We present the results of theoretical and simulation studies of the design and performance of a new type of FEL oscillator. This device, known by the acronym RAFEL for Regenerative Amplifier Free-Electron Laser, will be constructed in the space presently occupied by the Advanced FEL (AFEL) at Los Alamos, and will be driven by an upgraded (to higher average power) version of the present AFEL linac. In order to achieve a long-time-averaged optical output power of ∼1 kW using an electron beam with an average power of ∼20 kW, a rather high extraction efficiency η ∼ 5% is required. We have designed a 2 m long undulator to attain this goal; the first meter is untapered and provides high gain while the second meter is linearly tapered in magnetic field amplitude to provide high extraction efficiency in the standard K-M-R manner. Two-plane focusing and linear polarization of the undulator are assumed. Electron-beam properties from PARMELA simulations of the AFEL accelerator were used in the design. A large saturated gain, ∼ 500, requires a very small optical feedback to keep the device operating at steady state. However, the large gain leads to distorted optical modes which require two and three-dimensional simulations to adequately treat diffraction effects. This FEL will be driven by 17 MeV electrons and will operate in the 16 μm spectral region.

We have successfully operated the photoinjector and rf linear accelerator for the Los Alamos APEX free electron laser (FEL) at design energy, average macropulse current, and emittance. The accelerator, which operates at 1.3 GHz, consists of a 6 MeV photoinjector and three standing-wave structures with a total beam energy of 40 MeV. This paper presents performance characteristics of the APEX system. The results show that this technology is capable of providing reliable, high-peak current, ultra-high brightness electron beams.

We present the design and calculated performance for a 20-W XUV-FEL source for 60-nm projection lithography. The compact design features an 85-MeV rf linear accelerator with photoinjector, and a pulsed microwiggler with multifacet resonator optics.

In recent experiments at Los Alamos, we directly observed for the first time free-electron laser (FEL) fundamental wavelengths from 20 to 45{mu}m. Our 1988 facility's demonstrated wavelength span now extends from 9 to 45{mu}m. Upgrades are in progress to increase this span. This wavelength region uniquely complements existing FEL applications facilities. 5 refs., 6 figs., 1 tab.

In recent experiments at Los Alamos, we directly observed for the first time free-electron-laser (FEL) fundamental wavelengths from 20 to 45 /spl mu/m. Our 1988 facility's demonstrated wavelength span now extends from 9 to 45 /spl mu/m. Upgrades are in progress to increase this span. This wavelength region uniquely complements existing FEL applications facilities.

One-dimensional numerical studies have been made of free electron laser oscillators in which the incident electron energy varies (chirps) as a function of time over each micropulse. Optical radiation resonant with such micropulses is chirped in frequency. The highest calculated efficiency (up to 8.1% for wavelengths near 10 μm) has been obtained in cases where the optical pulse at saturation is short compared to the slippage.

Mirror surface quality requirements are theoretically investigated for XUV FEL oscillators. The resonator configuration considered is a grazing incidence ring using metal coatings. Starting at the end of the wiggler, the optical train consists of (1) a grazing incidence hyperboloid to magnify the optical beam expansion, (2) a multifaceted mirror to retroreflect the beam (the multifaceted mirror is composed of several flats and an off-axis paraboloid to collimate the expanding beam), (3) a scraper to output couple the optical energy, and (4) a second multifaceted mirror and hyperboloid pair to focus the optical beam to the wiggler entrance. For this resonator configuration, we evaluate FEL performance degradation due to mirror surface imperfections which can result from polishing errors and thermal distortion. The concept of mode matching is applied to the resonator design. Mode matching in an FEL oscillator is defined as despacing (translating) the paraboloidal mirrors to match the optical properties of the resonator to the FEL gain region (wiggler).

