Open Access
Add to BibSonomyAbstract
A setup of ultrafast transient infrared IR spectrometer is described in this paper that employed Schwarzschild objectives to focus the probe beam to a diffraction limited spot. Thus measurements were performed with very high spatial resolution in the mid-IR spectral region. Furthermore, modulating the polarization of the probe light enabled detecting transient dichroism of the sample. These capabilities of the setup were applied to study transient absorption of Photosystem II core complex and to image an organized film of methylene blue chloride dye. Moreover, a study of noise sources in a pump probe measurement is presented. The predicted noise level of the current setup was 8.25 μOD in 104 acquisitions and compared very well with the experimental observation of 9.6 μOD.
©2013 Optical Society of America
1. Introduction
Time resolved infrared (TRIR) laser spectroscopy can monitor changes in a structure of a molecule as it undergoes light induced transformations. This capability has been widely applied in a wide range of samples like dye sensitized solar cells [1,2], green fluorescent protein [3] or Photosystem II(PS II) [4]. However, spatial resolution in these studies was from 50 to 100 micrometers or lower. Low spatial resolution prevents from obtaining information about micro-scale objects or structures in the sample.
On the other hand, steady state IR micro-spectroscopic measurements have been successfully carried out with synchrotron sources [5]. The achieved spatial resolution is close to what is allowed by diffraction for a given wavelength [6,7]. This result cannot be accomplished with a conventional globar source even using focal plane array detector. In this case due to the low brightness of the source signal-to-noise ratio deteriorates rapidly as spatial resolution is increased [8].
Mid-IR quantum cascade lasers (QCL) have been successfully applied in the study of samples in vapour state [9,10]. Potentially they could also be used for TRIR measurements of liquid and solid samples. However the spectral width of a QCL output radiation is narrow compared to typical absorption lines of liquid or solid state molecules. Thus scanning has to be employed and collection of the whole spectrum becomes impractically long. However this limitation is not so detrimental for imaging applications [11].
Another type of a mid-IR light source is based on super-continuum generation in an optical fibre. This new and promising technique was used to generate light with spectrum spanning from near IR (1-2 μm) to mid-IR (3.5 – 4.5 μm) [12,13]. A successful application of this type of source in spectroscopy and microscopy has already been reported [13]. While these sources could replace synchrotrons in steady state micro-spectroscopy, time resolved measurements with super-continuum sources have not been yet demonstrated. Furthermore, wavelength range of the super-continuum has been limited to below 5 μm. Extending this range further into mid-IR could be technologically challenging.
Solid state laser based mid-IR sources usually employ difference frequency generation (DFG) by femtosecond near-infrared pulses generated by optical parametric amplification (OPA). It would be useful to compare characteristics of this type of source to synchrotron radiation. Typical spectral brightness (SB) of synchrotron radiation at mid-IR wavelengths is in the range of (1 ÷ 10) 1016 photons/(s mm2 sr) for 0.1% normalized bandwidth. Whereas DFG source can produce output radiation with SB of 1017 ÷ 1018 photons/(s mm2 sr) for the same bandwidth. For example, 2 mW average power of 6 μm laser light with bandwidth of 220 cm−1, 4 mm beam waist diameter and M2 = 1.5 is equivalent to SB of 5.65 1017 photons/(s mm2 sr) for 0.1% bandwidth. Therefore, DFG generated mid-infrared radiation is brighter by an order of magnitude. If only the spectral energy densities are compared, then the resulting signal to noise ratio is expected to be proportional and thus greater for laser generated mid-infrared by an order of magnitude. However, comparison of the noise characteristics of DFG and synchrotron sources is more complicated. Due to its broadband nature and high pulse repetition rate, IR synchrotron radiation is typically used in Fourier Transform IR setup, whereas dispersive and referenced absorption measurements are performed with DFG based sources as presented here. Beam and pulse instabilities affect the noise in the final spectrum very differently for these two methods. However, a relative comparison can be obtained from the variances of the final acquired spectra. The beam characteristics of IR Beamline B22 in Diamond Synchrotron (UK) state that spectral signal stability is 0.05% on 100% signal. This corresponds to 0.22 mOD noise in measured absorbance. Furthermore, random high intensity radio frequency noise sources with typical repetition timescale of tens to hundreds of seconds prevent long averaging times to be used at stations such as U10b at NSLS. On the contrary, laser setups allow long acquisitions. This, combined with their high pulse to pulse stability [14], results in ability to measure absorbance changes of 10 μOD or smaller [15,16].
