B I O G R A P H YVlad Stolojan obtainedhis BSc in Physics fromthe University of EastAnglia (1996) and hisPhD in physics from theUniversity of Cam-bridge. After working atthe University of Cambridge on electrontomography, he moved to the University ofSurrey where he is currently an RCUKResearch Fellow working on the characteri-zation of nano-electronic devices.

A B S T R A C TA dedicated scanning transmission electronmicroscope is ideally coupled with energydispersive x-ray and electron energy lossspectroscopies to obtain information aboutthe chemical composition, morphology andelectronic structure on the nanoscale. Withseveral signals being available simultane-ously with the pass of a sub-nanometre-sized beam, this instrument can answerquestions from a broad range of researchareas, in a timely fashion. The user-friendli-ness of the instrument comes at almost nocost in performance, making it an idealmulti-tool in a teaching environment.

K E Y W O R D Sscanning transmission electron microscopy,energy-dispersive X-ray spectroscopy, elec-tron energy loss spectroscopy, Ronchigram,nanotubes, viruses, semiconductors

A C K N O W L E D G E M E N T SThis work was partially supported by theEPSRC.

A U T H O R D E TA I L SDr Vlad Stolojan Advanced Technology InstituteUniversity of SurreyGuildford GU2 7XH, UKTel: +44 (0)1483 689411Email: V.Stolojan@surrey.ac.uk

Microscopy and Analysis 22(4):15-18 (EU),2008

STEM AP P L I C A T I O N S

I N T R O D U C T I O NThe majority of current microscopical researchis concerned with issues on the nanoscale,from adhesive joints in the aerospace industryto grain boundaries in metals and ceramics,and from almost all aspects of nanotechnol-ogy to plasmonics and biology. Electronmicroscopy is able to provide structural, chem-ical and physical information on that scalethrough the variety of instruments and associ-ated techniques available.

Scanning transmission electron microscopycombines some of the best characteristics ofscanning electron microscopy and transmis-sion electron microscopy. Similar to an SEM,the STEM forms a small probe which it thenrasters across a sample to acquire spatial infor-mation from surface-sensitive secondary elec-trons. However, the accelerating voltages arethose of the TEM, so that various signals can berecorded in transmission, using the electronsthat have passed through and interacted withthe sample. Thus one can form SEM-likeimages with secondary electrons, to gain infor-mation about the surface of the sample, aswell as TEM-like brightfield images (i.e. imagesformed with a parallel beam and an objectiveaperture placed in the back focal plane of theobjective lens).

However, the STEM is better than a simplesum of parts through a mode of imagingcalled annular darkfield (ADF) (Figure 1). Thisrefers to images formed with those electronsthat have interacted with and traversed thesample and have been recorded on a ‘dough-

nut-shaped detector. When the scatteringangles are high enough (typically semi-anglesfrom ~80 mrad to ~260 mrad) the image con-trast is incoherent and is much more easilyinterpretable than that in brightfield TEMimages, as the intensity depends linearly onthe density, the thickness and the square ofthe atomic number of the specimen. Oneimportant thing to note here is that all threeimaging modes (brightfield, annular-dark-field and secondary electron) can be acquiredat the same time, within the same scan, thelimitation being the number of video cardsavailable in the instrument (usually two).

The imaging information can also be furthercombined with spectroscopic informationfrom energy-dispersive X-ray spectroscopy(EDX) and electron energy loss spectroscopy(EELS). EDX looks at the energies of the X-raysgenerated by the incident electrons interact-ing inelastically with the sample, whilst EELSmeasures the energy lost by an incident elec-tron through inelastic scattering. EDX offerschemical composition information for ele-ments with medium to high atomic number,whilst EELS offers chemical composition forlight elements and some of the transition met-als, as well as electronic structure information,such as that related to collective excitations(e.g. plasmons), and joint and local density-of-states.

The spatial resolution of these techniques isrelated to the size of the electron probe itself(typically 2.5 to 10 Å for an instrument withoutspherical aberration correction), as well as the

Scanning Transmission Electron Microscopy:A Tool for Biology and Materials Science Vlad Stolojan, Advanced Technology Institute, University of Surrey, Guildford, UK

Figure 1: Schematic diagram of the signals available inthe scanning transmission electron micro-scope. Secondary electrons (SE) and X-raysgenerated by the scanned probe are collectedat detectors above the sample, whilst theannular darkfield (ADF or HAADF, dependingon the angle range) and the brightfield (BF)images are collected below the sample. EELS,diffraction and the brightfield signal are mutu-ally exclusive, as they involve the same scat-tered electrons.

