US 20060233554 A1
Described is an optical fiber system for delivering ultrashort pulses with minimal distortions due to nonlinearity. The system is based on delivering the optical pulses in a higher order mode (HOM) of a few-moded fiber. The fiber is designed so that the dispersion for the HOM is very large. This results in a dispersion length LD for the delivery fiber that is exceptionally small, preferably less than the non-linear length LNL. Under these conditions the system may be designed so the optical pulses experience minimum non-linear impairment, and short pulse/high peak power levels are reproduced at the output of the delivery fiber.
1. A method comprising:
(a) generating optical pulses, the optical pulses having a pulse width W,
(b) propagating the optical pulses through a pulse stretcher,
(c) converting the propagating mode of the optical pulses to a HOM,
(d) propagating the optical pulses along a length L of optical fiber to an output, where L is chosen such that the pulse width WO at the output is approximately equal to W.
2. The method of
3. The method of
4. An optical device comprising:
(a) a source of optical pulses,
(b) a pulse stretcher coupled to the source of optical pulses,
(c) a mode converter coupled to the pulse stretcher,
(d) an optical fiber coupled to the mode converter, the optical fiber supporting a HOM.
5. The optical device of
6. The optical device of
7. The optical device of
8. The optical device of
9. The optical device of
10. The optical device of
11. The optical device of
12. The optical device of
13. The optical device of
14. The optical device of
15. The optical device of
16. The optical device of
17. The optical device of
18. An optical fiber supporting a HOM, and having a dispersion length LD and a non-linear length LNL, where LD is less than LNL.
19. The optical fiber of
20. The optical fiber of
This invention relates to optical fiber systems that produce very short, high power, optical pulses.
(Parts of the following section may not be prior art.)
Optical fiber lasers are available that produce optical pulses with high pulse energy, good beam quality and excellent optical characteristics. Several applications for these optical pulse lasers exist, ranging from time-resolved near-field scanning optical microscopy (NSOM) pump-probe experiments for understanding ultrafast electronic processes in materials (see S. Smith, N. C. R. Holme, B. Orr, R. Kopelman and T. B. Norris, “Ultrafast measurement in GaAs thin films using NSOM,” Ultramicroscopy, vol. 71, pp. 213-223, 1998); for two-photon fluorescence of dyes (see A. Lago, A. T. Obeidat, A. E. Kaplan, J. B. Khurgin, P. L. Shkilnikov and M. D. Stern, “Two-photon-induced fluorescence of biological markers based on optical fiber,” Optics Letters, vol. 20, pp. 2054-2056, 1995); for studying biological processed in living tissues (see G. Alexandrakis, E. B. Brown, R. T. Tong, T. D. McKee, R. B. Campbell, Y. Boucher, and R. K. Jain, “Two-photon fluorescence correlation microscopy reveals the two-photon natures of transport in tumors,” Nature Medicine, vol. 10, pp. 203-207, 2004). The last application has potential impact on the prospects for non-invasive cancer detection schemes where the delivery fiber is an endoscope (see E. B. Brown, Y. Boucher, S. Nasser, R. K. Jain, “Measurement of macromolecular diffusion coefficients in human tumors,” Microvascular Research, vol. 67, pp. 231-236, 2004).
For the case of studying live tissues, the fs pulse acts as the pump beam that excites fluorescence mediated by a 2-photon process. Since multi-photo processes are by nature inefficient, high peak powers are needed. However, this cannot be achieved by increasing the average power of the source, because high average power will cause tissue damage. Hence, such applications typically require pulses of the duration of roughly 100 fs, with pulse energies as high as 1 nJ, while the average power is maintained at roughly 100 mW or lower. A commonly used laser source for such schemes is a mode-locked Ti:Sapphire laser that can output very high peak powers with repitition rates of ˜80 MHz.
