Cathodoluminescence 2011 A MAS Topical Conference, Co-Sponsored by AMAS

Conference Organizers Colin MacRae

Marion Stevens-Kalceff Scott Wight

Abstract Book Published by Microanalysis Society http://www.microbeamanalysis.org/

ISBN# 978-0-7334-3069-5

Thank You to the Workshop Sponsors!

Time Tuesday ProgramChair Marion Stevens-Kalceff

8:15-8:30 Introduction & Welcome8:30-9:15 Jens Gotze Application of cathodoluminescence microscopy and spectroscopy in geosciences9:15-9:45 David Stowe Contrast mechanisms in cathodoluminescence microscopy 9:45-10:15 Dom Drouin Characterization of semiconductor materials using low voltage Cathodoluminescence and Monte Carlo simulation.10:15-1030 Break

Chair Colin MacRae10:30-11:10 Paul Edwards Development of hyperspectral cathodoluminescence imaging systems for the characterization of semiconducting materials11:10-11:30 Eloise Gaillou Cathodoluminescence applied to natural, plastically deformed diamonds: low temperature and high spatial resolution matter.11:30-11:50 Scott Wight Lispix Tools for Probing CL Spectrum Images11:50-12:10 Paul Carpenter A Combined EPMA and Cathodoluminescence Study of Minerals from Franklin NJ 12:10-1:00pm Lunch1:00-1:40 Poster session Bill Horne, Marion Stevens-Kalceff, Huiliang Zhang, Sarah Brokus, Morten Raanes, Jens Wendler

Chair Scott Wight1:40-2:30 Vendor Presentations FEI (20min), JEOL (15 min), Attolight (15min)2:30-5:00 Lab demos/Software demo Casino demo, Lispix demo, JEOL/xCLent lab, Quanta/Gatan, Helios/Gatan

Wednesday ProgramChair Ed Vicenzi

8:15-9:00 Samuel Sonderegger Picosecond Time-Resolved Cathodoluminescence to Probe Exciton Dynamics in GaN and GaN based hetereostructures9:00-9:30 Marion Stevens-Kalceff Cathodoluminescence Microanalysis of Ultrapure Silicon Dioxide Polymorphs9:30-10:00 Jean-Daniel Ganiere Cathodoluminescence universal extension system + CCD detector to acquire hyperspectral images-a performant tool to investigate the nanworld10:00-10:20 Break

Chair Paul Carpenter10:20-10:50 Nick Wilson Peak fitting of hyperspectral cathodoluminescence data sets10:50-11:10 Paul Edwards High resolution hyperspectral cathodoluminescence imaging of semiconductor and plasmonic nanostructures11:10-11:30 Colin MacRae Luminescent Database – Tools to Determine the Cathodoluminescence Transition11:30-11:50 JoAnn Buscaglia Forensic Science Applications of Cathodoluminescence (CL) Microscopy and Spectroscopy11:50-12:10 Ed Vicenzi Cathodoluminescence Spectral Complexity in Growth-Zoned Jadeite12:10-1:10 Lunch

Chair Scott Wight1:10-2:20 Vendor Presentations Gatan (20min), Thermo (15min), Horiba (15min), Leica (10min)2:30-2:50 Heather Lowers Relationships between quartz trace elements and SEM-cathodoluminescence textures revealed using WDS and LA-ICP-MS mapping techniques2:50-3.05 Daniel Doutt The Role of Native Point Defects in Degenerately Doped ZnO Films3.05-3:20 Jonathan Poplawsky High resolution luminescence microscopy under simultaneous electron beam and laser irradiation3:20-3:35 Chung-Han Lin AlGaN/GaN high electron mobility transistor degradation investigation through nanoscale depth resolved cathodoluminescence spectroscopy3:40-4:00 Brian Rusk Cathodoluminescence textures and trace elements in hdyrothermal quartz4:00-4:20 Break4:20-5:00 Software Demonstrations Paul Edwards CHIMP, Leica prep demo, Nick Wilson Chimage5:00-5:15 Group photo

Thursday ProgramChair Nick Wilson

8:30-9:00 Colin MacRae Collection and Quantification of Hyperspectral Cathodoluminescence9:00-9:20 Danielle Silletti Cathodoluminescent Signatures of Neutron Irradiation9:20-9:35 Zhichun Zhang Thermal process dependence of Li configuration and electrical properties of Li-doped ZnO9:35-10:00 Nestor Zaluzec MSA/MAS Hyper-dimensional Spectral File Format10:00-10:20 Break10:20-10:50 Wolfgang Grunevald Preparation of High Quality Sample Surfaces Using Ion milling10:50-11:20 Natasha Erdman Advanced Sample Preparation for Analytical Microscopy11:20-11:40 David Stowe Revealing large area, artifact free cross sections using argon milling for cathodoluminescence studies11:40-12:00 Poster session Best poster award12:00-1:20 Lunch

Chair Jens Gotze1:20-1:50 Bill Leeman Quantitative Application of Cathodoluminescence (CL) to Natural Quartz with Application to Geothermometry1:50-2:05 Thomas Franklin Scanning Ionoluminescence Microscopy2:05-2:20 E.J. Katz Nanoscale Depth-Resolved Electronic Properties of PECVD SiO2 and SiOx Dielectrics for Device-Tolerant Electronics2:20-2:40 Sergui Maximenko Application of CL/ EBIC-SEM techniques for evaluation of crystallographic defects in thin films and nanostructured materials2:40:3:00 Break

Chair Paul Edwards3:00-3:35 Round Table3:35-3:50 Concluding remarks

Application of cathodoluminescence microscopy and spectroscopy in geosciences J. Götze* * TU Bergakademie Freiberg, Institute of Mineralogy, Brennhausgasse 14, D-09596 Freiberg, Germany Cathodoluminescence (CL) has developed into a powerful method for a lot of investigations in the geosciences during the last decades. Distinct luminescence properties of certain minerals such as elements (diamond), sulphides (sphalerite), oxides (periclase, corundum, cassiterite), halides (fluorite, halite), sulphates (anhydrite, baryte), wolframates (scheelite), phosphates (apatite), carbonates (calcite, aragonite, dolomite, magnesite, smithsonite, witherite, cerrusite) or silicates (feldspar, quartz, zeolites, kaolinite, dickite, pyrophyllite, zircon) allow not only a rapid identification and sometimes quantification of the different mineral constituents. The application of cathodoluminescence microscopy and spectroscopy, often coupled with other analytical methods (e.g. electron spin resonance), enables to reveal information concerning the real structure of minerals. In addition, valence and site occupancy of certain trace elements may be determined by luminescence spectroscopy (Fig. 1). The close relationship between crystal-chemical properties and luminescence characteristics of natural and synthetic minerals is the basis of detailed studies of internal structures, zonal growth and distribution of trace elements within solids (Fig. 2). Accordingly, luminescence is applied to investigate the homogeneity of natural and synthetic crystals. The results can provide information about the conditions during crystal growth, defect density or the incorporation of certain trace elements. These investigations can effectively be accompanied by additional analytical methods with high spatial resolution such as microprobe, PIXE, laser ablation ICP-MS or SHRIMP for micro-chemical analysis, or detailed isotope studies [1]. The CL properties of mineral phases are often highly variable depending on the specific conditions during formation. Certain ions can be encountered in natural and synthetic minerals (e.g., Eu2+, REE3+, Mn2+, Fe3+, Cr3+) in different combinations yielding characteristic sets of ions for various types of rocks or different regions [2]. On the other hand, certain geological processes may create specific defects in mineral phases, which reflect these specific conditions of formation. The knowledge of the typomorphic luminescence characteristics of minerals can be used to reconstruct the processes of mineral formation, alteration and diagenesis or effects such as radiation damage [1], [3]. Several studies during the last decades emphasized that this is also relevant for extraterrestrial materials [4]. In several fields which are closely related to geosciences it could be shown that cathodoluminescence can be an effective addition to conventional analytical methods such as X-ray diffraction, polarizing and electron microscopy, fluid inclusion studies or Raman spectroscopy. Different materials were successfully investigated by luminescence analyses such as synthetic minerals and gemstones, industrial raw materials, products of coal and waste combustion (fly ashes, slags), metallurgical slags and dust, structural materials (natural stones, bricks, mortar), ceramics, glasses, refractory materials and archaeological materials [5]. In addition, CL has been applied to prospecting and exploration purposes or to specific problems of mineral processing and sorting [6].

References [1] M. Pagel et al., Cathodoluminescence in Geosciences. Springer, Berlin Heidelberg New York,

2000. [2] M. Gaft et al., Luminescence Spectroscopy of Minerals and Materials. Springer, Berlin

Heidelberg, 2005. [3] D.K. Richter et al., Mineralogy Petrology 79 (2003) 127. [4] A. Gucsik, Cathodoluminescence and its Application in the Planetary Sciences. Springer,

Berlin Heidelberg, 2009. [5] J. Götze, Cathodoluminescence microscopy and spectroscopy in applied mineralogy. FFH C

485, Freiberg [6] B.S. Gorobets and A.A, Rogojine, Luminescent Spectra of Minerals. RPC VIMS, Moscow.

FIG. 1. Shift of the CL emission band of Mn2+ in different carbonates due to the influence of the crystal field. Aragonite (rhombic) shows greenish CL, whereas calcite (trigonal) has orange CL.

FIG. 2. Quartz crystal from Chemnitz, Germany in crossed polarized light (Pol) and cathodoluminescence (CL). The CL image reveals distinct growth zoning, which is not visible using conventional microscopy.

Contrast mechanisms in cathodoluminescence microscopy D.J. Stowe* and A. Pakzad** * Gatan U.K., 25 Nuffield Way, Abingdon, Oxon., OX14 1RL, U.K. ** Gatan Inc., 5794 W. Las Positas Blvd., Pleasanton, CA, 94588, U.S.A. The experimental technique of cathodoluminescence (CL) microscopy has found favor in many research environments world-wide due to its almost unique ability to provide optical information at high spatial and spectral resolution. From oil and gas exploration to high power lasers and high brightness LEDs CL has played a vital role in extending our understanding of a wide range of materials systems. Many materials exhibit the phenomenon of cathodoluminescence, the emission of a photon (light) when excited by an electron beam, including semiconductors, many minerals and some organic compounds. The emission of a CL photon in these materials is dependent on the manner in which the material returns to the equilibrium state after excitation by the electron beam. The energy of the emitted photons in cathodoluminescence is typically in the near-infra red, through visible to UVA wavelengths (0.3 to 6eV); compared to several keV for other micro-analytical techniques such as EDS or WDS. The low energy of CL emissions is related directly to the low energy electronic transitions taking place within an excited material. The low energy of these transitions mean that the probability of emitting a photon and the wavelength of an emitted photon can be influenced by many factors including: concentration of the optical centre, electronic structure of the material, alloy concentrations, impurity centers (intentional or otherwise), point defects, extended defects, stress, electric fields, temperature and more. This susceptibility of the CL emission (probability and wavelength) to these many factors and the ability to measure the changes in CL emission with high spatial and spectral resolution means that CL can provide a powerful tool in understanding a wide range of properties within a sample. We examine some of the key mechanisms that give rise to contrast in CL micrographs and demonstrate via research highlights from investigators around the world how CL microscopy is being used to extend the understanding of a range of materials including: provenance of oil and gas bearing shale (Fig. 1), the influence of extended defects on InGaN multi-quantum well emission and stress distribution in engineering ceramics. We also examine the phenomenon of photon emission from metal surfaces excited by an electron beam and demonstrate the ability of CL to visualize surface plasmon polariton resonance modes in metal nanoparticles (Fig. 2).

FIG. 1. Single source dominated provenance Barnett Shale (Miss.) Texas, Low Grade Met. Ouchita Orogeny. Grains concentrated by agglutinated forams. Courtesy of Dr J. Schieber, University of Indiana.

FIG. 2. A - Secondary electron image showing silver nanorod formed by FIB milling on silicon substrate. Scale bar is 250nm. B - Three band pass images extracted from the parent spectrum-image revealing three distinct surface plasmon resonance modes with emission at different wavelengths. Image courtesy of Dr A Polman, AMOLF F.O.M., Netherlands

Characterization of semiconductor materials using low voltage Cathodoluminescence and Monte Carlo simulation. D. Drouin CRN2 (Centre de recherche en nanofabrication et nanocaractérisation), Département de génie électrique et génie informatique, Université de Sherbrooke, Sherbrooke, Québec, J1K 2R1, Canada. The development of a wide variety of new materials and devices to be used in nanoelectronic and photonic applications has required powerful tools for the characterization of both bulk and microstructure samples. The Scanning Electron Microscope (SEM) plays a key role in the characterization of a wide variety of these materials. The Cathodoluminescence (CL) technique combined with simulation software will be presented as an investigative tool to characterize gallium nitride (GaN) and indium phosphide (InP) semiconductors. Gallium nitride materials have drawn considerable attention due to their remarkable optical properties and their application in high-energy photon emitters, such as light emitting diodes and laser diodes. Due to lattice mismatches between the growth substrate and the GaN epilayer, many strain-relieving threading dislocations (TDs) appeared and might significantly degrade the optoelectronic device's performance. A detailed investigation of the luminescence charge collection processes around TDs is thus pertinent, in order to determine how TDs act as efficient non-radiative traps. A direct measurements technique of carrier diffusion lengths on self-assembled quantum dots (QDs) by low-voltage cathodoluminescence will also be presented. At 1 kV, we spatially resolve individual QDs which appear as correlated bright spots. Analysis of CL intensity profiles across single QDs provide a direct measurement of the characteristic carrier diffusion length in the InAs WL before capture. Carrier injection, temperature and sample doping dependence will be discussed. Finally, the latest version of the Monte Carlo simulation software CASINO (www.gel.usherb.ca/casino) will be presented. This program has been adapted for CL applications with added absorbed energy distributions and flexibility to model more realistic SEM samples. It is thus possible to simulate a complete image of a complex sample featuring multiple regions of different chemical composition. Such image will show the CL intensity according to the amount of absorbed energy within the selected regions. It is thus possible, with such tool, to investigate the electron beam interaction effect. Another important factor affecting the achievable spatial resolution is the carrier diffusion within the material. The carrier diffusion can be modeled using an exponential decay as follows: 1/r*exp(r/L): where “r” is the distance and “L” is the diffusion length. This work will present the effect of carrier diffusion on CL images of GaN materials. These simulated images will be compared to experimental high spatial resolution images. Acknowledgement The author would like to acknowledge the work of all the contributors to CASINO: H. Demers, N.P. Demers, M. Phillips, N. Pauc, E. Dupuy.

1 keV 5K, GaN:S300 nm

Figure 1: Low temperature (5K) CL micrographs acquired at 1 keV on GaN:Si using near band edge luminescence.

Figure 2 (a) Zoom on a submicrometer bright spot associated to a single QD. (b) CL intensity profiles at 300 K for Ib=3 nA and 0,8nA respectively. The ambipolar diffusion length La is deduced from the csch(R/L) fitting curves (solid line).

Figure 3 Modeling threading dislocation (TD) with cylinders of infinite length (size varying from 10 nm – 50 nm to “model” carriers diffusion near TD) of same composition as the matrix in CASINO.

Figure 4 Simulation of deposited energy line scans across a threading dislocation (TD), located at 0 nm, in gallium nitride with a cylinder radius (modeling the diffusion length near TD) of 10 nm. The incident energy was varied from 1 to 5 keV. The number of electrons N0 was changed to keep the nominal deposited energy constant (E0×N0 = 10,000 keV)

References [1] D. Drouin, A. R. Couture, D. Joly, X. Tastet, V. Aimez and R. Gauvin, Scanning, vol. 29, pp. 92-101, 05. 2007

[2] N. Pauc, M. R. Phillips, V. Aimez, D. Drouin, Appl. Phys. Lett. 89 (2006), 161905.

[3] E. Dupuy, D. Morris, N. Pauc, V. Aimez, M. Gendry, and D. Drouin, Appl. Phys. Lett. 94 (2009) 032113.

[4] H. Demers, N.Poirier-Demers, A.R. Couture, D. Joly, M. Guilmain, N. de Jonge, D. Drouin, (2011). Scanning 33(3), 135-146.

TD

Development of hyperspectral cathodoluminescence imaging systems for the characterization of semiconducting materials

P.R. Edwards, K.J. Lethy, J. Bruckbauer, N. Kumar, F. Sweeney, C. Trager-Cowan, K.P. O’Donnell,

and R.W. Martin

Department of Physics, SUPA, University of Strathclyde, Glasgow G41 3EJ, Scotland, U.K.

While cathodoluminescence (CL) imaging and spectroscopy are well-established methods for

characterizing direct bandgap semiconductors, their combination into hyperspectral imaging [1] has

greatly increased the technique’s usefulness in this field. The added ability to map shifts in emission

wavelength—which can result from microscopic spatial variations in (for example) elemental

composition, elastic strain, carrier concentration, or quantum confinement—can provide information

directly relevant to material properties and device performance. In this work, we describe the

development of hyperspectral CL imaging systems for the analysis of group III nitride

semiconductors. The nitrides underpin a wide range of optoelectronic technologies, including blue

LEDs/lasers and solid state lighting, and much work is now focused on applications involving nano-

scale patterning of such materials, requiring the high spatial resolution which CL provides.

Collecting luminescence for hyperspectral analysis presents additional challenges compared with

conventional CL, and we will discuss optical design considerations in terms of light throughput, field

of view, and compatibility with the collection of other SEM signals [2]. Examples of exploiting this

multimode functionality of SEM include simultaneously acquiring CL and WDX [3], and our

continuing work into combining CL with electron channeling contrast imaging (ECCI) in order to

characterize dislocations [4].

This work will also discuss the data analysis methods employed, particularly those used to extract

CL wavelength shifts. While plotting the wavelength of the most intense pixel in each spectrum is

the simplest way of doing this, the results are often dominated by noise. Plotting the “centroid”, or

centre of mass (e.g. Fig 2), provides a fast alternative, but our preferred method is to use non-linear

least-squares curve fitting to fit one or more peaks to each spectrum (e.g. Fig 3). This allows very

subtle shifts to be observed, as well as the deconvolution of overlapping peaks.

Finally, we will present results of the application of multivariate statistical analysis to our datasets.

Such methods are particularly effective in the separation of fixed-wavelength emission peaks such as

those from rare-earth-doped materials like GaN:Er. Although the wavelength shifts described above

complicate such analysis in more general semiconductor cases, we will present work in which we

have used the technique effectively in the presence of moving peaks.

References

[1] J. Christen, M. Grundmann, D. Bimberg J. Vac. Sci. Technol. B 9 (1991) 2358

[2] P.R. Edwards, R.W. Martin , Semicond. Sci. Technol. 26 (2011) 064005

[3] P.R. Edwards, R.W. Martin , M.R. Lee, Amer. Miner. 92 (2007) 235

[4] C. Trager-Cowan, F. Sweeney et al., Phys. Rev. B 75 (2007) 085301

[5] Thanks to Prof T. Wang and co-workers at the University of Sheffield for providing samples.

This work was supported by the UK Engineering and Physical Sciences Research Council.

FIG. 1. Schematics of Strathclyde hyperspectral CL instruments: (left) high resolution FEGSEM-

based system, and (right) microprobe-based system for simultaneous compositional analysis.

FIG. 2. SE image of a collection of gallium nitride (GaN) nanorods (left); and centroid energy map

(right) calculated from a CL hyperspectral image of a single isolated rod.

0.0 0.2 0.4 0.6 0.8 1.0

distance (µm)

0.0

0.2

0.4

0.6

0.8

dis

tance (µ

m)

0.0 0.2 0.4 0.6 0.8 1.0

distance (µm)

2.79

2.80

2.81

2.82

2.83

centre

(eV

)

0.0 0.2 0.4 0.6 0.8 1.0

distance (µm)

0.0

0.2

0.4

0.6

0.8

dis

tance (µ

m)

0.0 0.2 0.4 0.6 0.8 1.0

distance (µm)

3.400

3.405

3.410

3.415

centre

(eV

)

FIG 3. Results of NLLS curve fitting of two Gaussian peaks to the same dataset as Fig 2, showing

(left) intensity and (right) peak energy. The top row is an InGaN/GaN quantum well emission peak,

and the bottom row is the intrinsic GaN band-edge emission.

