This page covers the degradation of light-emitting diodes: InGaN/GaN blue and phosphor-converted white LEDs, AlInGaP red LEDs and AlGaN ultraviolet (UVB and UVC) LEDs. It covers the gradual loss of light output and its projection, colour shift of white LEDs, the thermal and current acceleration of chip degradation, hydrogen and defect diffusion in the active region, and package effects such as encapsulant browning. The device physics is on the Light Emitting Diode page, the junction-temperature models on the Thermal Stress page and the lifetime statistics on the Failure Rate page; luminaire drivers and optics are outside the scope of the vault.

1. Why It Happens

Trigger: Drive current and junction temperature (Vazquez et al.; Glaab et al.); for UVC LEDs mainly the current, since temperature had no significant effect at 233 $nm$ (Glaab et al.); temperature alone for the package, which browns even without current (Vazquez et al.).
Mechanism, defects in the active region: Non-radiative recombination centres are created in or near the quantum wells and raise the SRH coefficient, often by recombination-enhanced defect generation; in UVC LEDs Wen et al. place them in the n-AlGaN below the wells (Casu et al.; Wen et al.).
Mechanism, hydrogen: Hydrogen diffuses from the p+ layer and passivates Mg acceptors, which lowers hole injection, raises the turn-on voltage and the resistance, and follows a square-root-of-time law (Meneghini et al.); in far-UVC LEDs a loss of hydrogen near the active region accompanies the early drop (Glaab et al.).
Mechanism, dislocations: In UVC LEDs on AlN/sapphire, defects activated during operation scale with the threading dislocation density (Ruschel et al.).
Mechanism, package: Epoxy browning, accelerated by temperature (Vazquez et al.); package-dependent colour shift of phosphor-converted white LEDs (DOE).
What sets the rate: The junction temperature, with 1.2 to 1.5 $eV$ for red LEDs (Vazquez et al.); the current, with an inverse-square law for red LEDs and a third-power loss for far-UVC LEDs (Vazquez et al.; Glaab et al.); the dislocation density (Ruschel et al.); the package type (DOE).
Failure criterion: The output falling to 70 per cent of its initial value, $L_{70}$ (TM-21; Vazquez et al.); a colour shift beyond the specification of the luminaire (DOE; background knowledge).

2. Where to Look

Location: The quantum wells and the layers next to them, the p-side first (Casu et al.); the n-AlGaN under the wells of UVC LEDs (Wen et al.); the p-GaN and its contact for hydrogen effects (Meneghini et al.); the encapsulant and the phosphor for package effects (Vazquez et al.; DOE).
Signature: An early rise of output (annealing) followed by exponential decay (Vazquez et al.); a stronger loss at low current, a higher ideality factor and more reverse leakage when defects grow (Wen et al.); a higher turn-on voltage and resistance with hydrogen passivation (Meneghini et al.); yellowing of the package or a colour shift whose direction depends on the package (DOE).
Confirm with: LM-80 lumen maintenance at several case temperatures and the TM-21 projection (TM-21); L-I and I-V curves before and after stress, with the slope of the log-log L-I curve (Meneghini et al.); cathodoluminescence and low-frequency noise (Wen et al.); unbiased high-temperature storage to separate package from chip effects (Vazquez et al.).
Look-alikes: Package browning, which degrades unbiased LEDs too, against chip degradation, which needs current (Vazquez et al.); a light loss from a higher junction temperature through a worse thermal path, which recovers when cooled (background knowledge).

3. Prevention

Design: Keep the junction cool and the current within the rating, for example below 200 $mA$ for mid-power packages and $T_{J} \leq 120$ $^{\circ} C$ for CSP LEDs (DOE); improve cooling, which lowered the thermal resistance by 15 to 20 per cent with forced convection (Vazquez et al.).
Material and process: Lower the dislocation density of UVC templates (Ruschel et al.); control hydrogen in the p-layers (Meneghini et al.; Glaab et al.); use packages and encapsulants that resist browning, such as ceramic or silicone instead of epoxy (Vazquez et al.; background knowledge).
Operation: Run UVC LEDs at moderate current, since the loss scales with the third power of the current (Glaab et al.).
Screening and test: LM-80 with at least 20 units for the 6 times projection limit and the TM-21 data window (TM-21); accelerated life tests at several currents and temperatures with Weibull analysis (Vazquez et al.).

