Vertical-External-Cavity Surface-Emitting Laser (VECSEL)

1. Theory

Definition: A vertical-external-cavity surface-emitting laser (VECSEL), or optically pumped semiconductor disk laser, is a VCSEL-like gain chip, a Bragg mirror with a few micrometres of quantum wells on it, with the second mirror placed outside the chip, so that the cavity includes a free-space region of a few millimetres to tens of centimetres.
Gain chip: Quantum wells at the antinodes of the standing wave (resonant periodic gain) separated by barriers that absorb the pump; one chip had 14 compressively strained InGaAs wells of 8 nm with GaAsP strain compensation on a 25-pair AlGaAs/AlAs mirror of more than 99.9% reflectivity, with a total semiconductor thickness of about 6.5 $μ$ m.
Pumping: A multimode diode laser or diode bar pumps the chip at a wavelength shorter than the laser wavelength (808 nm for 980 nm), and its beam quality hardly matters; electrically pumped versions (NECSELs) are limited to about 1 W.
Cavity: A concave output coupler, a few mm to tens of cm from the chip, sets the mode size on the chip; the beam is a single transverse mode ( $M^{2}$ of 1.05 to 1.15 at 1 to 3 W).
Thermal management: The chip is grown upside down, soldered epitaxial side down on a diamond heat spreader and the substrate is etched away, or a transparent heat spreader (diamond, SiC or sapphire) is attached to the emitting surface.
Wavelengths: About 960 to 1030 nm (InGaAs on GaAs), 1.3 to 1.55 $μ$ m (GaInNAs), about 1.5 $μ$ m (InGaAsP on InP), about 2 $μ$ m (GaInSb on GaSb), about 850 nm (GaAs wells) and 610 to 650 nm (GaInP wells).
Pulsed operation: Mode-locking, with a saturable absorber mirror or by self-mode-locking, gives picosecond pulses at repetition rates of a few GHz, and pulses of 290 fs have been reported.

2. Working Principle

Low gain: The wells are thin, so the gain per pass is only about a percent or a few percent and the Bragg mirror must reflect more than 99.9%, the output coupler transmitting typically 0.3 to 2%.
Resonant periodic gain: A thin well at an antinode sees twice the average intensity of the standing wave, so the single-pass gain is $2 N_{w} d_{w} g$ , and the small-signal gain scales with the confinement factor while the saturation fluence scales with its inverse.
Pump absorption: Pump photons are absorbed in the barriers within a micrometre or so ( $α_{p}$ of $1 0^{4} c m^{- 1}$ ) and the carriers diffuse into the wells.
Heat: The quantum defect and the nonradiative recombination heat the chip; the gain peak moves at about 0.3 nm/K and the resonance at about 0.1 nm/K, so the gain peak is made to lie on the short-wavelength side to delay the thermal rollover.
Power scaling: The pump power and the spot area are increased together at constant intensity, so that the output grows with the area while the temperature stays about constant; the best spot size depends on the heat spreaders.
Output: More than 3 W of continuous single-mode power with 44% slope efficiency at 980 nm, tens of watts with diode-bar pumping, and 106 W in a multimode beam.
Frequency doubling: The power circulating inside the cavity is much higher than the output, so a nonlinear crystal inside the cavity gives watt-level green, blue and red.

3. Applications

Pumps for fibre lasers and amplifiers (research and development in the sources read): A 3 W, 980 nm VECSEL pumped a 3 cm Er/Yb fibre to over 250 mW at 1535 nm, and an early 1004 nm device was proposed for erbium-doped fibre amplifiers.
Compact visible sources: Frequency-doubled VECSELs for projection displays.
Spectroscopy and laser cooling: Near-infrared single-mode devices.
Dual-comb spectroscopy (research): Mode-locked disk lasers as comb sources.
Industrial machining (development): High-power optically pumped devices.