The delivery fiber desirably propagates the high power, short pulses through a (typically) 1-2 meter-long fiber, and provides an output that is close in characteristics to the laser output. However, there are two physical constraints that affect the output from the delivery fiber. The dispersion of the fiber, due to material as well as waveguide dispersion, leads to pulse broadening that transforms the 100-fs pulse at the input of the fiber into 10-20 ps long pulse at the fiber-output. In addition, since the peak power levels are so high, nonlinear phase shifts due self-phase modulation (SPM) lead to a narrowing of the spectral width of the pulse, further broadening the pulse. The dispersion effect is linear, and thus may be compensated by a bulk linear chirp element between the Ti:Sapphire output and fiber input. The linear chirp element may be a bulk grating or prism pair used to stretch or compress pulses. Such elements are capable of providing arbitrary amounts of positive or negative dispersion. For this application, they may be adjusted to provide dispersion that is equal in magnitude, but opposite in sign to that of the specified length of the fiber endoscope/delivery medium.
Accordingly, while the dispersion problem may be addressed with some effectiveness, the nonlinear SPM effect is non-recoverable. Hence, a majority of fiber delivery schemes work with special fibers or complicated phase engineering of pulses to counteract the SPM effect.
In a high performance system the delivery fiber should also be a single mode fiber with low loss. Propagation in multiple modes degrades the ability to tightly focus the output from the fiber. A tightly focused output enables concentrating the high peak power pulse on a small region, thus enabling efficient 2-photon fluorescence, as well as ensuring high resolution for microscopy applications. Propagation in multiple modes also spreads the pulses due to modal dispersion, which lowers the peak power and reduces the efficiency of nonlinear measurement techniques.
Low bend losses are desirable in applications, such as endoscopes, where the fiber, even though short in overall length, may still undergo multiple bends. Connection losses include the concatenation of a collimating element at the fiber output. This element will focus the output to a tight spot. Candidates for the collimating element are fiber-GRIN lenses, or thin-film-based diffractive optic elements such as mode transformers, or other beam shaping elements. Normally, such miniature beam shaping elements can be epoxy-bonded to the tip of the fiber, but considering that very high peak powers will emanate from the fiber, it is desirable that a mode transforming element such a long-period grating or GRIN lens be used, since they are fiber-based, and can be easily fusion spliced to the output of the fiber, with low loss and high power handling capability.
The baseline candidate for the delivery fiber is a standard fiber (doped core, and silica cladding) which is single-moded at the desired wavelength of operation. A typical desired operating wavelength is ˜800 nm (for Ti:Sapphire lasers), and the effective area (Aeff) of a standard single mode fiber (SMF) at this wavelength is <25 μm2. This fiber satisfies all the above criteria, but at pulse energies >0.1 nJ, SPM severely distorts the pulse. The pulse width rapidly expands past 250 fs (desired pulse widths are <200 fs).
A variety of solutions to the SPM problem have been proposed. Among these are using a multimoded fiber, but forcing signal propagation in the fundamental mode to enable signal propagation in a large Aeff, thus decreasing SPM. See F. Helmchen, D. W. Tank and W. Denk, “Enhanced two-photon excitation through optical fiber by single-mode propagation in a large core,” Applied Optics, vol. 41, pp. 2930-2934, 2002. However, pulse widths obtained with this method are still undesirably large, especially for two-photon applications.
A variation of the above solution is to use a large core, multimoded microstructure fiber. See D. Ouzounov, K. Moll, M. Foster, W. Zipfel, W. W. Webb and A. L. Gaeta, “Delivery of nanojoule femtosecond pulses through large-core microstructured fibers,” Optics Letters, vol. 27, pp. 1513-1515, 2002. Microstructured fibers are guided by a photonic crystal of air holes running through the glass fiber, and this mitigates mode coupling problems. However, it appears that coupling effectively only into the fundamental mode in microstructured fibers is a problem, and significant power is lost to higher order modes. This causes unwanted modal noise in the system. In addition, microstructured fibers have poor geometric control compared to standard doped fibers, and a potential drawback is geometric ovalities that would cause large polarization mode dispersion (PMD), a source of additional noise.
Another option is the use of photonic bandgap fibers, where the signal propagates in a central air core. In this case, most of the signal energy resides in air, and hence undergoes negligible amounts of SPM-based pulse broadening. However, photonic bandgap fibers are difficult to manufacture in comparison to doped fibers, and hence are not a cost-effective solution. Geometric regularity problems are severely exacerbated, leading to the possibility of high PMD and associated problems. They also suffer from the inability to splice a mode-shaping element at the fiber output, because the splice causes the photonic bandgap effect to disappear, and will yield large losses.
