Patent Application
20100086251
The present invention relates to the production of optical pulses at a desired wavelength using soliton self-frequency shift in higher-order-mode fibers.
The phenomenon of soliton self-frequency shift (SSFS) in optical fiber in which Raman self-pumping continuously transfers energy from higher to lower frequencies (Dianov et al., JETP. Lett. 41:294 (1985)) has been exploited over the last decade in order to fabricate widely frequency-tunable, femtosecond pulse sources with fiber delivery (Nishizawa et al., IEEE Photon. Technol. Lett. 11:325 (1999); Fermann et al., Opt. Lett. 24:1428 (1999); Liu et al., Opt. Lett. 26:358 (2001); Washburn et al., Electron. Lett. 37:1510 (2001); Lim et al., Electron. Lett. 40:1523 (2004); Luan et al., Opt. Express 12:835 (2004). Because anomalous (positive) dispersion (β2<0 or D>0) is required for the generation and maintenance of solitons, early sources which made use of SSFS for wavelength tuning were restricted to wavelength regimes >1300 nm where conventional silica fibers exhibited positive dispersion (Nishizawa et al., IEEE Photon. Technol. Lett. 11:325 (1999); Fermann et al., Opt. Lett. 24:1428 (1999)). The recent development of index-guided photonic crystal fibers (PCF) and air-core photonic band-gap fibers (PBGF) relaxed this requirement with the ability to design large positive waveguide dispersion and therefore large positive net dispersion in optical fibers at nearly any desired wavelength (Knight et al., IEEE Photon. Technol. Lett. 12:807 (2000)). This allowed for a number of demonstrations of tunable SSFS sources supporting input wavelengths as low as 800 nm in the anomalous dispersion regime (Liu et al., Opt. Lett. 26:358 (2001); Washburn et al., Electron. Lett. 37:1510 (2001); Lim et al., Electron. Lett. 40:1523 (2004); Luan et al., Opt. Express 12:835 (2004)).
Unfortunately, the pulse energy required to support stable Raman-shifted solitons below 1300 nm in index-guided PCFs and air-core PBGFs is either on the very low side, a fraction of a nJ for silica-core PCFs, (Washburn et al., Electron. Lett. 37:1510 (2001); Lim et al., Electron. Lett. 40:1523 (2004)) or on the very high side, greater than 100 nJ (requiring an input from an amplified optical system) for air-core PBGFs (Luan et al., Opt. Express 12:835 (2004)). The low-energy limit is due to high nonlinearity in the PCF. In order to generate large positive waveguide dispersion to overcome the negative dispersion of the material, the effective area of the fiber core must be reduced. For positive total dispersion at wavelengths <1300 nm this corresponds to an effective area, Aeff, of 2-5 μm2, approximately an order of magnitude less than conventional single mode fiber (SMF). The high-energy limit is due to low nonlinearity in the air-core PBGF where the nonlinear index, n2, of air is roughly 1000 times less than that of silica. These extreme ends of nonlinearity dictate the required pulse energy (U) for soliton propagation, which scales as UD·Aeff/n2. In fact, most microstructure fibers and tapered fibers with positive dispersion are intentionally designed to demonstrate nonlinear optical effects at the lowest possible pulse energy, while air-core PBGFs are often used for applications that require linear propagation, such as pulse delivery. For these reasons, previous work using SSFS below 1300 nm were performed at soliton energies either too low or too high (by at least an order of magnitude) for many practical applications, such as multiphoton imaging where bulk solid state lasers are currently the mainstay for the excitation source (Diaspro, A., Confocal and Two-Photon Microscopy, Wiley-Liss:New York (2002)).
There are a number of biomedical applications that require femtosecond sources. Although applications requiring a large spectral bandwidth (such as optical coherence tomography) can also be performed using incoherent sources such as superluminescent diodes, techniques based on nonlinear optical effects, such as multiphoton microscopy and endoscopy, almost universally require the high peak power generated by a femtosecond source.
Molecular two-photon excitation (2PE) was theoretically predicted by Maria Goppert-Mayer in 1931 [1]. The first experimental demonstration of two-photon absorption [2], however, came nearly 30 years later, after the technological breakthrough of the invention of the ruby laser in 1960. It was almost another 30 years before the practical application of 2PE for biological imaging was demonstrated at Cornell University in 1990 [3]. Once again, this new development was propelled in large part by the rapid technological advances in mode-locked femtosecond lasers [4, 5]. Since then, two-photon laser scanning microscopy has been increasingly applied to cell biology and neurosciences [6-10]. A number of variations, including three-photon excitation (3PE) [1-14], second and third harmonic generation imaging [15-17], near-field enhanced multiphoton excitation [18] and multiphoton endoscopic imaging [19], have emerged and further broadened the field, which is currently known as multiphoton microscopy (MPM). Today, MPM is an indispensable tool in biological imaging. Like any nonlinear process, however, multiphoton excitation requires high peak intensities, typically 0.1 to 1 TW/cm2 (TW=1012 W). Besides tight spatial focusing, MPM typically requires pulsed excitation sources to provide additional temporal “focusing” so that efficient multiphoton excitation can be obtained at low average power. For example, a femtosecond laser with 100-fs pulse width (τ) at 100 MHz pulse repetition rate (f) will enhance the excitation probability of 2PE by a factor of 105, i.e., the inverse of the duty cycle (fτ). The development of multiphoton imaging depends critically on ultrafast technologies, particularly pulsed excitation source.
Endoscopes play an important role in medical diagnostics by making it possible to visualize tissue at remote internal sites in a minimally invasive fashion [20]. The most common form employs an imaging fiber bundle to provide high quality white light reflection imaging. Laser scanning confocal reflection and fluorescence endoscopes also exist [21, 22] and can provide 3D cellular resolution in tissues. Confocal endoscopes are now becoming available commercially (Optiscan Ltd, Australia, Lucid Inc, Rochester) and are being applied in a number of clinical trials for cancer diagnosis. Multiphoton excitation based endoscopes has attracted significant attention recently. There were a number of advances [23], including fiber delivery of excitation pulses [24], miniature scanners [25], double clad fibers for efficient signal collections [26], etc. Thus, just like MPM has proven to be a powerful tool in biological imaging, multiphoton endoscopes have great potentials to improve the capability of the existing laser-scanning optical endoscopes. It is quite obvious that a compact, fully electronically controlled, femtosecond system seamlessly integrated with fiber optic delivery is essential for multiphoton endoscopy in medical diagnostics, particularly to biomedical experts who are not trained in lasers and optics.
Perhaps the most promising and successful area in biomedical imaging that showcases the unique advantage of multiphoton excitation is imaging deep into scattering tissues [10]. In the past 5 to 10 years, MPM has greatly improved the penetration depth of optical imaging and proven to be well suited for a variety of imaging applications deep within intact or semi-intact tissues, such as demonstrated in the studies of neuronal activity and anatomy [27], developing embryos [28], and tissue morphology and pathology [29]. When compared to one-photon confocal microscopy, a factor of 2 to 3 improvement in penetration depth is obtained in MPM. Nonetheless, despite the heroic effort of employing energetic pulses (˜μJ/pulse) produced by a regenerative amplifier [30], MPM has so far been restricted to less than 1 mm in penetration depth. One promising direction for imaging deep into scattering tissue is to use longer excitation wavelength. Although the “diagnostic and therapeutic window,” which is in between the absorption regions of the intrinsic molecules and water, extends all the way to ˜1300 nm (see water absorption spectrum in
Shortly after the inception of MPM, mode-locked solid state femtosecond lasers, most commonly the Ti:S lasers [5, 31], have emerged as the favorite excitation sources to dominate the MPM field today. When compared to earlier ultrafast lasers, e.g., ultrafast dye lasers, the Ti:S lasers are highly robust and flexible. The concurrent development of the mode-locked Ti:S lasers was perhaps the biggest gift for MPM and enabled MPM to rapidly become a valuable instrument for biological research. Nonetheless, the cost, complexity, and the limited potential for integration of the hulk solid state lasers have hampered the widespread applications of MPM in biological research. The fact that a disproportionate number of MPM systems are located in physics and engineering departments [32], instead of the more biologically oriented institutions, reflects at least in part the practical limitations of the femtosecond pulsed source. Obviously, the requirement of a robust, fiber delivered, and cheap source is even more urgent for multiphoton endoscopy in a clinical environment.
