Fundamental Transverse Mode Selection (TMS#0) of Broad Area Semiconductor Lasers with Integrated Twice-Retracted 4f Set-Up and Film-Waveguide Lens

Previouslywe focused on fundamental transversemode selection (TMS#0) of broad area semiconductor lasers (BALs)with two-arm folded integrated resonators for Fourier-optical spatial frequency filtering. The resonator had a round-trip length of 4f, where f is the focal length of the Fourier-transform element (FTE), that is, a cylindrical mirror in-between the orthogonal resonator branches. This 4f set-up can be called “retracted once” due to the reflective filter after 2f ; that is, the 2f path was used forwards and backwards. Now the branches are retracted once more resulting in a compact 1f long linear resonator (called “retracted twice”) with a roundtrip length of 2f. One facet accommodates the filter, while the other houses the FTE, now incorporating a film-waveguide lens. The BAL facet with the filter represents both the Fourier-transform plane (after 2f, i.e., one round-trip) as well as the image plane (after 4f, two round-trips). Thus filtering is performed even after 4f, not just after 2f. Experimental results reveal good fundamental TMS for pump currents up to 20% above threshold and a one-dimensional beam quality parameterM 1D = 1.47.The BALs are made from AlGaInAsSb, but the concept can equally well be employed for BALs of any material system.


Introduction
Broad area (semiconductor diode) lasers (BALs) allow for the generation and extraction of higher light output powers as compared to conventional narrow stripe lasers.The higher power for larger pump currents is distributed over a larger cross-section thus avoiding catastrophic optical mirror damage (COMD).Typical BALs exhibit emitter widths of 50-200 m.The drawback is represented by the build-up and oscillation of larger transverse modes (in cases without further provisions) limiting the portion of the higher light power easily focused into a small volume or onto a small area.
Low spatial frequencies stand for low transverse mode numbers and vice versa.Thus, in order to achieve fundamental mode TMS a low-pass filter has to be applied.

Concepts
Formerly we have integrated a Fourier-optical 4f set-up into the semiconductor resonator with two orthogonal branches The spatial filter to support the fundamental mode is made up of a nontilted and several tilted/oblique flanks/segments, the latter intended to refract and reflect unwanted rays out of the resonator and to reflect the supported rays back into the resonator.Fresnel reflectivity of ≈31% at the semiconductor-air interface is used.In the bottom part of the figure there is an SEM (scanning electron microscope) image of the facet with filter of an actual TMS#0-BAL sample.[24,25], as sketched in Figure 1.The resonator length has been 2f and the round-trip path length has come up to 2f plus 2f after Fresnel-reflection at the spatial frequency filter.This 4f set-up might be called "retracted once" due to the reflective arrangement.The dry-etched bend in-between branches has been cylindrical and has officiated as reflective Fourier-transform element.One of the resonator facets has been used as the output plane of the laser; the other facet has housed the spatial frequency filter, realized by flanks for redirection of certain groups of light rays.
In detail: rays with low propagation angles with respect to the optical axis account for low spatial frequencies and thus for the fundamental transverse mode (#0).They are Fresnelreflected back into the resonator at the semiconductor-air interface of the central (nontilted) filter segment with a reflectivity of about 31%.Rays with larger propagation angles, which correlate with larger spatial frequencies and higher transverse modes, are refracted and reflected out of the resonator by tilted/oblique dry-etched flanks/segments of the spatial filter.
The principle has worked well.But the two-branch resonator has shown some disadvantages like a strong astigmatism due to the 45 ∘ tilt of the Fourier-transform mirror and an overall resonator length of ≈5 mm or more, too long for bestpossible laser efficiency [26].This statement touches a principle problem of the integration of a 4f set-up: The desired minimum mode width of several 10 m, about equal to the central filter segment width of 2 0 , necessitates a mirror focal length of several millimeters, considering the properties of the 1D spatial Fourier-transform, vacuum wavelength  0 , and (effective) refractive index  (eff) [9], but for best differential quantum efficiency  a semiconductor laser resonator should rather be short [26].Thus, currently, that is, for this contribution, we have retracted the branches (once more) so that they spatially coincide, giving a compact linear 2f geometry (1f plus the path after reflection, the resonator thus called "retracted twice"), as sketched in Figure 2.
One resonator facet contains the low-pass spatial frequency filter to support low spatial frequencies and hence the fundamental transverse mode (#0).The other resonator end accommodates the Fourier-transform element.
The BAL facet with the filter represents both the Fouriertransform plane (after 2f, i.e., after one round-trip, one Fourier-transformation) as well as the image plane (after 4f, two round-trips, both Fourier-transformations).
The laser is an edge-emitting pn diode with intrinsic active region.The p-and the n-doped regions above and below the active region also serve as the upper and the lower filmwaveguide cladding, respectively.The layer sequence is made from the AlGaInAsSb material system on GaAs substrate and incorporates eight active layers of Stransky-Krastanovgrown GaAsSb quantum dots [27] embedded in GaAs layers.Therefore, the laser emission wavelength is around 933 nm.But the TMS-BAL concept reported here can equally well be utilized for BALs of any material system and emission wavelength.
Ridge lasers are realized, as sketched in more detail in Figure 3.The laser ridge is 100 m wide, ≈2 m high, and ≈2.8 mm long.The central nonoblique segment of the spatial frequency filter is 2 0 = 30 m wide.As this segment is considerably narrower than the ridge (some) low-pass spatial filtering is performed even after 4f, not just after 2f, which is an advantage of this approach for TMS: due to the 30 m narrow central segment intensity maxima of potentially oscillating higher order transverse modes are stripped, inevitably resulting in suppression of these modes.
The reflective Fourier-transform element is made up of a cleaved plane semiconductor facet in combination with a collecting film-waveguide lens.The latter is realized by dryetching into the upper cladding of the layer sequence stopping 50 nm below the upper cladding (i.e., within the active region); see also Section 3.This way the effective refractive index  eff of any laser mode is reduced.(Due to the reduction of the refractive index in the region of the waveguide lens, its top-viewed shape has to resemble that of a diverging lens known from free-space optics in order to achieve a collective function.)Half of the lens function is achieved before reflection of the wave at the plane facet and the other half after this reflection on the wave's way back.The focal length  of the complete (biconcave) film-waveguide lens (with its two plane-concave halves) is supposed to be equal to the resonator length  ( = ).
The radius of curvature  of the waveguide lens is 46 m, giving a diameter 2 of 92 m, that is, nearly equal to the ridge width (see Figures 3 and 2 again).