It should be mentioned that TRIR spectra have been acquired using pulsed synchrotron radiation [17,18], but temporal resolution in these measurements was only around 100 ps and the measured pump-induced transmission changes are on the order of 5-30% with such sources. Laser sources usually have pulse durations of several hundreds of femtoseconds and as a result better resolved kinetic information can be obtained. All the above mentioned details indicate that laser mid-IR source could be used in TRIR micro-spectroscopic measurements with improved signal to noise ratio compared to a synchrotron source with the added benefit of increased time resolution and sensitivity.
An absorption spectrum acquired with polarized IR radiation can provide information about the direction of transition dipoles associated with the observed vibrational transitions, if the sample is either macroscopically oriented, or if a difference measurement is conducted by photoselection using linearly polarized optical excitation. Then, knowledge of the sample orientation regarding the probe light polarization is required. Directional information together with kinetic data could provide qualitatively new insights into the structure and functionality of materials. It is anticipated that diffraction limited pump-probe mid-infrared imaging, with polarization selection, will offer a new tool for material science (semiconductor materials), forensic science and biological sciences. Furthermore, directional excited state absorption spectrum may support information obtained by means of ground state absorption and facilitate analysis and identification of sample molecules.
Sufficient signal to noise ratio is required in order to resolve dichroic differences for pump-probe transient signals, which typically involve small transmission changes measured on top of a large background. It is therefore required to analyze and quantify the relevant sources of noise. There have been several theoretical descriptions presented for steady state spectroscopy [19–22]. In the case of time resolved pump-probe spectroscopy, the theoretical models presented in the literature are usually limited to a certain design of setup [23–25]. The results of these descriptions can easily be modified to accommodate broader range of setups.
This paper describes a TRIR spectroscopic setup that is capable to perform polarization resolved transient absorption measurements with diffraction limited spatial resolution. In addition, an improved theoretical model is presented that allows estimating influence of different noise sources in the TRIR setup. The prediction of the model is compared with experimental observations.
2. Theoretical model for noise in TRIR spectroscopy
The following description of the noise in pump probe measurement is based on the analysis presented by A.L. Dobryakov and co-authors [25] and extends it to include correlation between consecutive “pump on” and “pump off” pulses. Furthermore the expressions for different noise sources is taken from reference [22] and adapted for the TRIR measurement case. However it should be noted that this is by no means a complete and thorough statistical description of transient absorption signal and noise. We restrict our analysis to the case of an uncorrelated noise with a flat power spectral density (“white noise”) [26]. The goal of following section is to provide a platform for comparison of different techniques and to give practical tool to assess their performance. The change in sample absorbance due to pump excitation could be represented as:
where TON is sample transmittance with the pump radiation and TOFF is sample transmittance without the pump. A part of the probe beam is split off before the sample and detected by reference detector to produce a signal on one pixel R. The rest of the probe beam passes through the sample and is detected by the other detector. Further this detector will be called sample detector and the signal on one of its pixels will be S. Then sample transmittance can be expressed:The ratio Q does not have to be precisely equal to transmittance. Rather it should be directly proportional to it as the ratio of two transmittances is used in calculating ΔA. The noise in a TRIR signal could be described as a standard deviation of absorbance change ΔA:Several authors have reported that there exists substantial correlation between consecutive laser shots in the pulse train caused by variations in the laser or the OPA with different characteristic frequencies [24,25,27]. As a result, noise could be reduced if QON and QOFF are measured during two consecutive pulses. Then while evaluating SD(ΔA) it is important to take into account this correlation rQQ between the two measurements:Furthermore, the same argument applies to the ratio Q itself and correlation rSR between signal S and reference R should be also taken into account:If dominating noise is uncorrelated “white noise” and the measurement is repeated N times, the standard deviation of the estimated ΔA value decreases proportional to N-1/2. However if we assume that there exists correlation between consecutive pulses, this is no longer true. Furthermore, for uncorrelated data the standard deviation of ΔA is estimated using:This estimator becomes biased when there are significant correlations present in the system. Thorough description of this issue is beyond scope of this paper, but it is worth mentioning that there are several methods of getting better quality estimators in this case (see reference [28] and references therein). However, if correlation between two pulses decreases rapidly as separation between them increases, the estimator in Eq. (6) can still be applied and can provide an estimate with a negligible bias. Thus, we used Eq. (6) to obtain experimental values for SD(ΔA) and assumed that N-1/2 law holds too.There could be several noise components that contribute to the signals on the sample and reference detectors [22]. One of them is shot to shot fluctuations of laser pulse energy. In this case SD(S) and SD(R) could be expressed in terms of relative laser stability k:
It is reasonable to assume that k is the same for both the probe and the reference detector. If there are no other noise sources, SD(ΔA) becomes:Two additional sources of noise are the dark noise and analog-digital conversion (ADC) quantization noise. In the first case SD(ΔA) becomes:where DN is the dark noise defined as the standard deviation of the signal registered by the detector with no light falling onto the detector. The dark noise consists of several components arising from thermal fluctuations in the detector, electrical noise in amplification and ADC circuitry and others. It is assumed that DN is equal for both the sample and the reference detector, but this noise is not correlated. For the ADC quantization noise the SD(ΔS) and SD(ΔR) are equal to 12-1/2 of the least significant bit [22]. Thus if we assume that signals on both detectors are equal to S:Finally, noise will be added from changes in scattering, pointing and transmission due to physically moving or flowing the sample with the aim to replace the sample volume on a shot-to-shot basis. This is required by some applications of pump probe spectroscopy in order to avoid photodamage, photobleaching or formation of long lived photoproduct states. While the effect of bubbles and similar obstructions can be reduced by data discrimination techniques, smaller variations in the sample absorbance could easily be mistaken for pump pulse induced changes. The resulting fluctuation of sample detector signal will give rise to following noise of ΔA:Here it was assumed that SD(ASAMPLE) is the same for “pump on” and “pump off” pulses and that they are not correlated. There is a possible additional noise source arising from fluctuations of the pump pulse power. But the effect of this noise is difficult to quantify and it is not likely to be dominating. Figure 1 shows theoretically evaluated noise levels as function of signal level for a typical experimental situation. Furthermore a comparison with experimental data is presented. The data was obtained by repeating typical pump and probe measurements without a sample for required number of shots. Thus the standard deviation of the resulting signal represented a noise level. The data points in Fig. 1 correspond to the average value of the central part of the laser spectrum and the noise level should increase at the side of the spectrum due to reduced correlation rSR. The signal level was changed by reducing the intensity of IR radiation with a wire grid polarizer placed before the beamsplitter (BS in Fig. 2 ). As it can be seen from the Fig. 1, the correspondence between theory and the experiment (yellow line and orange points) is good and shows that the model is sufficient to describe the sensitivity of the instrument.
Fig. 1 Theoretical noise levels for different noise components as a function of signal amplitude. Parameter values were taken from experimental observations: Laser stability k = 0.85%, signal/reference correlation rSR = 0.93, consecutive pulse correlation rQQ = 0.8, dark noise level DN = 2 counts, sample absorbance variation dA = 0.1 mOD, number of averaged shots N = 10000. The sum of dark and laser noise theoretical values is compared to experimental data. The data was obtained by performing scans without a sample. The IR beam intensity was changed between the scans using a wire grid polarizer.
Fig. 2 General layout of the setup. OPG - optical parametric generator; DFG - difference frequency generator, CH – optical chopper, DL – delay line, AM – active mirror. Optical components of Low Resolution Branch: L1 – lens for visible pump, BS – beam splitter, RMI1, RM2, RM3 – removable mirrors for switching between Low Resolution and High Resolution Branches, OAP – Au coated 90 deg off-axis parabolic mirror, S – sample, L2 – ZnSe lens for IR probe beam. L3, L4 - Spectrometer input lenses. Optical setup of High Resolution Branch is presented in separate Figure.
3. Description of TRIR setup
The setup for TRIR absorption measurements was designed with a focus on high spectral and spatial resolutions as well as day to day stability and low noise. It was intended to be capable of measuring both large scale liquid or solid samples with sample translation and micro scale samples with dimensions of several to tens of microns. Figure 2 represents the general layout of all the components of the setup.
The primary laser source for the setup was a Ti:Sapphire regenerative amplifier (Spitfire PRO, Spectra Physics) which delivered 70 fs pulses at 800 nm with pulse repetition rate of 1 kHz. The energy of the generated pulses was up to 4 mJ and pulse to pulse fluctuations were typically lower than 0.5%. The pointing of the laser output beam was controlled by a stabilization system consisting of two quadrant detectors and two active mirrors with piezoelectric actuators (MRC Systems). The 800 nm output radiation was split between two optical parametric amplifiers (OPA, Topas-C, Light Conversion). One of the OPAs was equipped with difference frequency generation (DFG) module that enabled wavelength tuning in mid IR region (2.6 – 14 μm). The other OPA used additional frequency mixing stages to generate ultraviolet, visible and near IR light pulses (285 – 810 nm). The pulse to pulse stability of the OPA output before DFG or mixing stages was only slightly worse than the pump laser and was typically in 0.4 ÷ 0.7% region. The output of mid-IR OPA/DFG was directed to a specially designed polarization modulation setup. Figure 3 shows the layout of this setup. The main principle of its operation was following: the incoming radiation was split in two parts with a beamsplitter (BSW510 or BSW710, Thorlabs), polarization of one part was rotated by 90 degrees. The rotation was performed using image rotating periscope thus avoiding any transmission optics. The two parts were recombined using beam splitter while a delay line in one of the branches ensured precise overlap of the two pulses in time. By using an optical chopper (MC2000,Thorlabs) to block every other pulse in each branch, the setup produced sequence of pulses where every other pulse was of orthogonal polarization to the previous one. Although using this design entailed losing a significant part of the mid-IR pulse energy, it did not have any moving reflective or transmissive elements and proved to be very stable. The extinction ratio between the pulses of orthogonal polarizations in the sequence was better than 1:100. The polarization modulated mid-IR probe beam was subsequently split in two by another beamsplitter (BS in Fig. 2) to produce signal and reference beams. The reference beam was expanded to match the beam size and divergence of the signal beam before the detection system input optics.