Mode-locked femtosecond fiber lasers at 1.03 and 1.55 μm [33, 34] have been improving significantly in the last several years, mainly in the output pulse energy (from 1 to ˜10 nJ) [35]. Even higher pulse energy can be achieved in femtosecond fiber sources based on fiber chirped pulse amplification [36]. However, femtosecond fiber sources, including lasers and CPA systems, have seen only limited applications in multiphoton imaging. The main reason is that they offer very limited wavelength tunability (tens of nanometer at best), severely restricting the applicability of these lasers, making them only suitable for some special purposes. In addition, existing femtosecond fiber sources at high pulse energy (>1 nJ) are not truly “all fiber,” i.e., the output are not delivered through a single mode optical fiber. Thus, additional setup, typically involving free-space optics, must be used to deliver the pulses to the imaging apparatus, partially negating the advantages of the fiber source. Reports have demonstrated the possibility of propagating femtosecond IR pulses through a large core optical fiber at intensities high enough (˜1 nJ) for multiphoton imaging [24]. In addition, a special HOM fiber that is capable of delivery energetic femtosecond pulses (˜1 nJ) has been demonstrated [37]. However, both fibers have normal dispersion, and both require a free-space grating pair for dispersion compensation. Not only is such a grating pair lossy and complicated to align, it needs careful adjustment for varying fiber length, output wavelength, and output pulse energy, and falls short of the requirement for most biomedical research labs and future clinical applications.
The present invention is directed to overcoming these and other deficiencies in the art.
The present invention relates to an apparatus for producing optical pulses of a desired wavelength. The apparatus includes an optical pulse source operable to generate input optical pulses at a first wavelength. The apparatus further includes a higher-order-mode (HOM) fiber module operable to receive the input optical pulses at the first wavelength, and thereafter to produce output optical pulses at the desired wavelength by soliton self-frequency shift (SSFS).
The present invention also relates to a method of producing optical pulses having a desired wavelength. This method includes generating input optical pulses using an optical pulse source, where the input optical pulses have a first wavelength and a first spatial mode. The input optical pulses are delivered into an HOM fiber module to alter the wavelength of the input optical pulses from the first wavelength to a desired wavelength by soliton self-frequency shift (SSFS) within the HOM fiber module, thereby producing output optical pulses having the desired wavelength.
The present invention is useful in providing optical pulses that a tunable over a wide wavelength range. The present invention can be used in any application that involves optical pulses. Examples of such uses include, without limitation, spectroscopy, endoscopy, and microscopy applications. Such uses can involve medical, diagnostic, and non-medical applications. In one embodiment, the present invention provides wavelength tunable, all-fiber, energetic femtosecond sources. In another embodiment, the present invention provides femtosecond sources based on a new class of optical fiber (i.e., an HOM fiber) that was recently demonstrated, where, for the first time, a large anomalous dispersion was achieved at wavelengths below 1300 nm in an all-silica fiber.
a) Soliton self-frequency shifted spectra corresponding to different input pulse energies into the HOM fiber module. (b) High resolution trace of the initial input spectrum over a 30-nm span. (c) High resolution trace over a 100-nm span of the shifted soliton for 1.63-nJ input into the HOM fiber. (d) Solution self-frequency shifted spectra calculated from simulation using a 280-fs Gaussian pulse input and at shifted soliton energies comparable to those in (c). (c) Measured second-order interferometric autocorrelation trace of the output soliton at 1.39-nJ pulse input into the HOM fiber, corresponding to a deconvolved pulse width of approximately 50 fs (FWHM). The tall spike in the experimental spectra (a) is entirely due to the imperfection of our commercial fiber source, where a CW-like spike was present at 1064 nm (b).
The present invention relates to an apparatus for producing optical pulses of a desired wavelength. The apparatus includes an optical pulse source operable to generate input optical pulses at a first wavelength. The apparatus further includes a higher-order-mode (IIOM) fiber module operable to receive the input optical pulses at the first wavelength, and thereafter to produce output optical pulses at the desired wavelength by soliton self-frequency shift (SSFS).
In one embodiment, the HOM fiber module includes an HOM fiber. Suitable HOM fibers can include, without limitation, a solid silica-based fiber. In another embodiment, the HOM fiber module includes an HOM fiber and at least one mode converter. The at least one mode converter can be connectedly disposed between the optical pulse source and the HOM fiber. The HOM fiber module can also include an HOM fiber, a mode converter connectedly disposed between the optical pulse source and the HOM fiber, and also a second mode converter terminally connected to the HOM fiber. Suitable mode converters that can be used in the present invention are well known in the art, and can include, for example, a long period grating (LPG).
Suitable optical pulse sources that can be used in the present invention can include, without limitation, mode-locked lasers and chirped pulse amplification (CPA) systems. More particularly, the mode-locked laser can be a mode-locked fiber laser, and the CPA system can be a fiber CPA system. The optical pulse source used in the present invention can include those that generate input optical pulses having various pulse energies. In one embodiment, the optical pulse source generates a pulse energy of at least 1.0 nanojoules (nJ). In another embodiment, the optical pulse source generates input optical pulses having a pulse energy of between about 1.0 nJ and about 100 nJ.
The optical pulse source can also be one that generates input optical pulses such that the first wavelength is a wavelength within the transparent region of a silica-based fiber. In one embodiment, the optical pulse source is one that generates a first wavelength below 1300 nanometers (nm). In another embodiment, the optical pulse source is one that generates a first wavelength between the range of about 300 nm and about 1300 nm.
The optical pulse source used in the present invention can also be one that generates input optical pulses having a subpicosecond pulse width.
Suitable HOM fiber modules that can be used in the present invention can include, without limitation, HOM fiber modules that produce output optical pulses having a pulse energy of at least 1.0 nJ. Suitable HOM fiber modules can also be those that produce output optical pulses such that the desired wavelength is a wavelength within the transparent region of a silica-based fiber. In one embodiment, the HOM fiber module produces an output optical pulse having a desired wavelength that is below 1300 nm. In another embodiment, the HOM fiber module produces an output optical pulse having a desired wavelength between the range of about 300 nm and about 1300 nm. The HOM fiber module can also be such that it produces output optical pulses having a subpicosecond pulse width.
The apparatus of the present invention can further include a power control system connectedly disposed between the optical pulse source and the HOM fiber module. The power control system for use in the present invention can be one that achieves subnanosecond power tuning of the first wavelength. Suitable power control systems can include, without limitation, a lithium niobate (LiNbO3) intensity modulator device.
The apparatus of the present invention can further include a single-mode fiber (SMF) connectedly disposed between the optical pulse source and the HOM fiber module.
The apparatus of the present invention can be used in a variety of applications where optical pulses of a desired wavelength are needed. For example, the apparatus can be effective in producing output optical pulses that can penetrate animal or plant tissue at a penetration depth of at least 0.1 millimeters (mm).
The apparatus of the present invention can further be such that the HOM fiber module is terminally associated with medical diagnostic tools such as an endoscope or an optical biopsy needle.
The apparatus of the present invention can further be functionally associated with a multiphoton microscope system.
The apparatus of the present invention can also further be functionally associated with a multiphoton imaging system.
The present invention also relates to a method of producing optical pulses having a desired wavelength. This method includes generating input optical pulses using an optical pulse source, where the input optical pulses have a first wavelength and a first spatial mode. The input optical pulses are delivered into an HOM fiber module to alter the wavelength of the input optical pulses from the first wavelength to a desired wavelength by soliton self-frequency shift (SSFS) within the HOM fiber module, thereby producing output optical pulses having the desired wavelength.
The method of the present invention can involve the use of the apparatus described herein as well as the various aspects and components of the apparatus (e.g., the optical pulse source and the HOM fiber module) described herein.
In one embodiment, the method can further include converting the first spatial mode of the input optical pulses into a second spatial mode prior delivering the input optical pulses into the HOM fiber so that the output optical pulses have the second spatial mode, where the first spatial mode and the second spatial mode are different modes. This method can further include reconverting the second spatial mode of the output optical pulses back to the first spatial mode.
In another embodiment, the method can further include tuning the first wavelength of the input optical pulses to an intermediate wavelength prior to delivering the input optical pulses into the HOM fiber. The tuning can include, without limitation, power tuning. Such power tuning can include varying the power of the input optical pulses so as to vary the desired wavelength. In one embodiment, the power tuning can include subnanosecond power tuning using a power control system connectedly disposed between the optical pulse source and the HOM fiber module. Suitable power control systems can include, without limitation, a lithium niobate intensity modulator device. In another embodiment, the tuning can be achieved by varying the length of the HOM fiber so as to vary the desired wavelength.
Described in more detail below is the concept of SSFS in optical fibers and more particularly in HOM fibers.