Lumen Maintenance and Colour Shift

Name	Formula	Related Formulae	Variables/Constants
Lumen maintenance projection (TM-21)	$Φ (t) = B e^{- α t}, L_{p} = \frac{l n ( 100 B / p )}{α}$	$L_{p} \leq 6 D (n \geq 20), L_{p} \leq 5.5 D (10 \leq n \leq 19)$	$Φ (t) : n or ma l i z e d l u min o u s f l ux$ $B : p r o j ec t e d ini t ia l co n s t an t$ $α : d ec a y r a t e co n s t an t$ $L_{p} : t im e t o p p er ce n t o f t h e ini t ia l f l ux (L_{70} f or p = 70)$ $D : L M - 80 t es t d u r a t i o n$ $n : n u mb er o f u ni t s t es t e d; T M - 21 (M i l l er) : d a t a b e l o w 1000 h a r e e x c l u d e d, t h e f i t u ses t h e l a s t 5000 h f or 6000 t o 10000 h o f d a t a an d t h e l a s t 50 p er ce n t ab o v e 10 000 h, f e w er t han 10 u ni t s a r e n o t p r o j ec t e d, a l imi t e d r es u l t i s w r i tt e n w i t h^{'} >^{'}, f or e x am pl e L_{70} (6 k) > 36000 h, an d t h e co n f i d e n ce ban d i s o n e - s i d e d a t 90 p er ce n t; w or k e d e x am pl e w i t h 20 u ni t s an d 6000 h : B = 1.023 an d α = 1.042 \times 1 0^{- 5} h^{- 1} a t 55^{\circ} C c a se, B = 1.027 an d α = 1.598 \times 1 0^{- 5} h^{- 1} a t 85^{\circ} C$ $(co m p u t e d : L_{70} = 36412 h a t 55^{\circ} C (s l i d es 36392) an d 23987 h a t 85^{\circ} C (s l i d es 23 997); s in ce 36412 e x cee d s 6 x 6000 h, t h e 55^{\circ} C r es u l t m u s t b e r e p or t e d a s L_{70} (6 k) > 36000 h)$
Temperature interpolation of the decay rate	$α (T) = A e^{- E_{a} / (k_{B} T)}, \frac{E _{a}}{k _{B}} = \frac{l n ( α _{2} / α _{1} )}{1/ T _{1} - 1/ T _{2}}$	$L_{p} (T) = \frac{l n ( 100 B _{0} / p )}{α ( T )}$	$α (T) : d ec a y r a t e a t t h e in - s i t u c a se t e m p er a t u r e T (ab so l u t e)$ $A : p r e - e x p o n e n t ia l f a c t or$ $E_{a} : a c t i v a t i o n e n er g y$ $k_{B} : B o l t z mann co n s t an t (8.617 \times 1 0^{- 5} eV / K)$ $α_{1} an d α_{2} : d ec a y r a t es a t t h e b r a c k e t in g L M - 80 t e m p er a t u r es T_{1} an d T_{2}$ $B_{0} : in t er p o l a t e d ini t ia l co n s t an t; T M - 21 : u se t h e A r r h e ni u s in t er p o l a t i o n o n l y i f b o t h α a r e p os i t i v e, d o n o t e x t r a p o l a t e ab o v e t h e hi g h es t t es t t e m p er a t u r e, u se t h e l o w es t t es t t e m p er a t u r e b e l o w t h e r an g e; s l i d es : in t er p o l a t e d B = 1.025 an d L_{70} ab o u t 29277 h a t 70^{\circ} C, w i t h E_{a} / k_{B} g i v e n a s ab o u t 5167.5 K$ $(co m p u t e d : t h e tw o α v a l u es g i v e E_{a} / k_{B} = 1675 K, E_{a} = 0.144 eV, an d α (70^{\circ} C) = 1.303 \times 1 0^{- 5} h^{- 1}, w hi c h r e p r o d u ces 29279 h; 5167.5 K w o u l d g i v e 18388 h, so t h e s l i d e v a l u e o f E_{a} / k_{B} i s in co n s i s t e n t w i t h i t s o w n r es u l t (m y c h ec k))$
Chromaticity shift of white LEDs	$Δ u^{'} v^{'} = (u^{'} - u_{0}^{'})^{2} + (v^{'} - v_{0}^{'})^{2}$	$Δ u^{'} v^{'} \leq 0.007 (typical lighting limit)$	$Δ u^{'} v^{'} : d i s t an ce o f t h e co l o u r p o in t f r o m i t s ini t ia l v a l u e in t h e C I E 1976 u^{'} v^{'} d ia g r am, u^{'} an d v^{'} : c h r o ma t i c i t y coor d ina t es, u_{0}^{'} an d v_{0}^{'} : ini t ia l v a l u es; t h e 0.007 l imi t i s co mm o n in l i g h t in g s p ec i f i c a t i o n s (ba c k g r o u n d k n o w l e d g e); D O E (D a v i s an d H an se n, L M - 80 d a t a o f 223 p h os p h or - co n v er t e d w hi t e L E D s) : 96 p er ce n t r e a c h L_{70} b ey o n d 60000 h, mi d - p o w er (p o l y m er) p a c k a g es s hi f t t o w a r d b l u e, hi g h - p o w er (cer ami c) p a c k a g es t o w a r d y e l l o w, c hi p - o n - b o a r d a r r a y s in d i r ec t i o n s t ha t d e p e n d o n t h e mak er, c hi p - sc a l e p a c k a g es w i t h p o l y m er s i d e w a l l s t o w a r d g r ee n or b l u e an d w i t h r e f l ec t i v e s i d e w a l l s t o w a r d y e l l o w or b l u e, mi d - p o w er L E D s s h o u l d r u n b e l o w 200 mA, hi g h - p o w er an d C O B L E D s t o 1.5 A, an d C S P L E D s n ee d T_{J} a t or b e l o w 120^{\circ} C$ $(co m p u t e d : n o n e; t h e s hi f t d i r ec t i o n se p a r a t es p a c k a g e f r o m p h os p h or e f f ec t s o n l y q u a l i t a t i v e l y (m y r e a d in g))$

