Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
990 IEEE JOURNAL OF SELECTED TOPICS IN QUANTUM ELECTRONICS, VOL. 9, NO. 4, JULY/AUGUST 2003 Toward Single-Cycle Laser Systems Thomas R. Schibli, Onur Kuzucu, Student Member, IEEE, Jung-Won Kim, Student Member, IEEE, Erich P. Ippen, Fellow, IEEE, James G. Fujimoto, Fellow, IEEE, Franz X. Kaertner, Senior Member, IEEE, Volker Scheuer, and Gregor Angelow Invited Paper Abstract—Few-cycle pulse generation based on Ti:sapphire, Cr:forsterite, and Cr:YAG gain media is reviewed. The dynamics of these laser systems is well understood in terms of soliton and dispersion managed soliton formation stabilized by artificial saturable absorber action provided by Kerr-lens modelocking. These systems generate 5-, 14-, and 20-fs pulses with spectral coverages of 600–1150, 1100–1600, and 1200–1500 nm, respectively. The design of dispersion compensating laser optics providing high reflectivity and prismless operation over this bandwidth is discussed. A novel active synchronization scheme based on balanced optical cross correlation, the equivalent to balanced microwave detection, for synchronization of independently mode-locked lasers is introduced. Its use in synchronizing an octave-spanning Ti:sapphire laser and a 30-fs Cr:forsterite laser yields 300 attoseconds timing jitter measured from 10 mHz to 2.3 MHz. The spectral overlap between the two lasers is large enough to enable direct detection of the difference in the carrier-envelope offset frequency between the two lasers. These are the most important steps in the synthesis of single-cycle optical pulses with spectra spanning 600–1600 nm. Index Terms—Mode-locked lasers, synchronization, ultrafast optics. I. INTRODUCTION O VER THE last four decades, remarkable progress has been achieved in short-pulse generation directly from lasers. The shortest pulses are generated by the excitation and phase locking of many of the longitudinal modes of a laser. In conventional lasers with a typical repetition rate of 100 MHz, up to a few million of these modes can be coupled together to produce pulses as short as 5 fs at a center wavelength of 800 nm, corresponding to less than two optical cycles [1]. Several key developments have led to this result. First of all, we have the development of broad-band laser materials with full-width at half-maximum (FWHM) bandwidth of 100 to 200 nm in the Manuscript received March 3, 2003; revised June 16, 2003. This work was supported in part by the Massachusetts Institute of Technology Lincoln Laboratory under Grant ACC-334, the National Science Foundation under Grant ECS-0119452, the Air Force Office of Scientific Research under Grant F49620-01-1-0084 and Grant F49620-01-1-0186, and by the Office of Naval Research under Grant N00014-02-1-0717. T. R. Schibli, O. Kuzucu, J.-W. Kim, E. P. Ippen, J. G. Fujimoto, and F. X. Kaertner are with the Department of Electrical Engineering and Computer Science and Research Laboratory of Electronics, Massachusetts Institute of Technology, Cambridge MA, 02139 USA. V. Scheuer and G. Angelow are with Nanolayers Optische Beschichtungen GmBH, Rheinbreitenbach, Germany. Digital Object Identifier 10.1109/JSTQE.2003.819108 near infrared, centered around 800, 1300, and 1450 nm, which support the generation of few-cycle pulses. Today, the most important laser materials for few-cycle pulse generation are Ti:sapphire [2], Cr:LiSAF–LiCAF, Cr:forsterite, and Cr:YAG. The second ingredient was the discovery of a broad-band saturable absorber mechanism, Kerr-lens mode-locking (KLM), which works down to pulsewidths of only a few femtoseconds [3]–[7]. Very soon after discovery of the broad-band laser material Ti:sapphire and the modelocking mechanism KLM, the pulsewidth was limited by the means available for dispersion compensation [8] and finally the bandwidth of the available Bragg mirrors [9]–[12]. A remedy for bandwidth limitation and dispersion compensation was achieved by the introduction of chirped mirrors [13]–[15], the equivalent of chirped fiber Bragg gratings. Those mirrors were further refined by taking into account the impedance-matching problem which occurs at the air–mirror interface and the grating structure in the mirror [16]. Dispersion compensation over one octave was finally achieved [17], which led to the generation of 5-fs short pulses. Dispersion compensating mirrors are now widely used to exploit the full bandwidth of several broad-band laser materials. Fig. 1 shows the history of short-pulse generation in terms of pulsewidth achieved for the different laser materials. Fig. 1 suggests that the progress slowed down considerably in recent years as the pulsewidth approached the few-cycle regime. A solution to this dilemma can be achieved by shifting the center wavelength into the UV and XUV regime using high-harmonic generation, as suggested in the mid 1990s [18] and which has been observed experimentally quite recently [19]. Nevertheless, further pursuit of the generation of shorter pulses in the near infrared regime is equally important. Their broad spectra allow novel applications in frequency metrology, since they provide stabilized combs of millions of lines through the stabilization of a single laser. Further, trains of pulses that are phase stabilized provide electric fields with subcycle precision that open up novel experiments in condensed matter physics. Experiments using sub-two cycle pulses interacting with GaAs in the strong field limit revealed interesting dynamics of the two-level system called carrier-wave Rabi flopping [20], [21], where significant population dynamics occur on the time scale of one optical cycle. Other experiments requiring octave spanning spectra are the observation of quantum interferences via single- and two-photon absorption [22]. Very recently, phase controlled low-energy pulses lead 1077-260X/03$17.00 © 2003 IEEE SCHIBLI et al.: TOWARDS SINGLE-CYCLE LASER SYSTEMS 991 the high mirror and low , index materials used for the dielectric (1) Fig. 1. History of mode locking in terms of the years in which shorter and shorter pulses were achieved. Distinct branches pertaining to different laser types are shown. to the observation of a phase sensitive photoelectron emission from Gold-surfaces in a photomultiplier tube [23], which was already predicted to occur earlier in [24] and references therein. These are only a few examples of the new research area called phase senstive nonlinear optics. This paper reviews few-cycle pulse generation and shows a method to synthesize single-cycle pulses by coherent addition of the spectra emitted from two or more of those lasers. In Section II, we review the design of broad-band dispersion compensating laser optics, necessary for few-cycle pulse generation. In Section III, a compact prismless Ti:Sapphire laser and Cr:Forsterite and Cr:YAG lasers are demonstrated emitting octave spanning spectra or few-cycle pulses, respectively. In Section IV, a novel scheme for synchronization of independently modelocked lasers is described based on a balanced optical cross correlator. With this device it is possible to synchronize the ultrabroad-band Ti:sapphire laser with a 30-fs Cr:forsterite laser to within about 1/10 of an optical cycle. This is the first and most important step in the synthesis of single-cycle pulses from different independently mode-locked lasers. Considerable spectral overlap between the Ti:sapphire and Cr:forsterite laser enables the observation of the direct homodyne beat between both lasers. Locking of this beat to zero frequency is the final step for achieving a coherent pulse synthesis approaching the single-cycle regime. II. DISPERSION COMPENSATING MIRRORS The generation of few-cycle pulses via external compression [25]–[27] as well as direct generation from Kerr-lens modelocked lasers [1], [28]–[31] relies heavily on the availability of chirped mirrors [16], [32], [33] for dispersion compensation. There are two reasons to employ chirped mirrors. First, the highof a standard dielectric Bragg-mirror reflectivity bandwidth of [see Fig. 2(a)] is determined by the Fresnel reflectivity is the center frequency of the mirror. Metal The frequency mirrors are, in general, too lossy, when used as intracavity laser mirrors. For material systems typically used for broad-band optical coatings such as Silicon Dioxide and Titanium Dioxide and (these indexes may vary with depending on the deposition technique used), a fractional bandcan be covered. Furthermore, the variation width in group delay of a Bragg-mirror already affects pulses whose spectra fill half of the mirror spectral range A way out of this dilemma was found by the introduction of chirped mirrors [13], the equivalent of chirped