We demonstrate highly sensitive infrared spectroscopy of sample volumes close to the diffraction limit by coupling a femtosecond fiber-feedback optical parametric oscillator (OPO) to a conventional Fourier-transform infrared (FTIR) spectrometer. The high brilliance and long-term stable infrared radiation with -bandwidths up to 125 nm is easily tunable between 1.4 μm and 4.2 μm at 43 MHz repetition rate and thus enables rapid and low-noise infrared spectroscopy. We demonstrate this by measuring typical molecular vibrations in the range of 3 μm. Combined with surface-enhanced infrared spectroscopy, where the confined electromagnetic near-fields of resonantly excited metal nanoparticles are employed to enhance molecular vibrations, we realize the spectroscopic detection of a molecular monolayer of octadecanethiol. In comparison to conventional light sources and synchrotron radiation, our compact table-top OPO system features a significantly improved performance making it highly suitable for rapid analysis of minute amounts of molecular species in life science and medicine laboratories.
© 2015 Optical Society of America
Infrared (IR) spectroscopy allows for unambiguous identification and chemical characterization of molecular species based on their infrared vibrational bands. Unfortunately, such material-specific absorption signals are relatively small, which hampers the spectroscopic detection of minute amounts of substances, e.g., in sensing applications. Also infrared hyperspectral chemical imaging of unstained biological tissue sections  with micrometer resolution as it is used in medicine to discriminate between healthy tissue and tumors , suffers from small absorption signals and small signal-to-noise ratio (SNR) of conventional IR light sources. This is only possible on reasonable time scales with immense technical effort, e.g. at synchrotron light sources providing 2–3 orders of magnitude brighter IR radiation than thermal light sources [3, 4].
One way to overcome these limitations is the use of specially designed metal nanoparticles known as resonant plasmonic nanoantennas . Resonantly excited in the infrared, such nanoantennas confine infrared radiation on the nanometer scale and therefore provide locally high electromagnetic field intensities. Following this approach of surface-enhanced infrared absorption (SEIRA), vibrational signals of molecules located in the plasmonic hotspots can be enhanced up to five orders of magnitude in comparison to conventional infrared spectroscopy [6–8].
Another approach to achieve high intensities and thus enable broadband Fourier-transform infrared (FTIR) spectroscopy of minute sample volumes in reasonable accumulation times is the use of advanced light sources of excellent brilliance , defined as the number of photons per second, per unit source area, per unit solid angle and per 0.1% of the central frequency:3,10]. However, synchrotron light sources are extremely cost-intensive due to the high consumption of energy and deliver only a few mW of radiation power in the IR . Additionally, beam-time is rarely available. Besides these incoherent broadband sources, IR laser sources with narrow bandwidth such as CO2 lasers and quantum cascade lasers (QCLs) [11, 12] are available. Combined with SEIRA such QCLs provide an increased sensitivity in comparison to conventional infrared spectroscopy as recently shown in . Due to their limited bandwidth, sweeping techniques or even several lasers emitting at different wavelengths have to be applied to cover the infrared spectral range of interest, which enable sensitive diffraction-limited IR spectroscopy . However, also the generation of intrinsically broadband and tunable femtosecond radiation with mode-locked laser systems has been shown to be difficult due to the lack of suitable gain media and significant atmospheric absorption in the mid-infrared spectral region . Furthermore, difference frequency generation (DFG) has been employed to generate very broadband infrared supercontinua . Another approach involves supercontinuum generation in nonlinear fibers or waveguides, where recently a brilliance of 8 × 1020 ph/s/mm2/sr/0.1%BW has been reached, beating synchrotron sources by 2–3 orders of magnitude . The drawback of such super-continua, however, is their sensitivity to quantum noise in the process of spectral broadening, which significantly reduces the temporal coherence and leads to excess noise in spectroscopic precision measurements .
Optical parametric light sources, such as optical parametric oscillators (OPO) or amplifiers (OPA), provide coherent and easily tunable, extremely brilliant infrared radiation. In these devices, near-infrared femtosecond radiation is converted in nonlinear crystals to the mid-infrared spectral range. With bandwidths of several tens of nanometers in the mid-infrared, individual molecular vibrations can be easily measured, while the continuous tunability ensures the versatility to access other spectral regions.
