Transient Grating Photoluminescence Spectroscopy: An Ultrafast

May 1, 2014 - Pieter GeiregatJorick MaesKai ChenEmile DrijversJonathan De .... Hannah L. Stern , Alexandre Cheminal , Shane R. Yost , Katharina Broch ...
5 downloads 0 Views 1MB Size
Letter pubs.acs.org/JPCL

Transient Grating Photoluminescence Spectroscopy: An Ultrafast Method of Gating Broadband Spectra Kai Chen, Joseph K. Gallaher, Alex J. Barker, and Justin M. Hodgkiss* The MacDiarmid Institute for Advanced Materials and Nanotechnology, PO Box 600, Wellington 6010, New Zealand School of Chemical and Physical Sciences, Victoria University of Wellington, PO Box 600, Wellington 6010, New Zealand S Supporting Information *

ABSTRACT: Ultrafast photoluminescence (PL) spectroscopy can cleanly resolve excitedstate dynamics and coupling to the environment, however, there is a demand for new methods that combine broadband detection and low backgrounds. We present a new method, transient grating photoluminescence spectroscopy (TGPLS), that addresses this challenge by exploiting a focusing geometry where ultrafast broadband spectra are transiently diffracted away from the background PL. We show that TGPLS can resolve the complex spectral relaxation observed in conjugated polymer and oligomer solutions, with an essentially flat spectral response throughout the visible region and potentially beyond. The benefits we demonstrate using TGPLS could expand access to spectral information, particularly for other multichromophoric and heterogeneous materials where complex spectral relaxation is expected. SECTION: Spectroscopy, Photochemistry, and Excited States

U

with optimum cutting angle matched with IR-tuned tilted wavefront gate pulses and a noncollinear phase matching geometry.24,25 More complex setups reminiscent of PL-seeded noncollinear optical parametric amplifiers can also achieve ∼150 nm bandwidth in the visible; however, detection is constrained to a window that depends on the crystal cutting angle, and spectral correction procedures are even more complicated.26−28 Optical Kerr gate spectroscopy,29−31 illustrated in Figure 1b, is a broadband alternative to upconversion. The intense gate pulse induces transient birefringence to open a shutter constructed by two orthogonal linear polarizers spanning the Kerr medium. The optical Kerr effect is phase-matched for all signal wavelengths; however, in practice, its application is limited by the transmission bandwidth and extinction ratio of the thin polarizers required to suppress the PL background. Using high extinction polarizers, leakage of only 10−5 can easily overwhelm the ultrafast gated signal when the background is accumulated over a PL lifetime that is orders of magnitude longer than the gated signal. Thus, there is an unfulfilled need for broadband ultrafast PL spectroscopy techniques that can be applied to a broader range of samples. Here we present a novel ultrafast PL method called transient grating photoluminescence spectroscopy (TGPLS) that breaks free from these constraints. TGPLS addresses the limitations of existing methods using an approach inspired by a previously investigated method of frequency-resolved optical gating

ltrafast photoluminescence (PL) measurements have provided incisive probes of the active site environment of enzymes,1 photoisomerization,2 solvation,3,4 charge-5 and energy-transfer6 processes, and the photophysics of DNA7 and conjugated polymers,8−10 to name a few examples. The information sought is not limited to the kinetics of PL decay (related to population dynamics) but also the ultrafast evolution of PL spectra, which reports on the nature of excited states.11−14 For example, PL spectra with short-lived high energy intensity have been used to identify solvated electrons in hot charge-transfer states5 as well as hot delocalized excitons in organic photovoltaic devices.10 As the principal current ultrafast PL method, fluorescence upconversion spectroscopy13,15−21 (Figure 1a) and related downconversion spectroscopy22,23 achieve ultrafast time resolution with low backgrounds because the upconverted signal generated via sum-frequency generation is spatially and spectrally separated from the background PL and the gate beam. However, upconversion suffers from an intrinsic inability to detect broadband spectra because phase matching is only achieved for a specific PL wavelength and drops off sharply according to the different dispersion of the fluorescence and upconverted signal. Obtaining an ultrafast PL spectrum requires scanning the crystal angle to separately resolve different PL wavelengths12,19 and renormalizing via convoluted calibration procedures.3 Thinner upconversion crystals (∼100 μm LiIO3) are more tolerant to dispersion and offer somewhat improved bandwidth (∼100 nm) at the expense of efficiency.21 The stateof-the-art fixed geometry (simultaneous) broadband upconversion system achieves PL detection in the specific 425−750 nm window by balancing the various constraints using a thin crystal © 2014 American Chemical Society

