Characterization of RadiationCharacterization of Radiation Damage Using X-Ray Diffraction

I. C. NoyanColumbia University, Department of Applied Physics and Applied Mathematics

Thanks

Conal E. Murray, IBM T. J. Watson Research Laboratory;

Sean Polvino, Columbia University, APAM;

O K l iOzgur Kalenci,

Andrew Ying,

Zhonghou Cai, APS, 2 ID-D.

Diffraction techniques are very sensitive to the material symmetry within a probethe material symmetry within a probe volume.

If the material is truly perfect, and large, we can use dynamical diffraction theory to model scattered beams.

Imperfections or small material volumesImperfections, or small material volumes, change the mode of scattering to kinematicalkinematical.

Since radiation can cause local damage within the material the scattering signaturewithin the material, the scattering signature should change with exposure.

Thus, x-ray or neutron diffraction techniques should be able to yield “in-situ” data on the evolution of lattice damage.

This talk will illustrate an actual real timeThis talk will illustrate an actual real time measurement using “x-rays” to cause the damage and also to detect itdamage and, also, to detect it.

SynopsisStructural damage is observed in silicon-on-insulator (SOI) and SiGe on SOI samples illuminated with(SOI) and SiGe on SOI samples illuminated with monochromatic (11.2 keV) x-ray microbeams approximately 250 nm in diameter.

The x-ray diffraction peaks from the irradiated layers degrade irreversibly with time, indicating permanent g y , g pstructural damage to the crystal lattice.

We do not see any such damage on bulk samplesWe do not see any such damage on bulk samples.

The damage mechanism is not fully understood at this timetime.

Outline

Basic Diffraction Theory: yWhat do we expect to measure?

Experiments/Results: What we did measure?

Conclusions.

Basic Kinematical FormulationK

Construct the diffracted

K0Kh

wave from the coherently scattered waves from all crystal atoms; assume:crystal atoms; assume:

Photoelectric absorption only,No multiple scattering,

rr

p g,Negligible energy taken by the reflected beam.

•X rays are really scattered by electrons: write the electron•X-rays are really scattered by electrons: write the electron density distribution in the crystal as:

)()()( ryrr rrr = ρρ

)(rrρ

)()()( ryrr ∞= ρρ

El t d it di t ib ti f t i l i di)(r∞ρ : Electron density distribution of a triply-periodicinfinite medium.

1 r∑ ).2exp(v)( 1- rhiFrhkl

hklrrr ∑ −=∞ πρ

Fhkl: the structure factor (diffraction from a unit cell).

:)(ry renvelope function of the sample. .

outside0inside1

)(⎩⎨⎧

=ry vv: unit cell volume.

outside0⎩

The diffracted wave amplitude, , can be written as:

,)](2exp[)()( 3∫∫∫ ⋅= rdrkirk vvvvvπρ

,)(v)( 1- ∑ −=hkl

hkl hkYFkvvv

hkl

,])(2exp[)()( 3∫∫∫ ⋅−=− rdrhkiryhkY vvvvvvvπ

vvvK0, Kh: the incident and diffracted wave vectors, respectively.

,0KKk h

vvv−= is the wave vector.

)( hkYvv

Fourier transform of sample shape (the shape),( hkY − Fourier transform of sample shape (the shape function.)

])(2exp[)()( 3∫∫∫ ⋅−=− rdrhkiryhkY vvvvvvvπ ,])(2exp[)()( ∫∫∫ ⋅−=− rdrhkiryhkY π

Crystal size is infinite in 3-d:

thenhkhkYrrr

vvvv−=− :),()( δ

hKKd =− 0 for no-zero amplitude. -> Bragg’s law.

F lli fil f fi i hi k “ ”

∫ ∫ ∫∞ ∞

⋅−=−2/

])(2exp[)(t

dzrhkidydxhkY vvvvvπ

For a crystalline film of finite thickness “t”:

∫ ∫ ∫− ∞− ∞2/

])(p[)(t

zy

))(sin()( thkVhkYvr

vr −≈

π ,)(

)(thk

VhkY vr−

≈−π

Kinematic Diffraction Peak Profile from a thin film:

θ ,)(

)(sin)2( 2

222

ΘΘ

Δ hkl ttFNI

≈The expected peak shape of the SOI layer is given by the

f th li d

( )θθπ 2cos

)(

ΔΘ Bk =

square of the normalized sinc function.