1 µm

0 200 400 600 800 1000

distance (nm)

0

200

400

600

800

dis

tance (n

m)

2.95

3.00

3.05

3.10

3.15

centro

id (e

V)

Cathodoluminescence applied to natural, plastically deformed diamonds: low temperature and high spatial resolution matter. E. Gaillou,* J.E. Post,** T. Rose,** and J.E. Butler*** * Department of Terrestrial Magnetism, Carnegie Institution of Washington, 5241 Broad Branch Road N.W., Washington, DC 20015, USA. ** Department of Mineral Sciences, Smithsonian Institution, Washington, DC 20560, USA. *** Chemistry Division, Naval Research Laboratory, Washington DC 20375. Natural pink diamonds are among the rarest and most expensive diamonds, and the origin of their color is still a matter of speculation. For most of them, the pink color is restricted to lamellae oriented along 111 created by plastic deformation and the rest of the diamond is colorless. While cathodoluminescence (CL) on a standard luminoscope imagery system shows either an overwhelming signal from the plastically deformed lamellae (green CL, Fig. 1 left) or no signal at all, a CL spectrometer attached to a scanning electron microscope (SEM) revealed detailed features associated with the plastic deformation (Fig 1 right). Pink diamonds display a blue (or less frequently no) CL emission in the colorless areas and a patchy greenish-yellow emission in the pink lamellae, which are 1 µm or less in thickness (Fig. 1 right). Diamonds can exhibit sharp absorption and emission features at liquid nitrogen temperature, which is sufficiently low for the high Debye temperature of diamond. Each absorption / emission feature is generally related to one single structural defect, and consists of a zero phonon line (ZPL) and several phonon replica (or side bands) at low temperatures, but only a very broad band is observed at room temperature. Hence, room temperature CL experiments do not provide all of the information, and Gaussian or Lorentzian curves cannot appropriately be fitted to explain these CL emissions. CL spectroscopy was conducted on pink diamonds at liquid nitrogen using a cooling stage in an SEM. The colorless areas of the diamonds typically show a band centered at about 415 nm that is broad, even at low temperature. This is characteristic of the "blue band" also called "band A", and originates from dislocations inside the diamond structure [1]. Inside the pink lamellae, the CL signal is made of two structured emissions. The most intense emission has a ZPL at 503 nm, attributed to the H3 center, which is due to two nitrogen atoms around a vacancy. This emission is responsible for the greenish-yellow CL seen by imaging. The weaker emission has a ZPL at 405.5 nm and a structure similar to the N3 center (three nitrogen atoms surrounding a vacancy) at 415 nm but perhaps modified by a nearby unknown center [2]. CL of the peculiar pink diamonds from the Argyle Mine in Australia and Santa Elena in Venezuela will also be presented and discussed. Their unusual CL characteristics distinguish these pink diamonds from diamonds from other localities. These distinct CL signatures might be useful in resolving some issues related to sources of certain conflict diamonds, and also has implication for their mode of formation and aftergrowth processes. From the combined results of CL and other spectroscopies, microscopy and microanalyses methods, we can conclude that plastic deformation can create defects, most likely vacancies, in specific zones (lamellae) inside the diamonds. With time of residence and the high temperature inside the Earth's mantle, the vacancies might have migrated to pre-existing impurities, such as nitrogen, creating the new centers observed by CL, and most likely the still unknown center responsible for the pink color of such diamonds.

References [1] A.M. Zaitsev, Optical Properties of Diamond, Springer Edition, 2001. [2] Gaillou et al., Diam. Relat. Mater., 19, (2010) 1207.

FIG. 1. CL images of a pink diamond obtained with RGB color filters. Left: On a standard luminoscope system, the whole diamond emits a green CL, the pink lamellae of plastic deformation appearing just brighter. Right: On a SEM equipped with a CL system, it is possible to distinguish that the colorless areas of the diamond emit a blue CL, while only the pink lamellae display a green CL.

Lispix Tools For Probing Cathodoluminescence Spectrum Images S.A. Wight,* E.P. Vicenzi,** D.S. Bright,* J.M.. Davis,* and J.J. Donovan*** * National Institute of Standards and Technology, Gaithersburg, MD 20899 ** Museum Conservation Institute, Smithsonian Institution, Suitland, MD 20746 *** Dept. of Earth and Planetary Sciences, University of California, Berkley, CA 94720 The community is increasingly moving toward the collection of spectral image data sets for the collection of cathodoluminescence data. Improvements in technology and specifically to the desktop computing power and the reduction in the cost of storage have resulted in the widespread collection of large data sets by many techniques. Cathodoluminescence and x-ray spectrum mapping are two such techniques that are capable of generating large spectrum image data cubes. Tools are needed to efficiently mine and process these data sets. Lispix is a freely available public domain image analysis program for Windows, written and maintained by David Bright at NIST [1]. Lispix is based upon the Common Lisp language with the Windows specific code written in Allegro so the program is portable. It includes many of the basic image processing functions found in commonly used image processing applications. It also incorporates a collection of special purpose research tools for electron microscopy and spectral imaging. Lispix has proven useful to many researchers for processing and analyzing images, stacks of images and data cubes. Pixels can be bit, integer, real, complex and color (RGB) data types. There are extensive documentation, help files, example images and tutorials on the website. Lispix is widely used by the Microanalysis Group at NIST and a community of researchers worldwide. Array of Circles is a new data cube or spectral image manipulation tool in Lispix. The motivation to develop this tool came from a real world problem. There was a flat polished mineral specimen, Jadeite, which had been analyzed by electron microprobe in one lab and by cathodoluminescence in another lab. We wanted to be able to compare the results and look for correlations between the two types of data. The problem was that the electron microprobe data was collected as a series of spot analysis arranged in the shape of a grid on the surface of the specimen. The cathodoluminescence data was a rectangular spectrum image over nearly the same area except there was a slight rotation of the specimens between analyses so that they did not register with one another nicely. The solution was the Array of Circles tool; it extracts the sum spectrum comprising all pixels within the area under each circle from the data cube. Dave Bright developed this tool for us for maximum flexibility; it acts as a mask that is placed over top of a data cube that has been loaded into Lispix. The number of circles in the x and y directions is user defined as well as the number of pixels in the diameter of each circle. The center green handle allows the user to move the mask around on the face of the data cube for proper alignment and the right hand green handle allows the rectangular array of circles to be rotated in space to accommodate the misalignment of the specimen between instruments. This handle also has the ability to change the spacing between the circles to replicate the spacing used in the previous experiment. Once everything is set up correctly, the Save Spectra button extracts the spectra from under each of the circles, labels by column and row number and exports the spectra as EMSA/MAS formatted text files automatically to the hard drive [2]. It took about 30 seconds on my office desktop.

The extracted CL spectra are deconvolved into their contributing peaks with Optical Fit [3] and then the contributing peaks are plotted as a contour map over the sampled area. These plots can be visually compared with the corresponding intensity plots from the electron microprobe. References [1] http://www.nist.gov/lispix/ [2] http://www.nist.gov/lispix/doc/image-file-formats/MAS-spectrl-std-v1-0.htm [3] http://www.csiro.au/luminescence/opticalfit/index.html

FIG. 1. Electron probe analysis points over false colored panchromatic CL image (left), secondary electron image of CL spectrum image (center), and one slice of the CL spectrum image data cube (right).

1000 3000 5000eV

0

1

2

Inte

nsity

x10

4

OpticalFitPeak 1

Peak 2

Peak 3

spec00-07 Peak 1 Peak 2 Peak 3

Channels: 660 Range: 0...6590eV Marker: 359eV, 6839 (chan. 35) FIG. 2. Spectrum image data cube with overlaid array of circles mask (left) and one of the extracted spectrum fit to three Gaussians (right).

A Combined EPMA and Cathodoluminescence Study of Minerals from Franklin NJ P.K. Carpenter* and E.P. Vicenzi**

*Department of Earth and Planetary Sciences, CB 1169, Washington University, Saint Louis, MO, 63130 **Museum Conservation Institute, Smithsonian Institution, Suitland, MD 20746 Approximately 300 minerals are found at the Franklin and Sterling (FS) mineral deposits of New Jersey, of which 69 are unique to the locality and an impressive 89 minerals exhibit luminescence [1,2]. The de-posit is comprised of two zinc ore bodies formed by metamorphism of limestone and gneiss. Three basic mineral suites are present, banded Zn ore containing zincite, franklinite, willemite, Mn-bearing minerals, and calcite; calc-silicate assemblages which include Mn and Pb-bearing minerals, and diverse late-stage alteration assemblages. Most luminescence has been observed in hand specimens using UV sources and spectacular samples containing up to 5 luminescent minerals are not uncommon. Due to recrystallization, samples have fine and coarse grain sizes, so that characterization of the mineralogy and cathodolumines-cence (CL) requires a combined approach using quantitative electron-probe microanalysis (EPMA), stage mapping, and hyperspectral CL analysis. We report ongoing analysis of Franklin samples and applica-tions to microanalysis standards using these techniques. The origin of CL in Franklin minerals is primarily due to ppm to wt% concentrations of the activator Mn in minerals with low concentrations of the quenching element Fe. Pb acts as a sensitizer to Mn in general. However, it may be that several mechanisms are responsible for CL in general as emission could be pro-duced by elements in very low concentrations. Further, luminescence color has historically been used without supporting EPMA data to identify FS minerals and errors in identification are suspected. EPMA stage mapping using wavelength-dispersive (WDS) and silicon-drift (SDD) x-ray spectrometers coupled with a Gatan MonoCL system were used to map a Franklin ore sample. The map images were used to select areas of interest for hyperspectral CL and quantitative analysis (Fig. 1A and 1B). The se-lected area contains zincite, franklinite, CL-emitting willemite and hardystonite, and other non-CL silicate minerals and arsenian fluorapatite. The recrystallization of willemite is complex and results in textural changes that have been identified using backscattered-electron (BSE) and CL imaging (Fig. 1C and 1D). The willemite exhibits two distinct compositions with Mn that differ by a factor of 2. The hyperspectral CL data have been deconvolved with Gaussian peaks and used to generate maps which are in agreement with Mn compositional variation (Fig. 2 and 1D). Hardystonite exhibits permanent changes to the CL emission after exposure to a 10 m, 25-50 nA probe current during analysis and mapping. Fluorapatite and low Mn (and absent Fe) does not appear to emit CL though luminescence is reported for FS samples [2]. EPMA reference standards require a demonstration of homogeneity and detection of compositional zon-ing. A number of standards exhibit CL variation that is due to compositional variability (e.g., Natural Bridge diopside, Fig. 3A), growth zoning (REE phosphates, Fig. 3B), deformation (corundum, Fig. 3C), and beam damage (Wilberforce apatite, Fig. 3D). CL imaging could therefore be used for screening of potential standards as well as monitoring of beam damage during routine use. The remarkable sensitivity of CL to trace element concentrations awaits correlation with LA-ICPS and multiple-spectrometer WDS analysis. We acknowledge Dr. Pete Dunn for his significant contributions to the mineralogy of the Franklin and Sterling mineral deposits.

References [1] http://franklin-sterlinghill.com/dunn/index.shtml [2] http://sterlinghillminingmuseum.org/aboutus/fluorescentminerals.php

Zincite

Franklinite

Willemite

A

B50 µm

C

D

A B

C D

A B

FIG 1 (top). A. X-ray stage map of Franklin ore sample, Red-Ca, Green-Si, Blue-Zn. Study area outlined in red. Sample 24 x 44 mm. B. CL stage map with willemite and minor hardystonite. C. Backscattered-electron image of study area showing low (avg. 1.6 wt%, brighter BSE) and high (avg. 3.7 wt%, darker BSE) Mn-content of recrystal-lized willemite (150x magnification). D. CL spectrum image from outlined region in Fig. 1C at 564 nm with win-dow 1.7 nm. FIG 2 (lower left). A. Fitted Gaussian curves to CL spectrum. B. CL intensity curves for low Mn (lower curve, darker regions) and high Mn (upper curve, brighter regions) areas in Fig 1D hyperspectral image. FIG 3 (lower right). MonoCL images of EPMA standards. A. Natural Bridge diopside, B. GdPO4, C. Corundum, D. Wilberforce apatite.

A Combined Cathodoluminescence and X-ray Study of Franklin and Sterling Hill Minerals using a Field Emission EPMA M.M. Heines,* W.A. Lamberti,* and W.C. Horn* * ExxonMobil Research and Engineering Company, Corporate Strategic Research, 1545 US 22 East, Annandale, NJ 08801 Minerals from the Franklin and Sterling Hill mining district in northern New Jersey are unique in the world as some 80 different minerals are rich in fluorescent emission, with many containing several fluorescent species in a single specimen [1]. This feature has led many to study these minerals using cathodoluminescence (CL). The recently installed JEOL JXA-8530F Field Emission Electron Probe Microanalyzer (EPMA) at ExxonMobil’s Corporate Strategic Research facility offers novel capabilities for examining these minerals. The combined techniques of Wavelength Dispersive Spectrometry (WDS) x-ray analysis and CL spectroscopy allow for spatially-resolved correlation of specific CL spectral features with trace species. The CL spectrometer system employed (xCLent) is also well-suited to perform fluorescence lifetime studies across a wide range of time scales (limited by the response time of the CL spectrometer with a minimum exposure time of 30 ms). The field emission design of this instrument permits these studies to be performed with higher spatial resolution than what has previously been achieved [2-6]. The high performance of the field emission column also allows for studies to be carried out across a wide range of incident electron energies while maintaining high spatial resolution with high analytical beam currents. Additionally, this instrument utilizes five WDS spectrometers which include two large crystal spectrometers and one high intensity spectrometer for improved detection limits and counting times, respectively. Using a Franklin and Sterling Hills mineral sample rich in Franklinite, Willemite and Calcite (Table 1), CL resolution versus x-ray resolution at various incident electron energies will be discussed. The CL response will be compared to UV-induced fluorescence in the same sample (Figure 1). The CL response for each phase (Figure 2) versus concentration will also be investigated in order to create a calibration curve and estimate quantitative CL detection limits for this system. The authors would like to acknowledge Tom Bruno for providing the mineral samples for this study [7].

[1] P.J. Dunn, Franklin and Sterling Hill, New Jersey: The World’s Most Magnificent Mineral Deposits, Peekskill, New York, 1995.

[2] R.J.R.S.B. Bhalla et al., J. Lumin., 4 (1971) 194. [3] R.J.R.S.B. Bhalla et al., J. Electrochem. Soc., 119 (1972) 740. [4] H.G. Machel, Geosci. Can., 12 (1985) 139. [5] N.G. Hemming et al., J. Sediment. Petrol., 59 (1989) 404. [6] S. Cazenave et al., Min. Pet., 78 (2003) 243. [7] Mineral samples provided by Tom Bruno, ExxonMobil Research and Engineering Company. [8] Chapoulie, R. et al., Scanning Microsc. Suppl., 9 (1995) 225 [9] Gaft, M. et al., Luminescence Spectroscopy of Minerals and Materials, Springer, Berlin, 2005.

TABLE 1. Key Mineral Phases for this Study Mineral Phase Composition Emitter Previously Reported

CL Feature

Willemite Zn2SiO4 Mn2+ 525 nm [3] Calcite CaCO3 Mn2+ 610-620 nm [8], [9] Franklinite (Zn, Mn2+, Fe2+)(Fe3+, Mn3+)2O4 N/A N/A

FIG. 1. Comparison of a short wave UV-induced fluorescence image (a) and a CL image (b) of the area located within the white box of figure 1a.

FIG. 2. Region of interest analysis of cathodoluminescence hyperspectral map indicating major phases and their corresponding CL spectra. Note the unknown phase spectral features in the range of 700-730 nm.

1a 1b

Cathodoluminescence Microcharacterization of BaFCl:Sm Core Shell Nanocrystallites. Marion A. Stevens-Kalceff,* Hans Riesen,** Zhiqiang Liu,** *School of Physics, University of New South Wales, Sydney, 2052 NSW Australia. *Electron Microscope Unit, University of New South Wales, Sydney, 2052 NSW Australia. **School of Physical, Environmental and Mathematical Sciences, University of New South Wales, ADFA, Canberra ACT 2600, Australia The development of novel technologies based on nanostructured materials is facilitated by advanced microscopy and microanalysis techniques which provide spatially resolved high sensitivity information about the structure, composition and properties of materials. BaFCl:Sm3+ is a photoluminescent storage phosphor for ionising radiation [1,2]. The mechanism is based on the reduction of Sm3+ to Sm2+ by F-centres and/or free electrons that are created upon exposure to ionising radiation in the BaFCl host. The efficiency of this class of storage phosphors has been significantly improved by optimizing the preparation of core-shell nanoparticles comprising of a BaFCl core and a BaFCl:Sm3+ shell using a co-precipitation method [3]. High resolution powder X-ray diffraction techniques indicate average crystallite size of ~160 nm. The Sm3+Sm2+ conversion efficiency of these nanoparticles is more than four orders of magnitude greater than that of conventional high temperature (HT) sintered micro-crystals. Cathodoluminescence (CL) microanalysis is ideal for investigating the useful properties of these efficient nano-phosphors. Fig. 1 shows typical CL spectra from high temperature microcrystalline (HT), nanocrystalline and X-ray irradiated nanocrystalline core shell (cs) BaFCl:Sm3+. In Fig. 1(a) all spectra exhibit a broad emission centered at ~380nm which is associated with defects in the BaFCl host lattice. Fine structure in the CL spectra is associated with Sm. Major 4GJ

6HJ (Sm3+) and 5DJ7FJ (Sm2+) transitions are identified. There are

significant differences in the 4GJ6HJ (Sm3+) emission lines between the two materials,

indicating significantly different local environments for the Sm3+ ions. After ionizing radiation, there is a net increase in Sm2+ associated CL emission. After reduction to Sm2+ by ionizing radiation the resulting 5DJ

7FJ (Sm2+) emission lines coincide for both materials. A series of CL spectra excited by a stationary focused beam have been collected at 100nm intervals across individual particles. In Fig. 2, the CL from the host lattice BaFCl is ~uniform across the particle. In comparison the CL emissions associated with Sm2+/3+ are enhanced at the edges of the particle. These data indicate that the Sm is concentrated at the edges of this typical ~uniformly thick platelet-like particle prepared using the co-precipitation method. In contrast, CL intensity profiles from conventional high temperature (HT) sintered micro-crystals are approximately the same within experimental uncertainty for both BaFCl and Sm2+/3+

associated emissions (not shown here), indicating that Sm is likely to be distributed homogeneously through the micro-crystals. CL provides information on the synthesis dependent distribution of Sm in BaFCl:Sm particles. Sm is concentrated near the surface of the nanocrystalline particles but is uniformly distributed within the microcrystalline particles.

References [1] H. Riesen, W.A. Kaczmarek, Inorg. Chem. 46 7235-7237. (2007) [2] H. Riesen, W.A. Kaczmarek, Radiation storage phosphor & applications, International PCT

Application, WO2006063409-A1, 16 Dec. 2005. [3] H. Riesen, T. Massil, Z. Liu, Core-shell nanophosphors for radiation storage and methods,

International PCT Application, WO2011054050A1, 5 Nov, 2010. [3] This research was supported by the Australian Research Council LP110100451. The

Australian Microscopy and Microanalysis Research Facility at UNSW is gratefully acknowledged. Author contact: Marion.Stevens-Kalceff@unsw.edu.au

(a) (b)

Fig. 1. (a) Linear and log plots of typical CL spectra (10 keV, 10 nA), from undoped BaFCl, high temp (HT) BaFCl:Sm, cs BaFCl:Sm, and X-radiated cs BaFCl:Sm (50mGy). (b) CL spectra obtained from at 15 keV, 3 nA, from a HT sintered BaFCl:Sm3+, cs nanocrystalline BaFCl:Sm3+. Major 4GJ

6HJ (Sm3+) and 5DJ7FJ (Sm2+) transitions are shown.

