Wavelength dependent resonance Raman band intensity of broadband stimulated Raman spectroscopy of malachite green in ethanol Qiongyan Cen, Yuhan He, Mei Xu, Jingjing Wang, and Zhaohui Wang Citation: The Journal of Chemical Physics 142, 114201 (2015); doi: 10.1063/1.4914188 View online: http://dx.doi.org/10.1063/1.4914188 View Table of Contents: http://scitation.aip.org/content/aip/journal/jcp/142/11?ver=pdfcov Published by the AIP Publishing Articles you may be interested in Two-dimensional stimulated resonance Raman spectroscopy of molecules with broadband x-ray pulses J. Chem. Phys. 136, 174117 (2012); 10.1063/1.4706899 Lineshapes for resonant impulsive stimulated Raman scattering with chirped pump and supercontinuum probe pulses J. Chem. Phys. 129, 184504 (2008); 10.1063/1.3009221 Polarization dependence of vibrational coupling signals in femtosecond stimulated Raman spectroscopy J. Chem. Phys. 127, 124501 (2007); 10.1063/1.2780843 Femtosecond broadband stimulated Raman spectroscopy: Apparatus and methods Rev. Sci. Instrum. 75, 4971 (2004); 10.1063/1.1807566 Stimulated resonance Raman scattering from epitaxially oriented crystals of biphenyl-capped thiophene Appl. Phys. Lett. 84, 4783 (2004); 10.1063/1.1760223
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 150.216.68.200 On: Thu, 23 Apr 2015 06:00:38
THE JOURNAL OF CHEMICAL PHYSICS 142, 114201 (2015)
Wavelength dependent resonance Raman band intensity of broadband stimulated Raman spectroscopy of malachite green in ethanol Qiongyan Cen ( 岑琼燕 ),a) Yuhan He ( 何玉韩 ),a) Mei Xu ( 徐媚 ), Jingjing Wang ( 王静静 ), and Zhaohui Wang ( 王朝晖 )b) State Key Laboratory of Physical Chemistry of Solid Surfaces, The MOE Key Laboratory of Spectrochemical Analysis and Instrumentation, Xiamen University, Xiamen 361005, China and Department of Chemistry, College of Chemistry and Chemical Engineering, Xiamen University, Xiamen 361005, China
(Received 29 December 2014; accepted 25 February 2015; published online 16 March 2015) Resonance broadband stimulated Raman spectroscopy of malachite green in ethanol has been performed. With a tuning picosecond visible laser source and a broadband Raman probe, the Raman gain and loss spectra have been measured simultaneously. By scanning the Raman pump across the first absorption band of the molecule, we found that the resonant Raman bands could be only seen when the pump laser tuned in the range of the red edge of the S1 ← S0 transition. Dispersive lineshapes of resonant Raman bands have been observed in the Raman loss spectra, while the line shape is normal (same as spontaneous Raman) in the Raman gain spectra. Although, the resonant bands in the loss spectrum are usually stronger than that in the gain spectrum, the band intensities of both loss and gain linearly increase with the pump energy. The relative magnitude of each corresponding resonant band in the Raman loss and gain varies with the pump wavelength. Mode specified Raman excitation profiles have been obtained through broadband stimulated Raman measurement. C 2015 AIP Publishing LLC. [http://dx.doi.org/10.1063/1.4914188]
INTRODUCTION
Raman spectroscopy is a powerful tool to reveal molecular vibrations and their signatures, enabling the accessibility of a large amount of information about molecular structures and properties, the magnitude of the bond strengths, the nature of the chemical environment, or even the characteristics of the electronic excited state.1–4 However, spontaneous Raman scattering is typically very weak and easily shielded by fluorescence and other luminescent processes. Limited by the uncertainty principle, spontaneous Raman has limited temporal resolution while hold adequate spectral resolution when used in the field of ultrafast spectroscopy.1,5,6 Recently, with the advancement in ultrafast laser techniques, nonlinear Raman spectroscopic methods have been developed with large enhancement in sensitivity (up to 106 times) and great improvement in temporal resolution (down to tens of femtoseconds) to study some of the most challenging fundamental problems in natural science. Femtosecond (fs, 10−15 s) Stimulated Raman Spectroscopy (FSRS) is one of the most powerful nonlinear Raman techniques to access short-lived intermediates of molecules in condensed phase.2–4,7–28 In FSRS, a fs actinic pulse is used to create population in the electronic excited state; at certain delay time, a picosecond (ps, 10−12 s) Raman pump and a broadband fs Raman probe pulse interact with the molecules simultaneously to generate the stimulated Raman signals. Without the actinic pulse, FSRS provides conventional Raman spectra, to be more specific, we will call FSRS without the a)Q. Cen and Y. He contributed equally to this work. b)Author to whom correspondence should be addressed. Electronic mail:
[email protected] 0021-9606/2015/142(11)/114201/7/$30.00
actinic pulse as Broadband Stimulated Raman spectroscopy (BB-SRS) there-in-after. In BB-SRS, stimulated Raman bands appear at both the red (Stokes) and blue (anti-Stokes) side of the Raman pump due to exchange of energy between the two incident pulses. Unlike the spontaneous Raman, the anti-Stokes side BB-SRS signal represents the same feature as the Stokes side2,3,18,19 and provides almost exactly the same information as spontaneous Raman when the Raman pump is set at off-resonance away from the sample electronic transitions. At resonant condition, in conventional Raman, the incident light is set to match the electronic transition, which can greatly enhance the scattering intensity compared to the spontaneous Raman scattering.29,30 The resonance Raman process is coupled with the electronic excited states. Therefore, the structure information about the excited state can be derived from resonance Raman spectroscopy. Theoretical and experimental results showed that the resonance Raman was mainly controlled by Frank Condon effect, while other interacting mechanisms, such as Duschinsky mixing and Herzberg-Teller effect, contribute to the resonance Raman intensity as well.29–31 However, in BB-SRS, resonant condition becomes much more complicated, the resonance Raman signals can be observed when either or both of the Raman pump and Raman probe are set in resonance with the electronic transition of the molecule. In condensed phase, both the Raman pump and probe will fall into the broad absorption band of the sample in most of the cases. Therefore, we should expect for different spectral features in BB-SRS due to the stimulated nature and the double resonance condition because of the extra probe pulse added. (Only if we put the Raman pump at the very edge of the absorption band of the sample, we may
142, 114201-1
© 2015 AIP Publishing LLC
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 150.216.68.200 On: Thu, 23 Apr 2015 06:00:38
114201-2
Cen et al.
avoid the double resonant complexity, then the enhancement from the resonance will be trivial.) Recently, the Raman pump was set to on-resonance to enhance the excited state SRS signal in FSRS studies.7–16 Tahara et al.10,12 used ultraviolet resonance SRS to follow the early time photodynamics of yellow protein and obtained stronger SRS signal than that of using the near IR Raman pump. Dobryakov et al. observed excited state SRS spectra of trans-azobenzene even without the actinic pulse.8 Resonance BB-SRS measurement of Rhodamine 6G by Mathies et al. showed dispersive lineshapes of the resonant bands in both SRS loss and Gain spectra.13,15,16 Umapathy et al. reported mode dependent dispersive lineshapes of crystal violet at the loss side of their BB-SRS.14 Lee et al. simulated resonance BB-SRS spectra with the consideration of resonance Raman stimulated effects between the two electronically resonant states and predicted the dispersive lineshapes and the overall spectral feature in very well consistence with experiments for both off-resonance and on-resonance BB-SRS.20,23 However, there is only very few reported experimental results about resonance BB-SRS, detailed experimental and theoretical analyses are needed for better understanding of the resonance effects in the BB-SRS spectra. In this paper, we report a systematic measurement of resonance BB-SRS of malachite green (MG) in ethanol. By varying the Raman pump wavelength across the S1 ← S0 transition of malachite green, along with a very broadband probe, we can measure the BB-SRS gain and loss spectra simultaneously. Hence, the properties of the resonance Raman band for each vibrational mode appearing in the BB-SRS loss and Gain spectra can be compared under exactly the same experimental condition.
