A new optically pumped waveguide amplifier with ultra-large mode area is proposed. This amplifier is based on gain guiding in a transverse grating waveguide in which the pump is confined by the photonic bandgap while the signal is guided by optical gain. Characteristics of the propagating modes of the waveguide amplifier are analyzed theoretically using the transfer matrix method, indicating robust single-transverse-mode operation with large modal gain.
©2008 Optical Society of America
High-power lasers and amplifiers with excellent beam quality are important to many defense and civilian applications. Power scaling of waveguide-based lasers and amplifiers is often limited by optical nonlinearity and optical damage of host materials induced by the high peak power. To mitigate these catastrophic problems while maintaining large output power, waveguides with large mode area (LMA) are highly desired. Several schemes have been reported, beyond the conventional approach of index guiding, to achieve waveguide lasers and amplifiers with ultra LMA (ULMA), i.e., a modal diameter of 100 µm or above. Lang et. al. have proposed lateral grating-confined broad-area waveguides in which one-dimensional gratings are used laterally to confine the transverse modes of waveguides . Utilizing the narrow modal selectivity of gratings at oblique incident angle, waveguides with transverse gratings enable strong modal discrimination among higher-order modes (HOMs), and single HOM operation was demonstrated in waveguides with a core diameter of 300 µm. Yariv has proposed a similar configuration except the gratings contain gain to provide both lateral confinement and optical amplification for the signal . Due to the finite penetration of the Bloch modes into the gratings, the effective modal area of such waveguides can be much larger than the core dimension .
Recently Siegman has proposed the use of optical gain in index-antiguided waveguides to achieve single-transverse-mode operation with ULMA . Such a scheme has been experimentally demonstrated in a flash-lamp-pumped neodymium-doped phosphate fiber with a core diameter of a few hundreds microns [5,6]. Index antiguiding (IAG), although effectively reducing the threshold of gain guiding (GG), could trap pump light in the cladding leading to low optical gain in the core when the waveguides are end-pumped . This would be an unwanted situation because end pumping is one of the most desirable attributes of waveguide lasers and amplifiers. In this article I propose a new scheme that could potentially resolve the above issue. By enabling gain guiding in transverse grating waveguides, I show that the signal could have robust single-transverse-mode operation with ULMA, and at the same time the optical pump can be confined in the same core to achieve high optical gain and conversion efficiency. This scheme can be extended to other photonic bandgap based structures, and is potentially attractive for optically pumped high-power lasers and amplifiers.
2. Principle and analysis
The working principle of gain-guided transverse grating waveguides (GG-TGW) is as follows. Consider a one-dimensional transverse grating waveguide composed of a dielectric slab sandwiched by two semi-infinite gratings layered in the y-direction that is perpendicular to light propagation direction (Fig. 1). This waveguide resembles conventional distributed-Bragg-reflector (DBR) waveguides except the DBR is applied transversely rather than longitudinally. The core has a complex refractive index nc=nco+ini where ni is related to the power gain coefficient by g=2niko and ko=2π/λ is the vacuum wavenumber. The gratings are assumed to have very small index variation such that their optical properties can be well described using coupled mode theory in terms of average refractive index ncl and coupling constant κ.  Furthermore, I consider that nco is slightly less than ncl to exclude the total internal reflection guiding mechanism.
It is well known that a one-dimensional grating can resonantly reflect light with wavelength λ at an incident angle θR, provided the grating pitch Λ meets the Bragg condition Λ=λ/(2nclsinθR). At resonance the reflectivity of the grating equals to tanh2(κw), which approaches unity if the coupling constant κ is large and/or the length of grating w is long.  Given θR, the angular bandwidth of a grating resonance is approximately 2κ/(koncl cosθR), which could be very narrow if θR is small. Thus, GG-TGW is designed to have a strong and sharp resonance at the pump wavelength by employing large κw and small (but finite) θR. In such case the pump light can be effectively trapped inside the core via Bragg resonance, even though nco is smaller than ncl. On the other hand, the signal we are after has a single-transverse mode (i.e., the lowest order mode with mode order equals 0) that has a grazing incident angle in a large core. Due to its large mismatch from the grating resonance, the signal experiences the grating as a slab with refractive index ncl, which is slightly larger than nco as previously defined. Thus, the waveguide behaves like an IAG slab waveguide for the signal. With appropriate material gain in the core via optical pumping (i.e., ni>0), gain guiding is expected to take place and yield a single-transverse mode with ULMA .
