1. Introduction

Essentially, all fiber lasers in use today are made of glass. These glass structures normally involve a double-clad structure in which the core glass has been doped with a variety of rare-earth ions, most notably are ytterbium-doped fiber lasers. These glass fiber lasers deliver a very high cw power or pulsed energy, with a single mode (SM) output and broad gain bandwidths. Yet, there are some limitations to power scaling of glass fiber lasers which result from laser induced damage to the small cores, nonlinear effects, and thermal loading. The purpose of this work is to develop a new and novel class of high power fiber lasers based on crystalline materials rather than the conventional glass fiber structure. The basic premise of this work is the rather straightforward idea that rare-earth doped, crystalline materials such as YAG and other garnets are known to deliver extremely high laser powers. For example, the technology of Nd:YAG lasers employing conventional rods and disks is well established and reliable. Our approach is to draw from the broad knowledge base for Nd:YAG and other solid-state lasers to extend this technology to fabricate single-crystal (SC) fiber lasers. SC fiber lasers would be scalable to much higher power levels compared to their glass counterpart, largely due to the fact that SC fiber lasers have a significantly higher thermal conductivity and also significantly reduced nonlinear effects (i.e., SBS, SRS) compared to glass fiber lasers [1].

SC fibers have been grown since the early 1980s, with most of the work concentrating on passive (pure) SC sapphire fibers [2]. Both sapphire and YAG have melting points exceeding 1900 °C, high thermal conductivity, and relatively high refractive indices. The advantage of YAG over sapphire is that it is possible to dope YAG with rare-earth ions, making YAG crystals excellent laser hosts. For this reason, there has been more work of late on growing both pure and doped YAG SC fibers. Both sapphire and YAG fibers have essentially all been unclad, in the sense that there has been no true fiber cladding as is commonly associated with glass core/clad fiber lasers. Furthermore, some of the rare-earth doped YAG fibers have been rather large in diameter (~1,000 μm) and thus may be more properly regarded as a minirod lasers and not at all a flexible fiber laser [3]. In this paper, we have studied the basic optical properties of pure YAG fibers to better understand the higher losses in these fibers compared to the intrinsic losses. For these fibers to be a viable host for rare-earth dopants and laser applications, it is necessary to lower the losses significantly but also eventually clad the fibers in a manner similar to the double-clad structures of glass fiber lasers. The work done on rare-earth doped SC YAG fibers have been reported previously [4].

The approach we have used to fabricate SC fibers is Laser Heated Pedestal Growth (LHPG) [2]. In this well-established technique, the tip of a single-crystal or ceramic YAG preform is melted with a CO₂ laser and an SC fiber is pulled from the molten oxide. Using this technique we have grown SC YAG fibers with diameters between 90 to 400 μm and lengths as long as 1 m. In general most of these fibers have been unclad. However, there is ongoing research designed to achieve a cladding using either a sol-gel coating applied to the as-grown fiber or creating a graded index structure in YAG by using a specialty designed rod-in-tube preform for the LHPG fiber growth.

2. Background and experimental setup

SC fibers have been grown by several methods, including Edge-Defined Film Fed Growth (EFG) [5], micro-pulling-down [6], and the LHPG method [2]. The most common method used today is LHPG. This method was first developed by Haggerty [7] at MIT and then further refined by a number of researchers including those at Stanford University [8], Bell Labs [9], Rutgers University [10], University of South Florida [3], Shasta Crystals, Inc [11], and others [12]. LHPG is a crucibleless method much like the float zone technique. In the LHPG technique, a CO₂ laser beam is focused onto the tip of an oxide crystal source rod, creating a small molten bead. A seed fiber is dipped into the molten region and slowly pulled upward, forming the single-crystal fiber. The source rod, which may be single or polycrystalline (ceramic), is simultaneously fed upward to replenish the supply of molten material. A schematic diagram of our setup is shown in Fig. 1.

 

Fig. 1 Experimental set-up of the LHPG technique used to grow SC YAG fibers. The use of transmissive axicons rather than a reflective optics allows us more flexibility in adjusting the optical parameters of the ring of CO₂ laser light.