Name	Formula	Related Formulae	Variables/Constants
Single-pass gain and threshold of the gain chip	$2 N_{w} d_{w} g_{t h} = A_{i} + T_{m}$	$T_{m} = T_{oc} + T_{D B R}$ $Γ g L = Γ_{x y} L_{a} g [1 + cos (2 β z_{s}) \frac{s i n ( β L _{a} )}{β L _{a}}]$	$N_{w}, d_{w} : n u mb er o f q u an t u m w e l l s, a l l pl a ce d a t an t in o d es o f t h e s t an d in g w a v e, an d t hi c k n ess o f o n e w e l l$ $2 : g ain e nhan ce m e n t o f a t hin w e l l a t an an t in o d e, s in ce Γ L \to 2 L_{a} f or β L_{a} ≪ 1 an d z_{s} = 0, t h e w e l l s see in g tw i ce t h e a v er a g e in t e n s i t y o f t h e s t an d in g w a v e (t h e r eso nan t p er i o d i c g ain)$ $g_{t h} : ma t er ia l g ain o f t h e w e l l s a t t h r es h o l d, t h e t r an s v er se co n f in e m e n t b e in g t ak e n a s u ni t y b ec a u se t h e p u m p e d s p o t i s ma t c h e d t o t h e m o d e$ $A_{i} : p a ss i v e l oss p er p a ss o f t h e c hi p an d o f t h e c a v i t y$ $T_{oc}, T_{D B R} : t r an s mi ss i o n o f t h e o u tp u t co u pl er (0.3 t o 2 % in t y p i c a l d e v i ces) an d o f t h e B r a g g mi r r or o f t h e c hi p (b e l o w 0.1 %)$ $Γ_{x y}, L_{a}, z_{s}, β : t r an s v er se co n f in e m e n t f a c t or, t hi c k n ess o f t h e a c t i v e l a y er, p os i t i o n o f i t s ce n t r e an d p r o p a g a t i o n co n s t an t, in t h e g e n er a l e x p r ess i o n f or t h e a x ia l co n f in e m e n t f a c t or$
Pump absorption and carrier generation	$G_{p} (z) = \frac{q _{a} α _{p} λ _{p}}{h c} e^{- α_{p} z}$	$η_{ab s} = (1 - R_{p}) (1 - e^{- α_{p} d})$ $D \frac{d ^{2} n _{b}}{d z ^{2}} - n_{b} (A + B n_{b} + C n_{b}^{2}) + G_{p} (z) = 0$	$G_{p} : r a t e o f g e n er a t i o n o f e l ec t r o n - h o l e p ai r s p er u ni t v o l u m e a t t h e d e pt h z$ $q_{a} : p u m p p o w er d e n s i t y a t t h e s u r f a ce$ $λ_{p} : p u m p w a v e l e n g t h$ $α_{p} : ab sor pt i o n coe f f i c i e n t o f t h e ba r r i er ma t er ia l (1 0^{4} c m^{- 1} in o n e d es i g n, so t ha t 1/ α_{p} i s 1 μ m)$ $η_{ab s} : f r a c t i o n o f t h e p u m p p o w er ab sor b e d in a t hi c k n ess d, R_{p} b e in g t h e s u r f a ce r e f l ec t i v i t y a t t h e p u m p w a v e l e n g t h (27 % f or an u n co a t e d s u r f a ce in o n e d e v i ce)$ $n_{b} : c a r r i er d e n s i t y in t h e ba r r i er s$ $D : ambi p o l a r d i f f u s i o n coe f f i c i e n t (10 c m^{2} / s) an d A$ $B, C : m o n o m o l ec u l a r, bim o l ec u l a r an d A ug er r eco mbina t i o n coe f f i c i e n t s, t h e w e l l s b e in g f i l l e d b y d i f f u s i o n f r o m t h e ba r r i er s$
Quantum defect and heat generated	$η_{Q D} = 1 - \frac{λ _{p}}{λ}$	$P_{h e a t} = P_{ab s} - P_{o u t}$	$η_{Q D} : f r a c t i o n o f t h e ab sor b e d p u m p e n er g y l os t p er p h o t o n b ec a u se t h e p u m p p h o t o n i s m or e e n er g e t i c t han t h e l a ser p h o t o n$ $λ_{p}, λ : p u m p an d l a ser w a v e l e n g t h s$ $P_{h e a t} : p o w er d i ss i p a t e d in t h e c hi p, w hi c h a l so in c l u d es t h e n o n r a d ia t i v e r eco mbina t i o n o f t h e c a r r i er s$ $P_{ab s} : ab sor b e d p u m p p o w er an d P_{o u t} : l a ser o u tp u t (so P_{h e a t} e x cee d s t h e q u an t u m - d e f ec t p a r t)$
Thermal resistance and temperature rise	$R_{t h} = \frac{Δ T _{ma x}}{P _{h e a t}}$	$Δ T \approx \frac{d}{k} \frac{P _{h e a t}}{π w ^{2}}$	$R_{t h} : t h er ma l r es i s t an ce o f t h e c hi p w i t h i t s h e a t s p r e a d er an d h e a t s ink$ $Δ T_{ma x} : ma x im u m t e m p er a t u r e r i se in t h e p u m p e d s p o t$ $Δ T : t e m p er a t u r e r i se a cr oss a l a y er o f t hi c k n ess d an d t h er ma l co n d u c t i v i t y k w h e n t h e h e a t f l o w s in o n e d im e n s i o n, w hi c h h o l d s f or a p u m p s p o t o f r a d i u s w m u c h l a r g er t han d$ $k : t h er ma l co n d u c t i v i t y, ab o u t 2200 W / (m K) f or d iam o n d, 400.8 f or co pp er, 44.1 f or G a A s an d 4.5 f or G a I n N A s in o n e d es i g n s t u d y (ab o v e 1500 f or t h e C V D d iam o n d o f an o t h er d e v i ce)$