fiber Bragg gratings, which at that time were already well-developed components in fiber optics [34]. When the Bragg wavelength of the mirror stack is varied slowly enough and no limitation on the number of layer pairs exists, an arbitrarily broad-band mirror with high reflectivity can be engineered. The second reason for using chirped mirrors is based on their dispersive properties due to the wavelength dependent penetration depth of the light reflected from different positions inside the chirped multilayer structure. Mirrors are filters, and in the design of any filter, the control of group delay and group delay dispersion is difficult. This problem is further increased when the design has to operate over wavelength ranges up to an octave or more. A. Matching Problem Several designs for ultrabroad-band dispersion compensating mirrors have been discussed so far. For dispersion compensating mirrors which do not extend the high reflectivity range far beyond what a Bragg-mirror employing the same materials can already achieve, a multicavity filter design can be used to approximate the desired phase and amplitude properties [35], [36]. For dispersion compensating mirrors covering a high reflectivity , the concept of double-chirped range of up to mirrors (DCMs) has been developed [16], [42]. It is based on the following observations. A simple chirped mirror provides high reflectivity over an arbitrary wavelength range and, within certain limits, a controllable average group delay via its wavelength dependent penetration depth; see Fig. 2(b). However, the group delay as a function of frequency shows periodic variations due to the impedance mismatch between the ambient medium and the mirror stack, as well as within the stack [Fig. 2(b)]. A structure that mitigates these mismatches and gives better control of the group delay dispersion (GDD) is the DCM [Fig. 2(c)], in a way similar to that of an apodized fiber Bragg grating [37]. Fig. 3 shows the reflectivity and group delay of several Bragg and mirrors composed of 25 index steps, with , similar to the refractive indexes of and , which . The Bragg-mirror result in a Fresnel reflectivity of can be decomposed in symmetric index steps [16]. The Bragg and dewavenumber is defined as scribes the center wavenumber of a Bragg mirror composed of equal index steps. In the first case (Fig. 3, dashed-dotted 992 IEEE JOURNAL OF SELECTED TOPICS IN QUANTUM ELECTRONICS, VOL. 9, NO. 4, JULY/AUGUST 2003 Fig. 2. (a) Standard Bragg mirror. (b) Simple chirped mirror. (c) DCM with matching sections to avoid residual reflections causing undesired oscillations in the GD and GDD of the mirror. Fig. 4. Schematic structure of proposed broad-band dispersion compensating mirror, avoiding the matching problem to air: (a) front-tilted mirror; (b) back-side coated mirror, and (c) Brewster-angle mirror. Fig. 3. Comparison of the reflectivity and group delay of chirped mirrors with 25 layer pairs and refractive indexes n = 2:5 and n = 1:5. Upper portion shows the enlarged top percent of the reflectivity. Dashed-dotted curves show the result for a simply chirped mirror. Result for DCMs, where in addition to the chirp in the Bragg wave number k , the thickness of the high-index layers is also chirped over the first 12 layer pairs from zero to its maximum value for a linear chirp (dashed curves) and for a quadratic chirp (solid curves) [16]. line), only the Bragg wave number is linearly chirped from 6.8 m m over the first 20 index steps and held constant over the last five index steps. The reflectivity of the structure is computed assuming the structure imbedded in the low index medium. The large oscillations in the group delay are caused by the different impedances of the chirped grating and the surrounding low index material causing a strong reflection at the interface of the low index material and the grating stack. By adiabatic matching of the grating impedance to the low index material this reflection can be avoided. This is shown in Fig. 3 by the dashed and solid curves, corresponding to an additional chirping of the high index layer over the first 12 steps according with and , reto the law spectively. The strong reduction of the oscillations in the group delay by the double-chirp technique is clearly visible. Quadratic tapering of the high index layer, and, therefore, of the grating already eliminates the oscillations in the group delay completely, which can also be shown analytically by coupled mode analysis [42]. Because of the double chirp a high transmission window at the short wavelength end of the mirror opens up which is ideally suited for the pumping of Ti:sapphire lasers. So far, the DCM is only matched to the low index material of the mirror. Ideally, the matching can be extended to any other ambient medium by a properly designed anti-reflective (AR) coating. However, this AR-coating has to be of very high quality, i.e., very low residual reflectivity ideally a power reflectivity of 10 , i.e., an is required. The quality of amplitude reflectivity of the AR-coating can be relaxed, if the residual reflection is directed out of the beam path. This is achieved in so-called tilted front-side or back-side coated mirrors [38], [39] [Fig. 4(a) and (b)]. In the back-side coated mirror the ideal DCM structure, which is matched to the low index material of the mirror, is deposited on the back of the substrate, made of the same or at least very similar low index material. The AR-coating is deposited on the front of the slightly wedged substrate so that the residual reflection is directed out of the beam and does not deteriorate the dispersion properties. Thus, the task of the AR-coating is only to reduce the Fresnel losses of the mirror at the air-substrate interface, and, therefore, it is good enough for some applications, if the residual reflection at this interface is of the order of 0.5%. However, the substrate has to be very thin in order to keep the overall mirror dispersion negative, typically on the order of 200–500 m. Laser grade quality optics are hard to make on such thin substrates and the stress induced by the coating leads to undesired deformation of the substrates. The front-side coated mirror overcomes this shortcoming by depositing the ideal DCM-structure matched to the index of the wedge material on a regular laser grade substrate. A 100–200- m-thin wedge is bonded on top of the mirror and the AR-coating is then deposited on this wedge. This results in stable and octave spanning mirrors, which have been successfully used in external SCHIBLI et al.: TOWARDS SINGLE-CYCLE LASER SYSTEMS compression experiments [41]. Both structures come with limitations. First, they introduce a wedge into the beam, which leads to an undesired angular dispersion of the beam. This can be partially compensated by using these mirrors in pairs with oppositely oriented wedges. The second drawback is that it seems to be impossible to make high quality AR-coatings over one or more than one octave of bandwidth, which have less than 0.5% residual reflectivity [45], i.e., on one reflection such a mirror has at least 1% of loss, and, therefore, such mirrors insert high losses inside a laser. For external compression these losses are acceptable. A third possibility for overcoming the AR-coating problem is given by using the ideal DCM under Brewster-angle incidence (Fig. 4) [40]. In that case, the low index layer is automatically matched to the ambient air. However, under p-polarized incidence the index contrast or Fresnel reflectivity of a layer pair is reduced and more layer pairs are necessary to achieve high reflectivity. Also, the penetration depth into the mirror increased, so that scattering and other losses in the layers become more pronounced. On the other hand, such a mirror can generate more dispersion per bounce due to the higher penetration depth. For external compression such mirrors might have advantages because they can cover bandwidths much wider than one octave. This concept is difficult to apply to the fabrication of curved mirrors. There is also a spatial chirp of the reflected beam, which may become sizeable for large penetration depth and has to be removed by back reflection or an additional bounce on another Brewster-angle mirror that recombines the beam. For intracavity mirrors, a way out of this dilemma is found by mirror pairs, which cancel the spurious reflections due to an imperfect AR-coating and matching structure in the chirped mirror [17]. This design also has its drawbacks and limitations. It requires a high precision in fabrication and, depending on the bandwidth of the mirrors, it may be only possible to use them for a restricted range of angles of incidence. B. DCM Pairs