Download Full Size | PDF

A unique improvement of our LHPG apparatus is the use of a pair of AR-coated ZnSe axicons to convert the Gaussian CO₂ laser beam into a ring, rather than the more common method employing a reflective (refraxicon) focusing system [8]. The growth speed is between 1 to 4 mm/min and the source-to-fiber reduction for stable growth is about 3:1. Almost all of our starting source material is single-crystal YAG with dimensions 1 x 1 x 120 mm. The 25 W CO₂ laser is amplitude stabilized by taking a small power from the rear output mirror and inputting this signal to a LabView program (see Fig. 1). This program returns a signal to the power supply to control the power to ± 0.2%. In addition to amplitude stabilization, we also use a laser micrometer to measure and control the fiber diameter.

3. Losses in SC fibers

3.1 Spectral losses

The spectral absorption in YAG fibers was measured using an FTIR (Bruker Tensor) and visible-near IR spectrometer (Ocean Optics H4000). The spectrum of a 6-cm long, 330 μm diameter YAG fiber, not optimized to reduce background loss, is shown in Fig. 2 in arbitrary units. The spectrum is essentially featureless from the visible to about 4 μm. The absorption peak near 4.8 μm is an intrinsic multiphonon peak fundamental to all YAG crystals. The data shown in Fig. 2 is essentially what would be obtained for a bulk YAG crystal. One interesting aspect of this data is that it does not show a particularly high absorption in the visible region. This is in contrast to SC sapphire fibers, which exhibit a much greater increase in loss at short wavelengths [3,10].

 

Fig. 2 Spectral loss for a SC YAG fiber. The peak around 4.8 μm is intrinsic to YAG.

Download Full Size | PDF

3.2 Total losses in YAG fibers at different laser wavelengths

There has been considerable loss data taken on sapphire and other SC oxide fibers, but most of this data has been either broadband data taken using spectrometers or at selected laser wavelengths such as at the Er:YAG laser wavelength of 2.94 μm. The motivation for much of the early development of sapphire fibers was to use these fibers for the delivery of Er:YAG laser power in surgical applications [13]. Today there seems to be a greater emphasis on doped YAG fibers for laser applications. Therefore, one of the primary purposes of this work is to better understand the source of the losses in our undoped SC YAG fibers so that we may fabricate RE doped SC YAG fibers with lower loss.

We have taken a closer look at the losses for our YAG fiber through careful measurements at selected laser wavelengths using the cut-back method. For all the lasers, the light was coupled into the fiber using a 9 cm focal length lens, resulting in a spot size smaller than the fiber core. The high index of YAG and the absence of a material cladding allow for all angles to be accepted into the fiber. This, in addition to a large fiber diameter, results in a highly multimode fiber.

In this work we present, to our knowledge, the lowest published loss of SC YAG fibers. Two notable improvements to the LHPG process have led to loss reduction compared to previous work [14]. First, old degraded refraxicon optics were replaced with transmissive axicons and secondly, the usage of guide tubes whose inner diameters more closely matched the diameter of the fiber. During the past three years there has been an overall reduction in loss from over 1 dB/m at 1.06 μm to 0.3 dB/m. In Fig. 3 we show the total loss for one of our 330-μm diameter, 40-cm long SC YAG fibers at four laser wavelengths.

 

Fig. 3 Loss for SC YAG fiber at four laser wavelengths.

Download Full Size | PDF

The data in Fig. 3 are for an unannealed, as-grown fiber. As mentioned in the discussion above, sapphire fibers have a much greater increase in loss from the IR to visible, compared with the slight increase in YAG fibers seen here [10]. We have observed that the transmission of YAG fibers does not improve with annealing (12 hr at 1,200 °C in air) for visible or IR wavelengths. The only time that annealing has proven helpful for these fibers are when the as-grown fibers have high losses in the visible region. Then, we do see some improvement in loss after annealing. However, in general, we no longer anneal any of our fibers.

