We have experimentally demonstrated an efficient all-fiber passively Q-switched Yb-doped fiber laser with Samarium doped fiber as a saturable absorber. Average output power of 3.4 W at a repetition rate of 250 kHz and a pulse width of 1.1 microseconds was obtained at a pump power of 9.0 W. By using this fiber laser system and an MgO-doped congruent periodically poled lithium niobate (MgO:c-PPLN), second harmonic generation (SHG) output at 532 nm was achieved at room temperature. The conversion efficiency is around 4.2% which agrees well with the theoretical simulation.
1. Introduction
Over the last decades, many
different display technologies have been developed to replace the traditional
cathode-ray tubs (CRTs), especially in large screen projection market for
business and home theater. Laser-based projection display seems to be a superior alternative to existing
projection systems because laser light sources can provide extremely wide color
expression. Lasers have faster response time and longer life time than any
lamps. Moreover, the small system size with high brightness and high power
efficiency will also attract the industry [1, 2].
For laser-based projection systems, a
high power and high efficient green laser source with compact size is now one
of key components. Traditional gas- and dye-based green lasers are not suitable
for this application due to their large size, inefficiency, and cumbersomeness to maintain.
Although many efforts have been focused on developing semiconductor green laser
diodes, the performance achieved to date is still not satisfied because of the
material limitation at this wavelength region. Nonlinear frequency conversion
of high-power lasers around 1060 nm now seems to be the most practical method
for many applications requiring high quality green lasers.
Visible fiber lasers have attracted
a lot of attention recently due to potential applications in the fields of
industrial processing, bioinstrumentation, and display. Comparing with the
conventional Nd:YAG solid-state lasers, the ytterbium (Yb)-doped fiber lasers
have strong potential for the systems where air-cooling, long term stability
and maintenance-free operation are required. Using Yb-doped fiber lasers and
nonlinear crystals to generate CW green light has already been reported [3–5]. But for laser
display application, high repetition rate pulse input is more attractive for SHG
process. Because the conversion efficiency is proportional to the square of
input power, which means under the same average input power, the SHG conversion
efficiency of a pulsed input will be much higher than that of a CW input due to
the higher pulsed peak power.
In this study, we report an air-cooled
simple all-fiber, passively Q-switched Yb doped fiber laser utilizing Sm-doped
fiber as an SA. Then using this fiber laser and an MgO-doped periodically poled
lithium niobate (MgO-PPLN) crystal, we have achieved high repetition green pulses
output at room temperature. The characteristics of fiber laser output and SHG
light have been investigated.
2. All-Fiber Q-Switched Yb-Doped Fiber Laser
Compact Yb-doped Q-switched fiber
lasers have attracted significant interests as it could provide a high power,
high efficiency and low cost source for many applications. There are different
technologies to Q-switch fiber laser that includes active Q-switching using,
for example, piezoelectric modulator [6], electrooptic modulator [7], and
acoustic modulator [8]. The other alternative is the passive Q-switching using
a saturable absorber (SA) in laser cavity that offers much simpler design as
compared to the active Q-switching. Huang et al. reported recently a passively
Q-switched Yb-doped fiber laser with Cr4+: YAG as an SA and achieve a high peak
power of 20 kW [9]. However their laser consists of a Cr4+: YAG slab, three
collimating and focus lenses, and a mirror that complicated substantially the
laser cavity design. Dvoyrin et al. reported pulsed Yb-doped fiber lasers with
Bismuth-doped fiber as an SA [10]. They succeeded in Q-switching in wide
wavelength range from 1050 to 1200 nm with pulse width from 1 to 1.5 microseconds
at a repetition rate from 10 to 100 kHz. Nevertheless, the stable pulse
operation was only obtained in double-cavity design and the peak power was
limited to 40 W at 1064 nm. Fotiadi et al. described the dynamics of an
all-fiber self-Q-switched Yb-doped fiber laser with a samarium absorber [11].
Regular pulse operation was observed at discrete pump levels and the dynamics
was attributed to polarization mode switching. It appears that though saturable
absorbers have been successfully employed in Q-switching solid-state lasers, it
is still an interesting research topic to achieve high repetition rate, high
peak power Q-switching in Yb-doped fiber laser.
2.1. Experiments
The schematic of the fiber laser
configuration is shown in Figure 1. It consists of two pump lasers of 975 nm to
provide bidirectional pump, 10.5 meters of twin-core fiber, a piece of Sm-doped
fiber as an absorber and one highly reflective fiber Bragg grating (FBG). The
twin-core fiber has two optically contacted large cores, coated with a low
refractive index polymer as shown in Figure 2. The single-mode Yb-doped active
core has a diameter of 6.7m and an NA of 0.13. The multimode core has a diameter of 118m with a multimode absorption coefficient of 0.77 dB/m at the pump wavelength. The advantage of this coupling
method is the ability to increase the output power from the fiber laser by
coupling multiple pump sources at both ends, or by exploiting the physical
separation between the pump and active paths in which the independent access to
the pump path allows for power scaling by adding the pump power along the fiber
length. The efficient light coupling between the pump and active cores had been
described recently in [12]. The single mode Sm-doped SA was obtained from
CorAcitve (Sm119) with an NA of 0.14 and a core diameter of 6.3m. Its peak
absorption near 1064 nm was measured to be 8 dB/m.
