This page covers the degradation of lithium niobate electro-optic modulators, both titanium-diffused and proton-exchanged bulk devices and thin-film lithium niobate (TFLN): the drift of the DC bias point, its temperature acceleration, the thermal bias shift from pyroelectric charge, and leakage and domain inversion under continuous bias. Electroabsorption modulators share the mechanisms of the laser and photodiode pages, since they are reverse-biased multiple-quantum-well diodes that absorb light, and are not repeated here. The device physics is on the Electro-Optic Phase Modulator and electroabsorption modulator pages of this vault.

1. Why It Happens

Trigger: A DC bias held for long times, especially at high temperature (Mitsugi et al.; Sosunov et al.); temperature changes of the chip (Burns et al.).
Mechanism, charge redistribution: Under DC bias, charge moves in the buffer layer and in the crystal, so the field in the waveguide changes with time and the bias must be adjusted to hold the operating point; in the RC picture the time constant and the final field split are set by the resistances and capacitances of the layers (Burns et al.; Yu et al.; background knowledge for the model).
Mechanism, near-surface defects: A disturbed near-surface layer from processing enhances drift, and treating it reduced the short-term drift and raised its activation energy (Sosunov et al.).
Mechanism, pyroelectric charge: A temperature change alters the spontaneous polarization and leaves charge on the Z faces that neutralizes slowly (Burns et al.).
Mechanism, domain inversion: Long DC bias with leakage current creates micro-domains near the hot electrode (Mitsugi et al.).
What sets the rate: The temperature, with about 0.8 to 1.6 $eV$ (Mitsugi et al., computed; Sosunov et al.); the resistivity of the buffer (Burns et al.); the leakage current (Mitsugi et al.).
Failure criterion: A bias voltage outside the range of the bias controller within the service life, typically required for 15 to 20 years (Mitsugi et al.; background knowledge for the controller limit).

2. Where to Look

Location: The buffer layer under the electrodes and the near-surface region of the crystal (Burns et al.; Sosunov et al.); the hot waveguide electrode for domains (Mitsugi et al.); the SiO2 cladding and the SiO2-LN interface of TFLN (Yu et al.).
Signature: An overshoot at bias switching and a slow baseline shift (Yu et al.); a bias that drifts with temperature cycles (Burns et al.); a leakage current that falls slowly with occasional steps (Mitsugi et al.).
Confirm with: Bias-drift tests at several temperatures with a fit of the time constants (Sosunov et al.); bias step tests (Yu et al.); thermal cycling of the bias point (Burns et al.); leakage monitoring and domain imaging after long bias (Mitsugi et al.).
Look-alikes: The fast photovoltaic response to UV-generated electrons, which relaxes before testing (Sosunov et al.); thermal bias shift, which follows temperature rather than time (Burns et al.).

3. Prevention

Design: Use a buffer with controlled, moderate conductivity, such as SiTiON instead of SiO2 (Burns et al.); prefer X-cut crystals where pyroelectric charge is a concern (Sosunov et al.).
Material and process: Treat the near-surface layer after waveguide fabrication (Sosunov et al.); for TFLN, anneal treatments that raise the effective resistance, kept below the temperature that damages the SiO2-LN interface (Yu et al.).
Operation: Use a bias controller with enough range for the drift over life, and limit the DC bias and the temperature (Mitsugi et al.; background knowledge).
Screening and test: Long bias tests at elevated temperature with Arrhenius extrapolation (Mitsugi et al.; Sosunov et al.).