Another proposal is to use pulse shaping schemes to combat the nonlinear broadening in standard fibers. The pulse is temporally chirped, and spectrally narrowed before launching into the delivery fiber. While this produces short pulses, the power levels are low and not desirable for two-photon applications.
New approaches that can minimize the deleterious effects of nonlinear SPM, while maintaining the advantages of a standard fiber, such as low propagation and bend losses, low PMD, ability to splice to GRINs and other lenses, and high manufacturing yield and control, would represent a significant advance in the technology.
We have developed an optical fiber system for delivering ultrashort pulses with minimal distortions due to nonlinearity. The system is based on delivering the optical pulses in a higher order mode (HOM) of a few-moded fiber. The fiber is designed so that the dispersion for the HOM is very large. This results in a dispersion length LD for the delivery fiber that is exceptionally small, preferably less than the non-linear length LNL. Under these conditions the optical pulses experience minimum non-linear impairment, and short pulse/high peak power levels are reproduced at the output of the delivery fiber.
The relative magnitudes of dispersive and nonlinear effects in fibers used for short pulse propagation are succinctly described by two characteristic lengths, the dispersion length LD, and the nonlinear length LNL, given by:
In standard SMFs an 800-nm, 100-fs pulse with 1 nJ energy, typical values for the characteristic lengths are LD˜9 cm and LNL˜1.3 cm (the corresponding dispersion of SMF is −100 ps/nm-km). Thus LD>>LNL, and nonlinear effects dominate, yielding undistorted pulses only for energies as low 0.1 nJ (that is, only for pulse energies as low as 0.1 nJ, the LD/LNL ratio is substantially smaller than unity). Existing fiber designs to combat this problem, as mentioned earlier, concentrate on satisfying condition (2) by increasing the Aeff for signal propagation. This serves to make LNL significantly larger than LD (which is held nominally constant and similar to SMF).
The novel class of fiber designs proposed here yield an innovative means to satisfy condition (2). Instead of increasing Aeff (and thus LNL), the signal is propagated in a higher order mode (HOM) of a fiber specially designed to yield very high negative dispersions for one particular HOM. Hence, condition (2) is satisfied by holding LNL nominally constant and similar to that of SMF, but LD is significantly shortened by increasing the magnitude of (negative) dispersion provided by the HOM of the fiber.
HOMs of specially designed few moded fibers are especially suited for this application, because HOMs can offer very high dispersion values, while maintaining a large Aeff, and very low propagation and bend losses. It has been demonstrated that the LP11 mode of a fiber can have dispersions as high as −700 ps/nm-km at the operation wavelength of 1550 nm. It has also been shown that the LP02 mode at the operation wavelength of 1550 nm can have −210 ps/nm-km dispersion, and only 0.45 dB/km loss, yielding very high figures of merit (FOM=dispersion/loss) of 466 ps/nm-dB. This enables up to 50% longer transmission distances for communication pulses, because the large dispersion and Aeff of these fibers mitigate nonlinear distortions in comparison to a communications system that uses single mode dispersion compensating fibers.
Optical fibers suitable for use in the invention have low ratios of LD/LNL, which enables high power pulse propagation for fs laser pulse delivery systems, as described earlier. In the preferred embodiments, this ratio is less than 1, and preferably less than 0.5. The specific optical fibers described here utilize the LP02 of the fiber for pulse propagation, but similar designs can be achieved for any HOM. While such designs can be applied for any wavelength of operation, illustrative designs described below are optimized for fiber-delivery of Ti:Sapphire laser pulses, which nominally operate in the 800-nm wavelength range. That suggests that the wavelength range over which the devices of the invention preferably operate is 700-900. However, other wavelength regimes may also be found useful. As a reference, the specific optical fiber designs used to illustrate the invention can be compared to SMF, which has a LD/LNL ratio of ˜6.92, which yields undistorted pulses for pulse-energies up to 0.1 nJ (maximum undistorted pulse energy achievable with a fiber is roughly proportional to the LD/LNL ratio−value of LNL depends on pulse energy as well as undistorted width—this has been calculated for 1-nJ pulses of 100 fs width, in all cases illustrated here). The objective, in some preferred embodiments, is to achieve LD/LNL ratios smaller than unity. Few mode fibers supporting these specially designed HOMs can be distinguished from standard multimode fibers in two respects. Firstly, they are intentionally designed to be highly dispersive for one particular, desired HOM, in contrast with multimode fibers, where most of the modes experience negligible waveguide dispersion, and the dispersion of all modes is similar to the material dispersion of silica glass. Secondly, the fibers are designed such that the propagation constant of the desired HOM of propagation is sufficiently separated from other modes, so as to avoid intermodal coupling at bends.
