Open Access
Add to BibSonomyAbstract
Two simple external cavity diode laser designs using fibre pigtailed gain chips are tested and their properties compared with a high end DBR fibre laser. These ECDLs demonstrate a FWHM linewidth as low as 5.2kHz with a fitted Lorentzian FWHM linewidth as low as 1.6kHz. Tuning ranges of 200nm covering 1420nm to 1620nm were demonstrated. To the best of our knowledge these are the narrowest linewidth and most broadly tunable external cavity diode lasers reported to date. The improvement in linewidth is attributed to greatly enhanced acoustic isolation allowed by using fiber coupled gain chips and by replacing kinematic mounts with a pair of rotatable wedges for cavity alignment which eliminates acoustic resonances. A detailed description and discussion of techniques used to characterize the frequency noise and linewidths of these lasers is provided.
© 2014 Optical Society of America
1. Introduction
Narrow linewidth tunable laser sources are fundamental to a vast array of applications in fields including atomic physics, spectroscopy, quantum information, coherent communications, remote sensing and precision measurement. Across these fields, simpler, cheaper, lasers with narrower linewidths and increased tuning ranges continue to enable new applications and broader use of this technology.
Over the past 50 years the history of tunable lasers has largely mirrored the development of laser technology in general. Initial dye lasers have been replaced by External Cavity Diode Lasers (ECDLs), while higher power systems have been dominated by tunable solid state lasers such as Ti:Sapphire or frequency converted Nd:YAG lasers using optical parametric oscillators (OPOs) [1]. Diode lasers without stabilizing external cavities have filled the low cost, low performance end of the market with commercially available DFB and DBR diodes offering linewidths as narrow as 500kHz. More recently, fibre lasers and frequency converted fibre lasers have begun to replace many of the solid state systems with different designs offering higher powers and greater tunability [2–4], or narrower linewidths [5,6]. The advent of frequency combs now allows the synthesis of light almost anywhere with outstanding stability and accuracy [7]. Yet, for all this, the External Cavity Diode Laser retains its position as the workhorse of so many laboratories by virtue of its simplicity, versatility, respectable performance and very low cost.
External Cavity Diode Lasers (ECDLs) first emerged in the early 1980s as an alternative to complex and expensive dye and solid state lasers [8,9]. By the early 1990s simple designs which achieved linewidths of a few 100 kHz and many nm of tuning range were common and well understood [10,11]. Over the ensuing years many different configurations and designs have been demonstrated, each with their own advantages and disadvantages. Probably the most common configurations are the Littrow [8,10–12] and Littman [13] configurations. Today, diode gain media suitable for the construction of ECDLs are readily available covering the 400nm to 2μm range, while at 1.55μm tuning ranges up to 120nm (for a 10dB power reduction) are commercially available (eg TLK-L 1550R from Thorlabs).
The frequency noise in an external cavity diode laser primarily derives from 1/f and white noise. The resulting autocorrelation spectrum typically includes both Lorentzian and Gaussian aspects and may be described by a Voigt function [14]. At high frequencies the autocorrelation spectrum appears Lorentzian. This largely has its origins in the well understood spontaneous emission process which gives rise to the natural linewidth or Schawlow-Townes linewidth (with several broadening factors). This component depends inversely on the power in the lasing mode and ECDLs with fitted Lorentzian full width at half maximum (FWHM) linewidths around a kHz have been reported since the 1980s [15–18]. At low frequencies Gaussian noise dominates. This Gaussian noise is independent of the laser power, and increases with the measurement integration time, as it comes primarily from environmental sources [14]. Despite careful design aiming to keep mechanical resonances above typical environmental acoustic noise frequencies FWHM Gaussian linewidths have remained above 52kHz [19–21] (for a 0.33ms integration time) and typically above 100kHz. By actively locking ECDLs to an external reference such as a stabilized isolated high finesse cavity, it is possible to achieve linewidths as low as 1Hz [22] but the complexity and expense of such techniques limit their widespread use. This has left users who are seeking narrower linewidth sources primarily reliant upon fiber distributed feedback (DFB FL) and fibre distributed Bragg reflector (DBR FL) lasers, which achieve linewidths in the 1-10kHz range [5]. Although compact and narrow linewidth DFB FL and DBR FL sources are readily available, they have very small tuning ranges (typically 1nm) and they are only available at select wavelengths which limits their utility for many applications. Furthermore, in contrast to ECDLs they require specialized infrastructure for fabrication, limiting the ability of small laboratories to experiment with designs and optimize the performance for each application. Recently, fibre coupled external cavity diode lasers using feedback from a bragg grating written into a planar waveguide have demonstrated free running linewidths of the order of a kHz when acoustically isolated [23]. Since these lasers use feedback from a bragg grating in a waveguide tuning is extremely limited with up to a few tenths of a nm demonstrated. However, this tuning performance is similar to DFB FL and DBR FL systems without being limited to wavelengths corresponding to fibre gain media.