Accelerated Life of LED Chips

Name	Formula	Related Formulae	Variables/Constants
Exponential decay with an Arrhenius rate	$P (t) = P_{0} e^{- t / τ}, \frac{1}{τ} = C (I) e^{- E_{a} / (k_{B} T_{J})}$	$T_{J} = T_{A} + R_{t h} P_{D}$	$P (t) : o pt i c a l o u tp u t$ $P_{0} : ini t ia l o u tp u t a f t er t h e e a r l y r i se$ $τ : d ec a y t im e t o 0.368 o f P_{0}, C (I) : c u r r e n t - d e p e n d e n t p r e f a c t or$ $E_{a} : a c t i v a t i o n e n er g y$ $k_{B} : B o l t z mann co n s t an t (8.617 \times 1 0^{- 5} eV / K)$ $T_{J} : ab so l u t e j u n c t i o n t e m p er a t u r e$ $T_{A} : ambi e n t t e m p er a t u r e$ $R_{t h} : t h er ma l r es i s t an ce$ $P_{D} : d i ss i p a t e d p o w er; V a z q u ez e t a l . (5 mm A l I n G a P r e d L E D s, 15 p er t es t, 0 t o 40 mA a t 130, 150 an d 180^{\circ} C) : o u tp u t f i r s t r i ses f or a t im e t ha t d e p e n d s o n t h e s t r ess (d e f ec t ann e a l in g), t h e n d ec a y s e x p o n e n t ia l l y, u nbia se d L E D s a l so d ec a y t h r o ug h e p o x y b r o w nin g, E_{a} = 1.5 \pm 0.1 eV f or a l l c u r r e n t s (1.2 t o 1.5 eV e l se w h er e in t h e t e x t), T_{J} - T_{A} = 223.56 P_{D} (man u f a c t u r er 240 K/W), T_{J} = 160.6, 164.1 an d 169.0^{\circ} C a t 20, 30 an d 40 mA in 150^{\circ} C ambi e n t, f or ce d co n v ec t i o n 15 t o 20 p er ce n t l o w er t h er ma l r es i s t an ce, ln C l in e a r in c u r r e n t, an d an M T T F o f 3 \times 1 0^{6} h a t 65^{\circ} C an d 20 mA$ $(co m p u t e d : w i t h an a ss u m e d 2.0 V a t 20 mA, T_{J} = 73.9^{\circ} C a t 65^{\circ} C ambi e n t; t h e 10.6 K r i se a t 150^{\circ} C im pl i es 47 mW; f r o m T_{J} = 160.6^{\circ} C t h e a cce l er a t i o n f a c t or i s 2.2 \times 1 0^{4} f or 1.5 eV an d 3.0 \times 1 0^{3} f or 1.2 eV, t u r nin g t h e 847 h W e ib u l l sc a l e in t o 1.9 \times 1 0^{7} an d 2.6 \times 1 0^{6} h)$