MICROSCOPY AND ANALYSIS JULY 2008 15

size of the excitation event (the excitation vol-umes are ~100 Å when exciting a bulk plas-mon, <0.5 Å when exciting a single atom, and~10 to 30 Å for X-rays). When a STEM isequipped with EDX and EELS it is an extremelypowerful and versatile tool and this isreflected in the quality of results with andwithout spherical aberration correctors [1-6].

S T E M S P E C I F I C AT I O N SIn a broad research community like that withinan university, an investment in an expensiveinstrument such as a transmission electronmicroscope has two major requirements: thefirst is that it provides competitive informationacross a broad range of fields of study, and thesecond is that, as an instrument in a teachingenvironment, its user-friendliness allows stu-dents to divert more attention to the science(i.e. their sample) within the instrument andthe science behind the instrument, so that itbecomes a true multi-user microscope. TheSTEM is a natural choice for users that want tocombine classical electron microscopy withanalytical spectroscopic studies on thenanoscale.

At the University of Surrey, this is role is per-formed by a Hitachi HD2300A STEM with aSchottky field emitter, which can be operatedat accelerating voltages from 120 to 200 kV.The instrument is fitted with a Gatan Enfina

EEL spectrometer and an EDAX EDX spectrom-eter (Figure 2). The instrument has a pair ofcondenser lenses for forming the probe, anobjective lens for focusing and a projector lensfor changing the solid angles into the variousdetectors in normal imaging mode and forchanging the camera length in diffractionmode. The instrument has four main operat-ing modes, effectively equivalent to four dif-ferent probe sizes, from 10 Å down to 2.5 Åcalled EDX, Normal, High Resolution and UltraHigh Resolution.

There are also two extra modes: one is dif-fraction and the other is DECON(tamination).In the DECON mode, a broad beam is rasteredacross a region of interest of the sample toburn off any volatile carbon radicals presenton the surface of the sample, which couldharden into a carbon deposit as the electronsare formed into a very small and intense spot.

One important difference from TEMs is thatin a STEM the magnification is realizedthrough the way the sample is scanned, muchlike an SEM, as opposed to using several pro-jector lenses. This has the great advantagethat, once an operating mode is selected, thelens currents stay virtually constant. The disad-vantage is that, at medium to low magnifica-tions (<100,0003), the electron beam isscanned sufficiently far from the centre of theobjective lens that is is affected by the spheri-

cal aberration of the lens and this also affectsthe collection of energy loss spectra whilst thebeam is scanned.

One of the most useful features for theoperation of a STEM is the Ronchigram, whichis the shadow image of the stationary probegoing through the sample and projected ontoa CCD camera. The Ronchigram makes align-ment, stigmation and focusing intuitive, aswell as making the measurement of aberra-tions relatively straightforward [7]. It is no sur-prise that the Ronchigram underpins the suc-cess of the Nion spherical aberration corrector[8] and the Nion SuperSTEM II [9]. It is also thebasis of the recently proposed ‘apertureless’microscopy [10]. Furthermore, it is an excellentway to teach new users about electron micro-scopes, as it is a true ‘one-stop shop’ for opti-mal alignment [7]

A P P L I C AT I O N S I N B I O L O G Y Biological specimens are a particular casewhere the STEM can provide essentially unri-valled information through the mass-thicknessand Z2 contrast present in high-angle annulardarkfield (HAADF) images. The image of aDNA molecule supported on a carbon filmrevealed the power of this instrument as longago as 1971 [11]. Beam damage is reduced sig-nificantly as a result of rapidly scanning theintense probe across the sample, thus reducing

Figure 2: (a) The Hitachi HD2300A STEM atthe University of Surrey. It is fittedwith an EDAX EDX spectrometer andan Gatan Enfina EELS spectrometer.(b) Detail of the control console,showing two LCDs: the lefthand-sideone is for controlling the microscopeand the righthand-side one is usedinterchangeably between the twospectrometers. The operation of theSTEM is similar to that of a HitachiSEM, with an extremely simple ‘but-tons’ box and a separate stage con-trol unit.