In this paper we describe several simple ECDL designs and characterize their performance. Linewidths as low as 5.2kHz with a 1ms integration time are measured which is up to an order of magnitude better than results previously reported from free running ECDLs [19–21]. Increased tuning ranges up to 200nm are also demonstrated making these, to the best of our knowledge, the most broadly tunable ECDLs reported to date. The improvements in linewidth are attributed to high levels of acoustic isolation from the environment made possible by fibre coupled nature of the gain chip and by the use of a pair of rotatable wedges for cavity alignment which eliminates all the acoustic resonances normally associated with the use of kinematic mounts. These simple changes suggest an easy avenue by which to upgrade ECDLs employed in other laboratories. Our particular laser is implemented at 1560nm as a cheaper and vastly more tunable alternative to commercial fibre DFB FL and DBR FL lasers yet exhibits similar noise performance and is thus ideal for applications such as the laser cooling of Rubidium [4].
2. Laser design
It is often taken for granted that the pointing direction of a laser will not drift and that a laser may be aligned to a sensitive component such as a high finess modecleaner or a single mode fiber without fear that pointing drifts will undo all one’s work. We make this assumption because laser cavities are generally robustly fixed to their housings which in turn are fixed to a reference plane such as an optical table to which they are bolted. The cost of rigidly fixing a laser cavity to its housing and a reference plane is that all the environmental noise of that reference and on the housing will couple into the laser cavity to some extent. Great efforts have been employed to dampen vibrations, stiffen the system and shift mechanical resonances in some commercial ECDLs in order to minimize the laser coupling to the environment without compromising pointing stability. An alternative approach is to fibre couple the ECDL output allowing the laser cavity to be robustly acoustically and mechanically isolated from its environment while the pointing stability is maintained by rigidly fixing the fibre output end to the application system’s reference plane. Recently, polarization maintaining, fibre coupled, isolated, half-butterfly diode gain chips with a very broad gain bandwidth have become commercially available (eg SAF1550P2, Thorlabs). This has allowed fibre coupled ECDLs, which are strongly decoupled from their environment, to be very easily and inexpensively constructed.
Two different laser configurations have been implemented using such gain chips and their performance has been characterized. In the first grating based ECDL (G-ECDL) configuration (Fig. 1(a)), the tuning element consists of a holographic reflective diffraction grating with 1100 lines/mm and a diffraction efficiency around 90% at 1.56μm (05HG1100-900-1, Newport/ Richardson) mounted on a commercial kinematic mount (POLARIS-K1-H, Thorlabs). This is perhaps the simplest of possible designs and is implemented to test what advantages may be offered by simply moving to a fibre coupled output with strong environmental isolation and little additional effort.
In the second filter based ECDL (F-ECDL) configuration (Fig. 1(b)), tuning is accomplished by rotating the angle of an intracavity band pass dielectric filter. As the filter is rotated away from normal incidence the pass band is shifted to shorter wavelengths. The use of such an intracavity filter for tuning is not new. The advantages of this arrangement in combination with a cavity mirror retro-reflector for minimizing acoustic environmental coupling have been clearly described as far back as 1988 [24]. Since then, numerous implementations of this basic design have been demonstrated [25–29] including a 52kHz FWHM linewidth ECDL source [19]. Recently, an implementation using a fibre coupled gain chip in concert with a tunable filter and a fixed etalon suggested FWHM linewidths of 20kHz or better may be achieved. However, the strong structure observed in this measurement and the lack of details describing the measurement method made the results ambiguous [29].
The filter used in the F-ECDL design is a commercial DWDM filter manufactured by Lightwaves 2020 with a 100GHz pass band. Its transmission spectrum is shown in Fig. 2 and is broadly similar in performance to filters used in similar filter based ECDLs [19]. Rather than the traditional approach in which cavity alignment is achieved by adjusting the angle of a kinematically mounted cavity mirror or by translating the X-Y position of a cavity lens, our cavity is aligned by rotating a pair of 2° AR coated wedge prisms. The advantage of this approach is its insensitivity to vibration. Even under extreme vibration a wedge will not tend to oscillate or spin on its axis and position translation of a wedge has no effect to first order. On the other hand, a lens in an X-Y translator tends to vibrate in position and a kinematic cavity mirror will vibrate in angle. Using a pair of rotatable wedges for alignment allows us to directly glue our cavity mirror to a piezo-electric transducer (PZT) which in turn is glued to the laser housing. Similarly, the collimating lens is glued in its mount minimizing X-Y translation and vibration. Together, these changes eliminate most moving parts and ideally all significant acoustic resonances.