An important difference between incoherent and coherent sources is the fundamentally different nature of the noise characteristics. Incoherent sources are almost free from intensity noise, while coherent sources exhibit significant intensity noise on different time scales. In particular the sub-millisecond time scale as well as the long term drift properties are relevant for FTIR spectroscopy. Drifts in the intensity or wavelength will directly distort the baseline making it impossible to normalize spectra with reference measurements, while sub-millisecond fluctuations will perturb the acquisition of a single spectrum and will transform into random noise in the spectrum. These disadvantages are quite challenging and have so far hampered the use of coherent sources for FTIR spectroscopy, despite their great advantages in terms of brilliance.
In the present work, we overcome these limitations by employing a continuously tunable fiber-feedback OPO emitting highly brilliant and stable IR radiation with bandwidths up to 125 nm in the wavelength range between 1.4 μm and 4.2 μm. This radiation is coupled into a conventional FTIR spectrometer and subsequently into an attached microscope, selecting μm2-sized areas for infrared spectroscopy [Fig. 1]. To demonstrate highly sensitive and low-noise IR spectroscopy, we investigated micrometer-sized volumes of a polymer [Fig. 1(a)] and employed SEIRA to spectroscopically detect even smaller numbers of molecules [Fig. 1(b)]. The results are compared to FTIR spectroscopy with conventional thermal and synchrotron light sources. We further give a quantitative measure for the fiber-feedback OPO noise performance on different relevant time scales.
2. Experimental setup
The high-brilliance infrared source for FTIR spectroscopy is realized by coupling a fiber-feedback OPO  into an infrared microscope, which is attached to a conventional FTIR spectrometer [Fig. 2(a)].
The OPO is pumped by an Yb:KGW solid-state oscillator that provides 7 W average power at 1040 nm with a repetition rate of 43 MHz and a pulse duration of 450 fs . The nonlinear frequency conversion takes place in a 10 mm long, temperature controlled, periodically-poled lithium niobate (PPLN) crystal, which is pumped by a 2 – 4 W fraction of the pump laser. The generated signal is collimated with a curved mirror (ROC 150 mm). By employing a half-wave plate and a polarizing beam splitter, a variable share of this signal is fed back into a 4.2 m long SMF-28 fiber that contains most of the intra-cavity optical path length, making the setup extremely compact. Furthermore, the pulses are broadened temporally due to group velocity dispersion in the glass, which leads to a highly stretched feedback pulse. This pulse is delayed in order to be synchronized with the next pump pulse and act as seed for the subsequent parametric amplification step. The high chirp on the feedback pulse ensures both extremely precise wavelength tuning by matching a certain delay as well as excellent wavelength stability, which is further investigated later in this paper. Despite the significant chirp on the feedback, the time-bandwidth product of the emitted signal pulse is on the order of 0.42–0.45 in the whole signal tuning range and can be even lower if low pump power is used. The spectroscopically useful bandwidth ( ) is on the order of 20 nm at 1.4 μm and increases to 125 nm towards 4 μm. In wavenumbers, this corresponds to approximately 80 cm−1 over the whole tuning range. Including temperature tuning and changing of the poling period of the multi-channel PPLN crystal from 27.91–31.59 μm, a gap free tuning range from 1.38–2.03 μm as well as 2.12–4.17 μm is achieved [Fig. 2(b) and 2(c)]. Depending on the actual wavelength, the typical power level throughout the tuning range exceeds several hundreds of mW, which is absolutely sufficient for spectroscopic applications, where typically a few mW are required. The system reaches a brilliance of 3 × 1022 ph/s/mm2/sr/0.1%BW in the idler output, measured at 3.5 μm, which is about four orders of magnitude greater than that of synchrotron radiation. The tuning range of signal and idler would allow us to extend the spectral range by employing a DFG-scheme  into the 5–20 μm region with mW power levels.