Modal characteristics of the proposed structure are analyzed using the transfer matrix method, following Lang et. al with minor modification . As shown in Fig. 1, the waveguide consists of three layers, including two identical semi-infinite gratings (j=0,2) and a slab (j=1). I adopt a time factor of exp(iωt) so that a forward propagating wave (traveling in the +z direction) with a complex propagation constant βc=β+ig will have a spatial dependence of exp(-iβcz), where a positive g represents gain of the electric field. Within each layer, the transverse field profile can be written as a superposition of a forward-propagating wave E +(y) and a backward-propagating wave E -(y) that are the eigenfunctions of Helmholtz equations for that layer. That is,
where aj represents the complex field amplitude and yj is the beginning coordinate of the j-th layer. For a slab, the eigenfunctions are the well-known plane waves
where k ⊥=[(ncko)2-βc2]1/2 is the transverse wavevector of a plane wave traveling obliquely in the slab. For a grating, the forward and backward traveling waves are strongly coupled and can be calculated from coupled mode theory to be
where σ≡π/Λ is the grating resonant wavenumber, r≡κ/(S+k ⊥-σ), and S≡[(k ⊥-σ)2-κ 2]1/2. In the above definition, I have chosen positive roots for k ⊥ and S to allow a solution of leaky modes that have outgoing phase and are unbounded in the waveguides .
The field amplitudes aj + and aj - are related to those of the first layer via electromagnetic boundary conditions by 
where dj is the thickness of the j-th layer and Mj is the corresponding 2×2 transfer matrix. For a slab,
and for a grating,
A mode of the waveguide is a field distribution in which the fields in the boundary layers are purely outgoing, i.e., a pure backward-propagating wave in the j=0 layer and a pure forward-propagating wave in the j=2 layer. Assuming a0 -=1 and solving a2 - (βc)=0 will yield βc of the eigenmodes from which the modal index β/ko and the modal gain gm can be determined. The field distribution of the eigenmodes can then be calculated according to Eqs.(1)–(6).
3. Result and discussion
I consider a TGW with the following parameters: nco=1.5, core width d=125 µm, pump wavelength λp=980 nm, signal wavelength λs=1550 nm, and κ=100 cm-1. The grating also has an average refractive index ncl=1.5005 and a resonant incident angle θR=20°. Figure 2(a) shows the loci of the complex effective indexes neff=βc/ko of the eigenmodes at 980 nm for a TGW (red curve) and a conventional IAG-slab waveguide (blue curve) with otherwise identical parameters. All modes are essentially leaky due to index antiguiding. As expected, the conventional IAG-slab waveguide exhibits a monotonically increasing loss at higher mode order. The transverse grating waveguide, on the other hand, shows a similar trend at small mode order but exhibits a sharp increase in reflectivity near θR. The eigenmode that resides in the stopband of the gratings has essentially zero propagation loss and is hereafter named the “gap mode”. This gap mode, whose transverse wavevector is set by the grating period to be π/Λ, features a very large mode order (~250) with nearly uniform interference fringes across the entire core as shown in the inset of Fig. 2(b). The amplitude of the gap mode decays exponentially into the gratings, as is typical for a Bloch mode in the stopband of a grating . As a comparison, Fig. 2(b) also shows the fundamental mode (blue curve) of the IAG-slab waveguide. It has a relatively small loss coefficient ~0.05 cm-1 at 980 nm due to grazing incidence; its leaky nature, however, is evident as its field amplitude is nonzero and rises gradually in the cladding.
Figure 2(a) also shows the loci (green curve) of the eigenmodes at 1.55 µm in a passive TGW. The curve is smooth within the first few hundred modes since the corresponding θR at 1.55 µm occurs at an even larger incident angle and thus a smaller modal index. For a passive TGW, all modes are lossy (i.e., gm<0) due to radiation loss in an index antiguiding waveguide. As the material gain g in the core increases by optical pumping, it compensates for the radiation loss and gm increases linearly, as shown in Fig. 3(a) for the three lowest modes. For material gain g between 0 and 5 cm-1, modal gain gm,i equals g-gth,i, where gth,i is the threshold gain for the onset of gain guiding and equals 0.42, 1.66, and 3.64 cm-1, respectively, for the first three modes. In fact, gain does more than compensating for the radiation loss; it also substantially changes the modal confinement. Figure 3(b) shows the field amplitude of the fundamental mode at various pump gains. At g<gth,0=0.42 cm-1, the modes are leaky and their amplitudes rise exponentially in the cladding; at g>gth,0, the modes stay roughly the same in the core but change significantly in the cladding from diverging exponentially to converging exponentially. Behavior of this type is consistent with that in a conventional IAG-slab waveguide , indicating that the lower-order modes, which are off-resonant with the gratings, indeed experience the grating like a uniform slab as previously suggested.
Figure 3(a) also indicates that, for a given material gain g, the difference in modal gain between the fundamental (i=0) and first HOM (i=1) is 1.24 cm-1, which is nearly 3 times the threshold gain of the fundamental mode gth,0. This suggests potentially strong modal gain discrimination between the fundamental modes and other higher-order modes, and the signal can sustain robust single-transverse mode as long as g<gth,1. The largest modal gain for a pure single-transverse mode can be obtained at the onset of the first higher-order mode, i.e., gm,0max=gth,1-gth,0. In the present example, gm,0max equals 1.24 cm-1 at a material gain of 1.6 cm-1. Even larger gm,0max is possible by further reducing the index contrast between the core and the gratings . Thus, GG-TGW holds promise for robust single-transverse-mode operation with high gain.