Fig. 2 Insertion loss (IL) vs. wavelength at normal incidence for the commercial DWDM filters used in the F-ECDL. Indicative sample variation in the filter performance is plotted for two different loss scales.
A popular approach taken to improve stability in similar filter based ECDL designs, particularly those which use kinematically mounted cavity mirrors is the use of a retro-reflector or cat’s eye configuration for the cavity mirror [19,27–29]. Focusing and retro-reflecting the beam from the cavity mirror reduces sensitivity to fluctuations in the mirror’s angular position. In the case of kinematically mounted cavity mirrors such fluctuations are typically the dominant source of laser cavity instability while in the F-ECDL the cavity mirror is glued to the housing and is thus much more stable than a kinematically mounted mirror.
In both ECDL configurations a commercial fibre coupled diode gain chip (SAF1550P2, Thorlabs) is used as the gain medium. Collimation is achieved using a 2.97mm focal length, AR coated moulded aspheric lens (355660-C, Thorlabs) glued to a flexure mount. The driver used for current, temperature and piezo control was a MOGbox DLC-202 from MOGlabs. The piezo output was low pass filtered with fast tuning performed using the diode bias current. For acoustic isolation the lasers rested on 5cm of acoustic dampening foam.
Fast tuning of the lasers may be achieved by adjusting the diode bias current or by adjusting the cavity length with a piezo. Since the filter in the F-ECDL is not simultaneously angle tuned its mode hop free tuning range is limited to about one free spectral range. The optical path length of the cavity was around 6cm giving a free spectral range (FSR) of around 2.5GHz. In the G-ECDL a mode hop free tuning range of many free spectral ranges may be achieved by ensuring the grating is angle tuned at the same rate the cavity resonant frequency is tuned. In practice this is achieved by ensuring the pivot of the mount holding the grating lies at the intersection of the plane of the grating and a plane normal to the laser beam intersecting with the diode facet [12,30].
The isolator built into the output of the gain chip was found to be insufficient to prevent feedback in a number of situations so an additional dual stage 50dB fibre coupled polarization maintaining isolator with the fast axis blocked was incorporated after the gain chip in both configurations.
3. Characterization and discussion
The output was fibre coupled in single transverse mode polarization maintaining fibre with the fast axis blocked in the isolator and was thus robustly single mode and polarized. The measured threshold and slope efficiency of the G-ECDL is shown in Fig. 3(a) giving 36mW from a diode current of 340mA which was current limited by our diode driver. The gain chip itself is capable of operating at up to 500mA which gives an extrapolated output of around 50mW. This is similar to commercial ECDLs which use the same gain chip (TLK-L1550R, Thorlabs). The transmission losses on the filter and wedges used in the F-ECDL were around 20% double pass, higher than the 10% grating loss in the G-ECDL. This resulted in a higher threshold and lower slope efficiency producing an output of 20mW at 350mA when tuned to 1559nm. However, when the filter in the F-ECDL was tuned normal to the laser cavity a coupled cavity was formed in which some of the power reflected from the faces of the filter was coupled back into the gain chip. This significantly decreased cavity losses increasing the power output when tuned between 1559nm and 1561nm. The slope efficiency and threshold when tuned for maximum wavelength and power are shown in Fig. 3(b).
Fig. 3 Output power was measured as a function of wavelength (while being tuned over several tens of nm) for both the (a) G-ECDL and (b) F-ECDL configurations. Insets: Output power vs. diode current for each configuration is plotted at a given wavelength as indicated on each figure.
The tuning performance was characterized by manually adjusting the grating or filter angle while measuring the output power with a power meter and the wavelength using an optical spectrum analyzer (AQ-6315A). The results for the G-ECDL are plotted in Fig. 3(a) demonstrating a 10dB tuning range of just over 200nm. If the cavity is optimized for 1560nm without the fast tuning piezo sandwiched in the mount and the cavity alignment is left fixed the tuning range was slightly reduced to around 190nm. The tuning curve for the F-ECDL without any cavity optimization as it is tuned is shown in Fig. 3(b) and gives an 18nm 10dB tuning (relative to the 1559nm power level).
