Документ взят из кэша поисковой машины. Адрес оригинального документа : http://www.issp.ac.ru/lsc/files/core_doped_1998.pdf Дата изменения: Thu Feb 9 10:14:53 2012 Дата индексирования: Tue Oct 2 08:48:36 2012 Кодировка: Поисковые слова: arp 220

Journal of Crystal Growth 191 (1998) 520 -- 524

Growth of sapphire core-doped fibers
V.N. Kurlov*, S.N. Rossolenko, S.V. Belenko
Institute of Solid State Physics, Russian Academy of Sciences, Chernogolovka, Moscow District 142432, Russian Federation Received 4 October 1997; accepted 20 January 1998

Abstract Single-crystal sapphire core-doped fibers (Al O --Al O :Ti>) have been grown by modified EFG/Stepanov method using automated weight control. 1998 Elsevier Science B.V. All rights reserved. PACS: 81.05.!t; 81.10.!h; 81.10.Fq Keywords: Shaped crystal growth; Sapphire single crystals; Core-doped fibers; Titanium sapphire; Modulated structures; EFG/Stepanov method

1. Introduction Single-crystal fibers of a number of refractory materials for various applications have been grown over the past several years [1--4]. The following principal techniques have been used in the growth process: edge-defined film-fed growth (EFG) method [1], the micro-pulling-down ( -PD) method [2], the laser heated floating zone directional solidification process (LHFZ) [3] and the laser heated pedestal growth (LHPG) [4] method. The availability of high-quality fiber crystals doped with active laser ions only in a sharply separated inner core region [5,6] is of special interest. Because of the step-like change of the radial doping profile of non-linear optical core-doped fibers, the

pumping energy is absorbed only by the central part and an efficient laser mode translation takes place. Single-crystal sapphire-clad ruby fibers have been grown using a two-step laser melting technique [5]. Homogeneously doped ruby fibers were first grown by a floating-zone technique from small source rods, and then these fibers were carefully surface melted in the same CO laser apparatus to outdiffuse Cr in a region near the fiber surface. In this paper we report the preparing in situ of single-crystal sapphire core-doped fibers (Al O --Al O :Ti>) by modified EFG/Stepanov method using automated weight control. 2. Experimental procedure

* Corresponding author. Fax: #7 096 5764111; e-mail: kurlov@issp.ac.ru.

Single-crystal sapphire core-doped fibers (Al O --Al O :Ti>) were grown from the melt by

0022-0248/98/$19.00 1998 Elsevier Science B.V. All rights reserved. PII S 0 0 22-02 48( 9 8 )0 011 5 - 8


V.N. Kurlov et al. / Journal of Crystal Growth 191 (1998) 520­ 524

521

the modified EFG/Stepanov technique. These experiments were done in a 8 kHz induction-heated graphite susceptor/molybdenum crucible setup held within growth chamber. Fig. 1 is a schematic of the growth apparatus. The crucible and die were molybdenum in all experiments. For preparing in situ the core-doped fibers, a crucible combined with reservoirs for doped and undoped melts, and dies feeding these melts simultaneously to the crystallization front were used. The feed material was crushed Verneuil crystals containing Ti at less than 10\ wt% in undoped row material and up to 0.5 wt% in doped row material. The atmosphere was high purity argon. Experimental runs used a C-axis (10 0012) and perpendicular to C-axis sapphire seeds to initiate growth. Growth rates were varied from 1.0 cm/min to 0.1 cm/min. The doping concentrations in the grown crystals have been measured by an X-ray microanalyzer "Camebax" provided with semiconductive detector and "Link AN 10000" system. The observation of dopant modulation in the grown crystals has been made with the help of cathodoluminescence (scann-

ing electron microscope DSN-960 ("Opton")). The image contrast depends on the contents of the luminescent impurity (Ti>) in the matrix-activator couple. Sapphire core-doped fibers have been grown using automated crystal weight control.