Thermal detuning of the gain from the cavity resonance	$δ λ (Δ T) = (\frac{d λ _{g ain}}{d T} - \frac{d λ _{c a v}}{d T}) Δ T$	$\frac{d λ _{g ain}}{d T} \approx 0.3 nm/K$ $\frac{d λ _{c a v}}{d T} \approx 0.1 nm/K$	$δ λ : s hi f t o f t h e g ain p e ak r e l a t i v e t o t h e mi cr oc a v i t y r eso nan ce o f t h e c hi p w h e n t h e a c t i v e r e g i o n i s h o tt er b y Δ T$ $d λ_{g ain} / d T, d λ_{c a v} / d T : t e m p er a t u r e coe f f i c i e n t s o f t h e g ain p e ak (t h e ban d g a p) an d o f t h e r eso nan ce (t h e r e f r a c t i v e in d e x), t h e v a l u es b e in g t h ose o f o n e t h es i s f or q u an t u m - w e l l c hi p s$ $(t h e g ain p e ak i s t h er e f or e d es i g n e d t o l i e t o t h e s h or t - w a v e l e n g t h s i d e o f t h e r eso nan ce a t l o w p u m p p o w er, t o d e l a y t h e t h er ma l r o l l o v er o f t h e o u tp u t)$
Output power and slope efficiency	$P_{o u t} = η_{s} (P_{p} - P_{t h})$	$η_{s} ≲ η_{ab s} \frac{λ _{p}}{λ} \frac{T _{m}}{A _{i} + T _{m}}$	$P_{o u t}, P_{p}, P_{t h} : o u tp u t p o w er, in c i d e n t p u m p p o w er an d p u m p p o w er a t t h r es h o l d$ $η_{s} : s l o p e e f f i c i e n cy (44 % ab o v e 3 W co n t in u o u s w a v e a t 980 nm in o n e d e v i ce)$ $η_{ab s} : f r a c t i o n o f t h e p u m p ab sor b e d$ $λ_{p} / λ : r a t i o o f t h e p h o t o n e n er g i es o f t h e l a ser an d o f t h e p u m p, w hi c h i s 1 - η_{Q D}$ $T_{m} / (A_{i} + T_{m}) : s ha r e o f t h e l oss t ha t i s u se f u l o u tp u t co u pl in g, t h e in e q u a l i t y co min g f r o m t h e c a r r i er s l os t b y r eco mbina t i o n o t h er t han in t o t h e m o d e$
Mode spacing and repetition rate of the external cavity	$Δ ν = f_{r e p} = \frac{c}{2 L _{c a v}}$	$f_{r e p} \approx 15 GHz \times \frac{1 cm}{L _{c a v}}$	$Δ ν : s p a c in g o f t h e l o n g i t u d ina l m o d es o f a l in e a r c a v i t y w h ose l e n g t h i s m os tl y f r ee s p a ce (t h e r e f r a c t i v e in d e x b e in g ab o u t 1)$ $f_{r e p} : r e p e t i t i o n r a t e o f a m o d e - l oc k e d l a ser, w hi c h e q u a l s t h e m o d e s p a c in g f or o n e p u l se in t h e c a v i t y (t y p i c a l l y a f e w G H z, an d 50 G H z in o n e c a se w i t h 100 mW)$ $L_{c a v} : l e n g t h o f t h e c a v i t y, b e tw ee n a f e w mi l l im e t r es an d so m e t e n s o f ce n t im e t r es, se t b y t h e s t abi l i t y o f t h e r eso na t or w i t h t h e c u r v e d mi r r or$
Power scaling with the pump spot	$P_{o u t} \approx π w^{2} η_{s} (I_{p} - I_{t h})$	$Δ T \approx \frac{d}{k} η_{h} I_{p}$	$w : r a d i u s o f t h e p u m p e d s p o t$ $I_{p}, I_{t h} : p u m p in t e n s i t y an d i t s v a l u e a t t h r es h o l d (b o t h p er u ni t a r e a)$ $Δ T : t e m p er a t u r e r i se f or o n e - d im e n s i o na l h e a t f l o w, w hi c h d oes n o t d e p e n d o n w a t co n s t an t in t e n s i t y b ec a u se t h e h e a t f l ux i s η_{h} I_{p}, w i t h η_{h} t h e f r a c t i o n o f t h e p u m p t ha t i s co n v er t e d t o h e a t$ $(t h e o u tp u t sc a l es w i t h t h e a r e a o f t h e s p o t a s l o n g a s t h e h e a t f l o w i s o n e - d im e n s i o na l, an d f or a s ma l l s p o t t h e l a t er a l s p r e a d in g o f t h e h e a t mak es t h e t e m p er a t u r e r i se l o w er, so t h er e i s an o pt im u m s p o t s i z e t ha t d e p e n d s o n t h e h e a t s p r e a d er s)$