In the G-ECDL with the fast tuning piezo sandwiched between the fine pitched adjuster and the grating mount faceplate it was found that the G-ECDL misaligned significantly as it was manually tuned. For this reason, the piezo fast adjustment was removed for the remaining measurements. Using a commercially available fine adjuster with an integrated piezo (such as POLARIS-K1PZ, Thorlabs) would likely eliminate this issue.
Linewidth measurement
The linewidths of the two ECDLs, and for comparison a commercial DBR fibre laser (Rock RFLS-35-3-1560.48-NSI, NP Photonics), were measured using both the Delayed Self-Heterodyne Interferometer (DSHI) method and using an unbalanced-path Mach-Zehnder Interferometer (MZI). The ECDL linewidths were measured at a current of 300mA with a tuning of 1560nm for both the F-ECDL and G-ECDL.
The delayed self-heterodyne interferometer (DSHI) technique works by using a delay which is longer than the coherence length of a laser to produce a local oscillator with an uncorrelated phase. When mixed, the resulting spectrum is a convolution of the original frequency noise autocorrelation spectrum. In the case of a Lorentzian line shape, the FWHM linewidth is half the fitted Lorentzian FWHM linewidth. For a Gaussian line shape the FWHM linewidth is a factor of smaller than the fitted value. For real line shapes the difference will be somewhere in between [31]. A schematic of this measurement system is shown in Fig. 4. Using an 83km fibre delay line, the two ECDL lasers and the commercial DBR FL were each measured and the results plotted in Fig. 5. These results are summarized in Table 1. At high frequencies the noise spectra of both ECDLs was Lorentzian but the DBR FL showed strong sidebands at around 500kHz, overwhelming a very small Lorentzian component. The fitted Lorentzians for the F-ECDL and the G-ECDL were 3.28kHz and 6.72kHz FWHM respectively. These give deconvolved Lorentzian FWHM linewidths of 1.64kHz and 3.36kHz respectively. At low frequencies the noise had a Gaussian form for all three lasers. The fitted Gaussians gave 8.63kHz FWHM for the F-ECDL, 14.1kHz FWHM for the G-ECDL and 16.7kHz FWHM for the Rock DBR fibre laser. The corresponding deconvolved FWHM linewidths are 6.1kHz for the F-ECDL, 10.0kHz for the G-ECDL and 11.8kHz for the Rock DBR fibre laser. The true deconvolved FWHM linewidths will be a little less than the deconvolved Gaussian FWHM linewidths since the line shape is not Gaussian and the true deconvolution factor will be greater than . The resolution bandwidth used was 1kHz giving an effective integration time of 1ms. The shot noise intercepts with the fitted Lorentzians for the F-ECDL and the G-ECDL were at 11.2MHz and 22.9MHz respectively. It is important to validate the original assumption that the delay line is long compared with the coherence length. For a Lorentzian line shape the coherence time τc may be related to the FWHM linewidth νFWHM by τc = 1/(πνFWHM) and for a Gaussian line shape the relationship is [32]. For the F-ECDL we find a delay to coherence length ratio of 2.1 assuming a Lorentzian and 3.7 assuming a Gaussian line shape. A ratio of 6 has been suggested as a good rule of thumb [33] for obtaining clean reliable DSHI measurements, although the lack of any apparent delta function or modulation in the spectrum, together with the use of fitting to estimate linewidths gives confidence in the measurements made. Measuring lasers with linewidths in the kHz range requires extremely long delay lines. An alternative arrangement is the use of recirculating loops [34] although effective delays by this technique remain restricted to around 180km [35].
Fig. 4 Linewidth measurement configuration using the delayed self-heterodyne interferometer (DSHI) method with delay lines of 2.27km or 83km.
Fig. 5 Delayed self heterodyne measurement with 83km delay line, 1kHz resolution bandwidth and 150 averages. (left) 2.5MHz span with fitted Lorentzians - filter based ECDL (3.28kHz FWHM fit giving a 1.64kHz FWHM linewidth), Grating based ECDL (6.72kHz FWHM fit giving a 3.36kHz FWHM linewidth), The sidebands on the Rock prevent an accurate Lorentzian fit. (right) 100kHz span, with fitted Gaussians – filter based ECDL (8.6kHz FWHM fit giving a 6.1kHz FWHM linewidth), Grating based ECDL (14.1kHz FWHM fit giving a 10.0kHz FWHM linewidth), NP Photonics Rock (16.7kHz FWHM fit giving a 11.8kHz FWHM linewidth).
Table 1. Summary of linewidth results.