0≠2=∞→ θΔforAt 0,

T λ)cos( Bt

Tθ

=

T IT 39.122

m== θθπ Δcosk tforII Max

FWHM λ885.0≅

BtFWHM

θcos≅

SamplesAll SOI substrates were wafer-bonded; the SOI layer and underlying Si

SiNSiN[001]Sub

[001]SOI

layer and underlying Si substrate possess an offset of ~0.40o.

SOISOI

We looked at:

Bare SOIBOXBOX

Bare SOI

SOI/ SiN Si substrate0.1 μm Si substrate0.1 μm

SOI/SiGe

BulkSi/SiGe

Bulk Si

SiNSiN[001]Sub

[001]SOI

SiN

SOI

SiN

SOISOISOI

BOXBOX

Si substrate0.1 μm Si substrate0.1 μm

NSLS X20 Macrobeam Measurements

Si 004

8.05 keV (λ = 1.54 Å) ( )

~109 photons/sec

1 x 2 mm2 beam size

“t” from three techniques (nm)

TEM: 142 (1)

FWHM->143 (1)

The theory works fine for SOI samples

RADS->142.23 (0.02)

The theory works fine for SOI samples.J. Appl. Cryst. (2009). 42, 401–410 , Ying, et. al.

106 Scan 1S 11

The diffraction profile is invariant over 70

104

105

10

ps)

Scan 11 Scan 21 Scan 31Scan 41

over 70 measurements, two and a half d

102

103

10ns

ity (c

p Scan 41 Scan 51 Scan 61

days.

No radiation

100

101

10

Inte

n

damage at X20.

63 64 65 6610-1

10

2θ/θ (D )2θ/θ (Deg)

Microbeam measurementsAPS 2ID-D: Zone plate optics (~ 0.3 μm FWHM)

E = 11.2 keV (1.107 Å)

Slit optics on the diffracted beam path (widths of 100 µm and 10 mm), Measure SiGe (008) and Si (008) intensity (107o and 109o 2θB)Simultaneous measurement of diffraction and Ge Kα fluorescenceSimultaneous measurement of diffraction and Ge Kα fluorescence,The beam intensity can be attenuated by inserting Al foils in the beam path.

Bicron detectorBicron detector

Fluorescence detector(out of diffraction plane)

Attenuator box

zone plate

(Six circle Newport Kappa)

1010 photons/seczone plate ~1010 photons/sec

I id t f d b f ti f tt t thi kIncident focused beam as a function of attenuator thickness

1.73*10^9 Cps1.2 x 108 Grays

Calculated from XOP XPOWER pkg.Polvino, et. al. APL, 92, 224105, 2008

Repeated Diffraction ExperimentstSi3N4 ~ 105 nm

tSOI ~140nmtSOI 140nm

tBOX 140 nm z

Sample is irradiated with unattenuated beam for 10 minute intervals

After each irradiation attenuators are inserted (8 layers Al foil), and the sample is measured in an x-z topographic mesh around the area of irradiation and a θ/2θ scan at the center of irradiation.

z (μm)z (μm)

x (μm)

Variation of the 008Variation of the 008 diffraction peak profile from the SOI layer with exposure t di t bto direct beam .

The inset shows the profile at zero exposure with theexposure with the intensity in log scale.

The thickness fringes indicate afringes indicate a perfect Si layer.

These fringes disappear withdisappear with irradiation.

The initial diffracting layerThe initial diffracting layer thickness, calculated from the peak-breadth using the Scherrer formula, yields a value very close , y yto the physical thickness of the SOI layer.

The thickness after 60 minutes ofThe thickness after 60 minutes of irradiation, calculated from the broadened peak profile, drops to approximately half of its initial pp yvalue.

The integrated intensity of the broadened peak at this point isbroadened peak at this point is approximately 1/3rd of its initial value.

After irradiation, there is less crystalline Si in the irradiated volume.