There have been several proposals to increase the bandwidth of laser mirrors by mutual compensation of GDD oscillations [15], [43], [44] using computer optimization. These early investigations resulted in a rather low reflectivity of less than 95% over almost half of the bandwidth considered. The ideas leading to the DCMs help us to show analytically that a design of DCM pairs covering one octave of bandwidth, i.e., 600 to 1200 nm, with high reflectivity and improved dispersion characteristics is indeed possible [17]. Use of these mirror pairs in a Ti:sapphire laser system resulted in 5-fs pulses with octave spanning spectra directly from the laser [1]. Yet, the potential of these pairs is by no means fully exploited. The DCM, M1 [see Fig. 5(a)] consists of an AR coating and a double-chirped back-mirror MB with given wavelength-dependent penetration depth with suppressed spurious reflections. The reflectivity range of the back mirror can be easily extended to one octave by slow chirping and a sufficient number of layer pairs. However, the smoothness of the resulting GDD strongly depends on the quality of matching provided by the AR coating and the double-chirped section. The reflections occurring at the AR coating, which are similar to those occurring in a Gires–Tournois Interferometer (GTI), 993 Fig. 5. DCM-Pair (a) M1 and (b) M2. DCM M1 can be decomposed in a double-chirped back-mirror MB matched to a medium with the index of the top most layer. In M2, a layer with a quarter-wave thickness at the center frequency of the mirror and an index equivalent to the topmost layer of the back-mirror MB is inserted between the back mirror and the AR coating. New back mirror comprising the quarter-wave layer can be reoptimized to achieve the same phase as MB with an additional -phase shift over the whole octave of bandwidth. add up coherently when multiple reflections on chirped mirrors occur inside the laser (over one round trip). This leads to pre- and post-pulses, or even suppression of mode locking, if the mode-locking mechanism is not strong enough to suppress them sufficiently. As already discussed above, experimental results indicate that a residual reflection in the AR coating of or smaller, depending on the number of reflections per round trip, is required so that the pre- and post-pulses are sufficiently suppressed. This corresponds to an AR coating with less than 10 residual power reflectivity, which can only be achieved over a very limited range. Over a range of 350 nm centered around 800-nm residual power reflectivities as small as 10 have been achieved effectively after reoptimization of the AR-coating section and the double-chirped section to form a combined matching section of higher matching quality. For an even larger bandwidth approaching an octave, a residual power reflectivity of 10 is no longer possible [45]. A solution to this limitation is offered by the observation that a coherent subtraction of the pre- and post-pulses to first order in is possible by reflections on a mirror pair M1 and M2 [Fig. 5(a) and (b)]. A sequence of two reflections on mirror M1 and on a similar mirror M2 with an additional phase shift of between the AR coating and the back mirror, having the same reflectivity and AR coating, cancels the oscillations to first order in . Now, the GTI effects scale like the power reflectivity of the AR coating instead of the amplitude reflectivity , which constitutes a tremendous improvement, since it is possible to deof the mirror sign AR coatings for the low index material with a residual power reflectivity between 0.001 and 0.01 while covering one octave of bandwidth [45]. However, there does not exist a single physical layer which generates a phase shift during one passage for all frequency components conof tained in an octave. Still, a layer with a quarter-wave thickness at the center frequency is a good starting design. Then, the back-mirror MB in mirror M2 can be reoptimized to take care of the deviation from a quarter-wave thickness further away from 994 IEEE JOURNAL OF SELECTED TOPICS IN QUANTUM ELECTRONICS, VOL. 9, NO. 4, JULY/AUGUST 2003 Fig. 6. Reflectivity of the mirror with pump window shown as thick solid line with scale to the left. Group delay design goal for perfect dispersion compensation of a prismless Ti:sapphire laser is shown as a thick dashed-dotted line with scale to the right. Individual group delays of the designed mirrors are shown as thin line, and its average as a dashed line, which is almost identical with the design goal over the wavelength range, form 650–1200 nm. Measured group delay, using white light interferometry, is shown as the thick solid line from 600–1100 nm. Beyond 1100 nm the sensitivity of Si-detector used prevented further measurements. the center frequency, because the back mirror acts as a highly dispersive medium in which the phase or group delay can be designed at will. Fig. 6 shows in the top graph the designed reflectivity of both mirrors of the pair in high resolution taking into account the absorption in the layers. The graph below shows the reflectivity of the mirror, which has in addition high transmission between 510–550 nm for pumping the Ti:sapphire crystal. Each mirror and fabricated using consists of 40 layer pairs of ion-beam sputtering [46], [47]. Both mirror reflectivities cover more than one octave of bandwidth from 580 to 1200 nm or 250 to 517 THz, with an average reflectivity of about 99.9%, including the absorption in the layers. In addition, they correct for the second- and higher order dispersion of all intracvity elements such as the Ti:sapphire crystal and the thin BaF wedges for fine adjustment of the dispersion from 650 to 1200 nm within the 12 bounces occuring in one round trip. The choice for the lower wavelength boundary in dispersion compensation is determined and limited by the pump window of Ti:sapphire. The many bounces allow for building a compact all-mirror dispersion-controlled octave spanning Ti:sapphire laser even at moderate repetition rates of around 100 MHz. BaF was chosen because it has the lowest ratio between third- and second-order dispersion in this wavelength range of all the materials known to us. This is important, since it seems to be impossible to design octave-spanning chirped mirrors with large negative third-order dispersion as well as high reflectivity. Another advantage of BaF is, that 1 mm BaF has a similar dispersion as 2 m of air. Thus, one can replace air by BaF to scale the laser to higher repetition rates if needed, which is of importance for frequency metrology. The group delays of the individual mirrors of the pair, its average, and the measured average of the mirror pair M1 and M2 is shown underneath the reflectivity trace. The design was carried out according to the description given above. The dispersion measurement was performed using white light interferometry [48], up to about 1100 nm, because of the silicon detector roll-off. The oscillations in the group delay of each mirror are about ten times larger than those of high-quality DCMs covering 350 nm of bandwidth [30]. However, in the average group delay of both mirrors the oscillations are ideally suppressed due to cancellation by more than a factor of ten. Therefore, the effective residual reflectivity of the mirror pair covering one octave is even smaller than that of conventional DCMs. Because of slight fabrication errors, the oscillations in the GD still do not precisely cancel, especially close to 900 and 1000 nm deviations from the design goal on the order of 1–2 fs occur, which will lead to observable spectral features in the spectral output of the laser described below. Note that the fabrication of this new mirror pair has dramatically improved performance when compared to a earlier versions of such a pair [17]. In addition, those earlier pairs were also designed for use in combination with prisms. III. FEW-CYCLE LASERS Because of the limitations in bandwidth and dispersion control imposed by conventional laser mirrors, the generation of pulses with spectra covering one octave and pulsewidths as short as 5 fs (less than two optical cycles at 800 nm) was thought to be possible only by external compression. Impressive results have been achieved using external compression [50], [51]. Spectral broadening by self-phase modulation (SPM) is well known and was recently put to spectacular use with photonic bandgap fibers [52], [53] to generate spectra from about 500 nm to 1.5 m using the unamplified pulse train of a standard mode-locked Ti:sapphire oscillator. To date, recompression of these spectra has not been successful and recent characterization of these wide-band spectra has shown that this is a highly