Weibull life distribution of an LED population	$F (t) = 1 - exp [- (\frac{t}{η})^{β}], \overset{ˉ}{t} = η Γ (1 + \frac{1}{β})$	$t_{f} : P (t_{f}) = 0.7 P_{0}$	$F (t) : f r a c t i o n f ai l e d$ $η : sc a l e (63.2 p er ce n t f ai l e d)$ $β : s ha p e$ $\overset{ˉ}{t} : m e an l i f e$ $Γ : g amma f u n c t i o n$ $t_{f} : f ai l u r e t im e, d e f in e d b y t h e o u tp u t f a l l in g t o 70 p er ce n t; V a z q u ez e t a l . a t 150^{\circ} C : β = 4.1, 5.1, 5.6 an d 8.3 an d η = 1660, 847, 163 an d 85 h a t 10, 20, 30 an d 40 mA (t h e l a s t tw o r e a d a s g i v e n, t h o ug h t h ey b r e ak t h e t r e n d), c u r r e n t a cce l er a t i o n m o d e l l e d b y an in v er se p o w er l a w w i t h e x p o n e n t 2; a s ha p e w e l l ab o v e 1 m e an s w e a r - o u t (see t h e F ai l u r e R a t e p a g e)$ $(co m p u t e d : m e an l i f e 779 h a t 20 mA; t h e sc a l e f a l l s 2.0 t im es f r o m 10 t o 20 mA, c l ose t o t h e in v er se - s q u a r e l a w, b u t 5.2 an d 10 t im es f u r t h er a t 30 an d 40 mA)$
Current acceleration of UV-LED decay	$Δ P \propto I^{m}$	$\frac{Δ P ( I _{2} )}{Δ P ( I _{1} )} = (\frac{I _{2}}{I _{1}})^{m}$	$Δ P : l oss o f o pt i c a l p o w er a f t er a f i x e d t im e$ $I : s t r ess c u r r e n t$ $m : c u r r e n t e x p o n e n t; Gl aab e t a l . (233 nm f a r - U V C L E D s, 20 an d 100 mA a t 20 an d 75^{\circ} C) : t h e p o w er f a l l s t o ab o u t 50 t o 30 p er ce n t w i t hin ab o u t 100 h, a f t er 250 h ab o u t 40 p er ce n t r e main s a f t er 20 mA an d b e l o w 10 p er ce n t a f t er 100 mA s t r ess, t h e l oss sc a l es w i t h t h e c u r r e n t t o t h e t hi r d p o w er, t h e t e m p er a t u r e ha s n o s i g ni f i c an t e f f ec t, t h e e a r l y d r o p co m es w i t h a r i se o f t h e c u r r e n t b e l o w t h e d i f f u s i o n v o l t a g e (ab o u t 0.3 t o 1 μ A) an d a l oss o f h y d r o g e n o n t h e p - s i d e n e a r t h e a c t i v e r e g i o n, an d a s l o w er d ec l in e d o mina t es a f t er ab o u t 100 h; R u sc h e l e t a l . (265 nm, 200 mA = 60 A/c m^{2} a t 20^{\circ} C f or 2000 h, d i s l oc a t i o n d e n s i t i es 0.8 \times 1 0^{9} t o 5.8 \times 1 0^{9} c m^{- 2}) : l o w er d i s l oc a t i o n d e n s i t y g i v es m or e p o w er an d s l o w er d e g r a d a t i o n, e x t r a p o l a t e d L_{70} ab o v e 10000 h f or t h e b es t t e m pl a t e$ $(co m p u t e d : m = 3 mak es 100 mA s t r ess 125 t im es a s d ama g in g a s 20 mA a t e q u a l t im e)$