Computer simulations of FEL lasing differ in the degree to which they approximate real experiments. One of the FEL codes used extensively at Los Alamos takes account of the features of each electron micropulse and follows the growth and saturation of the optical micropulse. With no additional adjustments, this code displays the development of sidebands and demonstrates their control when optical filters of various kinds are used. Other codes that do not include a description of the micropulse do not automatically display sidebands but need to have artificial noise of some kind added. This is not unexpected because sidebands are generated by an FEL instability; instabilities, in general, need some kind of initiating disturbance.

This paper consists of two parts: the conceptual design and optical performance characteristics of a grazing angle-of-incidence ring resonator utilizing multifaceted metal mirrors for use with a 50 nm rf-linac-driven XUV FEL oscillator; and electron beam and wiggler requirements for a self-amplified spontaneous emission (SASE) amplifier to produce high power in the 20 to 40 nm wavelength range. The basis for these studies is the 3-d FEL simulation code FELEX which is used to derive tolerances on mirror figure and thermal distortion, alignment sensitivity, and alternative output coupling methods. The sensitivity of the output characteristics of an XUV FEL SASE amplifier to wiggler field errors is also studied. 17 refs., 8 figs., 2 tabs.

A comparison between some of the measurements made with a uniform wiggler during the Los Alamos free electron laser oscillator experiment and the results of one-dimensional pulse propagation calculations using the mathematical model of Colson and Ride will be presented. Small-signal gain data will be discussed. Calculated output power vs optical resonator length will be compared with data. Time-integrated optical spectral data, which show clear evidence for Raman sidebands, will be compared with calculated spectra as a function of resonator length. The calculated electron energy distribution function, using the measured electron-beam micropulse shape, and the associated extraction efficiency will be compared with measurements.

Operation of free-electron lasers at long optical wavelengths (≥600 nm) has now been successfully demonstrated at several laboratories. To operate a free-electron laser at shorter wavelengths imposes constraints on the brightness of the electron beam which are difficult to achieve. Until recently, it was perceived that only an electron storage ring could satisfy these beam requirements. However, our previous 1-D theoretical calculations revealed that modest improvements in the emittance available from rf-linear accelerators would be sufficient to allow operation of an FEL in the XUV spectral range. We shall present new theoretical results for the design of a linac-driven XUV FEL derived from an improved simulation model. The model is fully three-dimensional in its treatment of the undulator magnetic field, the optical radiation field, and the motion of electrons in a finite-emittance beam. Furthermore, the model computes self-consistently the motion of the electrons and the amplification, diffraction, and the refraction of the light within the undulator magnet. Propagation of the optical beam and reflection at the mirrors of the optical resonator are incorporated in the model so that a complete laser oscillator solution can be generated. The computed performance parameters of a particular XUV FEL oscillator design will be compared with the output of synchrotron radiation sources.

In an rf‐linac‐driven XUV free‐electron laser oscillator, the gain depends on the details of the shape of the electron beam’s phase‐space distribution, particularly the distribution of electrons in the transverse (to the direction of propagation) position and velocity coordinates. This strong dependence occurs because the gain in this device is inhomogeneously broadened. Our previous theoretical studies have assumed that the transverse phase space distribution is a product of uncorrelated Gaussian functions. In the present work, we present the results of a theoretical study of the gain for non‐Gaussian phase‐space distributions. Such distributions arise either from a better representation of the electron beam from an rf‐linac or from an emittance filter applied to the beam after the linac.

In a Compton-regime free electron laser a relativistic electron moves in the combined potential (ponderomotive potential) formed by the optical radiation field and the static magnetic field. If the energy of the electron, the wavelength of the light, and the period and strength of the static magnetic field are connected by the well-known resonance condition, the electron may exchange part of its energy with the radiation field during its transit through the interaction region (wiggler magnet). Whether the electron gains or loses energy is determined by the phase of its transverse motion with respect to the phase of the optical field. Satisfaction of the resonance condition implies that the relative phase between the light and the electron will vary slowly during the interaction time, thus allowing the possibility of net energy transfer.