MICROSCOPY AND ANALYSIS JULY 200816

Figure 3: (a) HAADF image of reovirus parti-cles within a mouse melanoma cell. The inset shows a typical reovirus,with its characteristic hexagonalbody. (b) Equivalent brightfield image ofthe cluster of the reo viruses. Thecontrast in this image is affected byinterfence (coherent contrast),which makes interpretation moredifficult. Sample courtesy of Dr Lucy Heine-mann, University of Surrey.

STEM AP P L I C A T I O N S

the overall damage. The Ronchigram allowsone to focus on the sample very quickly andeasily, even during the acquisition of animage.

Figure 3a shows a HAADF image of a cancercell infected with reo viruses and Figure 3b itsequivalent brightfield image, one that wouldbe obtained using a TEM (for comparison pur-poses, we have used the projector lens in ourSTEM to increase the acceptance angle intothe brightfield detector, so as to reproduceTEM conditions as closely as possible). It isworth noting that these images were col-lected under what are very harsh conditionsfor biological samples: room-temperaturespecimen holder and 200 kV accelerating volt-age. Without the Ronchigram, these imageswould have probably been nearly impossibleto achieve. These images show the character-istic accumulation of icosahedral reovirus par-ticles within the cytoplasm of the cell, in whatis referred to as a viral inclusion or factory. Theinset in Figure 3a shows a close-up of an indi-vidual reovirus particle with its characteristichexagonal head.

A P P L I C AT I O N S I N N A N O -T E C H N O L O G Y The simultaneous availability of the SE, BF andHAADF/ADF signals is advantageous whenstudying carbon nanotubes, particularly whenthey are embedded in polymers (Figure 4).Due to the different secondary electron cross-sections for carbon nanotubes and polymers,the carbon nanotubes are readily visible in theSE image, compared to BF and ADF images, asthey are virtually the same material (i.e. car-bon). The simultaneous availability of thesesignals makes the imaging and the analysis ofembedded carbon nanotubes a more straight-forward exercise, with added informationabout the surface morphology and/or the con-tamination of these nanostructures.

The difference between incoherent andcoherent imaging can be seen in Figure 5,where 5a is the HAADF image of a WS2 nan-otube with lattice planes 6.2 Å apart and 5b isthe simultaneously acquired brightfieldimage. In the brightfield image (Figure 5b),the position of the W planes is unclear as theposition and contrast of the fringes (bright ordark) depends on the defocus, whilst the con-trast in the HAADF image is mainly ~Z2.

A P P L I C AT I O N S I N M AT E R I A L SS C I E N C EThe last example presented here is a detailedexamination of amorphous silicon implantedwith manganese using ion-beam implantation(5x1016 Mn ions cm-2 implanted at 50kV) andannealed, with the goal of forming ferromag-netic silicon.

Manganese can also form three variants ofthe silicide: non-magnetic SiMn, antiferro-magnetic Mn5Si3 and magnetic Mn4Si7. TheHAADF image in Figure 6a shows that theimplanted region is crystallized, with Si crys-tallites ~30 nm in diameter, with increasedintensity in the grain boundary region. Theincreased intensity at the grain boundariescould be caused by densification as well as an

Figure 4: Low magnification images of carbonnanotubes dispersed in the polymer p-phenylene vinylene. Secondary elec-trons are generated preferentially inthe carbon nanotubes. The thicknessof the sample affects significantly thebrightfield image (a), so only nan-otubes at the edge of the sample arevisible. The HAADF image (b) showsthat the thickness and density of thefilm are uniform, whilst the SE image(c) shows the location and distributionof carbon nanotubes within the poly-mer. Sample courtesy of Dr Ross Hatton,University of Surrey.

MICROSCOPY AND ANALYSIS JULY 2008 17

increase in the Mn concentration at theboundaries.

EDX mapping (not shown here) confirmsthat Mn does indeed migrate to grain bound-aries and is confined to the crystallized region,as predicted from the binary phase diagram.EELS mapping along the green line in Figure6a shows the relative distribution of Mn con-fined to the crystallized region, as extractedfrom the area under the Mn white line. TheMn white line represents the spectral featureassociated with energy lost by the incidentelectrons to Mn atomic transitions from the

spin-orbit split 2p energy level to empty statesin the 3d band (called ‘white lines’ becausethat is how they appeared on the original pho-tographic recordings of energy loss spectra).

