Dopant segregation in YAG single crystal fibers grown by the laser heated pedestal growth technique

Highlights

• Single crystal fibers of good optical quality were grown by LHPG technique.

• Radial dopant segregation in single crystal fiber controlled by growth speed.

• Preferential radial segregation of Nd ions compared to Ho ions indicating dependence of segregation on size of dopant.

• Sol-gel based technique to add Ho dopants to single crystal YAG fiber is demonstrated.

• The calculated Δn due to segregation of Nd ions for such a fiber was determined to be 1.2 × 10⁻⁴.

Abstract

In this study, we report self-segregation of dopants in a crystal matrix within a single crystal (SC) fiber. Neodymium and holmium-doped yttrium aluminum garnet (YAG) fibers were grown using the Laser Heated Pedestal Growth (LHPG) technique and cross-sectional dopant concentration was measured using electron-probe micro-analysis. It was observed that the degree of auto-segregation of the rare-earth dopant depended on the difference in ionic size of the dopant ion and the Y³⁺ ion in the YAG matrix. While holmium showed little tendency to self-segregate, the concentration of neodymium ions varied as much as 25% across the cross-section of the fiber. Strong correlation between the dopant concentration profile and fiber draw speed was also demonstrated. Since the local refractive index depends on the concentration of dopants, a refractive index profile can be achieved by a dopant profile across the fiber cross-section. Engineered index profiles can be realized by varying growth conditions, dopants, crystal matrix, etc. Such an approach is promising in applications such as the development of monolithic SC fibers with graded-index profiles.

Introduction

Despite successful application of glass optical fibers in development of devices such as fiber optic sensors, fiber lasers, etc.; applications requiring extreme material properties have proven challenging given silica stability limits. Conditions such as high temperatures, radioactivity, high energy fluences, corrosion, etc., limit applicability of glass fibers. For such applications, single crystal (SC) fibers have shown exceptional promise. Crystalline fibers grown from materials like sapphire or YAG are thermally, chemically, and mechanically more robust compared to silica fibers, making them ideal for extreme environment sensing applications such as in boilers, jet engines, and nuclear reactors [1]. Additionally, high thermal conductivity and low nonlinear gain coefficients, coupled with high surface-area-to-volume ratio of the fiber geometry make SC fibers suitable for applications involving high optical power density [2], [3].

Single crystal fibers, grown by processes like the LHPG technique, do not intrinsically have a functional cladding. The cladding is not only essential to confine light and reduce frustrated total internal reflection that may arise due to contact with the surface when the fiber is deployed in various applications, but it also helps reduce the modal volume required for distributed sensing applications [4], and cladding pumping schemes in high power laser applications [5]. Glasses have been explored, to limited success, as potential cladding materials for crystal fibers [6], [7]. For high temperature or high-power laser applications, the differences in thermal and mechanical properties of glasses and crystals are exacerbated and lead to significant losses and ultimately catastrophic device failure. To preserve homogeneity of material, and exploit the advantages of a crystalline phase, growing a crystalline cladding is essential.

While a few post-cladding techniques (where a crystalline layer is grown on an existing fiber) have been proposed [8], these procedures usually involve two separate growth processes, which not only introduce scattering losses at the core-clad interface but also increase the time and cost involved for production. Most importantly, current reports on these alternative growth methods reveal limitations in either maximum length of clad fiber, maximum cladding thickness, temperature stability, or finished fiber quality [5], [8], [9], [10], [11]. Hence, a single step process is more desirable. Unlike drawing of glass fibers, starting from a rod-in-tube type preform is not effective as low viscosity in the liquid molten pool causes intermixing and homogenizes the melt composition [12]. Instead, in-situ fabrication of a graded index fiber can be accomplished by leveraging crystal growth melt dynamics.

Previous researchers have shown that certain LHPG-grown crystal fibers exhibit non-uniform radial distribution of a dopant species [13]. Depending on dopant ions, a tendency may be observed for the dopant species to segregate towards the center of the fiber or the periphery, creating a distribution profile. In magnesium (Mg) doped sapphire, for example, the center axis of the fiber showed a much higher concentration of Mg²⁺ ions compared to fiber edges [13]. Such an effect has also been observed in YAG SC fibers as well. While neodymium (Nd³⁺) ions were observed to concentrate at the center axis [14], [15], chromium (Cr⁴⁺) ions were observed to have higher concentration near the fiber periphery [16].

The segregation of the ions in the YAG matrix were argued to correlate with their size, as well as the differential solubility of the dopant in the solid and liquid phases of the host crystal matrix [17]. The latter is quantitatively described by the segregation coefficient (k) which is the ratio of the solubility in the solid phase to solubility in the liquid phase. Hence, if the value of k is 1, it implies there is no depletion or accumulation of dopants in the melt when the crystal grows, which is indicative of a homogenous solid solution in the crystal. The segregation coefficient also seems to be related to the relative size difference between the dopant ion and the host ion it replaces. Ions such as Nd³⁺ and cerium (Ce³⁺), which are larger in size compared to the yttrium (Y³⁺) ions (see Table 1), have k < 1 (see Table 2) and tend to segregate towards the center of the fiber. On the other hand, ions like Cr⁴⁺, which are smaller compared to Y³⁺ ions, have k > 1 and tend to segregate towards the outer rim of the fiber. In addition, ions such as Yb³⁺, which is very similar in size to Y³⁺ ion (see Table 1), has a reported k value of about 1 in YAG. During crystal growth, Yb ions indicate no tendency to self-segregate. The same should be also true for ions such as Ho, Er, Tm, etc. In short, the value of k determines where a particular dopant species will move radially with respect to the center axis of the fiber during growth. Since the refractive index of YAG changes linearly with addition of dopant ions (see Table 3), auto-segregation of dopant ions in YAG can be explored as a means to create an effective refractive index gradient to achieve a monolithic core-clad structured crystalline fiber.

In this work, we present a more careful look at growth of graded index single crystal fibers via adding dopants to pure fibers and re-growing them with LHPG. The selective segregation of dopants across the fiber cross-section demonstrates a concentration distribution profile, which is expected to also approximate a graded refractive-index profile. Control of the concentration distribution profile is engineered not only by varying dopant species and the fiber matrix material, but also growth conditions like fiber draw speed.