nontrivial task [54]. Possible reasons for this incompressibility are the complicated phase structures emerging from the interplay between SPM and group delay dispersion (GDD) in the fiber, which are highly sensitive to pulse energy fluctuations from pulse to pulse. Therefore, a source which directly generates octave spanning spectra and correspondingly short pulses can be superior. This is possible because a train of octave spanning pulses can be phase-stabilized directly by the 1f–2f technique described in [56]. A phase-stabilized pulse has a well-defined electric field even on a subcycle timescale, which can be used to map out phenomena occurring on an attosecond time scale [19]. Intracavity spectral broadening is advantageous in the sense that the spectral phase is cleaned up each round trip by dispersion-managed modelocking [49] as well as by KLM, making the extracavity pulse recompression more successful. Pulses in the two-cycle regime have already been generated directly from a Ti:sapphire oscillator [1], [30], [55] at a center wavelength of 800 nm using DCMs. The DCM technology described in Section II enables the fabrication of laser optics with high reflectivity and the possibility of precise dispersion compensation of up to a full octave bandwidth. This was demonstrated recently in combination with prism pairs [1]. For this paper, we used the new DCM pairs described in Section II to build the much simpler prismless laser shown in Fig. 7. Note further that the design shown in Fig. 7 dispenses with the second focus used previously for the enhancement of KLM [1], because the smoother GD requires less mode-locking strength. The dispersion of the mirrors has been SCHIBLI et al.: TOWARDS SINGLE-CYCLE LASER SYSTEMS Fig. 7. Schematic diagram of the octave spanning prismless Ti:sapphire laser. Z-fold cavity is astigmatically compensated. Twelve bounces on the DCMPs (six on each mirror type) provide a smooth and broad-band compensation of the dispersion of the laser crystal, the BaF wedges, and the air, which resides inside the laser cavity. BaF wedges are used to fine-adjust the dispersion. designed in such a way that the dispersion of all the intracavity elements, including the thin small angle BaF wedges that are used to fine-tune the dispersion of the resonator, is compensated. In combination with the broad-band laser materials discussed in Section I, a whole class of few cycle lasers has been developed. In this section, we present a detailed description of the afore-mentioned octave spanning Ti:sapphire laser employing double-chirped mirror pairs (DCMPs) and BaF wedges for dispersion compensation and we briefly mention the other results obtained with Cr:forsterite and Cr:YAG. The setup of the Ti:sapphire laser is shown in Fig. 7. All cavity mirrors except the output coupler are DCMs. The different shadings (gray and black) are used to denote the two different mirrors (M1 and M2) of the pair. Twelve bounces on these mirrors per round trip generate the necessary negative dispersion to compensate the positive second- and third-order dispersion of the laser crystal, the 3.7 m of air inside of the 82 MHz laser cavity, and the 11 mm of BaF per round trip which is used to fine-tune and balance the dispersion on each side of the laser crystal. It is important to note that the mirrors have to be placed in such a way that the same number of bounces on each mirror type occurs not only within one full round trip, but also on each side of the laser crystal, since the nonlinearity of the crystal would corrupt the cancellation of the dispersion oscillations. The highly doped Ti:sapphire laser crystal has a path length of 2 mm and it absorbs approximately 72% of the 532-nm pump light which is emitted by a frequency-doubled diode pumped Nd:YVO laser (Spectra Physics Millenia Xs) and is focused with a 60-mm focal length plano-convex lens into the Ti:sapphire crystal. The curved mirrors shown in Fig. 7 have a radius of curvature of 100 mm and the coating of these curved DCMs transmits more than 95% of the 532-nm pump light. The twelve bounces on the DCMPs enable a very compact cavity layout. Even for repetition rates as low as 82 MHz, it is possible to fit the resonator on a footprint comparable to the size of a standard sheet of letter paper if an arm-ratio of 2:1 is used. This compactness paired with the fact that a slight misalignment of the resonator does not lead to a change of the dispersion—as this was the case in a prism-compensated cavity—enables the construction of a particularly reliable system for the 995 continuum generation. With a layout shown in Fig. 7, we obtain stable long-term mode locking even if the laser is optimized for octave spanning operation, for which precise dispersion-compensation over a large bandwidth is absolutely crucial [1]. We found that it is possible to operate the laser without any realignment for several weeks. However, to maintain a constant output power, the pump power has to be adjusted to compensate for the cavity misalignment that naturally occurs over time. Another benefit of the prismless cavity is that fine tuning of the dispersion can be achieved by a simple adjustment of the insertion of one of the BaF wedges. This significantly simplifies the handling of the system, since the user has to control only one degree of freedom. This also greatly simplifies the use of such a laser in a synchronized system as the one described in Section IV, because in contrast to a variation of the prism-separation, the fine tuning of the dispersion in the cavity shown in Fig. 7 is done by a sole variation of the insertion of the BaF wedge, which typically changes the cavity length much less than 100 m. This cavity length change can be easily corrected by the use of a piezo-actuated resonator mirror. Another important element in an octave spanning laser is the output coupling mirror. Since a standard Bragg-mirror based on and can only cover about 200 nm around a center wavelength of 800 nm, materials with a higher index contrast are needed. The output coupler used in the laser presented here is the same as the one previously used [1]. The output-coupling mirror layer of MgF . consists of 5 ZnSe/MgF pairs topped with a This mirror has a high reflectivity between 700 and 1030 nm with a transmission of about 1% at 800 nm. The laser emits about 100 mW of mode-locked power through the 1% output coupler when pumped with 3.8 W of 532-nm light. The output power is limited by the low output coupling of the broad-band output coupler. The efficiency of the laser could be drastically increased by a higher output coupling. Due to the high transmission of the output coupler at 650 and 1100 nm, the flat-top intracavity spectrum of the laser is significantly emphasized in the wings of the output spectrum as shown in Fig. 8. This explains the strongly pronounced M-shape of the output spectrum. On one hand, this spectral shaping can be beneficial to obtain a wider spectrum in the output beam of the laser; but on the other hand, it leads to early pulse breaking due to strong spectral filtering, and, therefore, it can prohibit further broadening of the intracavity light. The small oscillations in the spectrum originate from the residual oscillations in the group delay of the DCMPs. The spectral shape significantly influences the number of optical cycles within the FWHM pulse duration. If we define the center frequency as the weighted average of the power spectrum (2) denotes the Fourier transform of the electric field where THz of of the pulse, we obtain the center frequency the spectrum shown in Fig. 8. Using Fourier transformation, we obtain a FWHM pulse duration of 3.4 fs assuming a flat spectral phase. However, it might be difficult to fully recompress the pulses emitted by this laser and therefore the real pulse duration that can be used for an experiment might be noticeably longer. 996 IEEE JOURNAL OF SELECTED TOPICS IN QUANTUM ELECTRONICS, VOL. 9, NO. 4, JULY/AUGUST 2003 Fig. 8. Typical output spectrum emitted by the laser shown in Fig. 7 on a linear (right) and logarithmic scale (left). Upper graph shows the spectral power per plotted as a function of wavelength . Lower graph shows the same spectrum as power per frequency interval plotted as a function of frequency. M-shape of the spectrum is due to the transmission characteristic of the output coupler. Small oscillations originate from the residual oscillations in the dispersion of the DCMPs. Theoretical FWHM duration of the corresponding pulses assuming a flat spectral phase is 3.4 fs, equivalent to the duration of less than 1.3 optical cycles at the weighted center frequency of 365 THz [see (2)]. 