(a) (b)

Fig. 2. (a) CL spectra obtained at 10 keV, 10 nA, from nanocrystalline cs BaFCl:Sm3+ Spectra are collected at 100nm intervals along the arrow as indicated,. (b) Normalized intensities of CL peaks in Fig. 2 (a) plotted as a function of position across the cs BaFCl:Sm particle. The 595, 640 and 688nm emissions are associated with Sm. The broad emissions <400nm are associate with the BaFCl host lattice. dimension marker = 1 µm

Bring Color to Electron Microscopy Using Nanoparticle Cathodoluminescence D.R. Glenn,* H. L. Zhang,** N. Kasthuri,*** R. Schalek, *** A.S. Trifonov, ** J.W. Lichtman, *** and R.L. Walsworth, *,** * Harvard-Smithsonian Center for Astrophysics, Cambridge, MA 02138 ** Department of Physics, Harvard University, Cambridge, MA 02138 *** Department of Molecular and Cellular Biology and Center for Brain Science, Harvard University, Cambridge, MA 02138 We demonstrate a multi-color nanoscale imaging technique based on cathodoluminescence (CL) emitted by color centers in nanodiamonds (NDs) [1] and other nanometer-sized crystals (e.g., nanophosphors[2]) under excitation by an electron beam in a scanning electron microscope (SEM). We have identified three classes of light-emitting nanoparticles that are spectrally distinct at room temperature and can be obtained with diameters on the order of 10 - 100 nm. Compared to CL markers based on organic molecules [3], these color centers are bright and highly stable under SEM excitation. In conjunction with appropriate functionalization of the nanoparticle surfaces, this approach will provide nanoscale information about molecular function to augment the structural information obtained with standard SEM techniques. References [1] Zaitsev, A.M., Optical properties of diamond : a data handbook. 2001, Berlin ; New York:

Springer. [2] Lim, S.F., et al., In Vivo and Scanning Electron Microscopy Imaging of Upconverting

Nanophosphors in Caenorhabditis elegans. Nano Letters, 2005. 6(2): p. 169-174. [3] Niitsuma, J.-i., et al., Cathodoluminescence investigation of organic materials. Journal of Electron Microscopy, 2005. 54(4): p. 325-330.

Cathodoluminescence Study of Feldspar Minerals S.A. Brokus,* J.D. Borycz,* J.M. Lunderberg,* G.F. Peaslee,* P.A. DeYoung,** D.E. Cooper,*** J. Buscaglia,**** and D.K. Silletti***** * Hope College, Chemistry Department, Holland, MI 49423 ** Hope College, Physics Department, Holland, MI 49423 *** FBI Laboratory, Latent Print Operations Unit, Quantico, VA 22135 **** FBI Laboratory, Counterterrorism and Forensic Science Research Unit, Quantico, VA 22135 ***** ORISE Visiting Scientist Program, FBI Laboratory, Quantico, VA 22135 A collection of 42 common feldspar mineral samples have been examined by cathodoluminescence (CL) spectroscopy and ion beam induced luminescence (IBIL) spectroscopy. Previously reported luminescent centers (Mn2+ and Fe3+) were observed and their UV-Visible spectral peak positions vary with stoichiometric changes in the Na-K-Ca composition of the feldspars, as expected. Similarly, Si-O and Al-O lattice defect luminescence in the UV-Visible spectrum were observed, as well as an unassigned IR luminescence. Additional analyses of the feldspar samples by x-ray diffraction (XRD), electron microprobe (EMP), micro x-ray fluorescence (µXRF), and particle induced x-ray emission (PIXE) were performed in an attempt to determine the mechanism for this unassigned IR peak, as well as the observed peak centroid shifts within luminescent peak signatures [1]. References

[1] This project was funded by National Science Foundation (RUI PHY-0651627; MRI 0319523), the Department of Homeland Security (2008-DN-077-ER0008), the FBI Laboratory’s Visiting Scientist Program (an educational opportunity administered by ORISE). and Hope College Chemistry and Physics Departments. Aid from Hope College Geology and Environmental Sciences Department and the Smithsonian Institution of Natural Sciences are gratefully acknowledged.

FIG. 1. The consistency of CL and IBIL peak positions was demonstrated with replicate spectra of the same zircon mineral sample. When CL spectra from members of the same feldspar are compared among different geographical locations, there are significant variations that result due to the difference in mineral composition, lattice size, and structure. FIG. 2. North America including contiguous United States, portions of Canada, and Mexico. Markers locate general site of samples (credit: Google Maps).

Characterization of Arendal quartz by combining EPMA Cathodoluminescence and SEM Electron Back-Scattering Diffraction Analyses B.E. Sørensen,* M.P. Raanes** and Y.D. Yu ** * Department of Geology, Norwegian University of Science and Technology, Sem Selands vei 1, NO-7491 Trondheim, Norway ** Department of Materials Technology, Norwegian University of Science and Technology, Alfred Getz vei 2, NO-7491 Trondheim, Norway Sørensen and Larsen [1] studied quartz deformation of quartz rich rock that had undergone partial recrystallisation at lower temperatures. Four quartz types, Qz1, Qz2, Qz3 and Qz4, are defined by gray-scale SEM -CL. They typically co-exist in a single sample, commonly in association with mm-scale deformation zones. Two-stage-infiltration-driven recrystallization explains the formation of Qz1, Qz2 and Qz3 [1]. The first infiltration stage happened when aqueous fluids infiltrated a dry quartzite at static conditions. This caused brecciation, dissolution and re-precipitation. Old brecciated grains were represented by Qz1, whereas newly precipitated quartz formed Qz2. A second fluid infiltration stage caused renewed recrystallization and formation of quartz type Qz3. This second stage of fluid infiltration is associated with strain softening and mm- to cm scale variations in deformation textures. Qz3 behaved plastically with subgrain rotation recrystallization (SGR), whereas Qz1 and Qz2 behaved more brittly during deformation. In this study, the further investigations are carried out by combining hyperspectral EPMA CL and SEM EBSD analyses [2] to unravel the relations between deformation microstructures and luminescence in the samples. An interesting observation was that an increase in luminescence of Qz1 inside relative to outside the zones of most intense deformation as shown in FIGs. 1 and 2. Electron probe micro analysis (EPMA) was carried out using a JEOL JXA-8500F combining cathodoluminescence and x-ray microanalysis (WDS) by use of Ocean Optics CL detector system and xCLent software. Electron back scatter diffraction (EBSD) measurement was performed in a Zeiss Supra 55VP FEG-SEM fitted with a Nordif EBSD detector and the TSL OIM EBSD software. References [1] B.E. Sørensen et al., Contrib. Mineral Petrol. 157 (2009) 147. [2] Y.D. Yu et al., EMC2008, DOI: 10.1007/978-3-540-85156-1, (2008) 513.

FIG. 1. EPMA-CL false colour image showing the quartz types described in text. Note how luminescence of Qz increase toward the lower part of image, where deformation is high.

FIG. 2(a) Al foil, double side carbon and Cu conducting tapes were used for increasing conductivity of thin quartz section on glass substrate during EBSD scan. (b) EBSD Inverse Pole Figure (IPF) from EBSD scan of the marked area in FIG. 1, also as a subarea inside the blue square of FIG. 2(a).

Using Cathodoluminescence Spectroscopy to distinguish biogenic from early-diagenetic Calcite in Cretaceous Microfossils from the Tanzanian Drilling Project

J.E. Wendler*, I. Wendler*, T. Rose**, B.T. Huber*

* Department of Paleobiology, Smithsonian Institution, PO Box 37012, MRC 121, Washington, DC 20560

** Department of Mineral Sciences, Smithsonian Institution, PO Box 37012, MRC 121, Washington, DC 20560

Microfossils (dinoflagellates, foraminifera) showing rare, spectacular preservation were recovered from sediments of the Cretaceous (92 Million years ago) by the Tanzanian Drilling Project (TDP). Cathodoluminescence (CL)-spectral analysis of this material was motivated by the fact that the pristine dinoflagellate calcite is comparatively Mg-rich; however, it undergoes significant early re-crystallization that likely was associated with a decrease in Mg content. We aimed at detecting this process by searching for a shift in wavelength of the ca. 630 nm CL-band that usually is related to changes in Mg content. We found that the material shows a different CL spectral property related to “intrinsic” luminescence bands making it impossible to detect a wavelength shift in the ca. 630 nm luminescence band. The unusually strong “intrinsic” CL-spectral pattern common to the Tanzanian material is characterized by a dominant peak around 395 nm. Previous studies [1] of pure calcite crystals (sinter calcite) reveal that this peak is associated with luminescence effects different from impurities or defects but rather due to lattice properties that cause an electron transition of Ca+ - to CO3

- - centers [2]. The best preserved (pristine) specimens of our study reveal delicate surface traits in SEM-imaging and have a glassy light-optical appearance. They represent the calcite that is closest to the original biogenic shell substance. These specimens exclusively show the 395 nm luminescence peak (Fig. 1) while towards re-crystallization and, finally, cementation the common 625 nm band develops (Fig. 2). Our findings have taxonomic implications as the surface crystal pattern that was thought to be an unaltered biogenic pattern actually represents a re-crystallized shell (Fig. 2) and therefore cannot be accorded the same taxonomic significance. Apparently, the ratio of “intrinsic”- versus Mn-activated luminescence can be taken as a potential measure of the degree of earliest-diagenetic re-crystallization of biogenically formed calcite. This is important to evaluate the reliability of geochemical proxies for paleo-environmental reconstruction drawn from the study of microfossils. Besides the dinoflagellate-calcite, pristine calcitic and aragonitic foraminifera also show a well-defined ca. 400 nm peak (Fig. 3). References [1] Habermann et al., Zbl. Geol. Palaeont. Teil 1. 10-12 (1999) 1275-1284. [2] Calderón et al., J. Phys. 17 (1984) 2027-2038. [3] This research was funded by the German Science Foundation (DFG) fund WE 4587/1-1. J.

Götze is acknowledged for inspiring comments.

FIG. 1. Calcareous dinoflagellate (SEM; scale: 20 µm) at best preservation. CL is dominated by the 395 nm band with generally weak relative intensity. CL-spectra acquired using an analytical FEI Nova NanoSEM 600 with a Gatan MonoCL spectrometer operating with a high sensitivity photomultiplier tube (HSPMT) detector and a Xiclone liquid nitrogen-cooled CCD array.

FIG. 2. Re-crystallized calcareous dinoflagellate (SEM; scale: 20 µm) with beginning late-diagenetic cementation. CL is composed of both the 395 nm (“intrinsic”) peak and the 625 nm (Mn-activated) peak.

FIG. 3. Calcitic foraminifer (upper panel) and aragonitic foraminifer (scales: 100 µm, left image: glassy appearance in reflected light; right image: polished specimens in CL; diagram: CL-spectra). Gaussian fitting of the aragonite spectrum reveals a dominant ~430 nm peak and a ~570 nm residual.

Picosecond Time-Resolved Cathodoluminescence to Probe Exciton Dynamics in GaN and GaN based hetereostructures P. Corfdir*, P. Lefebvre**, A. Dussaigne*, L. Balet*, S. Sonderegger***, T. Zhu*, D. Martin*, J.-D. Ganière*, N. Grandjean*, and B. Deveaud-Plédran* * Institute of Condensed Matter Physics, Ecole Polytechnique Fédérale de Lausanne (EPFL), 1015 Lausanne, Switzerland ** Instituto de Sistemas Optoelectrónicos y Microtecnología, ETSI Telecomunicación, Universidad Politécnica, 28040 Madrid, Spain *** Attolight AG, EPFL Innovation Square, PSE D, 1015 Lausanne, Switzerland Introduction Picosecond and femtosecond spectroscopy allow for a detailed study of carrier dynamics in nanosctructured materials [1]. In such experiments, a laser pulse usually excites several nanostructures at once. However, spectroscopic information may also be acquired using pulses from an electron beam in a modern scanning electron microscope (SEM), exploiting cathodoluminescence (CL) where electrons are promoted from the conduction band to the valence band upon impingement of the high energy electron beam onto a semiconductor. This approach offers several advantages over the usual optical spectroscopy. The multimode imaging capabilities of the SEM enable the correlation of optical properties (via CL) with surface morphology (secondary electron mode) at the nanometer scale [2] and the large energy of the electrons allows the excitation of wide-bandgap materials. Here, we present results obtained with an original time-resolved cathodoluminescence (TRCL) setup [3]. This setup uses ultrafast UV laser pulses to create short photoelectron pulses. The light pulses from an ultrafast UV laser illuminate a metal photocathode from which the electrons are extracted and accelerated inside the high voltage column of the microscope and focused on the sample surface. The collected CL signal is dispersed in a spectrometer and analyzed with an ultrafast STREAK camera to obtain high time resolution. Our current setup reaches combined space and time resolutions of 50 nm and 10 ps, respectively. Measurements can be carried out at temperatures between 25 K and 300 K. We will describe the TRCL setup in detail and will also present results obtained on a-plane GaN and a-plane (Al,Ga)N/GaN quantum wells (QW) [4]. Results We first study an epitaxial lateral overgrown (ELO) a-plane GaN grown by hydride vapor phase epitaxy on r-plane sapphire that has been studied at 27 K. Large densities of basal stacking faults (BSFs) are usually observed in a-plane GaN. These extended defects can be seen as a type-II QWs and give rise to a broad and intense emission at 3.42 eV (50 meV below the emission energy of the D°X of wurtzite GaN[6]). We evidence that exciton localization and recombination processes are strongly dependent on the local BSF density. In low-BSF-density zones, we show that the diffusion of free excitons towards BSFs is donor assisted. On the other hand, zones with BSF bundles present a totally inhibited D°X emission. The change in BSF-bound exciton luminescence decay time is explained through direct relation to the local BSF density.

As a next step, we proceeded to grow a (Al,Ga)N/GaN single QW by molecular beam epitaxy on the GaN sample studied above. Interest has been very high in a-plane GaN since Waltereit et al. demonstrated ten years ago the realization of polarization free (Al,Ga)N/GaN quantum wells (QWs) [5]. Built-in electric fields are indeed absent in a-plane GaN (non-polar GaN), which allows for the growth of wide QWs without decreasing the radiative recombination probability of electrons and holes. However, even when processing techniques such as epitaxial lateral overgrowth (ELO) are used, non-polar GaN grown on sapphire presents high densities of extended defects. While dislocations are considered as non-radiative recombination centers, basal plane stacking faults (BSFs) are optically active and give rise to an emission centered 50 meV below the excitonic bandgap of GaN [7]. We present a low-temperature TRCL study of exciton dynamics as a function of the local BSF density in a-plane (Al,Ga)N/GaN single QWs grown by molecular beam epitaxy on an ELO-GaN template. First, CL experiments demonstrate the existence of nearly BSF-free regions as well as the existence of regions with BSF bundles. This indicates that the BSF distribution of the underlying a-plane GaN template [8] is reproduced in the QW. We confirm the results obtained by Badcock et al., who demonstrated that the intersection of BSFs with the QW leads to the formation of quantum wires (QWR) [9]. We then study the local relaxation-recombination dynamics of excitons in both QW and QWRs. In particular, we show that the dynamics of QW excitons is dominated by their capture by the BSFs. The QW CL decay time therefore exhibits a strong spatial dependency, explaining the large range of values reported so far for exciton radiative lifetimes in non-polar (Al,Ga)N/GaN QWs [4]. We finally demonstrate that below 60 K, QWR excitons exhibit a zero-dimensional behavior, which we relate to their binding on localization centers such as QWR-width fluctuations. References [1] Shah, J. Ultrafast Spectroscopy of Semiconductors and Semiconductor Nanostructures, Ch. 8

(Springer, Berlin, 1999). [2] Reimer, L. Scanning Electron Microscopy, Ch. 1 (Springer, Berlin, 1998). [3] M. Merano et al., Nature 438, 479 (2005). [4] P. Corfdir et al., J. Appl. Phys. 107, 043524 (2010). [5] P. Waltereit et al., Nature 406, 865 (2000). [6] P. Corfdir et al., J. Appl. Phys. 105, 043102 (2009). [7] G. Salviati et al., Phys. Stat. Sol. (a) 171, 325 (1999). [8] P. Corfdir et al., Appl. Phys. Lett. 94, 201115 (2009). [9] T. J. Badcock et al., Appl. Phys. Lett. 93, 101901 (2008).

Cathodoluminescence Microanalysis of Ultrapure Silicon Dioxide Polymorphs. Marion A. Stevens-Kalceff School of Physics, University of New South Wales, Sydney NSW 2052 Australia Electron Microscope Unit, Mark Wainwright Analytical Centre, University of New South Wales, Sydney NSW 2052 Australia Ultrapure silicon dioxide (SiO2) is an key material in electronics, photonics, optics and piezoelectric technologies. Characterization of the defect microstructure of silicon dioxide allows the performance of technologically important applications to be optimized. A range of synthetic pure SiO2 polymorphs typically used in device applications including bulk single-crystal silicon dioxide (quartz), bulk Type I-IV amorphous silicon dioxide (a-SiO2), dry thermal a-SiO2 thin films on silicon (001) and buried a-SiO2 layer in silicon (001) have been systematically investigated under identical excitation conditions using cathodoluminescence (CL) spectroscopy techniques.CL from SiO2 is characterised by broad overlapping emissions associated with localised intrinsic defects in the tetrahedral SiO2 structure. Knowledge of the ultrapure polymorphs underpins the understanding of the more complex impurity defect structures of natural SiO2 polymorphs which are of critical interest in the geosciences. CL emissions from pure SiO2 polymorphs are generally related to local point defects in the tetrahedrally coordinated SiO2 host lattice. A single broad multi-component CL emission (FWHM 0.4 eV) at approximately 1.9 eV (650 nm) is observed from quartz (α-SiO2) under low dose excitation and is attributed to the Non-Bridging Oxygen Hole Centre (NBOHC) with intrinsic and hydrogen impurity-related precursors. CL emissions from bulk Type I-IV a-SiO2 and 50nm, 300nm and 900nm dry thermal thin films of a-SiO2 are identified with a range of native defect centers associated with point defects in the a-SiO2 tetrahedral structure including oxygen deficient defects, (e.g. oxygen vacancies), non bridging oxygen defects and self trapped excitons [1] or trace levels of residual impurities incorporated during synthesis. The relative intensities, peak widths and/ or irradiation kinetics differ between each bulk and thin film a-SiO2 polymorph. The CL emissions from bulk and thin film a-SiO2 polymorphs include the NBOHC at 1.9 eV (650 nm); the radiative recombination of the self trapped exciton (STE) at 2.2 eV (565 nm); and Oxygen Deficient Centers (ODC) at 2.7 eV (460 nm) and 4.5 eV (275 nm). CL emission at ~3.4 eV (365 nm) is observed from Type I and II a- SiO2 and is attributed to the charged compensated substitutional Al3+:M+ defect (where M+ is typically Li+, Na+, K+ or H+). The defect structure of buried oxides in silicon fabricated using ion implantation techniques is complex due to the presence of residual strain, interface states, implantation induced silicon nano-clusters formed during fabrication and native structural defects within the buried a-SiO2 layer. Separation by IMplantation of OXygen (SIMOX) fabrication processes produce a-SiO2 confined within the silicon wafer with residual compressive strain maximized at the two Si-SiO2 planar interfaces. The concentration of defects is enhanced at surfaces and interfaces. The formation of the SiO2 native defects which involve the relaxation of the surrounding oxide network are hindered in the confined strained buried oxide, while the formation of silicon nano-clusters and crystalline coesite platelets are facilitated. The CL emission from buried a-SiO2 is significantly different from bulk and thin film a-SiO2 and cannot simply be associated with analogous a-SiO2 native point defects[2]. CL emission from the confined strained buried oxide is dominated by defects associated with large surface to volume ratio nanoscale silicon clusters and their interfaces. Following

corrections for optical absorption by the Si top layer, CL microanalysis allows the defect microstructure of the buried a-SiO2 layer to be investigated in situ[2]. A minimum of six visible CL emission components are observed: The 1.65 eV (760 nm) emission is associated with passivated silicon nano-clusters and SiO2-Si particle surface states. Similarly the 2.15 eV (590 nm) and 2.35 eV (535 nm) emissions are associated with surface states of silicon nano-clusters near the Si - SiO2

interfaces. The 2.7 eV (460 nm) emission is associated with the oxidized silicon nano-clusters. The ~ 3 eV (415 nm) emission is associated with a silicon excess defect (e.g. H complexed oxygen vacancy) and/ or coesite crystalline platelets near the a-SiO2 - Si substrate interface[2]. The most significant physical processes contributing to the changes in the CL spectra from the SiO2

polymorphs are defect generation via radiolysis and local modification due to highly localized electric fields produced by charge trapping at defects within the SiO2. Electron irradiation induced dose dependent changes in the CL emission intensities are consistent with defect generation and transformation, resulting in oxygen vacancy generation in the bulk and thin film a-SiO2 polymorphs. Electron irradiation can produce localized trapped charge induced electric fields within the irradiated micro-volume of the SiO2 resulting in electro-migration of charged mobile defect species and local modification of the defect structure and/ or chemical composition. Charge trapping is enhanced within the strained confined buried a-SiO2 layer at interfaces and nanoscale silicon cluster defects. These enhanced electric fields can ultimately contribute to volume loss and breakdown of the irradiated confined buried oxide. Non-destructive depth resolved information about the defect distribution in the a-SiO2 thin films and buried a-SiO2 layer can also be determined by varying the electron beam energy and hence the depth at which the CL emission is generated. Analysis of the variation of the integrated emission intensity of a CL peak (ICL ) as a function of primary electron beam energy (E0) then allows the depth distribution of the associated defect center (n(z)) to be determined[3]; where dE/dz is the Energy Transfer Function and R is the maximum electron range, which can be deduced from Monte Carlo simulations. The distribution of defects is strongly influenced in the vicinity of the surface and Si-SiO2 planar interfaces; in particular the concentration of oxygen deficient defects is enhanced. CL techniques provide nondestructive, sensitive, high resolution methods of assessing the relative concentration and distribution of defects in SiO2 polymorphs. References [1] Skuja, L. Optical Properties of Defects in Silica. In: Pacchioni, G., Skuja, L. & Griscom, D. L.