Figure 1: Schematic of an all-fiber Q-switched fiber
laser design.
Figure 2: SEM image of the cross-section of
the twin-core fiber coated with lower index of polymer. The left core is for
pumping and the right one has a Yb-doped core.
We select Sm-doped absorber for its
fast response time [13]. In its near IR absorption band, the lifetime of upper
energy level is less than 5 nanoseconds that is essential to achieve narrow
pulse width and high repetition rate when used in Q-switching.
The
diode pump laser has a fiber-coupled output though a multimode fiber of 105/125m core/cladding diameter, capable of delivering up to 6.5 W of optical power at
975 nm. The
high reflectivity FBG was fabricated in house with a reflectivity of >99.9%
and a −3 dB bandwidth of 0.6 nm centered at 1063.37 nm. The cleaved end of the
active fiber served as a 4% output coupler.
The temporal behavior of pulse train
was recorded with an InGaAs photodiode with a bandwidth of 80 MHz (New Focus,
model: 2117) and a digital oscilloscope (Tektronix, model: TKD 3032B). The rise time of the InGaAs photodiode is
about 3 nanoseconds. Optical spectra were recorded with an optical spectral
analyzer (OSA) (Ando Corp. model: AQ-6315E).
A power meter (Coherent, model: FieldMaster-GS) was used to measure CW
laser power or average power of a Q-switched pulse train.
In our laser configuration, no
active cooling was required for all fiber components. Two pump diode lasers
were mounted on heat dissipate plates and fiber assembly used a standard
anodized aluminum fiber tray mounted on an aluminum metal plate, which
separates the electrical and optical compartments in the system. The power
supply, providing up to 10A of current, uses a conventional fan for convection
cooling.
2.2. Results and Discussion
Before inserting the Sm-doped SA,
the CW output of the laser, assembled with same components less than that of an
SA, was characterized by providing
a reference for the performance of Q-switched laser output. The CW output as a function of pump power is
plotted in Figure 3. In CW operation, the laser delivered 5 W at a pump power of
9 W with a slope efficiency reached 68%.
Figure 3: The average output power versus
pump power: for CW output (); for Q-switched output with a 0.45 m of
Sm-absorber (); that with a 0.6
m of Sm-absorber (◆).
After inserting 0.6 m of Sm-doped
SA, the Q-switched output gave an average power of 2.8 W at a pump power of 8.2 W with a power conversion efficiency of 41%.
The laser with a shorter Sm-absorber fiber length of 0.45 m improved the
slope efficiency and produced higher average power (see Table 1 for more
details). The average output power of the Q-switched laser as function of the
pumped power for two Sm-absorber lengths is also plotted in Figure 3. The Q-switching
efficiency (the ratio of the Q-switched slope efficiency to that of CW one) was
69% and 60% for 0.45 m and 0.6 m of Sm-absorber, respectively, indicating that
Sm-abosrber is a good candidate for Q-switching fiber laser.
Table 1: Parameters of Yb-Sm Q-switched fiber laser.
Stable Q-switched pulses were
obtained when the pump power was at much higher level than the threshold value.
Figure 4 displays the oscilloscope trace of the pulse train. The pulse shape
exhibits a fast leading edge and a slow falling edge that is typical for the
Q-switched pulse. The pulse width was measured to be 800 nanoseconds at FWHM
and the pulse duration is 3.6 microseconds for a 0.6 m long Sm-doped fiber at
8.4 W pump power. The pulse-to-pulse stability was better than ±5%. If the pump
power is increased over 9 W, the average output power that was attributed to the onset of nonlinear effects will start to decrease.
Figure 4: Typical Q-Switched pulse train
recorded with oscilloscope at a pump power of 8.4 W.
The spectrum of the Q-switched pulse
output was measured by an OSA with its resolution set at 0.05 nm. Figure 5
shows the spectrum at an average output power of 1.0 W for 0.45 m of
Sm-absorber. Its −3 dB bandwidth is as broad as 0.158 nm.
Figure 5: The output spectrum of the
Q-switched fiber laser.
3. Second Harmonic Generation
Single-pass wavelength conversion by
a nonlinear optical (NLO) crystal becomes one of the most attractive methods to
generate coherent radiation in various spectral domains from ultraviolet to
mid-infrared since QPM was first proposed in the early days of nonlinear optics
[14]. Especially, as periodically poling technology became commercially
available. Compared with other phase matching methods such as birefringent
phase matching (BPM) technique, QPM technique allows us to use the largest
nonlinear coefficient over the whole transparent spectral range of the crystal
without the walk-off effect.
Among all the periodically poled
crystals, periodically poled lithium niobate (PPLN) has been one of the most
attractive materials due to its large second-order nonlinear coefficient, wide
transparency band width covering UV to IR, mature and low cost fabrication
process [15]. However, PPLN will suffer from pointing instability at medium
powers and permanent photorefractive damage at high power [16]. Usually PPLN
crystals have to heat to the temperature over hundred degrees to reduce these
effects. This effect also can be reduced by doping some magnesium oxide (MgO)
into the crystal during growth. Compared to pure LiNbO3 crystals, 4%-5% MgO:c-PPLN crystals can increase the resistance against optical damage
hundred times higher, which allows devices working with high input and output
power at room temperature [17].