Fig. 3 Layout of polarization modulator. BS - beam splitter, CH - chopper, P - periscope, IRP - image rotating periscope, DL- delay line.
The pump beam was modulated by another chopper at a quarter of the original pulse repetition rate. The two choppers were electronically synchronized to the laser output as well as with each other using a D-type flip-flop TTL logic and their state was additionally read out by the data acquisition system via an external input. This readout was used by the acquisition software to identify pulses in the sequence. This sequence consisted of the first two pulses of orthogonal polarization coinciding with pump pulse on, while the following two probe pulses measured absorption of unpumped sample. The time delay between the probe and pump pulses was provided by a delay line. It consisted of a translation stage (M-IMS400LM, Newport) and a hollow retro-reflector (Edmund Optics). The polarization of the pump beam was controlled by a zero-order waveplate.
Depending on the nature, scale or volume of the sample used in experiment the signal and pump beams could be directed to one of two branches. The first branch was optimized for low spatial resolution measurements (spot size 75 μm) with the capability of sample translation. The other branch was based on a microscope that was intended for the measurement with diffraction limited resolution. Switching between the two optical paths could be performed easily and with high repeatability by means of inserting mirrors on magnetic bases. In the low resolution branch of the setup the sample holder consisted of a mount adapted to accept standard FTIR cells. For liquid samples a flow cell (Harrick Scientific Products) with two 2mm CaF2 widows and Teflon spacer between them was used. The sample mount was attached to a custom built motorised sample translator that was used to move the mount in Lissajous patterns and by adjusting the translation speed it was possible to adjust the number of pump pulses probing the same sample spot. The probe beam was focused onto the sample with off-axis parabolic mirror into a spot size of 75 μm diameter (full width at half maximum, FWHM), while the pump beam was focused with a fused silica lens to a spot size of 300 μm (FWHM).
The branch of the setup that was optimized for measuring micro scale samples was based on an all-reflective infrared microscope (HYPERION 1000, Bruker). The reflective surfaces within the microscope were coated with gold to maximize the transmittance in the mid-IR. The microscope was equipped with Schwarzschild condenser and objective (36x, NA = 0.5) which enabled focusing of the probe beam to a diffraction limited spot. A measurement of the typical spot size is presented in Fig. 4 and was performed using the knife-edge technique. The data was processed following a procedure described in Ref [7] and the resulting spot size of the central peak was consistent with the values reported previously for synchrotron radiation [6,7]. As it can be seen in Fig. 4, the use of Schwarzschild optics resulted in side lobes in the beam pattern. There are methods of reducing the side lobes reported in literature [6], but they result in decreased transmission of the microscope and thus were not used in this setup. It should be noted that by very nature of Schwarzschild optics the transmission loss is rather high (50% and higher). Yet there are ways of achieving higher transmission through the microscope at the expense of spatial resolution by changing the input beam input characteristics such as beam waist size or location. This arrangement was used for some samples in this setup.
A small right angle fused silica prism was mounted behind the primary mirror of the collecting objective. It was used to direct the pump beam on to the sample. The prism was placed in the blind spot of Schwarzschild objective and produced almost no additional losses for the probe beam. The layout of the objectives, the prism and beam path through both are represented in Fig. 5 .
Fig. 5 Beam paths in microscope with Schwarzschild objectives and additional right angle prism. The lens telescope at the input of the microscope could be use to adjust the input beam parameters and optimize the throughput.
The same prism could be employed to deliver high energy continuous-wave or pulsed radiation for photo-excitation of the sample. The laser source for this radiation was nanosecond lamp pumped Nd:YAG laser (Lab-150, Spectra Physics) equipped with harmonic generator and an optical parametric oscillator (basiScan/170, Spectra Physics). The mid-IR light exiting the microscope was collimated with spherical Au mirror and directed to the detection system.