(eds.) Defects in SiO2 and Related Dielectrics: Science and Technology. Kluwer, Dordrecht, Netherlands (2000).

[2] Stevens-Kalceff, M., A. Cathodoluminescence microcharacterization of the radiation-sensitive defect microstructure of in situ buried oxide in silicon. J. Physics D, 44, 255402 (2011)

[3] Goldberg, M., Barfels, T., Fitting, H. J. 1998. Cathodoluminescence depth analysis in SiO2-Si-systems. Fresenius Journal of Analytical Chemistry, 361, 560-561 (1998).

[4] Support from the Australian Research Council and the Australian Microscopy and Microanalysis Research Facility at UNSW is gratefully acknowledged.

Author contact: Marion.Stevens-Kalceff@unsw.edu.au

dzznEdz

dEI

R

CL )(0

0

Cathodoluminescence universal extension (CLUE / Horiba Jobin-Yvon) system + CCD array detector to acquire hyperspectral images: a performant tool to investigate the nanoworld J.D. Ganière* and D. Hocrelle ** * Institute of Condensed Matter Physics, Ecole Polytechnique Fédérale de Lausanne (EPFL) 1015 Lausanne, Switzerland ** Optical Spectroscopy Division, Horiba Jobin-Yvon, 91830 Chilly-Mazarin, France Cathodoluminescence (CL) has, in comparison to PL, the advantage of a much higher lateral and depth resolution due to the small spot size of the exciting electron beam. This makes CL an unique tool for luminescence mapping and for selected investigation of nanometer-sized structures such as pyramids or nanowires. The multimode imaging capabilities of the electron microscope enable the correlation of optical properties (via cathodoluminescence) with surface morphology (secondary electron mode) at the nanometre scale. We gain access to spectroscopic information within a single nano-object and correlation with the surface morphology is straightforward. Photomultipliers (PM) and/or InGaAs flux detectors are largely used to acquire the CL signal. PM are clearly the best choice to obtain CL images in real time (e.g. at TV speed) or in case where the dynamic is critical. Working in photon counting mode allows the users to increase arbitrarily the dynamics by increasing the acquisition time. Increased performances of array detectors (CCD or InGaAs) open new possibilities to acquire hyperspectral images. Using a CCD as a detector, at each point of the image, the full spectral information is available. Intensity coded images in black and white / color or even RGB images can be built using advanced image treatments ( e.g. integration over one or several spectral windows). These new options allow to highlight the contrasts which are difficult to observe by conventional methods, while retaining the ability to obtain mono / poly-chromatic maps. Different examples from the nanoworld are presented in the visible or infrared region of the spectrum to illustrate the potential of hyperspectral CL. The first example (Fig. 1) is a CL mapping of GaN microcolumns cover with InGaN quantum wells. RGB images are used to highlight non uniformity of thickness and/or changes in chemical composition of the QW. The second example (Fig. 2) shows how differential contast technique (the signal is integrated over two adjacent spectral windows and the difference is color coded to build a new image) can be used to highlight optical mode in ZnO nanowires. In the last example, taken in the infrared part of the spectrum, we investigate, at low temperature (10 K) InGaAs/AlGaAs tetrahedral pyramids[1], with a quantum dot located at the apex. Without any doubt the use of new and powerful array detectors allows to highlight contrasts that are difficult to observe with flux detectors and it makes easier the interpretation of CL images. References [1] Q. Zhu, J. D. Ganière, Z. B. He, K. F. Karlsson, M. Byszewski, E. Pelucchi, A. Rudra, and E. Kapon, Phys. Rev

B 82 (2010), 165315

Fig. 1. Microcolumns of GaN covered with an InGaN quantum well on the top. RGB image

highlights the local change of thickness and/or in chemical composition of the quantum well.

Fig. 2. ZnO nanorod, a) SE images - the insert shows the three spectral windows used to generate

images presented in b), c) and d). The CL signal is integrated over the full spectral band and the intensity is color coded (b), the signal is integrated over the red, green and blue channels to generate a RGB image (c), the signal is integrated over the blue and the green channel, the difference is color coded (d).

Peak fitting of hyperspectral cathodoluminescence data sets Nick C. Wilson*, Colin M. MacRae*, Aaron Torpy*, and Ed P. Vicenzi** nick.wilson@csiro.au *CSIRO Process Science and Engineering, Clayton South, VIC 3169, Australia **Department of Mineral Sciences, Smithsonian Institution, Washington, DC 20560 Hyperspectral cathodoluminescence (CL) data collection offers several advantages over flood gun, spectral region of interest (ROI), or serial spectral data collection. Firstly, a key reason to collect the whole spectrum in as short a period as possible is to avoid artefacts in the CL spectrum due to beam damage. Secondly, hyperspectral data collection gives the ability for post-hoc examination of the data to find unexpected phases (x-ray data) and defects or centres (CL data) within a sample[1,2]. However, the great advantage of spectral collection of CL is the ability to perform peak fitting on the data, a necessary step in performing quantification of trace element concentrations or defect concentrations using CL. It is possible in some systems to correlate peak height with concentration for a range of elements, including the transition metals and rare-earths, however, this can be challenging as the peaks for the elements are often overlapped [3]. This means that the count rate change in a simple ROI window around the peak energy of interest can be dominated by another peak in the spectrum, and hence the need for peak fitting to deconvolute the contribution of each peak/emitter? to the signal. For example in Fig. 1, the Dy3+ peak at 2.16 eV sits on the shoulder of the much larger WO −2

4 intrinsic peak, and thus can only be quantified by peak fitting. To deconvolute the spectrum, a set of Gaussian and/or Lorentizan functions are chosen and a non-linear least square minimization is performed. This is best done with some knowledge of the underlying physics and elemental transitions, to avoid the problem of over-fitting the spectrum by adding too many peaks. A good place to start looking for such information is a cathodoluminescence database[4]. However, mineral samples can quite often be complex, with different regions in a map having quite distinct CL spectra, such as in Fig. 2. The challenge is then how to define a set of peaks for peak fitting. Attempting to create a ‘super-set’ of Gaussians to fit the whole map inevitably leads to poor convergence in fitting. To tackle this problem, we use clustering algorithms as a means to explore these large datasets, and partition the data set into different spectral shapes, for which a tailored set of peaks can be generated and then fitted to the appropriate regions of the map.

1.5 2.0 2.5 3.0 3.5eV

0

50

100

150

200

250

300

350

400

450In

tens

ity

Dy3+Dy3+

sm3+

Eu3+Dy3+Nd3+ sm3+ Er3+Er3+ Eu2+Eu3+

Fig 1. A cathodoluminescence spectrum acquired from a scheelite (CaWO4) sample from the Golden Mile, Kalgoorlie, showing the main WO −2

4 peak and a set of peaks attributable to rare earth lines.

Fig 2. Cathodoluminescence spectra obtain from different regions within a carbonado diamond. References: [1] N. C. Wilson et al, Microscopy and Microanalysis 14 (Suppl. 2) (2008) 764 [2] N. C. Wilson et al, Microscopy and Microanalysis 16 (Suppl. 2) (2010) 266 [3] MacRae, C.M., et al., Microsc. Microanal. 15, 222–230, 2009 [4] MacRae, C.M., and Wilson, N.C., Microsc. Microanal. 14, 184–204, 2008

1.6 1.8 2.0 2.2 2.4 2.6 2.8 3.0 3.2 3.4eV

0

200

400

600

800

1k

Inte

nsity

Nv0

N3V

Band A

N2V

ZPL Nv0Nv0Nv0

Nv0Nv0 N2VZPL N2VN2VN2VN2VN2V

n3V ZPL

1.5 2.0 2.5 3.0 3.5eV

0

100

200

300

400

500

600

700

800

900

1k

Inte

nsity

Nv0

N2V

ZPL Nv0

Nv0

Nv0

Nv0

Nv0

N2VZPL N2VN2VN2V ZPL 3.188N2VN2V

High resolution hyperspectral cathodoluminescence imaging of semiconductor and plasmonic nanostructures

P.R. Edwards,* K.J. Lethy,* J. Bruckbauer,* A.W. Wark,** and R.W. Martin*

* Department of Physics, SUPA, University of Strathclyde, 107 Rottenrow, Glasgow G41 3EJ, U.K.

** Centre for Molecular Nanometrology, WestCHEM, Department of Pure and Applied Chemistry,

University of Strathclyde, 295 Cathedral Street, Glasgow G1 1XL, U.K.

There is an increasing trend within the semiconductor field to move towards feature sizes well below

the optical diffraction limit. This is motivated by different considerations, including: photonics

applications; improved light extraction for optical devices; and advanced methods for growth of low-

defect material. Cathodoluminescence (CL) provides the only far-field technique capable of spatially

resolving emission from such features. In this work, we present examples of high-resolution

hyperspectral CL imaging of various nanostructures using a field-emission SEM, and describe how

the technique has provided insights into the materials and devices studied.

As an example of high resolution CL imaging, Fig 1 shows emission from an array of gallium nitride

(GaN) nano-pyramids. These features are grown through a patterned array of holes; possible

applications include higher-efficiency LEDs, arrays of nano-scale light emitters, or—if the pyramids

are allowed to grow further and coalesce—films of material with reduced densities of lattice

dislocations [1]. Using hyperspectral imaging to map the peak wavelength emitted from the pyramid

facets shows the luminescence to blue-shift towards the apex, on a length scale of a few 10’s of nm.

This points towards a variation in composition during the sample growth.

An important aspect of semiconductor science is the identification, characterization, and control of

crystal defects. CL is useful in this regard by identifying, for example, regions of high non-radiative

recombination such as at dislocations, or the presence of impurity centers through the detection of

additional emission bands. Fig 2 shows the surface of an InGaN/GaN LED structure, where

dislocations threading through in the bulk material have resulted in the presence of surface defects

[2]. By peak fitting to the hyperspectral CL dataset, a previously undocumented trench-like feature

has been identified, associated with a higher luminescence intensity and red-shifted emission. This is

attributed to a variation in elastic strain, and the effect is observed on a length scale of only ~10 nm.

Finally, we present the results of using CL to investigate localized surface plasmons within silver

nanocubes. The very high electric fields induced near such nanoparticles have led to interest in using

them for molecule detection using surface-enhanced Raman spectroscopy. Cathodoluminescence

provides a means of investigating these surface plasmons by generating them via the beam/surface

interaction and observing their radiative decay. Using hyperspectral CL imaging we have

demonstrated sufficient spatial and spectral resolution to map the distribution of distinct spectral

features within cubes of only ~50 nm in size (Fig 3) [3].

References

[1] C. Liu, A. Šatka, L. Krishnan Jagadamma, P.R. Edwards, D.W.E. Allsopp, R.W. Martin, P.A.

Shields, J. Kováč, F. Uherek, W.N. Wang, Appl. Phys. Expr. (2009) 121002

[2] J. Bruckbauer, P.R. Edwards, T. Wang, R.W. Martin, Appl. Phys. Lett. (2011) 141908

[3] P.R. Edwards, D. Sleith, A.W. Wark, R.W. Martin, J. Phys. Chem. C 115 (2011) 14031

[4] Thanks to C. Liu, D. Allsopp, P. Shields and W. Wang of the University of Bath for samples.

This research was supported by the UK Engineering and Physical Sciences Research Council.

FIG. 1. SE image and fitted CL intensity (centre) and peak energy (right) from InGaN/GaN

quantum wells (QWs) grown on GaN nanopyramids.

FIG. 2. Surface defects on an InGaN/GaN QW sample, showing (left-right): SE image; fitted CL

intensity and peak energy maps; and linescans through the CL maps.

FIG 3. Intensity images and spectra corresponding to the first three principal components of a CL

hyperspectral image measured from 50 nm silver nanocubes (shown in SE micrograph, bottom left).

0 50 100 150 200

distance (nm)

0 50 100 150 200 0 50 100 150 200

400 500 600 700 800

CL inte

nsity (

a.u

., o

ffset)

wavelength (nm)

1st component

2nd component

3rd component

100 nm

6 kV

100 nm

6 kV

0.0 0.1 0.2 0.3 0.4 0.5

distance (µm)

0.0 0.1 0.2 0.3 0.4 0.5 0.6

distance (µm)

2.76

2.78

2.80

2.82

2.84c

en

tre (e

V)

Luminescent Database – Tools to Determine the Cathodoluminescence Transition Colin M. MacRae*, Aaron Torpy*, Nick C. Wilson* and Jackson Smith** *Microbeam Laboratory, CSIRO Process Science and Engineering, Clayton, Australia, 3168 ** RMIT Physics, Melbourne, Australia, 3001 Minerals and synthetic materials may emit luminescence in the infrared to ultra-violet range when exposed to a beam of electrons, ions or photons[1]. The most common luminescence activators found in minerals and synthetic materials are transition metal ions (Cr3+, Mn2+, Mn4+, Fe3+) and rare earth elements (REE2+/3+) [2]. The energy of the emission lines from these activators varies little between different host materials, and so can be used to uniquely identify luminescence activators and their charge states. A database of luminescence emissions was collated by CSIRO and made freely available both as a web resource[3] and as a component of the ‘OpticalFit’ spectrum fitting software for the MS Windows™ environment[4]. Since the database was released in 2005, the number of emission line entries has expanded from approximately 1200 to over 2500, and now comprises a total of 350 mineral and material host matrices. Each entry contains the host mineral or material name and associated chemical formula, the emitter type (including ions, intrinsic emissions, etc.), the wavelength of energy of line observed, and the full width at half maximum if reported. Wherever possible, the experimental temperature is included, as well as the type of technique used to excite the luminescence, and the reference to the original publication of the emission line. Example spectra are included when available as downloadable raw data files, in text or comma separated value formats. To better determine the emitters and resolve underlying spectral components the OpticalFit software can fit a number of Gaussian peaks to each spectra and the peak positions of the fitted Gaussians can be compared to those provided in the database. It is expected that these will not always align perfectly as spectrometers calibrations vary and subtle variations in bonding can emission peaks. In addition the OpticalFit software also differentiates between lines, bands and peaks with FWHM extracted from the database. An example of the luminescence database and the OpticalFit software is shown in figure 1, wherein the cathodoluminescence spectrum of the fluoride mineral Gagarinite-Y (NaCaYF6) from the Ilimaussaq Alkali Complex, Greenland, is overlaid with attributed lines, Eu3+, Sm3+, Dy3+ and Tb3+ taken from the luminescence database for rare earth emitters in Gagarinite, as reported by Ryabenko and Gorobets [5]. References 1. MacRae et al. Microscopy Research and Technique, 2005, vol. 67, number 5, pp 271-277 2. Gorobets, B.S. and Rogojine, A.A. Luminescent Spectra of Minerals- Reference Book.

(Translated from Russian, RPC Russia Institute of Mineral Resources (VIMS), Moscow, 2002) 3. http://www.csiro.au/luminescence/ 4. http://www.csiro.au/luminescence/opticalfit/ 5. Ryabenko S.V. and Gorobets B.S. Rare-earth fluorides from alkali-rare-earth metasomatites. -

Mineralogical Collected Papers ( Mineralogicheski Sbornik). - Lvov: Lvov State University. Vyshcha Shkola. - 1986. - No. 40 (1). - P. 77-80.

Figure 1. Cathodoluminescence spectrum of Gagarinite-(Y), (NaCaYF6), from the Ilimaussaq Alkali Complex in Kitaa, Greenland, showing line attributions as per Ryabenko and Gorobets.

Emitter Energy (eV) Eu 3+ 2.033 Sm 3+ 2.054 Dy 3+ 2.153 Tb 3+ 2.265 Dy 3+ 2.53 Gd 3+ 3.974

Figure 2. Cathodoluminescence emission lines for Gagarinite-(Y)[5].

1.4 1.8 2.2 2.6 3.0 3.4

eV

0

4k

8k

12k

Eu, Ryabenko et al. 1986 Sm, Ryabenko et al. 1986 Dy, Ryabenko et al. 1986 Tb, Ryabenko et al. 1986

Inte

nsit

y

Forensic Science Applications of Cathodoluminescence (CL) Microscopy and Spectroscopy J. Buscaglia*; C.S. Palenik**; S.A. Brokus***; D.K. Silletti****; D.E. Cooper*****; D.K. Purcell******; and G.F. Peaslee*** * FBI Laboratory, Counterterrorism and Forensic Science Research Unit, Quantico, VA 22135 ** Microtrace, LLC, Elgin, IL 60123 *** Hope College, Chemistry Department, Holland, MI 49423 **** ORISE Visiting Scientist Program, FBI Laboratory, Quantico, VA 22135 ***** FBI Laboratory, Latent Print Operations Unit, Quantico, VA 22135 ****** Graduate Center, The City University of New York, New York, NY, 10019 Cathodoluminescence (CL) microscopy and spectroscopy are applicable to a range of questions involving the forensic analysis of trace evidence and can aid in comparison, authentication, and provenance determinations of forensic geologic materials. The CL emission is characteristic of either the geological environment of formation of the mineral or, for a synthetic luminescent material, the manufacturing process. CL is observed in many materials routinely encountered as trace evidence, including sediment and rocks, building materials, glass, and polymers. The variation in luminescence for a particular mineral can therefore be used to discriminate among samples from different sources or, in certain cases, provide information about the provenance of a sample. CL is a relatively nondestructive method, which is an important consideration for the analysis of trace evidence, particularly when sample size is limited and court admissibility may restrict destruction, alteration, or consumption of evidence [1]. Within the category of geological evidence (e.g., sand, soil, and concrete (FIG. 1)), many of the most abundant minerals are cathodoluminescent (e.g., quartz, feldspar, and carbonate minerals). Traditionally, the use of these mineral components for forensic discrimination or sourcing has been limited due to their presence in nearly all samples. The variation in luminescence within a given mineral type can provide additional source discrimination beyond commonly used methods and offers the potential for improving the significance of geological evidence and evidence containing geologically derived additives. Research has demonstrated that cold cathode CL with light microscopy provides a relatively fast method to screen soil samples through visual identification of luminescent minerals (e.g., identification and classification of feldspars), the ability to determine if multiple populations of a given mineral exist (e.g., quartz), and a means to estimate the relative abundances of luminescent minerals in a sample. Zoning (i.e., compositional changes within a crystal), textures, and coatings are also revealed via CL imaging and can provide additional information about the origin of a sample. CL spectroscopy can offer more detailed information about specific activators in a given mineral. Together, visual and spectroscopic examination of mineral components can be combined to provide a variety of information about sediment samples that complement more traditionally used analytical techniques for forensic comparison and provenance determinations. Many synthetic or anthropogenically modified minerals such as pigments and fillers/extenders also luminesce (e.g., anatase, wollastonite, zincite and talc). Such minerals are used in the manufacturing of a variety of materials commonly encountered as forensic evidence including paint and duct tape. In both materials, CL provides a means to visualize details of the layer structure. In white architectural paints, for example, CL can be used to identify multiple layers that may not be visible by light microscopy (FIG. 2). The three main components of duct tape (i.e., adhesive, backing, and reinforcing fibers) all luminesce (FIG. 3). Within a given layer of paint or duct tape, CL can also be used to classify the major inorganic fillers/extenders and pigments, estimate the size of the inorganic component, and observe its distribution in a sample. The utility of CL for forensic trace evidence examinations will presented via research results for soil/sediments, mineral grains, tape, concrete, and paint. The combination of CL microscopy and spectroscopy with reflected light microscopy, image processing, and statistical methods will also be presented. A novel application of CL to detect neutron irradiation damage as a potential nuclear forensics tool will be introduced [2].

