Expand this Topic clickable element to expand a topic
Skip to content
Optica Publishing Group

Widely tuneable fiber optical parametric amplifier for coherent anti-Stokes Raman scattering microscopy

Open Access Open Access

Abstract

We present a narrow-bandwidth, widely tunable fiber laser source for coherent anti-Stokes Raman scattering (CARS) spectro-microscopy. The required, synchronized, two-color pulse trains are generated by optical-parametric amplification in a photonic-crystal fiber (PCF). The four-wave-mixing process in the PCF is pumped by a 140ps, alignment-free fiber laser system, and it is seeded by a tunable continuous-wave laser; hence, a high spectral resolution of up to 1cm−1 is obtained in the CARS process. Since the PCF is pumped close to its zero-dispersion wavelength, a broad parametric gain can be accessed, resulting in a large tuning range for the generated signal and idler wavelengths. CARS spectroscopy and microscopy is demonstrated, probing different molecular vibrational modes within the accessible region between 1200cm−1 and 3800cm−1.

© 2012 Optical Society of America

1. Introduction

Laser based spectroscopy and imaging have become important tools for biochemical and medical applications in the past decades. Nonlinear optical effects such as second (SHG) and third harmonic generation (THG) provide the opportunity for damage-free investigations in living cells (in-vivo). The missing chemical selectivity of these processes can be overcome by stimulated Raman scattering (SRS) and fluorescence microscopy, with the drawbacks of either a slightly lower imaging speed, due to a required lock-in data acquisition [1], or the sacrifice of label-free operation. Coherent anti-Stokes Raman scattering (CARS) provides both label-free chemical information and fast acquisition times [2,3]. CARS is a third order non-linear interaction, where the stimulated emission of an anti-Stokes signal is related to resonant interactions with molecular energy states. This fact confers CARS chemical sensitivity. Thus, a suitable incident frequency set (ωpump/probe, ωStokes) is required in order to collect intense anti-Stokes signals. This resonant characteristic makes CARS a competitive technology to gain specific tissue information with high speed.

Due to the need of two widely separated and narrowband frequencies, the usual laser systems of choice are solid-state laser pumped optical parametric oscillators (OPO), which provide a high tuning range [2]. These systems have megahertz repetition rates and typical pulse durations between sub-picoseconds and ten picoseconds. This range of pulse durations has been found to be a good compromise between reasonable spectral resolution and the required high peak power. However, adjusting temporal and spatial overlap of the two pulse trains increases experimental difficulties. In addition the costs and size of such table-spanning systems prevent the spread of CARS systems in biochemical sciences.

Over the past few years new approaches were proposed for generating the required wavelength set directly inside of optical fibers. Fiber technology offers the benefit of highly stable, compact and robust laser sources. The first fiber-based CARS pump sources used soliton self frequency shift (SSFS) [4] and super continuum (SC) generation [5, 6]. Despite of their good wavelength tunability, these systems come with a rather low spectral power density and poor spectral resolution. To overcome this problem, efficient spectral compression by SHG was demonstrated [7], leading to an increased system complexity.

As an alternative approach, efficient, degenerately pumped four-wave-mixing (DFWM) in endlessly single-mode PCFs could be shown [6, 811]. DFWM is an optical nonlinear third order process where one pump field interacts with the waveguide material generating both a signal and an idler field. The resulting frequency separation between the generated fields and the pump is given by the phase-matching condition; and, hence, it can be adjusted by the dispersion of the fiber and, consequently, by a proper design of the air-hole structure of the PCF. Using pump pulses with a width of tens of picoseconds, the DFWM provides temporally and spatially overlapping pulses out of one fiber end. The spectral components have a well-separated spectral distribution with bandwidths of few nanometers, leading to a spectral resolution of a few ten wavenumbers [10,11]. Such sources were successfully applied to CARS microscopy providing fast imaging with pixel-dwell-times of one microsecond.

As other parametric processes in bulk media, DFWM in PCFs also yields a spectrally distributed parametric gain [12]. Due to the high nonlinearity of small-core PCFs this gain is so high that the DFWM signal is generated out of quantum noise over fiber lengths of several ten centimeters. As optical parametric amplification (OPA) theory predicts [13], this gain can also be used for the amplification of an external seed signal. Especially a continuous wave (cw) seed signal is beneficial for CARS applications due to the generation of narrow, transform-limited signal and idler pulses. Furthermore, a cw-seed ensures permanent temporal overlap with the pump pulses, without any need of complicated locking or delay schemes. Parametric amplification of a cw seed signal could efficiently be demonstrated previously [1416]. Most recently a fiber optical parametric amplifier (FOPA) has been used for CARS [16]. By tuning the pump and seed wavelengths, excitation frequencies between 2700cm−1 and 3200cm−1 were potentially addressable.

