Many lasers allow the operator to tune or change the output wavelength from the UV to the IR on demand.
Ian Read, MKS/Spectra-Physics
Applications facilitated by tunable lasers fall into two categories: situations in which one or more discrete wavelengths are not available from any single- or multiline fixed-wavelength laser, or ones in which the laser wavelength must be tuned continuously during the experiment or test, such as spectroscopy and pump-probe experiments.
Many types of tunable lasers can produce tunable continuous-wave (CW), nanosecond, picosecond, or femtosecond output. Their output characteristics are governed by the laser medium used.
A fundamental requirement of tunable lasers is their ability to emit over an extended wavelength range. Special optical elements are used to select a specific wavelength or wavelength band from within this range. Various types of materials are used to generate tunable laser light, with the most common being organic dyes or crystals such as titanium sapphire (Ti:sapphire). In both cases, argon ion (Ar+) or frequency-doubled neodymium ion (Nd3+) pump lasers are employed due to efficient absorption at ~490 nm.
Dye molecules can be used to generate wavelengths in the UV through the visible range. However, accessing a wide tuning range requires changing between many different dye molecules, which can be cumbersome. Solid-state lasers provide wide tunability using a single laser gain material, such as dielectric crystalline, to amplify light power. This eliminates the need to perform lengthy dye changes.
Ti:sapphire has emerged as the leading tunable laser material, given its broad (680 to 1100 nm) emission profile that can be continuously tuned and its output that can be upconverted into the UV-VIS spectral range or downconverted into the IR spectral region. These features enable many applications in both chemistry and biology.
Figure 1. Schematic of the CW standing-wave laser based on Ti:sapphire. A birefringent tuning element is shown.
Tunable CW standing-wave lasers
Conceptually, the CW standing-wave laser is the simplest laser architecture. Consisting of a high reflector, a gain medium, and an output coupler (Figure 1), this laser provides CW output using various laser gain media. To achieve tunability, the gain medium is chosen to cover the wavelength range of interest.
Many fluorescent dyes are available to shift the laser wavelength to the desired region. Dye lasers have the advantage of covering a broad wavelength range across the UV-VIS spectrum but also the disadvantage of having narrow wavelength tunability using a single dye/solvent combination. The solid-state Ti:sapphire laser provides the advantage of having a broad wavelength lasing range using a single gain medium but the disadvantage of operating in the near-IR region of the spectrum from 690 to 1100 nm.
For both gain media, wavelength tuning is accomplished using passive wavelength-stabilizing elements. The first of these elements is a multi-plate birefringent or Lyot filter. This optical element modulates the gain by providing high transmission at a specific wavelength, thereby forcing the laser to operate at that wavelength.
Tuning is accomplished by rotating this birefringent filter. Although simple, CW standing-wave lasers allow for many longitudinal laser modes. This produces linewidths on the order of 40 GHz full width half maximum (<1.5 cm–1), which can be a limiting factor for some applications, such as Raman spectroscopy. To achieve narrower linewidths, a ring configuration is required.
Tunable CW ring lasers
Ring lasers have been used since the early 1980s to achieve tunable CW radiation from a single longitudinal cavity mode, with spectral bandwidths possible in the kilohertz region. Similar to their standing-wave counterparts, tunable ring lasers are available that use both dye and Ti:sapphire lasing media. The former is capable of very narrow <100 kHz linewidths, while the latter provides <30 kHz linewidths. The tuning range spans from 550 to 760 nm for the dye laser and from 680 to 1035 nm for Ti:sapphire version. Both outputs can be frequency doubled for access to the UV region of the spectrum.
According to Heisenberg’s uncertainty principle, as the energy is more precisely defined, the less precisely the pulse width can be determined. For the standing-wave CW laser, the cavity length defines the number of allowed energies as discrete longitudinal modes. When the cavity length is short, the number of allowed longitudinal modes increases, which leads to a broader and less-defined output linewidth.
In the ring configuration, the laser cavity can be treated as an infinite cavity length, and the energy will be precisely defined. Only a single longitudinal mode will be present in the cavity. To achieve the single-mode operation condition, several optical elements are required (Figure 2).