References [1] C.S. Palenik and J. Buscaglia. Applications of cathodoluminescence in forensic science in Forensic Analysis on the Cutting Edge. R.D. Blackledge (ed.) Wiley-Interscience R.D. New York, 2007. [2] This work was supported by the Counterterrorism and Forensic Science Research Unit of the FBI Laboratory Division, the FBI Laboratory Visiting Scientist Program, administered by the Oak Ridge Institute for Science & Education (ORISE), and Hope College. Additional funding support for Hope College from the Department of Homeland Security (2008-DN-077-ER0008) and the National Science Foundation (PHY-0969058) is acknowledged. The authors acknowledge the Missouri University Research Reactor for the use of their facilities and resources, and the Smithsonian Institution, National Museum of Natural History, Department of Mineral Sciences for their advice and generous donation of mineral specimens of known provenance. We also thank Robert Koons, Skip Palenik, Paul DeYoung, Dave Daugherty, Patrick Breen, Joshua Borycz, Rachel Driscoll, Elly Earlywine, J. Mark Lunderburg, and Ashley Standerfer for their contributions

FIG. 1. Concrete can be examined macroscopically to compare mineralogy, voids, and particle size in concrete-based materials. Shown above are two concrete samples and the differences observable by cold-cathode CL. The CL image of the left sample is dominated by calcite (orange), whereas the right sample contains mainly feldspar (blue).

FIG. 2. CL provides a means of visualizing paint layers that are not readily observable by traditional means. This cross section of architectural paint is shown in reflected light (left) and by cold-cathode CL (right). The paint is composed of four layers, three of which are dominated by calcite (orange). The fourth (blue) layer utilizes anatase as a white pigment.

FIG. 3. Shown are the cold-cathode CL images of duct tapes in cross section (oriented with the adhesive layer on the right) from 11 different sources. The luminescence of duct tape occurs due to the use of calcite, anatase, talc, and zincite as pigments and fillers/extenders. In addition to identification of the minerals used, CL provides useful visualization of morphological features such as layers, structure, voids, and pigment distribution.

Cathodoluminescence Spectral Complexity in Growth-Zoned Jadeite

E.P. Vicenzi,*,** A. Herzing,**, P.K. Carpenter,*** N. Wilson**** and C. MacRae****

* Museum Conservation Institute, Smithsonian Institution, Suitland, MD ** Surface and Microanalysis Science Division, NIST, Gaithersburg, MD *** Department of Earth and Planetary Sciences, Washington University, St Louis, MO ****CSIRO Minerals, Microbeam Laboratory, Clayton South, Victoria, AU

Jadeite is a sodium aluminum pyroxene (NaAlSi2O6) formed from high pressure and temperature

fluids within subduction zone complexes at tectonic plate boundaries [1]. Rocks composed of nearly

entirely jadeite are termed jadetites, and are a rare find at less than a dozen localities worldwide.

These rocks have played an important role as cultural heritage artifacts given their durability and

value as objects-of-art in Asia and Mesoamerica for the past millennia [2]. In addition to their

exceptional fracture toughness, these materials exhibit strong cathodoluminescence (CL). Because

jadeite CL is often intense it has been useful as a petrographic tool for metamorphic geologists, and

as such, many of these studies have been conducted using the CL luminoscope. Correlation of

elemental impurities with specific CL features has remained somewhat elusive though correlated

EPMA,SIMS, and LAICP-MS studies indicate that emission in the green region (~565 nm) is related

to Mn2+

while the blue region CL (~440 nm) is coupled with Al3+

or Eu3+

in some samples [3,4].

Because solid state interdiffusivities of cations in jadeite are slow even under geological conditions,

once an impurity element has been included into the mineral structure it cannot be readily

transported from its site of incorporation. As a result, complex CL zoning is not uncommon in jades.

In this study we seek to examine the nature and extent of such spectral complexity in a sample from

Guatemala using both a Gatan MonoCL4 and xCLent/JEOL systems.

Photomultiplier tube (PMT)-based imaging is an efficient method to reveal the strong heterogeneous

nature in our jadeite specimen (Figures 1A and B). However in order to discern spatially resolved

spectral features associated with the multifaceted core and rim zonation in these crystals we have

collected hyperspectral CCD-based data cubes.A maximum pixel spectrum of a spectrum image

requires at least 7 Gaussian peaks to describe the data (Figure 2B) and mapping-out the 3 main

deconvolved peaks allow one to begin unraveling the intracrystalline zoning (Figure 2A).

Multivariate statistical analysis is most useful tool in sorting through the 10s of thousands of CL

spectra collected during spectrum imaging. Reducing the dimensionality of these large data sets can

be achieved in an number of ways, and we have employed the spatial simplicity solution [5] to

compute just 9 components that can combined to mimic the hyperspectral CL data (Figure 3A).

These 9 end-member components and their distribution allow us to rigorously identify spectral

signatures that correlate with the crystal’s growth history, including late overprinting events.

References:

[1] Harlow et al., The Australian Gemmologist. 21 (2001) 7-10.

[2] Harlow et al., Canadian Mineralogist 44 (2006) 305-321.

[3] Sorensen et al., American Mineralogist 91 (2006) 979-996.

[4] Dopfel, BS Thesis Mount Holyoke College (2006) 61 pgs.

[5] Keenan, Surface & Interface Analysis 41 (2009) 79-87. Acknowledgements : We would like to thank Sorena Sorensen of the US National Museum of Natural History for the

specimen examined in this study. The thin section is part of her long term jadeitite research program conducted in

conjunction with George Harlow of the American Museum of Natural History.

data

Fit peaks

residual

Energy (eV) 1.6 2.0 2.4 3.6 4.0 4.4

1.61eVred

3.17eVgreen

1.92eVblue

CCD-based

Figure 1 (top) A) Panchromatic CL image of growth-zoned jadeite, B) PMT-based composite

overlay using red, green, and blue glass filters. Figure 2 (middle) A) False color RGB composite of

3 prominent Gaussian peaks fitted to the maximum pixel spectrum, B) Maximum pixel spectrum

(filled) fit with 7 Gaussian peaks. Figure 3 A) Multivariate statistical output overlay showing the

distribution of 9 components for the spatial simplicity solution, B) Spectral signatures of the 9 end-

member components in 3A.

A B

MVS Component overlay 260 360 460 560 660 760

Comp1

Comp2

Comp3

Comp4

Comp5

Comp6

Comp7

Comp8

Comp9

Inte

nsi

ty

Wavelength (nm)

A B

5 0 µ m

Spectrum Imagepanchromatic

PMT-based RGB

A B

spectrum image

1

2

3

Relationships between quartz trace elements and SEM-cathodoluminescence textures revealed using WDS and LA-ICP-MS mapping techniques Heather A. Lowers*, Alan E. Koenig*, Brian G. Rusk** *U.S. Geological Survey, MS 973, Denver, Colorado, 80225, USA ** Quartz is among the most common minerals in the crust and forms in many types of geologic environments. Scanning electron microscopy with cathodoluminescence (SEM-CL) reveals textures in quartz that are not observable by other methods such as transmitted or reflected light microscopy or backscattered electron imaging. Understanding the origin of CL textures and their relation to trace element chemistry is of broad interest in a wide range of geologic studies as it may reveal insights into the physical and chemical conditions of quartz crystallization. In this study, we examined quartz from various ore deposit types to determine the relationship between trace element abundances and CL variations. SEM-CL images were acquired at the USGS in Denver, Colorado, on a JEOL 5800LV SEM equipped with a Centaurus CL detector and photomultiplier. This photomultiplier has a spectral range from approximately 300 to 650 nm; however, it is more efficient at gathering light in ultraviolet and blue wavelengths than in red and infrared wavelengths. Quantitative analyses were obtained using a JEOL JXA 8900 electron microprobe operating at 20 kV, 50 nA (cup), a beam defocused to 5 µm to minimize specimen damage, and count times of 10 minutes on peak and 5 minutes on background. WDS mapping was performed using 20 kV accelerating voltage, 100 nA (cup), and a focused beam. Typical map sizes were 500 by 500 pixels with a one micrometer pixel size and dwell time of one second per pixel resulting in ~90 hour map runs. Elements mapped by WDS included Fe, Ca, Al, and Ti. Simultaneous mapping by EDS included Na, Mg, P, S, and K. Quantitative LA-ICP-MS trace element maps were constructed on polished 100 µm thick sections using a series of parallel line scans to produce a rectangular mapped area (e.g. [1]). A 193 nm (ArF excimer) LA system was used at an energy density of 8 J/cm2, a repetition rate of 5 Hz, and a scan speed of 3 µm/s. The beam size was 50 µm with a 60 µm spacing between adjacent line scans. A Perkin Elmer Elan DRC quadrupole ICP-MS was used. Individual time slices of the time-resolved spectra were converted to concentration using routine concentration calculations. Calibration was conducted using NIST612 and NIST610 glass reference materials, using GeoReM preferred values [2] with Si as the internal standard, assuming stoichiometric quartz. A concentration grid was created from each time slice and corresponding x-y location. Concentration maps were then constructed from the data using a commercial graphing program (Surfer, Golden Software). The following elements were analyzed by LA-ICP-MS (typical detection limits for each element are in parentheses in parts per million): Be (150), Li (5), B (50), Na (50), Mg (10), Al (25), K (25), Ti (10), Cr (50), Mn (20), Cu (20), Zn (50), Ga (5), Ge (20), As (40), Rb (5), Mo (15), Sn (15), Sb (5), Ba (5), Cs (3), W (15), Pb (10), Th (8), and U (5). The elements Li, Na, Al, K, Ti, Ga, Sb, and Ba were present in sufficient quantities and to be mapped in quartz from some samples. All other elements were consistently below the detection limits.

Comparisons of SEM-CL images and WDS x-ray intensity and LA-ICP-MS maps such as the one shown in Figure 1 were conducted. We find that variations in CL intensity correlate with trace elements yet the specific elements correlating with CL-bright or CL-dark zones vary from sample to sample. Elevated abundances of Al, for example, correlate with CL-dark zones in some samples and CL-bright zones in others. In some samples, Ca, Mg, K, and Ti also correlate with variations in CL intensity. This suggests that of the elements observed, none is consistently a CL-activator or a CL-quencher. It should be noted that the CL images obtained in this study are grayscale images even though CL emissions actually range from infrared to ultraviolet. In order to better understand the relationship between CL and trace element distribution, a study combining measurement of CL spectral emissions with WDS mapping and EPMA spot analyses of selected areas would be required. References [1] Koenig et al., 2009, Geology, 37, 511-514. [2] Jochum and Stoll, 2008, Laser Ablation ICP-MS in the Earth Sciences, 40, 147-168.

Figure 1. Comparison of the SEM-CL image and WDS x-ray intensity and LA-ICP-MS maps of hydrothermal quartz show the bright SEM-CL zones to be higher in Al, Li, and lower in Ga.

The Role of Native Point Defects in Degenerately Doped ZnO Films D. R. Doutt*, S. Balaz**, L. Isabella*, and L. J. Brillson*,**

*Department of Physics, The Ohio State University, Columbus, OH 43210 **Dept. of Electrical & Computer Engineering, Ohio State University, Columbus, OH 43210

We have used a complement of depth-resolved cathodoluminescence spectroscopy (DRCLS), optical absorption threshold, and Hall techniques to measure Fermi levels in degenerately Ga-doped ZnO (GZO) vs. growth temperature and their correlation with native point defects. DRCLS shows that Zn vacancy-related (VZn-R) defects increase monotonically as the ratio of acceptor to donor concentrations increases, demonstrating the delicate correspondence between donor and acceptor defects in highly n-type GZO. Furthermore, DRCLS reveals that Zn vacancy-related (VZn-R) defects decrease monotonically as free carrier concentration increases. Transparent conducting oxides (TCO) have found applications in many opto- and microelectronic applications such as transparent electrodes for flat-panel and flexible displays, photovoltaic cells, and LEDs. Of late, the demand for these applications and devices has grown, leading researchers to explore alternative materials for TCO applications. GZO has emerged as a strong TCO candidate due to its low cost and environmental safety. Until now, it has been unclear what kind of role native defects play in the resistivity, mobility, and carrier concentrations of these GZO films.

GZO films were grown by pulsed laser deposition (PLD) on quartz substrates [1]. Films were grown at substrate temperatures of 100, 200, 400, and 600 °C under both forming gas (FG: 5% H2 in Ar) and pure Ar ambient. DRCLS studies (fig. 1) reveal a dominant peak at 3.7 eV- 3.85 eV for all GZO films due to degenerate conduction band-valence band recombination. By using small probe energy increments we are able to plot the changes in Fermi level position with depth for these GZO films. We propose that the energy EFmax at which DRCLS conduction band emission decreases by 50% from its maximum intensity on a linear scale (fig. 2) provides the estimate (EF-EV)DRCLS. We can extract the charge density (n) from the shift in Fermi level position due to degenerate doping and observe how the carrier densities vary with depth and with sample. DRCLS of GZO films show emission from 1.7- 2.1 eV isolated or clustered VZn emphasizing that (VZn- R) are the only relevant acceptors available [2]. Plotting the ratio of the peak intensities I (VZn-R)/I(NBE) vs. the ratio of acceptor (NA) and donor (ND) densities for a characteristic probe energy shows that higher acceptor densities relate to lower doping and carrier densities. Furthermore, DRCLS free carrier densities inside these sub-micron-thick GZO films can vary with depth, introducing spatial variations in EF – EC that display the same (VZn-R) density dependence on a nanometer scale (fig. 3). These results suggest a general role of native point defects in semiconductor doping.

References [1] R.C. Scott, K.D. Leedy, B. Bayraktaroglu, D.C. Look, and Y-H Zhang, Appl. Phys. Lett. 97, 072113 (2010). [2] Y. Dong, F. Tuomisto, B.G. Svensson, A. Yu. Kuznetsov, and L.J. Brillson, Phys. Rev. B, Rapid Communications 81, 081201(R) (2010)

FIG 1: Typical DCLS for GZO shows 1.8 and 2.04 eV VZn-R features and emission from conduction band states extending below EC up to EFmax.

FIG 2. DRCLS spectrum of FG-grown GZO grown at 600°C versus incident beam energy EB. EFmax indicates energy at which intensity drops to 50% of its maximum value.

Figure 3: Nanometer scale anticorrelation of nDRCLS versus I(VZn-R)/I(NBE) within GZO film

High Resolution Luminescence Microscopy under Simultaneous Electron Beam and Laser Irradiation Jonathan D. Poplawsky and Volkmar Dierolf Lehigh Physics Department, 16 Memorial Drive East, Bethlehem, PA 18015 We have developed a fiber based confocal optical microscope that operates inside of a commercial SEM instrument (JEOL 6400) enabling the excitation of a sample either by a laser or by electron beam, and hence combining the complimentary techniques of photoluminescence (PL) and cathodoluminescence (CL). The module is attached to a cryogenic stage (Oxford) allowing temperature dependent measurements. The instrument uses single-mode fibers that enter the SEM by vacuum feedthroughs. The illumination and collection fibers operate as effective pinholes providing, in combination with a microscope objective (NA=0.3), high spatial resolution (~2µm) and excellent collection efficiency. The high spatial resolution ensures that the light collected from the sample is in a region of optimal laser beam and electron beam overlap. The setup is shown in Fig.’s 1 and 2. The capabilities of the instrument are demonstrated by experiments involving the excitation of europium ions in-situ doped in GaN thin films [1]. Rare earth (RE) ions have been investigated for their usefulness in LED lighting applications as well as high powered lasers. The link between a direct excitation of RE ions compared to excitation through energy transfer from (injected) electron-hole pairs (as required in a LED) is still not well understood. With this instrument, it is possible to combine both excitation schemes. Most notable, we find that the total light emission is reduced when the electron beam and a visible laser are combined and not increased as would be expected in simple excitation models. The experimental results indicate that the excitation is defect-trap mediated and also show that not all rare earth ions can be excited through the excitation of electron-hole pairs. The instrument will enable a wide variety of studies in which the link of laser and e-beam irradiation of samples is of interest. In addition to the combination of PL and CL studies, the structural information from SEM measurements can also be combined with those from Raman spectroscopy. This has been implemented in our instrument. Overall, we have shown the feasibility of a high spatial resolution fiber based combination of confocal microscopy/spectroscopy and electron beam irradiation, which opens up a host of studies that were not previously possible. References [1] J. Poplawsky et. al., Mater. Res. Soc. Symp. Proc. Vol.1342 (2011) [2] L. Bodiou and A. Braud, Appl. Phys. Lett. 93, 151107 (2008).

FIG 1: A schematic of the combined PL and CL experimental setup.

FIG 2: A picture of the confocal microscope mounted on an Oxford low temperature SEM stage. The fiber closest to the camera is the laser input fiber (1), which directs the collimated beam into a beam splitter (2), and is then reflected by two mirrors (3) through the objective (4). The collected light then follows the same path in the opposite direction, travels through the beam splitter, and into the collection fiber (5).

AlGaN/GaN high electron mobility transistor degradation investigation through nanoscale depth resolved cathodoluminescence spectroscopy Chung-Han Lin*, D. R. Doutt**, T. A. Mertz**, , Jungwoo Joh***, Jesus, A. del Alamo***, U. K. Mishra**** and L. J. Brillson*Department of ECE, The Ohio State University, Columbus, Ohio 43210, USA

*****

** Department of Physics, The Ohio State University, Columbus, Ohio 43210, USA *** Microsystems Technology Laboratories, MIT, Cambridge MA 02139, USA ****ECE and Materials Department, University of California-Santa Barbara, Santa Barbara, California

93106 USA *****

GaN-based high electron mobility transistors (HEMTs) are leading devices for high frequency, high power electronics due to their high current density, breakdown voltage and temperature capabilities. However, extreme operating conditions such as high current and high voltage challenge device reliability. Furthermore, the combination of these effects plus appearance at highly localized region not only impedes the investigation of failure mechanism but also the prediction of device failure. In this study, we use depth resolved cathodoluminescence spectroscopy (DRCLS), Atomic force microscopy (AFM) and Kelvin probe force microscopy (KPFM) to measure the 3-dimensional temperature variation, stress distribution, and native point defect generation inside operating GaN-based HEMTs under on- and off- state stress.