The high brilliance IR radiation generated with the described fiber-feedback OPO is coupled to a commercially available FTIR spectrometer (Bruker Vertex 80) by means of mirror optics. Subsequently, the IR beam is sent in an infrared microscope (Bruker Hyperion 2000) through a KRS-5 polarizer and focused to the sample by a 36× Cassegrain objective (NA = 0.52). An aperture is inserted after the sample to select μm2-sized areas. The transmitted IR radiation is detected with a liquid nitrogen cooled 250×250 μm2 mercury-cadmium-telluride (MCT) detector. For the synchrotron studies a similar setup is used at the IR2 beamline at the Angström source Karlsruhe (ANKA). Infrared radiation provided by electrons in the storage ring (2.5 GeV and a beam current of 120 mA) is coupled into an infrared microscope (Bruker IRscope II) attached to an FTIR spectrometer (Bruker IFS 66 V/S). A circular aperture in the microscope enables us to select areas down to approximately 50 μm2 from the natural spot size of the synchrotron beam (diameter of approximately 20 μm). All infrared spectra presented in the present study are recorded in transmission geometry with a typical accumulation time of 12 seconds and a resolution of 2 cm−1 with 20 kHz sampling rate unless specified otherwise.
3. Vibrational IR spectroscopy
To demonstrate the capability of our fiber-feedback OPO laser source for conventional FTIR spectroscopy, we performed surface-enhanced infrared spectroscopy using resonantly excited gold nanoantennas covered with a molecular monolayer of ODT (thickness 2.8 nm). Employing standard PMMA resist electron beam lithography and lift-off techniques, we fabricated 100×100 μm2-sized arrays of rectangularly shaped nanostructures (length L = 800 nm, height h = 50 nm, width w = 60 nm) as well as single nanoantennas (L = 900 nm, h = 60 nm, w = 20 nm). The spacings between the periodically arranged nanoantennas are 40 nm in direction along the long antenna axis and 1.5 μm perpendicular to it. Details on the selective ODT adsorption on gold surfaces are given in [6, 23]. Excited with light polarized parallel to the long antenna axis (E‖), the infrared transmittance of such gold nanoantenna arrays (normalized to the bare CaF2 substrate) reveals a broadband plasmonic response and enhanced molecular vibrations of ODT (3.425 and 3.51 μm) as seen in the globar measurements [Fig. 3(a)]. Due to the resonant nature of the plasmonic enhancement, SEIRA signals with Fano-type line-shapes only occur for a good spectral match of the plasmonic and molecular excitations, which is not given for perpendicular polarization (E⊥).
However, even for parallel polarization the enhanced molecular vibrations are hardly visible due to the low brilliance of the globar source, even for long accumulation times (t = 20 min) and relatively large apertures (20×20 μm2). In contrast to that, the relative transmittance recorded with the fiber-feedback OPO provides a significantly improved SNR in a 50 times reduced acquisition time as seen for the asymmetric [Fig. 3(b)] and symmetric [Fig. 3(c)] CH2 stretching vibrations of the ODT molecules . Due to the limited spectral bandwidth of the OPO radiation, which is however sufficient to investigate individual IR vibrational bands, the laser emission wavelength has to be tuned to the respective CH2 vibration. This reliable tuning is achieved by temperature adjustment of the PPLN crystal and/or cavity tuning of the OPO feedback in less than 2 min and can be easily automatized.
The brilliance of the respective light source becomes even more crucial for infrared spectroscopy with aperture sizes close to the diffraction limit, e.g., SEIRA using single nanoantennas. Such measurements of single nanostructures with thermal light sources are challenging and synchrotrons with 2–3 orders of magnitude brighter IR radiation have to be used [6,22]. As with our OPO setup, the synchrotron light is coupled to an infrared microscope, which allows the selection of μm2-sized areas for relative transmittance spectroscopy of an ODT-covered single gold nanoantenna [Fig. 3(d)]. In agreement with SEIRA using nanostructure arrays, the asymmetric ODT vibration is enhanced by the plasmonic near-field of the resonantly excited antenna and shows the expected polarization dependence. For synchrotron radiation [Fig. 3(e)], the enhanced vibrational signal is slightly larger than the noise, while in the OPO experiments the vibration is clearly visible [Fig. 3(f)]. Spectral differences in the background of the OPO and synchrotron measurements arise from a nonlinear response of the MCT detector, which becomes more pronounced for intensities close to the saturation level as present in the OPO measurements. Both measurements are performed on the very same nanoantenna with similar spot sizes (≈ 50 μm2), but a ten times longer acquisition time for the synchrotron measurements, thus demonstrating the impact of excellent brilliance and stability of the laser beam.
