Junior Engineer
3rd year bachelor thesis
A little hands-on experience with lasers
A little hands-on experience with lasers
My bachelor thesis describes the activity in the Optoelectronic Laboratory at Università degli Studi di Pavia. The characterization work has been performed on devices developed by ing. Marco Passerini at Glasgow University.
A Laser (Light Amplification by Stimulated Emission of Radiation) is an optical device that emits a coherent beam of light. Tipically it consists of an active medium, a pumping system and 2 mirrors. The pumping can be a source of light or an electrical source that gives energy to the electrons of the active medium. The active medium generates photons at a fixed wavelengh with the stimulated emission process. The mirrors provide a reflection of the photons inside the cavity to enhance the amplification of the beam. One of the mirrors is usually less reflective, providing laser emission.
A laser has namely infinite longitudinal modes, obtained by the relation f = mc/2nL , where m is an integer, c is the speed of light, n is the cavity refractive index and L is the cavity lenght. However, having a limited gain, the bandwidth is limited, as seen below.
In a semiconductor (SC) laser, the pumping system is usually a current injection, while the active medium is a semiconductor with direct band gap (the carrier momentum in the valence and conduction bands are the same). The mirrors are usually fabricated cleaving the faces of the semiconductor. A typical Power over Current characteristic is depicted in the figure below.
As we see, there is a zone below the threshold where the pumping is not effective, and a linear regime over threshold in which the optical power is proportional to the current. Also, the temperature greatly affects the semiconductor, thus they are often stabilized in temperature with Peltier cells.
To achieve high-frequency lasing, the two most common method for SC lasers are Q-switching and Mode-Locking. This work is focused on the latter. With M-L, the longitudinal modes of the SC laser are linked by phase matching, oscillating at the same time. This is possible inserting an optical switch in the cavity that modulates the optical losses. In fact this switch emits light with a period Δt equal to 2nL/c, that is exactly the inverse of the frequency separation of two longitudinal modes.
There are different ways for achieving Mode-Locking regime; in our case we will study a passive mode-locking, where the optical switch is a saturable absorber (SA).
I worked with 2-section SC lasers, lasing in passive Mode-Locking regime. The devices were fabricated using a GaAs/AlGaAs double quantum-well (DQW) material grown by metal-organic vapour-phase epitaxy (MOVPE). The semiconductor material structure consists of a heavily p-doped GaAs cap layer followed by 1 µm p-doped AlGaAs upper cladding layer. The active region is undoped and consists of two 10 nm GaAs quantum wells spaced by a 10 nm AlGaAs barrier. The quantum wells and the barrier are placed between two 0:25 µm undoped AlGaAs layers in a separate confinement structure. The lower cladding is a 1:6 µm n-doped AlGaAs layer and the substrate is formed by n-doped GaAs material. The double-section laser consists of a ridge monomodal waveguide with a width of 3 µm and a depth of 900 nm fabricated using two different processes:
Standard process
Self-alignment process
The first step consists in the definition of the ridge waveguide with lithographic technique, etched subsequently with SiCl4 with reactive ion etching (RIE). The photoresist is used as a mask for the RIE process. With a lift-off procedure, the photoresist is removed and a layer of SiO2 si deposited on the sample. With a second lithography step we define the current injection area, aligned to the ridge waveguide. The SiO2 dielectric is removed from this region via RIE process. Finally, the p and n-contacts are deposited.
Self-alignment stems from the need to avoid critical steps in the lithography process, in particular respect to the mask-aligner precision. This process consists into these steps:
A thick layer of photoresist is deposited on the substrate (around 1.8 µm) and it is exposed to define the ridge waveguide.
The exposed resist is removed with RIE.
Without removing the remaining photoresist, a layer of SiO2 is deposited.
Operating a lift-off, both the photoresist and the dieletric are removed, opening the access to the current injection. There is no need for a second lithographic step.
The metal electrode is divided in two sections: the longer is the gain area, the other is the saturable absorber. They are separated by 20 µm, providing a resistance of approximately 1.5 kΩ. The two zones can be defined with two different approaches.