3.3 Scattering losses in YAG fibers at different laser wavelengths

The total losses, α_T, are a sum of the scattering, α_s, and absorptive, α_a, losses or ${\alpha }_T={\alpha }_s+{\alpha }_a$. To better understand the nature of the losses for the measured α_T we have made separate measurements of the scattering losses. We used a two-inch integrating sphere (Thorlabs IS200) with a silicon photoconductive detector to measure the scattering contribution to the total loss. The same four lasers used for the data in Fig. 3 were used as the source lasers, and the integrating sphere was moved along the fiber from the input to the output end. YAG has a high refractive index, so there is a substantial signal from the output end reflection when the sphere is near the output end of the fiber. Therefore, scattering from the output end is omitted when α_s is computed from the data. The loss due to scattering is calculated using,

 

Fig. 4 (a) Scattering loss as a function of the distance of the integrating sphere along the length of the fiber. The data shown here were using a 532 and 1064 nm laser. (b) Scattering loss measured at the same four laser wavelengths as for the total loss in Fig. 3.

Download Full Size | PDF

4. Facet formation in SC fibers

One of the more unusual effects in SC fibers is the tendency for facets or small ridges to be present on the outer surface of the fiber. These facets have been seen by other investigators for YAG as well as for sapphire fibers [6,12,16,17]. We show some typical facets on the surface and also the cross section of our YAG fibers in Fig. 5. The fiber shown in Fig. 5(a) was grown from an unoriented seed fiber. The facets are spaced about 21 μm apart with a height of less than 40 nm for a 330-μm diameter fiber grown at a speed of 2 mm/min. The fiber shown in Fig. 5(b) was grown from a seed fiber oriented 15° from the [100] to [110] plane. This orientation of the seed fiber was chosen based on the work of Ishibashi, et al. [16] and Kitamura, et al. [18] who studied the growth of YAG fibers grown from different seed orientations. They showed that this particular orientation yielded essentially facet-free fibers but the cross section of the fibers is square shaped [19]. In general, faceting is a common occurrence in YAG crystal growth even for bulk crystals [20]. In addition, stress during bulk crystal growth leads to stress birefringence which is typically found in the neighborhood of the central core. For bulk crystals, the YAG rods or bars for laser applications can be mined outside this area of strain to yield a more efficient laser rod. We have looked for strain fields in our fibers by viewing the fibers under cross polarizers. At this point we do not observe any significant stress-induced birefringence in the fibers.

 

Fig. 5 Surface and cross section of (a) a fiber grown from an unoriented seed and (b) another fiber which was grown using an oriented seed.

Download Full Size | PDF

5. Summary and conclusions

SC YAG fibers are emerging as a potentially excellent source of fiber optics that may compete with glass fiber lasers. In this work, we have studied the basic optical properties of pure YAG fibers to better understand the mechanisms for the losses which exceed the intrinsic losses for YAG. For these fibers to be a viable host for rare-earth dopants and laser applications, it is necessary to lower the losses significantly but eventually also to clad the fibers in a manner similar to the double-clad structures of glass fiber lasers. While cladding is a challenge, there have been studies involving a glass overclad on YAG fibers [17]. While this technique does provide a cladding that can operate in the 1 to 2 μm region, it is not ideal from the viewpoint of exploiting the high power capability predicted for YAG, a capability based on a much higher SBS threshold and thermal conductivity compared to glass. Our work on a true crystal clad for YAG fibers is encouraging, but there is still much to be done.

We have studied the absorption losses for YAG fibers grown by the LHPG method. The total absorption was found to be highest at the shortest wavelength of 532 nm, decreasing to as low as 0.3 dB/m at 1064 nm, currently the lowest published loss. The scattering losses for the fibers were measured using an integrating sphere that was moved along the length of the fiber. When we omit the scattering from the output end of the fiber, which is abnormally high due to end reflections, we find that scattering contributes about 30% to the total loss. We are not certain of the origin of this scattering, but it is suspected that the scattering centers have a size larger than the wavelength of light.