Received: March 31, 2014 Accepted: May 1, 2014 Published: May 1, 2014 1732

dx.doi.org/10.1021/jz5006362 | J. Phys. Chem. Lett. 2014, 5, 1732−1737

The Journal of Physical Chemistry Letters

Letter

Figure 1. Schematic depiction of (a) fluorescence upconversion, (b) optical Kerr gating, and (c) TGPLS methods. TGPLS combines the low background detection of upconversion with the broadband capability of optical Kerr gating.

Figure 2. Optical setup of TGPLS from (a) top and (b) side views, where LF and SF are long- and short-pass filters. (c) Generation of ⃗ from gate vectors corresponding ⃗ to kG+ grating k vectors spanning kG− to gate beams focused with an internal angle of φ and cross at an angle ⃗ by ⃗ to kS+ of θ. (d) Phase-matched diffraction of signals spanning kS− using the range of grating k vectors in panel c.

(FROG) for characterizing ultrafast laser pulses.32,33 In TGPLS (Figure 1c), ultrafast PL spectra are sampled via diffraction from a transient phase grating generated via the interference of two ultrafast gate pulses, where the optical Kerr effect modulates the refractive index of the medium.34 Background PL is eliminated by spatial filtering, and broadband gating results from a focusing geometry that maximizes the range of available grating k vectors and a gating mechanism that has high tolerance to phase mismatch. As we demonstrate for a range of organic chromophores exhibiting ultrafast broadband dynamics, the detection bandwidth of TGPLS easily spans the visible region and is expected to extend into the UV and near-infrared (IR). Gating does not require polarizers, and in line with theoretical expectations, gating efficiencies in TGPLS are comparable to Kerr gating and broadband upconversion. Figure 2a,b shows the layout of the TGPLS system, which features a boxcar35 geometry to ensure spatial separation of the gated PL signal from the ungated background and the residual gate beams. Fused silica with highly polished surfaces is used as the gate medium owing to its combination of third-order susceptibility from a fast (electronic rather than nuclear) response, extremely low PL background, and UV-IR transparency, making it a good substrate (among many other possibilities) to induce the optical Kerr effect with amplified laser pulses. We first consider the TG efficiency, which in the low diffraction limit, can be estimated as η = (πn1d/λ cos θ)2,36 where n1 is the refractive index modulation depth, d is the grating thickness, and λ and θ are the incident wavelength and angle, respectively. We estimate a laser-induced refractive index modulation depth of ∼4 × 10−5 based on peak gate intensity of 160 GW/cm2 and nonlinear refractive index of fused silica 2.48 × 10−7 cm2/GW37). The appreciable modulation depth results from the TG gate using the on-diagonal elements of the χ(3)

tensor when gate beams are fixed to the same polarization as each other and the input fluorescence. Combining this estimate with our geometrical parameters results in an estimated ∼7% gate efficiency at 500 nm. We measured a diffraction efficiency of ∼5% for a 500 nm, 100 fs pulse, which agrees very well with the estimate and is comparable to broadband upconversion38 and optical Kerr gating.10,39 Compared with TGPLS using the same gate medium, optical Kerr gating requires higher gate intensities (and thus higher gate backgrounds) to compensate for the lower off-diagonal matrix elements associated with mixing polarization states in the χ(3) tensor.40 Because the Kerr-induced refractive index modulation interacts equally with all wavelengths, the broad bandwidth of TGPLS can be understood by considering the focusing geometry. A grating vector, kG⃗ , is generated as the difference between two interfering gate k vectors (Figure 2c). Upon interacting with the input signal kI⃗ , the diffracted signal kS⃗ is generated, whereby the phase-matching condition (kG⃗ + kI⃗ = kS⃗ ) satisfies the Bragg condition of the grating (Figure 2d). If the TG is formed via the interference of two plane wave gate beams, phase matching is only achieved for the degenerate case because at the boxcar incident angle, only an input signal with the gate wavelength matches the Bragg angle for kG⃗ . Here the gated bandwidth would match that of the gate beams. Two further considerations can significantly enhance the gated bandwidth. First, the tolerance to phase mismatch can be assessed using the coupled mode theory developed to describe volume phase gratings. The fwhm bandwidth centered at the Bragg condition can be estimated as Δλ1/2 ≈ λ cot θ(Λ/d).41 In our setup, where the grating thickness d = 1 mm, the grating period Λ = 13.5 μm, and the incident angle θ ≈ 2.5°, this is enough to 1733