The detection system consisted of two IR Czerny-Turner type grating spectrometers (Triax190, Horiba) with mercury cadmium telluride (MCT) 128 pixel array detectors (Infrared Systems Development Corporation) attached to them. Three gratings were installed in each of the spectrometers with groove densities of 300, 200 and 100 mm−1 and blazed at 4000, 5000 and 9000 nm respectively (Princeton Instruments). The diameter of the input beam was adjusted to match the numerical aperture of the spectrometers and a ZnSe lens (focal length of 75 mm) was used to focus the beam onto the spectrometer input slits. As the spectrometers image the slit onto the focal plane, the spot size on the input slit was chosen to be smaller than pixel size of the MCT array (0.2 mm). The input geometry for the signal and reference spectrometers was matched as closely as possible and fine adjustments were made according to the signal on the detectors. Rotational structure of water vapour absorption spectrum and polystyrene thin film absorption were used to fine tune the wavelength calibration and matching.
The two MCT arrays consisted of 128 pixels each 0.2 mm in width and 0.5 mm in height and with 50 μm spacing between active areas. According to the manufacturer specifications the uniformity among the pixels was better than 15% with typical D* values of 30 ÷ 40 109 cm Hz1/2 W−1 (Peak value at 9 μm, 1 kHz). The whole array assembly was cooled by LN2 dewar. Pre-amplification, integration and ADC conversion was performed by custom designed acquisition system (FPAS, Infrared Systems). This system allowed user to adjust gain and trim for each individual pixel or the whole array and setting integration gate width and position with provided software. Acquired array data was transferred to a PC via data acquisition card (PCI-DIO-32HS, National Instruments). The control of experiment, data manipulation and representation was performed by custom software developed using LabVIEW environment (National Instruments).
4. Results
Several demonstrative studies were carried out to illustrate the unique capabilities of the setup. As it has already been previously reported, complex kinetics of charge separation in the Photosystem II core complex could be studied by means of ultrafast transient IR absorption spectroscopy [4]. A similar experiment was performed with the current setup with high spatial resolution. A liquid solution of PS II core complex was pumped by visible 680 nm radiation and probed with IR light centred at around 6000 nm. The cell holding the static liquid sample was placed in the focus of the microscope and remained stationary during the measurement. Thus special care had to be taken to prevent sample photo damage and its effect on measured spectra. Relatively short scans of randomized time delay points were acquired, constantly monitoring change in the signal amplitudes. Lowest possible pump power was used and never exceeded 5 W/cm2. Moreover the same measurement was repeated outside microscope with sample translation. As the resulting spectra exactly match the ones obtained in the microscope, it was safe to conclude that no noticeable sample photodamage occurred within measurement time.
Acquired time resolved data for each probe polarization was further processed employing global analysis method [29]. Using data from previous reports in the literature, a sequential model with four exponential decays was used to fit the data. Resulting spectra are presented in the Fig. 6 . The spectra are very similar to the ones reported previously in Ref [4], where a detailed discussion of the band assignments is also presented. In summary, the bleaching of the 1640 – 1710 cm−1 band observed at short delay times has been assigned to keto C9 = O modes of several chlorophylls or pheophytins in different environments or in presence of different hydrogen bond strengths. Weak bands at 1715 and 1730 cm−1 were assigned to 10a-ester modes of the same molecules. The excited state of keto mode was mainly responsible for the positive band above 1640 cm−1. At longer delay times the most distinct feature is a band shift at 1665/1657 cm−1, which was assigned to amide C = O vibration and its response to the charge separation [4]. Despite substantially better spectral resolution of the current setup, no additional information can be obtained due to overlap of several broad absorption features of several different molecules or of identical molecules in different surroundings. The time constants obtained by fitting current data also coincide with the ones obtained in Ref [4] within experimental error. As far as sample dichroism is concerned, it could be seen from Fig. 6, that there are no structural differences between the spectra of two probe polarizations. The absorption in this spectral region is caused by multiple pigments in the reaction center core with overlapping absorbance, which are not linear absorbers, thus no directional information was obtained for this sample. A relatively large sample was used in above measurements and high spatial resolution of the microscope was not employed. However, the capability to focus to diffraction limited spot allows potentially obtaining similar results using micro-fluidic flow capillaries or micro-drops.
Fig. 6 Time resolved IR spectra of liquid PSII core complex. Each graph represents evolution associated difference spectra obtained by global analysis of the data. A sequential model with increasing lifetimes was used and resulting time constants are indicated in the legend.
In order to test the ability of the setup to acquire polarization resolved spectra, an oriented sample was prepared. Using the technique described in Ref [30]. an organized film of methylene blue chloride (CAS 61-73-4) dye was produced. The technique involved rubbing a small amount of dye powder in one direction with cotton on a CaF2 window. Resulting film had maximum visible absorption at 580 nm of 0.1 OD for the polarization parallel to film direction (i.e. rubbing direction) and 0.39 for the polarization orthogonal to film orientation. A very strong IR absorption band at 1600 cm−1, assigned to the vibration of heterocycle skeleton [31,32], showed less pronounced dichroism with absorption for orthogonal polarizations of 20.5 mOD and 37 mOD. The TRIR absorption spectrum of the dye film, acquired with 580 nm pump polarized along one of the probe beams, is shown in the Fig. 7 . The main feature of the spectra is the negative band centred at 1600 cm−1 corresponding to the same vibrations as in the FTIR spectrum. In line with expectations, substantial difference exists for the spectra of the orthogonal polarizations.