First, a Faraday isolator is inserted into the cavity to ensure that the intracavity photons always follow the same path. Intracavity etalons are used to further minimize the output linewidth. In the ring configuration, unlike with the standing-wave laser cavity, there are no end mirrors. The photons continuously circulate through the laser cavity. Second, the cavity length must be stabilized to correct for any mechanical changes caused by environmental fluctuations, such as heat or vibrations.
To achieve ultranarrow spectral bandwidths, the cavity must be stabilized, using one of two methods: employing mechanical piezo-driven mirrors to stabilize the cavity length for kilohertz response times, or using electro-optical (E-O) modulators to achieve megahertz response times. Several specialized laboratory setups have shown that spectral bandwidths are measurable in hertz. The key element for determining the ring cavity spectral resolution is the external frequency reference cavity. As shown in Figure 2, the reference cavity is used to generate the signal necessary to stabilize the laser cavity length. This external cell must be insulated from environmental fluctuations originating from temperature, mechanical vibrations, and acoustic noise. The reference cell should be well separated from the ring laser cavity itself to avoid inadvertent coupling between the two. The reference signal is processed using the Pound-Drever-Hall method.
Figure 2. The optical layout of a ring Ti:sapphire laser with external reference cell.
Mode-locked, quasi-CW lasers
For many applications, precisely defined temporal characteristics of the laser output are more important than a precisely defined energy. In fact, achieving temporally short optical pulses requires a cavity configuration in which many longitudinal modes resonate simultaneously. When these circulating longitudinal modes have a fixed-phase relationship in the laser cavity, the laser becomes mode locked. This results in a single pulse oscillating in the cavity, with a period defined by the laser cavity length.
Mode locking can be achieved either actively, using an acousto-optic modulator (AOM), or passively through Kerr-lens mode locking. The former, popular in the 1980s, utilizes an intracavity AOM as a transient shutter, opening and closing at half the cavity length frequency. Using this method, pulses in the hundreds of picoseconds can be achieved. Over the past several decades, scientific applications require improved time resolution, thus creating a need for shorter pulses.
Synchronously pumped dye lasers emerged as a method for tuning the center wavelength and shortening the optical pulse by an order of magnitude, in the range of tens of picoseconds. To achieve this state, the dye laser cavity must have the same cavity length as the mode-locked pumping laser. The pump and dye laser pulses meet at the gain medium to generate the stimulated emission from the dye molecule. The laser output is stabilized by adjusting the dye laser cavity length. The synchronous pump configuration can also be used to drive optical parametric oscillators (OPOs) (discussed below).
Ti:sapphire mode-locked lasers are an example of passive Kerr-lens mode locking (Figure 3). In this method, pulses are generated through the gain modulation and the intensity-dependent refractive index of Ti:sapphire.
In principle, as the pulse propagates through the gain medium, the peak intensity is higher when the pulse is present. This creates a passive lens that focuses the pulse beam more tightly and extracts the gain more efficiently until no gain is available to support simultaneous resonance of the CW mode in the cavity. A mechanical perturbation to the cavity is used to cause an intensity spike to initiate mode locking. With this method, pulses as short as 4 fs have been generated using Ti:sapphire.
Figure 3. In a mode-locked Ti:sapphire laser, the center wavelength is tuned by moving a tuning slit, located between two dispersive prisms.
It is important to note that over 300 nm of bandwidth can be combined into a single pulse. In accordance with Heisenberg’s uncertainty principle, shorter pulses require more longitudinal modes. Because of this, the laser cavity must have sufficient dispersion compensation from the cavity optics to maintain the phase relationship necessary for stable mode locking. Shown in Figure 3, compensating prisms are added to the cavity to ensure a constant phase relationship. Using this method, pulses as short as 20 fs can be achieved. To generate shorter pulses, higher-order contributions to the dispersion must also be compensated. This compensation is accomplished using special mirrors that introduce optical chirp to maintain the phase relationship necessary for stable mode locking.
Because Kerr-lens mode locking is most efficient with shorter pulses (higher intensity), this method mostly applies to generating femtosecond pulses. In the intermediate range between 100 fs and 100 ps, a hybrid approach called regenerative mode locking can be used. This method employs both an intracavity AOM and the Kerr effect. The AOM drive frequency is derived from a real-time measurement of the cavity repetition frequency, and the amplitude depends upon the pulse duration. As the desired pulse width increases and the Kerr effect decreases, the stabilizing AOM amplitude increases to support mode locking. As a result, regenerative mode locking is capable of providing stable, tunable output across a wide range from 20 fs to 300 ps, while using a single laser system.