Important Questions

Ques: Why is the gain per pass of a VECSEL so small, and why must the mirrors be so good?

The gain region is only a few micrometres thick and holds a few thin wells, so the single-pass gain is $2 N_{w} d_{w} g$ . For the 14 wells of 8 nm in the 980 nm device the thickness is $2.24 \times 1 0^{- 5}$ cm, and with an output coupler of 4% (96% reflectivity) at least $g \approx 1800$ cm $^{- 1}$ is needed even without passive loss (computed), which is a high gain and needs strong pumping. The threshold condition is that the gain equals the loss per pass, so the passive loss and the mirror transmission must be a fraction of a percent to a few percent, and the Bragg mirror must reflect more than 99.9%. This is the same condition as in a VCSEL, which the book puts typically above 99%, and the external cavity has the advantage that the output coupler can be chosen freely, between 0.3% and 2% in typical devices. Reference: Coldren, Corzine and Mašanović, Diode Lasers and Photonic Integrated Circuits, 2nd ed. (Wiley, 2012), Section 1.8, Appendix 5 and Problem 2.12; Z. L. Fan et al., Appl. Phys. Lett. 86, 211116 (2005); Vertical External-cavity Surface-emitting Lasers, RP Photonics Encyclopedia.

Ques: Why are the quantum wells placed at the antinodes of the standing wave, and what does this cost?

A well at an antinode sees the maximum field of the standing wave, so its modal gain is twice that of a well in a traveling wave of the same power. In the book’s expression the axial confinement factor of a thin active layer at the antinode is $2 L_{a} / L$ , from the factor $1 + sin (β L_{a}) / (β L_{a}) \to 2$ . The wells are separated by barriers that absorb the pump. The cost is that more wells make the structure thicker and more sensitive to growth errors, strain and temperature, and the inhomogeneous saturation of the gain can become a problem for narrow-linewidth or mode-locked lasers. The small-signal gain scales with the confinement factor and the saturation fluence with its inverse. Reference: Coldren et al., 2nd ed., Appendix 5 and Problem 2.12; Vertical External-cavity Surface-emitting Lasers, RP Photonics Encyclopedia; thesis, University of Southampton (eprints.soton.ac.uk/381283), Chapter 2.

Ques: Why is the pump wavelength much shorter than the laser wavelength, and how much heat does this cause?