METHOD
SRS spectroscopy has been thoroughly presented in the literature both experimentally and theoretically.2–4,7–28,32–41 We will briefly review the main aspects related to our work in this paper. In BB-SRS, we can observe Raman bands on both the lower energy side (Stokes) and higher energy side (anti-Stokes) with respect to the Raman pump. The signals on the Stokes side appear as gain (positive) of the Raman probe, usually called SRS gain. On the anti-Stokes side, the Raman signals emerge as loss (negative) of the Raman probe, known as SRS loss, as displayed in Fig. 1, respectively. The peak frequencies of the Raman bands both in SRS gain and loss agree with that in
J. Chem. Phys. 142, 114201 (2015)
FIG. 2. Four-wave mixing energy level diagram for SRS. (a) and (b) corresponding to the gain process, and (c) and (d) for the loss.
spontaneous Raman within the experimental error, and the SRS gain spectrum matches the spontaneous Raman closely. Although there are different terms for the loss spectra with consideration of contributions from different mechanisms in the literature, such as ultrafast Raman loss,14,32,33 inverse Raman,20 simulations and experiments showed that loss and gain signals are with equal magnitude for off-resonance excitation.16,18 The BB-SRS provides the same structure information as conventional spontaneous Raman spectroscopy under electronically off-resonance condition. Fig. 2 shows the four-wave mixing energy level diagram of the major stimulated processes which contribute to the SRS gain and loss signals. From the classical point of view, the two incident pulses, Raman pump and probe, interact with the sample in two steps: step 1, one of the interaction brings the molecules to an electronically upper state (imaginary if off-resonance), and the other interaction pulls the molecules back through stimulated emission and creates coherence in the ground state with the Raman active vibration modes. The coherence undergoes dephasing with a time constant characterized by each vibration mode. This step is an analog as IR absorption which populates the molecules to the vibrational excited states. Step 2, during the dephasing of the coherence generated in the first step, the third interaction from the Raman pump or probe reads out the coherence and generates SRS signal. The readout procedure is a sum frequency or difference frequency generation process. We only plotted diagrams of the main contribution mechanisms to the BB-SRS spectra under our experimental geometry. Detailed theoretical approaches can be found in Refs. 23–28. Even in these diagrams, although stimulated processes of both (a) and (b) will transfer energy from the Raman pump to the Raman probe and create Raman gain signal in the Stokes side of the BB-SRS spectrum, process (a) is the dominating mechanism because of Ipu/Ipr = ∼105 in our measurement. The same situation for (c) and (d), (d) is the major channel for the Raman loss signal. The SRS generation is coherent, and the phase match condition for the SRS Gain (Fig. 2(a)) and loss (Fig. 2(d)) can be written as SRS Gain : kpu − kpr − k pu = −ksig. ⇒ k sig. = kpr,
FIG. 1. Illustration of broadband stimulated Raman spectroscopy.
SRS Loss : kpr − kpu + kpu = −ksig. ⇒ ksig. = −kpr .
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 150.216.68.200 On: Thu, 23 Apr 2015 06:00:38
114201-3
Cen et al.
Both the SRS gain and loss will emit in the probe direction, the gain will be in phase with the probe and cause the increase of the probe light intensity, and the SRS loss will be out of phase with the probe and cause the probe light intensity to decrease. From the phase match condition, we can see that SRS is a self phase matching method. The SRS signal always emits along with the Raman probe, which implies that the experimental geometry can be very flexible and fluorescence rejection can be easily realized.2–4 Off resonance, all the bands in the SRS loss spectrum are equal in magnitude and linewidth to the SRS gain spectrum. On resonance, transition possibilities will strongly depend on the relative position of the Raman pump and probe at the absorption spectrum, thus wavelength dependent intensity should be expected. Additionally, the mechanisms which are insignificant for off-resonance SRS, such as diagram (b) for the SRS gain and (c) for loss in Fig. 2, could not be neglected anymore because of the on-resonance enhancement, and these mechanisms may contribute to the SRS signal with different phases, which will lead to complicated lineshapes of the BBSRS bands.
J. Chem. Phys. 142, 114201 (2015)
EXPERIMENT
and combined with a small offset vertically, and focused into the sample with a small displacement (beam diameter at focus ∼70 µm). One of the two white light beams overlaps with the pump beam inside the sample as the Raman probe, and the other beam is used as a reference. The Raman pump beam is focused into ∼250 µm diameter with another lens. The cross angle between the pump and probe beam is about 5◦, and both the pump and probe are set to p-polarized. The probe and reference beams are collected and collimated, then sent into a spectrograph and recorded with a CCD (Charge Coupled Device) camera. Most of the spectra shown in this paper have been taken with a 300 g/mm grating to catch broad enough spectra in a single measurement and save acquisition time. A 1200 g/mm grating was used to get higher spectral resolution for lineshape analysis of the resonant bands. The probe and reference spectra with and without the Raman pump were recorded with 10 s acquisition time and averaged over 5 accumulations (longer acquisition or >5 accumulations for better S/N ratio), and then the BB-SRS spectrum was calculated by the division of the Raman probe spectra with and without the pump IProbe − Ibkg ÷ Iref − Ibkg Raman pump on . (1) ISRS = IProbe − Ibkg ÷ Iref − Ibkg Raman pump off
The experimental setup is schematically shown in Fig. 3. Briefly, a fs Ti:Sapphire laser system provides fundamental output at 800 nm (100 fs, 1 kHz) (Legend Elite-Duo-FS, Coherent). Part of the 800 nm output is used to pump a visible optical parametric amplifier (OPA) (Topas400-WL, Light conversion) after second harmonic bandwidth compressor (SHBC) to create the tunable ps Raman pump. The pulse duration is about 2.5 ps within the tuning range from 470 to 900 nm (with SHG 300-900 nm), and the line width is ∼7 cm−1. A super continuum white light generated by focusing a small portion of the fs 800 nm output into a 2 mm thick sapphire window is used as the Raman probe. The white light spectrum covers from 420 nm to 800 nm, so that with proper choice of the Raman pump wavelength, full Raman spectrum in both the Stokes and anti-Stokes sides (gain and loss) can be observed. In this paper, most of the spectra showed are below 1800 cm−1 (−1800 cm−1 to 1800 cm−1) to focus on the resonant SRS bands. The white light is split into two beams,
Iprobe is the white light intensity on the CCD detector and Ibkg is the detector background measured with the Raman probe and reference blocked. The calculated ISRS is similar to the transmittance in absorption spectroscopy. ISRS is dimensionless and sometimes labeled as percentage. MG (purchased from Sinopharm Chemical Reagent Co., Ltd, and used without further purification) was dissolved in specpure ethanol. The concentration used in most of our measurement is 1.2 × 10−4 mol/L. Pump power used is 30 mW (30 µJ/pulse) if not specified, which is chosen after we tested the dependence of the BB-SRS signal on the Raman pump energy to make sure that the BB-SRS spectra in both gain and loss are not saturated. We closely compared the spectra obtained with a 1 mm quartz cuvette and a flow-cell, and found no difference. Then, the 1 mm cuvette was used without circulation of the sample. The spontaneous Raman spectra measured with a confocal Raman system (Xplora, Horiba), with 532 nm excitation (0.25 mW) and 10 s accumulation.
RESULTS AND DISCUSSION
FIG. 3. Schematic of broadband stimulated Raman setup.
Fig. 4(a) shows the absorption spectrum of malachite green in ethanol. The first absorption peaked at 621 nm indicates with the dashed line. Figs. 4(b) and 4(c) present the BB-SRS spectra of the sample and pure solvent (ethanol), respectively. From the BB-SRS spectrum of ethanol, Raman bands at the Stokes side of the Raman pump laser are positive peaks as we may observe in a regular or conventional Raman spectroscopy, and this is the SRS gain spectrum; at the antiStokes side, the BB-SRS bands appear as negative peaks, and this is the SRS loss spectrum. We have to point out that both the gain and loss spectra show almost exactly the same feature as that in conventional Raman spectroscopy. As for the BB-SRS spectra of the MG solution in Fig. 4(b), the Raman
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 150.216.68.200 On: Thu, 23 Apr 2015 06:00:38
114201-4
Cen et al.