When the self-heterodyne delay time is comparable to the coherence time (or smaller) the resulting power spectrum appears strongly modulated with a delta function at the heterodyne frequency. Measuring linewidths as the 3dB bandwidth with any delta function present would be highly inaccurate. DSHI measurements using short delay lines can lead to confusing or deceptive results [14,33,36] yet this practice remains common. The power spectral density at frequency ω, with heterodyne frequency Ω, a delay time τ and a laser coherence time τc is given by Eq. (1) (note that brackets have been added to the original expression given in [33]).
The self-heterodyne power spectrum for the F-ECDL was measured for a 2.27km delay line with the results plotted in Fig. 6 showing the characteristic delta function and modulation profile. For comparison we plot Eq. (1) filtered by a 1kHz Gaussian resolution bandwidth filter for a coherence time τc = 1.43ms showing excellent agreement. This corresponds to a FWHM Lorentzian linewidth 1/(πτc) [32] of 220Hz which is a great deal smaller than the 1.64kHz FWHM Lorentzian linewidths obtained using the 83km delay line. Furthermore, it is difficult to observe let alone make any reliable measurement of the Gaussian noise contribution which is effectively filtered out for small delay times [14]. This highlights the complexity of correctly interpreting linewidth measurements using self-heterodyne measurements with short delays.
Fig. 6 Delayed self heterodyne measurement with 2.27km delay line, 1kHz resolution bandwidth and 150 averages and 2.5MHz span. Also plotted is the calculated spectrum from Eq. (1) using a 1kHz FWHM Gaussian filter assuming a coherence time τc = 1.43ms corresponding to a Lorentzian FWHM linewidth 1/(πτc) [32] of 220Hz.
The practical difficulties of using suitably long delay lines for measuring very narrow linewidth lasers have led to a rise in the use of unbalanced path length Michelson or Mach Zehnder interferometers (MZI) as frequency discriminators particularly in the fibre DBR and fibre DFB communities [5,6,18]. An unbalanced interferometer produces a sinusoidal transmission response as the input laser frequency is varied and thus may be used as a frequency discriminator. A full fringe spacing corresponds to a frequency difference of 1/τd where τd is the delay time of the path imbalance. If a laser sits at the mid-fringe of the interferometer, the frequency as a function of time and thus the frequency noise may be directly mapped to the interferometer output. To maintain a calibrated response it is important that the frequency excursions of the laser from the mid-fringe frequency remain small compared to the fringe spacing. This is usually achieved by locking the interferometer or laser to mid-fringe. It is important that the locking loop bandwidth be small compared to the linewidth of the laser as any laser frequency noise within the locking loop bandwidth will be strongly suppressed, invalidating measurements at these frequencies. It is also important that such a MZI be acoustically isolated as environmental fluctuations in the interferometer path length at frequencies outside the locking bandwidth will be indistinguishable from laser frequency noise potentially increasing the measured laser frequency noise. A Michelson interferometer employing a faraday rotator offers the added advantage of suppressing polarization noise in the delay fibre, although this architecture was not used here.
A schematic of the MZI used is included in Fig. 7. The path length imbalance used was 300m giving a fringe spacing of 680kHz. In order to maintain a calibrated linear response, frequency deviations from the mid-fringe frequency must be small compared to this value, of the order of 70kHz. From the self-heterodyne linewidth measurements in Fig. 5, it is clear that the structure of the ECDLs fits this criterion, however the side bands present on the DBR FL are a cause for concern. Ideally one would use a 10m path imbalance giving a 20MHz fringe spacing but the signal from a given frequency deviation would be 30 times smaller which, in our case resulted in signals comparable with the detection noise floor. For this reason we may expect the DBR FL noise measured by this system to be smaller than the actual value.
Fig. 7 Linewidth measurement using an unbalanced Mach-Zehnder Interferometer (MZI) with a path imbalance (delay line) of 300m.
The signal is calibrated by offsetting the voltage controlled oscillator (VCO) driving the acousto-optic modulator (AOM) in the interferometer by a few Hz relative to the oscillator used to mix down the output resulting in a fringe scan. The RMS signal voltage is measured which corresponds to of the fringe spacing or a 120kHz frequency excursion. The measured noise spectrum for each laser and the detection noise floor is plotted in Fig. 8.
Fig. 8 Frequency noise power spectrum in Hz/Sqrt(Hz) using the unbalanced Mach Zehnder Interferometer with a 300m path imbalance and the interferometer locked to the respective lasers using a PID locking loop with a 30Hz bandwidth. The peaks at 30Hz results from ripple in the locking loop at the 30Hz corner frequency. The beta separation line described in [37] is also shown (dashed). Noise at frequencies less than the intersection with the beta separation line contributes primarily to Gaussian noise while noise at higher frequencies contributes primarily to Lorentzian noise in the wings of an autocorrelation spectrum.
