SEI Detector Everhart-Thornley detector Microscope chamber wall Faraday cage Scintillator Electrons in Light pipe Photomultiplier Electrical signal out Screen Quartz window +200 V +10 kv Always contains some back-scattered component Robinson detector bulky, restricts WD, retracted for EDS 100µm SEM Optics (v 1 + u 2 ) = constant diameter of filament = d 0 Strong C1 (small v 1 ) means large u 2 and small d. Apertures used to reduce spherical aberration i = i 0 (β/α 0 ) 2 Physically fixed distance β demagnified source image diameter d 1 at cross-over d 1 = d 0 x v 1 /u 1 Minimised for large α o (strong C1) or small β (small objective aperture) WD α ½A A tanα = 2WD ~ α final probe diameter d d = d 1 x WD/u 2 = d 0 v 1 WD u 1 u 2 Can decrease probe size by either decreasing v 1 and increasing u 2 (increase strength of CL or increase V) or decreasing WD (increase strength of OL or increase V).
SEM Performance 100 µm Pixels (picture elements) ~ 100 µm, Specimen pixel size, p = M Resolution can never be better than p Best to use a probe size, d, equal to p Probe size should be reduced as magnification increases. s = h α s = beam diameter defocus h = depth of field Assuming s p h = s µm α =100 Mα A We already know that α ~ tanα ~ 2WD WD α ½A A tanα = 2WD ~ α 100 µm So, h = M ( ) = 0.2WD 2WD A AM mm Can increase h by: 1. Increasing WD - Degrades resolution 2. Decreasing size of objective aperture May degrade resolution 3. Decreasing M Resolution Probe Size B = brightness = beam current density per unit solid angle [Am -2 sr -1 ] So, resolution is improved by: 1. Using a small λ (high kv - but probably increases E too) 2. Using a large aperture (increasing β) decreases h, increases aberration 3. Using a small probe current, i (may require longer exposure times) 4. Using a bright electron gun (typically FEG)** As β appears in all three terms as either numerator or denominator, its effect is ambiguous. Considering its very large effect on both r d and r s, the minimum probe diameter for best resolution (d 1 = 0) is: d opt = Kλ ¾ C s ¼ K ~ 1.22 C s ~ 20 mm d opt ~ 2.3 nm (20 kv) As d decreases so does i. According to the Pease-Nixon equation for thermionic emission: ( ) d = d it opt 7.92x10 9 + 1 j [ ] 3/8 i = probe current [A] T = temperature of filament [K] j = current density at filament surface [A/cm 2 ] (=Bπα 02 )
Resolution-Beam Current To resolve two points, there must be a visible difference in the signal generated from them. n = average number of electrons detected from a point, varies by n n 1 n 2 C = = n 1 n n 1 (n 1 > n 2 ) C = natural contrast (0 C 1) According to Rose, the eye can only distinguish between two adjacent points if n 1 > n 2 + 5 (signal 1 is greater than signal 2 by five times the noise in signal 1) n 1 Minimum observable contrast level: C > 5 n 1 n 1 = 5 n 1 n 1 > 25 C 2 i c > 4 x 10-12 qfc 2 n 0 = it if x 10 = -6 e e [A] n = qn 0 n 0 = number of electrons incident on a given pixel i = beam current t = time the beam stays at a given pixel F = entire frame scan time (assuming 1000x1000 pixels) n = number of electrons actually detected from a pixel q = product of detector efficiency and electron yield (0.1 < q < 0.2 for secondary electrons) i c = critical current needed to see a contrast level C Actual contrast increases as n increases by: 1. Increasing beam current 2. Increasing scan time Resolution = f(c) Resolution-Voltage photoresist on silicon
Topographical Images Kanter, 1961 δ = δ 0 cosθ δ = Secondary electron coefficient δ 0 = value of δ at normal incidence (θ = 0) Smallest for normal incidence (20-40 is common) Does not account for increased secondary electron emission caused by backscatter, which also increases with θ. Not valid for very low kv, in which case all secondary electrons escape δ Beam δ 0 0 20 40 60 80 Tilt (deg) R θ R 0 θ Stereomicroscopy Two images are taken, one tilted 10-15 with respect to the other. Both are viewed simultaneously with a stereo viewer or anaglyph glasses. p h = 2Msin(θ/2) h = vertical height difference p = relative lateral displacement M = magnification θ = parallax angle Also possible (but highly unusual) to get topographical information using a small backscattered detector. - Only line-of-sight electrons detected, so emphasises shadows. - Modern BSI detectors are annular, so the effect is cancelled.