J. Chem. Phys. 142, 114201 (2015)
FIG. 4. (a) Absorption spectrum of malachite green in ethanol. BB-SRS spectra pumping at 690 nm, (b) malachite green/ethanol solution (c = 1.2 × 10−4 mol/L), (c) ethanol. Dashed line in (a) shows the absorption maximum, and arrow indicates the pump wavelength at the absorption spectrum.
pump wavelength is on-resonant with the S1 ← S0 transition of MG. Resonant Raman bands from the MG molecule appear in both the SRS loss and gain spectra. There is no resonant band observed above 1800 cm−1 in the BB-SRS spectra, so that we will only present data between ±1800 cm−1 in what follows. Fig. 5 shows a closer comparison of the conventional Raman spectrum of solid MG and the BB-SRS spectra of the MG solution. Fig. 5(b) is the SRS gain spectrum enlarged 10 times. Fig. 5(c) is the loss spectrum plotted in reverse sign of the Raman shift. The solvent Raman bands marked with dashed box (positive peaks in gain, and negative peaks in loss). The assignment of Raman bands is summarized in Table I.42,43 All the resonant bands appeared in the BB-SRS spectra related to the vibrations of aromatic ring structure of MG. It is worthwhile to point out several different aspects in BB-SRS: (i) the two low frequency modes at 226 cm−1 and 430 cm−1 overlapped with the solvent bands and it is not clear that these two bands are present or not in the BB-SRS spectra, and the 530 cm−1 mode only appears in the loss spectrum; (ii) the resonant peak intensity of the same mode in the BB-SRS gain spectrum is significantly smaller than that of
FIG. 5. (a) Conventional Raman spectrum of solid MG excited at 532 nm; (b) Raman gain (enlarged 10 times) and (c) Raman loss spectra of BB-SRS pumping at 690 nm. Spectra are offset vertically, and the loss spectrum plotted in reverse sign for better comparison. Solvent bands marked with dashed box.
the loss spectrum; (iii) the lineshapes in the gain spectrum are normal, and all resonant bands are positive peaks as those of the solvent. In the loss spectrum, most of the resonant Raman bands are positive peaks (opposite from those of the solvent), and some of them are dispersive. By tuning the pumping wavelength (λpu) across the absorption band from 760 nm to 550 nm (with a 5 nm interval), we measured the wavelength dependence BB-SRS spectra of MG in ethanol in order to analyze the properties of the resonant Raman bands. Fig. 6 shows the BB-SRS spectra with different λpu, all spectra normalized by the band intensity of the 883 cm−1 of the solvent Raman band. There are three distinguishable ranges in the BB-SRS: (1) with λpu longer than 700 nm, the resonant gain bands fall around 800 nm (for instance, when λpu = 700 nm, the resonant gain band of 1619 cm−1 mode is at 790 nm), which will interfere with the fs laser for generation of the white light. Since cutoff edge filters were used to remove the 800 nm residual (spectrum shown in the supplementary material Fig. S1),44 the gain spectra in this range are of very poor quality. Thus, we did not present the BB-SRS spectra with λpu longer than 700 nm; (2) with λpu between 690 and 640 nm, in the red side of the first absorption corresponding to the rising of the S1 ← S0 transition of MG, the resonant SRS bands can
TABLE I. Assignment of the BB-SRS bands. This work SRS (cm−1) 807 918 1174 1224 1297 1368 1399 1485 1595 1619 a Data b Data
Conventional Raman (cm−1)
Reporteda (cm−1)
Calculatedb (cm−1)
Assignment
804 917 1174 1219 1295 1365 1395 1488 1592 1617
803 918 1177 1223 1300 1374 1403 1494 1594 1620
... ... ... ... 1306 1366 1413 ... 1572 1594
γ(C–H)ringa Ring skeletal vibrationa δ(C–H)ringa δ(C–H)ringa ν(C–C)ring ν s(N–Ph–C)a δ(C–H)ringa and ν(C–C)ringb ν(C–C)ringa and δ (CH3)b ν(C–C)ringa,b ν(C–C)ringa,b
from Ref. 42. from Ref. 43.
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 150.216.68.200 On: Thu, 23 Apr 2015 06:00:38
114201-5
Cen et al.
J. Chem. Phys. 142, 114201 (2015)
FIG. 7. BB-SRS loss spectra with resonant bands at different pump wavelengths.
FIG. 6. Wavelength dependent BB-SRS spectra. (a) Resonant bands appear in both loss and gain, (b) no resonant band. Normalized with the 883 cm−1 solvent band intensity and offset for comparison. Concentration of MG is 1.2 × 10−4 mol/L.
be seen in both the loss and gain spectra, as shown in Fig. 6(a); (3) with λpu shorter than 630 nm, around or in the blue side of the absorption peak, only solvent SRS bands can be observed, there is no resonant SRS signal of the MG solute in both the loss and gain spectra, as displayed in Fig. 6(b). In Fig. 7, we plotted the BB-SRS loss spectra under the λpu with resonant bands. Although it is difficult to distinguish lineshapes for the resonant bands between 1100 and 1500 cm−1 because of too many vibrations next to each other, the intense resonant band at 1619 cm−1 and the rather weak band at 807 cm−1 show clear dispersive lineshapes. The lineshape of the 1619 cm−1 band is the same as that of the offresonant solvent bands at λpu = 760 nm. With shorter λpu, the 1619 cm−1 band becomes dispersive with positive at higher frequency edge and negative at low frequency edge (H+/L−). Around 700 nm, this band turns into a positive peak, then becomes dispersive again but in different phase H−/L+. At 640 nm, it turns into regular negative peak as that of the solvent again. Although the lineshapes change drastically with pump wavelength, the peak position keeps constant within the limit of our spectral resolution. Similar dispersive lineshapes of the resonant SRS spectra of crystal violet were reported by Umapathy et al.14 and Rhodamine 6G by Mathies et al.,15,16 and possible explanations and simulations for the lineshapes had been provided by these authors and other researchers.13,20
However, it is still far away from fully understanding the nature of the lineshape complexity. We believe that this is a typical Fano lineshape introduced by contributions of more than one SRS mechanisms to the BB-SRS spectra. (Detailed analysis of the lineshape will be discussed elsewhere.) Fig. 8 presents the pump energy dependence of the resonant BB-SRS bands. Fig. 8(a) is the loss spectra, and Fig. 8(b) is the gain spectra enlarged 10 times for better view of the resonant bands. The peak intensities of the 1619 cm−1 band vs. pump energy displayed in Fig. 8(c). The SRS gain intensity is amplified 15 times for comparison. The pump energy dependence of the resonant BB-SRS signals is similar to that of the non-resonant SRS as reported by Mathies et al.4 Both the loss and gain SRS signals are linearly increasing with the pump energy, but with different slopes. This implies that there is most likely no higher order nonlinear process involved in the resonance BB-SRS under the current experimental conditions. To understand the wavelength dependence resonant band intensity difference of each vibrational mode in the gain and loss spectra, we need to take the transition possibilities at λpu (νpu), λGain (νpu − νvib), and λLoss (νpu + νvib) into consideration. Since the gain and loss bands generated from the same pump pulse, we can approximately assume the contributions from the pump to the loss and gain to be the same and only take the probe to account for the difference of the gain and loss resonant bands. We can estimate the relative band intensity ILoss and IGain at each λpu from the following formula:44 ( ) (Ptr)λLoss (Ipr)λLoss ILoss ∝ × . (2) IGain on−resonance (Ptr)λGain (Ipr)λGain In Eq. (2), Ptr is the transition possibility, and Ipr is the intensity of the white light at the corresponding wavelength and can be derived from the white light spectrum. Although the absolute value of Ptr is not available, we can estimate the relative transition possibility from the absorption spectrum of the sample because of Ptr ∝ absorbance at each corresponding wavelength. However, the absorbance in the UV-VIS absorption spectrum only represents the sum of the electro-vibronic
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 150.216.68.200 On: Thu, 23 Apr 2015 06:00:38
114201-6
Cen et al.
J. Chem. Phys. 142, 114201 (2015)
FIG. 9. Pump wavelength dependent peak intensity of BB-SRS loss bands. Scattered symbols in (a)-(c) are the data points and solid lines are from fitting. (d) is the absorption spectrum of malachite green in ethanol. Dashed vertical lines indicate the peak position of each Raman mode.
FIG. 8. Pump energy dependent BB-SRS (a) loss and (b) gain spectra pumping at 690 nm. (c) Resonant SRS intensity of the 1619 cm−1 mode in the gain (square, enlarged 15 times) and loss (dot) vs. pump energy. Solid lines are from linear fitting. Concentration of MG is 1.2 × 10−4 mol/L.
coupling factors of all the vibration modes involved, so that it may not be specific enough. Fortunately, we can get a better estimation of the relative transition possibility from our BB-SRS spectroscopy. Fig. 9 presents the band intensities of the 1619 cm−1, 1174 cm−1, and 807 cm−1 Raman modes in the BB-SRS loss spectra (as shown in Fig. 7) at different λpu along with the absorption spectrum of the sample. The peak position of each band is marked with dashed vertical line, and all of them fall in the range of the rising side of the S1 ← S0 absorption. These absorption-like spectra are the Raman excitation profiles (REP) for the corresponding vibrations. We can use the fitting curves in Fig. 9 to derive the specified relative transition possibility of each Raman mode by measuring the amplitude at the corresponding wavelength. Fig. 10 shows the pump wavelength dependent ratio of the resonant band intensity, ILoss/IGain, for (a) 807 cm−1, (b) 1174 cm−1, and (c) 1619 cm−1 Raman modes (solid squares). The solid line was calculated with Eq. (2) by using the mode specific REP from Fig. 9, and dashed line was from the absorption spectrum of the sample. Both the calculated curves were rescaled to match the data. It is clearly shown that using the mode specific REP obtained from the BB-SRS spectroscopy to estimate the coupling factor is much better than simply using
the absorption spectrum of the sample. These results proved that mode specific measurement of the Frank-Condon activity may be realized by BB-SRS spectroscopy. We should point out that the simplified model of Eq. (2) can only give the qualitative trend of the wavelength dependence behavior, and the calculated curves have to be rescaled to match the data. In order to quantitatively interpreting these results, detailed analysis in terms of the molecular dynamics and excited state potential surface should be carried out.