In the late 1990s, regenerative mode locking enabled the first tunable, one-box computer-controlled Ti:sapphire laser. This innovation made the technology more accessible to a wider range of researchers and applications. Advancements in multiphoton imaging have been largely driven by technological advancement. Femtosecond laser pulses are now accessible for biologists, neuroscientists, and physicians. For example, over the years, several enhancements have made the Ti:sapphire laser ubiquitous in the bioimaging field.
Ultrafast ytterbium lasers
Despite the utility of Ti:sapphire, some bioimaging experiments demand longer wavelengths. A typical two-photon absorption process is initiated by a 900-nm photon. Because longer wavelengths mean less scattering, biological experiments requiring a deeper imaging depth are more efficiently driven by longer excitation wavelengths.
It is also important to consider the wavelength of the subsequent fluorescence photon from the dye attached to the biological sample. This fluorescence photon is typically emitted in the 450- to 550-nm wavelength range, which will be more susceptible to scattering. For this reason, several fluorescent markers have been developed that absorb further into the infrared wavelength range. To address this requirement, one-box, computer-controlled, synchronously-pumped OPOs driven by 1045-nm ytterbium lasers were developed. This new class of lasers produces output across the 680- to 1300-nm range. For multiphoton imaging, this architecture offers a significantly higher performance alternative to Ti:sapphire.
Ultrafast amplifiers
The above examples produce ultrafast pulses in the nanojoule energy range. However, many applications require higher-energy tunable light sources. Because wavelength conversion is a nonlinear process, efficiency depends on available energy. For these applications, several technologies are used to increase the energy and tunability of ultrafast lasers.
Amplification of ultrafast pulses falls into two categories: multipass and regenerative amplifiers. The former has the advantage of achieving very high energies (100 mJ) with very low background, but the repeated passes through the amplification stage can degrade the output beam quality. For this reason, regenerative amplification is the preferred method for generating pulse energies in the microjoule or millijoule range.
In general, ultrafast-pulse amplification is achieved through the chirped pulse amplification method (Figure 4). The process starts with a mode-locked oscillator with a femtosecond pulse duration — the seed laser. It is important for the seed laser to have enough bandwidth so that the pulse duration can be temporally stretched or chirped in time. Optical chirp is generated because different colors of light travel at different velocities through optical materials. In general, red wavelengths will propagate faster than blue wavelengths. The stretcher grating introduces positive-chirp red ahead of blue, for example, to separate the wavelength components in time and space. Stretching is necessary to reduce the intense peak power of a millijoule-level femtosecond pulse. Following stretching, the nearly 300-ps pulse is directed to a secondary, regenerative laser cavity. The final step is to use a second grating to introduce negative chirp and reconstruct the amplified pulse. This process is shown schematically in Figure 4.
Most regenerative amplifiers today use Ti:sapphire, but other gain media, such as ytterbium, are gaining popularity. In both cases, the amplifiers have narrow tunability, about 780 to 820 nm for Ti:sapphire, which limits their utility when applied to spectroscopy. To overcome this limitation, several frequency-conversion options are available.
Figure 4. A schematic representation of chirped pulse amplification.
Harmonic frequency conversion is the simplest method for tuning the wavelength from either an ultrafast oscillator or an ultrafast amplifier system. In principle, the incident photon is upconverted to an integer multiple of the fundamental frequency. For Ti:sapphire — with a fundamental tuning range from 700 to 1000 nm — the second harmonic translates to a tuning range of 350 to 500 nm, the third harmonic to 233 to 333 nm, and the fourth harmonic to 175 to 250 nm. In practice, the fourth harmonic range is limited to 200 nm due to harmonic crystal absorption. For applications requiring wavelengths outside this range, parametric conversion options are necessary.
Ultrafast OPOs and OPAs
While the pulsed ultrafast output can be frequency doubled and even tripled, the 700- to 1000-nm tuning range of Ti:sapphire leaves wavelength gaps across the UV-VIS and IR spectral regions. For experiments that require ultrafast pulses in these “gap” spectral regions, parametric downconversion is necessary. This method converts a single high-energy photon into two low-energy photons: signal and idler (Figure 5).