DC Bias Drift

Name	Formula	Related Formulae	Variables/Constants
Two-layer RC model of DC bias drift	$V_{L N} (t) = V_{0} [\frac{C _{b}}{C _{b} + C _{L N}} e^{- t / τ} + \frac{R _{L N}}{R _{b} + R _{L N}} (1 - e^{- t / τ})]$	$τ = \frac{R _{b} R _{L N}}{R _{b} + R _{L N}} (C_{b} + C_{L N}), τ_{d i e l} = ε_{0} ε_{r} ρ$	$V_{L N} (t) : p a r t o f t h e a ppl i e d D C bia s V_{0} t ha t f a l l s a cr oss t h e l i t hi u m ni o ba t e a f t er t im e t$ $C_{b} an d C_{L N} : c a p a c i t an ces o f t h e b u f f er l a y er an d o f t h e cr y s t a l$ $R_{b} an d R_{L N} : t h e i r r es i s t an ces$ $τ : d r i f t t im e co n s t an t$ $τ_{d i e l} : d i e l ec t r i c r e l a x a t i o n t im e o f o n e l a y er$ $ε_{0} : v a c uu m p er mi tt i v i t y$ $ε_{r} : r e l a t i v e p er mi tt i v i t y$ $ρ : r es i s t i v i t y (e q u i v a l e n t - c i r c u i t m o d e l, ba c k g r o u n d k n o w l e d g e); Y u e t a l . (T F L N) : t h e R C n e tw or k i s t h e m o d e l u se d, an d t h e f l a tt e nin g o f d r i f t a f t er t r e a t m e n t i s r e a d a s a hi g h er e f f ec t i v e r es i s t an ce; B u r n s e t a l . : bia s d r i f t co m es f r o m c ha r g e v a r ia t i o n s in t h e b u f f er or t h e c ha r g e imba l an ce in t h e cr y s t a l, an d a S i O 2 b u f f er o f ab o u t 2 \times 1 0^{11} Ω cm w a s r e pl a ce d b y s i l i co n t i t ani u m o x y ni t r i d e o f ab o u t 1 \times 1 0^{5} t o 1 \times 1 0^{10} Ω cm (t es t d e v i ces 8.0 \times 1 0^{4} t o 2.5 \times 1 0^{5} Ω cm) w h ose r es i s t i v i t y s t a y e d s t ab l e f or m o n t h s an d a f t er 125^{\circ} C o v er ni g h t in ai r$ $(co m p u t e d : w i t h ε_{r} = 3.9 (a ss u m e d), τ_{d i e l} = 0.069 s f or 2 \times 1 0^{11} Ω cm, 3.5 \times 1 0^{- 3} s f or 1 \times 1 0^{10} Ω cm an d 2.8 \times 1 0^{- 8} t o 8.6 \times 1 0^{- 8} s f or t h e t es t f i l m s, so a co n d u c t i v e b u f f er se ttl es a t o n ce an d t h e f ina l s pl i t i s se t b y t h e r es i s t an ces, n o t b y t h e c a p a c i t an ces)$
Arrhenius acceleration of the drift rate	$\frac{1}{τ ( T )} = \frac{1}{τ _{0}} e^{- E_{a} / (k_{B} T)}, E_{a} = \frac{k _{B} l n ( r _{2} / r _{1} )}{1/ T _{1} - 1/ T _{2}}$	$A F = exp [\frac{E _{a}}{k _{B}} (\frac{1}{T _{u se}} - \frac{1}{T _{a cc}})]$	$τ (T) : d r i f t (r e l a x a t i o n) t im e a t ab so l u t e t e m p er a t u r e T$ $τ_{0} : p r e f a c t or$ $E_{a} : a c t i v a t i o n e n er g y$ $k_{B} : B o l t z mann co n s t an t (8.617 \times 1 0^{- 5} eV / K)$ $r_{1} an d r_{2} : r a t es a t T_{1} an d T_{2}$ $A F : a cce l er