FIG. 10. Pump wavelength dependent ratio of band intensity of the BB-SRS loss and gain of (a) 807 cm−1, (b) 1174 cm−1, and (c) 1619 cm−1 Raman mode. Solid squares are the data points. Solid line is calculated with Eq. (2) by using values from Fig. 9, and dashed line is from the absorption spectrum, respectively.
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 150.216.68.200 On: Thu, 23 Apr 2015 06:00:38
114201-7
Cen et al.
J. Chem. Phys. 142, 114201 (2015)
CONCLUSION
8A. L. Dobryakov, M. Quick, I. N. Ioffe, A. A. Granovsky, N. P. Ernsting, and
By using a wide tuning range picosecond visible source, we performed pump wavelength dependent resonance broadband stimulated Raman spectroscopy on malachite green in ethanol. The resonance BB-SRS spectra measured with the pump wavelength tuning across the first visible absorption peak in a 5 nm interval, corresponding to the S1 ← S0 transition of the molecule. Due to the fluorescence rejection nature of the BB-SRS spectroscopy, we obtained high quality Raman spectra even at pump wavelength on the absorption maximum without the interference from the fluorescence. With the broadband Raman probe, we can catch the resonant BBSRS bands in both the gain and loss spectra up to ±1800 cm−1 simultaneously. The resonant BB-SRS bands could be seen only when the pump laser tuned in the range of the red edge of the visible absorption spectrum. Dispersive lineshapes of resonant Raman bands have been observed in the SRS loss spectra, while the lineshapes are normal (same as the offresonance SRS) in the Raman gain spectra. The pump energy dependence measurement showed that the resonance BB-SRS band intensity linearly increases with the pump energy, which implies that there is not likely any higher order nonlinear process involved in resonance BB-SRS. For each active resonant mode, the BB-SRS band intensity in the loss side is usually stronger than that in the gain side and varies with the pump wavelength significantly. We believe that the wavelength dependence behavior of the resonant BB-SRS bands comes from the difference of the transition possibilities at λLoss (νpu + νvib) for loss and λGain (νpu − νvib) for gain. From BB-SRS loss spectra, we can obtain the mode-specified absorption spectra for each resonant Raman active vibration, which can be used to estimate the mode specified Frank-Condon activity. Benefited from the high sensitivity and fluorescence-free nature, the broadband stimulated Raman spectroscopy is a powerful tool to perform resonance Raman measurement. BBSRS not only can be used to follow ultrafast dynamics but is also a unique characterization tool.
9J. Lee, J. R. Challa, and D. W. McCamant, J. Raman Spectrosc. 44(9), 1263-
S. A. Kovalenko, J. Chem. Phys. 140(18), 184310 (2014).
ACKNOWLEDGMENTS
This work was supported by the National Nature Science Foundation of China (Grant No. 21273179). 1H.-o. Hamaguchi and T. L. Gustafson, Annu. Rev. Phys. Chem. 45, 593-622
(1994). Kukura, S. Yoon, and R. A. Mathies, Anal. Chem. 78(17), 5952-5959 (2006). 3P. Kukura, D. W. McCamant, and R. A. Mathies, Annu. Rev. Phys. Chem. 58, 461-488 (2007). 4D. W. McCamant, P. Kukura, S. Yoon, and R. A. Mathies, Rev. Sci. Instrum. 75(11), 4971-4980 (2004). 5Z. Wang, Y. Pang, and D. D. Dlott, J. Phys. Chem. A 111(17), 3196-3208 (2007). 6J. C. Deàk, Y. Pang, T. D. Sechler, Z. Wang, and D. D. Dlott, Science 306(5695), 473-476 (2004). 7T. M. Kardas, B. Ratajska-Gadomska, A. Lapini, E. Ragnoni, R. Righini, M. Di Donato, P. Foggi, and W. Gadomski, J. Chem. Phys. 140(20), 204312 (2014). 2P.
1272 (2013). 10M. Chergui, S. Takeuchi, H. Kuramochi, T. Tahara, A. Taylor, S. Cundiff, R.
de Vivie-Riedle, and K. Yamagouchi, EPJ Web Conf. 41, 08002 (2013). Varghese, S. J. Yoon, E. M. Calzado, S. Casado, P. G. Boj, M. A. DiazGarcia, R. Resel, R. Fischer, B. Milian-Medina, R. Wannemacher, S. Y. Park, and J. Gierschner, Adv. Mater. 24(48), 6473-6478 (2012). 12H. Kuramochi, S. Takeuchi, and T. Tahara, J. Phys. Chem. Lett. 3(15), 20252029 (2012). 13D. W. McCamant, J. Phys. Chem. B 115(29), 9299-9305 (2011). 14S. Umapathy, B. Mallick, and A. Lakshmanna, J. Chem. Phys. 133(2), 024505 (2010). 15S. Shim, C. M. Stuart, and R. A. Mathies, ChemPhysChem 9(5), 697-699 (2008). 16R. R. Frontiera, S. Shim, and R. A. Mathies, J. Chem. Phys. 129(6), 064507 (2008). 17H. Ando, B. P. Fingerhut, K. E. Dorfman, J. D. Biggs, and S. Mukamel, J. Am. Chem. Soc. 136(42), 14801-14810 (2014). 18U. Harbola, S. Umapathy, and S. Mukamel, Phys. Rev. A 88(1), 011801 (2013). 19B. Zhao, Z. Sun, and S.-Y. Lee, J. Chem. Phys. 134(2), 024307 (2011). 20X. Qiu, X. Li, K. Niu, and S. Y. Lee, J. Chem. Phys. 135(16), 164502 (2011). 21X. Qiu, X. Li, K. Niu, and S.-Y. Lee, J. Raman Spectrosc. 42(4), 563-575 (2011). 22K. Niu, B. Zhao, Z. Sun, and S. Y. Lee, J. Chem. Phys. 132(8), 084510 (2010). 23K. Niu, S. Cong, and S. Y. Lee, J. Chem. Phys. 131(5), 054311 (2009). 24Z. Sun, X. Q. Qiu, J. Lu, D. H. Zhang, and S.-Y. Lee, J. Raman Spectrosc. 39(11), 1568-1577 (2008). 25Z. Sun, J. Lu, D. H. Zhang, and S. Y. Lee, J. Chem. Phys. 128(14), 144114 (2008). 26Z. Sun, Z. Jin, J. Lu, D. H. Zhang, and S. Y. Lee, J. Chem. Phys. 126(17), 174104 (2007). 27S. Yoon, D. W. McCamant, P. Kukura, R. A. Mathies, D. Zhang, and S.-Y. Lee, J. Chem. Phys. 122(2), 024505 (2005). 28S. Y. Lee, D. Zhang, D. W. McCamant, P. Kukura, and R. A. Mathies, J. Chem. Phys. 121(8), 3632-3642 (2004). 29A. Baiardi, J. Bloino, and V. Barone, J. Chem. Phys. 141(11), 114108 (2014). 30A. M. Kelley, J. Phys. Chem. A 112(47), 11975-11991 (2008). 31R. Horvath and K. C. Gordon, Coord. Chem. Rev. 254(21-22), 2505-2518 (2010). 32B. Mallick, A. Lakhsmanna, and S. Umapathy, J. Raman Spectrosc. 42(10), 1883-1890 (2011). 33S. Umapathy, A. Lakshmanna, and B. Mallick, J. Raman Spectrosc. 40(3), 235-237 (2009). 34B. Zhao, K. Niu, X. T. Li, and S. Y. Lee, Sci. China: Chem. 54(8), 1989-2008 (2011). 35A. W. Parker, R. H. Bisby, G. M. Greetham, P. Kukura, K. M. Scherer, and M. Towrie, J. Phys. Chem. B 118(42), 12087-12097 (2014). 36S. Funkner, K. Saito, G. Niehues, Y. Yazawa, T. Furuya, K. Yamamoto, and M. Tani, Appl. Phys. Lett. 105(2), 021103 (2014). 37L. D. Zhu, W. M. Liu, and C. Fang, Appl. Phys. Lett. 105(4), 041106 (2014). 38W. Yu, P. J. Donohoo-Vallett, J. Zhou, and A. E. Bragg, J. Chem. Phys. 141(4), 044201 (2014). 39M. Creelman, M. Kumauchi, W. D. Hoff, and R. A. Mathies, J. Phys. Chem. B 118(3), 659-667 (2014). 40E. Pontecorvo, C. Ferrante, C. G. Elles, and T. Scopigno, J. Phys. Chem. B 118(24), 6915-6921 (2014). 41D. P. Hoffman, O. P. Lee, J. E. Millstone, M. S. Chen, T. A. Su, M. Creelman, J. M. Fréchet, and R. A. Mathies, J. Phys. Chem. C 117(14), 6990-6997 (2013). 42G.-N. Xiao and S.-Q. Man, Spectrosc. Lett. 46(8), 577-582 (2013). 43G. B. S. Schneider and P. Freunscht, Phys. Status Solidi B 189(1), 37-42 (1995). 44See supplementary material at http://dx.doi.org/10.1063/1.4914188 for Raman probe spectrum and background subtraction information. 11S.