BEI Detector Slow response time, so slow beam scan rates are necessary Electron beam Final probe-forming aperture Annular solid-state p-n diode creates cascade of electrons Electrical signal out Solid-state detector Specimen surface Backscattered electrons For good compositional contrast images, a polished specimen is required Compositional Images η 0 = -0.0254 + 0.016Z 1.86x10-4 Z 2 + 8.31x10-7 Z 3 η 0 = backscattered electron coefficient, untilted Z = atomic number Partly because of diffraction, η also varies with θ, but in a more complicated way. Several empirical equations exist to describe this relationship, one of which is: η = 0.89(η 0 /0.89) cosθ Contrast is typically weak and easily drowned out by topographical effects, so best practice is to use polished specimens. C = η 1 η 2 η 1 (η 1 > η 2 ) Weak contrast large I c large d (poorer resolution) Al Pt: C = 0.685, d = 3.6 nm α-brass β-brass: C = 0.003, d = 186 nm 10 µm Fraction of incident beam which is backscattered decreases as E 0 increases, so can improve contrast by reducing kv.
Channelling Paths of low atomic density - channels Electrons that penetrate deeply are less likely to escape from the surface once scattered Strong when Bragg condition is satisfied for just one set of planes. Much weaker than Z contrast: 0.0005 < C < 0.05, so requires: 1. High brightness (FEG) 2. Small semi-angle (α < 10 mrad) 3. BSE detector with large solid angle of collection 4. Clean, smooth sample surface 5. High probe current (I > 10 na) Optimum for low kv and large Z Used in metallurgical and geological samples to: detect small orientational changes defects (eg, dislocations) Channelling in Electrons out ECCI detector Sample http://ssd.phys.strath.ac.uk/index.php/electron_channeling_contrast_imaging Channelling Paths of low atomic density - channels FIB-induced channeling contrast SE image of steel grains Comparison of ECCI and SE image of a GaN thin film
EBSD Components include: Sample tilted 70 from horizontal Phosphor screen (like TEM) fluoresced by electrons to form diffraction pattern CCD camera for viewing pattern Peltier cooled CCD camera Phosphor screen Beam control Sample Phenomenon Electrons first inelastically scatter then are Bragg diffracted to form Kikuchi bands Applications: Measure crystal orientation/misorientation Discriminate between different materials Texture map grain morphology, orientations, boundaries Forward scatter electron detector Stage Control PC Digital stage and beam, control unit EBSD Width of Kikuchi bands = w = R ~ L2θ ~ Lλ d (Rd = Lλ) Ni: Fm3m Ni: Fm3m True symmetry of crystal 4-fold symmetry of [001] Planes with large d-spacings produce thinner Kikuchi bands
EBIC Electron Beam Induced Current Incident electron knocks out hundreds of thousands of electrons from their atoms, leaving positively-charged holes behind Bias voltage separates e - and h + to form current Image is formed with the current signal Contrast corresponds to differences in: Conductivity Electron-hole pair lifetime Electron-hole mobility Used to study semiconductors Imaging of perpendicular p-n junctions in a MOSFET transistor as a function of V (therefore depth) R.F. Egerton Charging Samples in the SEM must typically be conductive or be made conductive May need to sputter coat thin (~10 nm) layer of C, Au, or other conductive material to prevent charging Total electron yield 1 charging + charging Self-stabilises More electrons leaving the sample surface than are arriving on it More electrons arriving at sample surface than are leaving it charging Alternatives to coating: Operate at low kv (where yield ~ 1) Leak air into vacuum (charge is dissipated by electron-gas interactions) For some samples, must also consider: Degradation (beam damage) Radiation damage Volatility V 1 Accelerating Voltage V 2 ~1 kv ~5 kv
Cross section of coriander seed Environmental SEM (ESEM) Can be used to image wet, volatile, or non-conductive samples Everhart-Thornley SEI detector cannot be used, as high bias would cause breakdown in poor vacuum Gas-phase detectors are used instead for SEI ESEMs have three separate vacuum systems: 1. Gun (high vacuum) 2. Column (lower vacuum) 3. Sample Chamber(1 10 torr) (~10 2 10 3 Pa) Loss of resolution due to scattering of electrons off gas molecules BSE ESEM image of live Leptospermum flavesces stem cells with water film End of Lecture