Technological Work
The monocrystalline layer sequences are epitaxially grown with our molecular beam epitaxy (MBE) system R450 by DCA Instruments Oy, Turku, Finland.The central nontilted segment of the spatial frequency filter and the facet on the Fourier element side of the resonator are cleaved.The other (i.e., tilted) flanks are implemented with standard photolithography, using a shared dry-etch step.For the dry-etch processes (reactive ion etching (RIE)) a MicroSys 350 machine by Roth & Rau, Wuestenbrand, Germany, is used.
A second dry-etch step has to be employed for etching into the cladding in the area of the film-waveguide lens.The replacement of the original cylindrical mirror (the curved semiconductor facet) by the combination of the cleaved plane facet with the dry-etched film-waveguide lens is done for technological reasons: for the hypothetic case with a curved (nonoblique) mirror the curvature would have to be so small (the radius of curvature  so large) that the distance along the optical axis between the mirror edges and the center of the mirror would be around 225 nm only, which is not suitable for standard photolithography.
The desired etch-depth for the waveguide lens has to be hit with an accuracy of better than 25 nm.Otherwise the focal length of the waveguide lens would differ from the resonator length by more than 30% [28], resulting in nonexact 1D spatial Fourier-transformation and nonoptimal spatial filtering.
In order to have an in situ (real-time) etch-depth control with this demanding accuracy we have successfully employed an adequate measurement techniques, by transferring the concept of reflectance anisotropy spectroscopy (RAS), which is well-known from epitaxy for growth control by now [29][30][31][32][33][34][35], to reactive ion etching (RIE) of monocrystalline semiconductor layer sequences [36][37][38][39].We have used argon as the plasma gas and 2 volume-% of chlorine as the reactive gas.This way the RIE-RAS signals and signal transients have been meaningful and we have achieved an accuracy in in situ etch-depth control of 16 nm.(In cases with low etch rates (below ≈75 nm/min) we have even resolved monolayer etch ablation [39].)The details of the RIE-RAS principle and of its experimental results during film-waveguide etching are given elsewhere [28,38].Figure 4: Experimental results, that is, near-(a) and far-field (b) intensity distributions, for a BAL without spatial filter or waveguide lens for com parison.The laser has exactly the same layer sequence as that with the results in Figure 5, since both lasers have been made from the same wafer.The measurements are performed at 91 K device temperature in continuous wave operation.As expected there is no evidence of fundamental transverse mode selection (TMS#0).Figure 5: Experimental results for a TMS#0-BAL with spatial filter to support the fundamental transverse mode (#0) and with film-waveguide lens.The intensity profiles are taken on the filter side of the resonator.The laser has exactly the same layer sequence as that from Figure 4, since both lasers have been made from the same wafer.Again (a) gives the near-field, (b) the far-field intensity distribution.The measurements are performed at 91 K device temperature in continuous wave operation.Obviously the fundamental transverse mode is strongly supported.But filamentation is not suppressed.Gaussian fits are introduced into the diagrams as red lines.