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 150.216.68.200 On: Thu, 23 Apr 2015 06:00:38
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 150.216.68.200 On: Thu, 23 Apr 2015 06:00:38
THE JOURNAL OF CHEMICAL PHYSICS 142, 114201 (2015)
Wavelength dependent resonance Raman band intensity of broadband stimulated Raman spectroscopy of malachite green in ethanol Qiongyan Cen ( 岑琼燕 ),a) Yuhan He ( 何玉韩 ),a) Mei Xu ( 徐媚 ), Jingjing Wang ( 王静静 ), and Zhaohui Wang ( 王朝晖 )b) State Key Laboratory of Physical Chemistry of Solid Surfaces, The MOE Key Laboratory of Spectrochemical Analysis and Instrumentation, Xiamen University, Xiamen 361005, China and Department of Chemistry, College of Chemistry and Chemical Engineering, Xiamen University, Xiamen 361005, China
(Received 29 December 2014; accepted 25 February 2015; published online 16 March 2015) Resonance broadband stimulated Raman spectroscopy of malachite green in ethanol has been performed. With a tuning picosecond visible laser source and a broadband Raman probe, the Raman gain and loss spectra have been measured simultaneously. By scanning the Raman pump across the first absorption band of the molecule, we found that the resonant Raman bands could be only seen when the pump laser tuned in the range of the red edge of the S1 ← S0 transition. Dispersive lineshapes of resonant Raman bands have been observed in the Raman loss spectra, while the line shape is normal (same as spontaneous Raman) in the Raman gain spectra. Although, the resonant bands in the loss spectrum are usually stronger than that in the gain spectrum, the band intensities of both loss and gain linearly increase with the pump energy. The relative magnitude of each corresponding resonant band in the Raman loss and gain varies with the pump wavelength. Mode specified Raman excitation profiles have been obtained through broadband stimulated Raman measurement. C 2015 AIP Publishing LLC. [http://dx.doi.org/10.1063/1.4914188]
INTRODUCTION
Raman spectroscopy is a powerful tool to reveal molecular vibrations and their signatures, enabling the accessibility of a large amount of information about molecular structures and properties, the magnitude of the bond strengths, the nature of the chemical environment, or even the characteristics of the electronic excited state.1–4 However, spontaneous Raman scattering is typically very weak and easily shielded by fluorescence and other luminescent processes. Limited by the uncertainty principle, spontaneous Raman has limited temporal resolution while hold adequate spectral resolution when used in the field of ultrafast spectroscopy.1,5,6 Recently, with the advancement in ultrafast laser techniques, nonlinear Raman spectroscopic methods have been developed with large enhancement in sensitivity (up to 106 times) and great improvement in temporal resolution (down to tens of femtoseconds) to study some of the most challenging fundamental problems in natural science. Femtosecond (fs, 10−15 s) Stimulated Raman Spectroscopy (FSRS) is one of the most powerful nonlinear Raman techniques to access short-lived intermediates of molecules in condensed phase.2–4,7–28 In FSRS, a fs actinic pulse is used to create population in the electronic excited state; at certain delay time, a picosecond (ps, 10−12 s) Raman pump and a broadband fs Raman probe pulse interact with the molecules simultaneously to generate the stimulated Raman signals. Without the actinic pulse, FSRS provides conventional Raman spectra, to be more specific, we will call FSRS without the a)Q. Cen and Y. He contributed equally to this work. b)Author to whom correspondence should be addressed. Electronic mail:
[email protected] 0021-9606/2015/142(11)/114201/7/$30.00
actinic pulse as Broadband Stimulated Raman spectroscopy (BB-SRS) there-in-after. In BB-SRS, stimulated Raman bands appear at both the red (Stokes) and blue (anti-Stokes) side of the Raman pump due to exchange of energy between the two incident pulses. Unlike the spontaneous Raman, the anti-Stokes side BB-SRS signal represents the same feature as the Stokes side2,3,18,19 and provides almost exactly the same information as spontaneous Raman when the Raman pump is set at off-resonance away from the sample electronic transitions. At resonant condition, in conventional Raman, the incident light is set to match the electronic transition, which can greatly enhance the scattering intensity compared to the spontaneous Raman scattering.29,30 The resonance Raman process is coupled with the electronic excited states. Therefore, the structure information about the excited state can be derived from resonance Raman spectroscopy. Theoretical and experimental results showed that the resonance Raman was mainly controlled by Frank Condon effect, while other interacting mechanisms, such as Duschinsky mixing and Herzberg-Teller effect, contribute to the resonance Raman intensity as well.29–31 However, in BB-SRS, resonant condition becomes much more complicated, the resonance Raman signals can be observed when either or both of the Raman pump and Raman probe are set in resonance with the electronic transition of the molecule. In condensed phase, both the Raman pump and probe will fall into the broad absorption band of the sample in most of the cases. Therefore, we should expect for different spectral features in BB-SRS due to the stimulated nature and the double resonance condition because of the extra probe pulse added. (Only if we put the Raman pump at the very edge of the absorption band of the sample, we may
142, 114201-1
© 2015 AIP Publishing LLC
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 150.216.68.200 On: Thu, 23 Apr 2015 06:00:38
114201-2
Cen et al.
avoid the double resonant complexity, then the enhancement from the resonance will be trivial.) Recently, the Raman pump was set to on-resonance to enhance the excited state SRS signal in FSRS studies.7–16 Tahara et al.10,12 used ultraviolet resonance SRS to follow the early time photodynamics of yellow protein and obtained stronger SRS signal than that of using the near IR Raman pump. Dobryakov et al. observed excited state SRS spectra of trans-azobenzene even without the actinic pulse.8 Resonance BB-SRS measurement of Rhodamine 6G by Mathies et al. showed dispersive lineshapes of the resonant bands in both SRS loss and Gain spectra.13,15,16 Umapathy et al. reported mode dependent dispersive lineshapes of crystal violet at the loss side of their BB-SRS.14 Lee et al. simulated resonance BB-SRS spectra with the consideration of resonance Raman stimulated effects between the two electronically resonant states and predicted the dispersive lineshapes and the overall spectral feature in very well consistence with experiments for both off-resonance and on-resonance BB-SRS.20,23 However, there is only very few reported experimental results about resonance BB-SRS, detailed experimental and theoretical analyses are needed for better understanding of the resonance effects in the BB-SRS spectra. In this paper, we report a systematic measurement of resonance BB-SRS of malachite green (MG) in ethanol. By varying the Raman pump wavelength across the S1 ← S0 transition of malachite green, along with a very broadband probe, we can measure the BB-SRS gain and loss spectra simultaneously. Hence, the properties of the resonance Raman band for each vibrational mode appearing in the BB-SRS loss and Gain spectra can be compared under exactly the same experimental condition.
METHOD
SRS spectroscopy has been thoroughly presented in the literature both experimentally and theoretically.2–4,7–28,32–41 We will briefly review the main aspects related to our work in this paper. In BB-SRS, we can observe Raman bands on both the lower energy side (Stokes) and higher energy side (anti-Stokes) with respect to the Raman pump. The signals on the Stokes side appear as gain (positive) of the Raman probe, usually called SRS gain. On the anti-Stokes side, the Raman signals emerge as loss (negative) of the Raman probe, known as SRS loss, as displayed in Fig. 1, respectively. The peak frequencies of the Raman bands both in SRS gain and loss agree with that in
J. Chem. Phys. 142, 114201 (2015)
FIG. 2. Four-wave mixing energy level diagram for SRS. (a) and (b) corresponding to the gain process, and (c) and (d) for the loss.