With regard to the application of the FOPA to CARS, two main aspects are of interest - the spectral resolution and the accessible tuning range. In this contribution we present a tuneable FOPA system with an increased accessible frequency range, designed to address both the characteristic fingerprint region and the excitation region for the CH stretching modes with a high spectral resolution of a few wavenumbers. In contrast to the source presented in [16] significantly longer pulses are used giving some major advantages. First of all a significantly increased spectral resolution is obtained, which improves image contrast and is also advantageous for differentiating structurally similar molecules. In addition, the potential measurement of subtle frequency changes caused by the chemical environment of the molecule would require high spectral resolution. Further significant advantages of the longer pulse duration are the more compact setup and simplified experimental handling. The increased robustness of the pulses against nonlinear spectral broadening and temporal walk-off allows for alignment-free all-fiber implementation and fiber pulse delivery. The simultaneous emission of all-pulses from a single PCF with intrinsic temporal and spatial overlap, as presented herein, simplifies handling as no adjustable delay line is required, furthermore, the alignment-free, perfect spatial overlap allows to illuminate a larger field of view more homogeneously.

The presented source consists of a pulsed all-fiber laser which provides parametric gain inside of a PCF. This gain is used to amplify a cw Ti:Sapphire seed signal generating a narrow-band signal and its idler, respectively. This setup will be presented in section 2 while the output characteristics of the FOPA will be described in section 3. The different combination possibilities of the three output wavelengths give access to both of the above mentioned excitation bands. By tuning the cw seed within the parametric gain bandwidth, high resolution CARS spectroscopy and multi excitation CARS microscopy are demonstrated in section 4.

2. Setup of the fiber optical parametric amplifier

To drive the fiber optical parametric amplifier (FOPA) used in our experiments two laser systems are applied - a pulsed fiber laser for pumping and a bulk laser for seeding the nonlinear process in the photonic crystal fiber (PCF). Figure 1 shows the scheme of the setup.

 figure: Fig. 1

Fig. 1 Schematic setup of the seeded fiber optical parametric amplifier. Pump and seed will be coupled into a PCF with a length of 0.35m. Labels: BP - band pass, DM - dichroic mirror, HWP - half-wave plate. The graphs below show the characteristics of the incident laser pulses: (a) spectrum of the Ti:Sa seed signal; (b) spectrum of the fiber laser pump; (c) temporal shape of the pulsed pump

Download Full Size | PDF

A spectrally filtered and amplified mode-locked fiber laser similar to the one presented in [10] is used to pump the parametric process. The emission wavelength is selected by an internal FBG, which is centred at 1038nm. By mechanical expansion or compression such a grating can easily be tuned by ±3nm when placed in a suited mounting [17]. This approach provides a simple tuning mechanism of the central wavelength of the pump laser system. The grating, moreover, determines the bandwidth and pulse duration of the laser. For the chosen grating bandwidth of 15pm, a pulse duration of 140ps was obtained (Fig. 1(c)). The pulses are transmitted through a band-pass filter, which blocks unwanted ASE, and a polarizer, which ensures a correct excitation of the PM fiber. Due to self-phase-modulation within the amplifier fibers, the bandwidth of the pump is broadened to 0.08nm. The resultant spectrum, visible in Fig. 1(b), shows beginning modulations (double peak structure). Finally, an average power of 400mW is coupled into the PCF, corresponding to a peak power of about 2.7kW. This generates the necessary parametric gain in the FOPA. Including the PCF, the whole system is polarisation maintaining, which helps in mitigating parasitic effects such as nonlinear polarisation rotation [18].

For efficient frequency conversion, the selection of a suitable nonlinear fiber is important. The process of DFWM works most efficiently for perfect phase-matching, e.g. κ = 2kpkiks + 2γPp = 0 with 2ωp = ωi + ωs [12]. Consequently, the spectral distribution of the effective refractive index of the chosen fiber determines the signal and idler wavelengths. The resulting parametric gain g = g(κ) is dependent on the net-phase-matching κ. The resulting spectral gain distribution leads to the line shape of the generated FWM bands in unseeded FWM laser sources [10]. The bandwidth of this shape can significantly be broadened by pumping closer to the zero dispersion wavelength (ZDW) of the fiber. With some orders of magnitude the parametric gain enables the amplification of an external seed signal with only very low power. This allows the use of a continuous wave (cw) seed source which is beneficial due to the automatic temporal overlap with the pump pulses. The insets in the setup in Fig. 1 sketch the temporal pulse trace in front of and behind the PCF in case of cw seeded FOPA. Only in presence of pump intensity the cw signal is amplified and a respective idler pulse is generated. The non-amplified temporal regions between the signal pulses yield in a low power background not having any effect in a CARS sample.

We decided to use NKT Photonic’s SC-5.0-PM (pitch: 3.22, doL: 0.5, ZDW: 1040nm, mode field diameter: 5μm) for frequency conversion because it has its ZDW close to the emissions bandwidth of Ytterbium. Due to the flat dispersion characteristics near the ZDW, small frequency separations between pump and signal and broad gain bandwidths can be expected. This is confirmed by the numerical calculations shown in Fig. 2. The narrow bandwidth of the fiber pump output was neglected in the numerical calculations. Figure 2(a) shows two typical features: the signal wavelength gets closer to the pump wavelength when the pump wavelength of the fiber laser system approaches the ZDW of the PCF. At the same time the signal becomes spectrally broader, since the phase-matching condition is fulfilled for a broader spectral region. Choosing a pump wavelength around 1040nm we were expecting CARS excitation frequencies around 1450cm−1 with a bandwidth of several hundred wavenumbers. In the following experiments the DFWM signal of the three-color output is always used as CARS pump, while either the residual DFWM pump or the idler serves as the Stokes wave for the CARS process. Therefore two excitation frequency bands are available from the FOPA as depicted in Fig. 2(b). Thus with this system we should be able to generate two tuning bands - the soft tuning of the input seed signal’s wavelength and the hard tuning by slightly tuning the pump laser’s FBG by ±3nm around its center wavelength. By using both bands, the system theoretically provides an overall tuneabilitiy of the CARS excitation frequency from 1200cm−1 up to 3800cm−1 with a 400cm−1 wide gap around 2100cm−1, which contains only a few molecular modes of minor importance.

