Active Pixel Sensors for electron microscopy

https://doi.org/10.1016/j.nima.2007.05.308Get rights and content

Abstract

The technology used for monolithic CMOS imagers, popular for cell phone cameras and other photographic applications, has been explored for charged particle tracking by the high-energy physics community for several years. This technology also lends itself to certain imaging detector applications in electron microscopy. We have been developing such detectors for several years at Lawrence Berkeley National Laboratory, and we and others have shown that this technology can offer excellent point-spread function, direct detection and high readout speed. In this paper, we describe some of the design constraints peculiar to electron microscopy and summarize where such detectors could play a useful role.

Introduction

Transmission Electron Microscopy (TEM) presents interesting detector design challenges due to the significant multiple scattering of electrons in the energy region of interest (roughly 80–400 keV) along with large variations of range and energy loss as a function of energy. Furthermore, certain applications and imaging modes can present very large ionizing radiation doses to the detector, so radiation hardness is required.

Different applications also place varied performance demands on the detector. Simultaneous imaging of multiple copies of the same biological structure requires large number of pixels. At the same time, due to the radiation sensitivity of the biological sample, high detection efficiency is desirable. Exposure times are generally long (seconds) but can be shorter in some applications. In materials science, although many in-situ applications would benefit from an ability to read out images at arbitrarily high speeds, the ability to read out at >100 Hz would open up the possibility of visualizing catalysis on the atomic scale. As an example, Fig. 1 shows time-resolved images at atomic scale of gold particles coalescing [1] obtained with a video rate TEM detector. Even at 30 Hz, the readout speed is too slow to be able to deduce the trajectories of individual particles. Further, the spatial resolution of the detector is too low to visualize individual particles. A detector with the spatial resolution of a high-quality CCD-based system and a high readout rate would be needed.

For ultimate resolution, photographic film optimized for electron microscopy is still unsurpassed (for an excellent comparison of conventional TEM detectors, see Ref. [2]). Film for TEM can have sub-micron grain size (although there is a trade-off between grain size and sensitivity) and thus excellent spatial resolution, but besides the obvious speed disadvantage, film is a digital medium: a grain is either converted (e.g. Ag+ sites are converted to Ag0 sites) or not. For practical reasons, film is thus augmented by a detector capable of faster, electronic readout—a thin layer of scintillating phosphor powder coupled (usually by a fiber-optic taper, although in some cases by a mirror and lens optical system) to a Charge-Coupled Device (CCD) readout. As multiple scattering dominates the point-spread function (PSF) in the fiber, a thin fiber is needed (and optical coupling is a way to avoid the PSF degradation due to the fiber-optic coupling).

Direct electron detection in silicon is an obvious way to overcome many of the limitations listed above. Direct detection with CCDs has been attempted [3], but it suffers from radiation damage (with the exception of very low energy electrons in thick, fully depleted CCDs). Paralleling the developments in high-energy physics, several authors have investigated the equivalent of silicon strip [4] or hybrid pixel detectors [5], [6]. In a similar vein, for several years, we have been investigating the use of monolithic CMOS Active Pixel Sensors (APS), and discuss here the motivation and performance.

Section snippets

Analysis

Monolithic APS, first described in 1967 [7], [8], have enjoyed a significant re-birth as cheap digital cameras and rivals to CCDs. This same technology is potentially interesting for electron microscopy, as APS offer excellent PSF, direct detection and high readout speed. An APS is a CMOS integrated circuit, with an imaging region divided into pixels, and a signal processing region. The pixels are “active” as they contain transistors used for buffering the charge collected and steering it.

Conclusions

CMOS APS offer revolutionary detection capabilities for TEM. As the detector can be thin, effects from multiple scattering at higher energies can be reduced. Since readout electronics is integrated on the same chip as the detector, high-speed readout is possible. Deep sub-micron feature size CMOS has demonstrated considerable radiation resistance, which bodes well for high-flux TEM applications. APS for TEM do not yet approach the spatial resolution of film, but could do so as technology

Acknowledgments

The authors would like to acknowledge the skilled technical help of John Turner, Mark West and George Zizka in preparation of apparatus and execution of tests, and C.E. Nelson of NCEM for the use of images in Fig. 1. This work was supported by the Director, Office of Science, of the US Department of Energy under Contract No. DE-AC02-05CH11231.

References (16)

  • M. Haider

    Ultramicroscopy

    (1994)
  • G.Y. Fan

    Ultramicroscopy

    (1998)
  • A.R. Faruqi

    Nucl. Instr. and Meth.

    (2003)
  • Jarron

    Nucl. Phys. B

    (1999)
  • Faruqi

    Nucl. Instr. and Meth. A

    (2005)
  • C.E. Nelson, et al., 2005, Unpublished...
  • J. Zuo

    Microsc. Res. Tech.

    (2000)
  • D.G. Stearns et al.

    Rev. Sci. Instrum.

    (1989)
There are more references available in the full text version of this article.

Cited by (0)

View full text