Fig. 7 TRIR spectra of oriented methylene blue chloride dye. One probe polarization (Pol 1) was parallel to the pump radiation polarization, while the other (Pol 2) – orthogonal. The pump polarization was orthogonal to the dye orientation direction and corresponded to the orientation with higher pump absorption. The inset shows standard deviation of N = 10000 measurements of ΔA at −200 ps delay which corresponds to noise level of the setup. The noise at the centre of the spectrum is 9.6 μOD. It increases to the side of the spectrum as signal amplitude decreases and the dark noise contribution becomes noticeable.
Furthermore, high spatial resolution of this setup was further employed for imaging the edge of the previously mentioned dye thin film. Namely, a linear 450 μm path going from film to clean substrate was chosen with some random dye particles scattered along the way. The path is indicated with a grey line in the visible image of the sample obtained with a CCD camera (See Fig. 8A ). A series of scans was performed every 11 μm, each scan containing only two delay points: at −200 ps and 1 ps. The first point served as a background reference and the second point was chosen to avoid any interference from time zero artefacts, but still able to provide a strong signal.
Fig. 8 Time resolved imaging of the dye thin film edge. Left (a): Visible image of the sample obtained with a CCD camera. Grey line indicates region scanned with TRIR. Right (b): Mapping of the signal amplitude at 1600 cm−1 for two orthogonal polarizations. Dark spots on the image (a) are dye particles of around ten microns in size. As in Fig. 7, Polarization 1 was parallel to the pump radiation polarization, while Polarization 2 – orthogonal. The pump polarization was orthogonal to the dye orientation direction and corresponded to the orientation with higher pump absorption. 1000 shots were acquired at each spot
The resulting dependence of the signal at 1600 cm−1 on probe spot position for orthogonal polarizations is presented in Fig. 8. Clearly the difference in the signal amplitude between the two polarizations indicates presence of oriented structure in the sample. However this is not true for bulk dye particles that are not orientated – the two signals match in this case. As it can be seen from Fig. 8, dye particles appear as a micron sized randomly oriented objects in otherwise oriented film and could be identified from the obtained scan.
Finally to determine systems noise level, a series of acquisitions (N = 10000) before time zero was used. The standard deviation of the signal after the time zero represented the noise level of the system and is presented in the inset of Fig. 7. As current samples were of low optical density, the dominating source of noise was laser intensity fluctuations (see Fig. 1). The following parameters were taken from the thin film experiment and used in the calculations: N = 10000 acquisitions, 0.95 average signal and reference correlation coefficient (rRS), 0.75 consecutive pulse correlation (rQ) and laser stability of 0.85% (k) Then the theoretically predicted noise level is 8.25 μOD (see Eq. (8)). This predicted value compares well with experimentally observed noise level of 9.6 μOD at the centre of the spectrum.
5. Conclusions
Despite being an established technique TRIR spectroscopy has not been previously implemented with diffraction limited spatial resolution. Furthermore, polarization resolved TRIR instrumentation has been implemented here to provide the highest possible sensitivity by alternating modulation of the polarization of the pulse train. The setup described in this paper was capable of acquiring transient absorption spectra with diffraction limited spatial resolution. Namely, for 6 μm radiation measured beam spot was below 5 μm. In addition to that, an IR probe polarization modulator was employed that enabled obtaining polarization resolved absorption characteristics of the sample. The performance test of the setup showed that it is capable of measuring liquid samples at high spatial resolution without producing significant photodamage. Moreover, ultrafast changes in IR transient absorption were monitored for oriented dye thin film with polarization resolution. From the measured spectra it can be clearly seen that transition dipoles have distinct directionality that is related to dye orientation. Furthermore, the two unique capabilities of the instrument were combined to produce a TRIR scan of the film that helped indentify random particles inside otherwise oriented film. The setup also showed good noise characteristics that corresponded well to the ones predicted by presented theoretical model. The instrumentation provides new capabilities to perform time resolved mid-infrared imaging with increased S/N and time resolution as compared to synchrotron sources, and allows imaging on the basis of lifetime, polarisation, amplitude and spectral content.
These unique capabilities could be employed to investigate a broad range of objects: from flowing samples in capillaries to static liquid in micron-sized drop, from oriented films to structures formed of nano-scale objects and many more. We believe that this setup is capable of providing qualitatively new information about the structure of the samples in question and changes they undergo after excitation by pump pulse.