 figure: Fig. 2

Fig. 2 Simulation results for a 0.35m long 5μm core PCF pumped with 2.5kW. (a) shows the signal and idler wavelength position for the gain maximum as a function of the pump wavelength. The coloured areas illustrate the theoretical FWHM bandwidth of each band. (b) shows the calculated frequency separation between signal and pump or signal and idler as a function of the pump wavelength. The blue boxes illustrate the theoretically accessible Raman excitation regions by mechanically tuning the internal FBG by ±3nm (yellow box).

Download Full Size | PDF

To sample the spectral accessibility of the predicted parametric gain, a continuously emitting Ti:Sapphire oscillator is used. This system provides a tuneable continuos wave output from 850 up to 950nm at constant narrow bandwidth of <0.03nm (resolution limit of the optical spectrum analyzer employed) (Fig. 1(b)).

3. Fiber optical parametric amplification

The FOPA system described above has four input parameters available which were adjusted to the intended application - pump and seed power as well as pump and seed wavelength. First, the basic influence of a seed on the spectral distribution is investigated. The fiber laser pump provides the necessary peak power to drive optical parametric generation (OPG) from quantum noise along the 35cm fiber. The corresponding broadband signal and its respective idler spectrum are plotted light gray in Fig. 3(a). As expected, the frequency separation between the two possible wavelength combinations touches the fingerprint region as well as the CH stretching bands. The average power of the signal is 33mW which is perfectly suited for doing CARS spectro-/microscopy. With a pump power of 390mW the energy conversion efficiency can be calculated to be 9.4% which compares well with former results [10].

 figure: Fig. 3

Fig. 3 (a) Spectrum measured at the output of the PCF for three cases: high pump power without seed (light gray), high pump power with 5mW seed (dark gray) and lower pump power with 5mW seed (red). In the first and in the last case the signal output power was the same (33mW). The spectra below show (b) signal (including unamplified seed spectrum), (c) pump and (d) idler for the last case in more detail. The frequency separation between signal and pump is within the so-called fingerprint region. This frequency separation can be increased by a factor of two when using signal and idler.

Download Full Size | PDF

In presence of a 5mW narrowband seed, the signal average power (without cw underground) increases to 55mW. Simultaneously, in the spectrum the maximum power density increases more than one order of magnitude whereas the spectral width of the signal decreases significantly from several nanometers to few tens of picometers (dark grey spectrum of Fig. 3(a)). A pedestal of spontaneous OPG is still visible, which could deteriorate the spectral resolution of the presented CARS pump system. To prevent this pedestal from appearing, the pump power is reduced to 280mW, where the signal power is the same as in the unseeded case (33mW). By doing this, the pedestal strongly reduced while the peak spectral power density remains nearly the same (compare with red spectrum in Fig. 3).

Figure 3(b–d) show the different spectral bands in more detail. All three bands provide a remarkable narrow FWHM bandwidth. The broadest line corresponds to the pump, whose bandwidth increases from initially 0.08nm to 0.24nm due to self-phase modulation (SPM) in the single-mode PCF. However, the SPM-induced spectral structure (Fig.3(c)) is stable over time and its full width corresponds to a still small value of 2.2cm−1. Even narrower are the signal and the idler with 0.06nm and 0.14nm or 0.8cm−1 and 1.0cm−1, respectively. The spectrum of the signal pulses is broader as the calculated time-bandwidth product due to SPM and parametric processes taking place near the signal wavelength. Nevertheless, this FOPA system provides a spectral resolution of few wavenumbers for both usable spectral band combinations. Especially in case of using signal and idler, the spectral resolution is more than one order of magnitude lower than other optical parametric fiber laser sources demonstrated previously [10, 11, 16]. Considering these bandwidths, the system performs even better than commercial table top bulk OPO systems (typically >10cm−1).

Such a high spectral resolution requires a fine tunability to adjust the desired CARS excitation frequency exactly. The tuning range of the FOPA is based on the accessible parametric gain of the PCF. The shape of the signal or idler spectrum in absence of seed gives a first estimation of the spectral gain distribution. The statistical nature of the quantum noise leads to the amplification of all frequencies located inside the parametric gain spectrum. Thus the OPG spectrum resembles directly the shape of the gain. Figure 4(a) shows the OPG spectral density of the idler as a function of the calculated frequency separation measured for a pump power of 350mW (dashed line).

 figure: Fig. 4

Fig. 4 (a) Measured OPG spectrum (350mW pump power) (dashed line, gray area) and measured idler power (300mW pump power, 5mW seed power) at the output of the PCF for different signal seed wavelengths (red circles) with respective Gaussian fit (red). Depending on the chosen wavelength pair, two excitation frequency bands (lower and upper abscissa) are addressable. (b) OPG spectra for different pump wavelengths and 350mW pump power.