This high quality beam in combination with SEIRA enables low-noise detection of about 10,000 ODT molecules located in the hotspots of a single nanoantenna. We estimated this number assuming an homogeneous ODT adsorption on gold surfaces with an area of 21.6 Å2 per molecule  and a SEIRA active area of two times the end faces (20 nm × 60 nm) of the gold antenna according to . To enable spectroscopy below the diffraction limit, our approach could be combined with a scanning near-field optical microscope [26–28].
In microscopic IR spectroscopy typical aperture sizes range from 100×100 μm2 down to the diffraction limit, enabling chemical imaging with micrometer-scale resolution, e.g., of biological tissue sections . However, particularly at small aperture sizes this method suffers from the low brilliance of conventional thermal light sources. As a consequence, the intensity acquired with a standard MCT detector decreases rapidly with shrinking aperture size (see Fig. 4(b), black curves). In contrast, the decrease of the OPO intensity is much less pronounced due to the excellent beam profile. Moreover, the nitrogen cooled MCT detector can always be used at the optimum power level close to saturation by exploiting the 107 times higher brilliance of the OPO compared to the globar. As a further consequence of the high brilliance and excellent stability, especially on the sub-millisecond time scale, the OPO RMS noise level remains almost constant with varying aperture size, whereas the noise level of the globar increases by orders of magnitude. This effect is illustrated in Fig. 4(c), where spectrally resolved stability measurements in air are shown. The presented relative transmittance spectra are calculated from two subsequent measurements with typical acquisition times of 12 s each. For larger apertures, both globar and fiber-feedback OPO feature a similar RMS noise level whereas for smaller apertures the relative transmission remains smooth for the OPO. However, the RMS noise level of the OPO changes slightly over time, which arises from spectral and intensity drifts on the time scale of seconds and thereby influences successive measurements. The underlying mechanisms are discussed in the introduction. The same behavior is observed in a representative infrared absorption measurement of a 5 μm thick photoresist layer (XARP 3100/10, purchased from Allresist) spincoated on a CaF2 substrate . As depicted in Fig. 4(d) and 4(e), the relative transmittance of a 40×40 μm2 area reveals a clearly visible vibrational feature at 3.31 μm for both sources. For decreasing apertures, however, the vibrational signal is below the noise level in the globar experiments, but remains unchanged in the OPO studies. This confirms the ability of our fiber-feedback OPO to acquire high quality and low noise spectra at the diffraction limit, thus enabling rapid spectroscopic imaging applications with resolution only limited by the diffraction of light. Furthermore, the high power in principle allows the use of low-cost deuterated triglycine sulfate (DTGS) detectors operating at room temperature. In contrast to nitrogen-cooled MCT detectors, DTGS detectors feature a lower sensitivity, but do not suffer from undesired nonlinearities. Such nonlinearities lead to slightly different absolute absorption values in our spectra in comparison to the globar measurements if the detector is exposed to high intensities close to the saturation level.
In more detail, the long term performance of the OPO is evaluated by a series of measurements over 15 min in air, where we investigate the spectroscopically relevant properties of the mid-IR beam [Fig. 5]. We constantly monitor the OPO spectrum with the FTIR detector at one frame every 12 s in order to detect long term drifts that would detrimentally impact spectroscopic measurements by changing the baseline. The analysis is performed by fitting the measured spectrum with a Gaussian peak and extracting the peak intensity, center wavelength and spectral full width at half maximum (FWHM) bandwidth. Fluctuations in peak intensity as low as 0.23% RMS are achieved. Excellent spectral stability with a variance of less than 0.09 nm in center wavelength and about 0.02 nm in spectral bandwidth is observed. There are almost no drift features visible. These values are on the level of the near-infrared solid-state pump laser, and, with our tunable OPO, available over a broad spectral range.
5. Conclusion and outlook
We demonstrate ultra-sensitive infrared spectroscopy near the diffraction limit by coupling a high-brilliance fiber-feedback OPO light source to a conventional FTIR spectrometer. Combining the setup with surface-enhanced infrared spectroscopy, we detect approximately 10,000 ODT molecules attached to a resonantly excited single gold nanoantenna in a ten times faster acquisition time compared to synchrotron light sources. This excellent performance is enabled by the four orders of magnitude higher brilliance as well as the low-noise nature of the tabletop OPO with respect to both the sub-millisecond time scale as well as the long term drift.