The GG-TGW amplifiers proposed here could be suitable as power amplifiers in masteroscillator- power-amplifier systems where the last-stage amplifiers need to deliver very high power output while maintaining an excellent beam quality. Such devices could be potentially realized in rare-earth-ion-doped glass waveguide amplifiers. Take the present TGW amplifier as an example and assume sinusoidally modulated gratings used as cladding. A grating with κ=100 cm-1 and θR=20° implies an index modulation with an amplitude of ~0.001 and a period of ~0.96 µm. Assuming the device is 5 cm long and the target total gain is 20, this requires a modal gain gm,0=0.6 cm-1. Given a core width of 125 µm and an index difference nco-ncl=-5×10-4, this corresponds to a pump gain coefficient ~1 cm-1 that is well below the onset of the first higher-order mode (gth,1=1.66 cm-1) and the device will therefore operate with a pure single-transverse mode. Such specifications can be most probably achieved in Er:Yb co-doped phosphate glasses (Er3+ density ~1021 cm-3) where a device length up to 4 cm  and a net gain of more than 2 cm-1 has been experimentally demonstrated . The vertical confinement of the guided modes can be achieved by the ion-exchange technique which allows precise tuning of the index contrast between the core and the gratings by controlling the salt concentration, exchange time, or temperature.  The transverse grating layers can be fabricated using reactive ion etching  or by UV-induced densification of phosphate glass followed by chemical etching.  Both methods are suitable for large-area patterning of uniform structures and have been applied to fabricate longitudinal DFB/DBR gratings in such materials.
The bandwidth of a Bragg grating at its resonance is Δλ=λp2κ/(πnclsinθR), and this in principle imposes a minimum spectral separation between the pump and signal used in the proposed TGW scheme. For the present example, this corresponds to 15 nm which is smaller than the spectral separation in practical rare-earth-ion-doped laser/amplifier systems. In practice, however, the spectral separation is desired to be a few times larger than Δλ to avoid the finite reflectivity at the side lobes of the gratins. On the other hand, Δλ in principle can be made very small if a weak grating (and correspondingly a long grating length) is employed.
I propose a new optically-pumped waveguide laser amplifier based on gain guiding in transverse grating waveguides to support an ultra-large mode area with a high modal gain. In this configuration, the pump is guided in the core via the Bragg resonance while the signal is confined by gain guiding. Modal analysis based on the transfer matrix method confirms robust single-transverse mode operation over a large span of material gains. Such a scheme is attractive for high-power waveguide-based lasers and amplifiers.
The author thanks Dr. Lee Casperson at UNC Charlotte for his valuable inputs and proofreading of this manuscript. The author also acknowledges discussions with Dr. Kirankumar Hiremath at UNC Charlotte. This work was sponsored by Department of Defense Joint Technology Office Multidisciplinary Research Initiative contract W911NF-05-1-0517.
References and links
1. R. J. Lang, K. Dzurk, A. A. Hardy, S. Demars, A. Schoenfelder, and D. F. Welch, “Theory of gratingconfined broad-area lasers,” IEEE J. Quantum. Electron. 34, 2196–2210 (1998). [CrossRef]
4. A. E. Siegman, “Propagating modes in gain-guided optical fibers,” J. Opt. Soc. Am. A. 20, 1617–1628 (2003). [CrossRef]
5. A. E. Siegman, Y. Chen, V. Sudesh, M.C. Richardson, M. Bass, P. Foy, W. Hawkins, and J. Ballato, “Confined propagation and near single-mode laser oscillation in a gain-guided, index antiguided optical fiber,” App. Phys. Lett. 89, 251101 (2006). [CrossRef]
7. V. Sudesh, T. McComb, Y. Chen, M. Bass, M. Richardson, J. Ballato, and A.E. Siegman, “Diode-pumped 200 µm diameter core, gain-guided, index-antiguided single mode fiber laser,” Appl. Phys. B 90, 369–372 (2008). [CrossRef]
8. A. Yariv and P. Yeh, Optical Waves in Crystals: Propagation and Control of Laser Radiation (John Wiley & Sons, 2003).
9. P. Yeh and A. Yariv, “Bragg reflecting waveguides,” Opt. Commun. 19, 427–430 (1976). [CrossRef]
10. A. E. Siegman, “Gain-guided, index-antiguided fiber lasers,” J. Opt. Soc. Am. A 24, 1677–1682 (2007). [CrossRef]
11. K. Liu and E. Y. B. Pun, “Modeling and experiments of packaged Er3+-Yb3+ co-doped glass waveguide amplifiers,” Opt. Commun. 273, 413–420 (2007). [CrossRef]
13. D. L. Veasey, D. S. Funk, N. A. Sanford, and J. S. Hayden, “Arrays of distributed-Bragg-reflector waveguide lasers at 1536 nm in Yb/Er co-doped phosphate glass,” Appl. Phys. Lett. 74, 789–791 (1999). [CrossRef]
14. S. Pissadakis and C. Pappas, “Planar periodic structures fabricated in Er/Yb co-doped phosphate glass using multi-beam ultraviolet laser holography,” Opt. Express 15, 4296–4303 (2007). [CrossRef] [PubMed]