SSFS is a well-known and well-understood phenomenon. The concept of SSFS was first discovered ˜20 years ago in fiber optic communications, and most of the past experiments on SSFS relates to telecom. Optical soliton pulses generally experience a continuous downshift of their carrier frequencies when propagating in a fiber with anomalous dispersion. This so-called soliton self-frequency shift originates from the intra-pulse stimulated Raman scattering which transfers the short wavelength part of the pulse spectrum toward the long wavelength part [38] (SSFS sometimes is also called Raman soliton shift). Through the balancing of optical nonlinearity and fiber dispersion (i.e., soliton condition), the pulse maintains its temporal and spectral profiles as it shifts to the longer wavelengths. Although the physics of SSFS was well known for the last 20 years, its practical application was limited because the use of conventional fibers for generating wavelength-shifting solitons has major limitations. However, several new classes of optical fibers, such as photonic crystal fibers [39] (PCF, sometimes also known as microstructure fiber) and solid-core or air-core band gap fibers (BGF) [40], has generated enormous excitement in the last 5 years and greatly improved the feasibility of SSFS. Indeed, there are a number of experimental demonstrations of SSFS in PCF and BGF [41, 42, and 43]. However, none of the previous work can generate soliton energies that are of practical interest to biomedical research, i.e., solitons with pulse energies between 1 to 10 nJ and at wavelengths below 1300 nm. As we will elaborate below, the pulse energies produced in previous works are either one to two orders of magnitude too small or several orders of magnitude too large.
Because material nonlinearity for silica glass is positive at the relevant spectral range, the fundamental condition to form an optical soliton in silica fiber is anomalous dispersion. In addition, the existence of an optical soliton requires exact balance between fiber nonlinearity and dispersion. Thus, the energy of an optical soliton (Es) is determined by material nonlinearity and dispersion, and scales as [44]
Es∝λ3·D·Aeff/n2τ. (1),
where n2 is the nonlinear refractive index of the material, τ is the pulse width, D is the dispersion parameter, Aeff is the effective mode field area, and λ is the wavelength. Although standard single mode fibers (SMF) cannot achieve anomalous dispersion at λ<1280 nm, it was realized that the total dispersion (D) in a waveguide structure such as an optical fiber consists contributions from the material (Dm), the waveguide (Dw), and the bandgap (in the case of BGF). By appropriately engineering the contributions of the waveguide and/or the bandgap, it is possible to achieve anomalous dispersion (D>0) at virtually any wavelength, thus, enabling soliton and SSFS at wavelengths below 1280 nm. (It is worth noting that the dispersion parameter D is actually positive for anomalous dispersion.) Previously, there were two approaches to achieve anomalous dispersion, and therefore soliton propagation and SSFS, at λ<1280 nm:
(1) Small-core PCF can achieve anomalous dispersion for wavelengths down to ˜550 nm [45]. When the waveguide is tightly confining, with the air-silica boundary defining the confinement layer, the waveguide dispersion (Dw) is akin to that of microwave waveguides with perfectly reflecting walls. Hence, large positive waveguide dispersion may be realised by tightly-confined LP01), (fundamental) modes in PCFs. However, the associated trade-off is with Aeff, and designs that yield dispersion >+50 ps/nm/km in the wavelength ranges of 800 nm or 1030 nm typically have Aeff of 2-5 μm2. Because the soliton energy scales with the value of D*Aeff, a small Aeff will severely limit the pulse energies that can be obtained with PCFs. For example, in one experiment using a special PCF structure performed, a soliton pulse energy of ˜20 pJ was obtained at 800 nm [46], orders of magnitude smaller than practical for imaging. Indeed, most PCF structures are designed to demonstrate nonlinear optical effects at the lowest possible pulse energy.
(2) Air-guided BGFs potentially can offer anomalous dispersion at any wavelength [47], but the extremely low nonlinearities in these fibers (the nonlinearity of air is ˜one thousand times smaller than silica glass) make them impractical for a device that utilises a nonlinear interaction to achieve the frequency shift. In one demonstration, a MW (˜μJ pulse) optical amplifier is needed for observing SSFS in air-guiding fiber [43]. Not only is such a high power unnecessary for most biomedical applications, the cost and complexity of the high power amplifier also makes it completely impractical as a tool for biomedical research.
Although SSFS provides a convenient mechanism for wavelength tuning of a fixed wavelength fiber laser, previous works in SSFS were performed at soliton energies either too low or too high (by at least an order of magnitude) for practical use. Thus, it is essential to invent a new fiber structure, with just the right amount of optical nonlinearity and dispersion (i.e., D·Aeff/n2) in order to produce soliton pulses of practical utility for biomedical imaging.
An optical fiber generally propagates a number of spatial modes (electric field states). Because of modal dispersion and interference, however, only single mode fibers (i.e., fibers with only one propagating mode) are of interest for applications such as high speed data transmission and pulse delivery for imaging. It was realized, however, a multimode fiber can propagate only one mode if two conditions are met: 1. the input field is a pure single mode and (2) the couplings between various modes during propagation are small. In the case that the one propagating mode is not the fundamental mode, the fiber is called a HOM fiber. HOM fibers first attracted attention in optical communications nearly ten years ago. The main motivation was for dispersion compensation of high bit-rate optical communications. The advantage of HOM fibers is to provide another degree of freedom in the design space to achieve the desired dispersion characteristics. There were a number of devices invented using HOM fibers [48]. In fact, dispersion compensators based on HOM fibers have been commercially available for several years [49].
We realized that the design freedoms enabled by HOM fibers are exactly what is needed for achieving the desired soliton energy at wavelength below 1300 nm for biomedical imaging: (1) A higher order mode can achieve anomalous dispersion at wavelength below 1300 nm, a condition necessary for soliton and impossible to obtain in a conventional silica SMF. (2) A higher order mode typically has a much larger Aeff than that of PCF for achieving higher soliton energy. (3) The silica core of the HOM fiber retains just enough nonlinearity to make SSFS feasible at practical energy level. (4) The all silica HOM fiber retains the low loss properties (for both transmission and bending) of a conventional SMF, and allows easy termination and splicing. (5) A HOM fiber leverages standard silica fiber manufacturing platform, which has been perfected over the course of 30 years with enormous resources. Thus, an appropriately designed HOM fiber can provide the necessary characteristics desired for biomedical imaging, and can be manufactured immediately with high reliability.
The Examples set forth below are for illustrative purposes only and are not intended to limit, in any way, the scope of the present invention.
Soliton-self frequency shift of more than 12% of the optical frequency was demonstrated in a higher-order-mode (HOM) solid, silica-based fiber below 1300 nm. This new class of fiber shows great promise of supporting Raman-shifted solitons below 1300 nm in intermediate energy regimes of 1 to 10 nJ that cannot be reached by index-guided photonic crystal fibers or air-core photonic band-gap fibers. By changing the input pulse energy of 200 fs pulses from 1.36 nJ to 1.63 nJ, clean Raman-shifted solitons were observed between 1064 nm and 1200 nm with up to 57% power conversion efficiency and compressed output pulse widths less than 50 fs. Furthermore, due to the dispersion characteristics of the HOM fiber, red-shifted Cherenkov radiation in the normal dispersion regime for appropriately energetic input pulses were observed.
In this example, soliton self-frequency shift from 1064 nm to 1200 nm with up to 57% power efficiency in a higher-order-mode (HOM) fiber is demonstrated (Ramachandran et al., Opt. Lett. 31:2532 (2006), which is hereby incorporated by reference in its entirety). This new class of fiber shows great promise for generating Raman solitons in intermediate energy regimes of 1 to 10 nJ pulses that cannot be reached through the use of PCFs and PBGFs. The HOM fiber used in the experiments of this example was shown to exhibit large positive dispersion (˜60 ps/nm-km) below 1300 nm while still maintaining a relatively large effective area of 44 μm2 (Ramachandran et al., Opt. Lett. 31:2532 (2006), which is hereby incorporated by reference in its entirety), ten times that of index-guided PCFs for similar dispersion characteristics. Through soliton shaping and higher-order soliton compression within the HOM fiber, clean 49 fs pulses from 200 fs input pulses were generated. Due to the dispersion characteristics of the HOM fiber, red-shifted Cherenkov radiation in the normal dispersion regime for appropriately energetic input pulses was also observed.