Experimental TMS Results and Discussion
The devices show clear laser characteristics.The best specimen emits an optical power of about 10 mW/facet continuous wave at room temperature and at a pump current of 20% above the laser threshold current  th .This very sample shows a good differential quantum efficiency of  = 2.5%.
The TMS measurements were performed at a device temperature of 91 K. To illustrate the TMS results the emission of two BAL devices made from the same wafer (identical layer sequences) will be compared.In Figure 4 the near-and far-field intensity distributions are reproduced for the case neither with a spatial frequency filter nor with a waveguide lens.It is obvious from the results depicted in Figure 4 that there is no transverse mode selection at all, as expected.
On the contrary, in Figure 5 the corresponding near-and far-field intensity distributions are given for a TMS#0-BAL, that is, a laser with spatial frequency filter for mode #0 and film-waveguide lens.The intensity profiles are taken on the filter side of the resonator.The fundamental transverse mode (#0 with one intensity maximum) is strongly supported.
But dips and bumps in the envelope of the near-field intensity distribution are discernible as compared to the Gaussian fit (red line).In principle they might either be related to the additional (weak) oscillation of an undesired higher transverse mode (i.e., #4 with five intensity maxima here) or the occurrence of filamentation [40][41][42][43][44][45].But the filter plane also serves as the image plane of the 4f set-up, where, due to the 30 m narrow central segment, intensity maxima of potentially oscillating higher order transverse modes are stripped, inevitably resulting in suppression of these modes.Thus the deviations from the desired smooth envelope are likely to be attributable to filamentation.
The intensity distributions allow for calculation of a 1D beam quality parameter, that is, with a full far-field angle of 2 1D = 4.1 ∘ .These (best) results are taken at 10% above laser threshold, but similar values are achieved for pump currents of up to 20% above threshold.At pump currents more than 25% above threshold TMS is not observed.

Conclusions
A concept for fundamental spatial transverse mode selection (TMS#0) of edge-emitting broad area (semiconductor diode) lasers (BAL) is presented, which employs a twiceretracted 4f set-up with an actual length of 1f integrated into the laser resonator.One resonator facet incorporates the spatial frequency filter; the other one houses a plane facet in combination with a film-waveguide lens as the Fouriertransform element.Experimental results show good fundamental transverse mode selection for pump currents of up to 20% above threshold, but filamentation is not suppressed.The best one-dimensional beam quality parameter measured is  2 1D = 1.47.

Figure 1 :
Figure1: Former concept (4f set-up "retracted once"): sketch of a BAL laser with integrated 4f set-up in the configuration with two orthogonal branches.The spatial filter to support the fundamental mode is made up of a nontilted and several tilted/oblique flanks/segments, the latter intended to refract and reflect unwanted rays out of the resonator and to reflect the supported rays back into the resonator.Fresnel reflectivity of ≈31% at the semiconductor-air interface is used.In the bottom part of the figure there is an SEM (scanning electron microscope) image of the facet with filter of an actual TMS#0-BAL sample.

Figure 2 :Figure 3 :
Figure 2: Current concept: sketches and SEM micrographs (top-views) of a TMS#0-BAL with compact linear f long laser resonator (4f set-up "retrac ted twice") and spatial filter to support the fundamental transverse mode (#0).The central filter segment has a width of 2 0 = 30 m.

1 0
Normalized near-field intensity Near-field position on laser facet (m)