Download Full Size | PDF

The accurate way to measure the shape of the gain spectrum is to detect the idler power for different seed wavelengths. As the non-seeded part of the parametric conversion, the idler is obtained without any cw background power. The idler measurement shown in Fig. 4(a) was done with 300mW of pump power and a 5mW seed power. The measured power distribution represents the accessible soft tuning range of the FOPA. Compared to the OPG spectrum mentioned above, the maximum of the curve is shifted due to the different pump power used in both experiments. This observation fits to numerical calculations predicting that the frequency distance between signal and pump decreases by 180cm−1 for a 50mW lower pump power. In good agreement a shift of 200cm−1 was measured experimentally.

Furthermore, it has to be noticed that the gain bandwidth remains nearly constant in the pump peak power range of 2.5kW to 3.5kW. The behaviour of the measured idler power (red) closely follows the OPG spectrum (gray). The FWHM bandwidth of both (OPG and seeded idler power) amounts to 230cm−1. This tuning range is doubled when signal and idler are used for CARS as signal and idler shift in opposite directions during tuning. The tuning range can be expanded further by tuning the FBG of the fiber pump laser as demonstrated by the OPG spectra shown in Fig. 4(b) exemplarily for three pump wavelengths and a constant pump power of 350mW. By detuning the FBG from 1038nm to 1040nm, the frequency difference between signal and pump decreases from 1740cm−1 down to 1590cm−1. In addition, the FWHM width increases slightly from 175cm−1 up to 230cm−1 in good agreement with the theoretical expectations.

Not only spectral purity but also pulse peak power is important to excite nonlinear processes such as CARS. Thus, a measurement of the temporal pulse shape is mandatory. For both cases - absence (OPG) and presence of a 5mW seed signal (OPA) - the temporal pulse traces were measured with a 18ps response time photodiode and a 70GHz sampling oscilloscope. For both measurements the pump power was chosen to achieve the same signal average power of 33mW. As can be seen in Fig. 5, less pump peak power is needed in the seeded case to reach the same signal power level as in the unseeded case. The measured pulse widths are very similar with 55ps (OPG case) and 65ps (OPA case). The difference of the inverse group velocity between signal and pump is numerically calculated to be 2.3ps/m and between signal and idler to be just 0.8ps/m. Hence, the usage of a few tens of picoseconds long pulses and short fiber length leads to an automatic temporal overlap of all three colors out of one fiber end, as demonstrated before in [10]. In addition, such long pulse widths are beneficial in terms of application with standard microscope optical lenses and filters due to a relatively small group velocity mismatch in glass (lower than 10ps/m for both possible wavelength combinations).

 figure: Fig. 5

Fig. 5 OPG: temporal pulse shape of DFWM pump (black, red) and signal (orange) in case of optical parametric generation (no seed). OPA: temporal pulse shape of DFWM pump (black, red) and signal (orange) in case of optical parametric amplification of a 5mW cw seed signal.

Download Full Size | PDF

With the measured signal pulse width, the peak power of the FOPA signal output can be estimated to be 530W. This corresponds to an amplification of 50dB in peak power. Furthermore 28% of pump peak power could be converted into the signal.

Besides the ease of use of the presented FOPA system, the seed power is one critical parameter that the user has always to take care of. Due to saturation effects, increasing the seed power will not lead in every case to a higher energy of the signal pulses. Figure 6(a) shows the development of the amplified net signal average power as a function of the incident seed power. As can be seen, DFWM gets saturated already for a few milliwatts of seed power. Any further increase of the seed power leads only to a drop of the signal output power due to saturation effects. Furthermore, a degradation of the temporal pulse shape with increasing seed power (see Fig. 6(b)) could be observed. Simulations show that back conversion in the most intense regions of the pulse is one reason for the pulse deformation. This results not only in a broadening of the pulses but also in a significant loss in peak power.

 figure: Fig. 6

Fig. 6 (a) Power of the amplified signal (cw background subtracted) over incident seed power. (b) Development of the temporal signal pulse shape with increasing seed. (c) Development of the signal output spectrum with increasing pump power and constant 5mW seed power.

Download Full Size | PDF

Consequently, tuning the seed power above the saturation limit is not an option for up-scaling the signal peak power. Only by keeping the input seed power below this limit, the peak power of Gaussian-like signal pulses increases with higher input seed as expected. To raise the saturation threshold, the pump power can be increased. Doing so one has to be aware of a further spectral broadening of all three wavelengths due to stronger self- and cross-phase-modulation effects as exemplarily shown in Fig. 6(c) for the signal. A possibility to avoid further spectral broadening while up-scaling the signal peak power is to use a shorter PCF together with higher pump powers.