Given the properties of the novel HOM fiber illustrated in
The flexibility of this design space is further illustrated with the theoretically designed fiber whose refractive profile is illustrated in
HOMs also provide a greater degree of design-freedom to achieve desired dispersion profiles, in addition to the large dispersion magnitudes. Note that the dispersion profiles in
The inventive fiber designs illustrated here can be utilized in a high power pulse delivery system.
For most applications, the high power pulse at the output of the fiber is focused on to a small spot size to obtain large 2-photon fluorescence. As mentioned earlier, the HOM fiber should not be confused with a multimode fiber. The signal exits the fiber in a single well-defined mode. Hence it can be focused with lenses in a manner identical to conventional Gaussian beams, to achieve any desired spot size.
While the mode converters described in connection with
While different types of mode converters may be used for the invention, as indicated above, a preferred means to obtain the mode-converting device functionality is with a broadband long period fiber grating (LPG). The LPG may be induced in the HOM fiber itself, enabling a low cost, low loss, mode-converting device. Broadband mode converters are known that cover a wavelength range as large as 500 nm. For more details see S. Ramachandran, M. Yan, E. Monberg, F. Dimarcello, P. Wisk and S. Ghalmi, “Record bandwidth microbend gratings for spectrally flat variable optical attenuators,” IEEE Photon. Tech. Lett., vol. 15, pp. 1561-1563, 2003; S. Ramachandran, U.S. Pat. No. 6,768,835, both of which are incorporated by reference herein.
Whereas it is shown or may be inferred that the output from the short pulse device of the invention is propagated in free space, using standard collimating devices, it may also be coupled to other forms of media.
Methods for making optical fibers with profiles like those in
Optical fibers as described above that are specially designed to support HOMs may be construed as meaning that a substantial portion, typically a predominant portion, of the optical energy propagating in the optical fiber is in a mode higher than the fundamental mode LP01. Preferred HOMs are LP02 through LP0,10; and LP11 through LP1,10.
The element used to chirp the pulses, in the systems described here, is referred to as a pulse stretcher, which is a term familiar to those skilled in the art. For a another description of these elements see http://www- phys.llnl.qov/Orqanization/VDivision/Research/USP/USPFacilityVirtualTour/cpa.h tml incorporated herein by reference. The preferred choice of pulse stretchers are those operating on bulk optics, i.e. the optical pulses propagate through the stretching element. High-quality gratings and prisms are in this category.
The operation of the devices described above relies in part on having relatively high dispersion in the HOM fiber. While the actual dispersion value will vary, the typical dispersion value will be less than (more negative than) −150 ps/nm-km. The length of the delivery fiber will in part be determined by the dispersion value. In a qualitative sense, that length is where the dispersion in the HOM fiber compensates for the nominal dispersion from the pulse stretcher that appears at the input of the HOM fiber, but before the optical pulses undergo significant non-linear distortion. That length is typically from 1-20 meters. It should be evident that this relatively short length distinguishes in the usual sense this fiber from a transmission fiber.
While in principle the devices described here may function over a wide band of pulse frequencies and pulse length, the invention is preferably directed to devices where the pulses are femtosecond pulses (i.e. less than 1 picosecond), or shorter. In preferred embodiments the pulses are less than 200 femtoseconds.
Various other modifications of this invention will occur to those skilled in the art. All deviations from the specific teachings of this specification that basically rely on the principles and their equivalents through which the art has been advanced are properly considered within the scope of the invention as described and claimed.