The experimental setup is shown in
57% power conversion from the input pulse spectrum to the red-shifted soliton was measured for the case of 1.39 nJ input pulses to achieve ˜0.8 nJ output soliton pulses,
Finally, the appearance of Cherenkov radiation centered about 1350 nm for 1.45 nJ and 1.63 nJ input pulse energies,
Though not demonstrated in this example, light can be easily coupled back into the fundamental mode using another LPG at the output end. Previous work showed that by using a dispersion-matching design, ultra-large bandwidths can be supported by a LPG (Ramachandran, S., Journal of Lightwave Technology 23:3426 (2005), which is hereby incorporated by reference in its entirety). Recently, conversion efficiency of 90% over a bandwidth of 200 nm was obtained for a similar fiber structure (Ramachandran et al., Opt. Lett. 31:1797 (2006), which is hereby incorporated by reference in its entirety). Such a LPG will ensure the output pulse is always converted back to a Gaussian profile, within the tuning range. An important consideration for the output LPG is its length. Since the energetic output pulses are solitons for a specific combination of dispersion and Aeff of the L02 mode, nonlinear distortions may occur when the energetic pulse goes to the (smaller Aeff) fundamental LP0l mode at the output. However, the length over which the signal travels in the LP01 mode, and hence the distortion it accumulates, can be minimized because the high-index core of the HOM fibers enable LPG lengths of <5 mm. This implies that light can reside in the LP01 mode for <2.5 mm, hence largely avoiding nonlinear distortions. Note that the requirement for short LPGs actually complements the need for broad-bandwidth operation, since the conversion bandwidth is typically inversely proportional to the grating length (Ramachandran, S., Journal of Lightwave Technology 23:3426 (2005), which is hereby incorporated by reference in its entirety).
Both the wavelength shift and pulse energy can be significantly increased beyond what has been demonstrated through engineering of the fiber module. For example, simple dimensional scaling of the index profile can be used to shift the dispersion curve of the LP02 mode. Numerical modeling shows that an output soliton energy of approximately 2 nJ can be realized if the dispersion curve is shifted ˜100 nm to the longer wavelength side. Additionally, pulse energy can be scaled by increasing D·Aeff. Aside from increasing the magnitude of dispersion through manipulation of the index profile and dimensions of the fiber, the effective area can be significantly enhanced by coupling into even higher-order modes. An effective area of ˜2000 μm2 (more than 40 times this HOM fiber) was recently achieved by coupling to the LP07 mode (Ramachandran et al., Opt. Lett. 31:1797 (2006), which is hereby incorporated by reference in its entirety).
In summary, SSFS between 1064 nm and 1200 nm has been demonstrated in a higher-order-mode, solid silica-based fiber. 49 fs Raman-shifted solitons were obtainable at 0.8 nJ with up to 57% power conversion efficiency. Due to the dispersion characteristics of the HOM fiber, Cherenkov radiation was also observed for appropriately energetic input pulses. It is believed that HOM fiber should provide an ideal platform for achieving soliton energies from 1 to 10 nJ for SSFS at wavelengths below 1300 nm, filling the pulse energy gap between index-guided PCFs and air-core PBGFs. This intermediate pulse energy regime which could not be reached previously for SSFS could prove instrumental in the realization of tunable, compact, all-fiber, femtosecond sources for a wide range of practical applications.
To emphasize the significance of the proposed femtosecond sources, we compare our proposed sources with the existing mode-locked Ti:S laser, Ti:S pumped OPO and femtosecond fiber sources.
Table 1 compares some of the key characteristics of the existing and our proposed femtosecond sources. The proposed systems would be much less expensive than the currently used state-of-the-art single box Ti:S lasers (Spectra-Physics Mai Tai and the Coherent Chameleon), probably ⅓ to ¼ the cost. The telecom manufacturing platform employed in the proposed fiber sources provides an inherent opportunity for further cost reduction by volume scaling. In addition, there are the practical advantages offered by the all-fiber configuration, such as a compact foot print and a robust operation. However, what truly sets the proposed femtosecond sources apart from other existing fiber sources is performance. Table 1 shows that the proposed all-fiber sources will achieve comparable or better performances in terms of output pulse energy, pulse width, and wavelength tuning range when compared to bulk solid-state mode-locked lasers. We note that the output characteristics of the proposed sources listed above are delivered through an optical fiber. The elimination of the free-space optics makes the proposed fiber sources more efficient in delivering power to an imaging setup. Thus, even at a slightly lower output power, the imaging capability of the proposed sources will likely be close to that of the free-space Ti:S laser. It is worth emphasizing that significant research and development efforts have been devoted to femtosecond fiber sources in the last 15 years or so. However, femtosecond fiber lasers have so far failed to have a major impact in biomedical research. We believe the reason for the low penetration of fiber femtosecond sources in the biomedical field is precisely due to various performance handicaps (such as pulse energy, wavelength tunability, pulse width, fiber delivery, etc.) that kept existing fiber sources from being the “complete package.” It has nothing to do with the lack of demand or interest from biomedical researchers. Leveraging major technological advances in the fiber optic communication field and recent fiber laser developments, we believe we have finally arrived at the stage where all-fiber femtosecond sources can be realized without sacrificing performance. The successful completion of this research program will make femtosecond sources truly widely accessible to biologists and medical researchers and practitioner.
This program explores a new route for generating energetic femtosecond pulses that are continuously tunable across a wide wavelength range, where, in contrast to previous approaches, ultrafast pulses are wavelength shifted in a novel HOM fiber module by SSFS. By eliminating the constraint of a broad gain medium to cover the entire tuning range, our approach allows rapid, electronically controlled wavelength tuning of energetic pulses in an all-fiber configuration.
A feature of the proposed research is to harvest the recent development in femtosecond fiber sources and the latest breakthrough in fiber optic communication industry. During the course of our research and development in both academia and industry over the last 5 years, we have accumulated significant amount of preliminary data to support our approach. Specifically, we present below our studies on femtosecond fiber sources, HOM fibers, and SSFS, three key components of the proposed femtosecond source.
The performance of fixed wavelength femtosecond fiber sources at 1030 and 1550 nm have been improved significantly in the last several years. In fact, cost effective (˜$50 k) commercial fiber sources that are capable of delivering ˜10-nJ pulse energies at 40 MHz repetition rate or higher already exist. These sources are mostly based on fiber chirped pulse amplification (CPA), where a low pulse energy oscillator serves as a seed source for the subsequent optical fiber amplifier. Examples of such sources are offered by PolarOnyx Inc. and several other companies.
One of the draw backs of the commercial fiber sources is that they employ CPA technique to achieve the pulse energies required for our application. The combination of oscillator and amplifier inevitably increases the cost of the system. Obviously, a lower cost approach will be to build a fiber oscillator that can achieve the pulse energy directly. A series of advances in femtosecond fiber lasers at wavelengths around 1 μm, based on ytterbium-doped fiber (Yb:fiber) have been reported. These include some of the best performances reported for femtosecond fiber lasers [51, 52], such as the highest pulse energy (14 nJ), highest peak power (100 kW), highest average power (300 mW) and highest efficiency (45%). These are the first fiber lasers with pulse energy and peak power comparable to those of solid-state lasers. These lasers are diode-pumped through Fiber spliced to the gain fiber, and are therefore already stable and reliable laboratory instruments. Uninterrupted operation for weeks at a time is routine, except when the performance is pushed to the extremes of pulse energy or pulse duration.
The science that underlies the increases in pulse energy and peak power listed above is the demonstration of pulse propagation without wave-breaking [51]. The theoretical and experimental demonstration of “self-similar” evolution of short pulses in a laser [53] is a major breakthrough. This is a completely new way to operate a mode-locked laser. The laser supports frequency-swept (“chirped”) pulses that avoid wave-breaking despite having much higher energies than prior fiber lasers. The pulses can be dechirped to their Fourier-transform limit (
The high-energy lasers described above are experimental systems. They employ some bulk optical components in the cavity, such as diffraction gratings for anomalous group-velocity dispersion. These components naturally detract from the benefits of the fiber medium, and integrated versions of these devices will be needed for most applications. Virtually all of the components of the lasers are now available in fiber format, and several advances toward the ultimate goal of all-fiber and environmentally-stable devices were made in the past few years. The first step is to replace the diffraction gratings with a fiber device. Microstructure fibers, which have become commercially-available in the past couple years, offer new combinations of dispersion and nonlinearity. The demonstration of dispersion control with a PCF [55] was the first such application of microstructure fibers. The resulting laser is limited to low pulse energies by the small Aeff of the PCF. The extension of this approach to air-core PBF [47] is quite promising, as it will enable all-fiber lasers capable of wave-breaking-free operation.
Lasers with segments of ordinary fiber are susceptible to environmental perturbations such as strain or temperature changes. For ultimate stability, it will be desirable to construct lasers with polarization-maintaining fiber. We exploited the fact that photonic-bandgap fiber is effectively a polarization-maintaining fiber owing to the high index contrast, to build the first environmentally-stable laser at 1 μm wavelength [56]. This laser operates stably when the fiber is moved, twisted or heated. All the components of the laser (which was a testbed for new concepts) now exist in fiber format. It is therefore now possible to design lasers in which the light never leaves the fiber, and which are impervious to environmental perturbations.
