photon imaging. In practical terms, the dwell time –the illumination time required per pixel to yield an image with a high enough signal-to-noise ratio for subsequent analysis- limits the acquisition of a full frame to a time period that is usually hundreds of millisecond or longer. On the other hand, current high-gain femtosecond amplifiers running at kilohertz repetition rate usually supplies pulse energies (of several milijoules) that exceed by far the amount of energy needed in most of the applications where they are
Figure 5.7 Normalized spatiotemporal light intensities for the 0, +1, +2 and +3 diffraction orders coming from the diffraction grating and focused by an achromatic lens. In the left column the compensated measurements taken with the DCM are shown and compared to those without DCM (right column).
88
used. In particular, for nonlinear microscopy, 90-95% of the light is discarded. Therefore, some experiments have been performed in the fields of multiphoton microscopy to improve the temporal resolution by splitting the excitation beam into multiple beamlets by means of DOEs encoded onto a SLM. In this way, the incoming laser beam is shaped into a user-defined light pattern [23, 143] making possible a scanless two-photon microscopy. However, as previously commented, splitting a short pulse by means of a DOE is problematic because of wavelength dependence of the diffraction phenomenon. Light patterns in the sample appear naturally blurred, thus reducing image resolution, and temporally stretched, which dramatically influences the efficiency of the TPA process. To get an idea, the two-photon signal intensity is reduced to 0.5% of the available signal for a spatial frequency of 30 lp/mm encoded onto the DOE [148], due to spatial and temporal stretching of the pulsed beam at the sample plane.
Taking all this into account, we have proposed a proof of concept optical system for efficient generation of wide-field fluorescence signals in two-photon microscopy, reducing spatiotemporal distortions associated with the use of ultrashort light sources and DOEs [III]. The main idea is to encode properly designed CGHs onto a SLM in order to generate user- defined extensive light patterns in two dimensions. To counteract spatiotemporal distortions, we propose the use of the DCM, which is inserted between the SLM and the sample. It supposes a dispersion first- order correction, obtaining a two-photon fluorescence signal with higher spatial resolution and improved efficiency compared to the case without DCM.
Our experimental setup is shown in Fig. 5.8. The amplifier output of our FemtoPower compact Pro laser is used as pulsed source. The pulsed laser beam impinges by means of a beam splitter onto a Fourier CGH encoded onto a phase-only SLM. The CGHs are calculated by using the well-known Gerchberg–Saxton IFTA, but carried out in two stages as proposed by Wyrowski [43]. The dashed box in Fig. 5.8 is the DCM. As it was introduced before, it is made up of an achromatic lens L1 (focal length 𝑓1=300mm) coupled to a diffractive lens pair, DL1 and DL2. The focal
89
lengths of DL1 and DL2, for the central wavelength of the laser are 𝑓0,1 = −𝑓0,2= −150 𝑚𝑚. After the DCM, in order to properly excite the fluorescence signal in Rhodamine B (RB), we use a telescope with a refractive lens L2 (focal length 𝑓2=100mm) and a 20x microscope objective MO1 with focal distance 10 mm. To observe the fluorescence signal, the RB plane is imaged onto a conventional CCD sensor by means of a 50x microscope objective MO2. We place a suited filter F before the CCD camera to prevent from propagation of the infrared signal.
Proof of concept is provided by sending different light patterns to the RB sample. In this sense, several CGHs are generated to create user- defined laser profiles. Bike, spiral and a 5x5 matrix shapes are generated, and their computer reconstructions shown in Figs. 5.9a, 5.9d and 5.9g. The spatial spectrum of the samples ranges from frequencies of around 25 lp/mm to about 38 lp/mm. Within this interval, the effects of theoretical spatial and temporal broadening enlarge foci by a factor of 25 to 40 times in both spatial and temporal domains with respect to that of the unaffected zero diffraction order (SLM acting as a mirror).
In a first configuration, the DCM is removed from the setup and the fluorescence signal recorded. This is shown in Figs. 5.9b, 5.9e and 5.9h. As expected, the fluorescence pattern is blurred due to the spatial spreading of each frequency contained in the finite spectrum of the femtosecond pulse.
Figure 5.8 Schematic of the diffractive-refractive optical system used to improve TPA and fluorescence emission in RB.
90
Moreover, in order to detect significant signal from the RB, average beam power is set to 5 mW. Temporal stretching causes a loss of fluorescence signal. From Eq. (2.25) it is clear that the number of emitted photons at the sample depends inversely on the pulse width.
In a second configuration, the DCM is placed and fluorescence signal recorded again as shown in Figs. 5.9c, 5.9f and 5.9i. As done in the previous configuration, in order to ensure wide-field fluorescence signaling all over the sample plane for the repetition rate and the pulse width of our laser system the average power has to be adjusted to 3mW. The dispersion compensation abilities of the DCM preserve the spatial and temporal width of the laser pulse at the sample plane, and consequently the fluorescence
Figure 5.9 Light patterns used for TPA in RB. Left column (a, d and g) corresponds to computer reconstructions of the encoded holograms. In the central column (b, e and h), the recorded light pattern without DCM are shown. Right column (c, f and i) corresponds to the improved fluorescence signal recorded when DCM is implement in the optical system.
91
signal is obtained for a 40% less energy than in the previous case. The transverse spatial resolution for the CGH reconstruction is maintained due to the spatial chirp compensation capacity of the module, and so the blurring distortion is eliminated in a big degree.
In summary, this method evidences a great correction of both angular and temporal dispersion, which in practical terms means that it applicability can be relevant in techniques such as nonlinear microscopy or material machining with high bandwidth lasers.
93