4. Application of the FOPA system to CARS spectroscopy and imaging

We could show previously that the FOPA provides a self-synchronized, very narrowband three-color output with peak powers of at least several hundred Watts. This performance is perfectly suited for CARS applications. Especially, doing CARS spectroscopy provides the possibility to probe the actual spectral resolution and the tunability of the system. Accordingly we applied the FOPA to a mixture of n-hexane (C6H14) and toluol (C7H8) by focusing the collimated output into a cuvette containing both liquids. The beam is filtered previously to exclude any parasitic, non-phase-matched FWM between all three generated wavelengths in the PCF. This spurious signal would induce a broad CARS background in the sample, which would deteriorate the spectral characteristics of the resonance and therefore it has to be avoided. A second filter selects two of the three DFWM output wavelengths. Thus the CARS pump, which is always the DFWM signal (893–913nm), excites, together with one of the filtered Stokes wavelength (DFWM pump at 1039nm or the DFWM idler at 1205–1242nm), molecular vibrational modes in the sample. Behind the glass cuvette the generated CARS signal is filtered out of all other spectral components and it is measured with a multimode fiber connected to an digital spectroscope.

The CARS spectrum for different excitation frequencies was integrated and associated with the corresponding excitation wavenumber calculated from the CARS pump and Stokes wavelengths. This leads to a spectral distribution reflecting the Raman response of the sample. The tuning could be realized within the fingerprint region 1325cm−1– 1580cm−1 and the CH stretching regime 2670cm−1– 3160cm−1, respectively, while keeping the pump wavelength constant at 1039nm. The seed power was adjusted to keep the excitation power constant during the scan. To this end, for every wavelength setting, the CARS pump and Stokes power as well as their spectral widths were monitored to ensure constant excitation conditions. Thus the FOPA is able to provide 255cm−1 of tuning bandwidth for the first region, without wavelength tuning of the fiber laser, and 490cm−1 in the second one.

Figure 7 shows the spectral CARS power distribution within selected parts of both interesting frequency regions. Whereas Fig. 7(b) pictures a wide wavelength scan range over the CH stretching resonances of toluol and hexane, Fig. 7(a) details the CH deformation vibration of hexane. In both regions high resolution Raman spectra are displayed for comparison. These are shifted a few tens of wavenumbers to lower frequencies. Although a direct comparison of CARS lineshapes with pure Raman spectra is only hardly possible due to the complex single-resonance distribution of the CARS signal, it is obvious that the spectral CARS distribution fits roughly to the Raman measurements. Furthermore, with the aid of Fig. 7(a) or the inset of Fig. 7(b), it is possible to see that the system is able to resolve single resonances with bandwidths lower than 10cm−1. This demonstrates the high spectral resolution of the presented FOPA system.

 figure: Fig. 7

Fig. 7 Raman and CARS intensity of a mixture of n-hexane and toluol as a function of the wavenumber. The left panel displays the deformation vibration spectrum around 1430cm−1 belonging to n-hexane, whereas the right one corresponds to the CH stretching around 2900cm−1. All curves are normalised to their maxima.

Download Full Size | PDF

For image acquisition the laser was coupled into a home-built laser scanning microscope shown schematically in Fig. 8. The collimated beam is focused onto the sample using a NIR corrected microscope objective with a numerical aperture of 0.4 (Plan Apo NIR, Mitutoyo, Japan). The sample is imaged by raster scanning the laser focus across the sample by a galvanometric mirror. The CARS pump and Stokes wavelengths are filtered in a similar way as described for the spectroscopy experiments above. The power at the sample was kept low (30mW) in order to avoid photodamage. However, the peak power of signal, residual pump and idler pulses is sufficiently high to generate besides CARS also second harmonic generation (SHG) and two photon excited fluorescence (TPEF) signals in tissue. The light emitted in forward direction was recollimated and separated from the incident laser light using two short pass filters. For detection of the SHG and TPEF signals an additional short pass filter with a cut-off wavelength of 600nm was used. The CARS and SHG signals were then detected by a photomultiplier tube.

 figure: Fig. 8

Fig. 8 Schematic setup of the CARS microscope. L: achromatic lenses, LP: long pass filter at 850nm, BP: band pass filter at 1040nm, SP: short pass filter at 900nm, MM: moveable xy-scanning mirror, PMT: photo multiplier tube.

Download Full Size | PDF

For a microscopic image two frames of size 2048 × 2048 pixels were averaged with a pixel dwell time of 1μs. The system provides diffraction limited spatial resolution which results in a lateral resolution of 1μm. The presented light source is, due to its small footprint and compact and robust design, potentially suited for implementation into biomedical instruments. In order to demonstrate the capabilities of the described light source for simultaneous SHG, TPEF and spectrally highly resolved CARS imaging, a human arterial wall with artherosclerotic plaque deposition has been subsequently imaged. The excitation frequencies were chosen to excite the symmetric aliphatic methylene stretching vibration at 2845cm−1 and the methyl stretching vibration at 2930cm−1, in order to visualize the protein and lipid distribution.