Our development in robust femtosecond fiber lasers has already attracted significant commercial interests. PolarOnyx, Inc. (Sunnyvale, Calif.), and Clark/MXR, Inc. (Dexter, Mich.) have introduced products based on the lasers described above (see
Recently, an exciting new fiber type was demonstrated that yields strong anomalous dispersion in the 1-μm wavelength range, from an all-solid silica fiber structure where the guidance mechanism is conventional index-guiding [57]. This represents a major breakthrough in fiber design because it was previously considered impossible to obtain anomalous dispersion at wavelength shorter than 1300 nm in such an all-silica fiber. The key to the design was the ability to achieve strong positive (anomalous) waveguide dispersion (Dw) for the LP02 mode of a specially designed HOM fiber. We demonstrated a fiber that had +60 ps/nm-km dispersion for the LP02 mode in the 1060-nm wavelength range. Combined with in-fiber gratings, this enabled constructing an anomalous dispersion element with low loss (˜1%), and an Aeff (44 μm2) that is 10 times larger than PCF. Significantly, the guidance mechanism was index-guiding, as in conventional fibers. Hence, it retains the desirable properties of conventional fibers, such as low-loss, bend-resistant, and lengthwise invariant (in loss, dispersion, etc), making them attractive for a variety of applications.
In contrast, the LP02 mode may be designed to have the mode evolution shown in
There are a number of theoretical and experimental works on SSFS in the past [60-64], including some targeting biomedical applications [65, 66]. Reports have demonstrated SSFS in a number of fiber structures within the last 5 years. Previously, a novel tapered air-silica microstructure fiber was fabricated [41, 67] and demonstrated SSFS within the telecom window of 1.3 μm to 1.65 μm in a 10-cm long tapered microstructure fiber (inset in
Our recent breakthrough in the HOM fiber provides an exciting new opportunity for SSFS at the practical pulse energies of 1 to 10 nJ and at wavelength below 1300 nm. We have experimentally investigated the behavior of SSFS at Cornell using the HOM fiber module provided by OFS.
1. A continuous wavelength shift of ˜130 nm (1060 to 1190) was achieved.
2. An output pulse energy of 0.84 nJ was obtained at 1.39-nJ input pulse.
3. A high quality output pulse with ˜50-fs FWHM and a high conversion efficiency (i.e., the amount of optical power that is transferred to the wavelength shifted soliton) of ˜60% were obtained despite of the low quality input pulse.
4. Remarkable agreement between experiments and numerical modeling were achieved despite of the non-ideal input, demonstrating the robustness of soliton pulse shaping.
We note that at the highest input pulse energy, a new spectral peak appeared at much longer wavelength (˜1350 nm). This is the well-known resonance Cerenkov radiation of the soliton due to the negative dispersion slope [68], which is also predicted by our simulation (
Our initial success of SSFS in a HOM fiber module, and our proven capability to numerically predict the behavior of SSFS in a HOM fiber give us a high degree of confidence to achieve the stated goals. Through extensive numerical simulations, we have already determined the required dispersion (
In addition to the power tuning of the output wavelength, an alternative method for wavelength tuning is simply using different fiber length.
Perhaps the most promising and successful area in biomedical imaging that showcases the unique advantage of multiphoton excitation is imaging deep into scattering tissues. One of the promising approaches for imaging deep into scattering biological tissue is using longer excitation wavelength. It is well known that the scattering mean free path is proportional to the fourth power of the excitation wavelength in the Rayleigh region, where the size of the scatterer (α) is much smaller than the wavelength, i.e., 2πα/λ<0.1. When the size of the scatterer becomes comparable to the wavelength, i.e., in the Mie scattering region, the scattering mean free path (MFP) has a weaker dependence on the wavelength. Nonetheless, the MFP increases with increasing excitation wavelength. Although there is little data for tissue scattering beyond 1.1 μm, the available data at shorter wavelengths clearly indicates the general trend that the scattering MFP increases as one uses longer excitation wavelength [69]. In fact, the “diagnostic and therapeutic window,” which is in between the absorption regions of the intrinsic molecules and water, extends all the way to ˜1280 nm (see
There are a few experimental demonstrations of imaging at longer wavelengths by several groups [12, 70]. We have also carried out detailed studies of multiphoton excitation of fluorophores within the spectral windows of 1150 to 1300 nm, and have found useful multiphoton cross sections (10 to 100 GM, comparable to fluorescein at shorter wavelength [71]) exist for a number of long wavelength dyes (
Our overall approach to wavelength-tunable sources is to develop fiber sources of 10- to 25-nJ and ˜300-fs pulses, which will propagate in HOM fiber modules as Raman solitons to produce the desired outputs. Starting with pulses at 775 nm (1030 nm), pulses tunable from 775 to 1000 nm (1030 to 1280 nm) will be generated. The source development that we propose is enabled by the coincident advances in short-pulse fiber lasers and propagation of higher-order modes, along with the commercial development of semiconductor structures for stabilizing short-pulse lasers (to be described below). The availability of excellent fibers and continued improvement in the performance and cost of high-power laser diodes provide the technical infrastructure needed to support the development of short-pulse fiber devices.
Aim 1: Single Wavelength all-Fiber Femtosecond Sources.
We will develop single wavelength all-fiber femtosecond sources at 1030 nm and 775 nm with pulse energies at 10 and 25 nJ at repetition rates of 40 to 100 MHz.
Our first step is to modify and optimize commercially available femtosecond fiber sources to achieve ˜10-nJ pulses, which will be sufficient to achieve our first goal of 1- to 2-nJ output pulses. Although we are fully capable of building such sources ourselves, we aim to jump start the program by fully leveraging existing commercial technologies. The main task during this stage is to make the commercial sources truly all-fiber. We realized that one of the main drawbacks of existing commercial fiber sources is that they are not all-fiber. For example, the PolarOnyx system (
Energetic femtosecond fiber sources (either from an oscillator or a CPA system) have typically chirped output to avoid optical nonlinearity, and therefore, external dispersion compensation is required to recover the femtosecond pulses. The main reason for the required free-space grating compressor in the current fiber source is the lack of low nonlinearity anomalous dispersion fiber, i.e., fibers with large Aeff and large positive D value. Although airguided BGF can be used for dispersion compensation, there were a number of practical issues such as termination, fusion splice, birefringence, loss, etc. On the other hand, the proposed HOM fiber can easily perform dispersion compensation in addition to SSFS, by simply adding HOM fiber length in the HOM fiber module. For example, with a typical chirp of 0.24 ps2/nm from a fiber source (the amount of chirp caused by ˜12 m of SMF at 1030 nm), our simulation shows that a HOM fiber length of ˜6 m will produce the output nearly identical to that shown in
The second step, which involves our own laser and source development, aims to improve the pulse energy to ˜25 nJ in an all fiber design. Such pulse energies are necessary for achieving a final tunable output of 5- to 10-nJ pulses. There are two approaches to achieve our aim.
The first approach closely follows the strategy of the existing commercial devices using CPA. In a realistic fiber amplifier capable of the needed performance, a pulse is taken from an oscillator by splicing on an output fiber (tens of meters in length) where the pulse is highly stretched temporally. The stretched pulse is then amplified to high pulse energy by a fiber amplifier. Nonlinear effects that could distort the pulse are avoided because the pulse is stretched, which reduces the peak power. The output from the amplifier will be an amplified version of the same chirped pulse. The pulse is contained in ordinary single-mode fiber throughout the device. The above described CPA scheme has enabled significantly improved pulse energy in fiber amplifiers. Even μJ pulse energies can be obtained (although at much lower repetition rate). For our proposed sources, we will amplify pulses to 25 nJ at 1030. We aim to amplify to 50 nJ at 1550 nm in order to obtain ˜25-nJ pulses at 775 nm. Commercial fiber amplifier modules already exist to delivery the necessary power for our applications. In addition, methods for overcoming fiber nonlinearity in a fiber CPA system have been demonstrated [75, 76]. Thus, we do not anticipate any difficulty in achieving these design goals.
The combination of a laser and an amplifier in our first approach allows both to be designed easily, and is certain to meet or exceed our design specifications. Indeed, it is highly likely that commercial femtosecond fiber sources based on the CPA technique can deliver the necessary pulse energy (25 to 50 nJ) and power (1 to 2 watts) within the grant period. Thus, there is a possibility that we can continue leveraging commercial femtosecond fiber sources. On the other hand, the addition of an amplifier adds cost and complexity to the source (at least one more pump laser and driver will be required), and always adds noise to the output. Ultimately, it will be desirable to reach the needed pulse energies directly from oscillators. Thus, as an alternative and lower cost approach, we will pursue the development of high-energy oscillators in parallel with the construction of low-energy oscillators that are amplified to the required energies.