Department of ECE, Department of Physics, and Center for Materials Research, The Ohio State University, Columbus, Ohio 43210, USA

The shift of the near band edge emission (NBE) obtained by DRCLS can be used to extract both temperature and stress. According to our DRCL results, we found that there is still temperature variation (~ 30) at gate-drain spacing (~ 600nm) even at low power operation (fig. 1(a)). The shift NBE with gate-drain voltage VDG show that the field-induced stress increases up to 0.27 GPa, affecting threshold characteristics (fig. 1(b)). The DRCLS also show that from both the “channel” and “under gate”, the defect generation after on-state and off-state operation of the GaN-based HEMTs. Three major signals were found which NBE, blue band (BB) and yellow band (YB) (fig. 2(a)). The averaged YB/NBE ratio shows that YB at under gate –drain side region increase the most (~ 3.5 times) compared with under gate –source side (~ 2.2 times) and channel region(fig. 2(c)). Fig. 2(d) and (e) show numbered KPFM patches of an Off-state stressed transistor and corresponding changes in DRCLS defect intensity for each patch, respectively. Fig. 2(f) shows the variation in surface potential with VDG. For low potential regions such as areas 9 and 10, there is a striking change in the rate of surface potential decrease above Off-state stress VDG ~ 20-25 V. The corresponding increase in gap state defect density with lower EF can be calculated using a self-consistent electrostatic model. In general, negatively charged acceptors induce a dipole that shifts EF lower in the band gap, depending on the defect density, charge separation, dielectric permittivity, and background doping. As defect density increases, EF moves closer to the gap state. Using deep level spectroscopy measured trap densities of high 1012 cm-2

for similar stressed HEMTs, we fit the potential variation to defect densities increasing by nearly two orders of magnitude with a charge separation of approximately 8 nm, corresponding to defects located at or below the AlGaN surface. In summary, DRCLS techniques combined with other scanning probe techniques can be used not only to reveal the failure mechanisms, but also improve device performance for future applications.

References: [1] C.-H. Lin et al., Appl. Phys. Lett., 95, 033510-1 (2009) [2] C.-H. Lin et al., Appl. Phys. Lett., 97, 223502-1 (2010) [3] This work is supported by the Office of Naval Research Grant No. N00014-08-1-0655 (Dr. Paul Maki and Harry Dietrich)

Fig. 1 (Color) DRCLS results show (a) Temperature distribution of a GaN HEMT operated at VDS = 3.5 V and VGS

= -2 V. P1 is nearest the gate while P10 is furthest away. Inset photo shows SEM image of source-drain region measured [1]. (b) External stress at gate edge drain side caused by applied voltage under Off-state stress [2]. A 30 ˚C difference and a 0.27 GPa increase can be observed.

Fig. 2 (Color) DRCLS results for a GaN HEMT, (a) DRCLS spectrum from under the gate overhang before operation; Inset: schematic of device cross section. Arrow indicates incident electron beam path through gate overhang at EB = 20 keV. (b) increase in 2.2 eV Yellow Band (YB) emissions before (also (a)) and after 12 hours, VD= 10 V, VG= -2 V, 476 mA/mm operation. (c) Position-dependent average YB/NBE ratios showing largest increase under drain-side gate after 12 hour on-state operation. Inset shows SEM image and definition of Area 1-5 [1]. (d) Off-state KPFM maps showing numbered “upper” and “lower” low potential regions and (e) corresponding surface potential and average YB/NBE intensity ratio increases with Off-state stress. From color potential scale (red higher, blue lower), YB/NBE increases most at lowest potential regions 6, 8, 9, and 10. Higher potential patches display slower changes. 1 and 2 correspond to extrinsic drain and drain-side gate edge of an unstressed reference device. BB/NBE also increases in Off-state in regions 9 and 10. (f) Surface potential variation vs. VDG and corresponding self-consistent electrostatic defect density [2].

Cathodoluminescent textures and trace elements in hydrothermal quartz B.G. Rusk* * School of Earth and Environmental Sciences, James Cook University, Townsville, QLD, Australia and Department of Geology, Western Washington University, Bellingham, WA, 98225

When viewed with scanning electron microscope-cathodoluminescence (SEM-CL), hydrothermal vein quartz displays textures that are unobservable using any other technique (Fig.1). These textures provide unique views into the sequence of quartz precipitation and dissolution events during hydrothermal vein formation. Such textures relate specific quartz generations to specific mineralization events or fluid inclusion populations. In addition to illustrating temporal relations among multiple generations of quartz in a single vein, CL textures reflect fluctuations in pressure, temperature, or composition of hydrothermal fluids.

CL textures and their spectra result from defects in the quartz lattice, including those caused by trace element concentration variations. Trace element abundance and distribution result from variations in the physical and chemical conditions of quartz precipitation, followed by any subsequent solid-state changes in quartz chemistry. Here we show that CL textures, CL spectra, and trace element concentration vary systematically between different types of hydrothermal ore deposits.

Porphyry copper deposits, which form over a temperature range between ~350° and ~700°C, are characterized by fracturing, dissolution, and recrystallization textures that reflect multiple generations of fluid flow in a single volume of rock. Early quartz in porphyry copper deposits is CL-bright, and is typically overgrown by CL-dark quartz that is related to sulfide precipitation [1, 2]. Ti concentration is correlated with CL intensity in quartz veins from porphyry copper deposits, such that early CL-bright quartz contains up to ~250 ppm Ti, and later CL-dark quartz contains <10 ppm Ti [1, 2, 3]. Al concentrations in porphyry copper quartz are typically between 50 and 200 ppm and most other trace elements are present in concentrations below ~5 ppm. Ti-rich CL-bright quartz has a strong emission band at ~450 nm as well as a broad band at ~600 nm. In CL-dark quartz with less Ti, the 450 nm spectral band is small or absent and the broad 600 nm band dominates.

In epithermal quartz that forms at much lower temperatures (~150°-300°C), CL textures are dominated by growth textures that vary from microcrystalline spheres to centimeter-scale euhdral crystals with sharply defined growth zones of oscillating CL intensity. Replacement textures, such as quartz replacing calcite are common as are recrystallization textures, but textures that suggest diffusion or dissolution are absent. Trace element concentrations in epithermal quartz are distinctly different from those in porphyry copper quartz (Fig. 2). Ti concentrations are below 5 ppm and Al concentrations range from a few ppm to several thousand ppm. Lithium is correlated with Al and ranges up to several hundered ppm in quartz where Al is in the range of several thousand ppm [4]. Antimony is also abundant in epithermal quartz (up to ~200 ppm). No consistent progressive change in CL intensity or trace element concentration can be generalized in epithermal quartz, nor is there a consistent relationship between trace element abundance and CL intensity or CL spectra. A broad band with a peak at ~600 nm is the most common CL spectral emission in epithermal quartz. References [1] B. Rusk et al., Am. Min. 91 (2006) 1300.

[2] A. Mueller et al., Min. Dep. 45 (2010) 707. [3] M. Landtwing and T. Pettke, Amer. Min. 90 (2005) 122. [4] B. Rusk et al., Am. Min. 96 (2011) 703.

FIG 1. A) Plane polarized light; B) cross polarized light; C) Back-scattered electron; and D) SEM-CL image of vein of a quartz vein from the epithermal deposit in McLaughlin, CA. Transmitted light images show typical mosaics of quartz with little obvious textural information to differentiate multiple generations of quartz growth. SEM-CL reveals micro-textures that are not visible using any other analytical technique. SEM-CL textures are key for deciphering the history of vein formation. Pyrite precipitated last in the banded microcrystalline quartz, but both pyrite and microcrystalline quartz are overgrown by CL-bright euhedral quartz (lower left).

FIG 2. Titanium versus Aluminum concentrations in hydrothermal quartz from epithermal, orogenic Au, and porphyry-type deposits. The different deposit types can be distinguished from one another based on trace element concentrations alone. Quartz trace elements reflect the physio-chemical conditions of quartz growth and subsequent modifications.

Collection and Quantification of Hyperspectral Cathodoluminescence Colin M. MacRae, Nick C. Wilson, Aaron Torpy and Cameron Davidson colin.macrae@csiro.au Microbeam Laboratory, CSIRO Process Science and Engineering, Clayton, Australia Advances in hardware and software associated with cathodoluminescence (CL) detection have resulted in improved light sensitivity, faster data collection, better system stability, and enabled hyperspectral data collection and analysis[1]. In this presentation we describe further developments that enable the quantitative analysis of CL emitters in the wavelength range from ultraviolet (UV) to near infra-red (NIR). A dual CL spectrometer collection system was integrated with an electron microprobe to collect luminescence generated by the electron beam across the wavelength range from 200 nm to 1700 nm. The CL signal is collected and focused through a series of mirrors and lenses into two Peltier cooled CCD spectrometers, which simultaneously measure the UV/Visible (UV/Vis) spectrum from 200 to 980 nm, and the NIR spectrum from 900 to 1700 nm. The spectrometers overlap in the range 900-980 nm, which allows an intensity scaling factor to be calculated to give a smooth transition across the overlap region. The ability to simultaneously measure UV/Vis and NIR transitions is advantageous for measuring CL transitions associated with rare earths ions, which have emission lines spread across the spectrum from visible to NIR that are detectable and quantifiable (e.g. Nd3+ in Scheelite[2,3], Tb3+ and Dy3+in Fluorite[3,4]). This broad-spectrum collection strategy allows a greater range of emitters to be analysed, and also provides a means of easier peak identification. To quantify the concentration of rare earth elements within an area mapped by CL, a set of Gaussian / Lorentzian peaks are first fitted to the spectra of every pixel in the map. An independent measurement of trace elemental abundances from areas within the mapped region is then performed by electron probe microanalysis or laser induced inductively coupled mass spectrometry (LA ICP-MS). From this data, a calibration line is calculated for each CL peak by regression of the trace analyses against the fitted CL intensities from the same areas. This calibration line allows quantitative speciation maps to be plotted with a sensitivity of parts per million[5]. A grain of Wilberforce apatite (Ca5[PO4]3[F,OH]) has been investigated using the dual CL spectrometer system. The combined UV/Vis/NIR cathodoluminescence spectrum from the grain is shown in figure 1. A map collected at 20kV and 40nA shows clear differences in spectral response between the UV/Vis spectrum that is dominated by the intrinsic response (figure 2a) and the NIR region that is dominated by Nd3+ peaks (figure 2b). This new technique offers the ability to characterise a greater range of CL transitions and will provide new information in many materials and mineral problems.

References [1] MacRae, C.M., et al., Microscopy Research and Technique, 67(5): 271-277, 2005. [2] MacRae, C.M., et al., Microsc Microanal., 13 (Supp2): 1394CD-1395CD, 2007. [3] MacRae, C.M., and Wilson, N.C., Microsc. Microanal. 14, 184–204, 2008 [4] Pagel, M., Barbin, V., Blanc, P. & Ohnestetter, D. (2000). Cathodoluminescence

in Geosciences. Berlin; New York: Springer. [5] MacRae, C.M., et al., Microsc. Microanal. 15, 222–230, 2009 Figures

200 400 600 800 1000 1200 1400 1600nm

0

40

80

120

Inte

nsity

Figure 1. Cathodoluminescence spectra from a Wilberforce-apatite grain* collected by combined UV-VIS and VIS-NIR spectrometers. The broad peak centred at 420 nm is the intrinsic apatite peak while peaks at 870, 1070 and 1340 nm are attributed to Nd3+.

\ Figure 2.(a) Visible sum-spectra map and, (b) map of the Nd3+ peak centred at 1050 nm, collected on Wilberforce-apatite at 20 kV, 2 micron step size and 80 ms dwell per pixel. * Wilberforce apatite grain (Ontario, Canada) was provided by Museum Victoria.

a b

Nd3+

Nd3+

Nd3+

Cathodoluminescent Signatures of Neutron Irradiation Danielle Silletti1,2, Sarah Brokus1, Elly Earlywine1, Joshua Borycz1, Paul A. DeYoung3, Graham F. Peaslee1, and JoAnn Buscaglia4

1 Chemistry Department, Hope College, Holland, MI 49423 2 Federal Bureau of Investigation Laboratory, ORISE Visiting Scientist Program, Counterterrorism and Forensic Science Research Unit, Quantico, VA 22135 3 Physics Department, Hope College, Holland, MI 49423 4 Federal Bureau of Investigation Laboratory, Counterterrorism and Forensic Science Research Unit, Quantico, VA 22135

Nuclear proliferation and the potential threat to national security from unsecured special nuclear materials have renewed our national interest in not only detecting the presence of these materials, but also in detection of materials’ pathways into this country. Currently, the only method to identify where special nuclear materials have been stored involves measuring induced radiation in adjacent materials, which is usually short-lived (n, gamma) radiation. We have identified a permanent change to the luminescent properties of certain common minerals that is due to neutron irradiation that could potentially be developed into a nuclear forensics tool. Feldspars and calcite are ubiquitous minerals that are known to luminesce under electron bombardment. The UV-Visible spectra of hundreds of individual potassium feldspar and calcite grains were measured with cathodoluminescence (CL) spectroscopy before and after neutron irradiation. CL excitation uses an electron beam to induce fluorescence in certain minerals due to their chemical composition and defects in their crystal lattice structure; the resultant emission spectrum is acquired with a UV-Visible spectrometer. The presence of ionizing radiation causes additional crystal lattice defects that leave a permanent CL signature. A spectroscopic signature is described that increases proportionately to neutron dose in both calcite and feldspars. Preliminary dose-response results from a neutron source study and a reactor study are presented. There is also an orientation dependence in the luminescence measurement technique that complicates the analysis, but when fully understood could allow not only the total dose to be estimated, but also the direction of origin of the neutrons to be determined. References [1] Moxon, T. and Reed, S. J. Mineral. Magazine 70(5) (2006) 485-498. [2] Palenik, C.S. and Buscaglia J, Forensic Analysis on the Cutting Edge, R.D. Blackledge (ed.)

Wiley-Interscience, 2007. [3] This work was funded by The Department of Homeland Security (2008-DN-077-ER0008), the

National Science Foundation (PHY-0969058) with support from the Counterterrorism and Forensic Science Research Unit of the FBI Laboratory Division and Hope College. This research was supported in part by the FBI Laboratory Visiting Scientist Program, administered by the Oak Ridge Institute for Science & Education (ORISE). We would like to acknowledge the Missouri University Research Reactor for the use of their facilities and resources. We would also like to thank Dave Daugherty and Robert Koons for their contributions.

FIG. 1. CL spectra of a single calcite grain in 3 random orientations

FIG. 2. Calcite dose-response curve

FIG. 3. CL spectra of a single potassium feldspar grain in 3 random orientations

FIG. 4. Feldspar dose-response curve

Thermal process dependence of Li configuration and electrical properties of Li-doped ZnO

Z. Zhang,* T. Merz,** K-E. Knutsen,*** A. Yu. Kuznetsov,*** B. G. Svensson,*** L. J. Brillson****

* Department of Electrical and Computer Engineering, The Ohio State University, Columbus, OH 43210 ** Department of Physics, The Ohio State University, Columbus, OH 43210 *** Department of Physics, University of Oslo, P.O. Box 1048 Blindern, N-0316 Oslo, Norway **** Department of Electrical and Computer Engineering and Physics, The Ohio State University, Columbus, OH 43210

We used nanoscale depth-resolved cathodoluminescence spectroscopy (DRCLS) coupled with other surface science techniques to explore the relationship between the electrical properties and defects in Li doped ZnO. As a leading candidate for next generation opto- and micro-electronics, ZnO has attracted tremendous interest. However, despite extensive research in past decades, several fundamental issues remain unsolved, such as the control of amphoteric Li in ZnO. Li is unintentionally incorporated during HT-ZnO synthesis, or intentionally doped for p-type ZnO. However, HT-ZnO wafers and Li-doped ZnO are usually highly resistive as a result of Fermi-level dependent and low energy barrier to switch between interstitial LiI and substitutional LiZn as well as high compensation of acceptors and donors.[1, 2] Previous work showed our ability to control Li concentration by high temperature treatment or vacancy clustering.[2] Also our group correlated the 1.9-2.1eV and 2.3-2.5eV luminescence emissions with zinc vacancy (VZn) and oxygen vacancy (VO) related defects, respectively.[3] In this work, Li-doped melt-grown (MG)-ZnO samples were annealed 10 min in 5% Li2O and ZnO powder from 5000C to 6000C, with some quenched in DI water and the others slow cooled in air. Positron annihilation spectroscopy (PAS) of the slow cooled sample showed the formation of LiZn acceptor with large quantity.[4] DRCLS reveals an inverse relationship between the optical emission densities of lithium as interstitial versus on zinc vacancy sites, demonstrating the time dependence of Li interstitials to combine with zinc vacancies in order to form substitutional Li acceptors.

As shown in Fig.1, DRCLS spectra on Li-doped ZnO samples showed only a near band-edge emission in as-grown ZnO. After annealing at 500oC, the samples have higher intensity of 2.06 eV peak in quenched sample, which has been determined as VZn-related defects.[4] In contrast, a new peak around 3.0 eV stands out clearly in slow cooled sample. Four-point probe measurements showed that the quenched sample has low resistivity, 0.94 Ω•cm, however, the slow cooled samples are semi -insulating. It indicates that during anneal at 500oC Li diffuses into the sample, ionizing and acting as interstitial Li donor. When quenched in DI water, the interstitial Li doesn’t have time to diffuse and stays as a LiI donor. However, during slow cooling process in air, the LiI donor has enough time to fill in VZn and becomes substitutional LiZn, highly compensated by LiI and other intrinsic and extrinsic donors, resulting in high resistivity. This decreases the 2.06 eV (Vzn-related) peak intensity and increases 3.0 eV (Lizn-related) in Fig. 1(b). SIMS profile (Fig. 2b) of Li doped ZnO samples demonstrates that quenched and slow cooled ZnO samples after 500oC anneal have similar Li profiles, which indicates different cooling processes didn’t change the Li concentration in the samples. Scanning Spreading Resistivity Measurement (SSRM) profiles in Fig 2 (a) shows that Li diffuses deeper at higher temperatures, and more importantly, the resistivity of slow cooled samples follows the Li concentration. Fig. 3 shows the slow cooled Li-d6 after 600oC anneal. It’s clear that the 3.0eV peak in DRCLS spectra, LiZn acceptor as indicated, follows the Li concentration in the SIMS profile of the sample (inset of Fig. 3). Surface photovoltage spectroscopy (SPS) in Fig 4 confirmed the acceptor state of 3.0 eV peak. These studies identify the LiZn acceptor energy level, reveal thermal process dependence of Li configurations and electrical properties in ZnO, and promote low resistivity auto-doped n-ZnO and realization of stable p-type ZnO.

[1] M. G. Wardle, J. P. Goss and P. R. Briddon, Phys. Rev. B 71, 155205 (2005). [2] E. V. Monakhov, A. Yu. Kuznetsov and B. G. Svensson, J. Phys. D: Appl. Phys. 42, 153001 (2009).

[3] Y. Dong, F. Tuomisto, B. G. Svensson, A. Yu. Kuznetsov, and Leonard J. Brillson, Phys. Rev. B 81, 081201 (2010). [4] K. M. Johansen, A. Zubiaga, I. Makkonen, F. Tuomisto, P. T. Neuvonen, K. E. Knutsen, E. V. Monakhov, A. Yu. Kuznetsov, and B. G. Svensson, Phys. Rev. B 83, 245208 (2011). [5] The authors gratefully acknowledge support from the National Science Foundation Grant No. DMR-0803276 (Charles Ying), ECCS-0923805, and the Norwegian Research Council

2.0 2.5 3.0 3.5

Slow cooled

Semi-insulating

As grown1.2 ohm*cm

Quenched0.94 ohm*cm

2.0 2.5 3.0 3.5

1 2 3 4 5

0.04

0.08

0.12

EB (keV)

I LiZ

n/IV

Zn (A

.U.)

725032187Sampling Depth (nm)

Quenched0.94 ohm*cm

Slow cooledSemi-insulating

As grown1.2 ohm*cm

Photon Energy (eV)

3.02eV2.07eV

Slow cooled

Quenched

(b)(a)

Uo(nm)

18

725032

1 2 3 4 5

7E

B(keV)

2.07eV

Photon Energy (eV)

Inte

nsity (arb

. units)

3.00eV

Fig.1 80K CL spectra for Li-doped ZnO (a) as-grown, quenched in DI water and slow cooled in air after annealed at 500oC, (b) 5 keV CL spectra comparison and (inset) relative intensity changes of VZn and LiZn related defects in ZnO after quenching and slow cooling process.

0 1 2 3 4 5 6

1015

1016

1017

1018

1019

0 1 2 3 4 5 6

101

103

105

107

109

500oC Quenched

550oC Slow cooled

500oC Quenched

550oC Slow cooled

500oC Slow cooled

525oC

Slow cooled

Li conctr

(cm

-3)

Depth (um)

ρ (O

hm

*cm

)

525oC Slow cooled

Fig.2. (a) SSRM of slow cooled ZnO after 500oC, 525oC, and 550oC annealing, (b) SIMS profile of quenched ZnO after 500oC annealing, slow cooled ZnO samples after 500oC, 525oC and 550oC annealing.

