re-absorbed by the moisture molecules in the outer part of the molding, thereby increasing the rate of hydrolysis exponentially after some point of time [35]. It was hence found that the sudden increase in lumen degradation was due to simultaneous hydrolysis and condensation reactions as reported by Shabir et al. [35].

Figure 5.

Optical micrographs of white LEDs with varying test time intervals. The blue colored circle in the fresh sample represents the inner molding part close to LED chip periphery and the green colored circle represents the outer molding part which is far from the LED chip [34]. (reproduced with permission from [34]).

As the temperature rises in the sites of the molding part where condensation has occurred, the decrease in Si-O peak and the increase in H₂O peaks indicates that a different chemical reaction occurs where the involvement of H₂O is reduced. It was found that silicone network crosslinking can occur producing silicone ring-like structure due to thermal oxidation at temperatures of 0-200°C which is exothermic in nature [34, 38, 39]. The occurrence of thermal oxidation in those sites after condensation at 48 hours of test is believed to lead to crack depth formation which is the dominant degradation mechanism until 105 hours.

The phenomenon of crack recovery after 105 hours was seen. Silicone loses its elasticity after degradation especially in harsh environmental conditions as observed by many researchers in their studies [40]. Thus, possibility of molding part’s crack recovery due to its elastic nature is not valid. Another possibility is due to the diffusion of low molecular weight species such as Si-O (also known as Si oligomers or silica nano-fillers) from the interior of silicone to its surface as it could be invoked by the high temperature of the silicone due to thermal oxidation. The Si oligomers cover the cracks area, renders an apparent crack depth recovery [39]. This is also observed as an increase in Si-O peak in FTIR results as seen in Figure 6. EDS results also supports this theory showing an increase in the Si atoms after 45 h of testing appear to confirm the presence of Silicon oligomers. In other words, as cracks are generated due to thermal oxidation, Silicon oligomers can now diffuse out from the crack surfaces, filling the cracks and thus the cracks depth decrease.

Figure 6.

Overall peaks intensity variation for different samples with varying test time [34]. (reproduced with permission from [34]).

As the LED package degrades, the temperature of the device also increases as shown in Ref. [34]. At higher temperature, thermal aging of silicone occurs as evidenced by the increase in Si-CH₃ molecules in the FTIR results. Although thermal aging is an endothermic reaction as thermal energy is required to break the chemical cross-linking bonds in the silicone structure, the temperature of the LED does not decrease when thermal aging step in. This could be due to the degradation at either LED chip site or at the die attach as observed by many [24, 26] which can result in additional temperature rise and thus offset the decrease in temperature due to thermal aging reaction. In fact, this additional temperature rises at the LED chip or die attach sites will lead to rapid diffusion of oligomers in the region closer to the LED chip than the area far away from the chip region, and this is indeed observed by Singh et al. [34] where higher amount of crack depth recovery is observed for inner region cracks when compared with outer region cracks from 105 to 150 h duration.

However, the bond dissociation energy of Si-O-Si bond varies between 12 and 162 KJ/mol as reported in many literatures [41, 42, 43, 44, 45, 46]. So, if we consider the energy of ambient temperature of 85°C, it only corresponds to 2.97 kJ/mol. If the LED temperature is also added with the LEDs powered ON, it only corresponds to 3.89–5.7 kJ/mol. It is therefore clear that the energy from temperature cannot be the sole driving force for such kinds of degradations.

To provide detailed understanding of the driving force of the degradation mechanisms as mentioned above, Shabir et.al performed ab-initio calculations on the hydrolysis and condensation of silicone polymers [35]. They found that the activation energy E_a of hydrolysis of silicone was 123.2 kJ/mol while that for condensation it was −6.9 kJ/mol. The E_a of hydrolysis computed in this study was found to be in excellent agreement with the E_a of hydrolysis of silicone polymers present in the LED molding experimentally. The driving force of such reactions was found dominantly from the mixture of blue and yellow light emitted from the LEDs during the test [35]. This also agrees excellently with the test reports where white LEDs in OFF conditions exhibited no lumen degradations as mentioned earlier [26, 28].