The essential physical processes in a femtosecond laser are nonlinear phase accumulation, group-velocity dispersion, and amplitude modulation produced by a saturable absorber. A real or effective saturable absorber preferentially transmits higher power, so it promotes the formation of a pulse from noise, and sharpens the pulse. Once the pulse reaches the picosecond range, group-velocity dispersion and nonlinearity determine the pulse shape. In the steady state, the saturable absorber thus plays a lesser role, stabilizing the pulse formed by dispersion and nonlinearity. It is known that the pulse energy is always limited by excessive nonlinearity. This limitation is manifested in one of two ways:
(1) A high-energy pulse accumulates a nonlinear phase shift that causes the pulse to break into two (or more) pulses. This is referred to as “wave-breaking.”
(2) To date, the best saturable absorber for fiber lasers is nonlinear polarization evolution (NPE), which produces fast and strong amplitude modulation based on polarization rotation. It was employed in the Yb fiber lasers described in our preliminary results. A disadvantage of NPE is that the transmittance is roughly a sinusoidal function of pulse energy; the transmittance reaches a maximum and then decreases with increasing energy. Once the NPE process is driven beyond that maximum transmittance, pulses are suppressed because lower powers experience lower loss and are thus favored in the laser. This situation is referred to as “over-driving” the NPE.
Thus, eliminating “wavebreaking” and “over-driving” are essential in order to achieve high pulse energy from a fiber laser. We have shown that the first limitation, which is the more fundamental of the two, can be avoided [51, 53] using self-similar pulse evolution. We have calculated the energy of stable self-similar pulses and the result is plotted in
Semiconductor saturable absorbers (SSA's) are based on saturation of an optical transition, and in contrast to NPE (which is based on interference) they cannot be overdriven. Therefore, it should be possible to obtain much higher pulse energies in fiber lasers if NPE is replaced by a SSA. Historically, this was not feasible, because semiconductor structures capable of producing the large modulation depth (>10%) needed in a fiber laser did not exist. In addition, a practical impediment in the past was the lack of a commercial source of such structures—painstaking research was required to develop new ones. However, significant progress in the modulation depth has been made in the last several years and there is now a commercial company that sells SSA's. BATOP GmbH (Weimar, Germany) has emerged as a reliable source of SSA's, with a variety of designs at reasonable prices (<$1 k/piece). In particular, structures with 80% modulation depth are available as standard designs. It will be reasonably straightforward to incorporate these structures in our lasers in place of NPE. The main work will be optimizing the design of the structure for the target performance levels.
A second major advantage of SSA's is that they are compatible with integrated designs. The development of saturable absorbers that provide fast and deep modulation will significantly facilitate the design of all-fiber and environmentally-stable lasers. In principle, a femtosecond laser could be constructed of segments of polarization-maintaining fiber that provide gain and anomalous dispersion, and the saturable absorber. Fiber-pigtailed versions of SSA's are already commercially available. Such a laser would be as simple as possible, with no adjustments other than the pump power. We will design, construct and characterize high-power fiber lasers based on SSA's. Although the incorporation of SSA with large modulation depth in a mode-locked fiber laser is relatively new and there may be a number of practical issues to be addressed in this work, the fundamental basis of the approach is established theoretically, and initial experiments in our lab with structures from BATOP confirm that they perform as advertised. The promise of 25- to 50-nJ pulses directly from a robust and cost effective fiber oscillator is highly significant. Thus, we will include this development effort as a more exploratory component of this research program, complementing our reliable (may even be commercially available), but inherently more expensive, approach of a fiber CPA system.
We will design and develop novel HOM fiber modules for SSFS at input wavelengths of 1030 nm and 775 nm. We will start by modifying the existing HOM fiber design, and fabricate new fibers with the goal of achieving 1- to 2-nJ output pulse energy. We will then extend the design space to create new fibers that is capable of delivering 5- to 10-nJ output pulses. OFS Laboratories has powerful, proprietary design tools to design highly complex fibers—indeed, its market leadership in dispersion compensating fibers was enabled by its ability to provide robust solutions for managing dispersion of the multitude of transmission fibers used today. We realized that the design and fabrication of the HOM fiber module is the key enabling component for achieving our aims. We anticipate that several iterations will probably be needed in the design and fabrication of the device before we can achieve the optimum performance. We have therefore set aside sufficient budget to cover for the design and development cost for the HOM fiber modules. We note, however, the manufacturing process of the HOM fiber is entirely compatible with commercial silica fibers, making it an intrinsically low-cost approach. Thus, a low-cost device with telecom-grade reliability is possible.
The physics of SSFS, as also seen from our preliminary results, dictates that the wavelength tuning range is limited by the dispersion-zero crossings of the curves shown in
Dimensional scaling of the preform can also be used to shift the waveguide dispersion Dw. This is known for optical waveguides as complimentary scaling, which states that wavelength and dimension play a complimentary role in the wave equation, and hence are interchangeable. However, note that this is true only for the waveguide component of dispersion Dw. Changes in material dispersion entail that the total attained dispersion (D) is not wavelength scalable. In other words, to move the dispersion curve that provides satisfactory operation in the 1030-nm wavelength range to the ˜800-nm spectral range, we would need Dw high enough to counteract the strong negative trend for Dm as wavelength decreases. Hence, achieving similar properties at lower wavelengths would need both the use of dimensional scaling and the dispersion-increasing recipes described above.
Our preliminary experimental results and numerical simulations showed that HOM fiber modules for delivering 1- to 2-nJ output pulses can certainly be made within the first 18 months of the grant period. Although these pulse energies are already sufficient for some biomedical applications, our ultimate aim is to produce fiber sources that are capable of delivering 5- to 10-nJ pulse, making them credible replacements of the bulk solid state lasers.
To achieve 5- to 10-nJ output pulse energies, the fiber design will have to be more aggressive than the existing design-class. Preliminary studies of fiber design show that D*Aeff values of 5 to 10 times the existing HOM module are achievable (
1. Split the tuning range into two segments, and perform sequential shifting with two different HOM fiber modules.
2. Increase ΔNcore and/or ΔNring—in general, increasing these values will lead to larger tuning ranges.
3. Operate in even higher order modes—in general, D*Aeff monotonically increases with mode order. An even higher order mode leads to a much higher D*Aeff value.
In approach 1, two different HOM fibers will be fabricated. The first HOM fiber is optimized for the first half of the tuning range only. The second HOM fiber module consists of the first HOM fiber fusion-spliced to another HOM fiber that is optimized for the second half of the tuning range. Thus, the longest wavelength output from the first HOM fiber will be used as the input to the second HOM fiber to achieve tuning in the second half of the wavelength range. To minimize perturbation to the soliton, the D-values of the two HOM fibers at the transition wavelength should approximately be the same, i.e., the transition wavelength should be located at the cross-over region of the two dispersion curves.
Approaches 2 and 3 are both lower cost alternatives that can achieve the required pulse energy in one HOM fiber module. The associated drawback for approaches 2 and 3 is that the HOM fiber would guide many higher order modes, which may make it susceptible to mode coupling. Conventional wisdom states that fiber should be strictly single-moded to avoid modal interference problems. However, it had been demonstrated in a variety of applications, that specially designed HOM fibers are extremely robust to mode coupling, especially when the HOM is excited with the very high efficiencies that in-fiber gratings afford. As seen in our preliminary results with the existing HOM fiber, measured conversion efficiencies of up to 99.9% are regularly achieved with this technology. Hence, the key to achieving operation with negligible modal-interference is (a) utilising the extreme efficiencies of LPGs to excite only the desired HOM with purities exceeding 20 dB (99%), and (b) designing the fiber such that effective index spacing between the desired mode and other parasitic/unwanted modes is large enough (the exact value depends on the application—km length propagation usually requires modal index separations of ˜10−3, while an order-or-magnitude decrease in this value can be tolerated when propagating over only 10 s to 100 s of meters). For example, we have achieved mode coupling levels ˜0.1% in an LP07 mode in a fiber that guided at least 49 other modes. This mode was found to be robust over lengths as long as 20 m [37], which is well beyond the length required (<10 m) for our applications here.
We further note that femtosecond pulses are generally quite tolerant to mode interferences due to the short coherence length. Modal dispersion will generally separate the soliton pulse and the other modes in time so that no interference can occur. Such a phenomenon has been observed in femtosecond pulse propagation in a large mode area fiber. We further note that a small amount of residue power in the other modes is typically not a concern for multiphoton excitation because of the quadratic (or even higher order) power dependence of the excitation process. Thus, it is entirely feasible to design a HOM fiber module based on even higher order modes such as the LP07 mode that we have demonstrated in the past. We believe that approaches 2 and 3 both have a high probability for success.
Our design process will also consider several practical issues, such as deviations of fabricated profiles from the ideal design, sensitivity to various index and dimensional perturbations etc. This latter aspect is an important highlight of the design space we propose—since the HOM fiber is index guided, as opposed to band-gap guided, the dispersive effect is not strictly resonant in nature, and is much less sensitive to perturbations of the profile.