In order to access the molecule-specific fingerprint region, which is relevant in almost all spectroscopic applications, we plan to extend the accessible spectral range to 5–12 μm. This will enable high speed infrared spectroscopy in chemistry, biology, medicine and pharmaceutical science even in standard laboratories and will allow for broadband spectroscopic scanning near-field optical microscopy.
The authors gratefully acknowledge financial support by ERC Advanced Grant COMPLEX-PLAS as well as DFG, BMBF, GIF, Carl Zeiss foundation, and MWK Baden-Württemberg and the Baden-Württemberg Stiftung (PROTEINSENS, Spitzenforschung II). This work was supported by the German Research Foundation (DFG) within the funding program Open Access Publishing. We thank D. Floess for nanofabrication and J. Weis for cleanroom access. We acknowledge the Synchrotron Light Source ANKA for provision of instruments at the IR2 beamline and we would like to thank D. Moss, Y-L. Mathis and M. Süpfle for their assistance. We further thank Mr. G. Zachmann (Bruker Optik GmbH) for fruitful discussions.
References and links
1. M. Diem, M. Romeo, S. Boydston-White, M. Miljkovic, and C. Matthaus, “A decade of vibrational micro-spectroscopy of human cells and tissue (1994–2004),” Analyst. 129, 880–885 (2004). [CrossRef] [PubMed]
3. C. Petibois, G. Deleris, M. Piccinini, M. Cestelli-Guidi, and A. Marcelli, “A bright future for synchrotron imaging,” Nature Photon. 3, 179 (2009). [CrossRef]
4. M. Martin, U. Schade, P. Lerch, and P. Dumas, “Recent applications and current trends in analytical chemistry using synchrotron-based Fourier-transform infrared microspectroscopy,” Trends Anal. Chem. 29, 453–460 (2010). [CrossRef]
5. L. Novotny and N. van Hulst, “Antennas for light,” Nature Photon. 5, 83–90 (2011). [CrossRef]
6. F. Neubrech, A. Pucci, T. W. Cornelius, S. Karim, A. Garcia-Etxarri, and J. Aizpurua, “Resonant plasmonic and vibrational coupling in a tailored nanoantenna for infrared detection,” Phys. Rev. Lett. 103, 157403 (2008). [CrossRef]
7. R. Adato and H. Altug, “In-situ ultra-sensitive infrared absorption spectroscopy of biomolecule interactions in real time with plasmonic nanoantennas,” Nat. Commun. 4, 2154 (2013). [CrossRef] [PubMed]
9. P. Willmott, An Introduction to Synchrotron Radiation: Techniques and Applications (John Wiley & Sons, 2011). [CrossRef]
10. P. Dumas, F. Polack, B. Lagarde, O. Chubar, J. Giorgetta, and S. Lefrancois, “Synchrotron infrared microscopy at the French synchrotron facility SOLEIL,” Infrared Phys. Technol. 49, 152–160 (2006). [CrossRef]
11. Y. Yao, X. Wang, J.-Y. Fan, and C. F. Gmachl, “High performance ’continuum-to-continuum’ quantum cascade lasers with a broad gain bandwidth of over 400 cm−1,” Appl. Phys. Lett. 97, 081115 (2010). [CrossRef]
12. Y. Yao, A. J. Hoffman, and C. F. Gmachl, “Mid-infrared quantum cascade lasers,” Nature Photon. 6, 432–439 (2012). [CrossRef]
13. A. Hasenkampf, N. Kröger, A. Schönhals, W. Petrich, and A. Pucci, “Sensitive trace gas detection with cavity enhanced absorption spectroscopy using a continuous wave external-cavity quantum cascade laser,” Opt. Express 23, 5670–5680 (2015). [CrossRef] [PubMed]
14. P. Bassan, M. J. Weida, J. Rowlette, and P. Gardner, “Sensitive trace gas detection with cavity enhanced absorption spectroscopy using a continuous wave external-cavity quantum cascade laser,” Appl. Phys. Lett. 103, 131114 (2014).