The process for obtaining a linewidth from the frequency noise spectrum is nontrivial but a simple method for estimating the Gaussian linewidth is described in [37]. Noise at frequencies below the intersection with the beta separation line given by (where f is the frequency) contributes primarily to the Gaussian component of the line shape, while noise at frequencies above the intersection contributes primarily to the Lorentzian component and the spectral wings. Integrating above 80Hz to avoid regions affected by the locking loop we obtain 12.5ms integration time FWHM Gaussian linewidths of 12.7kHz for the F-ECDL, 17.7kHz for the G-ECDL and 16.0kHz for the DBR FL. If we integrate above 1kHz for the 1ms FWHM Gaussian linewidths we obtain 5.2kHz for the F-ECDL, 8.7kHz for the G-ECDL and 12.2kHz for the DBR FL. These are consistent with the Gaussian linewidths obtained using the 83km DSHI measurement of 6.1kHz, 10.0kHz and 11.8kHz for the F-ECDL, G-ECDL and DBR fibre lasers respectively.
The frequency noise spectrum provides a great deal more information than can be obtained from a single linewidth measurement allowing the origins of noise sources to be identified and eliminated. There are two main differences between the spectra of the G-ECDL and the F-ECDL. Firstly, the overall noise level on the G-ECDL is higher and secondly there are a number of acoustic resonances around 1kHz in the G-ECDL.
The G-ECDL is a highly conventional ECDL design. The only significant distinction this ECDL has from other systems reported is that the output from the diode is fibre coupled allowing the cavity to be strongly acoustically isolated from its environment. There remain some acoustic resonances visible in the G-ECDL frequency noise spectrum which we attribute to resonances in the kinematically mounted grating. On the other hand, the F-ECDL has no acoustic resonances visible in its frequency noise spectrum which can be attributed to the removal of the kinematic mirror mount in favour of using a rotatable wedge pair for alignment. It is an open question whether the performance may be further improved using a retro-reflector design for the rear cavity mirror in either laser configuration.
5. Summary
Simple grating and filter based ECDL designs using a fibre coupled gain chip have been demonstrated and characterized. The 10dB tuning range of the grating and filter ECDLs was 200nm and 18nm respectively. The output powers measured were up to 36mW and 45mW respectively. The linewidths of each ECDL were measured and compared to a commercial DBR fibre laser (NP Photonics Rock) using an 83km delayed self-heterodyne measurement and a 300m unbalanced Mach Zehnder interferometer. The linewidth results are summarized in Table 1. The FWHM Gaussian linewidth for the filter ECDL and grating ECDLs were 5.2kHz and 8.7kHz respectively for a 1ms integration time and 12.7kHz and 17.7kHz for a 12.5ms integration time. The fitted FWHM Lorenztian linewidths were 1.64kHz and 3.36kHz for the filter and grating based ECDLs respectively. The FWHM linewidths achieved are to our knowledge the lowest reported linewidths for free running ECDLs and were superior to the commercial DBR fibre laser used for comparison. These results show a simple inexpensive design path for the construction of widely tunable ECDLs with linewidths in the kHz range. This opens the way for the use of ECDLs in place of DBR and DFB fibre lasers in applications requiring narrower linewidths than have traditionally been possible from ECDLs.
Acknowledgments
The authors would like to thank Alexei Tikhomirov and Scott Foster from DSTO for helpful comments on linewidth measurement and Timothy Lam from ANU for the loan of equipment. J. E. Debs would like to acknowledge financial support from the IC postdoctoral fellowship program. C. C. N. Kuhn would like to acknowledge financial support from CNPq (Conselho Nacional de Desencolcimento Cientifico e Tecnologico). Product specifications are for clarity only.