1 Finally, 3.4 fs correspond to the duration of less than 1.3 cycles at the weighted center frequency (2) of 365 THz. As found earlier [1], [31] the M-shape of the output spectrum has a beneficial influence on the FWHM pulse duration. If we do the same calculation for a rectangular shaped spectrum with the same FWHM bandwidth from 645 to 1100 nm, we obtain a 4.6-fs FWHM pulse duration again assuming a flat spectral phase. This corresponds to the duration of 1.7 optical cycles at the spectrum’s center frequency of 370 THz. Therefore, in an experiment that is sensitive to the electric field rather than to the pulse envelope one might expect a notable difference between a flat-top and an M-shaped spectrum with the same FWHM bandwidth. So far, DCMPs have been only applied to the Ti:sapphire system. However, conventional DCMs have been used to extract ultrabroad-band spectra from Cr:forsterite [57], Cr:YAG [58], and Cr:LiCAF [59] lasers. The Cr:forsterite and Cr:YAG laser systems suffer from a large third-order dispersion, since they operate close to the point where the second-order dispersion is zero at 1490 nm for Cr:forsterite and 1580 nm for Cr:YAG. Currently, only chirped mirrors can compensate for the limiting third-order dispersion in addition to second-order dispersion over a large bandwidth. The spectra and interferometric autocorrelation traces of these laser systems are shown in Figs. 9 and 10. To overcome the bandwidth limitation posed by the finite-gain bandwidth of the laser material and by the difficulty Fig. 9. Optical power spectrum (top) and measured interferometric fit reveals a autocorrelation of the Cr:forsterite laser presented in [57]. 14-fs pulse duration. sech Fig. 10. Top: Optical power spectrum on a logarithmic (dotted) and on a linear scale (black) of the Cr:YAG laser presented in [58]. Bottom: measured interferometric autocorrelation of the same laser. Phase-retrieval algorithm reveals a 20-fs pulse duration. of compensating higher order dispersion over a larger spectral bandwidth than one octave, one could think of a phase-coherent synchronization of two or more of the broad-band lasers demonstrated before with overlapping spectra. Such a coherent pulse synthesis using longer 20–30-fs pulses from two independent Ti:sapphire lasers with slightly shifted center SCHIBLI et al.: TOWARDS SINGLE-CYCLE LASER SYSTEMS 997 frequency has been already demonstrated by Shelton et al. [60]. This is the subject of Section IV using the broad-band Ti:sapphire laser and a 30-fs Cr:forsterite laser together with a novel synchronization scheme. IV. TOWARDS SINGLE-CYLCE PULSE GENERATION As mentioned in Section III, single-cycle optical pulses may be achieved through phase coherent superposition of several spectrally overlapped few-cycle lasers. The synchronization of pulse trains from independently mode-locked lasers with subcycle timing fluctuations is the most important and the most challenging step in this synthesis process. Ideally, the relative timing jitter should be less than one-tenth of an optical cycle for a high-quality synthesized pulse stream. At a center wavelength of 900 nm, this condition restricts the timing jitter to 300 attoseconds or less, measured over the full Nyquist bandwidth, i.e., half the laser repetition rate. Several groups have investigated the possibility of active [61] and/or passive [62], [63] synchronization of multiple lasers. However, a subfemtosecond timing jitter over the Nyquist bandwidth has not been achieved to date. In this section, a method of synchronization is demonstrated in which the timing jitter between two passively mode-locked lasers is detected by a balanced cross correlator (see Fig. 12), the optical equivalent of a balanced microwave phase detector. The signal is then fed back via an electronic control loop in order to keep the two lasers synchronized. This method enables a drift-free and temperature-independent synchronization between two individual lasers, a task that is difficult to achieve with all-electronic schemes. To ensure the long-term stability of the system against thermal drifts, the two laser beams are combined inside the control loop. To further improve the stability of the system, we used prismless lasers, as discussed in Section III, to generate the corresponding parts of the continuum. The fine adjustment of the dispersion does not change the repetition rate significantly as it was previously the case in the prism-compensated cavities. In [64], it was already demonstrated that it is possible to generate ultrabroad spectra with high efficiencies in purely DCM-compensated Ti:sapphire lasers. To improve the long-term stability and mode locking of the Cr:forsterite laser presented in [57], we used a novel broad-band InGaAs saturable absorber on a large area, high-index contrast AlGaAs/Al O mirror [65]. Due to the high-index contrast between AlGaAs and Al O , the mirror’s reflectivity extends from roughly 1100 to 1500 nm, which enables the generation of sub-30-fs pulses at 1230-nm wavelength in a self-starting configuration. Even though the spectral width of this laser is significantly narrower than previously reported in [57], the setup does not require the cavity to be purged with dry nitrogen and the mode locking does not rely on a critical alignment of the cavity. This simplifies the operation of this laser. Furthermore, it is possible to operate the laser without any cavity realignment over several months. The output power of the laser is roughly 60 mW of mode-locked power through a 3% output coupler when pumped with approximately 2–3 W of 1064-nm light and is currently limited by the break-up of the pulse into multiple pulses. The tendency for pulse breakup could be reduced by introducing a saturable absorber with more saturable loss or by use of addi- Fig. 11. Optical spectra of the mode-locked Ti:sapphire and Cr:forsterite lasers on a logarithmic scale. Dashed lines indicate the spectra of the individual lasers in the vicinity of the spectral overlap. Theoretical FWHM duration of the corresponding pulses assuming a flat spectral phase and a perfect phase-locking between the two combs is 3.0 fs corresponding to the duration of a single optical cycle at the weighted center frequency [see (2)] of 320 THz. Fig. 12. Experimental setup of the synchronized lasers. Cr:fo: passively mode-locked Cr:forsterite laser, Ti:sa: passively mode-locked Ti:sapphire laser; SFG: sum-frequency generation; all bandpass filters transmit only the sum-frequency (1=496 nm = 1=833 nm + 1=1225 nm). Two beam splitters consist of thin fused silica glass substrates coated with a semi-transparent metal film. Third correlator is used to generate the graphs shown in Fig. 14. tional KLM. To improve the optical-to-optical efficiency, the Cr:forsterite crystal was cooled to 25 C. Fig. 11 shows the total spectrum of the two lasers (solid line) at the output. The dashed lines indicate the extent of the individual laser spectra in the vicinity of the overlap region. The shaded region indicates the spectral region filtered out to record the difference in carrier envelope offset frequency between the two lasers (see Fig. 15). For a phase-coherent superposition of the two lasers, the pulse envelopes of the two lasers must be synchronized. In addition, the difference in the carrier envelope offset frequency between the two lasers must be set to zero with a phase lock. Synchronizing the pulse trains with subcycle precision is the most challenging step in the synthesis process. Ideally, timing accuracy of less than one-tenth of an optical cycle should be achieved. In our case, this requires a timing jitter of 300 attoseconds or less measured over the full Nyquist bandwidth. To overcome the typical problems posed by balanced microwave mixers previously used for this task [61], we developed the optical equivalent of such a device, a balanced cross correlator. As shown in Fig. 12, the outputs of the two lasers are combined on a broad-band metallic beam splitter. One part of the combined beam is directed to two nearly identical cross corre- 998 IEEE JOURNAL OF SELECTED TOPICS IN QUANTUM ELECTRONICS, VOL. 9, NO. 4, JULY/AUGUST 2003 Fig. 13. Output of the balanced cross correlator as a function of time difference between the two laser pulses. Trace was taken when the two lasers were unlocked. Time axis was calibrated by measuring the frequency difference between the two lasers and measuring the cavity round-trip time. lators using 1-mm-thick LBO crystals phase matched for SFG of 833-nm light from the Ti:sapphire laser and 1225-nm light from the Cr:forsterite laser. The only difference between the two correlators is a 3-mm-thick fused silica window in the optical path of one of them. This glass inserts a group delay between 833 