The pump has to be absorbed by barrier layers, which are then spacers that pass the carriers to the wells, so its photon energy must exceed the barrier gap and be well above the laser photon energy. The difference is released as heat. With a pump of 808 nm and a laser of 980 nm the photon-energy ratio is 0.825, so the quantum defect is $1 - 808/980 = 17.6%$ of the absorbed power (computed). The nonradiative recombination adds to it. In-well pumping would reduce the defect but needs multipass absorption, because a few thin wells absorb little. The barrier absorption is strong, with $α_{p} = 1 0^{4}$ cm $^{- 1}$ in a design study, so that 1 $μ$ m absorbs 63% and 3 $μ$ m 95% of the pump (computed). Reference: Vertical External-cavity Surface-emitting Lasers, RP Photonics Encyclopedia; K. Sokół and R. P. Sarzała, Int. J. Electron. Telecommun. 60(3), 239 (2014); Fan et al., Appl. Phys. Lett. 86, 211116 (2005).

Ques: Why must the substrate be removed, or a heat spreader be added, in a VECSEL?

The heat is produced within a few micrometres of the surface and has to pass through everything between it and the heat sink. For one-dimensional flow, $Δ T \approx (d / k) P_{h e a t} / (π w^{2})$ . With an assumed 4 W of heat in a spot of 220 $μ$ m diameter, a 6.5 $μ$ m layer of GaAs ( $k$ = 44.1 W/(m K)) gives 15.5 K, but a 450 $μ$ m GaAs substrate in the heat path would give about 1070 K (computed), which is why a 0.5 mm substrate limits the power. The two remedies are to etch the substrate off the chip soldered to a diamond, leaving about 6.5 $μ$ m, or to bond a transparent heat spreader of diamond, SiC or sapphire to the emitting surface. The conductivity of GaInNAs, 4.5 W/(m K), shows why the active layer itself must stay thin. Reference: Fan et al., Appl. Phys. Lett. 86, 211116 (2005); Sokół and Sarzała, Int. J. Electron. Telecommun. 60(3), 239 (2014); Vertical External-cavity Surface-emitting Lasers, RP Photonics Encyclopedia.

Ques: Why is the gain peak of a VECSEL chip made shorter in wavelength than the cavity resonance?

Heating moves the gain peak (the bandgap) faster than the resonance (the refractive index), so the two can be aligned at the operating point only if the gain starts on the blue side. A chip of 980 nm emission had two microcavity photoluminescence peaks, at 946 nm and 980 nm, the stronger at 946 nm at low pump, and the wavelength moved only 0.22 nm per kW/cm $^{2}$ of pump intensity. The sources disagree on the rate of the gain peak: one thesis gives about 0.3 nm/K, the book (in its VCSEL discussion) about 0.5 nm/K. At 980 nm, an assumed 0.45 meV/K for the bandgap of a GaAs-type material gives 0.35 nm/K (computed), which supports 0.3 nm/K, whereas 0.5 nm/K corresponds to a wavelength of about 1.2 $μ$ m, so the thesis value is the more likely for this chip. Reference: Fan et al., Appl. Phys. Lett. 86, 211116 (2005); thesis, University of Southampton (eprints.soton.ac.uk/381283); Coldren et al., 2nd ed., Section 1.8.

Ques: Why can a VECSEL give a watt-level diffraction-limited beam when a VCSEL cannot?

In a VCSEL the Bragg mirrors are on the chip and the transverse mode is fixed by a small aperture, so the single-mode output is of the order of milliwatts. In a VECSEL the external mirror and the spot of the pump set the mode, so a large area of the chip, a spot of 220 $μ$ m in one device, contributes to a single TEM $_{00}$ mode. The gain region is also so thin that it adds little antiguiding nonlinearity, so the beam stays a large single-mode beam. The device gave $M^{2}$ of 1.05 at 1 W and 1.15 at 3 W, after the cavity length was tuned for maximum TEM $_{00}$ power. The external resonator, not the chip, sets the mode size, and for good beam quality its length should not be much smaller than the Rayleigh length of the beam. Reference: M. Kuznetsov, F. Hakimi, R. Sprague and A. Mooradian, “High-power (>0.5-W CW) diode-pumped vertical-external-cavity surface-emitting semiconductor lasers with circular TEM $_{00}$ beams,” IEEE Photon. Technol. Lett. 9, 1063 (1997), the first VECSEL; Wikipedia, Vertical-external-cavity surface-emitting-laser (accessed October 2026); Fan et al., Appl. Phys. Lett. 86, 211116 (2005); Vertical External-cavity Surface-emitting Lasers, RP Photonics Encyclopedia.