The key to achieving the desired properties is a mode that can transition (as a function of wavelength) through well-defined, sharp, index steps in the profile. Therefore, the fabrication process must be capable of producing both large index steps as well as steep index gradients (See
Once a preform is fabricated, standard fiber-draw processes will be used to obtain the fiber. The flexibility of the fiber draw process allows for drawing to a variety of non-standard fiber diameters—this afford dimensional scaling of the index profile, which in turn will allow for precisely tuning the zero-dispersion wavelengths.
For device operation, a mode converter is needed, which will convert the incoming Gaussian-shaped, LP01 mode into the desired LP02 mode. We achieve this with in-fiber LPGs. LPGs are permanently induced in fibers by lithographically transferring a grating pattern from an amplitude mask to the fiber using a UV laser [78]. For efficient grating formation, the fiber is saturated with deuterium, which acts as a catalyst for the process which results in UV-induced index changes in Germanosilicate glasses. LPGs offer coupling between co-propagating modes of a fiber and have found a variety of applications as spectral shaping elements and mode-conversion devices. But LPGs are traditionally narrow-band (as expected of any interferometric device), and while they offer strong (>99%) mode coupling, the spectral width of such coupling was typically limited to a range of 0.5 to 2 nm, too narrow for a femtosecond pulse. To overcome the spectral limitation, reports have shown that the LPG bandwidth can be extended to >60 nm [79] (˜500 nm, in some cases [80]) if the fiber waveguide were engineered to yield two modes with identical group velocities. An example of a pair of broadband mode-converter gratings employed with positive dispersion HOM fibers was shown in FIG. 9—note that the large (51-nm) bandwidth was uniquely enabled by the dispersive design of the fiber which enabled matching the group velocities of the two coupled modes.
All the HOM fiber modules fabricated in this project will be similar our existing HOM fiber modules shown in
At the output, a second LPG will be used to convert the beam back to a Gaussian output. We will use the dispersion-matching designs that can yield ultra-large bandwidths. This will ensure the output pulse is always converted back to a Gaussian profile, within the tuning range of ˜250 nm. An important consideration for the output LPG is its length—since the energetic output pulses are solitons for the specific combination of dispersion and Aeff of the LP02 mode, nonlinear distortions may occur when the signal goes to the (smaller Aeff) fundamental LP01 mode at the output. However, the length over which the signal travels in the LP01 mode, and hence the distortion it accumulates, can be minimized—the high-index core of these HOM fibers enable LPG lengths of <5 mm, which implies that light resides in the LP01 mode for <2.5 mm, hence largely avoiding nonlinear distortions. Note that the requirement for short LPGs actually complements the need for broad bandwidth operation, since the conversion bandwidth is typically inversely proportional to the grating length.
We aim to demonstrate two all-fiber femtosecond sources with wavelength tuning ranges of (1) 775 nm to 1000 nm and (2) 1030 nm to 1280 nm. The output pulse energies will be first at 1 to 2 and then at 5 to 10 nJ. We will combine the femtosecond sources and the HOM fiber modules developed in Aims 1 and 2 into an all-fiber system. The fully integrated source is schematically shown in
The intrinsic chirp from the fiber source (either a mode-locked fiber lasers or a CPA system), which was a major limitation in previous fiber systems, provides several key advantages for our system. First, it allows ˜10 m in the total length of the output fiber. Second, the highly chirped pulse makes the length of the single mode fiber pigtail inconsequential, eliminating the practical difficulties in cleaving and splicing. Finally, the longer single mode fiber pigtail can also accommodate additional fiber devices such as a variable fiber attenuator and/or a fiber optic switch.
Second harmonic generation (SHG) will be employed to generate femtosecond pulses at 775 nm. It was previously known that SHG with a linearly chirped fundamental pulse will result in a linearly chirped SH pulse [81], which can be subsequently compressed using linear dispersion. Interestingly, the final chirp-free SH pulse width is independent of whether the compression is carried out before or after the SHG [81]. The conversion efficiency, however, is obviously higher if the chirped fundamental pulse is compressed before SHG. Because the designed pulse energy at the fundamental wavelength is high (at 10 to 50 nJ/pulse), the conversion efficiency for SHG with the proposed excitation source will be limited mostly by the depletion of the fundamental power, not by the available pulse peak intensity. Thus, SHG will be highly efficient even with a chirped fundamental pulse with durations of the order of several picoseconds if efficient doubling crystals are employed [33, 82]. For example, with a periodically poled LiNbO3 (PPLN), a conversion efficiency of 85%/nJ was demonstrated with 230 fs pulses at 1550 nm [83]. Single-pass conversion efficiencies (energy efficiency) as much as 83% [84] and 99% [85] are demonstrated for bulk and waveguide PPLN devices, respectively. Thus, SHG with a chirped fundamental pulse can be used with the proposed femtosecond pulse source without the reduction in conversion efficiency, and, as discussed in the previous paragraph, has significant advantage in the subsequent fiber optic delivery process. In addition, chirped SHG also eliminates the possibility of damaging to the doubling crystal due to the high peak power of a femtosecond source. Photorefractive effects of the PPLN device is a concern at high average power (>500 mW), but such effects typically only occur at wavelength below 700 nm, and can be mitigated to a large extend by increasing the temperature of the crystal and/or by doping the crystal with Magnesium. To be conservative, we are targeting a power conversion efficiency of ˜50% on a routine basis.
We will design systems using two different tuning mechanisms: 1. power tuning, and 2. length tuning. As shown in the preliminary results, both tuning mechanisms offer similar tuning range (
The fiber-length tuning described above is obviously similar to the sequential tuning described in Aim 2 (approach 1) to achieve high pulse energy. Both require multiple HOM fiber modules. In length tuning, however, the same HOM fiber of different lengths are used; while in sequential tuning, two or more different HOM fibers are required.
Both power tuning and segmented length tuning require a mechanism to control the incident power. SSFS is a nonlinear optical effect and effectively happens instantaneously (<1 ps). Thus, the rate of the wavelength tuning of the proposed fiber source can be ultrafast, and is completely determined by the rate of power change. There are two approaches to adjust the power into the HOM fiber module. Mechanical in-line fiber attenuators can achieve a tuning speed of ˜10 Hz, several orders of magnitude faster than any existing laser systems. Because only a small range of power adjustment is necessary for achieving the entire range of wavelength tuning (less than a factor of 4 for power tuning), variable fiber attenuators that based on microbending can easily provide the speed and modulation depth required. Such a variable attenuator can be calibrated so that rapid, electronically controlled wavelength tuning can be achieved. We note that compact, electronically controlled variable fiber attenuators are widely available commercially. Most commercial attenuators can provide modulation depth of ˜1000. Thus, we do not anticipate any difficulty implementing the power control mechanism. An alternative approach will be to use a fiber coupled electro-optic modulator (EOM). Although such an approach will be more expensive (˜$2 k), it can easily provide nanosecond (i.e., pulse-to-pulse) wavelength switching speed. In addition, such a device also provides the capability for fast (ns) laser intensity control. To overcome the insertion loss of the electro-optic modulator, it can be placed before the fiber amplifier in a CPA system. We also note that these EOMs are routinely used in telecommunications and are highly robust (telecom certified) and compact (the size of half a candy bar). Our proposed source can be readily configured to provide this high speed tuning capability.
We will perform detailed system testing and characterization, providing feedbacks for iteration and optimization of our development efforts in Aims 1 and 2. In particular, we will assess the wavelength and power stability of the system. We are well aware the fact that SSFS is a nonlinear optical effect; and nonlinear optical effects are generally sensitive to fluctuations in input power, pulse width, and pulse spectrum. We have taken this stability issue into our design considerations. First, we start with an all-fiber, single wavelength femtosecond source. One of the salient features of an all-fiber design is its stability. It is well known that a fiber laser is more stable than a bulk solid state laser. Second, our fiber sources are specifically designed for biomedical imaging applications. Because of the broad output pulse spectrum (10 to 20 nm) and the broad excitation peaks of fluorescent molecules (tens of nm), a few nm of wavelength shift is generally inconsequential. This is in sharp contrast to applications such as precision frequency metrology, where even a small fraction of an Angstrom spectral shift cannot be tolerated. Finally, the soliton pulse shaping process is robust against fluctuations in the input, which is one of the main reasons that solitons were used in long haul communication systems. Our preliminary results in
Polarization control is another issue of practical concern. For applications that demand a linear input polarization, polarization maintaining (PM) fibers can be used throughout the system. Because the HOM fiber is fabricated within the conventional silica fiber platform, PM HOM fibers can be made using the same method designed for conventional PM fibers (such as adding stress rods to form a Panda fiber). For applications that demand adjustable input polarization, non-PM HOM fibers can be used and a simple in-line fiber polarization controller can be used to adjust the output polarization state, eliminating the conventional free-space wave plate and/or polarizer.
