Femtosecond illumination moves from technology to microscopy

As with general laser applications, wavelength tuning of ultrafast lasers has had to become a hands-free, push-button operation.

Aug 1st, 2004
Th 155225

As with general laser applications, wavelength tuning of ultrafast lasers has had to become a hands-free, push-button operation.

Multiphoton-excitation (MPE) microscopy places stringent demands on laser performance, but the biologists who use it generally have limited laser expertise. To overcome this lack of expertise, a new generation of ultrafast lasers with "hands-free" operation and specifically optimized output characteristics has been developed.

In MPE, a near-IR femtosecond laser beam is focused to a small spot within the sample by an objective of a high-numerical-aperture microscope. In a small region around the beam waist, the peak laser intensity is high enough to drive multiphoton absorption, in which two or three photons are simultaneously absorbed by molecules whose conventional (one-photon) absorption is in the green, blue, or even UV region of the spectrum. Resultant fluorescence is collected by the same microscope optics and detected by a low-noise photodetector. Scanning the focused laser across an x-y plane in the sample provides a two-dimensional, cross-sectional image; and a three-dimensional image can be built from a series of two-dimensional slices taken at different depths along the z-axis.

FIGURE1. A two-photon image excited at 930 nm using DiI staining lipids in NIH 3T3 cells was taken using a Chameleon laser with Zeiss LSM 510 META microscope system.
Click here to enlarge image

Since it was first demonstrated in 1990, MPE has grown to become one of the most popular methods for high-resolution, three-dimensional imaging of biological tissue (see Fig. 1). Moreover, because it involves only limited absorption of the laser light, it represents a relatively benign imaging environment, making it compatible with live cell (in vivo and in vitro) studies. The technique has also been expanded beyond two-photon excitation of fluorophores. Variant processes include, three-photon fluorophore excitation, CARS (coherent anti-Stokes Raman-shifted) imaging, harmonic generation to image transparent membranes, and multiphoton excitation of fluorescence from endogenous (native) cellular materials.

Early MPE microscopy experiments were performed using full-featured femtosecond lasers, such as the Coherent Mira, which provide numerous options, adjustments, and flexible performance characteristics. However, these lasers had to be carefully aligned with and pumped by a continuous-wave (CW) green laser, presenting a level of performance and complexity unnecessary for MPE and often presenting a hindrance to the typical biologist user.

Ultrafast-laser manufacturers, therefore, now support MPE with a new generation of simplified femtosecond lasers, such as the Chameleon XR, that are factory-sealed, "one-box" lasers. The solid-state pump laser and the Ti:sapphire laser are integrated in a compact, monolithic laser head, with all performance controlled by a single computer interface. With no need to physically align or manually adjust the laser optics, even a laser novice can use this type of turnkey tool in an MPE experiment. Moreover, the small footprint of this type of laser affords simple integration with optical microscopes designed to support MPE imaging.

Optimizing Pulsewidth

The goal of MPE microscopy is to generate high-resolution, high-contrast images in the shortest possible acquisition time, while minimizing any laser damage to the sample. The multiphoton-excitation process, and hence image contrast, increases nonlinearly with peak laser power. As a general rule, damage due to linear absorption increases with the average laser power because it is a function of the total energy deposited in the sample. The situation is complicated by the onset of damage also related to multiphoton (nonlinear) absorption near the beam waist. Consequently, the optimal MPE laser must deliver peak power high enough to efficiently excite the fluorescence, while remaining below the nonlinear damage limit and maintaining relatively low average power. This can be achieved by delivering pulses at the sample that are as short as possible. But achieving the shortest pulsewidth at the sample does not mean using a laser with the shortest possible pulsewidth because of the effect of group velocity dispersion (GVD) on ultrafast pulses in microscope optics.

In so-called "transform-limited" pulses, fundamental physics dictates that the pulse duration is inversely proportional to the spectral bandwidth of the pulse. When such a pulse passes through transmissive materials, such as glass lenses (and even mirror coatings), the different wavelength components of the pulse travel at different speeds. As a result, pulsewidth is lengthened (see Fig. 2). This process is sometimes referred to as chirping.

FIGURE 2. A GVD in a glass lens or other transmissive optics causes the various wavelength components in an ultrafast-laser pulse to travel at different speeds, thereby stretching the overall pulsewidth. For transform-limited pulses, this stretching is proportionally greater for a shorter pulse because a shorter pulse contains a broader spread of wavelengths.
Click here to enlarge image

The pulsewidth at the sample is a combination of the starting pulsewidth and the amount of GVD stretching, which is proportional to the spectral bandwidth of the input pulse. This means, for a transform-limited pulse, GVD stretching is inversely proportional to the pulsewidth (see Fig. 3). A very short input pulse (say 50 fs) is undesirable because it stretches substantially due to GVD. Conversely, a long input pulsewidth (hundreds of femtoseconds) is undesirable; it will only be stretched a little, but is too long to start with. For a specific value of GVD, there is an optimum input pulsewidth that will result in the shortest final pulsewidth at the sample. Most commercial optical microscopes used for MPE have GVD values between 4500 and 11,000 fs2 at a wavelength of 800 nm. This translates into an optimum laser pulsewidth of 120 to 175 fs.

