Because the optical sources differ widely among various biophotonics imaging modalities, it is desirable to construct a versatile fiber-based optical source that accommodates maximum number of the imaging modalities. Our idea is to start with a high-power near-infrared (1050±50 nm) Ytterbium ultrafast fiber oscillator, and employ photonic crystal fiber (PCF) nonlinear frequency conversion to generate wavelengths outside the laser emission spectrum. We focus on three key techniques: (1) optical frequency up-conversion; (2) optical frequency down-conversion; and (3) high-quality supercontinuum generation.
Optical frequency up-conversion
One decade ago a seemingly straightforward attempt of passing near-IR femtosecond laser pulses through a microstructured pure-silica fiber by several Bell Lab investigators initiated the so-called supercontinuum revolution in the rapidly evolving field of nonlinear fiber optics. Today, fiber supercontinuum generation has become commonplace due to its relatively simple setup, and has found numerous interdisciplinary applications. Theoretically, supercontinuum generation has been deconstructed into a mixture of well-known nonlinear optical processes through which the energy of an input femtosecond pulse is spread rather uniformly across a continuum of spectral bands that include the starting spectrum of the pulse. The initial surprise associated with the fiber supercontinuum has long passed. So, what next?
Perhaps new insights can be gained from the original observation of the fiber supercontinuum phenomenon. What makes this experiment visually striking is that an invisible infrared laser beam is transformed inside the microstructured fiber into an intense white light that lights up the entire fiber in a dark ultrafast laboratory. Would it be equally striking if the infrared laser beam can be transformed inside a microstructured fiber into an intense (multimilliwatt-level) colored light of choice, rather than a broad spectrum of white light? Or more technically, can the energy of the laser pulse (pump) be routed to a targeted anti-Stokes band (signal) having a wavelength shorter than the laser wavelength, rather than be nonspecifically spread into a supercontinuum?
It is obvious that such targeted routing of femtosecond pulse energy (TRFPE) is mutually exclusive with the supercontinuum generation. The promotion of the former necessitates the suppression of the later, and vice versa. The key strategy to promote the supercontinuum generation is to match the zero-dispersion wavelength (ZDW) of the fiber with the laser wavelength, so that the input pulse sustains a high peak intensity over a long (> 1m) fiber length to enhance all the nonlinear optical processes. Thus, TRFPE may be achieved by detuning the ZDW of the fiber far away (> 100 nm) from the laser wavelength. This operation regime, termed as the short nonlinear-interaction condition, has been highly uncommon because it could similarly suppress the intended nonlinear process that promotes the TRFPE. Somewhat surprisingly, our recent studies have realized robust TRFPE under the short nonlinear-interaction condition through intermodal four-wave mixing and soliton-mediated Cherenkov radiation.
Intermodal four-wave mixing
Four-wave mixing (FWM) is a well-known physical process to achieve TRFPE. Unfortunately, the phase-matching condition of conventional fiber-optic FWM forbids the efficient stimulation of the process under the short nonlinear-interaction condition. Thus, the supercontinuum onset has largely restricted the pump-to-signal conversion efficiency to within 2%, preventing the generation of a multimilliwatt-level signal. Interestingly, the above fundamental restriction can be lifted by fulfilling the phase-matching condition through an intermodal scheme in which the pump and the signal propagate in different fiber. The FWM with such phase-matching scheme, termed as the intermodal FWM, is stimulated at a deeply normal dispersion regime of a microstructured fiber. In other words, the microstructured fiber is designed to have a large mode-area (~40 μm2) so that its ZDW (~1200 nm) is detuned far to the red from the Ti:sapphire laser wavelength of ~800 nm. Over 7% conversion efficiency from the pump input to yellow 586-nm signal has been attained under such short nonlinear-interaction condition, enabling the generation of a multimilliwatt-level signal without noticeable contamination from the supercontinuum generation. The signals of different colors can be generated by a series of similarly microstructured fibers having slightly different ZDWs.
Why can the intermodal FWM be so efficient under the short nonlinear-interaction condition? It should be noted that the efficiency of the conventional single-mode FWM is usually limited by the walk-off length of the ultrashort pump and signal (or idler) pulses, which can be comparable or shorter than the nonlinear-interaction length of the pump pulse. Thus, the short nonlinear-interaction condition itself does not necessarily lead to low FWM efficiency. On the other hand, this condition minimizes all nonspecific frequency conversion processes so that the input pulse predominately releases its energy through the intermodal FWM. In other words, such condition favors the phase-matched (i.e., resonant) frequency conversion process of the four-wave mixing over all non-phase-matched (i.e., nonresonant) frequency conversion processes of the supercontinuum generation.