a t i o n f a c t or f r o m t h e t es t t e m p er a t u r e T_{a cc} t o u se a t T_{u se}; M i t s ug i e t a l . (T i : L i N b O 3, 10 V a cr oss a 25 μ m g a p, 2.05 an d 2.24 MΩ c hi p s a t 80 an d 100^{\circ} C f or 100 h) : t h e l e ak a g e c u r r e n t d ec a y e d 4.2 t im es f a s t er a t 100 t han a t 80^{\circ} C, w hi c h t h ey g i v e a s ab o u t 0.9 eV, c l ose t o t h e ab o u t 1 eV a ssoc ia t e d w i t h D C d r i f t, an d t h e r e q u i r e m e n t i s o v er 15 (in t r o d u c t i o n) or 20 y e a r s (ab s t r a c t); S os u n o v e t a l . (ann e a l e d p r o t o n - e x c han g e w a v e g u i d es) : t h e a c t i v a t i o n e n er g y o f t h e o p er a t in g p o in t r ose f r o m 0.9 t o 1.6 eV a f t er t r e a t m e n t o f t h e n e a r - s u r f a ce l a y er, t h e s h or t - t er m r e l a x a t i o n t im e f e l l f r o m 286 t o 116 s, an d t h e f i t coe f f i c i e n t B_{0} a t 25^{\circ} C f e l l f r o m 0.0046 t o 0.00004 h^{- 1}$ $(co m p u t e d : a r a t i o o f 4.2 b e tw ee n 80 an d 100^{\circ} C g i v es 0.81 eV; f r o m 85 t o 40^{\circ} C, A F = 66 a t 0.9 eV, 105 a t 1.0 eV an d 1720 a t 1.6 eV; B_{0} f e l l 115 t im es an d t h e r e l a x a t i o n t im e 59 p er ce n t)$
Short- and long-term drift components	$Δ V_{bia s} (t) = A_{1} (1 - e^{- t / τ_{1}}) + A_{2} (1 - e^{- t / τ_{2}})$	$Δ V_{bia s} in units of V_{π} or as optical drift in dB$	$Δ V_{bia s} (t) : s hi f t o f t h e bia s n ee d e d t o h o l d t h e o p er a t in g p o in t$ $A_{1} an d A_{2} : am pl i t u d es$ $τ_{1} an d τ_{2} : s h or t an d l o n g t im e co n s t an t s (bim o d a l f or m, ba c k g r o u n d k n o w l e d g e f or t h e s u m), V_{π} : ha l f - w a v e v o l t a g e; Y u e t a l . (600 nm X - c u t T F L N o n 4.7 μ m S i O 2 w i t h an 800 nm S i O 2 c l a dd in g) : u n t r e a t e d d e v i ces o v er s h oo t a t bia s s w i t c hin g, a 0.005 T t r e a t m e n t g i v es τ_{1} = 1 s an d τ_{2} = 10 s, s t r o n g er f i e l d s s u pp r ess t h e t r an s i e n t b u t mak e t h e s l o w d r i f t m or e p r o n o u n ce d, ab o u t 1 dB d r i f t a t 8 V o v er an h o u r a f t er ann e a l in g a t 723 K, an d b e l o w 0.3 dB o v er s h oo t an d b e l o w 0.1 dB l o n g - t er m d r i f t a f t er 823 K, t h e hi g h es t t e m p er a t u r e b e f or e S i O 2 - L N in t er f a ce d ama g e; S os u n o v e t a l . : s h or t - t er m d r i f t r e l a x es in min u t es an d l o n g - t er m d r i f t o v er h o u r s an d d a y s, V_{π} o f 3.5 t o 6 V, 0.13 t o 0.17 dB / c m l oss$ $(co m p u t e d : n o n e)$