spontaneous Raman within the experimental error, and the SRS gain spectrum matches the spontaneous Raman closely. Although there are different terms for the loss spectra with consideration of contributions from different mechanisms in the literature, such as ultrafast Raman loss,14,32,33 inverse Raman,20 simulations and experiments showed that loss and gain signals are with equal magnitude for off-resonance excitation.16,18 The BB-SRS provides the same structure information as conventional spontaneous Raman spectroscopy under electronically off-resonance condition. Fig. 2 shows the four-wave mixing energy level diagram of the major stimulated processes which contribute to the SRS gain and loss signals. From the classical point of view, the two incident pulses, Raman pump and probe, interact with the sample in two steps: step 1, one of the interaction brings the molecules to an electronically upper state (imaginary if off-resonance), and the other interaction pulls the molecules back through stimulated emission and creates coherence in the ground state with the Raman active vibration modes. The coherence undergoes dephasing with a time constant characterized by each vibration mode. This step is an analog as IR absorption which populates the molecules to the vibrational excited states. Step 2, during the dephasing of the coherence generated in the first step, the third interaction from the Raman pump or probe reads out the coherence and generates SRS signal. The readout procedure is a sum frequency or difference frequency generation process. We only plotted diagrams of the main contribution mechanisms to the BB-SRS spectra under our experimental geometry. Detailed theoretical approaches can be found in Refs. 23–28. Even in these diagrams, although stimulated processes of both (a) and (b) will transfer energy from the Raman pump to the Raman probe and create Raman gain signal in the Stokes side of the BB-SRS spectrum, process (a) is the dominating mechanism because of Ipu/Ipr = ∼105 in our measurement. The same situation for (c) and (d), (d) is the major channel for the Raman loss signal. The SRS generation is coherent, and the phase match condition for the SRS Gain (Fig. 2(a)) and loss (Fig. 2(d)) can be written as SRS Gain : kpu − kpr − k pu = −ksig. ⇒ k sig. = kpr,
FIG. 1. Illustration of broadband stimulated Raman spectroscopy.
SRS Loss : kpr − kpu + kpu = −ksig. ⇒ ksig. = −kpr .
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 150.216.68.200 On: Thu, 23 Apr 2015 06:00:38
114201-3
Cen et al.
Both the SRS gain and loss will emit in the probe direction, the gain will be in phase with the probe and cause the increase of the probe light intensity, and the SRS loss will be out of phase with the probe and cause the probe light intensity to decrease. From the phase match condition, we can see that SRS is a self phase matching method. The SRS signal always emits along with the Raman probe, which implies that the experimental geometry can be very flexible and fluorescence rejection can be easily realized.2–4 Off resonance, all the bands in the SRS loss spectrum are equal in magnitude and linewidth to the SRS gain spectrum. On resonance, transition possibilities will strongly depend on the relative position of the Raman pump and probe at the absorption spectrum, thus wavelength dependent intensity should be expected. Additionally, the mechanisms which are insignificant for off-resonance SRS, such as diagram (b) for the SRS gain and (c) for loss in Fig. 2, could not be neglected anymore because of the on-resonance enhancement, and these mechanisms may contribute to the SRS signal with different phases, which will lead to complicated lineshapes of the BBSRS bands.
J. Chem. Phys. 142, 114201 (2015)
EXPERIMENT
and combined with a small offset vertically, and focused into the sample with a small displacement (beam diameter at focus ∼70 µm). One of the two white light beams overlaps with the pump beam inside the sample as the Raman probe, and the other beam is used as a reference. The Raman pump beam is focused into ∼250 µm diameter with another lens. The cross angle between the pump and probe beam is about 5◦, and both the pump and probe are set to p-polarized. The probe and reference beams are collected and collimated, then sent into a spectrograph and recorded with a CCD (Charge Coupled Device) camera. Most of the spectra shown in this paper have been taken with a 300 g/mm grating to catch broad enough spectra in a single measurement and save acquisition time. A 1200 g/mm grating was used to get higher spectral resolution for lineshape analysis of the resonant bands. The probe and reference spectra with and without the Raman pump were recorded with 10 s acquisition time and averaged over 5 accumulations (longer acquisition or >5 accumulations for better S/N ratio), and then the BB-SRS spectrum was calculated by the division of the Raman probe spectra with and without the pump IProbe − Ibkg ÷ Iref − Ibkg Raman pump on . (1) ISRS = IProbe − Ibkg ÷ Iref − Ibkg Raman pump off
The experimental setup is schematically shown in Fig. 3. Briefly, a fs Ti:Sapphire laser system provides fundamental output at 800 nm (100 fs, 1 kHz) (Legend Elite-Duo-FS, Coherent). Part of the 800 nm output is used to pump a visible optical parametric amplifier (OPA) (Topas400-WL, Light conversion) after second harmonic bandwidth compressor (SHBC) to create the tunable ps Raman pump. The pulse duration is about 2.5 ps within the tuning range from 470 to 900 nm (with SHG 300-900 nm), and the line width is ∼7 cm−1. A super continuum white light generated by focusing a small portion of the fs 800 nm output into a 2 mm thick sapphire window is used as the Raman probe. The white light spectrum covers from 420 nm to 800 nm, so that with proper choice of the Raman pump wavelength, full Raman spectrum in both the Stokes and anti-Stokes sides (gain and loss) can be observed. In this paper, most of the spectra showed are below 1800 cm−1 (−1800 cm−1 to 1800 cm−1) to focus on the resonant SRS bands. The white light is split into two beams,
Iprobe is the white light intensity on the CCD detector and Ibkg is the detector background measured with the Raman probe and reference blocked. The calculated ISRS is similar to the transmittance in absorption spectroscopy. ISRS is dimensionless and sometimes labeled as percentage. MG (purchased from Sinopharm Chemical Reagent Co., Ltd, and used without further purification) was dissolved in specpure ethanol. The concentration used in most of our measurement is 1.2 × 10−4 mol/L. Pump power used is 30 mW (30 µJ/pulse) if not specified, which is chosen after we tested the dependence of the BB-SRS signal on the Raman pump energy to make sure that the BB-SRS spectra in both gain and loss are not saturated. We closely compared the spectra obtained with a 1 mm quartz cuvette and a flow-cell, and found no difference. Then, the 1 mm cuvette was used without circulation of the sample. The spontaneous Raman spectra measured with a confocal Raman system (Xplora, Horiba), with 532 nm excitation (0.25 mW) and 10 s accumulation.
RESULTS AND DISCUSSION
FIG. 3. Schematic of broadband stimulated Raman setup.
Fig. 4(a) shows the absorption spectrum of malachite green in ethanol. The first absorption peaked at 621 nm indicates with the dashed line. Figs. 4(b) and 4(c) present the BB-SRS spectra of the sample and pure solvent (ethanol), respectively. From the BB-SRS spectrum of ethanol, Raman bands at the Stokes side of the Raman pump laser are positive peaks as we may observe in a regular or conventional Raman spectroscopy, and this is the SRS gain spectrum; at the antiStokes side, the BB-SRS bands appear as negative peaks, and this is the SRS loss spectrum. We have to point out that both the gain and loss spectra show almost exactly the same feature as that in conventional Raman spectroscopy. As for the BB-SRS spectra of the MG solution in Fig. 4(b), the Raman
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 150.216.68.200 On: Thu, 23 Apr 2015 06:00:38
114201-4
Cen et al.
J. Chem. Phys. 142, 114201 (2015)
FIG. 4. (a) Absorption spectrum of malachite green in ethanol. BB-SRS spectra pumping at 690 nm, (b) malachite green/ethanol solution (c = 1.2 × 10−4 mol/L), (c) ethanol. Dashed line in (a) shows the absorption maximum, and arrow indicates the pump wavelength at the absorption spectrum.
pump wavelength is on-resonant with the S1 ← S0 transition of MG. Resonant Raman bands from the MG molecule appear in both the SRS loss and gain spectra. There is no resonant band observed above 1800 cm−1 in the BB-SRS spectra, so that we will only present data between ±1800 cm−1 in what follows. Fig. 5 shows a closer comparison of the conventional Raman spectrum of solid MG and the BB-SRS spectra of the MG solution. Fig. 5(b) is the SRS gain spectrum enlarged 10 times. Fig. 5(c) is the loss spectrum plotted in reverse sign of the Raman shift. The solvent Raman bands marked with dashed box (positive peaks in gain, and negative peaks in loss). The assignment of Raman bands is summarized in Table I.42,43 All the resonant bands appeared in the BB-SRS spectra related to the vibrations of aromatic ring structure of MG. It is worthwhile to point out several different aspects in BB-SRS: (i) the two low frequency modes at 226 cm−1 and 430 cm−1 overlapped with the solvent bands and it is not clear that these two bands are present or not in the BB-SRS spectra, and the 530 cm−1 mode only appears in the loss spectrum; (ii) the resonant peak intensity of the same mode in the BB-SRS gain spectrum is significantly smaller than that of
FIG. 5. (a) Conventional Raman spectrum of solid MG excited at 532 nm; (b) Raman gain (enlarged 10 times) and (c) Raman loss spectra of BB-SRS pumping at 690 nm. Spectra are offset vertically, and the loss spectrum plotted in reverse sign for better comparison. Solvent bands marked with dashed box.