Defect and Hydrogen Kinetics

Name	Formula	Related Formulae	Variables/Constants
Diffusion-limited degradation	$\frac{Δ P}{P _{0}} \propto t, c (x, t) = c_{s} erfc (\frac{x}{2 D t})$	$Δ R_{o n} \propto Δ P$	$Δ P / P_{0} : r e l a t i v e l oss o f o pt i c a l p o w er$ $t : s t r ess t im e, c (x, t) : co n ce n t r a t i o n o f t h e d i f f u s in g s p ec i es (h y d r o g e n) a t d e pt h x$ $c_{s} : so u r ce co n ce n t r a t i o n$ $D : d i f f u s i o n coe f f i c i e n t$ $Δ R_{o n} : in cr e a se o f t h e ser i es r es i s t an ce; M e n e g hini e t a l . (G a N - o n - G a N p n d i o d es, 110 μ m d iam e t er, M g ab o v e 1 \times 1 0^{19} c m^{- 3}, 0.7 kA/c m^{2} f or 36000 s) : t h e t u r n - o n v o l t a g e r ose a f t er 300 s, t h e g e n er a t i o n - r eco mbina t i o n c u r r e n t a f t er 6300 s an d t h e r e v er se l e ak a g e a f t er 31 500 s, t h e e l ec t r o l u min esce n ce f e l l g r a d u a l l y u p t o 13500 s, t h e n t h e l o g - l o g L - I s l o p e r ose f r o m ab o u t 1 t o 1.7 (m or e S R H r eco mbina t i o n), o pt i c a l l oss an d r es i s t an ce r i se sc a l e d w i t h t h e s q u a r e r oo t o f t im e an d cor r e l a t e d l in e a r l y, w hi c h f i t s h y d r o g e n d i f f u s in g f r o m t h e p + l a y er an d p a ss i v a t in g M g a cce pt or s; C a s u e t a l . : in a tw o - w e l l I n G a N L E D t h e w e l l n e a r er t h e p - s i d e d e g r a d e d f i r s t, e x p o n e n t ia l l y t h e n s a t u r a t in g, an d t h e n - s i d e w e l l l a t er, w hi c h f i t s a d i f f u s i v e d e f ec t f r o n t m o v in g f r o m t h e p - s i d e$ $(co m p u t e d : a l oss t ha t g r o w s a s t d o u b l es w h e n t h e t im e g r o w s f o u r t im es)$
Non-radiative recombination in the ABC model	$η_{I QE} = \frac{B n ^{2}}{A n + B n ^{2} + C n ^{3}}$	$I_{l e ak} (V_{r}), n_{i d} rise with aging$	$η_{I QE} : in t er na l q u an t u m e f f i c i e n cy$ $A : S R H (d e f ec t) coe f f i c i e n t$ $B : r a d ia t i v e coe f f i c i e n t$ $C : A ug er coe f f i c i e n t$ $n : c a r r i er d e n s i t y$ $I_{l e ak} : r e v er se l e ak a g e a t r e v er se v o l t a g e V_{r}$ $n_{i d} : i d e a l i t y f a c t or (A B C m o d e l, ba c k g r o u n d k n o w l e d g e); W e n e t a l . (275 nm A l G a N U V C L E D s, 250 x 500 μ m, 100 mA (ab o u t 200 A/c m^{2}) a t r oo m t e m p er a t u r e f or 7056 h) : A r ose b y 13.5 p er ce n t, B b y 8.2, C b y 6.3 an d t h e e x t r a t er m D b y 6.4 p er ce n t, t h e f or w a r d v o l t a g e a t 100 mA r ose b y 0.2 V, t h e l e ak a g e a t - 5 V r ose f r o m 1.2 \times 1 0^{- 7} t o 1.5 \times 1 0^{- 5} A, t h e t u r n - o n v o l t a g e coe f f i c i e n t f r o m 1.6 t o 3.5 mV/K, l o w - f r e q u e n cy n o i se b y o n e t o tw o d ec a d es, t h e e mi ss i o n r e d - s hi f t e d b y ab o u t 0.8 nm, t h e q u an t u m - w e l l c a t h o d o l u min esce n ce p e ak a t 281 nm d i s a pp e a r e d an d t h e Q W - t o - d e f ec t in t e n s i t y r a t i o f e l l f r o m 0.258 t o 0.014, w hi c h t h ey a tt r ib u t e t o d e f ec t s cr e a t e d in t h e n - A l G a N l a y er b e l o w t h e w e l l s$ $(co m p u t e d : t h e l e ak a g e g r e w 125 t im es an d t h e Q W - t o - d e f ec t r a t i o f e l l 18 t im es; a l a r g er A l o w er s η_{I QE} m os t a t l o w c u r r e n t, w h er e t h e A n t er m d o mina t es, a s t h e s t r o n g er l o w - c u r r e n t E L l oss s h o w s)$