References and links
1. F. J. Duarte, Tunable Lasers Handbook (Academic, San Diego, 1995).
2. J. Nilsson, W. A. Clarkson, R. Selvas, J. K. Sahu, P. W. Turner, S. U. Alam, and A. B. Grudinin, “High-power wavelength-tunable cladding-pumped rare-earth-doped silica fiber lasers,” Opt. Fiber Technol. 10(1), 5–30 (2004), http://www.sciencedirect.com/science/article/pii/S1068520003000464. [CrossRef]
3. D. Y. Shen, J. K. Sahu, and W. A. Clarkson, “High-power widely tunable Tm:fibre lasers pumped by an Er,Yb co-doped fibre laser at 1.6 mum,” Opt. Express 14(13), 6084–6090 (2006). [CrossRef] [PubMed]
4. S. S. Sané, S. Bennetts, J. E. Debs, C. C. N. Kuhn, G. D. McDonald, P. A. Altin, J. D. Close, and N. P. Robins, “11 W narrow linewidth laser source at 780 nm for laser cooling and manipulation of Rubidium,” Opt. Express 20(8), 8915–8919 (2012). [CrossRef] [PubMed]
5. C. Spiegelberg, J. Geng, Y. Hu, Y. Kaneda, S. Jiang, and N. Peyghambarian, “Low-noise narrow-linewidth fiber laser at 1550 nm (June 2003),” J. Lightwave Technol. 22(1), 57–62 (2004). [CrossRef]
6. S. Foster, G. A. Cranch, and A. Tikhomirov, “Experimental evidence for the thermal origin of 1/f frequency noise in erbium-doped fiber lasers,” Phys. Rev. A 79(5), 053802 (2009), http://link.aps.org/doi/10.1103/PhysRevA.79.053802. [CrossRef]
7. T. Udem, R. Holzwarth, and T. W. Hänsch, “Optical frequency metrology,” Nature 416(6877), 233–237 (2002), doi:. [CrossRef] [PubMed]
8. C. E. Wieman and L. Hollberg, “Using diode lasers for atomic physics,” Rev. Sci. Instrum. 62(1), 1–20 (1991), http://scitation.aip.org/content/aip/journal/rsi/62/1/10.1063/1.1142305. [CrossRef]
9. J. C. Camparo, “The diode laser in atomic physics,” Contemp. Phys. 26(5), 443–477 (1985), http://www.tandfonline.com/doi/abs/10.1080/00107518508210984. [CrossRef]
10. K. B. MacAdam, A. Steinbach, and C. Wieman, “A narrow‐band tunable diode laser system with grating feedback, and a saturated absorption spectrometer for Cs and Rb,” Am. J. Phys. 60(12), 1098–1111 (1992), http://scitation.aip.org/content/aapt/journal/ajp/60/12/10.1119/1.16955. [CrossRef]
11. L. Ricci, M. Weidemüller, T. Esslinger, A. Hemmerich, C. Zimmermann, V. Vuletic, W. König, and T. W. Hänsch, “A compact grating-stabilized diode laser system for atomic physics,” Opt. Commun. 117(5–6), 541–549 (1995), http://www.sciencedirect.com/science/article/pii/003040189500146Y. [CrossRef]
12. L. Nilse, H. J. Davies, and C. S. Adams, “Synchronous tuning of extended cavity diode lasers: the case for an optimum pivot point,” Appl. Opt. 38(3), 548–553 (1999). [CrossRef] [PubMed]
13. S. E. Park, T. Y. Kwon, E.-J. Shin, and H. S. Lee, “A compact extended-cavity diode laser with a Littman configuration,” IEEE Trans. Instrum. Meas. 52(2), 280–283 (2003), http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=1202029.
14. L. B. Mercer, “1/f frequency noise effects on self-heterodyne linewidth measurements,” J. Lightwave Technol. 9(4), 485–493 (1991), http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=76663. [CrossRef]
15. R. Wyatt and W. J. Devlin, “10 kHz linewidth 1.5μm InGaAsP external cavity laser with 55 nm tuning range,” Electron. Lett. 19(3), 110–112 (1983), http://digital-library.theiet.org/content/journals/10.1049/el_19830079. [CrossRef]
16. R. Wyatt, “Spectral linewidth of external cavity semiconductor lasers with strong, frequency-selective feedback,” Electron. Lett. 21(15), 658–659 (1985), http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=4250665. [CrossRef]
17. N. A. Olsson and J. P. van der Ziel, “Performance characteristics of 1.5μm external cavity semiconductor lasers for coherent optical communication,” J. Lightwave Technol. 5(4), 510–515 (1987), http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=1075530. [CrossRef]
18. J. M. Kahn, C. A. Burrus, and G. Raybon, “High-stability 1.5μm external-cavity semiconductor lasers for phase-lock applications,” IEEE Photon. Technol. Lett. 1(7), 159–161 (1989), http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=36024. [CrossRef]
19. D. J. Thompson and R. E. Scholten, “Narrow linewidth tunable external cavity diode laser using wide bandwidth filter,” Rev. Sci. Instrum. 83(2), 023107 (2012), http://scitation.aip.org/content/aip/journal/rsi/83/2/10.1063/1.3687441. [CrossRef] [PubMed]