The SA faces one of the two cavity mirrors. in this configuration, the longitudinal modes are separated by a Δf of c/2nL, as previously described.
The SA is fabricated in the middle of the cavity, thus providing a Δf equal to c/nL.
The lasers that I analyzed are fabricated to emit in M-L regime at 14, 35 and 60 GHz. The quality of fabrication is evaluated from three parameters:
Threshold current. The lower, the better, so that it limits the Joule heating.
Dynamic Resistance. As for the current, Rs should be low to avoid high heating over threshold.
Differential efficiency. Obtained from the characteristc I over P, indicates how much of the electrical power is converted into optical power. Depends on the mirrors efficiency and quality of the material.
It may happen that no threshold current is measured, that is the laser emits only with spontaneous emission. Or we can have an infinite Rs due to contact oxidation or bad annealing.
The SC lasers are mounted in group of 6 on the same slab. This slab is screwed to a copper support which is stabilized in temperature by a Peltier cell. The copper support is linked to a 10 kΩ thermistor that converts temperature variations into resistance variations. To test the lasers, a metallic probe is mounted on a magnetic base and connected to a current generator. The probe tip injects the current in the gain section of the laser. A power meter device is used to acquire the optical power exiting from the lasers.
Once everything is set, we determine the threshold current by increasing the current from the probe and reading the value of optical power. This operation will give us a rough range in which the laser passes from spontaneous to stimulated emission. At this point we refine the measurement with a LabView program, acquiring the values of optical power and laser voltage for a range of currents. Then, with a Matlab program, we create the graphs I over P and I over V. Thanks to this analysis, we divided the lasers into three categories:
As seen in the graph, those lasers have low threshold and the voltage increases slowly, meaning a few ohm resistance.
Despite a low threshold current, the voltage increases rapidly over threshold. This means that there have been small fabrication defects.
The fabrication was not succesful, some of them don't have stimulated emission or have a very high resistance.
The setup has been modified to characterize and measure fast optical signals. As before, we're using a current generator to inject current in the laser and a Peltier cell to keep the temperature at 25°C. The laser is inversely polarized in the gain section from 0 to -2 V by a voltage generator. An amperometer reads the current in the saturable absorber. The optical signal exiting from the laser is coupled to an optical fiber, single mode with 8° angle cut to avoid backreflections. The correct coupling is evaluated by an optical spectrum analyzer (OSA) that reads the optical power in real time. The optical fiber has been splitted in two branches to perform an efficient characterization:
First branch: Electrical spectrum analysis. The fiber end is coupled to a fast photodiode with 40 GHz bandwidth. This diode converts the optical signal into electrical signal, with a resolution of 0.05 A/W. The converted signal is read by a radio-frequency spectrum analyzer (RFSA) that dislay a graph P over f. For frequencies higher than 40 GHz, a frequency mixer is added to the setup.
Second branch: Optical spectrum analysis. The fiber tip is focused and the emission is collected by a Fabry-Perot interferometer. A ramp generator is connected to a piezoceramic on which a mirror of the interferometer is mounted. Changing the ramp value we move the mirror, thus changing the transmission peak of the Fabry Perot cavity. The beam exiting from the interferometer is then collected by a photodiode that converts the optical signal into a voltage, measured by an oscilloscope. So, changing the interferometer mirror position, we were able to measure the optical spectrum of the laser.
With the aforementioned setup, we analyzed the spectra of CPM lasers, obtaining different regimes. As previously seen, with low values of injected current, the laser is under threshold and the emission spectrum is continuous.
There is only one dominant frequency mode. The electrical spectrum, instead, is flat (a part from the noise).
Increasing the current up to a value two times the threshold, the laser enters the Mode-Locking regime. As seen below, the longitudinal modes are linked and they have a Δf equal to c/2nL.
The mode matching creates a pulsed electrical signal with a frequency around 16 GHz, as seen from the graph below, measured with the RFSA.
When the fabrication is well done, the laser can enter the CPM regime, in which the longitudinal modes are alternated, with a matching frequency equal to c/nL.
The electrical spectrum has a peak around 57 GHz, as shown below
During the thesis a total of 233 lasers have been charcaterized.