3.1. Experiments
An electric-field poling technique
was applied to fabricate the periodic domain structure on a 0.5 mm thick 3-inch
5% MgO:c-PPLN wafer. Typically, the applied voltage is around 21 kV. A uniform
periodically poled structure with periods of 6.5m was obtained with a 50%
duty ratio on the +Z face.
The experimental setup is
schematically shown in Figure 6. Fundamental light from Q-switched fiber laser
is collimated by a 10x objective lens (Lens-1). Since only the TM-polarization can satisfy the
QPM condition, a polarizer is used to make the fundamental light’s polarization
parallel with the Z-direction of PPLN. A
small part of fundamental light is separated by a beam splitter and focused by
Lens-4 onto an InGaAs detector (Thorlabs, model: PDA10CS). The main stream of
fundamental beam is focused by two lens (Lens-2&3). These two lenses can
adjust the beam diameter to satisfy the optimum confocal condition for
different sample length. In this paper, the sample length is 15 mm so the beam
is focused to a 50m spot radius in the center of the sample.
Figure 6: Second harmonic generation experimental setup.
The sample was mounted in a
temperature-controlled oven (Thorlabs, model: TC200) and the temperature was
set to 34.5°C which optimizes the QPM condition of the sample. SHG light was
focused by Lens-5 and a bandpass filter is used to block the fundamental light.
An Si detector (Thorlabs, model: PDA10A) is used for the measurement of the
pulses trains. Both two detectors are connected to a digital oscilloscope
(LeCroy, Model: LT224) for pulses trains recording. In this way, we can record
the fundamental pulses and SHG pulses synchronously in order to accurately compare
each other. The responsivities of two detectors were calibrated by two CW DPSS
lasers at 1064 nm and 532 nm, respectively.
3.2. Results and Discussion
Figure 7 shows the
comparison of a fundamental pulse and an SHG pluse. The fiber laser used in SHG
experiments is connected with 0.45 m Sm-absorber, as described before; the
fundamental pulse width is about 1.1 microseconds. Becasue the SHG power is
proportional to the squre of fundamental pulse, the SHG pulse FWHM width should
be 0.707 fundamental pulse. As shown in Figure 7, the SHG pulse width is about
0.75 microsecond.
Figure 7: Fundamental pulse and SHG pluse.
Figure 8 shows the output spectrum
of SHG light at 531.64 nm. The −3 dB bandwidth is about 0.09 nm which is
slightly narrower than that of the fundamental light at 1063.25 nm.
Figure 8: The output spectrum of SHG light.
The SHG peak power is plotted
against the incident fundamental peak power as shown in Figure 9. The SHG peak
power of 152 mW was achieved when the fundamental peak power was 3.6 W (average
power is 1 W). The conversion efficiency (SHG pulse power/fundamental pulse
power) is 4.2% which agrees with the theoretical simulation result. The conversion efficiency can be
expected to be above 90% when the input fundamental peak power is in kW level.
Figure 9: SHG peak power as a function of the fundamental peak power: the triangle dots
and solid line represent experimental data and the theoretical simulation
result.
4. Conclusion
We have
demonstrated an all-fiber, high repetition rate and high peak power passively
Q-switched Yb-doped fiber laser using Sm-doped fiber as an SA. With a pump
power of 8.4 W, the fiber laser outputs stable pulses with a pulse width of 1.1
microseconds at a repetition rate of 250 kHz, producing an average output power
of 3.4 W. All fiber connections were fusion spliced and fiber was looped in a
cm3 compact fiber tray. This laser design showed a significant
improvement as compared with other Q-switched fiber lasers and has the
following features: (a) all fiber components were splice connected, which
avoided the use of any bulk optics, (b) realized bidirectional pumping using a
twin-core fiber that separated pump path from the active signal path, (c) demonstrated
high repetition rate in the order of 200 kHz.
Using
this novel fiber laser and an MgO-PPLN wavelength conversion device, we have
obtained the green laser with peak power of 150 mW at room temperature. The
conversion efficiency is 4.2% which agrees well with the theoretical value. The
MgO:c-PPLN sample is 15 mm long and the temperature is controlled at
34.5°C to
satisfy QPM condition. The conversion efficiency will be higher if we can
achieve high peak power for fundamental light from fiber laser. Further increasing the peak
power is currently limited by the nonlinear effects in fiber laser and will be
improved by the use of larger core active fiber. The all-fiber configurations provide a
simple, low cost and easy to manufacturing solution for high power Q-switched
pulse laser that should be attractive to micromachining, medical treatment, and
other applications.
Acknowledgments
This work is partially supported by Natural Sciences and
Engineering Research Council of Canada (NSERC), Canadian Institute for
Photonics Innovations (CIPI), Ontario Centres of Excellent (OCE), and Canadian
Foundation for Innovation (CFI).