FIGURE 3. For a given microscope GVD value, there is an optimum input pulsewidth that will result in the shortest final pulsewidth at the sample.
Click here to enlarge image

It is important to note that this calculation is based on a transform-limited pulse. For example, if the laser produces pulses of say 50 fs it is pointless to stretch this before insertion into the microscope. This would deliver the worst of both worlds: the chirped pulse would start off longer but still undergo GVD stretching in the microscope as though it were a 50-fs pulse. The end result would be a very long pulsewidth at the sample.

Broad Tunability

Cell biologists use a wide range of fluorophores, stains, and indicators. Optimum one-photon excitation wavelengths for these materials span from the near-UV through the red. The ideal MPE laser thus needs to have quite broad tunability (see "The need for broad tunability," [below]).

The Chameleon XR covers a range from 705 to 980 nm, based primarily on two design factors: broadband coatings for the cavity mirrors enable reflectivity over the entire range and an automated tuning system maintains its own calibration and beam alignment over long-term use.

Broadband performance requires a multilayer coating and, in conventional coatings, light scatter increases with layer count. In addition, the reflectivity characteristics must be adjusted so that the reflectance bandwidth does not vary too much over the tuning curve. Otherwise the spectral bandwidth, and therefore pulsewidth, would vary as the laser is tuned.

The automated tuning is based on a single prism motion in conjunction with a fixed slit. The wavelength calibration of the prism position is stored in the laser computer to enable highly accurate tuning or random access to any wavelength. The slit width is designed to deliver the spectral bandwidth corresponding to the target laser pulsewidth, as already discussed. With this automated tuning mechanism, the laser cavity can be tuned from 705 to 980 nm with beam movement of less than 1/20 of the beam width.

Output Characteristics

Achieving minimum focused spot size with the microscope's high-NA objective requires filling its aperture with a symmetric TEM00 input beam. This is not trivial because a Ti:sapphire laser oscillator relies on a crystal with facets cut at Brewster's angle, which would naturally produce an elliptical beam. The answer is to design the rest of the cavity optics to cancel out the asymmetry created by the Brewster facets, producing a beam with almost perfect circularity (see Fig. 4).

FIGURE 4. This circular (TEM00) beam profile from a Chameleon XR laser allows the high-NA optics in an MPE microscope to be fully utilized, hence enabling maximum spatial resolution.
Click here to enlarge image

The newest ultrafast lasers for MPE also feature higher average output power. This enables a recently developed technique called multifoci MPE microscopy. Here, the beam is split into multiple beams, which are focused in the sample at the same z depth, but with a fixed x-y separation. This array of beams is then scanned across the sample, with the fluorescence detected by a CCD camera rather than a PMT. Because multiphoton excitation requires a certain threshold power to achieve acceptable signal to noise, this dictates the use of a laser with higher overall power. The latest MPE lasers deliver peak power levels of more than 1.5 W, which meet the needs of these applications. Of all the performance specifications, this proved to be the simplest issue to resolve—by building the one-box laser around a pump laser of higher output power. Traditional MPE users who want to avoid sample damage due to high average power can simply use a filter to limit the amount of laser light reaching the sample.

Ultrafast lasers have been used for many years in various research applications, and the scientists who traditionally performed these investigations were willing, and in some cases happy, to master the intricacies of ultrafast technology. Now, applications outside of pure science, such as MPE microscopy, femtosecond-laser vision correction, and micromachining are maturing. The success of these areas and the growth of others will depend upon the development of new lasers specifically matched to the needs of each application, together with enhanced reliability and ease-of-use.

CHRIS DORMAN is a product-line manager at Coherent Scotland, Todd Campus West of Scotland Science Park, Maryhill Road, Glasgow G20 OUA, Scottland; e-mail: chris.dorman@coherent.com.

The need for broad tunability

The ever-expanding range of fluorophores, stains, and indicators used by cell biologists includes green-fluorescent-protein derivatives to label specific gene products, caged compounds to release calcium ions upon laser excitation, and site-specific fluorophores.

At the shorter end of the Ti:sapphire gain curve, the 705 to 710-nm range is useful for excitation of DAPI (a widely used DNA stain) and Fura. Longer wavelengths (930 to 980 nm) enable three-dimensional imaging at much greater depths in living tissue because light scatter (the limiting factor) decreases at longer wavelengths. These long excitation wavelengths also allow cell biologists to perform optimum MPE imaging of cell and tissue samples with state-of-the-art fluorophores such as YFP and DsRed. In addition, biologists sometimes need to scan wavelengths to study fluorophore response as a function of wavelength.

More in Optics