the loss spectrum; (iii) the lineshapes in the gain spectrum are normal, and all resonant bands are positive peaks as those of the solvent. In the loss spectrum, most of the resonant Raman bands are positive peaks (opposite from those of the solvent), and some of them are dispersive. By tuning the pumping wavelength (λpu) across the absorption band from 760 nm to 550 nm (with a 5 nm interval), we measured the wavelength dependence BB-SRS spectra of MG in ethanol in order to analyze the properties of the resonant Raman bands. Fig. 6 shows the BB-SRS spectra with different λpu, all spectra normalized by the band intensity of the 883 cm−1 of the solvent Raman band. There are three distinguishable ranges in the BB-SRS: (1) with λpu longer than 700 nm, the resonant gain bands fall around 800 nm (for instance, when λpu = 700 nm, the resonant gain band of 1619 cm−1 mode is at 790 nm), which will interfere with the fs laser for generation of the white light. Since cutoff edge filters were used to remove the 800 nm residual (spectrum shown in the supplementary material Fig. S1),44 the gain spectra in this range are of very poor quality. Thus, we did not present the BB-SRS spectra with λpu longer than 700 nm; (2) with λpu between 690 and 640 nm, in the red side of the first absorption corresponding to the rising of the S1 ← S0 transition of MG, the resonant SRS bands can
TABLE I. Assignment of the BB-SRS bands. This work SRS (cm−1) 807 918 1174 1224 1297 1368 1399 1485 1595 1619 a Data b Data
Conventional Raman (cm−1)
Reporteda (cm−1)
Calculatedb (cm−1)
Assignment
804 917 1174 1219 1295 1365 1395 1488 1592 1617
803 918 1177 1223 1300 1374 1403 1494 1594 1620
... ... ... ... 1306 1366 1413 ... 1572 1594
γ(C–H)ringa Ring skeletal vibrationa δ(C–H)ringa δ(C–H)ringa ν(C–C)ring ν s(N–Ph–C)a δ(C–H)ringa and ν(C–C)ringb ν(C–C)ringa and δ (CH3)b ν(C–C)ringa,b ν(C–C)ringa,b
from Ref. 42. from Ref. 43.
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 150.216.68.200 On: Thu, 23 Apr 2015 06:00:38
114201-5
Cen et al.
J. Chem. Phys. 142, 114201 (2015)
FIG. 7. BB-SRS loss spectra with resonant bands at different pump wavelengths.
FIG. 6. Wavelength dependent BB-SRS spectra. (a) Resonant bands appear in both loss and gain, (b) no resonant band. Normalized with the 883 cm−1 solvent band intensity and offset for comparison. Concentration of MG is 1.2 × 10−4 mol/L.
be seen in both the loss and gain spectra, as shown in Fig. 6(a); (3) with λpu shorter than 630 nm, around or in the blue side of the absorption peak, only solvent SRS bands can be observed, there is no resonant SRS signal of the MG solute in both the loss and gain spectra, as displayed in Fig. 6(b). In Fig. 7, we plotted the BB-SRS loss spectra under the λpu with resonant bands. Although it is difficult to distinguish lineshapes for the resonant bands between 1100 and 1500 cm−1 because of too many vibrations next to each other, the intense resonant band at 1619 cm−1 and the rather weak band at 807 cm−1 show clear dispersive lineshapes. The lineshape of the 1619 cm−1 band is the same as that of the offresonant solvent bands at λpu = 760 nm. With shorter λpu, the 1619 cm−1 band becomes dispersive with positive at higher frequency edge and negative at low frequency edge (H+/L−). Around 700 nm, this band turns into a positive peak, then becomes dispersive again but in different phase H−/L+. At 640 nm, it turns into regular negative peak as that of the solvent again. Although the lineshapes change drastically with pump wavelength, the peak position keeps constant within the limit of our spectral resolution. Similar dispersive lineshapes of the resonant SRS spectra of crystal violet were reported by Umapathy et al.14 and Rhodamine 6G by Mathies et al.,15,16 and possible explanations and simulations for the lineshapes had been provided by these authors and other researchers.13,20
However, it is still far away from fully understanding the nature of the lineshape complexity. We believe that this is a typical Fano lineshape introduced by contributions of more than one SRS mechanisms to the BB-SRS spectra. (Detailed analysis of the lineshape will be discussed elsewhere.) Fig. 8 presents the pump energy dependence of the resonant BB-SRS bands. Fig. 8(a) is the loss spectra, and Fig. 8(b) is the gain spectra enlarged 10 times for better view of the resonant bands. The peak intensities of the 1619 cm−1 band vs. pump energy displayed in Fig. 8(c). The SRS gain intensity is amplified 15 times for comparison. The pump energy dependence of the resonant BB-SRS signals is similar to that of the non-resonant SRS as reported by Mathies et al.4 Both the loss and gain SRS signals are linearly increasing with the pump energy, but with different slopes. This implies that there is most likely no higher order nonlinear process involved in the resonance BB-SRS under the current experimental conditions. To understand the wavelength dependence resonant band intensity difference of each vibrational mode in the gain and loss spectra, we need to take the transition possibilities at λpu (νpu), λGain (νpu − νvib), and λLoss (νpu + νvib) into consideration. Since the gain and loss bands generated from the same pump pulse, we can approximately assume the contributions from the pump to the loss and gain to be the same and only take the probe to account for the difference of the gain and loss resonant bands. We can estimate the relative band intensity ILoss and IGain at each λpu from the following formula:44 ( ) (Ptr)λLoss (Ipr)λLoss ILoss ∝ × . (2) IGain on−resonance (Ptr)λGain (Ipr)λGain In Eq. (2), Ptr is the transition possibility, and Ipr is the intensity of the white light at the corresponding wavelength and can be derived from the white light spectrum. Although the absolute value of Ptr is not available, we can estimate the relative transition possibility from the absorption spectrum of the sample because of Ptr ∝ absorbance at each corresponding wavelength. However, the absorbance in the UV-VIS absorption spectrum only represents the sum of the electro-vibronic
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 150.216.68.200 On: Thu, 23 Apr 2015 06:00:38
114201-6
Cen et al.
J. Chem. Phys. 142, 114201 (2015)
FIG. 9. Pump wavelength dependent peak intensity of BB-SRS loss bands. Scattered symbols in (a)-(c) are the data points and solid lines are from fitting. (d) is the absorption spectrum of malachite green in ethanol. Dashed vertical lines indicate the peak position of each Raman mode.
FIG. 8. Pump energy dependent BB-SRS (a) loss and (b) gain spectra pumping at 690 nm. (c) Resonant SRS intensity of the 1619 cm−1 mode in the gain (square, enlarged 15 times) and loss (dot) vs. pump energy. Solid lines are from linear fitting. Concentration of MG is 1.2 × 10−4 mol/L.
coupling factors of all the vibration modes involved, so that it may not be specific enough. Fortunately, we can get a better estimation of the relative transition possibility from our BB-SRS spectroscopy. Fig. 9 presents the band intensities of the 1619 cm−1, 1174 cm−1, and 807 cm−1 Raman modes in the BB-SRS loss spectra (as shown in Fig. 7) at different λpu along with the absorption spectrum of the sample. The peak position of each band is marked with dashed vertical line, and all of them fall in the range of the rising side of the S1 ← S0 absorption. These absorption-like spectra are the Raman excitation profiles (REP) for the corresponding vibrations. We can use the fitting curves in Fig. 9 to derive the specified relative transition possibility of each Raman mode by measuring the amplitude at the corresponding wavelength. Fig. 10 shows the pump wavelength dependent ratio of the resonant band intensity, ILoss/IGain, for (a) 807 cm−1, (b) 1174 cm−1, and (c) 1619 cm−1 Raman modes (solid squares). The solid line was calculated with Eq. (2) by using the mode specific REP from Fig. 9, and dashed line was from the absorption spectrum of the sample. Both the calculated curves were rescaled to match the data. It is clearly shown that using the mode specific REP obtained from the BB-SRS spectroscopy to estimate the coupling factor is much better than simply using
the absorption spectrum of the sample. These results proved that mode specific measurement of the Frank-Condon activity may be realized by BB-SRS spectroscopy. We should point out that the simplified model of Eq. (2) can only give the qualitative trend of the wavelength dependence behavior, and the calculated curves have to be rescaled to match the data. In order to quantitatively interpreting these results, detailed analysis in terms of the molecular dynamics and excited state potential surface should be carried out.
FIG. 10. Pump wavelength dependent ratio of band intensity of the BB-SRS loss and gain of (a) 807 cm−1, (b) 1174 cm−1, and (c) 1619 cm−1 Raman mode. Solid squares are the data points. Solid line is calculated with Eq. (2) by using values from Fig. 9, and dashed line is from the absorption spectrum, respectively.
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 150.216.68.200 On: Thu, 23 Apr 2015 06:00:38
114201-7
Cen et al.