2.0 2.5 3.0 3.5

101

102

103

104

105

0 40 80 120 160 2005E17

1E18

5E18

1E19

Depth (nm)

Li C

onc. (c

m-3

)

0.05

0.1

0.5

1

I(LiZn)/I(NBE)

I(LiZn)/I(VZn)

Inte

nsity

(A.U

.)

3.00 eV2.07eV

Photon Energy (eV)

Inte

nsity (arb

. units)

1 2 3 4 5

EB(keV) Uo(nm)

18

725032

7

Fig.3. 80K CL spectra for Li-doped ZnO annealed in 600oC and slow cooled in air, and relative intensity changes of VZn and LiZn related defects following SIMS profile of Li concentration (inset). 3.0 eV peak rises up from 32 nm below the surface, and peaks above 72 nm, corresponding to the Li profile.

0.5 1.0 1.5 2.0 2.5 3.0 3.5

0.365

0.370

0.375

0.380

0.385

0.390

0.395

0.400

0.405

EVacEC

EV

ED EF

EF’3.0 eV

ΔΦ

Conta

ct P

ote

ntial D

iffe

rence (V

)

Incident Photon Energy (eV)

data Interpolated data

2.45 eV

3.0 eV

3.3 eV

2.15 eV

Fig.4. SPS spectrum of flash-annealed, slow-cooled Li-implanted HT-ZnO showing onset of photodepopulation from 3.0 eV acceptor state, decreased band bending, and higher Fermi level.

MSA/MAS Hyper-dimensional Spectral File Format Nestor J. Zaluzec***, , Nick Wilson**, Aaron Torpy**, and Mike Kundmann* ***Electron Microscopy Center, Argonne National Laboratory, Argonne, IL 60439 USA **CSIRO Process Science and Engineering, Clayton South, VIC 3169, Australia *e-Metrikos, P.O. Box 5506, Pleasanton, CA 94566, USA The last decade has seen a rapid and expanded growth of analytical methodologies to address characterization of technologically important materials. In these technologies, multi-dimensional data collection and analysis is a common aspect of our work and spans across a wide range of hardware and software used by the microanalysis and microscopy community. Some of these methods include: XEDS, EELS and CL spectroscopy and spectrum imaging, tomography in TEM/STEM/FIB, , in-situ dynamic time series, as well as many other modes. While the prevalence of these experimental multi-dimensional measurements is growing, there is, as yet, no commonly recognized standard file format for this type of data. At present, microscopy and microanalysis data sets of this type are saved in many different formats, some of which are proprietary. This poses problems for the long-term archiving of the data, as well as the sharing and comparative analysis of results between different labs and software packages. In the early 1990’s, the simpler problem of XEDS and EELS spectral file sharing led to the development of the EMSA/MAS spectral file format [1, 2], which is now an established ISO standard [3]. This format was optimized for the relatively small spectral data sets of two decades ago, but it is no longer well-suited to the gigabyte (and larger) data sets of today. The MSA Standards Committee, which is composed of members from both the Microscopy Society of America (MSA) and the MicroAnalysis Society (MAS), is in the process of defining a file format better matched to the demands of present-day microscopy and microanalysis techniques and has developed an outline for a hyper-dimensional spectral file format aimed at the broadest range of microanalysis applications. This format is open for comment by the community so as to maximize it’s effectiveness. Preliminary input has already been received from both researchers and vendors at the recent Microscopy & Microanalysis-2011 Meeting in Nashville, Tn held in August of 2011. In designing this format, there is always a tradeoff of design goals such as simplicity, speed of access, and size. The goal of the new format is to facilitate data sharing and long term access, not efficiency, thus the key parameters are simplicity and long term readability. The file structure should be easy to interrogate so that users can read or process data using scripts or coding in readily available programming languages. In addition, as in the original EMSA/MAS spectral file format it should be self-descriptive so that even someone unfamiliar with the format specification can easily and correctly interpret the stored data and its parameters. A fully text-based (e.g. ASCII) file format, such as the original EMSA/MAS format is no longer suitable due to the large nature of hyper-dimensional data sets. The committee has therefore concluded that the hyper-dimensional data of such a set is best stored at least in some part in compact binary form. Pure binary format is also be impractical, as one cannot read it without a priori knowledge of the format specification. It has therefore been proposed to split the data into two files, one containing the experimental data in binary form and another in XML format [5] describing the binary data layout and containing all other ancillary information about the measurements. The association of a file pair is maintained

by two independent mechanisms to minimize the risk of information loss if files are separated or renamed. At the file system level, the pair have the same name with different extensions (.hmsa for the binary data, .xml for the descriptor). Internally, each file contains a (practically) unique identifier, a 64-bit number randomly generated for each pair. XML has been chosen for the descriptor file because it is a text-based format that is easily viewed with a web browser or edited with basic text editing programs, such as NotePad or TextEdit. There is wide support for XML, both for users of such files and software developers [4]. XML also offers the advantage of supporting a broad range of text encodings, including Unicode, which had to be added to the original EMSA/MAS format when it was accepted as an ISO standard. The descriptor file contains a variety of XML tags, some required, some optional. A core set of required tags specify the dimensionality and binary type of the measured data. A larger set of optional tags contain descriptive metadata about the measurements, including specimen information, experimental conditions, instrumental setup, processing parameters, and their origination. Although the format is designed for broad applicability and routine practical use, a number of potential features have been purposely left out to maintain simplicity. There is no provision for minimizing file size through compression. This can be applied by users, as desired, via freely and commercially available compression utilities. This format also does not address the much larger problem of shared hierarchical databases for microscopy and microanalysis informatics [6]. It is a modest proposal for a standard container for moderately sized data sets of finite scope. The Standards Committee is preparing a specification document that provides an in-depth review of the motivation, goals, and underlying considerations of this file format design effort, as well as a detailed description of the format and its XML tags, including an XML schema file and several example files. We will be making these materials available for public review and comment [7] and seek feedback from all stakeholders in the microscopy and microanalysis community. Interested parties wishing to participate in detailed discussions should contact the sub-committee’s chair (zaluzec@aaem.amc.anl.gov) or attend its annual meeting held during the M&M conference. References [1] R.F. Egerton, C.E. Fiori, J.A. Hunt, M.S. Isaacson, E.J. Kirkland, N.J. Zaluzec, Proceedings of

the Electron Microscopy Society of America, San Francisco Press, (1991) 526. [2] R.F. Egerton, et al, EMSA Bulletin, 21, (1991) 35. [3] International Organization for Standardization, standard ISO 22029:2003. [4] J.H.J. Scott, S.A. Wight, B.B. Thorne, Microscopy and Microanalysis, 8 Suppl. 2, (2002) 646. [5] XML is a standard maintained by the World Wide Web Consortium at www.w3.org/TR/xml/ [6] J.H.J. Scott and N Ritchie, Microscopy and Microanalysis, 15 Suppl. 2, (2009) 540. [7] http://www.amc.anl.gov/ANLSoftwareLibrary/MSAMASFormat/

Preparation of High Quality Sample Surfaces Using Ion milling W. Gruenewald* * Leica Microsystems, 1170 Vienna, Austria Today ion milling is the most common method to prepare samples for Electron Microscopy. The method can be used to modify sample surfaces through ion polishing, cleaning or contrast enhancement. There is a wide range of parameter setting to optimize each process. Preparation artifacts can be reduced by using the optimum parameters for milling angle, ion energy and milling time. Surface sensitive methods like CL or EBSD need a perfect sample surface, free of any damage or contamination. Consequently, these methods need a sample preparation that meets this requirement. Many conventional preparation techniques deform the crystal lattice. Mechanical polishing of soft materials or a hard / soft material combination leads to smearing. Even a final ion polishing is not able to remove the induced damage completely. To overcome this problem we have used ion beam slope cutting. This is a method to prepare high quality sample surfaces with almost no limitation. Even brittle materials or very difficult material combinations can be easily prepared. The sample surface is partly protected by a sharp edged mask. The unprotected part of the sample will be milled by a high energy Ar ion beam. The result is a flat and polished surface almost free of any preparation artifacts. Normally ion beam slope cutting is carried out with one ion beam and an oscillated sample. We have used a unique sample preparation method. Instead of one ion beam and sample oscillation we use three ion beams hitting the sample from 3 different directions (Fig.1). This method provides a lot of advantages: The sample can be observed under perfect conditions while milling. The heat transfer between sample and stage is much better compared to that of an oscillated sample. Re-deposition and shadowing effects can be almost completely suppressed. We will show the advantages of ion milling over mechanical polishing. Several preparation examples (Fig. 2 and Fig. 3) show proof the surface quality. The examples include semiconductor materials, geological and metal samples.

FIG. 1. Unique triple ion beam technique for ion beam slope cutting

FIG. 2. Cross sections of a gold wire bond (left) and clay (right)

FIG. 3. Cross section of a Si / solder / Cu multilayer structure and the corresponding EBSD pattern of Cu and solder (right image) proving the surface quality.

Solder

Cu

Mask

Cross section Sample

Advanced Sample Preparation for Analytical Microscopy

N. Erdman and A.R. Campbell

JEOL USA Inc. 11 Dearborn Rd, Peabody, MA 01960

Examination of material cross sections often provides essential information about the crystal

structure, layer or film thicknesses, existence of voids or cracks, and other properties that might

impact performance and reliability. A variety of mechanical methods are available to prepare

specimen cross sections for scanning electron microscope (SEM) observation, particularly for

metallographic sample preparation. However, mechanical polishing presents several problems: a)

surface unevenness due to different hardness values of the composite constituents; b) abrasive

embedding into soft materials; c) strain, materials deformation; d) smearing of the fine structure. In 2006 JEOL introduced a new concept in specimen preparation apparatus, Cross Section Polisher

(CP), that utilizes a variable voltage broad argon ion beam that mills the specimen and eliminates

problems associated with the conventional methods of specimen cross sectioning for SEM (Fig. 1).

The region of interest to be polished is selected via a CCD camera, then a masking plate is placed

across the region and the milling is performed in unattended fashion. During ion beam milling, the

specimen stage can be rocked ±30 degrees, or continuously rotated 360 deg depending on the type of

application. This equipment is suitable for preparation of almost any type of material for SEM

observation.

We will show several examples of utilization of CP preparation for analysis of materials with SEM

and CL. Examples will include shale, a sedimentary rock comprised of various types of minerals

notoriously difficult for mechanical preparation, clays and organic deposits (Fig. 2 left); solar thin

films, composites, diamond/carbide (Fig. 2 right) interfaces and other materials. We will

demonstrate that CP is particularly effective in preparation of materials where strain effects can be

studied via EBSD and CL.

References

[1] N. Erdman, K. Ogura and A.R. Campbell, Microscopy and Microanalysis 2011

[2] N. Erdman, A.R. Campbell and S. Asahina, Microscopy Today, p.22, May 2006.

Fig. 1 JEOL Cross-section polisher (left) and the ion-beam preparation method (right).

Fig. 2 CL images of calcite grain in CP’d shale (left) and WC (right).

Revealing large area, artifact free cross sections using argon milling for cathodoluminescence studies D.J. Stowe,* M. Hassel-Shearer**, D. Erwin** and A. Pakzad** * Gatan U.K., 25 Nuffield Way, Abingdon, Oxon., OX14 1RL, U.K. ** Gatan Inc., 5794 W. Las Positas Blvd., Pleasanton, CA, 94588, U.S.A. It is often of great interest to reveal and interrogate hidden regions of specimens and be confident that the specimen preparation technique has had little or no impact on the properties bring investigated. This is particularly true for the cases of cathodoluminescence studies where the experimental results are sensitive to even small changes in the physical and/or electronic properties of the sample. Many commercially important devices consist of a number of critical interfaces joining together many materials with extremely different thermal, electrical and hardness characteristics. Therefore making cross sections while preserving the interfacial structure is extremely difficult using either FIB, cleaving or mechanical polishing. An alternative method will be demonstrated using a collimated Argon beam with a shielding blade to produce an undamaged interface suitable for high resolution SEM imaging and analytical characterization. Furthermore, this paper will describe results using this argon ion milling process for the preparation of large, artifact free cross sections along with cathodoluminescence results from a variety of materials including oil and gas shale, high brightness LEDs and a number of semiconductor films.

Quantitative Application of Cathodoluminescence (CL) to Natural Quartz with Application to Geothermometry William P. Leeman,* Colin M. MacRae,** Nick C. Wilson, ** Aaron Torpy,** Cin-Ty A. Lee,*** James J. Student,**** and Edward P. Vicenzi***** * Earth Science Division, National Science Foundation, Arlington, VA 22230 ** Microbeam Laboratory, CSIRO Process Science and Engineering, VIC 3169, Australia *** Dept. of Earth Science, Rice University, Houston, TX 77005 **** Dept. of Geology, Central Michigan University, Mount Pleasant, MI 48859 ***** Smithsonian Institution, Museum Conservation Inst., Suitland MD 20746 Recent experimental calibrations [1-2] of trace element solubilities in naturally occurring minerals as functions of temperature and pressure make it possible to investigate conditions of mineral formation. For example, Ti solubility in quartz ideally allows quantitative estimation of magmatic temperatures and/or pressures if temperature is known independently from other geothermometers. In practice, such applications are hindered by complex Ti-zonation in natural quartz as well as uncertainties in the activity of Ti in natural magmas. At natural concentration levels (ppm), characterization of Ti-zonation by cathodoluminescence (CL) provides details of internal compositional variability and structure but is generally qualitative whereas direct microbeam analysis is time-consuming and costly. To combine advantages of both approaches, we characterized a variety of natural and synthetic quartz crystals using both high-resolution SEM hyperspectral CL imaging (Fig. B) [using a JEOL 8500F EMPA with integrated grating spectrometer] and direct analysis via laser ablation inductively coupled plasma mass spectrometry (LA-ICPMS; using a ThermoFinnigan Element 2 mass spectrometer) for contents of Ti (Fig. A) and selected other elements. Hyperspectral data (250-990 nm) were first collected in 2 µm steps over freshly polished mineral surfaces. Spectral intensities were deconvolved into three principal contributions (1.93, 2.19, and 2.72 eV) of which the first two correspond to intrinsic quartz and non-bridging oxygen hole (NBOH), respectively, and the third correlates directly with Ti4+ content (Fig. C) and can be used to quantify spectral mapping of Ti content as well as derivative parameters such as temperature of quartz formation from rhyolitic magma (e.g., method of [1] assuming aTi = 1.0 in the melt phase). Well-characterized glass standards were used for analytical calibration and precision and accuracy (ca. ± 5%) verified by replication of data obtained by other methods and labs on the same sample (e.g., Fig. A) and by analysis of synthetic Ti-doped quartz grains previously analyzed by EMPA. Because many samples exhibit significant zoning or heterogeneity in Ti, spatial resolution of LA-ICPMS (typically with 30-50 µm spot sizes) can significantly limit analytical precision. Quantitative maps provide critical information on Ti variation in individual quartz grains. For example, in the images provided (Figs. A, B) it is readily seen that a low-Ti core (i.e., low-T) is overgrown by several zones of quartz having increasing Ti content (higher T) and an outer rim with intermediate Ti and T. Such patterns are common in rhyolitic volcanic rocks and reflect complexities encountered during magma storage prior to eruption, including injection of hotter magma into crustal magma reservoirs [3-6]. Applications of these data to understanding magmatic processes will be reviewed [7]. References [1] D.A. Wark, E.B. Watson, Contrib. Mineral. Petrol. 152 (2006) 743. [2] J.B. Thomas et al., Contrib. Mineral. Petrol. 160 (2010) 160. [3] D.A. Wark et al., Geology 35 (2007) 235. [4] J.J. Student et al., EOS, Trans. Amer. Geophys. Union 87 (2006) Abstract V23C-0619 [5] P. Shane et al., Contrib. Mineral. Petrol. 156 (2008) 397-411. [6] J. Vazquez et al. J. Volcanol. Geotherm. Res. 188 (2009) 186.

A. CL image of quartz showing core-to-rim Ti analysis transects (see inset) for ICPMS [RU=Rice Univ.; VT= Virginia Tech] and EMPA [RPI=Rennselaer Polytechnic Inst.]. B. Map of Ti content and estimated T C. CL intensity (2.7 eV) vs. RU based on calibration in (C) ICPMS Ti (ppm) [7] The authors thank Ilya Bindeman, Bill Bonnichsen, and Jay Thomas for providing some of the analyzed samples. This work was supported by CSIRO and research grants to C.-T.A. Lee. Leeman acknowledges Smithsonian Inst. for laboratory support and National Science Foundation for providing time to conduct this research.

CMB-AIIph10 Keweenawan

RU

VT

RPI

Comparative analyses

0

50

100

150

200

250

300

350

0 200 400 600 800 1000

Distance (µm)

EMPA (RPI)ICPMS (RU)ICPMS (VT)

Ti (ppm

290

0

T (°C) 890

675 50 1000 µm

y = 1.505x + 33.77

R 2 = 0.984

0

10

20

30

40

50

0 10

20

30

40

Ti/10

Int 2.7

Introduction to Scanning Ionoluminescence Microscopy T.M.W. Franklin* * Optoelectronics Research Centre, University of Southampton, Southampton, UK The recent installation of an ORION PLUS® [1] scanning helium ion microscope (SHIM) at the University of Southampton Nanofabrication Centre opens up new possibilities with scanning microscopy. The SHIM operates in a similar way to a conventional scanning electron microscope (SEM). However, the critical difference is the use of a beam of helium ions for imaging rather than electrons. The SHIM offers a 0.3nm resolution, superior material contrast and a higher secondary electron yield then a SEM. Equipping the SHIM with a Gatan cathodoluminecence system allows, for the first time, investigation into scanning ionoluminescence microscopy.

An overview of the current system will be presented; this will include the imaging and spectral acquisition capabilities of the system along with specifics of the SHIM. Particular attention will be given to the unique aspects of the SHIM in contrast to an SEM. A summary of results obtained from the system will be provided, investigations have looked at bulk semiconductors such as GaAs, quantum dots and garnets such as Ce:YAG. Defects in direct band gap semiconductors induced by collisions between the incident ion and atoms within the sample appear to cause rapid quenching of luminescence. In extreme cases no luminescence can be detected from the sample at all.

Much of the investigation has focused on developing the tool’s potential to image

immunofluorescence markers, with this in mind, organic fluorescent dyes such as Alexa Fluor® 488 and lanthanum based rare earth doped nanoparticles (LaPO4:Eu, LaF3:CeTb & LaPO4:CeTb) have been studied in depth, both on their own and in the case of Alexa Fluor® 488, conjugated to antibodies and used in real world biological samples. Initial results have shown great promise of using rare earth doped nanoparticles as immunofluorescence markers as they are resilient to beam damage that results in bleaching and have achieved ionoluminescent images with a resolution down to 20nm, Fig. 1. References [1] B.W. Ward, J.A. Notte, and N.P. Economou, Journal of Vacuum Science and Technology B. 24

(2006) 2871-2874. [2] With thanks to employees of Carl Zeiss and Gatan for installing the system and their continued

support

(a) (b) FIG. 1. Micrographs of LaF nanoparticles in; (a) secondary electron imaging; (b) ionoluminescence imaging.