Ques: How much does the output power grow when the pump spot is enlarged?

In the ideal case the power grows with the area, because at constant pump intensity the heat flux and the temperature rise are the same. In a design study of a 1.32 $μ$ m chip the optimum spot width was 50 $μ$ m with a bottom heat spreader, 125 $μ$ m with an upper diamond and 270 $μ$ m with two spreaders. The ideal gain of going from 75 $μ$ m to 270 $μ$ m is the ratio of the areas, $(270/75)^{2} = 13$ (computed), while the paper says about 11; the arithmetic supports 13, and it reports an actual gain of 3.5, so thermal limits and not the pump area set the optimum. Threshold also rises with the spot, by 4.8 times from 50 $μ$ m to 100 $μ$ m, against 4 times ideally. Reference: Sokół and Sarzała, Int. J. Electron. Telecommun. 60(3), 239 (2014); Vertical External-cavity Surface-emitting Lasers, RP Photonics Encyclopedia.

Ques: What is the highest output power of a VECSEL?

The sources give different numbers because they refer to different beams and years. Wikipedia quotes 45 W continuous single-mode power announced by Coherent in 2004, the RP Photonics article up to 30 W (Chilla et al., 2004), and a thesis 106 W average in multimode operation (Heinen et al., 2012), 72 W with cryogenic cooling (Chernikov et al., 2011), and for mode-locking 5.1 W with 682 fs pulses (Scheller et al., 2012) and 4.35 kW peak at 3.3 W with 400 fs (Wilcox et al., 2013). The 106 W is the highest continuous value but is multimode, so for a diffraction-limited beam the 30 W to 45 W of 2004 is the relevant record, and the two numbers for 2004 probably refer to different devices, or to an announcement and a publication. Reference: Wikipedia, Vertical-external-cavity surface-emitting-laser (accessed October 2026); Vertical External-cavity Surface-emitting Lasers, RP Photonics Encyclopedia; thesis, University of Southampton (eprints.soton.ac.uk/381283).

Ques: Why do mode-locked VECSELs run at gigahertz repetition rates with only nanojoule pulses?

The repetition rate of a pulse circulating in a linear cavity is $c /2 L$ , so a few-millimetre to centimetre cavity gives 15 GHz at 1 cm and 3 GHz at 5 cm (computed), and 50 GHz was reached. The pulse energy is the average power divided by this rate. A peak power of 4.35 kW in 400 fs corresponds to about 1.7 nJ (computed, ignoring the pulse shape factor), and with 3.3 W average this implies a rate of about 2 GHz (inferred), which is in the usual few-GHz range. The pulse energies are therefore low, except with synchronous pumping, and the pulse duration is limited near 100 fs by kinetic hole burning. Reference: Vertical External-cavity Surface-emitting Lasers, RP Photonics Encyclopedia; A. Rahimi-Iman, Self-Mode-Locking and Frequency-Modulated Comb Semiconductor Disk Lasers, Photonics 12(7), 677 (2025); thesis, University of Southampton (eprints.soton.ac.uk/381283).

Ques: Why does frequency doubling work well inside the cavity of a VECSEL?

The light circulating in the cavity is much stronger than the output, because the output coupler transmits only a small part. For a lossless coupler of 1% transmission the circulating power is about 100 times the output (computed, assumed value), and the second-harmonic power grows as the square of the fundamental power, so a nonlinear crystal placed in the cavity converts efficiently even with a modest output. This gives watt-level continuous green, blue and red, and compact blue and green sources for projection displays. The sources give no conversion efficiencies, so the benefit is stated only qualitatively here. Reference: Vertical External-cavity Surface-emitting Lasers, RP Photonics Encyclopedia; Wikipedia, Vertical-external-cavity surface-emitting-laser (accessed October 2026).

Quartz 5

Explorer