The results of investigating the protein and lipid distribution at the inner arterial wall are displayed in Fig. 9. In panel (a) the combined SHG, TPEF and CARS signals are displayed simultaneously showing the lipid and protein distributions, while the separate channels are displayed in panels (b)–(d) on the right: SHG and TPEF (red), CARS 2850cm−1 (blue) and CARS 2930cm−1 (green) channels. Imaging at 2850cm−1 and 2930cm−1 allows qualitatively distinguishing proteins from lipids, since the ratio of CH2/CH3 functional groups is significantly higher in lipids than in proteins. Large lipid clusters are visible, coloured in turquoise due to the combination of intense CARS signals at 2850cm−1 (CH2, blue channel (Fig. 9 (b)) and weaker signals at 2930cm−1(CH3, green channel (Fig. 9 (c)). The areas of high protein content appear greenish, due to a lower CH2/CH3 ratio than in lipids. The protein distribution is very similar to the collagen and elastin distribution of the elastic fibers of the inner arterial wall visualized in the SHG and TPEF channel. Thus, these elastic fibers are the major source of the observed CARS protein signal.

 figure: Fig. 9

Fig. 9 Multimodal nonlinear microscopic image of an atherosclerotic plaque deposition at a human artery wall. In panel (a) the combined CARS microscopic images acquired at the aliphatic methylene stretching vibration at 2845cm−1 (blue) and the methyl stretching vibration at 2930cm−1 (green) for imaging the lipid and protein distribution are contrasted to the combined TPEF and SHG signal of collagen and elastin (red). Lipids are coloured turquoise due to their high CH2/CH3, while proteins appear greenish. The single channels are separately displayed in panels (b)–(d). The images of the area of 450 × 450 μm2 were acquired (2048 × 2048 pixels) illuminating the sample with 30mW of laser power.

Download Full Size | PDF

The CARS contrast, i.e. the intensity ratio of the CARS signal in resonance to the off resonant signal, determines the detection limit in CARS microscopy and can therefore be used for evaluating the image quality of the CARS light source. In order to demonstrate high CARS image contrast using a light source with high spectral resolution, intensity profiles of vibrationally resonant (2850cm−1) and off-resonant (3000cm−1) illuminated images were compared. The position of the profile is indicated in Fig. 10 (right) in case of the image at 2850cm−1. For improved visualization of the changing average intensities, the profiles were filtered using a 10 point moving average filter. Only the strongest features - the brightest lipid deposits - are visible in the off-resonant image. Thus, the image contrast is 30 times higher in the resonant case. This improvement has been estimated by calculating the ratio of the difference of the signal peak to the background intensity in the resonant and off resonant image. The high contrast was additionally verified with an off-resonant measurement at 2740cm−1. In the presented experiment the image contrast depends primarily on the concentration of the vibrational modes under investigation in contrast to lower spectral resolution light sources according to previous reports [19]. Therefore, no complicated detection schemes are required e.g. frequency modulation CARS in order to detect molecular species at low concentration.

 figure: Fig. 10

Fig. 10 Line intensity profiles across the selected region of the sample for vibrationally resonant and off-resonant excitation.

Download Full Size | PDF

5. Conclusion

In this work we have demonstrated a widely tuneable, fiber-based CARS laser source with high spectral resolution. An efficient fiber optical parametric amplifier generates a three-color output (pump, signal, idler) based on four-wave-mixing in a photonic crystal fiber. The nonlinear frequency conversion is pumped by a 1040nm fiber laser while it is seeded by a wavelength tune-able cw Ti:Sapphire system. The whole system is polarisation maintaining and, therefore, fully controllable and reproducible in its output power, wavelength set and polarization. A 5μm-core PCF provides a several 100cm−1 broad parametric gain for amplification of the tuneable seed. Either combination of the three wavelengths can be used for CARS, resulting in two accessible excitation bands (signal & pump or signal & idler). The use of 140ps pump pulses leads to an intrinsic temporal overlap of all three generated wavelengths, which simplifies the experimental handling compared to previously presented FOPA systems [16]. Beside its temporal alignment-free output the whole system is potentially all-fiber and thus prospectively alignment-free, if using a fiber-coupled, tunable, external-cavity diode laser (ECDL) as seed source.

With the chosen fiber and DFWM pump source, CARS generation is theoretically possible for Raman modes between 1200 – 1900cm−1 and 2300 – 3800cm−1 by hard- (DFWM pump wavelength) and soft-tuning (seed wavelength) touching the fingerprint region and CH stretching molecular modes. Actually both excitation bands could be applied to CARS spectro- and microscopy with one fixed DFWM pump wavelength showing a tuning bandwidth of 255cm−1 around 1450cm−1 and 490cm−1 around 2915cm−1. This spectral region can be further extended by adapting the pump wavelength. Significant resonances of Hexane and Toluol with a bandwidth narrower than 10cm−1 could be resolved. With approx. 1cm−1, the actual spectral resolution of the FOPA system is higher than that of typical commercially available picosecond OPO systems (typically >10cm−1 bandwidth of signal output). Finally, the tuning abilities of the FOPA system could be applied to acquire chemically selective image information based on the CARS excitation of two spectrally separated CH stretching bonds of both lipids and proteins. Due to the narrow excitation bandwidth the contrast of resonant signal and non-resonant background is as high as 30, which results in high image quality and, moreover, gives the opportunity to resolve even closely neighbouring resonances. The capabilities could be demonstrated by clearly resolving lipids and proteins in an aorta section.