20. H. Talvitie, A. Pietiläinen, H. Ludvigsen, and E. Ikonen, “Passive frequency and intensity stabilization of extended-cavity diode lasers,” Rev. Sci. Instrum. 68(1), 1–7 (1997), http://scitation.aip.org/content/aip/journal/rsi/68/1/10.1063/1.1147810. [CrossRef]
21. S. D. Saliba and R. E. Scholten, “Linewidths below 100 kHz with external cavity diode lasers,” Appl. Opt. 48(36), 6961–6966 (2009). [CrossRef] [PubMed]
22. T. Nazarova, C. Lisdat, F. Riehle, and U. Sterr, “Low-frequency-noise diode laser for atom interferometry,” J. Opt. Soc. Am. B 25(10), 1632–1638 (2008). [CrossRef]
23. K. Numata, J. Camp, M. A. Krainak, and L. Stolpner, “Performance of planar-waveguide external cavity laser for precision measurements,” Opt. Express 18(22), 22781–22788 (2010). [CrossRef] [PubMed]
24. P. Zorabedian and W. R. Trutna Jr., “Interference-filter-tuned, alignment-stabilized, semiconductor external-cavity laser,” Opt. Lett. 13(10), 826–828 (1988). [CrossRef] [PubMed]
25. E. Luvsandamdin, S. Spießberger, M. Schiemangk, A. Sahm, G. Mura, A. Wicht, A. Peters, G. Erbert, and G. Tränkle, “Development of narrow linewidth, micro-integrated extended cavity diode lasers for quantum optics experiments in space,” Appl. Phys. B 111(2), 255–260 (2013), doi:. [CrossRef]
26. T. Hieta, M. Vainio, C. Moser, and E. Ikonen, “External-cavity lasers based on a volume holographic grating at normal incidence for spectroscopy in the visible range,” Opt. Commun. 282(15), 3119–3123 (2009), http://www.sciencedirect.com/science/article/pii/S0030401809004180. [CrossRef]
27. X. Baillard, A. Gauguet, S. Bize, P. Lemonde, P. Laurent, A. Clairon, and P. Rosenbusch, “Interference-filter-stabilized external-cavity diode lasers,” Opt. Commun. 266(2), 609–613 (2006), http://www.sciencedirect.com/science/article/pii/S0030401806004561. [CrossRef]
28. M. Gilowski, C. Schubert, M. Zaiser, W. Herr, T. Wübbena, T. Wendrich, T. Müller, E. M. Rasel, and W. Ertmer, “Narrow bandwidth interference filter-stabilized diode laser systems for the manipulation of neutral atoms,” Opt. Commun. 280(2), 443–447 (2007), http://www.sciencedirect.com/science/article/pii/S0030401807008577. [CrossRef]
29. N. Wang, M. Feng, Z. Feng, M. Y. Lam, L. Gao, B. Chen, A. Q. Liu, Y. H. Tsang, and X. Zhang, “Narrow-Linewidth Tunable Lasers With Retro-Reflective External Cavity,” IEEE Photon. Technol. Lett. 24(18), 1591–1593 (2012), http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=6253231. [CrossRef]
30. P. McNicholl and H. J. Metcalf, “Synchronous cavity mode and feedback wavelength scanning in dye laser oscillators with gratings,” Appl. Opt. 24(17), 2757–2761 (1985). [CrossRef] [PubMed]
31. T. Okoshi, K. Kikuchi, and A. Nakayama, “Novel method for high resolution measurement of laser output spectrum,” Electron. Lett. 16(16), 630–631 (1980), http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=4244210. [CrossRef]
32. B. E. A. Saleh and M. C. Teich, Fundamentals of Photonics (Wiley, New York, 1991).
33. L. Richter, H. I. Mandelberg, M. Kruger, and P. McGrath, “Linewidth determination from self-heterodyne measurements with subcoherence delay times,” IEEE J. Quantum Electron. 22(11), 2070–2074 (1986), http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=1072909. [CrossRef]
34. J. W. Dawson, N. Park, and K. J. Vahala, “An improved delayed self-heterodyne interferometer for linewidth measurements,” IEEE Photon. Technol. Lett. 4(9), 1063–1066 (1992). [CrossRef]
35. H. Tsuchida, “Limitation and improvement in the performance of recirculating delayed self-heterodyne method for high-resolution laser lineshape measurement,” Opt. Express 20(11), 11679–11687 (2012). [CrossRef] [PubMed]
36. P. Horak and W. H. Loh, “On the delayed self-heterodyne interferometric technique for determining the linewidth of fiber lasers,” Opt. Express 14(9), 3923–3928 (2006). [CrossRef] [PubMed]
37. G. Di Domenico, S. Schilt, and P. Thomann, “Simple approach to the relation between laser frequency noise and laser line shape,” Appl. Opt. 49(25), 4801–4807 (2010). [CrossRef] [PubMed]