Important Questions

Ques: How does TM-21 turn an LM-80 test into a lifetime, and why is the result often given as ‘>36 000 h’?

TM-21 fits the last part of the LM-80 data with $Φ (t) = B e^{- α t}$ and computes $L_{p} = ln (100 B / p) / α$ . The fit uses the last 5000 $h$ for 6000 to 10 000 $h$ of data and the last half beyond, and ignores data below 1000 $h$ . The projection may not exceed six times the test duration for 20 or more units, or 5.5 times for 10 to 19. In the NIST example with 6000 $h$ of data, the 55 $^{\circ} C$ fit gives $L_{70} = 36412$ $h$ (computed), more than six times 6000 $h$ , so it must be reported as $L_{70} (6 k) > 36000$ $h$ , while the 85 $^{\circ} C$ value of about 24 000 $h$ is within the limit. The rule prevents a slow exponential fit from claiming decades of life from months of data. Reference: C. Miller (NIST), ‘IES TM-21-11 Overview, History and Q&A Session’, EPA ENERGY STAR Lamp Round Table, San Diego, 24 October 2011 (slides, fetch-tool summary).

Ques: Is the activation energy in the TM-21 worked example 0.14 or 0.45 eV?

The slides give $E_{a} / k_{B} \approx 5167.5$ $K$ , which is 0.45 $eV$ , but the two decay rates of the example, $1.042 \times 1 0^{- 5}$ per hour at 55 $^{\circ} C$ and $1.598 \times 1 0^{- 5}$ at 85 $^{\circ} C$ , give 1675 $K$ , or 0.144 $eV$ (computed). The interpolated result in the slides, about 29 277 $h$ at 70 $^{\circ} C$ , is reproduced only with 1675 $K$ , which gives 29 279 $h$ , while 5167.5 $K$ would give 18 388 $h$ (computed). The value from the decay rates is therefore the correct one for this example, and 5167.5 $K$ is probably a transcription error in the slides (my check). The low value is typical of lumen depreciation of packaged white LEDs, which mixes chip, phosphor and encapsulant effects, and is far below the 1.2 to 1.5 $eV$ found for red LED chips. Reference: C. Miller (NIST), ‘IES TM-21-11 Overview, History and Q&A Session’, EPA ENERGY STAR Lamp Round Table, San Diego, 24 October 2011 (slides, fetch-tool summary); M. Vazquez, N. Nunez, E. Nogueira and A. Borreguero, ‘Degradation of AlInGaP red LEDs under drive current and temperature accelerated life tests’, Universidad Politecnica de Madrid (year not stated, after 2009; some equations garbled, fetch-tool summary).