15. S. Mirov, V. Fedorov, I. Moskalev, D. Martyshkin, and C. Kim, “Progress in Cr2+ and Fe2+ doped mid-IR laser materials,” Laser & Photon. Rev. 1, 21–41 (2010). [CrossRef]
16. S. Amarie and F. Keilmann, “Broadband-infrared assessment of phonon resonance in scattering-type near-field microscopy,” Phys. Rev. B 83, 045404 (2011). [CrossRef]
17. Y. Yu, X. Gai, P. Ma, D.-Y. Choi, Z. Yang, R. Wang, S. Debbarma, S. J. Madden, and B. Luther-Davies, “A broadband, quasi-continuous, mid-infrared supercontinuum generated in a chalcogenide glass waveguide,” Laser Photon. Rev. 8, 792–798 (2014). [CrossRef]
18. G. Genty, S. Coen, and J. M. Dudley, “Fiber supercontinuum sources,” J. Opt. Soc. Am. B 24, 1771–1785 (2007). [CrossRef]
19. T. Südmeyer, J. Aus der Au, R. Paschotta, U. Keller, P. G. R. Smith, G. W. Ross, and D. C. Hanna, “Femtosecond fiber-feedback optical parametric oscillator,” Opt. Lett. 26, 304–306 (2001). [CrossRef]
20. A. Steinmann, B. Metzger, R. Hegenbarth, and H. Giessen, “Compact 7.4 W femtosecond oscillator for white-light generation and nonlinear microscopy,” in Proceedings of ClEO 2011, (Optical Society of America, 2011), CThAA5.
21. R. Hegenbarth, A. Steinmann, Sergey Sarkisov, and H. Giessen, “Milliwatt-level mid-infrared difference frequency generation with a femtosecond dual-signal-wavelength optical parametric oscillator,” Opt. Lett. 37, 3513–3515 (2012). [CrossRef] [PubMed]
22. C. Huck, F. Neubrech, J. Vogt, A. Toma, D. Gerbert, J. Katzmann, T. Hartling, and A. Pucci, “Surface-enhanced infrared spectroscopy using nanometer-sized gaps,” ACS Nano 8, 4908–4914 (2014). [CrossRef] [PubMed]
23. F. Schreiber, “Structure and growth of self-assembling monolayers,” Prog. Surf. Sci. 65, 151–257 (2000). [CrossRef]
24. D. Lin-Vien, N. W. Colthup, W. G. Fateley, and J. G. Grasselli, The Handbook of Infrared and Raman Frequencies of Organic Molecules (John Wiley & Sons, 1991).
25. C. D’Andrea, J. Bochterle, A. Toma, C. Huck, F. Neubrech, E. Messina, B. Fazio, O. M. Marag, E. Di Fabrizio, M. Lamy de La Chapelle, P. G. Gucciardi, and A. Pucci, “Optical nanoantennas for multiband surface-enhanced infrared and raman spectroscopy,” ACS Nano 7, 3522–3531 (2013). [CrossRef]
26. R. Hegenbarth, A. Steinmann, S. Mastel, S. Amarie, A. Huber, R. Hillenbrand, S. Sarkisov, and H. Giessen, “High-power femtosecond mid-IR sources for s-SNOM applications,” J. Opt. 16, 094003 (2014). [CrossRef]
27. S. Bensmann, F. Gaussmann, M. Lewin, J. Wuppen, S. Nyga, C. Janzen, B. Jungbluth, and T. Taubner, “Near-field imaging and spectroscopy of locally strained GaN using an IR broadband laser,” Opt. Express 22, 22369–22381 (2014). [CrossRef] [PubMed]
28. X. G. Xu, M. Rang, I. M. Craig, and M. B. Raschke, “Pushing the sample-size limit of infrared vibrational nanospectroscopy: from monolayer toward single molecule sensitivity,” J. Phys. Chem. Lett. 3, 1836–1841 (2012). [CrossRef]
29. S. Bagheri, H. Giessen, and F. Neubrech, “Large-area antenna-assisted SEIRA substrates by laser interference lithography,” Adv. Opt. Mater. 11, 1050–1056 (2014). [CrossRef]