The energy partitioning between these two photons can be user-configured. In a typical parametric configuration based on Ti:sapphire, the incident 800-nm photon can be continuously tuned from about 1200 to 2600 nm. Because parametric downconversion is a nonlinear process, conversion efficiency can become an issue. To overcome this limitation, optical parametric oscillators (OPOs) are used in the nanojoule energy regime, and optical parametric amplifiers (OPAs) are used with millijoule energy levels.
Figure 5. A schematic representation of parametric downconversion.
Inside the OPO cavity, the light consists of a short pulse that travels back and forth through the cavity. However, unlike the dye laser configuration described above, the active medium is a nonlinear crystal that cannot store gain. The OPO crystal only converts photons when a pump pulse is present. The successful operation of an ultrafast OPO requires the pulses from the pump source to arrive at the crystal at the same time as the idler and signal photons that are circulating around the OPO cavity. In other words, the fixed-wavelength Ti:sapphire laser and the ultrafast OPO must have the exact same cavity length.
The layout of a typical ultrafast OPO is shown in Figure 6. Phase matching and cavity length are automated to select the desired wavelength and ensure the cavity round-trip time at that wavelength remains at 80 MHz, which is the same as the Ti:sapphire pump laser. In this example, the OPO is driven by the second harmonic of the Ti:sapphire pump laser. The resultant 400-nm beam generates signal and idler outputs with a total wavelength coverage from 490 to 750 nm (signal output) and 930 nm to 2.5 µm (idler output), with pulse widths below 200 fs. When coupled with the tuning range of the Ti:sapphire fundamental at 690 to 1040 nm, the system covers the wavelength range from 485 nm to 2.5 µm. Typical applications include soliton studies, time-resolved vibrational spectroscopy, and ultrafast pump-probe experiments.
Figure 6. In a synchronously pumped optical parametric oscillator (OPO), the center wavelength is changed by adjusting the phase-matching angle of the nonlinear crystal.
The OPA makes use of the same nonlinear optical process, but because the pump pulses have a higher peak power, an optical resonator is not required to achieve efficient wavelength conversion. A small part of the beam from an ultrafast amplifier is focused into a sapphire plate to generate a white-light continuum. This is used to seed the OPA crystals, which are usually barium borate crystals, pumped by the rest of the ultrafast amplifier beam. This is also where the beam undergoes orders of magnitude amplification at the signal and idler wavelengths in a single pass. The center wavelength of the output is again controlled by the phase-matching conditions of the crystal, and the spectral bandwidth is generally determined by the bandwidth of the pump and seed beams or the acceptance bandwidth of the crystal.
This type of OPA can be operated in the femtosecond or picosecond range, with pulse energies as high as several millijoules per pulse. At these energy levels, the resulting signal and idler beams can be converted to their harmonics or through sum-frequency and/or difference-frequency mixing.
An OPA that is pumped with millijoule pulse energies is capable of producing photons that span from the deep UV at 190 nm through the far-infrared spectral regions. These devices facilitate many spectroscopic applications, such as transient absorption spectroscopy, fluorescence upconversion, 2D infrared spectroscopy, and high harmonic generation.
Tunable lasers are now employed in many important applications, ranging from fundamental scientific research to laser manufacturing and life and health sciences. A wide scope of technologies is currently available. Starting from the simple CW tunable systems, whose narrow linewidth is used for high-resolution spectroscopy, molecular and atomic trapping, and quantum optics experiments — providing key information to modern researchers.
The more complex ultrafast amplifier systems utilize high-energy, picosecond, and femtosecond laser pulses to generate laser output from the UV through the far infrared. These ultrafast lasers are critical for understanding high-energy physics, high harmonics, and transient spectroscopies. The broad tuning range means that the same laser system can be used to study an infinite range of experiments in electronic and vibrational spectroscopy. Today’s laser manufacturers offer turnkey solutions that provide access to a laser output that spans more than 300 nm in the nanojoule energy regime. The more complex systems span an impressive range from 200 to 20,000 nm in the microjoule and millijoule energy regime.