Ch 09 - Metrology - Current IIT12

9-1

Chapter 9

Metrology Methods for Ion Implantation

Process Controls

Michael Current Current Scientific, San Jose, CA USA

[email protected]

New metrology tools have been developed for characterization of shallow (≈10 nm) and

deep (2 to 10 um) junctions as well as to map dose and annealing variations on test

wafers at sub-mm resolution. Well-established tools, in particular 4-point probes (4PP),

have been extended with the development of “elastic material” probe tips and MEMS-

based micro-4PP with micron scale probe spacing. Surface resistivity probes, coupled

with AFM scanning methods, now provide 2D device maps with nm scale resolution.

New physical methods include junction imaging with holographic-TEM and atom-by-

atom chemical analysis of device structures by field evaporation atom probes that provide

2D and 3D maps of IC devices. And new types of optical, based on reflectivity and

photoluminescence, and photo-voltage measurements provide non-contact tools

sensitivity to details of dopant and damage distributions as well as electrical

measurements of sheet resistance and carrier recombination leakage currents.

Chapter Outline

1. Metrology challenges for doping applications: 9-2

2. Electrical methods:

4-Point Probe (4PP) 9-3

micro-4-Point Probe (micro 4PP) 9-6

Active Surface Dopant levels (Nsurf) 9-10

Scanning Spreading Resistance Microscopy (SSRM). 9-11

3. Photo-Voltaic (PV) Methods:

Surface Charge 9-15

Carrier Spreading Sheet Resistance (RsL) 9-8

4. Optical methods:

Modulated Optical Reflectance (MOR) 9-24

Photo-Luminescence (PL) 9-27

5. 2D & 3D Physical Methods:

Holography (3D Image) with TEM (holo-TEM) 9-30

Atom Probe Tomography (APT) 9-32

6. Summary: 9-36

References: 9-37

9-2

1. Metrology Challenges for Doping Applications: The diverse requirements for metrology for ion implantation process have their roots in the wide

range of implant process conditions, spanning many decades in ion energy and dose (Fig. 1-1)

and the somewhat contradictory implications of the high levels of precision which have become

standard expectations for the placement and numbers of atoms implanted into a surface, often in

the range of 1%, with incident beam angle control at a fraction of a degree for all locations and

devices on a 300 mm wafer.

Fig. 1-1. Energy and dose regions for doping of CMOS and photovoltaic (PV) devices and

fabrication of SOI and other laminated materials.

During the last few years, a high level of innovation has been demonstrated in the range and

capabilities of metrology tools available for process development and production controls of ion

implantation and annealing. New methods have been developed for “classic” tools such as 4-

point probe (4PP) sheet resistance measurements, now available with micron-scale probes, and

“spreading resistance” techniques, now applied to 2-D analysis of CMOS doping and contacts.

Micro-probes developed for analysis of dielectric films on product wafer test pads have been

adapted to measure active surface dopant levels (Nsurf). Optical probes, such as modulated-

optical-reflectance (MOR), also known as “ThermaWave”, and photo-luminescence (PL), have

been enhanced by new understanding of fundamental interactions of light and doped

semiconductors and have been used in to provide new measurement capabilities for process

evaluation with micron-scale spatial resolution. Detailed understanding of surface and junction

photo-voltage effects have led to the development of various non-contact electrical probes for the

surface charge, sheet resistance and junction leakage currents capable of providing whole wafer

maps with pixel resolution of less than 1 mm. Physical analysis tools, such as secondary ion

mass spectroscopy (SIMS) and transmission electron microscopy (TEM) have been adapted to the

analysis of shallow (≈10 nm) doped layers and to provide 2-D images of local junction potentials,

such as with “holographic-TEM” imaging. Field-evaporation “atom probe” techniques have been

combined with focused ion beam sputtering and sample manipulation techniques developed for

TEM analysis to provide 3-D mapping of CMOS transistor chemical information on an atom-by-

atom basis. The use of these metrologies for implantation process is the topic of this chapter.