Ques: Does a red AlInGaP LED degrade with 1.2 or 1.5 eV?

Vazquez et al. give both values. One section states 1.5 plus or minus 0.1 $eV$ from a least-squares fit for all currents, another 1.2 to 1.5 $eV$ depending on the current. Their extrapolated MTTF of $3 \times 1 0^{6}$ $h$ at 65 $^{\circ} C$ and 20 $mA$ is the useful check: from the 847 $h$ Weibull scale at 160.6 $^{\circ} C$ junction, 1.5 $eV$ gives an acceleration of $2.2 \times 1 0^{4}$ and a life of $1.9 \times 1 0^{7}$ $h$ , while 1.2 $eV$ gives $3.0 \times 1 0^{3}$ and $2.6 \times 1 0^{6}$ $h$ , close to their value (computed, with an assumed 2.0 $V$ forward voltage). The lower value, or a current-dependent one, is therefore more likely behind the published MTTF (my reading). For a design, 1.2 $eV$ is also the safer choice, since it predicts the shorter life. Reference: M. Vazquez, N. Nunez, E. Nogueira and A. Borreguero, ‘Degradation of AlInGaP red LEDs under drive current and temperature accelerated life tests’, Universidad Politecnica de Madrid (year not stated, after 2009; some equations garbled, fetch-tool summary).

Ques: Why does an LED’s light output first rise and then fall during a life test?

Vazquez et al. observed in AlInGaP red LEDs that the output first rises for a time that depends on the stress conditions, which they attribute to the annealing of defects, and then decays exponentially, $P = P_{0} e^{- t / τ}$ . The rise matters for the analysis: a fit that includes it underestimates the decay rate. TM-21 deals with this by excluding the first 1000 $h$ and fitting only the last part of the data. The decay itself has two sources. LEDs stored without current also decayed, through browning of the epoxy, which temperature accelerates, while the chip degradation needs current and shows up in the I-V curves as more non-radiative current at low voltage and more series resistance at high voltage. Unbiased storage therefore separates the package from the chip. Reference: M. Vazquez, N. Nunez, E. Nogueira and A. Borreguero, ‘Degradation of AlInGaP red LEDs under drive current and temperature accelerated life tests’, Universidad Politecnica de Madrid (year not stated, after 2009; some equations garbled, fetch-tool summary); C. Miller (NIST), ‘IES TM-21-11 Overview, History and Q&A Session’, EPA ENERGY STAR Lamp Round Table, San Diego, 24 October 2011 (slides, fetch-tool summary).

Ques: How does hydrogen degrade a GaN LED or pn diode?

Meneghini et al. stressed GaN-on-GaN pn diodes at 0.7 $kA/c m^{2}$ and saw the turn-on voltage rise after 300 $s$ , then the generation-recombination current, the reverse leakage and the electroluminescence change. The optical loss and the rise of the series resistance both grew as the square root of time and were linearly correlated, which fits a diffusion process: hydrogen moves from the heavily Mg-doped p+ layer toward the junction and passivates Mg acceptors, so fewer holes are injected. In multi-quantum-well InGaN LEDs Casu et al. likewise found that the well nearer the p-side degrades first and the n-side well later, consistent with a front moving from the p-side. In far-UVC LEDs Glaab et al. link the early drop to a loss of hydrogen on the p-side near the active region. Reference: M. Meneghini, E. Fabris, M. Ruzzarin, C. De Santi, K. Nomoto, Z. Hu, W. Li, X. Gao, D. Jena, H. G. Xing, M. Sun, T. Palacios, G. Meneghesso and E. Zanoni, ‘Degradation mechanisms of GaN-based vertical devices: a review’, physica status solidi (a), doi:10.1002/pssa.201900750 (accepted manuscript, fetch-tool summary); C. Casu, M. Buffolo, A. Caria, C. De Santi, E. Zanoni, G. Meneghesso and M. Meneghini, ‘Impact of Generation and Relocation of Defects on Optical Degradation of Multi-Quantum-Well InGaN/GaN-Based Light-Emitting Diode’, Micromachines 13(8), 1266 (2022) (abstract only); J. Glaab, J. Ruschel, N. Lobo Ploch, H. K. Cho, F. Mehnke, L. Sulmoni, M. Guttmann, T. Wernicke, M. Weyers, S. Einfeldt and M. Kneissl, ‘Impact of operation parameters on the degradation of 233 nm AlGaN-based far-UVC LEDs’, Journal of Applied Physics 131, 014501 (2022) (abstract only).