3. Results and discussion To grow high-performance shaped crystals with different compositions in cross-section it was necessary to solve the problem of elimination of forced mass transfer in meniscus between the various composition melts (the problem of "form") and to control the prevention of the gas bubbles and solid inclusions in the volume of the crystal (the problem of "quality"). In order to solve the first problem it is necessary to form and maintain a preset spatial component distribution in the meniscus and on the front of the crystallization in the growth process, which is determined by thermal convection, mass exchange by diffusion and convection induced by a gradient of the surface tension of the meniscus. The main parameters which determine an intensity of the mixing between the different content melts in the meniscus are the pulling rate, the meniscus height dependent on thermal gradient, the geometric sizes of the capillary channels of the dies and the form of the die top. We used the results of the calculations carried out to estimate the spreading region of the core-doped part in the meniscus depending on the pulling rate and the meniscus height with regard to diffusion in the meniscus: d "d #2(h D/»), K where d is the core diameter, d the diameter of the inner die bore, D the diffusion coefficient of the dopant in the melt, h the meniscus height and » the pulling velocity [6]. To grow a crystal with a small zone of mixing between doped and undoped parts it is necessary to achieve a small meniscus height and great pulling rate simultaneously that displaces the condition on the crystallization front to the supercooling region. Supercooling is followed by the appearance of the cellular structure and facets on the crystallization front that results in the mass capture of bubbles and solid phase inclusions.

Fig. 1. Schematic of the experimental arrangement for the growth of single-crystal fibers.


522

V.N. Kurlov et al. / Journal of Crystal Growth 191 (1998) 520­ 524

In order to control the condition on the crystallization front and to prevent formation of the related defects we have applied an automated control system using the crystal weight sensor in the process of shaped crystal growth. The principles of the control system and realization of checking the condition on the crystallization front in the growth process were described in detail in Ref. [7]. The meniscus height and the conditions on the melt-crystal interface, including overcooling and overheating, are closely related to the amplitude of oscillations of the deviation M of mass rate. Fig. 2 shows the cross-sections of fibers grown under the different thermal conditions on the crystallization

front and correspondingly under the different M Q amplitudes. These fibers of 1.2 mm in diameter have been grown using a pulling rate of 2 mm/min and program mass rate of 9;10\ g/min. Coredoped fiber of high quality (Fig. 2a and Fig. 2b) was achieved at the range of the M amplitudes from Q $0.03 g/min to $0.05 g/min. Relatively large M Q amplitudes are connected with permanent noise in the melt-crystal system independently of the quality of the pulling mechanism. Conditions of overheated melt-crystal interface resulted in smaller M Q amplitudes (less than $0.03 g/min) for the same growth rate, (Fig. 3). This leads to the mixing of the doped and undoped melts in the meniscus. Conditions of overcooled melt-crystal interface resulted in larger M amplitudes (more than $0.05 g/min), Q see Fig. 4. On the first stage of the instability process development, under the values of M ampliQ tude near $0.05 g/min, areas of cellular structure appear on the smooth crystallization front that results in the increase in the probability of impurity capture, Fig. 4a. Then, while instability increases, orientation of the cells approach singular forms. In the next stages the crystallization front is faceted that leads to the mass capture of bubbles and solid-phase inclusions, Fig. 4b. According to the analysis of the M amplitude, Q the control of the crystallization front condition was realized. In the process of the growth of variable composition fibers the control was based on

Fig. 2. The cross-section of the sapphire fiber grown at optimal M amplitude (a) and distribution of the cathodoluminescence Q intensity corresponding to the content of the luminescence impurity (b).

Fig. 3. The cross-section of the sapphire fiber grown under overheating conditions on the crystallization front.


