Edge-emitting laser

An edge-emitting laser is a laser diode in which the light travels along a waveguide parallel to the wafer surface and exits through a facet formed by cleaving the chip. The cavity is 250 µm to 5 mm long, a few micrometers wide in a single-mode ridge or up to hundreds of micrometers in a broad-area device, and the optical mode is confined vertically to a few hundred nanometers by the layers around the active region. The alternative geometry is the VCSEL, whose short cavity is perpendicular to the wafer and which emits through the top or bottom surface. Edge emitters dominate where power, wavelength coverage or narrow linewidth matter: telecom transmitters at 1.3–1.55 µm, pump lasers at 9xx nm, and diode bars of tens to hundreds of watts.

Cavity types

In a Fabry–Pérot edge emitter the two cleaved facets form the mirrors. The semiconductor–air interface reflects about 0.3 of the guided mode (the plane-wave Fresnel value for an index of 3.2 is 0.27), enough for a long, high-gain cavity. A 300 µm cavity with group index 3.6 has longitudinal modes spaced by 1.11 nm (139 GHz) at 1550 nm, and the laser usually runs on several of them. A DFB laser etches a grating along the waveguide to select a single longitudinal mode, and a DBR laser places gratings in separate passive sections that can be tuned. All three share the in-plane waveguide, the cleaved output and the beam shape.

Beam shape: fast and slow axes

The emitting aperture at the facet is roughly a micrometer high and a few micrometers wide, so diffraction spreads the beam more in the vertical direction. The fast axis, perpendicular to the layers, typically diverges at 25–40° FWHM and the slow axis, parallel to them, at 6–12°, as described under far-field divergence. The beam is also astigmatic: the fast-axis waist sits at the facet while the slow-axis waist of a gain-guided or broad-area device lies several to tens of micrometers behind it, so a single spherical lens cannot focus both axes to the same plane; in an index-guided ridge the two waists nearly coincide.

Treating each axis as a Gaussian beam gives the mode sizes at the facet. A FWHM full angle $\theta$ corresponds to a $1/e^2$ half-angle $\theta_0 = \theta/1.177$, and the waist radius is

$$w_0 = \frac{\lambda}{\pi\,\theta_0}.$$

At 1550 nm, 35° gives $w_0 \approx 0.95$ µm and 10° gives $w_0 \approx 3.3$ µm, an aspect ratio of 3.5. The paraxial formula is only approximate at 35°, but it is adequate for estimating coupling.

Coupling to fiber

Standard single-mode fiber at 1550 nm has a mode radius of about 5.2 µm. The Gaussian overlap of the elliptical laser mode with the round fiber mode, with no gap and perfect alignment, is

$$\eta = \frac{4\,w_x w_y w_f^2}{(w_x^2 + w_f^2)(w_y^2 + w_f^2)},$$

which for the radii above gives 0.32 (−4.9 dB) for butt coupling. A lens system with equal magnification in both axes can raise this to 0.69 (−1.6 dB) at a magnification near 2.9, and the remaining loss is the ellipticity. Recovering it requires anamorphic optics, a lensed fiber with an elliptical tip, or an on-chip spot-size converter that expands the mode before the facet. Silicon photonic circuits receive edge-emitter light through an edge coupler for the same reason.

Facets, coatings and damage

Facet coatings set the mirror loss and the split of power between front and back. A common choice is a low-reflection front coating and a high-reflection back coating, which sends most of the output through one facet; the trade-offs are worked through in facet coatings and what they do to a laser diode. The facet is also the weak point for power. Absorption at the cleaved surface heats it, and above an optical power density of order 1–10 MW/cm² in unprotected GaAs facets the heating runs away into catastrophic optical damage. Spread over a 100 µm × 1 µm broad-area aperture, that range corresponds to 1–10 W, which is why high-power emitters use wide stripes, long cavities with low internal loss, and passivated or non-absorbing facets.

Edge emitters and VCSELs compared

The long cavity gives high gain per pass and high power from one emitter; the VCSEL gives a round beam, low threshold, and on-wafer testing before the chips are separated.

Common questions

What is the difference between an edge-emitting laser and a VCSEL?

The edge emitter's cavity lies in the wafer plane and emits from a cleaved side facet; the VCSEL's cavity is vertical and emits from the surface. Edge emitters reach higher power and cover longer wavelengths more easily; VCSELs have round beams, lower threshold and can be tested on the wafer.

Why is the beam of an edge-emitting laser elliptical?

The emitting aperture is much thinner perpendicular to the layers than parallel to them, and diffraction spreads light faster from the narrower dimension. The fast axis therefore diverges three to five times more than the slow axis.

References: L. A. Coldren, S. W. Corzine and M. L. Mašanović, Diode Lasers and Photonic Integrated Circuits, 2nd ed. (Wiley, 2012); G. P. Agrawal and N. K. Dutta, Semiconductor Lasers, 2nd ed. (Van Nostrand Reinhold, 1993); B. E. A. Saleh and M. C. Teich, Fundamentals of Photonics, 3rd ed. (Wiley, 2019).