dx.doi.org/10.1021/jz5006362 | J. Phys. Chem. Lett. 2014, 5, 1732−1737

The Journal of Physical Chemistry Letters

Letter

Figure 3. (a) TGPLS surface of a white light continuum pulse generated by focusing weak 800 nm pulses in a 1 mm thick CaF2 substrate placed at the sample position. (b) TGPLS spectrum extracted from panel a, with the time response at 550 nm shown as an inset. (c) Chirp-corrected TGPLS surface of β-carotene in hexane following 480 nm excitation (10 nJ/pulse). (d) Spectrally integrated kinetics from panel c fitted to a 150 fs decay convoluted with a Gaussian IRF, with the 0−100 fs spectrum shown as an inset. (Note that the excitation filter obscures the blue edge.) (e) Spectra at various time delays after 400 nm excitation (1.7 nJ/pulse) for a dilute solution of P3HT in chloroform. (f) Spectral relaxation for the same data in panel e represented by the time-dependent spectral peak from femtosecond to nanosecond time scales.

boost the expected bandwidth to Δλ1/2 ≈ 200 nm, centered at λ = 800 nm. Second, extra gate k vectors are provided by focusing the gate beams,33 which is already required to achieve sufficiently high peak power to modulate the refractive index of the gate medium. As illustrated in Figure 2c, gate beam k vectors spanning an angle of φ can generate grating vectors spanning ⃗ with the same absolute value of kG⃗ and with variable ⃗ to kG− kG+ components in the z direction. It is noted that pairing gate vectors in this way results in a broader spectrum of grating k vectors than is obtained when gate k vectors with the same z component are paired to generate parallel grating k vectors with variable length. Thus, tightly focused gate beams transiently generate a multiplexed volume phase grating with a superposition of grating k vectors that satisfy phase matching for a range of signal wavelengths, as illustrated in Figure 2c,d. Applying this analysis to the geometry in Figure 2a, the ratio of ⃗ ) to the degenerate ⃗ (kS− the blue (red) edge signal k vectors kS+ ⃗ /kS⃗ = sin(θ/2)/sin((θ ∓ φ)/ k vector kS⃗ can be expressed as kS± 2). The phase-matched bandwidth for our configuration therefore spans from 340 to 1300 nm and results in spectral resolution of a gated spectrum with a given input angle. We observe that the position of gated signals depends on their wavelength, which is evidence that grating k vectors with different z components are generated as described. Although the spectral resolution of TGPLS ultimately comes from redispersing the entire gated signal in a spectrometer, a beneficial consequence of the large range of available grating k vectors is a wider signal acceptance angle. The combination of generating a multiplexed volume grating and tolerance to phase mismatch is expected to produce a broad spectral response spanning the visible region and beyond. We first employed the TGPLS setup to gate white-light supercontinuum pulses to test its detection bandwidth and time response. Figures 3a,b shows that TGPLS resolves the

supercontinuum spectrum throughout the visible region, limited below 400 nm only by the optics not optimized for the UV and beyond 725 nm by a short-pass filter. The ultrafast time response at a given wavelength reflects the 200 fs time resolution, and the pulse chirp is clearly resolved in the surface, in line with previous TG-based characterization of optical pulses.33 TG-FROG measurements with