12000

ty (c

ount

s)

8000

grat

ed In

tens

it

-1804 -1800 -1796

4000

Inte

g

( )x (μm)

Si 008 intensity map of the radiation-damaged region . In this figure x and z

Variation of integrated intensity across the minor diameter. The

coordinates define the sample surface. red line is a Gaussian fit to the damaged region, centered at x=-1799.5 μm, with FWHM of 1 78 μmDamaged region is ~6x larger than the beam. 1.78 μm.

The damage modifies the elastic strain field set-up pdue to the nitride overlayer.

•The 2θ peak valuesThe 2θB peak values change to lower values on either side of the damage region.

X (μm)

damage region.

•There are two intensity maxima bracketing both sides of the peaksides of the peak.

•Two new boundaries have been created. Radial scans as a function of x

2θ/θ

Radial scans as a function of x position (minor diameter)

Other SamplesWe observed similar damage when we i di t d SOI i 10 minirradiated an SOI region with no SiN cover, and one with epitaxial SiGe

10 min

one with epitaxial SiGe.

We able to locate the d d i d 60 mindamaged regions a day after the initial scans were run

60 min

were run.

There was no change in intensities or dimensionsintensities or dimensions.

The growth rates of the damaged regions were ~ parabolic.g g g p

Growth saturated after ~ 60 minutes.

4

DMajor (with Si3N4) DMinor (with Si3N4) DMajor (no overlayer) DMinor (no overlayer)

2

3

er (μ

m)

1diam

ete

0 10 20 30 40 50 60

0

time (min)time (min)

Samples Investigated

tSubstrate tBOX tSOI Capping Layer(µm) (nm)

Sample A 700 140 25 nm 55nm Heteroepitaxial SiGeSiGe

Sample B 700 140 140 nm 100 nm Tensile Si3N4Film

Sample C 700 140 140 nm 100 nm Compressive Si3N4 Film

Sample D 700 N/A N/A 80 nm HeteroepitaxialSi0.85Ge0.15

The control sample shows no degradation after 9 hr of direct beamdirect beam.

]

10000

tens

ity [c

ps]

Ge

(008

) Int Scan 305

Scan 315

106 0 106 5 107 0 107 5 108 0 108 5

1000

SiG

106.0 106.5 107.0 107.5 108.0 108.5

2θ [deg]

•Sample D: 80 nm Heteroepitaxial Si0.85Ge0.15 on 700 μm Si.•We observed no damage on uncoated Si wafers either.

SiThe crystal lattice of the SOI layer suffers damage as a

Bulk Si

SiO2layer suffers damage as a function of irradiation:

• Thickness fringes Bulk Sigdisappear;

• Peak broadens

• Integrated intensity drops.

These changes occur irrespective of the presence of airrespective of the presence of a capping layer.

The damaged region is stable.

We could not resolve the damage with AFM. SEM or IR microscopymicroscopy.

Conclusions

X-ray diffraction techniques are excellent tools for detecting damagetools for detecting damage.

These tools are non-contact and “~non-destructive” and can be very useful for the study of radiation damage evolutionbe very useful for the study of radiation damage evolution.

Diffraction data analysis involves solving an i bl D M d linverse problem: Data → Model parameters

existence, uniqueness, stability of the solution

Defining a suitable “test” sample where the problem is constrained might be usefulproblem is constrained might be useful.

Integrated modeling tools are needed to understand the damage mechanismunderstand the damage mechanism.

A rigorous combination of scattering theory and multi-scale damage modeling might yield exciting results.

ACKNOWLEDGEMENTSWe would like to thank Adrian Chitu of Columbia University for his help with performing microscopy on the samples,

Drs Katherine Saenger and Guy Cohen of IBM Research Division for manufacturing the samples for this study.

Use of the Advanced Photon Source was supported by the U. S. DepartmentUse of the Advanced Photon Source was supported by the U. S. Department of Energy, Office of Science, Office of Basic Energy Sciences, under Contract No. DE-AC02-06CH11357.

Use of the National Synchrotron Light Source Brookhaven NationalUse of the National Synchrotron Light Source, Brookhaven National Laboratory, was supported by the U.S. Department of Energy, Division of Materials Sciences and Division of Chemical Sciences under Contract No. DE-AC02-98CH10886DE AC02 98CH10886.