As LEDs in outdoor applications are designed to perform for longer periods of time, lifetime prediction of LEDs is done with accelerated stress levels of temperature and humidity know as accelerated life tests (ALTs), and then lifetime at such conditions are extrapolated to estimate their lifetimes under normal operating conditions. The underlying idea of such extrapolation is that the devices under test should follow the same degradation mechanisms in both accelerated and normal operating conditions. The common acceleration models used for reliability test of LEDs are the Peck models which includes the extrapolation of test data from 85%/85°C to normal operating conditions and the Arrhenius equation is included for temperature effect therein. However, it was found by Tan et.al that the degradations are completely different for humidity level varying between 70% RH, 85% RH and 95% RH [24]. This renders the inapplicability of the Peck model for extrapolation of test data to normal operating condition for LEDs. Peck model also does not consider the effect of light energy on the degradation of LEDs. On the temperature extrapolation, Arrhenius equation is commonly used, and the assumption of constant activation energy is made, which indicate that the degradation mechanism is invariant as it is the basis of extrapolation as mentioned earlier. However, our study also found that this is not true. Hence there is a need for a new model which quantitatively covers all the factors for silicone degradation, and this model was developed by Shabir et.al from detailed understanding of DFT simulations and experimental data as well as the of physics of degradation from various reports [35]. The new model related the growing of the area of discoloration A_d with time (t) is given in Eq. (3) below.

$$A_d\left(t\right)=A_0+A_2{e}^{A_{1}t}$$E3

where

$$A_1=\frac{{{\displaystyle \sum }}_{i=1}^2{\beta }_i{P}_i}{\alpha }\text{;}$$E4

P₁ represents the power density of the blue light and P₂ represents the power density of the yellow light. The corresponding β represents the absorption coefficient of the different lights by the discolored region, respectively, which was assumed to be constant in this work. As the discolored region is dark brown in color, it was considered that all the blue light is absorbed, i.e., β₁ = 1.

$\alpha =\left(k_1{E}_a-\frac{k_{B}T}{V}-{\alpha }_{\mathit{condensation}}{\beta }_3\right)\delta$; Here, δ is the effective penetration depth of light. At a given ambient temperature, α is approximately a constant, E_a is the activation energy, k_B is Boltzmann constant, T is temperature and V is volume of the silicone molding the meaning of other parameters can be found in Ref. [35].

A₂ is related to the quality of the silicone. Specifically, if the silicone has lager pore size or trapped particles that can absorb light, the value of A₂ will be larger; $A_0=-A_2\text{.}$

The model fits very well to the experimental data at different light intensities and drive currents as shown in Figure 7. It also predicts no degradation if no light is present which was the case of white LEDs with power OFF during testing as reported by Singh et.al [26]. The model allows us to predict LED molding discoloration in the future and simultaneously quantify the quality of the silicone material of the LED molding, which is a factor that will affect the variation of the degradation time of LEDs.

Figure 7.

Model fitting with experimental data for 350 mA and 300 mA drive currents [35]. (reproduced with permission from [35]).

5. Conclusion

Reliability tests and causes of degradation have become important criteria of focus for large scale product manufacturing companies. Proper reliability test and analysis of the related cause of degradation shall help increase the product lifetime and market value. Hence, a descriptive review on the various degradation analysis of LEDs is presented here.

After detailed analysis using experimental and DFT techniques, it was found that the degradation of the encapsulant and molding part material, both make of silicone, are the primary sources of lumen degradation. The inapplicability of Peck’s model was also found as it failed to factor-in the primary driving force of silicone degradation which is the light emitted from the LED chip itself. A new model for prediction of silicone degradation was hence developed which determines the degradation of silicone as a function of time, considering all its driving forces. This model allows prediction of the quality of silicone used for LED packaging which in turn can improve the LED reliability. It can also help to predict the rate of discoloration of the LED which relate to its lumen degradation.