Thermal Shift and Bias Damage

Name	Formula	Related Formulae	Variables/Constants
Thermal bias shift	$Δ ϕ = α_{T} Δ T$	$Δ P_{s} = p Δ T (pyroelectric charge on the Z faces)$	$Δ ϕ : s hi f t o f t h e o pt i c a l p ha se a t t h e o p er a t in g p o in t$ $α_{T} : t h er ma l se n s i t i v i t y$ $Δ T : t e m p er a t u r e c han g e$ $Δ P_{s} : c han g e o f t h e s p o n t an eo u s p o l a r i z a t i o n$ $p : p y r oe l ec t r i c coe f f i c i e n t; B u r n s e t a l . : a t e m p er a t u r e c han g e a l t er s t h e s p o n t an eo u s p o l a r i z a t i o n an d l e a v es a c ha r g e imba l an ce b e tw ee n t h e Z f a ces t ha t t h e hi g h r es i s t i v i t y o f L i N b O 3 n e u t r a l i z es o n l y s l o w l y, an d S i T i O N - b u f f er e d d e v i ces s h o w e d ab o u t 2 t o 2.5 deg/K f r o m 25 t o 45^{\circ} C, ab o u t t e n t im es l ess t han a S i O 2 - o n l y co n t r o l; S os u n o v e t a l . : in X - c u t cr y s t a l s t h e p y r oe l ec t r i c L i + r e d i s t r ib u t i o n o f Z - c u t i s n o t e x p ec t e d$ $(co m p u t e d : a 20 K s w in g m o v es t h e o p er a t in g p o in t b y 40 t o 50 d e g r ees o f p ha se w i t h t h e S i T i O N b u f f er an d b y ab o u t 400 t o 500 d e g r ees w i t h S i O 2 a l o n e (t e n t im es, a ss u m e d e x a c tl y))$
Leakage current and domain inversion under continuous bias	$E \approx \frac{V}{g}, I_{l e ak} (t) ↓ with occasional steps$	$N_{d o main} \propto I_{l e ak}$	$E : f i e l d b e tw ee n t h e e l ec t r o d es$ $V : bia s$ $g : e l ec t r o d e g a p$ $I_{l e ak} : l e ak a g e c u r r e n t b e tw ee n h o t e l ec t r o d e an d g r o u n d$ $N_{d o main} : n u mb er o f mi cr o - d o main s (p r o p or t i o na l i t y i s t h e a u t h or s^{'} q u a l i t a t i v e o b ser v a t i o n); M i t s ug i e t a l . : 205 kΩ an d 700 kΩ c hi p s a t 10 an d 15 V an d 80^{\circ} C f or 100 h f or m e d mi cr o - d o main s, m or e in t h e l o w er - r es i s t an ce c hi p w i t h ab o u t tw i ce t h e l e ak a g e, t h e l e ak a g e d ecr e a se d g r a d u a l l y an d f a s t er a t 100^{\circ} C, o v er 1000 h a t 80^{\circ} C mi cr o - d o main s g r e w o n t h e h o t w a v e g u i d e f or b o t h p o l a r i t i es, m or e f or - 10 V, an d t h e l e ak a g e r ose in s u dd e n s t e p s an d t h e n s l o w l y d ecr e a se d$ $(co m p u t e d : 10 V o v er 25 μ m i s 4 \times 1 0^{5} V/m)$

Important Questions

Ques: Is the activation energy of LiNbO3 DC drift 0.8, 0.9, 1.0 or 1.6 eV?

The values come from different devices and treatments. Mitsugi et al. found that the leakage current of Ti:LiNbO3 chips decayed 4.2 times faster at 100 than at 80 $^{\circ} C$ and quote about 0.9 $eV$ , close to the about 1 $eV$ associated with DC drift; their own ratio gives 0.81 $eV$ (computed). Sosunov et al. measured 0.9 $eV$ for untreated proton-exchanged modulators and 1.6 $eV$ after treating the near-surface layer. For an untreated device about 0.9 $eV$ is therefore the more likely value, and 1.6 $eV$ applies only to the treated process. The choice changes the extrapolation strongly: from an 85 $^{\circ} C$ test to 40 $^{\circ} C$ , the acceleration is 66 at 0.9 $eV$ but 1720 at 1.6 $eV$ (computed). Reference: N. Mitsugi, H. Nagata and J. Minowa, ‘The Influence of Continuous DC Bias Application on Properties of Ti:LiNbO3 Waveguide Devices’, Transactions of the Materials Research Society of Japan 20, 373-376 (1996) (OCR partly garbled, fetch-tool summary); A. Sosunov, R. Ponomarev, A. Zhuravlev, S. Mushinsky and M. Kuneva, ‘Reduction in DC-Drift in LiNbO3-Based Electro-Optical Modulator’, Photonics 8, 571 (2021) (fetch-tool summary).

Ques: Why does a more conductive buffer layer reduce the bias drift?

Because the drift is the slow redistribution of the bias between the buffer and the crystal. In a two-layer RC model the field in the lithium niobate starts at the capacitive split and ends at the resistive split, with a time constant set by the resistances and capacitances (background knowledge). Burns et al. replaced a SiO2 buffer of about $2 \times 1 0^{11}$ $Ω cm$ with silicon titanium oxynitride of about $1 \times 1 0^{5}$ to $1 \times 1 0^{10}$ $Ω cm$ . The dielectric relaxation time of a layer is $ε_{0} ε_{r} ρ$ , 0.069 $s$ for the SiO2 and below $1 \times 1 0^{- 7}$ $s$ for the test films, assuming a permittivity of 3.9 (computed), so the conductive buffer settles at once and no slow transfer remains. Its resistivity also stayed stable for months, which matters because a changing buffer would itself cause drift. Reference: W. K. Burns, L. A. Hess and V. Agarwal, ‘Buffer layer structures for stabilization of a lithium niobate device’, US Patent 6,925,211 B2 (2005) (fetch-tool summary); Y. Yu, Y. Ren, N. Chen and T. Chu, ‘Suppressing DC Drift in Thin-Film Lithium Niobate Modulators via Multiferroic Skyrmion Excitation’, Zhejiang University, arXiv:2503.19044 (journal not stated; annealing temperatures inconsistent in the text, fetch-tool summary).