and 1225 nm to offset the pulses emitted by the Cr:forsterite and Ti:sapphire lasers by about 45 fs with respect to each other. Due to the low temperature dependence of the chromatic dispersion (less than 1 attosecond per C), it is legitimate to use this 45-fs delay as a reference for timing offset measurements. For small time differences between the two laser pulses, the difference between the currents of the two photo detectors at the end of each correlator is nearly proportional to the time difference between the two pulses. Furthermore, in the vicinity of zero timing offset this detector acts like a balanced phase detector operating in the multiple terahertz range if the signal amplitudes of the two correlators are balanced against each other. At the zero crossing of the difference of the photo currents, this detector delivers a perfectly balanced signal, and, therefore, amplitude noise of each laser does not affect the detected error signal. The output of this balanced cross correlator as a function of time difference between the Cr:forsterite and the Ti:sapphire pulses is shown in Fig. 13. The signal from the balanced mixer is used to lock the repetition rates of the two lasers by controlling the cavity length of the Ti:sapphire laser with cavity mirrors mounted on piezo-electric transducers in a manner similar to that discussed in [61]. This finally closes the control loop. The first beam splitter used to combine the two output beams from the lasers is inside this control loop. Since the output beam shown in Fig. 11 originates from this beam splitter, temperature drifts, acoustic noise, or beam fluctuations always affect both laser beams in the same way as they both travel along identical paths. Therefore, external noise cannot corrupt the relative jitter, and the output behaves as if it originated from the same source. Even environmental noise which influences the optical length of the cross correlators cannot corrupt the timing, since the group delay of the 3-mm fused silica, measured between 833 and 1225 nm, is the only timing reference in this system. Fig. 14 shows the resulting timing jitter measurement made with the out-of-loop cross correlator shown in Fig. 12. The residual timing jitter over the detector’s bandwidth of 2.3 MHz is 300 as 100 as. The stated error is determined from the Fig. 14. Timing jitter determined from the amplitude noise of the SFG of the out of loop cross-correlator (see Fig. 12). Time delay in the correlator was generated by a dispersive medium in front of the SFG crystal. rms jitter measured in a 2.3-MHz BW results in 300 as 100 as. 6 Fig. 15. Heterodyne beat between the Cr:forsterite and the Ti:sapphire lasers obtained behind a 10-nm-wide optical band-pass filter centered at 1130 nm. Two beat signals below the repetition rate of 82 MHz represent the difference in the carrier envelope offset frequency of both lasers. RF-analyzer filter-bandwidth was set to 30 kHz. Noise floor is caused by the uncompensated transimpedance amplifier and poses only a technical limitation. amplitude noise measured at the peak of the cross correlation. This trace is also shown in Fig. 14. As in most passively mode-locked laser systems, the main contribution to the timing jitter has frequency components up to a few times the relaxation oscillation frequency of the laser [66]. In the current system, the relaxation oscillation frequencies are roughly 70 kHz for the Ti:sapphire laser and 140 kHz for the Cr:forsterite laser. Therefore, we assume that noise above 2.3 MHz is negligible. As soon as the two lasers are locked to each other, we observe a strong beat signal in the overlap region of the optical spectrum. The beat signal shown in Fig. 15 was detected with an InGaAs PIN-diode connected to a transimpedance amplifier. To avoid saturation of the detector, only a small part of the optical spectrum was directed to the diode. The transmission of the 10-nm-wide band-pass filter has its maximum at 1130 nm. As described in [67], the beat signal represents the difference between the two in the carrier-envelope offset frequency lasers. In contrast to previous results, it is now possible to obtain this beat without the use of additional spectral broadening. This helps to provide an exceptionally large signal-to-noise ratio of about 50 dB in a 30-kHz bandwidth. For completion of the SCHIBLI et al.: TOWARDS SINGLE-CYCLE LASER SYSTEMS pulse synthesis process, it is necessary to phase-lock this difference offset frequency to dc. The widely used offset-locking technique, which was also employed in [60], could be applied; however, it would not lead to a long term stable phase coherent output, since any change in the path length of the offset lock would be transfered to the output beam. We currently investigate the use of a homodyne technique using a fraction of each of the two output beams of the first in-loop beamsplitter, that is superimposing the Ti:sapphire and Cr:forsterite beam.This would automatically stabilize the phase difference between the two spectral components constituting the output beam. V. CONCLUSION We have shown that broad-band dispersion compensating mirror technology has been developed to a point where two few-cycle lasers, a Ti:sapphire laser and a Cr:forsterite laser, can potentially span a spectrum from 600 to 1600 nm. The balanced cross-correlation technique demonstrated can be used to synchronize both lasers to a tenth of an optical cycle and most likely even further after optimization of all components of the overall system. In addition, balanced homodyne detection of the interference beat of both lasers can be used to stabilize the phase between the two lasers such that true single-cycle optical pulses can be generated in the near future. Using optical cross correlation instead of synchronization in the microwave region leads to a long term drift free and stable overall laser system. ACKNOWLEDGMENT The authors would like to thank S. Tandon, G. Petrich, and L. Kolodziejski for fabrication of the broad-band saturable absorbers; P. O’Brien for fabrication of the broad-band beam splitters; and H. Haus, J. Ye, S. Cundiff, and D. Jones for many fruitful discussions. The authors also would like to acknowledge the work on the broad-band Cr:YAG and the Kerr lens mode-locked Cr:forsterite lasers by D. Ripin and C. Chudoba as well as many discussions with U. Morgner and R. Ell. REFERENCES [1] R. Ell, U. Morgner, F. X. Kärtner, J. G. Fujimoto, E. P. Ippen, V. Scheuer, G. Angelow, and T. Tschudi, “Generation of 5 fs pulses and octavespanning spectra directly from a Ti:sapphire laser,” Opt. Lett., vol. 26, pp. 373–375, 2001. [2] P. F. Moulton, “Spectroscopic and laser characteristics of Ti:Al O ,” J. Opt. Soc. Amer. B, vol. 3, pp. 125–132, 1986. [3] D. E. Spence, P. N. Kean, and W. Sibbett, “60-fsec pulse generation from a self-mode-locked Ti:Sapphire laser,” Opt. Lett., vol. 16, pp. 42–44, 1991. [4] D. K. Negus, L. Spinelli, N. Goldblatt, and G. Feugnet, “Sub-100 femtosecond pulse generation by Kerr Lens modelocking in Ti:Sapphire,” in Advanced Solid-State Lasers, G. Dubé and L. Chase, Eds. Washington, DC: Optical Soc. Amer., 1991, pp. 120–124. [5] F. Salin, J. Squier, and M. Piché, “Mode locking of Ti:Al O lasers and self-focusing: A Gaussian approximation,” Opt. Lett., vol. 16, pp. 1674–1676, 1991. [6] M. Piché, “Beam reshaping and self-mode-locking in nonlinear laser resonators,” Opt. Commun., vol. 86, pp. 156–160, 1991. [7] U. Keller, G. W. ’tHooft, W. H. Knox, and J. E. Cunningham, “Femtosecond pulses from a continuously self-starting passively mode-locked Ti:Sapphire laser,” Opt. Lett., vol. 16, pp. 1022–1024, 1991. 999 [8] J. M. Jacobson, K. Naganuma, H. A. Haus, J. G. Fujimoto, and A. G. Jacobson, “Femtosecond pulse generation in a Ti:sapphire laser by using second- and third-order intracavity dispersion,” Opt. Lett., vol. 17, pp. 1608–1610, 1992. [9] M. T. Asaki, C.-P. Huang, D. Garvey, J. Zhou, H. C. Kapteyn, and M. N. Murnane, “Generation of 11-fs pulses from a self-mode-locked Ti:Sapphire laser,” Opt. Lett., vol. 18, pp. 977–979, 1993. [10] J. Zhou, G. Taft, C.