A Laser-Driven Light Source (LDLS™) With Multiple Fiber Outputs

 

Author:X. Ye*, Q. Wang, M. Dube, D. Gustafson, H. Zhu

Abstract: A laser-driven light source (LDLS™) system with multiple fiber outputs is developed for photonic or semiconductor processing applications. Simulation and measurement results of fiber- to-fiber uniformity from 380nm to 1000nm are presented.

1. Introduction

As the demands for ever-smaller device features in integrated circuits continue to increase, the needs for high brightness illumination sources used for inspection of these ever-shrinking devices continue to grow. One such illumination source is a laser-driven light source (LDLS™) [1], which is capable of producing high brightness broadband light output. LDLS operates by focusing laser radiation into a gas volume to excite the xenon gas fill into a plasma state, which could emit light for various applications.Due to its unique physics principles, the light output from an LDLS plasma features a higher spatial and temporal stability, longer lifetime [2], a much smaller plasma size with a much higher radiance [3]. These features make LDLS a better choice for applications in fiber-based light sources. This paper presents the design features and performances for an LDLS system having six fiber outputs. Spectrum uniformities among the six output fibers are discussed using experimental results.

 

2. Operation Principles

A schematic diagram for the LDLS system having a multi-fiber output is shown in Fig. 1. The Xe plasma output is collected and collimated by OAP-1 and focused by OAP-2 to a light pipe input end. The input end of the light pipe is positioned at the focal point of OAP-2 to homogenize the beam. A fiber bundle with seven channels is attached to the output end of the light pipe to divide the light into seven channels. The outer six channels have uniform output.

 

 Fig. 1 The sketch of the operation for the LDLS system with multiple outputs.

 

3. Design Parameters and Simulation Results

In the LDLS system, the light collection NA for OAP-1 is around 0.37, to form a collimated beam. The focal length of OAP-2 is equal to two times as OAP-1, which reduces focusing NA to 0.18 and increases the plasma image size by 2x on the OAP-2 focal plane. The aperture of the hexagon light pipe has a 2mm dimension. The fiber bundle contains seven fibers with a 400µm core diameter.

Fig. 2 shows the Zemax simulation results of the proposed design. In the simulation, the same experimental OAP parameters were used. Both of the OAPs have a 1.5-inch diameter, with aluminum coatings on their reflecting surfaces. The hexagon light pipe made of fused silica has a 25mm length. The fiber bundle comprises seven channels, which are made of 400µm core diameter broad-band multimode fibers with a 0.22NA. Fig. 2 (a) and (b) show the incoherent irradiance view on the output end of the light pipe and the input surface for the fiber bundle respectively. Fig. (2) (c) shows simulated results for the spatial irradiance variation on the light pile exit surface. From calculation, the spatial irradiance variation (defined as (I_max – I_min)/ I_mean) for the proposed design is only 11% in an area of 2mm by 2mm on the output end of the 25mm hexagon light pipe.

 INSERT FIGURES

 Fig. 2 The sketch of the operation for the LDLS system with multiple outputs.

4. Experimental Results

Experiments were conducted by using reflective optics to eliminate chromatic aberrations, which includes two OAP mirrors with the same optical specs as the optical simulations discussed in part 3. As shown in Fig. 3 (b), the six outer fibers have about +/-7% spectral flux variations. Spectral flux from the center fiber was about 30% higher than the average of the six outer channels.       

INSERT FIGURES

Fig. 3 (a) shows the relative spectral fluxes for the six outer channels from 380nm to 1000nm. As seen in Fig. 3 (b), the ratios of the six outer channels vs their average are within 95% to 110% in the 380nm to 1000nm wavelength range.

 

Fig. 4 Temporal variation for one channel output in a 280nm band, (a) Wavelength band used for the temporal stability measurement: channel 4 with a 280nm+/-10nm bandpass filter; (b) Temporal variation for normalized flux

 

Irradiance stability is critical for most applications. Fig. 4 (a) shows the wavelength band used for temporal stability of channel 4 with a 280nm center wavelength filter. The spectral peak shift from nominal 280nm to near 290nm is due to the combination of spectral slope from LDLS, grating spectral efficiency, transmission slope from spectrometer detector, and the AOI variation of a divergent beam. The inband flux is logged for 2 hours after fully stabilized for stability analysis. Fig. 4 (b) shows the variation for normalized inband flux over a 2-hours period is about 0.5%.