Ques: Why does a lithium niobate modulator drift with temperature?

Because lithium niobate is pyroelectric. Burns et al. explain that a temperature change alters the spontaneous polarization and leaves a charge imbalance between the Z faces of the crystal, which the high resistivity neutralizes only slowly, so the electrodes lose control of the waveguide field for a while. Their SiTiON-buffered devices showed about 2 to 2.5 degrees of phase per kelvin from 25 to 45 $^{\circ} C$ , about an order of magnitude less than a SiO2-only control, so a 20 $K$ swing moves the operating point by 40 to 50 degrees instead of several hundred (computed). Sosunov et al. note that X-cut crystals, whose surface normal is nonpolar, do not show the pyroelectric Li+ redistribution expected in Z-cut, which is one reason X-cut devices are preferred. Reference: W. K. Burns, L. A. Hess and V. Agarwal, ‘Buffer layer structures for stabilization of a lithium niobate device’, US Patent 6,925,211 B2 (2005) (fetch-tool summary); A. Sosunov, R. Ponomarev, A. Zhuravlev, S. Mushinsky and M. Kuneva, ‘Reduction in DC-Drift in LiNbO3-Based Electro-Optical Modulator’, Photonics 8, 571 (2021) (fetch-tool summary).

Ques: Does thin-film lithium niobate solve the drift problem?

Not by itself. Yu et al. tested X-cut TFLN modulators, 600 $nm$ of lithium niobate on 4.7 $μ m$ of thermal SiO2 with an 800 $nm$ SiO2 cladding, under bias steps of $\pm 4$ $V$ and a long hold at 8 $V$ . Untreated devices overshot strongly at each bias switch, and devices given magnetic-field and anneal treatments showed time constants of 1 and 10 $s$ , then after an 823 $K$ anneal an overshoot below 0.3 $dB$ and a long-term drift below 0.1 $dB$ in an hour, near the noise floor, against about 1 $dB$ after 723 $K$ . The authors interpret the improvement as a higher effective resistance in the RC model, and they limit the anneal by damage to the SiO2-LN interface. The text gives several anneal temperatures inconsistently. Reference: Y. Yu, Y. Ren, N. Chen and T. Chu, ‘Suppressing DC Drift in Thin-Film Lithium Niobate Modulators via Multiferroic Skyrmion Excitation’, Zhejiang University, arXiv:2503.19044 (journal not stated; annealing temperatures inconsistent in the text, fetch-tool summary).

Ques: What does a continuous DC bias do to a Ti:LiNbO3 waveguide?

Mitsugi et al. applied 10 to 15 $V$ across a 25 $μ m$ gap, a field of about $4 \times 1 0^{5}$ $V/m$ (computed), at 80 and 100 $^{\circ} C$ . Micro-domains of inverted polarization appeared, more in the chip with the lower resistance and about twice the leakage current, and over 1000 $h$ at 80 $^{\circ} C$ they grew on the hot waveguide for both polarities, more for -10 $V$ . The leakage current fell gradually but rose in sudden steps from time to time, which suggests that conductive paths form and heal. The paper leaves the mechanism open, citing absorption currents in dielectrics. Domain inversion changes the local electro-optic response, so continuous bias at high temperature is both a drift and a damage stress. Reference: N. Mitsugi, H. Nagata and J. Minowa, ‘The Influence of Continuous DC Bias Application on Properties of Ti:LiNbO3 Waveguide Devices’, Transactions of the Materials Research Society of Japan 20, 373-376 (1996) (OCR partly garbled, fetch-tool summary); Electro-Optic Phase Modulator page of this vault.

Quartz 5

Explorer

Lithium Niobate Electro-Optic Modulators