Figure 6c shows a trace of the relative Mncomposition across a grain boundary, takenwith steps of 7 Å, showing a Mn-rich region of~5 nm width. This allows us to switch our STEMinto nanodiffraction mode and collect a dif-fraction pattern from a region ~3 nm at thegrain boundary (Figure 6d). Comparing thediffraction data with that from the man-ganese silicides [12-14], we conclude that the

MICROSCOPY AND ANALYSIS JULY 200818

Figure 6: (a) An HAADF image of an amorphous Si layer deposited on a Si substrate which has been ion-beam-implanted with Mn and annealed with a pulsedexcimer laser. Upon annealing, the amorphous Si crystallizes in grains ~30 nm in diameter and the Mn migrates to the grain boundaries. (b) EELS spectrum acquired at positions along the green line reveals that Mn is distributed in the crystallized layer. The plot has been oriented so thatthe spatial dimension corresponds to that in the image in panel (a). (c) A detailed profile collected across the boundary confirms that Mn decorates the grain boundaries, in a region ~6 nm thick. (d) This allows us to collect diffraction patterns with a small ~3nm diameter probe from the region and show that the Mn does not form the non-mag-netic MnSi. Sample courtesy of Dr Nianhua Penh, University of Surrey.

spots marked by the red circle in 6d are due to220 planes in the antiferromagnetic Mn5Si3.

C O N C L U S I O N SScanning transmission electron microscopy isthe ideal technique for probing the chemical,compositional and electronic structure infor-mation on the nanoscale with spectroscopicand imaging techniques. At the University ofSurrey, we have opted for a dedicated STEMinstrument with EDX and EEL spectrometersfor providing in-depth information on thenanoscale for a wide range of samples, char-acteristic of the broad research efforts usuallyassociated with an university: from biology tomaterials science, from nanostructures tointerfaces.

We have found the SEM-like user friendli-ness of our instrument particularly attractiveto student users, and the simultaneous avail-ability of several imaging and spectroscopicsignals very useful in answering questionsabout the science and processes related tosamples in a timely fashion. Although STEM asan instrument has always had a niche follow-ing, the quality of the results obtained in thelast two decades justify it as a stand-aloneinstrument.

R E F E R E N C E S1. Muller, D. A. et al. The electronic structure at the atomic

scale of ultrathin gate oxides. Nature 399:758-761, 1999.2. Muller, D. A. et al. Atomic-scale chemical imaging of

composition and bonding by aberration-correctedmicroscopy. Science 319:1073-1076, 2008.

3. Nellist, P. D. et al. Direct imaging of the atomicconfiguration of ultradispersed catalysts. Science 274:413-415, 1996.

4. Nellist, P. D. et al. Direct sub-angstrom imaging of a crystallattice. Science 305:1741-1741, 2004.

5. Li, Z. Y. et al. Three-dimensional atomic-scale structure ofsize-selected gold nanoclusters. Nature 451:46-48, 2008.

6. Nelayah, J. et al. Mapping surface plasmons on a singlemetallic nanoparticle. Nature Physics 3:348-353, 2007.

7. The Ronchgram: www.rodenburg.org/STEM/t200.html. 8. The spherical aberration corrector: www.nion.com/tech.html9. The Daresbury Laboratory SuperSTEM facility:

www.superstem.dl.ac.uk.10. Rodenburg, J. M. et al. Transmission microscopy without

Figure 5: (a) HAADF image of a WS2 nan-otube showing the position ofthe W concentric shells. Therepeat unit is -S-W-S-, with twolayers of S atoms between theW atoms. Here contrast is pro-portional to Z2 and the W layersare directly identified. (b) The equivalent brightfieldimage is more difficult to inter-pret due to the coherent con-trast effects. The location of theW layers cannot be directly iden-tified from this image as thecontrast of the fringes dependson the amount of defocus andcan be reversed (arrow). Samplecourtesy of Dr Jeremy Sloan,University of Surrey and QueenMary’s College London.

lenses for objects of unlimited size. Ultramicroscopy107:227-231, 2007.

11. Crewe, A. V. et al. Visibility of a single atom. Science 168:1338-1340, 1970.

12. Gottlieb, U. et al. Magnetic properties of single crystallineMn4 Si7. J. Alloys Compd. 361:13-18, 2003.

13. Jorgensen, J. E. and Rasmussen, S. E. Refinement of the

structure of MnSi by powder diffraction. Powder Diffraction6:194-195, 1991.

14. Brown, P. J. and Forsyth, J. B. Antiferromagnetism in Mn5 Si3:the magnetic structure of the A F2 phase at 70K. J. of Phys.:Cond. Matter 7:7619-7628, 1995.

©2008 John Wiley & Sons, Ltd