4472
J. Phys. Chem. B 2006, 110, 4472-4479
Quantitative Multiplex CARS Spectroscopy in Congested Spectral Regions Hilde A. Rinia,† Mischa Bonn,‡ and Michiel Mu1 ller*,† Swammerdam Institute for Life Sciences, UniVersity of Amsterdam, P.O. Box 94062, 1090 GB Amsterdam, The Netherlands, and FOM Institute for Atomic and Molecular Physics (AMOLF), Kruislaan 407 Amsterdam, The Netherlands ReceiVed: NoVember 9, 2005; In Final Form: January 9, 2006
A novel procedure is developed to describe and reproduce experimental coherent anti-Stokes Raman scattering (CARS) data, with particular emphasis on highly congested spectral regions. The approach, exemplified here with high-quality multiplex CARS data, makes use of spontaneous Raman scattering results. It is shown that the underlying vibrational Raman response can be retrieved from the multiplex CARS spectra, so that the Raman spectrum can be reconstituted, provided an adequate signal-to-noise ratio (SNR) is present in the experimental data and sufficient a priori knowledge of the vibrational resonances involved exists. The conversion of CARS to Raman data permits a quantitative interpretation of CARS spectra. This novel approach is demonstrated for highly congested multiplex CARS spectra of adenosine mono-, di-, and triphosphate (AMP, ADP, and ATP), nicotinamide adenine dinucleotide (NAD+), and small unilamellar vesicles (SUVs) of 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC). Quantitative determination of nucleotide concentrations and composition analysis in mixtures is demonstrated.
Introduction Recently, coherent anti-Stokes Raman scattering (CARS) microscopy has received significant attention in areas as diverse as biophysics, biology, and the material sciences.1-3 This nonlinear optical analogue of spontaneous Raman scattering provides vibrational contrast, and thus chemical specificity, with excellent spatial resolution. Whereas spontaneous Raman scattering generally suffers from low signal levels which are readily overwhelmed by luminescence from the sample, the nonlinear and coherent interaction in CARS increases the signal levels by several orders of magnitude and permits straightforward discrimination from background luminescence. Unique to vibrational spectroscopy is the high level of specificity of the information contained in the spectra, in terms of both chemistry and the physics of the sample, without the need of labeling, with, for example, fluorescent chromophores. Many of the important vibrational spectral features remain resolvable even at room temperature and in complex samples such as live cells.4,5 A requirement to utilize the full potential of CARS microscopy and spectroscopy is the acquisition of spectra with a sufficiently high signal-to-noise ratio. Most CARS microscopy applications (e.g., refs 6-10) are based on two picosecond lasers with a difference in frequency that matches a particular vibrational mode. This mode of operation, which will be referred to as “single frequency CARS” in the following, optimizes signal generation efficiency at one particular Raman shift frequency. A clear advantage of this approach is that it permits very rapid image acquisition in CARS microscopy.11,12 To obtain a full CARS spectrum, however, one of the lasers has to be tuned relative to the other. Not only is this method of acquiring spectral information relatively time-consuming, but also, power, pulse duration, and spectral fluctuations, as well as timing jitter * Corresponding author. E-mail:
[email protected]. † University of Amsterdam. ‡ FOM Institute for Atomic and Molecular Physics (AMOLF).