There are several methods to remove the residue input light at the output of the HOM fiber module. Perhaps the simplest approach is to directly deposit a dichroic coating (long wavelength pass) on the output face of the fiber. Such coatings were often done for fiber lasers with linear cavities and the deposition techniques were similar to that on a conventional glass substrate. After all, a silica fiber is a piece of glass with a small diameter.
We will demonstrate the significance of our new femtosecond laser sources for biomedical applications of multiphoton microscopy, spectroscopy and endoscopy.
Our first stage demonstration involves “routine” multiphoton imaging and spectroscopy. We will compare the capability of the proposed tunable fiber source with our existing Ti:S lasers. We will verify the stability of the sources in imaging, especially since the two (and three) photon-dependence of excitation “amplifies” the effects of a fluctuating laser. In addition to multiphoton imaging, a potentially even more sensitive means to judge stability would be to test the laser as an excitation source in fluorescence correlation spectroscopy (FCS) experiments, where laser noise (e.g. oscillations) would be very obvious (e.g. FCS measurements on the same sample made with our Ti:S compared to ones carried out with our prototype laser). By installing our prototype laser source on one or more multiphoton systems we will test the laser in the most practical way by using it in a routine day-to-day fashion for a variety of imaging projects. Our main objective is to compare the imaging performance of MPM with the proposed sources and with conventional Ti:S lasers.
Our second stage demonstration experiments are designed to showcase the unique advantages of the proposed femtosecond sources, which have several important functional attributes for multiphoton imaging not found in the commonly used Ti:S laser. A review of the properties of the lasers being developed in Aims 1-3 and their importance to multiphoton imaging include: (1) all-fiber sources with integrated fiber delivery, (2) rapid, electronically controlled wavelength tuning, and (3) energetic pulses, particularly at the longer wavelength window of 1030 to 1280 nm.
(1) All-Fiber Sources with Integrated Fiber Delivery.
The two femtosecond sources proposed are both all-fiber sources with integrated single mode fiber-delivered (with >5 m of fiber length); that is, the output of the sources could be directly fed into a microscope scanbox, an endoscope scanning system, or through a biopsy needle for tissue spectroscopy. For multiphoton microscopy this would greatly simplify installation and maintenance of the system since alignment would be trivial; and for endoscopic imaging and spectroscopy applications fiber-delivered illumination is clearly essential.
A stable fiber-delivered femtosecond source in the 780-850 nm range can be directly incorporated in our current endoscope scanner design. We also note that the HOM fibers are highly resistant to bending loss, a characteristic that is impossible to obtain in the large area mode fiber previously demonstrated for pulse delivery [24]. Thus, it is particularly suited for small diameter, flexible endoscopes where bend radius as small as ˜1 cm is necessary. Although no clinical experiment is planned within the scope of this program, the long fiber delivery length (˜6 m) allows the source to be at a remote location away from the operating room. In a clinical environment, such a physical separation offers major practical advantages, such as eliminating the complications of sterilization, ultimately leading to a much reduced cost.
The HOM fiber that provides the dispersion compensation and wavelength tuning through SSFS can also be simultaneously used as the delivery and collection fiber for tissue spectroscopy. The diameter of the optical fiber is ˜0.125 mm (standard size for a single mode fiber), which is much smaller than the inside diameter of an 18 or 20 gauge needle that is routinely used for core biopsy. The excited signal will be collected by the same fiber. A fiber wavelength division multiplexer (WDM) can be placed between the fixed wavelength femtosecond source and the HOM fiber module to direct the collected signal to the detecting unit, which consists a grating and a CCD. In addition, the rapid wavelength tuning capability allows the emission spectrum of the tissue to be recorded as a function of the excitation wavelength. These multiphoton excited fluorescence excitation-emission matrix (EEM) can potentially provide unique diagnostic signatures for cancer detection just as one-photon EEM does [87, 88].
One potential complication of the proposed tunable source for multiphoton EEM is the power and pulse width variation across the entire tuning range. Calibration using a known multiphoton excitation standard, such as fluorescein dye, will be carried out before experimentation on biological samples. Such a calibration procedure is routinely used in previous multiphoton spectroscopy work. Multiphoton excitation standards have been established in the past and has extensive experience in multiphoton spectroscopy [71]. We don't expect significant problems in the calibration of the instrument.
Application of MPM in early cancer detection using a transgenic mouse line in which tumor formation is initiated by the conditional inactivation of the p53 and Rb1 genes by Adenovinis-Cre-mediated recombination has been reported [90]. The experiment on endoscopes and tissue spectroscopy through needle biopsies is highly synergistic with the on-going cancer research and provides an ideal platform for showcasing the “all-fiber” characteristics of the proposed femtosecond source.
A unique capability of the proposed sources is the ability to rapidly tune the wavelength much faster than currently possible with single box Ti:S systems. Rapid wavelength tuning would allow for line by line switching between excitation wavelengths during scanning, or for collecting excitation spectra, a potentially important parameter for biomedical applications that may utilize intrinsic fluorophores with overlapping emissions, but differing excitation spectra.
By synchronizing the wavelength control with the scanning and acquisition, we will modify one of our imaging systems to enable one wavelength during the “forward” line and a second during the return (without changing the Y position). This is analogous to what is now standard on modern AOM-equipped confocal microscopes, where, for example, a green dye is excited with 488 nm excitation in one direction and 547 nm excitation to excite a different dye during the return. In this way a two-color image can be collected using dyes with different excitation maximums and separable emissions. The temporal aspect eliminates problems with spectral cross-talk in many cases. Although multiphoton cross-sections for many dyes are broad often allowing for excitation of different dyes at the same wavelength (usually due to overlapping UV bands, so this normally only works at 800 nm or shorter), the ability to rapidly switch between wavelengths anywhere between 780 and 1000 nm would be an important enhancement for many dyes pairs. After interfacing the wavelength control with our scanning systems we will apply this capability in pilot experiments with fluorophores such as CFP and GFP which have different two photon excitation maxima, but partially overlapping emission spectra (
As an added benefit, the EOM device that enables rapid wavelength tuning can also be used to provide fast switching and modulation of the excitation beam. At a minimum this functionality should be comparable to what we currently achieve using our 80-mm resonance-dampened KTP* Pockel cells for routine beam blanking and intensity control (microsecond switching). Available fiber-coupled EOMs can switch in the sub-nanosecond range and should allow for a laser with a built-in modulator that would enable the user to reduce the effective laser repetition rate for measurements of fluorescent decays times and fluorescent lifetime imaging (FLIM), as well as for the more standard modulation needs. After implementing the required control electronics, we will use this functionality for routine beam blanking and control, photobleaching recovery measurements, and FLIM.
Another intriguing possibility provided by SSFS is that multiple wavelength tunable pulses can be obtained from the same fixed wavelength fiber source. For example, the output of the fixed wavelength femtosecond fiber source can be split into two halves and each half propagates through a HOM fiber module. The two HOM fiber modules can be the same (use power tuning) or of different lengths (length tuning). Such a multi-color femtosecond source opens a range of new opportunities, such as two-color two-photon excitation [9-94] and coherent anti-Stokes Raman scattering (CARS) imaging [95], where two synchronized ultrafast sources are needed previously. The spectral bandwidths directly from the proposed sources will likely be too large for CARS, possibly requiring spectral filtering or shaping.
The proposed longer wavelength femtosecond source offers unprecedented capability at the wavelength window of 1030 to 1280 nm. Although there are only a few experimental works for multiphoton imaging beyond 1100 nm, longer wavelength multiphoton imaging is feasible and can potentially offer significant advantage in deep tissue imaging, particularly with the high pulse energy we will be able to obtain. Efforts are underway on exploring this new spectral window for MPM, using the existing Ti:S pumped OPO. We will demonstrate the capability of the proposed femtosecond source for imaging in the 1.27 μm region using indicators such as shown in
All of the references listed below are hereby incorporated by reference in their entirety. These references are indicated herein above as being enclosed by brackets.
Although preferred embodiments have been depicted and described in detail herein, it will be apparent to those skilled in the relevant art that various modifications, additions, substitutions, and the like can be made without departing from the spirit of the invention and these are therefore considered to be within the scope of the invention as defined in the claims which follow.
| Filing Document | Filing Date | Country | Kind | 371c Date |
|---|---|---|---|---|
| PCT/US07/82625 | 10/26/2007 | WO | 00 | 12/7/2009 |
| Number | Date | Country | |
|---|---|---|---|
| 60863082 | Oct 2006 | US | |
| 60896357 | Mar 2007 | US |