Figure (above): Output spectra of a 20-cm LMA-10 fiber from a spectrometer (blue traces) and an optical spectrum analyzer (red traces) at pump power of 75 mW, 110 mW, and 140 mW; the green trace is the spectrum of the incident pump. Inset: (1) (2) observed far-field patterns of the pump and the anti-Stokes signal exiting from the fiber, respectively; (3) anti-Stokes signal power as the function of the pump power (dot) and the corresponding exponential fit of 0.00527exp(Ppump/18.4) (line).
|Tu H, Jiang Z, Marks DL, Boppart SA. Intermodal four-wave mixing from femtosecond pulse-pumped photonic crystal fiber. Applied Physics Letters, 94:101109, 2009.||PubMed Abstract|
Soliton-mediated Cherenkov radiation
The intermodal FWM is useful to generate large pulse-energy signal because amplified microjoule pulses are needed to stimulate the process in large mode-area microstructured fibers. Since the corresponding laser source is bulky, efforts should be taken to realize the TRFPE from compact femtosecond lasers producing unamplified nanojoule pulses. To stimulate nonlinear effects from these low-energy pulses, the microstructured fiber is designed to have a small mode-area (~2 μm2) so that its ZDW becomes shorter than the laser wavelength, or equivalently, the fiber is pumped at an anomalous dispersion regime where optical solitons rule. On one hand, the temporal compression of the solitons can emit light at a phase-matched spectral band, termed as Cherenkov radiation (CR), which may be used to achieve the TRFPE. On the other hand, CR has been widely accepted as the mechanism shaping the short-wavelength edge of the supercontinuum. This intriguing controversy is resolved recently by generating CR under the short nonlinear-interaction condition attained at a deeply anomalous dispersion regime of the fiber. This experiment definitely reveals the narrowband nature of CR that has long been obscured by the supercontinuum generation under long nonlinear-interaction conditions.
The reasons why the CR can be so efficient under the short nonlinear-interaction condition parallel those explaining the intermodal FWM. Since the CR is typically completed within the first few centimeters of the fiber, the short nonlinear-interaction condition has little influence on its efficiency. On the other hand, this condition adversely affects all the competing nonresonant frequency conversion processes of the supercontinuum generation to gain the energy from the pump pulse. The pump-to-signal conversion efficiency routinely surpasses 10%, enabling the generation of multimilliwatt CR signal.
While the tunability of the CR can be facilitated by tuning the pump wavelength of a widely-tunable unamplified Ti:sapphire laser, an alternative approach is to pump a series of PCFs with ZDWs ranging from 700 to 950 nm at a constant wavelength of ~1050 nm afforded by a compact Nd:Glass femtosecond laser or ytterbium-doped femtosecond fiber laser. The compact module of parallel-mounted short PCFs may add to such source the capability of a widely-tunable visible picosecond laser, or femtosecond laser if the CR pulse is compressible.
Figure (above): Visible laser-like output of fiber optic Cherenkov radiation generated by passing 1020-nm 200 fs 80 MHz input pulses in a PCF. The color of the output can be chosen by selecting the PCF with the proper size of the guiding core located in the center of the cross sectional SEM image of the PCF (black-and-white insets). The colored insets show the far-field images of the output from seven PCFs, illuminating a paper screen. The autofluorescence from the paper gives the bottom image a false blue color.
|Tu H and Boppart SA. Optical frequency up-conversion by supercontinuum-free widely-tunable fiber-optic Cherenkov radiation. Optics Express, 17:9858-9872, 2009.||PubMed Abstract|
|Tu H, Boppart SA. Ultraviolet-visible non-supercontinuum ultrafast source enabled by switching single silicon strand-like photonic crystal fibers. Opt. Exp., 17(20):17983-17988, 2009.||PubMed Abstract|
Optical frequency down-conversion
We intend to frequency down-convert the ~1050 nm pulses into narrowband fs IR (1100- 1500 nm) pulses, using the solition emission at the deeply anomalous dispersion regime of a PCF. Most of the PCFs employed in the Cherenkov radiation can also be employed to generate fiber solitons. Once a particular PCF and a specific fiber length are selected, it can be incorporated into a compact fiber-launching device requiring little day-to-day optical alignment. Soliton pulses of particular central wavelength can be obtained if an appropriate optical filter can be found. These spectrally isolated solitons are transform-limited fs pulses that can be used for nonlinear microscopy. The optimum operation for soliton emission depends on several factors. It should be noted that the soliton in the red edge is the most useful because it can be easily separated from the other solitons by a cutoff filter. To increase the conversion efficiency from the pump to this soliton, the condition of lower order soliton excitation is favored.