J. Chem. Phys. 142, 114201 (2015)
CONCLUSION
8A. L. Dobryakov, M. Quick, I. N. Ioffe, A. A. Granovsky, N. P. Ernsting, and
By using a wide tuning range picosecond visible source, we performed pump wavelength dependent resonance broadband stimulated Raman spectroscopy on malachite green in ethanol. The resonance BB-SRS spectra measured with the pump wavelength tuning across the first visible absorption peak in a 5 nm interval, corresponding to the S1 ← S0 transition of the molecule. Due to the fluorescence rejection nature of the BB-SRS spectroscopy, we obtained high quality Raman spectra even at pump wavelength on the absorption maximum without the interference from the fluorescence. With the broadband Raman probe, we can catch the resonant BBSRS bands in both the gain and loss spectra up to ±1800 cm−1 simultaneously. The resonant BB-SRS bands could be seen only when the pump laser tuned in the range of the red edge of the visible absorption spectrum. Dispersive lineshapes of resonant Raman bands have been observed in the SRS loss spectra, while the lineshapes are normal (same as the offresonance SRS) in the Raman gain spectra. The pump energy dependence measurement showed that the resonance BB-SRS band intensity linearly increases with the pump energy, which implies that there is not likely any higher order nonlinear process involved in resonance BB-SRS. For each active resonant mode, the BB-SRS band intensity in the loss side is usually stronger than that in the gain side and varies with the pump wavelength significantly. We believe that the wavelength dependence behavior of the resonant BB-SRS bands comes from the difference of the transition possibilities at λLoss (νpu + νvib) for loss and λGain (νpu − νvib) for gain. From BB-SRS loss spectra, we can obtain the mode-specified absorption spectra for each resonant Raman active vibration, which can be used to estimate the mode specified Frank-Condon activity. Benefited from the high sensitivity and fluorescence-free nature, the broadband stimulated Raman spectroscopy is a powerful tool to perform resonance Raman measurement. BBSRS not only can be used to follow ultrafast dynamics but is also a unique characterization tool.
9J. Lee, J. R. Challa, and D. W. McCamant, J. Raman Spectrosc. 44(9), 1263-
S. A. Kovalenko, J. Chem. Phys. 140(18), 184310 (2014).
ACKNOWLEDGMENTS
This work was supported by the National Nature Science Foundation of China (Grant No. 21273179). 1H.-o. Hamaguchi and T. L. Gustafson, Annu. Rev. Phys. Chem. 45, 593-622
(1994). Kukura, S. Yoon, and R. A. Mathies, Anal. Chem. 78(17), 5952-5959 (2006). 3P. Kukura, D. W. McCamant, and R. A. Mathies, Annu. Rev. Phys. Chem. 58, 461-488 (2007). 4D. W. McCamant, P. Kukura, S. Yoon, and R. A. Mathies, Rev. Sci. Instrum. 75(11), 4971-4980 (2004). 5Z. Wang, Y. Pang, and D. D. Dlott, J. Phys. Chem. A 111(17), 3196-3208 (2007). 6J. C. Deàk, Y. Pang, T. D. Sechler, Z. Wang, and D. D. Dlott, Science 306(5695), 473-476 (2004). 7T. M. Kardas, B. Ratajska-Gadomska, A. Lapini, E. Ragnoni, R. Righini, M. Di Donato, P. Foggi, and W. Gadomski, J. Chem. Phys. 140(20), 204312 (2014). 2P.
1272 (2013). 10M. Chergui, S. Takeuchi, H. Kuramochi, T. Tahara, A. Taylor, S. Cundiff, R.
de Vivie-Riedle, and K. Yamagouchi, EPJ Web Conf. 41, 08002 (2013). Varghese, S. J. Yoon, E. M. Calzado, S. Casado, P. G. Boj, M. A. DiazGarcia, R. Resel, R. Fischer, B. Milian-Medina, R. Wannemacher, S. Y. Park, and J. Gierschner, Adv. Mater. 24(48), 6473-6478 (2012). 12H. Kuramochi, S. Takeuchi, and T. Tahara, J. Phys. Chem. Lett. 3(15), 20252029 (2012). 13D. W. McCamant, J. Phys. Chem. B 115(29), 9299-9305 (2011). 14S. Umapathy, B. Mallick, and A. Lakshmanna, J. Chem. Phys. 133(2), 024505 (2010). 15S. Shim, C. M. Stuart, and R. A. Mathies, ChemPhysChem 9(5), 697-699 (2008). 16R. R. Frontiera, S. Shim, and R. A. Mathies, J. Chem. Phys. 129(6), 064507 (2008). 17H. Ando, B. P. Fingerhut, K. E. Dorfman, J. D. Biggs, and S. Mukamel, J. Am. Chem. Soc. 136(42), 14801-14810 (2014). 18U. Harbola, S. Umapathy, and S. Mukamel, Phys. Rev. A 88(1), 011801 (2013). 19B. Zhao, Z. Sun, and S.-Y. Lee, J. Chem. Phys. 134(2), 024307 (2011). 20X. Qiu, X. Li, K. Niu, and S. Y. Lee, J. Chem. Phys. 135(16), 164502 (2011). 21X. Qiu, X. Li, K. Niu, and S.-Y. Lee, J. Raman Spectrosc. 42(4), 563-575 (2011). 22K. Niu, B. Zhao, Z. Sun, and S. Y. Lee, J. Chem. Phys. 132(8), 084510 (2010). 23K. Niu, S. Cong, and S. Y. Lee, J. Chem. Phys. 131(5), 054311 (2009). 24Z. Sun, X. Q. Qiu, J. Lu, D. H. Zhang, and S.-Y. Lee, J. Raman Spectrosc. 39(11), 1568-1577 (2008). 25Z. Sun, J. Lu, D. H. Zhang, and S. Y. Lee, J. Chem. Phys. 128(14), 144114 (2008). 26Z. Sun, Z. Jin, J. Lu, D. H. Zhang, and S. Y. Lee, J. Chem. Phys. 126(17), 174104 (2007). 27S. Yoon, D. W. McCamant, P. Kukura, R. A. Mathies, D. Zhang, and S.-Y. Lee, J. Chem. Phys. 122(2), 024505 (2005). 28S. Y. Lee, D. Zhang, D. W. McCamant, P. Kukura, and R. A. Mathies, J. Chem. Phys. 121(8), 3632-3642 (2004). 29A. Baiardi, J. Bloino, and V. Barone, J. Chem. Phys. 141(11), 114108 (2014). 30A. M. Kelley, J. Phys. Chem. A 112(47), 11975-11991 (2008). 31R. Horvath and K. C. Gordon, Coord. Chem. Rev. 254(21-22), 2505-2518 (2010). 32B. Mallick, A. Lakhsmanna, and S. Umapathy, J. Raman Spectrosc. 42(10), 1883-1890 (2011). 33S. Umapathy, A. Lakshmanna, and B. Mallick, J. Raman Spectrosc. 40(3), 235-237 (2009). 34B. Zhao, K. Niu, X. T. Li, and S. Y. Lee, Sci. China: Chem. 54(8), 1989-2008 (2011). 35A. W. Parker, R. H. Bisby, G. M. Greetham, P. Kukura, K. M. Scherer, and M. Towrie, J. Phys. Chem. B 118(42), 12087-12097 (2014). 36S. Funkner, K. Saito, G. Niehues, Y. Yazawa, T. Furuya, K. Yamamoto, and M. Tani, Appl. Phys. Lett. 105(2), 021103 (2014). 37L. D. Zhu, W. M. Liu, and C. Fang, Appl. Phys. Lett. 105(4), 041106 (2014). 38W. Yu, P. J. Donohoo-Vallett, J. Zhou, and A. E. Bragg, J. Chem. Phys. 141(4), 044201 (2014). 39M. Creelman, M. Kumauchi, W. D. Hoff, and R. A. Mathies, J. Phys. Chem. B 118(3), 659-667 (2014). 40E. Pontecorvo, C. Ferrante, C. G. Elles, and T. Scopigno, J. Phys. Chem. B 118(24), 6915-6921 (2014). 41D. P. Hoffman, O. P. Lee, J. E. Millstone, M. S. Chen, T. A. Su, M. Creelman, J. M. Fréchet, and R. A. Mathies, J. Phys. Chem. C 117(14), 6990-6997 (2013). 42G.-N. Xiao and S.-Q. Man, Spectrosc. Lett. 46(8), 577-582 (2013). 43G. B. S. Schneider and P. Freunscht, Phys. Status Solidi B 189(1), 37-42 (1995). 44See supplementary material at http://dx.doi.org/10.1063/1.4914188 for Raman probe spectrum and background subtraction information. 11S.
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 150.216.68.200 On: Thu, 23 Apr 2015 06:00:38