Nanoscale Depth-Resolved Electronic Properties of PECVD SiO2 and SiOx Dielectrics for Device-Tolerant Electronics * E.J. Katz,* L.J. Brillson,** H. L. Hughes,*** K. B. Chung, *** G. Lucovsky * Department of Electrical and Computer Engineering, The Ohio State University, Columbus, OH 43210 USA ** Naval Research Laboratory, Washington, DC 20735 USA *** Department of Physics, North Carolina State University, Raleigh, NC 27695 USA

We have combined nanoscale depth-resolved cathodoluminescence spectroscopy (DRCLS) and spectroscopic ellipsometry (SE) to measure the energies and depth distribution of charge traps in plasma enhanced chemical vapor deposited (PECVD) SiO2, PECVD SiOx and SiO2/SiOx/SiO2 dielectrics for device-tolerant electronics. It is now known that suboxide layers within silicon-on-oxide (SOI) structures result in a-Si nanoclusters that can reduce the shift of the flatband voltage by trapping electrons and blocking radiolytic proton movement to the SiO2/Si interface after ionizing irradiation [1]. Previous studies based on Si+ ion implantation showed that radiolytic proton trapping occurs at network oxygen atoms having large Si-O-Si bond angles [1]. Such traps can be used to reduce the net positive charge during exposure to ionizing radiation that result in negative flatband voltage shifts and pathways for charge leakage. We first prepared ~200 nm PECVD SiO2 and SiOx films to determine the energies and distributions of defects inherent to both these samples, and any defects that may be unique – such as defects characteristic of a-Si nanoclusters in SiOx. Figure 1 displays the DRCLS measurements for PECVD SiO2 and SiOx. Both samples contain similar ~1.90 eV and ~2.80 eV peaks that are characteristic of the non-bonding oxygen hole center (NBOHC) and E’ center, respectively. The SiOx DRCLS shows a unique peak at 2.20 eV and a shoulder at 3.70 eV. The 2.20 eV peak represent Si-hexamer rings that are the building blocks for a-Si nanocluster growth, and the 3.70 eV shoulder can be deconvolved into a 3.62 eV and 3.92 eV peak. With the use of DRCLS and spectroscopic ellipsometry, we were able to determine that the 3.92 eV peak is a trap located 3.92 eV below the conduction band of SiO2. From figure 2, the 3.92 eV trap is located near the midgap of SiO2 and should be a strong electron and hole trap; hence we have associated it with the a-Si nanocluster electron trap. We then combined PECVD SiO2, PECVD SiOx and thermal SiO2 to create an ultra-thin 12 nm SiO2/ 40 nm SiOx/ 6 nm SiO2 dielectric stack. From our DRCLS results we observed that annealing causes a decrease in the ratio of SiOx related defects in relation to SiO2 defects. This decrease is an indication that for rapid thermal anneals of >900°C, a-Si nanoclusters phase segregate into nanocrystalline Si grains inside a new a-SiO2 matrix [2]. In addition to monitoring changes with annealing, we were able to create a highly-resolved defect depth profile for the dielectric stack. DRCLS spectra initially contain unwanted cathodoluminescence from lower energy back-scattered electrons. To create a more resolved spectra we first deconvolved lower energy backscattered electrons for our Monte Carlo Simulation, EB = 0.6 keV-1 keV, shown in figure 3a, and then out of the DRCLS data. Figure 3b shows a waterfall plot containing the deconvolved DRCLS data. Figure 3c is a plot of the intensity of the a-Si nanocluster related 3.90 eV peak and the NBOHC realted ~2.00 eV peak vs. depth. Increase in 2.00 eV and 2.15 eV with depth indicates more Si-hexamer ring formation, therefore, more Si-nanocluster growth. On the other hand, the 3.90 eV peak is maximum near the SiO2/SiOx top interface, and then unexpectedly decreases. This may be an indication of the strongest electron traps being near the top of the SiOx layer. In the future we plan to to extend our DRCLS measurements to measure defect densities and distributions in SiOx dielectrics deposited in various ways, such as high density plasma CVD. Furthermore, DRCLS can be used for HfO2 samples that have nitrogen compounds incorporated within them. Nitrogen may reduce interface SiOx formation and oxygen vacancies/interstitials, both of which can be monitored by our DRCLS techniques.

References [1] B.J. Mrstik, et. al., IEEE Trans. Nucl. Sci., 50, (2003) 1947. [2] B.J. Hinds, et. al., J. Vac. Sci. Technol. B, 16, (1998) 2171. [3] This research was supported by the Defense Threat Reduction Agency, through Grant No. HDTRA1- 08-1-0037 and MIPR 10-2197M, and the National Science Foundation, Grant No. ECCS-0823805.

Figure 1) Near IR-UV optical emission spectra for (a) 202 nm PECVD SiO2 and (b) 175 nm deposited PECVD SiOx.

vacuum continuum

SiO2 ~9 eV

~1.1 eV

3 eV

~4 eV

1.0 eV

Spectroscopic

Ellipsometry

4.8-4.95 eV

EC(Si)

EV(Si)

EV (SiO2)

EC (SiO2)

nano-Si

0.3 eV

DRCLS

3.9eV

vacuum continuum

SiO2 ~9 eV

~1.1 eV

3 eV

~4 eV

1.0 eV

Spectroscopic

Ellipsometry

4.8-4.95 eV

EC(Si)

EV(Si)

EV (SiO2)

EC (SiO2)

nano-Si

0.3 eV

SiO2 ~9 eV

~1.1 eV

3 eV

~4 eV

1.0 eV

Spectroscopic

Ellipsometry

4.8-4.95 eV

EC(Si)

EV(Si)

EV (SiO2)

EC (SiO2)

nano-Si

0.3 eV

DRCLS

3.9eV

Figure 2) Spectroscopic ellipsometry and DRCLS of PECVD SiOx confirm a trap state ~3.90 eV below the conduction band of SiO2. The location of the ~3.90 eV trap is an indication of the a-Si nanocluster electron trap, due to strong trapping expected near the SiO2 midgap. Figure 3) (a) Deconvolved Monte Carlo simulation for EB = 0.6 keV – 1 keV. (b) Waterfall plot showing the near-IR to UV optical emission spectra for SiO2/SiOx/SiO2 after deconvolution. (c) Peak intensity versus depth plot for the 3.90 eV and 2.00 eV peaks.

1 2 3 4 5 6 7

500

1000

1500

2000

2500

1 keV 1.5 keV 2 keV 2.5 keV 3 keV

Inte

nsity (A

.U.)

Energy (eV)

1.92 eV

2.91 eV4.21 eV

4.79 eV

202 nm Deposited SiO2 (HTO4)

3.56 eV

(a)

1 2 3 4 5 6 70

1000

2000

3000

4000

5000

1 keV 1.5 keV 2 keV 2.5 keV 3 keV

Inte

nsity (A

.U.)

Energy (eV)

2.20 eV

1.95 eV

2.83 eV

3.70 eV Shoulder

4.23 eV

175 nm Deposited Silicon Rich SiOx (HTO6)

(b)

1 2 3 4 5 6 70

1000

2000

3000

4000

Inte

nsity (A

.U.)

Energy (eV)

deconvolved 175 nm deposited Si-Rich SiO

x

3.62 eV

3.92 eV

Deconvolved 175 nm Deposited Silicon Rich SiOx

2 keV

0 5 10 15 20 25 30 35 400.000

0.001

0.002

0.003

0.004

0.005

SiOx

Inte

nsity (A

.U.)

Depth (nm)

.5kV new .6kV new .7kV new .8kV new .9kV new 1kV new

SiO2

Deconvolved Monte Carlo Spectra for 12 nm SiO2/40 nm SiO

x/6 nm SIO

2

1 2 3 4 5 6 70

2000

4000

6000

8000

10000

12000

14000

16000

Inte

nsity (A

.U.)

Energy (eV)

0.5kV 0.6kV 0.7kV 0.8kV 0.9kV 1.0kV

Waterfall plot of non-annealed and irradiated data

2.0 eV 2.15 eV

3.40 eV3.94 eV

4 6 8 10 12 14 16 18 20 220

1000

2000

3000

4000

5000

Peak Inte

nsity

Depth (nm)

3.90 eV peak intensity 2.00 eV peak intensity

3.90 eV and 2.00 eV Peak Intensities vs. Depth

(a) (b) (c)

Application of CL/ EBIC-SEM techniques for evaluation of crystallographic defects in thin films and nanostructured materials

S.I. Maximenko,* S. R. Messenger,** J. A. Freitas, Jr.,** P. B. Klein,** L. Mazeina,***** Y. N. Picard,**** R. L. Myers-Ward,** D. K. Gaskill,** C. R. Eddy, Jr.** C. D. Cress,** V. M. Bermudez,** S. M. Prokes,** P. G. Muzykov,*** T. S. Sudarshan*** and R. J. Walters** * Sotera Defense Solutions, Crofton, Maryland 21114 ** Naval Research Laboratory, Washington, DC 20375 *** Department of Electrical Engineering, University of South Carolina, Columbia, South Carolina 29208 **** Department of Materials Science and Engineering, Carnegie Mellon University, Pittsburgh, Pennsylvania 15213 ***** Technology Assessment and Transfer, Inc, Annapolis, Maryland 21401

For many years it has been known that crystallographic defects introduce energy levels which degrade the properties of semiconductors. Therefore, the identification and control of crystallographic defects are of particular importance. Among other characterization techniques, the cathodoluminescence (CL) and electron beam induced current (EBIC) modes of scanning electron microscopy (SEM) are very useful tools to study the opto-electronic properties of defects due to their high spatial resolution down to submicron level.

In the present work, we show the results of an investigation of the role of extended and point defects in the properties of epitaxial films and nanostructures. Particularly, to identify the different types of performance limiting stacking faults (SFs) in 4H-SiC epi-films, real-color CL imaging and spectroscopy was carried out. Based on the emission peak position at room temperature and the excitonic band gap (Egx) energies determined from spectra acquired at low temperature (~5K), several distinct types of SFs were found, as shown in Fig. 1(a). Time-resolved CL measurements reveal that, compared to defect-free regions, the effective carrier lifetimes are severely reduced by the presence of stacking faults and three-dimensional 3C-SiC inclusions (Figure 1(b)) [1]. To identify the effects of screw and edge dislocations on the transport properties (e.g. minority carrier diffusion length) of 4H-SiC epi-films, an EBIC technique was also applied. For this purpose, Ti/Ni Schottky contacts were deposited on the epitaxial layers. The diffusion length of the minority carriers at both defects was extracted by analyzing charge collection efficiencies at defect sites as a function of the beam accelerating voltage (Fig. 2(a)). For the first time, our results clearly showed that screw dislocations have shorter diffusion lengths, thereby demonstrating stronger recombination activity than due to edge dislocations [2].

We highlight the use of CL/EBIC techniques for local characterization of radiation-induced defects in InGaP2/GaAs/Ge multijunction solar cells for space application. Particularly, from analysis of excitonic emission from GaAs base region (Fig. 2(b)) and EBIC scans of cleaved samples subjected to varying irradiation levels it was directly shown that high energy electrons and protons can shift the position of GaAs p-n junction and degrade the minority carrier lifetime. The observed changes are attributed to generation of point defects serving as compensation defect centers [3]. In addition, we report on an application of the CL technique for the detection and identification of local changes of nanostructured material properties. Specifically, the effect of various Sn-doping levels on the properties of β-Ga2O3 nanowires (grown for gas sensing applications) determined by real-color CL imaging/spectroscopy in combination with electron backscatter diffraction (EBSD) and energy dispersive spectroscopy (EDS) will be presented [4].

References [1] S. I. Maximenko et al., Appl. Phys. Lett. 94 (2009) 092101. [2] S. I. Maximenko et al., J. Appl. Phys. 108 (2010) 013708. [3] S. I. Maximenko, et al., IEEE Trans. Nucl. Sci., 57 ( 2010) 3095. [4] S. I. Maximenko, et al., Nano Letters, 9 (2009) 3245. [5] Some parts of this research were supported by the Office of Naval Research and National

Research Council

(a) (b)

FIG. 1. (a) Room temperature CL real-color imaging of SFs and (b) CL decay of the near band edge luminescence intensity [~390 nm] showing reduction of the carrier lifetime on 4SF[4,4] site versus defect-free region.

(a) (b) FIG. 2. (a) Plot showing EBIC collection efficiency as a function of accelerating beam voltage computed for various diffusion lengths (solid lines) and acquired experimental points for screw/edge dislocation and defect free regions [2]. (b) Dependence of the exciton emission at 1.49 eV with displacement damage dose DDD for electron-and proton-irradiated samples captured at the center of GaAs base region (15-kV SEM beam) [3]. The data were normalized to the pre-irradiation values. The insert depicts low-temperature (8 K) CL spectroscopy emission spectra from GaAs middle cell before and after 0.4-MeV proton irradiation.

MAXIMENKO et al.: CL/EBIC-SEM TECHNIQUES FOR CHARACTERIZATION OF RADIATION EFFECTS IN MJ SOLAR CELLS 3099

Fig. 8. Cross-sectional real-color CL images of the cell irradiated by 0.4-MeVenergy protons, and corresponding cross-sectional SEM image of the structure.

Fig. 9. Dependence of the exciton emission at 1.49 eV with DDD for electron-and proton-irradiated samples captured at the center of GaAs base region (15-kVSEM beam). The data were normalized to the pre-irradiation values. The insertdepicts low-temperature (8 K) CL spectroscopy emission spectra from GaAsmiddle cell before and after 0.4-MeV proton irradiation.

produce a conductivity change. More research is needed to con-firm this. Note that these data qualitatively follow the maximumpower degradation data given in Table I.

The observed strong degradation of NBE CL emission is aresult of the generation of nonradiative recombination centersunder irradiation found previously by EBIC. The cathodolumi-nescence efficiency is defined in general as

(1)

where and are the radiative and nonradiative recombina-tion lifetimes, respectively. Since the total recombination life-time of material at low-level excitation conditions is deter-mined by , the characteristics of the CL NBEemission trend, following the functional form of (1), show ev-

idence that a strong degradation of the recombination lifetimeoccurs in the GaAs base region. In the presence of irradiation-in-duced defects, the luminescence efficiency can be expressedas [21]

(2)

where and are the initial nonradiative lifetime and life-time damage coefficients, respectively. Performing a fit to thedata in Fig. 9 (e.g., CL intensity versus DDD) and using thevalue of s for GaAs [20], the lifetime damagecoefficient is determined to be 8.2 10 g/s/MeV, which is ingood agreement with the analysis of GaAs electroluminescencein irradiated InGaP /GaAs/Ge solar cells [21].

IV. CONCLUSION

To supplement the usual – and QE solar cell character-ization measurements needed for space qualification, EBICand CL measurements have been performed on monolithicallygrown InGaP /GaAs/Ge multijunction solar cells. These mea-surements represent the first reported data performed directlyon the monolithic multijunction structure. The QE and –data did show interesting behavior at large DDD levels, butthe damage mechanisms could not be readily established.Using EBIC and CL measurements, it was directly shown thatelectrons and protons with high DDD values can act to shiftthe location of the GaAs p–n junction toward the Ge subcell.This phenomenon is consistent with lifetime degradation andcarrier removal mechanisms in which high concentrationsof compensation defect centers are introduced, which act totype-convert the p-type GaAs subcell base region to n-type,thereby changing the physical structure of the GaAs subcellfrom an n /p/p to an n /n/p structure. The CL intensitydegradation of the exciton and donor-to-acceptor emissionprocesses also supports this claim. For the largest DDD value(0.4-MeV proton irradiation), the donor-to-acceptor emissionintensity exceeds the exciton emission intensity. The CL inten-sity degradation data (1.49-eV peak) also confirm that the GaAssubcell is dominant in the overall degradation in the 3J device.

It has been shown that EBIC and CL techniques can signifi-cantly add to the understanding of the different damage mech-anisms involved in solar cell degradation. Even though thesemeasurements did provide a lot of useful information, the lim-ited number of samples (corresponding to the final fluence foreach radiation particle) used in this study prohibits a detailedunderstanding. Future experiments would warrant a fluence de-pendence for each irradiating particle, thereby enabling a moredynamic understanding of the damage mechanisms. These datawould also support the development of a physics-based modelto describe solar cell degradation in a space environment.

REFERENCES

[1] S. R. Messenger et al., “Modeling solar cell degradation in space:A comparison of the NRL displacement damage dose and the JPLequivalent fluence approaches,” Prog. Photovolt., Res. Appl., vol. 9,pp. 103–121, 2001.

[2] S. R. Messenger, E. A. Burke, G. P. Summers, and R. J. Walters,“Application of displacement damage analysis to low-energy protonson silicon devices,” IEEE Trans. Nucl. Sci., vol. 49, no. 6, pt. 1, pp.2690–2694, Dec. 2002.

Cathodoluminescence Spectroscopy of Nanodiamonds in Experiment: An Implication for the Astromineralogical Study of the NGC 7027 Nebula A. Gucsik,* H. Nishido,** K. Ninagawa,*** M. Kayama,** A. Tsuchiyama,* and I. Simonia**** * Graduate School of Science, Osaka University, 1-1 Machikaneyama, Toyonaka, Osaka 560-0043, Japan ** Research Institute of Natural Sciences, Okayama University of Science, 1-1 Ridai-cho, Okayama 700-0005, Japan *** Department of Applied Physics, Okayama University of Science, 1-1 Ridai-cho, Okayama 700-0005, Japan **** School of Graduate Studies of Ilia State University, Tbilisi, Georgia Introduction: Circumstellar dust of the NGC 7027 nebula undergoing permanent influence of electromagnetic and corpuscular radiation from the central star can carry information on the origin and evolution of the present system [1-and references therein]. We have conducted experiments on excitation and registration of cathodoluminescence spectroscopical properties of nanodiamond particles obtained at liquid nitrogene temperature (LNT). The main goal of this study is to understand the formation conditions of the diamond particles, especially in the planetary nebulae. Samples and Experimental Procedure: K2 (Ultradispersed Detonation Diamonds-UDD) nanodiamonds were mounted in the non-radiative epoxy material (Fig. 1A). Their size dimensions were grouped into the following groups: less than 3 nm (Diamond: A/1 and A/2), between 3-7 nm (Diamond: B/1 and B/2), bigger than 7 nm (Diamond: C/1 and C/2), and a mixture (Diamond: D/1 and D/2) all of them. CL measurements were obtained on carbon-coated, polished (by silicon colloids) thin sections. Color CL imaging was performed on a CL microscope. The system was commonly operated at 15 kV accelerating voltage and a beam current of 0.5 mA. High-resolution CL images and spectra were acquired using a scanning electron microscope (SEM at Okayama University of Science, Okayama, Japan) equipped with an grating-type monochromator, SEM-CL system (Figs. 1B and C). This system was operated at 15 kV accelerating voltage and a probe current of 1.5 nA. CL spectra were recorded in the wavelength range of 350-800 nm with 1 nm spectral resolution and a dwell time of 1 second per step by photon counting. A) B) C)

FIG. 1. The UDD-nanodiamond particles were mounted in the two-component epoxy (A), SEM-CL data were obtained at LNT temperatures (B) and the analyzing area was selected from the CL image (C).

0

2000

4000

6000

8000

10000

12000

350 400 450 500 550 600 650 700 750 800

Wavelength (nm)

Inte

nsit

y

Diamond/A1Diamond/A2Diamond/B1Diamond/B2Diamond/C1Diamond/C2Diamond/D1Diamond/D2

378-388- 392

N-center

450-452- 454

A-center

Results and Discussion: Cathodoluminescence spectral features of all K2 samples show two broad bands centered at around 388 (3.1 eV; A-center) and 452 (2.69 eV; N-center) nm (Fig. 2). According to Pratesi [2] in an agglomerated state: such agglomerates are called "A Centers" when occur as pairs of nitrogen atoms (type IaA), "B centers" when occur as four nitrogen atoms surrounding a common vacancy (type IaB) and mixtures of them may also occur. Type Ib – mostly represented in synthetic diamonds but very rare (about 0.1%) among natural diamonds – contain nitrogen as isolated single nitrogen atoms called "C Centers". Sometimes type I diamonds may also contain clusters of three nitrogen atoms called "N3 Centers". FIG. 2. Cathodoluminescence spectra of the UDD-nanodiamond particles obtained at liquid nitrogene temperature show two significant broad emission centers at 388 nm (N-center) and 452 nm (A-center).

Conclusion: According to this preliminary laboratory astrophysical experiment, diamond particles in nebula NGC7027 may be originated due to ejection of the outer parts of the Red Giants during planetary nebula formation. Further comparative studies will be done using meteoritic nanodiamonds samples (e.g., yielded from Allende meteorite). Acknowledgement: AG was supported by the Japan Society for Promotion of Science (JSPS) Long-Term Visiting Professorship and Max Planck Society (Max Planck Institute for Chemistry, Mainz, Germany) References [1] I. A. Simonia and K. M. Mikailov Astronomy Reports, 50 (2006) 960-964.

[2] G. Pratesi, In: A. Gucsik (ed) Cathodoluminescence and its application in the Planetary Sciences, Springe, pp. 160. (2009)