V.N. Kurlov et al. / Journal of Crystal Growth 191 (1998) 520­ 524

523

Fig. 4. The distribution of gas bubbles and solid inclusions in the cross-section of the sapphire fiber grown with supercooling on the crystallization front: a) initial stage of cellular structure generation; b) mass capture of inclusions resulting in the faceted crystallization front.

the maintenance of the amplitude of the M oscillaQ tion in a certain, very narrow, range of values to support the condition on the melt-crystal interface close to the supercooling condition. Proportional-integral-differential (PID) procedure relating to M was used in the closed-loop of Q the automated system. The change of the heating power at a period of control resulting from PID processing of the deviation M was limited to a cerQ tain value " P" . Geometrical parameter r of the crystal (e.g. radius of the rod or thickness of ribbon, tube) involved in the calculation of the program mass and thus in the evaluation of the deviation

M as well as heating power limitation " P" were Q used mainly for the influence on the M amplitude Q and its maintenance in a certain range. We have the automatic evaluation of the geometrical parameter r from the weight signal in the controlling software during growth run [7]. So, the user has a possibility of taking into account the evaluation of the parameter r and to change its program value r without interruption of the process control. The geometrical parameter r is the main parameter for change of the M amplitude to the necessary range. Varying Q r we dislocate the middle value of the M oscillaQ tions and thus change the slope of the heating power curve. This results in the overheating or overcooling of the meniscus and the melt--crystal interface, and M amplitude decreases or increases, Q respectively. The value of the heating power limitation " P" also influences the character of the heating power curve. Sufficiently small " P" results in the in crease of auto-oscillations in the heating power (in our case of its permanent increase during the growth run) and, consequently, in larger oscillations of M, i.e. in overcooling regime. Larger Q " P" leads to more freedom in the closed-loop and to more accurate portions of power change at each step of control and to smaller M amplitude. Q Therefore, variation of limitation " P" during the growth process was also used for changing the M Q amplitude under other constant and optimal parameters of the PID controller. But sometimes it is necessary to change the parameters of the PID procedure, namely, its integral coefficient (integration time). The value of total heating power change during the whole process depends on the dimensions of the crystal profile. The greater the profile of the crystal, the greater is the total power change in our case. Larger necessary power change requires a larger integral part in the PID "mixture" to provide the same, sufficiently small, regulation error. Supercooling on the crystallization front in the process of growing the variable composition crystals can be nonuniform because of the constitutional supercooling in the zone of more refractory melt, that results in the appearance of solid-phase inclusions in the doped part of crystal, Fig. 5. To obtain the optimal distribution of temperature in


524

V.N. Kurlov et al. / Journal of Crystal Growth 191 (1998) 520­ 524

(Ti>) part was 0.4--1.2 mm in diameter. The fibers free of the bubbles and solid-phase inclusions have been prepared using automated weight control. Concentration of Ti> was about 10\ wt% in the undoped part and about 0.1 wt% in the core part of crystal.

Acknowledgements We are indebted to A.B. Ermolaev and P.I. Antonov for the assistance and a number of helpful discussions relating to this work.
Fig. 5. The constitutional supercooling in the core part of fiber.

References
[1] H.E. LaBelle Jr., A.I. Mlavsky, Mater. Res. Bull. 6 (1971) 571. [2] D.H. Yoon, I. Yonenaga, N. Ohnishi, T. Fukuda, J. Crystal Growth 142 (1994) 339. [3] J.S. Haggerty, W.P. Menashi, NASA, Final Report Contract No. NAS 3-13479, February, 1971. [4] R.S. Feigelson, J. Crystal Growth 79 (1986) 669. [5] C.A. Burrus, L.A. Coldren, Appl. Phys. Lett. 31 (1977) 383. [6] P. Rudolph, K. Shimamura, T. Fukuda, Crystal. Res. Technol. 29 (1994) 801. [7] V.N. Kurlov, S.N. Rossolenko, J. Crystal Growth 173 (1997) 417.

the crystallization zone, the heights of the edge, the inside and outside parts of the die were varied.

4. Conclusions The single-crystal sapphire core-doped fibers have been grown by the modified EFG/Stepanov technique using automated weight control. The outer diameter was 1.0--1.8 mm and the core-doped