-P. Huang, M. M. Murnane, H. C. Kapteyn, and I. P. Christov, “Pulse evolution in a broad-bandwidth Ti:sapphire laser,” Opt. Lett., vol. 19, pp. 1149–1151, 1994. [11] C. Spielmann, P. F. Curley, T. Brabec, and F. Krausz, “Ultrabroad-band femtosecond lasers,” IEEE J. Quantum Electron., vol. 30, pp. 1100–1114, 1994. [12] A. Stingl, C. Spielmann, and F. Krausz, “Generation of 11-fs pulses from a Ti:Sapphire laser without the use of prisms,” Opt. Lett., vol. 19, pp. 204–206, 1994. [13] R. Szipöcs, K. Ferencz, C. Spielmann, and F. Krausz, “Chirped multilayer coatings for broadband dispersion control in femtosecond lasers,” Opt. Lett., vol. 19, pp. 201–203, 1994. [14] A. Stingl, M. Lenzner, C. Spielmann, F. Krausz, and R. Szipöcs, “Sub-10-fs mirror-dispersion-controlled Ti:Sapphire laser,” Opt. Lett., vol. 20, pp. 602–604, 1995. [15] R. Szipöcs and A. Kohazi-Kis, “Theory and design of chirped dielectric laser mirrors,” Appl. Phys. B, vol. 65, pp. 115–135, 1997. [16] F. X. Kärtner, N. Matuschek, T. Schibli, U. Keller, H. A. Haus, C. Heine, R. Morf, V. Scheuer, M. Tilsch, and T. Tschudi, “Design and fabrication of double-chirped mirrors,” Opt. Lett., vol. 22, pp. 831–833, 1997. [17] F. X. Kärtner, U. Morgner, T. R. Schibli, E. P. Ippen, J. G. Fujimoto, V. Scheuer, G. Angelow, and T. Tschudi, “Ultrabroadband double-chirped mirror pairs for generation of octave spectra,” J. Opt. Soc. Amer. B, vol. 18, pp. 882–885, 2001. [18] P. Corkum, “Breaking the attosecond barrier,” in Optics Photonics News, 1995, pp. 18–21. [19] H. Hentschel, R. Kienberger, C. Spielmann, G. A. Reider, N. Milosevic, T. Brabec, P. Corkum, U. Heinzmann, M. Drescher, and F. Krausz, “Attosecond metrology,” Nature, vol. 414, pp. 509–513, 2001. [20] S. Hughes, “Breakdown of the area theorem: Carrier-wave rabi flopping of femtosecond optical pulses,” Phys. Rev. Lett., vol. 81, pp. 3363–3366, 1998. [21] O. D. Mücke, T. Tritschler, M. Wegener, U. Morgner, and F. X. Kärtner, “Signatures of carrier-wave rabi-flopping in GaAs,” Phys. Rev. Lett., vol. 87, pp. 057 401–4, 2001. [22] J. M. Fraser, A. I. Shkrebtii, J. E. Sipe, and H. M. van Driel, “Quantum interference in electron-hole generation in noncentrosymmetric semiconductors,” Phys. Rev. Lett., vol. 83, pp. 4192–4195, 1999. [23] P. Dombi, A. Apolonski, G. G. Paulus, M. Kakehata, R. Holzwarth, T. Udem, C. Lemell, J. Burgdoerfer, T. W. Haensch, and F. Krausz, “Solidstate light phase detector,” in Conf. Lasers and Electro-Optics CLEO 2003, Baltimore, MD, June 1–6, 2003. [24] C. Lemell, X.-M. Tong, F. Krausz, and J. Burgdoerfer, “Electron emission from metal surfaces by ultrashort pulses: Determination of the carrier-envelope phase,” Phys. Rev. Lett., vol. 90, pp. 076 403-1–4, 2003. [25] A. Baltuska, Z. Wei, M. Pshenichnikov, and D. Wiersma, “Optical pulse compression to 5 fs at 1 MHz repetition rate,” Opt. Lett., vol. 22, pp. 102–105, 1997. [26] A. Baltuska, Z. Wei, M. Pshenichnikov, D. Wiersma, and R. Szipöcs, “All-solid-state cavity dumped sub-5-fs laser,” Appl. Phys. B, vol. 65, pp. 175–188, 1997. [27] M. Nisoli, S. De Silvestri, O. Svelto, R. Szipöcs, K. Ferenz, C. Spielmann, S. Sartania, and F. Krausz, “Compression of high energy laser pulses below 5 fs,” Opt. Lett., vol. 22, pp. 522–525, 1997. [28] A. Stingl, M. Lenzner, C. Spielmann, F. Krausz, and R. Szipöcs, “Sub-10 fs mirror-dispersion-controlled Ti:Sapphire laser,” Opt. Lett., vol. 20, pp. 602–604, 1995. [29] I. D. Jung, F. X. Kärtner, N. Matuschek, D. H. Sutter, F. Morier-Genoud, U. Keller, V. Scheuer, M. Tilsch, and T. Tschudi, “Self-starting 6.5 fs pulses from a KLM Ti:Sapphire laser,” Opt. Lett., vol. 22, pp. 1009–1011, 1997. [30] U. Morgner, F. X. Kärtner, S. H. Cho, Y. Chen, H. A. Haus, J. G. Fujimoto, and E. P. Ippen, “Sub-two cycle pulses from a Kerr-Lens modelocked Ti:Sapphire laser,” Opt. Lett., vol. 24, pp. 411–413, 1999. [31] D. H. Sutter, G. Steinmeyer, L. Gallmann, N. Matuschek, F. MorierGenoud, U. Keller, V. Scheuer, M. Tilsch, and T. Tschudi, “Semiconductor saturable absorber assisted Kerr-Lens mode-locked Ti:Sapphire laser producing pulses in the two-cycle regime,” Opt. Lett., vol. 24, pp. 631–633, 1999. 1000 IEEE JOURNAL OF SELECTED TOPICS IN QUANTUM ELECTRONICS, VOL. 9, NO. 4, JULY/AUGUST 2003 [32] R. Szipocs, K. Ferencz, C. Spielmann, and F. Krausz, “Chirped multilayer coatings for broadband dispersion control in femtosecond lasers,” Opt. Lett., vol. 19, pp. 201–203, 1994. [33] E. J. Mayer, J. Möbius, A. Euteneuer, W. W. Rühle, and R. Szipöcs, “Ultrabroadband chirped mirrors for femtosecond lasers,” Opt. Lett., vol. 22, pp. 528–531, 1997. [34] K. O. Hill, F. Bilodeau, B. Malo, T. Kitagawa, S. Theriault, D. C. Johnson, J. Albert, and K. Takiguch, Opt. Lett., vol. 19, pp. 1314–1316, 1994. [35] A. V. Tikhonravov, M. K. Trubetskov, and A. A. Tikhonravov, “To the design and theory of chirped mirrors,” in OSA Topical Meeting Optical Interference Coatings, Tucson AZ, Feb. 7–12, 1998. [36] B. Golubovic, R. R. Austin, M. K. Steiner-Shepard, M. K. Reed, S. A. Diddams, D. J. Jones, and A. G. Van Engen, “Double-gires-tournois interferometer negative-dispersion mirrors for use in tunable mode-locked lasers,” Opt. Lett., vol. 25, pp. 175–278, 2000. [37] M. Matsuhara and K. O. Hill, “Optical -waveguide bandrejection filters: Design,” Appl. Opt., vol. 13, pp. 2886–2888, 1974. [38] N. Matuschek, L. Gallmann, D. H. Sutter, G. Steinmeyer, and U. Keller, “Back-side-coated chirped mirrors with ultra-smooth broadband dispersion characteristics,” Appl. Phys. B, vol. 71, pp. 509–522, 2000. [39] G. Tempea, V. Yakovlev, B. Bacovic, F. Krausz, and K. Ferencz, “Tilted-front-interface chirped mirrors,” J. Opt. Soc. Amer. B, vol. 18, pp. 1747–1750, 2001. [40] G. Steinmeyer, “Brewster-angled chirped mirrors for smooth octave spanning dispersion compensation,” in Conf. Lasers and Electro-Optics (CLEO 2003), Baltimore, MD, June 2–6, 2003. [41] A. Apolonski, A. Poppe, G. Tempea, C. Spielmann, T. Udem, R. Holzwarth, T. Hänsch, and F. Krausz, “Controlling the phase evolution of few-cycle light pulses,” Phys. Rev. Lett., vol. 85, pp. 740–743, 2000. [42] N. Matuschek, F. X. Kärtner, and U. Keller, “Theory of double-chirped mirrors,” IEEE J. Select. Topics Quantum Electron., vol. 4, pp. 197–208, 1998. [43] V. Laude and P. Tournois, “Chirped-mirror-pairs for ultrabroadband dispersion control,” in Proc. Conf. Lasers and Electro-Optics (CLEO 1999), Baltimore, MD, 1999, Paper CTuR4. [44] R. Szipöcs, A. Köhazi-Kis, S. Lakó, P. Apai, A. P. Kovácz, G. DeBell, L. Mott, A. W. Louderback, A. V. Tikhonravov, and M. K. Trubetskov, “Negative dispersion mirrors for dispersion control in femtosecond lasers: Chirped dielectric mirrors and multi-cavity gires-tournois interferometers,” Appl. Phys. B, vol. 70, pp. 51–57, 2000. [45] J. A. Dobrowolski, A. V. Tikhonravov, M. K. Trubetskov, B. T. Sullivan, and P. G. Verly, “Optimal single-band normal-incidence antireflection coatings,” Appl. Opt., vol. 35, pp. 644–658, 1996. [46] V. Scheuer, M. Tilsch, and T. Tschudi, “Reduction of absorption losses in ion beam sputter deposition of optical coatings for the visible and near infrared,” Proc. SPIE Conf. , vol. 2253, pp. 445–454, 1994. [47] M. Tilsch, V. Scheuer, and T. Tschudi, “Direct optical monitoring instrument with a double detection system for the control of multilayer systems from the visible to the near infrared,” Proc. SPIE Conf., vol. 2253, pp. 414–422, 1994. [48] K. Naganuma, K. Mogi, and H. Yamada, “Group-delay measurement using the fourier-transform of an interferometric cross correlation generated by white light,” Opt. Lett., vol. 15, pp. 393–395, 1990. [49] Y. Chen, F. X. Kärtner, U. Morgner, S. H. Cho, H. A. Haus, J. G. Fujimoto, and E. P. Ippen, “Dispersion-managed mode locking,” J. Opt. Soc. Amer. B, vol. 16, pp. 1999–2004, 1999. [50] M. Pshenichnikov, A. Baltuska, R. Szipöcs, and D. Wiersma, “Sub-5-fs pulses: Generations, characterization, and experiments,” in Ultrafast Phenomena XI, vol. 3, Springer, Berlin, Germany, 1998, pp. 3–7. [51] Z. Cheng, G. Tempea, T. Brabec, K. Ferencz, C. Spielmann, and F. Krausz, “Generation of intense diffraction-limited white light and 4-fs pulses,” in Ultrafast Phenomena XI, vol. 8, 1998, pp. 8–10. [52] J. Ranka, R. Windeler, and A. Stentz, “Visible continuum generation in air silica microstructure optical fibers with anomalous dispersion at 800 nm,” Opt. Lett., vol. 25, pp. 25–28, 2000. [53] T. Birks, W. Wadsworth, and P. Russel, Conf. Lasers and Electro-Optics, vol. 59, Washington, DC, 2000. [54] X. Gu, L. Xu, M. Kimmel, E. Zeek, P. O’Shea, A. P. Shreenath, R. Trebino, and R. S. Windeler, “Frequency-resolved optical gating and single-shot spectral measurements reveal fine structure in microstructure-fiber continuum,” Opt. Lett., vol. 27, pp. 1174–1176, 2002. [55] D. Sutter, G. Steinmeyer, L. Gallmann, N. Matuschek, F. Morier-Genoud, U. Keller, V. Scheuer, M. Tilsch, and T. Tschudi, “Semiconductor saturable-absorbermirrorassisted Kerr-Lensmode-locked tisapphire laser producing pulses inthe two-cycle regime,” Opt. Lett., vol. 24, pp. 631–633, 1999. [56] D. Jones, S. Diddams, J. Ranka, A. Stentz, R. Windeler, J. Hall, and S. Cundiff, “Carrier-envelope phase control: Femtosecond mode-locked lasers with direct optical frequency synthesis,” Science, vol. 288, p. 635, 2000. [57] C. Chudoba, J. G. Fujimoto, E. P. Ippen, H. A. Haus, U. Morgner, F. X. Kaertner, V. Scheuer, G. Angelow, and