Consequently, the FOPA system demonstrated represents a robust, quasi alignement-free CARS laser source providing a high spectral resolution, tuneable three-color wavelength set out of one single fiber end. With this performance, this fiber source directly competes with state-of-the-art OPO systems, while, at the same, it offers a compact size and simplified handling.

Acknowledgments

This work was supported by the German Federal Ministry of Education and Research (BMBF) under contract [ 13N10773] and [ 13N10774]; an the European Network of Excellence Photonics4Life; M.B. acknowledges support from the Carl-Zeiss-Stiftung. The authors thank Christian Matthäus for providing the biological samples.

References and links

1. B. G. Saar, C. W. Freudiger, J. Reichman, C. M. Stanley, G. R. Holtom, and X. S. Xie, “Video-rate molecular imaging in vivo with stimulated Raman scattering,” Science 330(6009), 1368–1370 (2010). [CrossRef]   [PubMed]  

2. C. L. Evans and X. S. Xie, “Coherent Anti-Stokes Raman Scattering Microscopy: Chemical Imaging for Biology and Medicine,” Annual Rev. Analyt. Chem. (Palo Alto Calif) 1(1), 883–909 (2008). [CrossRef]  

3. C. Krafft, B. Dietzek, and J. Popp, “Raman and CARS microspectroscopy of cells and tissues,” The Analyst 134(6), 1046–1057 (2009). [CrossRef]   [PubMed]  

4. G. Krauss, T. Hanke, A. Sell, D. Träutlein, A. Leitenstorfer, R. Selm, M. Winterhalder, and A. Zumbusch, “Compact coherent anti-Stokes Raman scattering microscope based on a picosecond two-color Er:fiber laser system,” Opt. Lett. 34(18), 2847–2849 (2009). [CrossRef]   [PubMed]  

5. R. Selm, M. Winterhalder, A. Zumbusch, G. Krauss, T. Hanke, A. Sell, and A. Leitenstorfer, “Ultrabroadband background-free coherent anti-Stokes Raman scattering microscopy based on a compact Er:fiber laser system,” Opt. Lett. 35(19), 3282–3284 (2010). [CrossRef]   [PubMed]  

6. T. Gottschall, M. Baumgartl, A. Sagnier, J. Rothhardt, C. Jauregui, J. Limpert, and A. Tünnermann, “Fiber-based source for multiplex-CARS microscopy based on degenerate four-wave mixing,” Opt. Express 20(11), 12004–12013 (2012). [CrossRef]   [PubMed]  

7. M. Marangoni, A. Gambetta, C. Manzoni, V. Kumar, R. Ramponi, and G. Cerullo, “Fiber-format CARS spectroscopy by spectral compression of femtosecond pulses from a single laser oscillator,” Opt. Lett. 34(21), 3262–3264 (2009). [CrossRef]   [PubMed]  

8. W. Wadsworth, N. Joly, J. Knight, T. Birks, F. Biancalana, and P. Russell, “Supercontinuum and four-wave mixing with Q-switched pulses in endlessly single-mode photonic crystal fibres,” Opt. Express 12(2), 299–309 (2004). [CrossRef]   [PubMed]  

9. D. Nodop, C. Jauregui, D. Schimpf, J. Limpert, and A. Tünnermann, “Efficient high-power generation of visible and mid-infrared light by degenerate four-wave-mixing in a large-mode-area photonic-crystal fiber,” Opt. Lett. 34(22), 3499–3501 (2009). [CrossRef]   [PubMed]  

10. M. Baumgartl, M. Chemnitz, C. Jauregui, T. Meyer, B. Dietzek, J. Popp, J. Limpert, and A. Tünnermann, “All-fiber laser source for CARS microscopy based on fiber optical parametric frequency conversion,” Opt. Express 20(4), 4484–4493 (2012). [CrossRef]   [PubMed]  

11. M. Baumgartl, T. Gottschall, J. Abreu-Afonso, A. Díez, T. Meyer, B. Dietzek, M. Rothhardt, J. Popp, J. Limpert, and A. Tünnermann, “Alignment-free, all-spliced fiber laser source for CARS microscopy based on four-wave-mixing,” Opt. Express 20(19), 21010–21018 (2012). [CrossRef]   [PubMed]  

12. G. P. Agrawal, Nonlinear Fiber Optics (Academic, 2001).

13. R. W. Boyd, Nonlinear Optics (Academic, 2008).

14. S. Hädrich, T. Gottschall, J. Rothhardt, J. Limpert, and A. Tünnermann, “CW seeded optical parametric amplifier providing wavelength and pulse duration tunable nearly transform limited pulses,” Opt. Express 18(3), 3158–3167 (2010). [CrossRef]   [PubMed]  

15. P. J. Mosley, S. A. Bateman, L. Lavoute, and W. J. Wadsworth, “Low-noise, high-brightness, tunable source of picosecond pulsed light in the near-infrared and visible,” Opt. Express 19(25), 25337–25345 (2011). [CrossRef]  

16. S. Lefrancois, D. Fu, G. R. Holtom, L. Kong, W. J. Wadsworth, P. Schneider, R. Herda, A. Zach, X. S. Xie, and F. W. Wise, “Fiber four-wave mixing source for coherent anti-Stokes Raman scattering microscopy,” Opt. Lett. 37(10), 1652–1654 (2012). [CrossRef]   [PubMed]  3382054

