There is a large interest in the production of high power, ultra-short, 1 fs or less, coherent X-ray pulses, for atomic physics and other applications. However, the present design of X-ray SASE FELs leads to an X-ray pulse about 100 times longer. The time structure of the X-ray pulse is determined by several factors, mainly the electron bunch length and the FEL cooperation length. Until now all FELs have been working in the regime where the electron bunch length is much larger than the cooperation length. In the case of an X-ray SASE FEL this means that the X-ray pulse consists of a series of spikes, a few hundred in the case of the LCLS, and the bunch length is of the order of 200 fs. Several methods to reduce the bunch length to the 10 - 1 fs region have been proposed. These methods are based on electron bunch manipulation to cut the lasing part of the bunch to a fraction of the total length, thus reducing the X-ray pulse length. We are considering here a different method, using ultra-short, very low charge electron bunches, with a length of the order or shorter than the FEL cooperation length. In this case the X-ray pulse length after amplification in the undulator is a few times the electron bunch length. Our simulations show that in the case of LCLS and SPARX, operating with a 1 pC electron beam, we obtain X-ray pulses as short as 300 as, transform limited in the SPARX case, and almost transform limited in the LCLS case. Compared to other modes of producing short X-ray bunches this one offers the advantage of very small background signal, since all electrons are used to lase, and the charge is very small.
Read full abstract