T. Tschudi, “All-solid-state Cr:Forsterite laser generating 14-fs pulses at 1.3 m,” Opt. Lett., vol. 26, pp. 292–294, 2001. [58] D. J. Ripin, C. Chudoba, J. T. Gopinath, J. G. Fujimoto, E. P. Ippen, U. Morgner, F. X. Kaertner, V. Scheuer, G. Angelow, and T. Tschudi, “Generation of 20-fs pulses by a prismless Cr YAG laser,” Opt. Lett., vol. 27, pp. 61–63, 2002. [59] P. C. Wagenblast, U. Morgner, F. Grawert, T. R. Schibli, F. X. Kaertner, V. Scheuer, G. Angelow, and M. J. Lederer, “Generation of sub-10-fs pulses from a Kerr-Lens mode-locked Cr :LiCAF laser oscillator by use of third-order dispersion-compensating double-chirped mirrors,” Opt. Lett., vol. 27, pp. 1726–1728, 2002. [60] R. K. Shelton, L. S. Ma, H. C. Kapteyn, M. M. Murnane, J. L. Hall, and J. Ye, “Phase-coherent otpical pulse synthesis from separate femtosecond lsaers,” Science, vol. 293, pp. 1286–1289, 2001. [61] R. K. Shelton, S. M. Foreman, L. Ma, J. L. Hall, H. C. Kapteyn, M. M. Murnane, M. Notcutt, and J. Ye, “Subfemtosecond timing jitter between two independent, actively synchronized, mode-locked lasers,” Opt. Lett., vol. 27, pp. 312–314, 2002. [62] A. Leitensdorfer, C. Fuerst, and A. Laubereau, “Widely tunable twocolor mode-locked Ti:Sapphire laser with pulse jitter of less than 2 fs,” Opt. Lett., vol. 20, pp. 916–918, 1995. [63] Z. Wei, Z. Kobayashi, Z. Zhang, and K. Torizuka, “Generation of two-color femtosecond pulses by self-synchronizing tisapphire and Crforsterite lasers,” Opt. Lett., vol. 26, pp. 1806–1808, 2001. [64] T. R. Schibli, L. M. Matos, F. J. Grawert, and F. X. Kaertner, “Continuum generation from a prism-less Ti:Sapphire laser,” in Ultrafast Phenomena 2002, Vancouver, Canada, May 12–17, 2002. [65] D. J. Ripin, J. G. Gopinath, H. M. Shen, A. A. Erchak, G. S. Petrich, L. A. Kolodziejski, F. X. Kaertner, E. P. Ippen, and P. Ippen, “Oxidized GaAs/AlAs mirror with a quantum-well saturable absorber for ultrashort-pulse Cr :YAG laser,” Opt. Comm., vol. 214, pp. 285–289, 2002. [66] J. Son, J. V. Rudd, and J. F. Whitaker, “Noise characterization of a selfmode-locked Ti:Sapphire laser,” Opt. Lett., vol. 17, pp. 733–735, 1992. [67] Z. Wei, Y. Kobayashi, and K. Torizuka, “Relative carrier-envelope phase dynamics between passively synchronized tisapphire and Cr-forsterite lasers,” Opt. Lett., vol. 27, pp. 2121–2123, 2002. Thomas R. Schibli was born in Zürich, Switzerland, in 1972. He received the Diploma degree in physics in 1999 from the Swiss Federal Institute of Technology, Zürich, Switzerland (ETHZ), and the Ph.D. degree in electrical engineering in 2001 from the University of Karlsruhe, Germany. In 2001, he joined the Department of Electrical Engineering and Computer Science at MIT, Cambridge, MA, as a Postdoctoral Associate, where he currently works on coherent pulse synthesis and ultrashort pulse generation for frequency metrology and time-resolved measurements. Dr. Schibli is a member of the Optical Society of America. Onur Kuzucu (S’01) received the B.S. degree in electrical and electronics engineering from Middle East Technical University, Ankara, Turkey, in 2001. He is currently working toward the S.M. degree in electrical engineering and computer science at Massachusetts Institute of Technology, Cambridge, MA. His research interests include ultrashort pulse generation with emphasis on octave-spanning lasers and frequency metrology. Mr. Kuzucu is a recipient of the IEEE Antennas and Propagation Society Undergraduate Scholarship in 2000 and Presidential Fellowship from MIT in 2001. He is student member of the Optical Society of America. SCHIBLI et al.: TOWARDS SINGLE-CYCLE LASER SYSTEMS Jung-Won Kim (S’00) received the B.S. degree in electrical engineering from Seoul National University, Seoul, Korea, in 1999. He is currently pursuing the S.M. and Ph.D. degrees in electrical engineering and computer science at the Massachusetts Institute of Technology, Cambridge, MA. From 1999 to 2002, he worked as a Research Engineer at FiberPro, Korea, where he developed fiber optic communication and measurement systems. His current research interests include synchronization of multiple mode-locked lasers and generation of single-cycle optical pulses and its application to nonlinear optics. Mr. Kim is a student member of the Optical Society of America. Erich P. Ippen (A’66–M’69–SM’81–F’84) received the S.B. degree in electrical engineering from the Massachusetts Institute of Technology, Cambridge, MA, in 1962 and the M.S. and Ph.D. degrees in the same field from the University of California in 1965 and 1968, respectively. He was a Member of the Technical Staff at Bell Laboratories in Holmdel, NJ, from 1968 to 1980. In 1980, he joined the faculty of the Massachusetts Institute of Technology where he is now the Elihu Thomson Professor of Electrical Engineering and Professor of Physics. His research interests have included nonlinear optics in fibers, femtosecond pulse generation, ultrafast processes in materials and devices, photonic-bandgap structures, and ultrashort-pulse fiber devices. Dr. Ippen is a Fellow of the American Physical Society and the Optical Society of America. He is a member of the National Academy of Sciences, the National Academy of Engineering, and the American Academy of Arts and Sciences. James G. Fujimoto (M’86–SM’91–F’96) was born in Chicago, IL, in 1957. He received the B.S., M.S., and Ph.D. degrees from the Massachusetts Institute of Technology (MIT), Cambridge, MA, in 1979, 1981, and 1984, respectively. Since 1985, he has been a Professor in the Department of Electrical Engineering and Computer Science at MIT. His research interests include the development and application of femtosecond laser technology and studies of ultrafast phenomena. He is also active in laser medicine and surgery, including the development of optical coherence tomography imaging. He was also Co-Founder of Advanced Ophthalmic Devices and LightLab Imaging. Dr. Fujimoto was awarded the Discover Magazine Award for Technological Innovation in medical diagnostics in 1999, was co-recipient of the Rank Prize in Optoelectronics in 2002, and received the IEEE Streifer Award in 2002. He is in the National Academy of Engineering and the American Academy of Arts and Sciences and is a Fellow of the Optical Society of America. 1001 Franz X. Kaertner (S’87–M’89–SM’02) was born in Cham, Germany, in 1961. He received the Diploma and Ph.D. degrees in electrical engineering in 1986 and 1989, respectively, from the Technical University, Munich, Germany. From 1991 to 1993, he was a Feodor-Lynen Research Fellow of the Alexander von Humboldt Foundation at the Massachusetts Institute of Technology (MIT), Cambridge, MA. From 1993 to 1997, he was a Principal Investigator at the Swiss Federal Institute of Technology (ETH), Zurich, Switzerland. After spending 1998 as a Visiting Assistant Professor at MIT, he joined the Department of Electrical Engineering at the University of Karlsruhe (TH), where he held the chair for Photonics and Terahertz Technology and headed the High-Frequency and Quantum Electronics Laboratory. In 2001, he joined the Department of Electrical Engineering and Computer Science, MIT. His current research interests include ultrashort pulse generation, ultrafast phenomena, frequency metrology, and noise in microwave oscillators and optical devices. Dr. Kaertner is a member of the German Scholarship Foundation, the German Physical Society, and the Optical Society of America. Volker Scheuer was born in Harxheim, Germany, in 1951. He received his state examination as a teacher for secondary school in 1978 and his diploma in physics in 1986 and the Ph.D. degree in 1994 from the University of Technology, Darmstadt, Germany. After spending one year at the University of Kaiserslautern, Germany, he continued as a Researcher at the University of Technology, Darmstadt. From 1994 to 1998, he worked on ion beam sputtering at the University of Technology. In 1998, he become Co-Founder of NanoLayers, Optical Coatings GmbH, Rheinbreitbach, Germany. Gregor Angelow was born in Darmstadt, Germany, in 1958. From Darmstadt University of Technology he received the Diploma in 1987 and Ph.D. degree in physics in 1998 for research on aspherical mirrors generating non-Gaussian beams in high power lasers. From 1987 until 1994, he worked at Darmstadt University on nonlinear optical crystals and laser beam shaping in the UV by means of the photorefractive effect. From late 1994 to 1998, he worked as an R&D Consultant for industrial laser applications. In 1998, he became Co-Founder of NanoLayers Optical Coatings GmbH, Rheinbreitbach, Germany. He is responsible for the design, development, and sale of optical coatings for high-end applications.