17. G. A. Ball and W. W. Morey, “Compression-tuned single-frequency Bragg grating fiber laser,” Opt. Lett. 19(23), 1979–1981 (1994). [CrossRef]   [PubMed]  

18. D. G. Steel, R. C. Lind, J. F. Lam, and C. R. Giuliano, “Polarization rotation and thermal motion studies via resonant degenerate four-wave-mixing,” Appl. Phys. Lett. 35(5), 376–379 (1979). [CrossRef]  

19. G. Bergner, S. Chatzipapadopoulos, D. Akimov, B. Dietzek, D. Malsch, T. Henkel, S. Schlücker, and J. Popp, “Quantitative CARS microscopic detection of analytes and their isotopomers in a two-channel microfluidic chip,” Small 5(24), 2816–2818 (2009). [CrossRef]   [PubMed]  

Cited By

Optica participates in Crossref's Cited-By Linking service. Citing articles from Optica Publishing Group journals and other participating publishers are listed here.

Alert me when this article is cited.


Figures (10)

Fig. 1
Fig. 1 Schematic setup of the seeded fiber optical parametric amplifier. Pump and seed will be coupled into a PCF with a length of 0.35m. Labels: BP - band pass, DM - dichroic mirror, HWP - half-wave plate. The graphs below show the characteristics of the incident laser pulses: (a) spectrum of the Ti:Sa seed signal; (b) spectrum of the fiber laser pump; (c) temporal shape of the pulsed pump
Fig. 2
Fig. 2 Simulation results for a 0.35m long 5μm core PCF pumped with 2.5kW. (a) shows the signal and idler wavelength position for the gain maximum as a function of the pump wavelength. The coloured areas illustrate the theoretical FWHM bandwidth of each band. (b) shows the calculated frequency separation between signal and pump or signal and idler as a function of the pump wavelength. The blue boxes illustrate the theoretically accessible Raman excitation regions by mechanically tuning the internal FBG by ±3nm (yellow box).
Fig. 3
Fig. 3 (a) Spectrum measured at the output of the PCF for three cases: high pump power without seed (light gray), high pump power with 5mW seed (dark gray) and lower pump power with 5mW seed (red). In the first and in the last case the signal output power was the same (33mW). The spectra below show (b) signal (including unamplified seed spectrum), (c) pump and (d) idler for the last case in more detail. The frequency separation between signal and pump is within the so-called fingerprint region. This frequency separation can be increased by a factor of two when using signal and idler.
Fig. 4
Fig. 4 (a) Measured OPG spectrum (350mW pump power) (dashed line, gray area) and measured idler power (300mW pump power, 5mW seed power) at the output of the PCF for different signal seed wavelengths (red circles) with respective Gaussian fit (red). Depending on the chosen wavelength pair, two excitation frequency bands (lower and upper abscissa) are addressable. (b) OPG spectra for different pump wavelengths and 350mW pump power.
Fig. 5
Fig. 5 OPG: temporal pulse shape of DFWM pump (black, red) and signal (orange) in case of optical parametric generation (no seed). OPA: temporal pulse shape of DFWM pump (black, red) and signal (orange) in case of optical parametric amplification of a 5mW cw seed signal.
Fig. 6
Fig. 6 (a) Power of the amplified signal (cw background subtracted) over incident seed power. (b) Development of the temporal signal pulse shape with increasing seed. (c) Development of the signal output spectrum with increasing pump power and constant 5mW seed power.
Fig. 7
Fig. 7 Raman and CARS intensity of a mixture of n-hexane and toluol as a function of the wavenumber. The left panel displays the deformation vibration spectrum around 1430cm−1 belonging to n-hexane, whereas the right one corresponds to the CH stretching around 2900cm−1. All curves are normalised to their maxima.
Fig. 8
Fig. 8 Schematic setup of the CARS microscope. L: achromatic lenses, LP: long pass filter at 850nm, BP: band pass filter at 1040nm, SP: short pass filter at 900nm, MM: moveable xy-scanning mirror, PMT: photo multiplier tube.
Fig. 9
Fig. 9 Multimodal nonlinear microscopic image of an atherosclerotic plaque deposition at a human artery wall. In panel (a) the combined CARS microscopic images acquired at the aliphatic methylene stretching vibration at 2845cm−1 (blue) and the methyl stretching vibration at 2930cm−1 (green) for imaging the lipid and protein distribution are contrasted to the combined TPEF and SHG signal of collagen and elastin (red). Lipids are coloured turquoise due to their high CH2/CH3, while proteins appear greenish. The single channels are separately displayed in panels (b)–(d). The images of the area of 450 × 450 μm2 were acquired (2048 × 2048 pixels) illuminating the sample with 30mW of laser power.
Fig. 10
Fig. 10 Line intensity profiles across the selected region of the sample for vibrationally resonant and off-resonant excitation.
Select as filters


Select Topics Cancel
© Copyright 2024 | Optica Publishing Group. All rights reserved, including rights for text and data mining and training of artificial technologies or similar technologies.