Acknowledgments
We would like to thank Karim Maghlaoui for providing Photosystem II sample and Josef Wachtveitl and Karsten Neumann (Goethe-Universität Frankfurt, Germany) for sharing data collection code written in LabVIEW which inspired the data processing part of our data acquisition program. This work was supported by BBSRC via award BB/H004238/1
References and links
1. R. J. Ellingson, J. B. Asbury, S. Ferrere, H. N. Ghosh, J. R. Sprague, T. Lian, and A. J. Nozik, “Dynamics of electron injection in nanocrystalline titanium dioxide films sensitized with [Ru (4, 4’-dicarboxy-2, 2'-bipyridine) 2 (NCS) 2] by infrared transient absorption,” J. Phys. Chem. B 102(34), 6455–6458 (1998). [CrossRef]
2. J. Asbury, R. Ellingson, H. N. Ghosh, S. Ferrere, A. J. Nozik, and T. Lian, “Femtosecond IR study of excited-state relaxation and electron-injection dynamics of Ru (dcbpy) 2 (NCS) 2 in solution and on nanocrystalline TiO2 and Al2O3 thin films,” J. Phys. Chem. B 103(16), 3110–3119 (1999). [CrossRef]
3. J. J. van Thor, G. Y. Georgiev, M. Towrie, and J. T. Sage, “Ultrafast and low barrier motions in the photoreactions of the green fluorescent protein,” J. Biol. Chem. 280(39), 33652–33659 (2005). [CrossRef] [PubMed]
4. N. P. Pawlowicz, M.-L. Groot, I. H. M. van Stokkum, J. Breton, and R. van Grondelle, “Charge separation and energy transfer in the photosystem II core complex studied by femtosecond midinfrared spectroscopy,” Biophys. J. 93(8), 2732–2742 (2007). [CrossRef] [PubMed]
5. P. Dumas and L. Miller, “The use of synchrotron infrared microspectroscopy in biological and biomedical investigations,” Vib. Spectrosc. 32(1), 3–21 (2003). [CrossRef]
6. G. Carr, “Resolution limits for infrared microspectroscopy explored with synchrotron radiation,” Rev. Sci. Instrum. 72(3), 1613–1619 (2001). [CrossRef]
7. H. A. Bechtel, M. C. Martin, T. E. May, and P. Lerch, “Improved spatial resolution for reflection mode infrared microscopy,” Rev. Sci. Instrum. 80(12), 126106 (2009). [CrossRef] [PubMed]
8. L. M. Miller and R. J. Smith, “Synchrotrons versus globars, point-detectors versus focal plane arrays: Selecting the best source and detector for specific infrared microspectroscopy and imaging applications,” Vib. Spectrosc. 38(1-2), 237–240 (2005). [CrossRef]
9. A. A. Kosterev and F. K. Tittel, “Chemical sensors based on quantum cascade lasers,” IEEE J. Quantum Electron. 38(6), 582–591 (2002). [CrossRef]
10. J. Manne, O. Sukhorukov, W. Jäger, and J. Tulip, “Pulsed quantum cascade laser-based cavity ring-down spectroscopy for ammonia detection in breath,” Appl. Opt. 45(36), 9230–9237 (2006). [CrossRef] [PubMed]
11. M. Weida and B. Yee, “Quantum cascade laser-based replacement for FTIR microscopy,” in SPIE BiOS (2011), p. 79021C–79021C.