between the pulses from the two lasers, all negatively affect the signal-to-noise ratio (SNR) of the CARS spectrum acquired using the single frequency CARS approach. In contrast, multiplex CARS microscopy13-15 allows the measurement of a significant part of the vibrational spectrum simultaneously, by using a combination of a broad-band laser and a narrow-band laser. Since in this mode of operation every pair of laser pulses contributes to the whole spectrum, the SNRs of the acquired multiplex spectra have been demonstrated to be limited only by Poisson noise.16,17 This yields CARS spectra of unprecedented quality that permit detailed analysis in terms of both chemistry15 and the physical state of the sample.14,17,18 Multiplex CARS has recently also been used to study the vibrational structure of protein complexes under resonant excitation conditions using a polarization sensitive multiplex CARS scheme.19 Moreover, several novel approaches have recently been developed to implement broad-band multiplex CARS.20-22 CARS spectra generally have complex shapes due to the coherent addition of both resonant contributions from different vibrational modes and a nonresonant (NR) background contribution. The latter can be reduced in whole or in part by various methods including the use of specific polarization conditions,7,23 time-resolved measurements,10 a combination of phase and polarization control,24-26 and, in the case of thin samples, epidetection of the CARS signal.8,27 In most cases, however, the coherent addition of resonant and nonresonant contributions results in complex line shapes where vibrational resonances show apparent frequency shifts and differences in relative peak amplitude compared to the spontaneous Raman spectrum. As such, the coherent nature of the multiplex CARS signal prevents the direct determination of the concentration of the individual components. Nonetheless, most notably since the CARS signal can be measured relative to a NR reference signal, the signal-to-noise ratio for multiplex CARS is generally sufficient enough to be
10.1021/jp0564849 CCC: $33.50 © 2006 American Chemical Society Published on Web 02/07/2006
Quantitative Multiplex CARS Spectroscopy
J. Phys. Chem. B, Vol. 110, No. 9, 2006 4473
Figure 1. Schematic representation of the experimental setup for multiplex CARS spectroscopy. The pump/probe and Stokes lasers emit 10 ps and 90 fs pulses, respectively (as sketched in the temporal intensity profile inset). Symbols used: Q1 and Q2, achromatic λ/2 waveplates; P1 and P2, polarizers; F1, 710 nm Notch filter; F2, 710 nm short-wave pass filter. The tandem cuvette consists of two separate compartments for CARS measurement of both the sample and a nonresonant reference.
related quantitatively to the concentration of the species of interest. Actually, the interference of the resonant signal with the NR contribution increases the detection sensitivity through a homodyne type amplification of the resonant contribution. Indeed, multiplex CARS spectroscopy28-30 is well established for the measurement of dynamic termperature and composition distributions in combustion environments and plasma (see, e.g., refs 31-36). In this case, generally well-separated vibrational resonances are analyzed and concentration quantities are retrieved from the measured CARS spectra by comparison with calculations.37,38 However, such a quantitative analysis is no longer straightforward when different vibrational resonances become entangled and overlap in the spectral region of interest, even for the high-quality data obtained with multiplex CARS. In this paper, we address the issue of retrieving quantitative information from multiplex CARS spectra, under electronically nonresonant conditions, of various biologically relevant species (ATP, NAD+, and POPC) in water, that show characteristically congested vibrational spectra. We show that a quantitative determination of solute concentrations is possible down to the millimolar level. The complex spectra can be interpreted using fitting procedures that allow retrieval of the Raman response and reconstitution of the corresponding spontaneous Raman scattering spectra. This in turn permits the quantitative analysis of solute mixtures. We discuss in detail both the potentials and challenges encountered in extracting the full information that is inherently present in multiplex CARS spectroscopic data. Experimental and Theoretical Methods Multiplex CARS Spectroscopy. In multiplex CARS, a significant range of vibrational frequencies is addressed simultaneously through a four-wave-mixing process of narrow-band pump (pu) and probe (pr) lasers with a broad-band Stokes (S) laser. The bandwidth of the generated anti-Stokes (as) signal at frequency ωas ) ωpu - ωS + ωpr is determined by the Stokes spectral intensity profile, whereas the inherent spectral resolution is determined by the pump/probe bandwidth (see refs 14 and 15 for more details). In practice, the pump and probe are derived from the same laser. The experimental setup, which is schematically depicted in Figure 1, has been described in detail elsewhere.18 Briefly, two tunable mode-locked Ti:sapphire lasers (Tsunami, Spectra Physics) with a repetition rate of 82 MHz are made collinear and synchronized (“lok-to-clock”, Spectra Physics). An additional home-built, long-term feedback system ensures a timing
Figure 2. Energy level diagram depicting the resonant (a) and nonresonant (b) contribution to the multiplex CARS signal. The thick solid line represents the electronic ground state, and the thin solid lines indicate the vibrational levels of different vibrational modes. The various frequency contributions to the anti-Stokes signal are weighted by the intensity profile of the broad-band Stokes laser, as indicated by the gray curve. The transitions indicated in part b are far away from oneor two-photon resonances with electronically excited states.
jitter between the lasers of