Figure (above): Soliton generation from a 1-m PCF using two different pump lasers
|Tu H, Boppart SA. Versatile photonic crystal fiber-enabled source for multi-modality biophotonic imaging beyond conventional multiphoton microscopy. Proc. SPIE, Vol. 7569, 75692D, 2010.||n/a||n/a|
High-quality supercontinuum generation
The common practice to generate broadband supercontinuum is to match the ZDW of the fiber with the laser wavelength. However, the generated supercontinuum has several problems, such as the presence of spectrally-fine structure, lack of spectral flatness, noise, and poor coherence. The lack of spectral flatness has been attributed to the generation of Raman solitons and the wave trapping. Thus, high-quality supercontinuum generation must avoid fibers having ZDW(s). Following this hint, we have produced broadband supercontinuum from ultrahigh numerical aperture (UHNA) fibers and PCF using the ~800 nm pulses from a regular Ti:sapphire laser.
Figure (above): Comparison of the input and output spectrum from a prestine UHNA3 fiber, and the output spectrum after light treatment (800 nm, 400 mW, 1 hr); (b) stabilized broadband continuum generation from a light-pretreated (910 nm, 900 mW, 1 hr) UHNA3 fiber with tunable center wavelength across 800-910 nm; (c) continuum generation from the light-pretreated fiber with tunable bandwidth by adjusting the input power.
|Marks DL, Oldenburg AL, Reynolds JJ, Boppart SA. Study of an ultrahigh numerical aperture fiber continuum generation source for optical coherence tomography. Opt. Ltr. 27:2010-2012, 2002.||n/a|
|Tu H, Marks DL, Koh YL, Boppart SA. Stabilization of continuum generation from normally dispersive nonlinear optical fibers for a tunable broad bandwidth source for optical coherence tomography. Opt. Lett., 32(14):2037-2039, 2007.||PubMed Abstract|
To further enhance the spectral broadening, we seek to produce a broadband (740-1300 nm) supercontinuum using an all-normal dispersion-flattened PCF. We use a rigorous multipole method to calculate the modes of the most commonly used PCFs. With the appropriate selection of the pitch () and the hole diameter (d) of the PCF, the dispersion of the PCF can be all-normal with a minimum dispersion located around 1060 nm. We use the full power of our Ti:sapphire laser (Mai Tai HP, Spectra-Physics) at 1020 nm to pump the closest commercial available PCF to our design (NL-1050-NEG-1, Crystal Fibre A/S, Denmark), and obtained the broadband supercontinuum We have conducted various standard measurements to confirm the excellent coherence and noise properties of this supercontinuum.
Figure (above): Dispersion relation (a) and supercontinuum profile (b) of an NL-1050-NEG-1 PCF. The inset shows the cross-sectional image of the fiber.
Characterization and analysis of relative intensity noise in broadband optical sources
Relative Intensity Noise (RIN) is one of the most significant factors limiting the sensitivity of an optical coherence tomography (OCT) system. The existing and prevalent theory being used for estimating RIN for various light sources in OCT is questionable, and cannot be applied uniformly for different types of sources. The origin of noise in various sources differs significantly, owing to the different physical nature of photon generation. Shot noise limited detection has been the ultimate goal in Time-Domain, Spectral-Domain and Swept-Source OCT system. However the assumption behind the shot noise limited detection requires perfect cancellation of excess photon noise, which is only applicable in ideal cases. Thus, it is important to characterize and compare the excess photon noise which exists with SLD based optical sources. SLDs are the most commonly used optical source for OCT owing to the advantages in price and portability.
Excess photon noise for a purely spontaneous source is given by while we introduce an empirical noise suppression factor η when measuring RIN, modifying the equation to .
In this study we characterize and measure the noise from optical sources and compare with the theoretically estimated values. Findings include a noise suppression factor η which depends on the different gain dynamics, waveguide structures and properties between different ASE based sources. Based on the physical interpretation we have on the measured noise data we propose an easily configurable low noise optical source, which is an amplified SLD source.
a) RIN vs. linewidth for an SLD source (Covega Inc.) at 100 kHz. b) RIN vs linewidth for an EDFA at 100 kHz.
(Upper) Schematic of a conventional SLD source. (Lower) Schematic of a proposed amplified SLD source.
Using the proposed amplified SLD source we could observe the relative intensity noise decreasing up to 9dB. This easily configurable SLD source setup will have an advantage especially on the SD-OCT and SS-OCT systems, as heterodyne detection for excess photon noise cancellation is not easily applicable for those systems. Also the advantage of low price optical source setup based on SLDs provides benefit for compact and low-priced optical source.
Shin S, Sharma U, Tu H, Jung W, Boppart SA. Characterization and analysis of relative intensity noise in broadband optical sources for optical coherence tomography. IEEE PTL vol. 22, pp. 1057-1059, 2010.