12. O. P. Kulkarni, V. V. Alexander, M. Kumar, M. J. Freeman, M. N. Islam, F. L. Terry Jr, M. Neelakandan, and A. Chan, “Supercontinuum generation from ~19 to 45 μmin ZBLAN fiber with high average power generation beyond 38 μm using a thulium-doped fiber amplifier,” J. Opt. Soc. Am. B 28(10), 2486–2498 (2011). [CrossRef]
13. S. Dupont, C. Petersen, J. Thøgersen, C. Agger, O. Bang, and S. R. Keiding, “IR microscopy utilizing intense supercontinuum light source,” Opt. Express 20(5), 4887–4892 (2012). [CrossRef] [PubMed]
14. P. Hamm, R. A. Kaindl, and J. Stenger, “Noise suppression in femtosecond mid-infrared light sources,” Opt. Lett. 25(24), 1798–1800 (2000). [CrossRef] [PubMed]
15. G. M. Greetham, P. Burgos, Q. Cao, I. P. Clark, P. S. Codd, R. C. Farrow, M. W. George, M. Kogimtzis, P. Matousek, A. W. Parker, M. R. Pollard, D. A. Robinson, Z. J. Xin, and M. Towrie, “ULTRA: A unique instrument for time-resolved spectroscopy,” Appl. Spectrosc. 64(12), 1311–1319 (2010). [CrossRef] [PubMed]
16. M. Towrie, D. C. Grills, J. Dyer, J. A. Weinstein, P. Matousek, R. Barton, P. D. Bailey, N. Subramaniam, W. M. Kwok, C. Ma, D. Phillips, A. W. Parker, and M. W. George, “Development of a broadband picosecond infrared spectrometer and its incorporation into an existing ultrafast time-resolved resonance Raman, UV/visible, and fluorescence spectroscopic apparatus,” Appl. Spectrosc. 57(4), 367–380 (2003). [CrossRef] [PubMed]
17. R. P. S. M. Lobo, J. D. LaVeigne, D. H. Reitze, D. B. Tanner, and G. L. Carr, “Subnanosecond, time-resolved, broadband infrared spectroscopy using synchrotron radiation,” Rev. Sci. Instrum. 73(1), 1–10 (2002). [CrossRef]
18. L. Carroll, P. Friedli, P. Lerch, J. Schneider, D. Treyer, S. Hunziker, S. Stutz, and H. Sigg, “Ultra-broadband infrared pump-probe spectroscopy using synchrotron radiation and a tuneable pump,” Rev. Sci. Instrum. 82(6), 063101 (2011). [CrossRef] [PubMed]
19. B. Wybourne, “Optimum optical density for shot noise limited spectrophotometers,” J. Opt. Soc. Am. 50(1), 84–85 (1960). [CrossRef]
20. R. Cole, “Optimum optical density in spectrophotometry,” J. Opt. Soc. Am. 41(1), 38–40 (1951). [CrossRef]
21. L. D. Rothman, S. R. Crouch, and J. D. Ingle, “Theoretical and experimental investigation of factors affecting precision in molecular absorption spectrophotometry,” Anal. Chem. 47(8), 1226–1233 (1975). [CrossRef]
22. H. Mark and J. Workman, Chemometrics in Spectroscopy (Academic Press, 2007).
23. G. Moore and K. Koch, “Phasing of tandem crystals for nonlinear optical frequency conversion,” Opt. Commun. 124(3-4), 292–294 (1996). [CrossRef]
24. C. Schriever, S. Lochbrunner, E. Riedle, and D. J. Nesbitt, “Ultrasensitive ultraviolet-visible 20 fs absorption spectroscopy of low vapor pressure molecules in the gas phase,” Rev. Sci. Instrum. 79(1), 013107 (2008). [CrossRef] [PubMed]
25. A. L. Dobryakov, S. A. Kovalenko, A. Weigel, J. L. Pérez-Lustres, J. Lange, A. Müller, and N. P. Ernsting, “Femtosecond pump/supercontinuum-probe spectroscopy: optimized setup and signal analysis for single-shot spectral referencing,” Rev. Sci. Instrum. 81(11), 113106 (2010). [CrossRef] [PubMed]
26. A. G. P. Julius and S. Bendat, Random Data: Analysis and Measurement Procedures (Wiley, 2011).
27. D. Polli, L. Lüer, and G. Cerullo, “High-time-resolution pump-probe system with broadband detection for the study of time-domain vibrational dynamics,” Rev. Sci. Instrum. 78(10), 103108 (2007). [CrossRef] [PubMed]
28. P. M. T. Broersen, “Estimation of the accuracy of mean and variance of correlated data,” IEEE Trans. Instrum. Meas. 47(5), 1085–1091 (1998). [CrossRef]
29. L. J. G. W. van Wilderen, C. N. Lincoln, and J. J. van Thor, “Modelling multi-pulse population dynamics from ultrafast spectroscopy,” PLoS ONE 6(3), e17373 (2011). [CrossRef] [PubMed]
30. L. Wong, C. Hu, R. Paradise, Z. Zhu, A. Shtukenberg, and B. Kahr, “Relationship between tribology and optics in thin films of mechanically oriented nanocrystals,” J. Am. Chem. Soc. 134(29), 12245–12251 (2012). [CrossRef] [PubMed]
31. E. Sáez and R. Corn, “In situ polarization modulation—Fourier transform infrared spectroelectrochemistry of phenazine and phenothiazine dye films at polycrystalline gold electrodes,” Electrochim. Acta 38(12), 1619–1625 (1993). [CrossRef]
32. O. V. Ovchinnikov, S. V. Chernykh, M. S. Smirnov, D. V. Alpatova, R. P. Vorob’eva, N. Latyshev, B. Evlev, N. Utekhin, and N. Lukin, “Analysis of interaction between the organic dye methylene blue and the surface of AgCl(I) microcrystals,” J. Appl. Spectrosc. 74(6), 809–816 (2007). [CrossRef]