Ques: Why do UVC LEDs degrade so much faster than blue LEDs?

Because their AlGaN layers have far more defects and their p-side is harder to dope. Glaab et al. found that 233 $nm$ LEDs lose 50 to 70 per cent of their output within about 100 $h$ , and that the loss scales with the third power of the current, so 100 $mA$ is 125 times as damaging as 20 $mA$ (computed), while temperature had no significant effect. Ruschel et al. showed at 265 $nm$ that a lower threading dislocation density, from $5.8 \times 1 0^{9}$ down to $0.8 \times 1 0^{9}$ $c m^{- 2}$ , gives more power and slower decay, with an extrapolated $L_{70}$ above 10 000 $h$ for the best template. Wen et al. found after 7056 $h$ at 275 $nm$ a 13.5 per cent larger SRH coefficient, 125 times more leakage and new defect emission in the n-AlGaN. Reference: J. Glaab, J. Ruschel, N. Lobo Ploch, H. K. Cho, F. Mehnke, L. Sulmoni, M. Guttmann, T. Wernicke, M. Weyers, S. Einfeldt and M. Kneissl, ‘Impact of operation parameters on the degradation of 233 nm AlGaN-based far-UVC LEDs’, Journal of Applied Physics 131, 014501 (2022) (abstract only); J. Ruschel, J. Glaab, N. Susilo, S. Hagedorn, S. Walde, E. Ziffer, H. K. Cho, N. Lobo Ploch, T. Wernicke, M. Weyers, S. Einfeldt and M. Kneissl, ‘Reliability of UVC LEDs fabricated on AlN/sapphire templates with different threading dislocation densities’, Applied Physics Letters 117, 241104 (2020) (abstract only); P. Wen, Y. He, M. Feng, J. Dong, X. Zou, J. Chen, H. Qu, X. Sun, Q. Sun and H. Yang, ‘Unraveling the Degradation Mechanisms of AlGaN-Based UVC-LEDs: A Comprehensive Analysis’, IEEE Transactions on Electron Devices 73(5), 2901-2906 (2026) (fetch-tool summary).

Ques: Why do white LEDs shift colour in different directions?

Because the shift comes from the package and the phosphor, not only from the blue chip. The DOE analysis of LM-80 data for 223 phosphor-converted white LEDs found predictable lumen maintenance, with 96 per cent reaching $L_{70}$ beyond 60 000 $h$ , but colour shifts whose direction depends on the package: mid-power polymer packages shift toward blue, high-power ceramic packages toward yellow, chip-on-board arrays in directions that depend on the maker, and chip-scale packages toward green, blue or yellow depending on their sidewalls. A blue shift suggests a loss of phosphor conversion or a darkening of the polymer that scatters light into it, a yellow shift more conversion or a blue loss, but the article read does not give the mechanisms (my reading). Lumen maintenance alone therefore does not guarantee colour stability. Reference: L. Davis (RTI International) and M. Hansen (LED Lighting Advisors), U.S. DOE analysis of LM-80 data for 223 phosphor-converted white LED packages, as reported in ‘DOE publishes research on LED lumen and chromaticity maintenance or shift’, LEDs Magazine (report title and year not given in the article read) (fetch-tool summary).

Quartz 5

Explorer

Light Emitting Diodes (InGaN, AlInGaP, UV and White)