Ion Implantation Dose & Energy

0

1

10

100

1,000

10,000

1.E+10 1.E+11 1.E+12 1.E+13 1.E+14 1.E+15 1.E+16 1.E+17 1.E+18 1.E+19

Ion Dose (atoms/cm2)

Ato

m E

ne

rgy

(k

eV

)

Channel(Vth)

Steep Retrograde Channel (Vth)

Mid-well(Vpunch thru)

Deep-well(Latch up, SER)

Super-deep-well(CCDs)

H-cut(Laminatedmaterials)

Halo(Vth, SCE)

SD Extension(Vth, Ids)

S/D Contact

(Ids)

Poly-Si Gate(Vth, Ids)

H-cut(Si PV membranes)

SIMOX(SOI)

PV junctions

PV emitters

9-3

2. Electrical methods: Direct contact 4-poiint probes (4PP) have been standard tools for measuring the sheet resistance

of implanted and annealed junctions since the introduction of configuration switching techniques

in the early 1980’s. The 4PP technique was first developed for geological explorations, where a

linear array of 4 evenly spaced metal rods are used to measure the bulk resistivity of the local

ground (Fig. 2-1). When a linear array with rod spacing, s, is used where a current (I) is

conducted through the outer rods and a voltage (V) is measured between the 2 inner rods, the bulk

resistivity of the ground (assumed to be much thicker than the rod spacing) is:

(Ohm-cm) = 2*s *(V/I). [Eq. 1]

Figure 2-1. Linear 4 Point Probe measurements of sheet resistance.

For the case of ion implanted junctions, the junction depth, Xj, is usually much less than probe

spacing, so the current (in the absence of strong leakage current or probe penetration effects) is

conducted in rings through the thin junction layer. In this case, for a linear 4PP as shown in Fig.

2-1, the sheet resistance, Rs, of the junction is:

Rs(Ohms/square) = / Xj = (2/ ln(bc/ad))*(V/I) [Eq. 1-2]

where a, b, c and d are the distances indicated in Fig. 2-1. For a perfectly equally spaced linear

4PP array, b = c = 2a = 2d, so the sheet resistance is:

Rs(Ohms/square) = / Xj = (/ ln(2))*(V/I) = 4.532*(V/I) [Eq. 1-3].

For the more general case, the relationship between the junction sheet resistance and the

measured (V/I ratio depends on the probe configuration (usually linear or corners of a square) and

which probes are used for current and voltage connections. The detailed values also depend on

the orientation and distance of the 4PP from the wafer (or junction edges). For instance, for a

4PP placed along the edge of a “semi-infinite” conducting layer, the current spreads only into the

conducting side so,

Rs (at the edge) ≈ (1/2)*4.53*(V/I) ≈ 2.27*(V/I) [Eq. 1-4].

9-4

Probe spacing errors and configuration switching

A significant source of errors in 4PP measurements that use only a single configuration for

current and probe connections, as in Fig. 2-1, is the variations in the relative probe spacing in

contact with the junction. For macroscopic spring-loaded probes, with spacing of ≈1 mm, the

variations in probe contact locations for each probe results in 1 to more than 5% Rs variations for

single configuration measurements, depending on the quality and age of the probe tips and

mounting. Even for micro-4PPs, with probe spacing in the micron range, positional variations in

the contact locations of ≈20 nm results in systematic variation in Rs measurements which

increase as the probe spacing is reduced (Fig. 2-2).

Figure 2-2. Multi-probe array (left) with spacing ranging from 1.5to 18 um and Rs standard

deviations Rs for various probe spacing corresponding to a probe contact point variation of 20 nm

(right) [Peterson03].

Rs variations arising from probe spacing variations are removed by a combination of V/I

measurements using two of the three “configurations” of current flows and voltage measurements

for linear 4PP (Fig. 2-3). For Rs measurements using 4PPs and 4-terminal resistors, a

combination of two configurations results in a measurements which are independent of the probe

spacing (for a 4PP) or the resistor area (or a symmetric “van der Pauw” resistor) [van der

Pauw58]. Use of the various configurations results different (V/I) ratios which are linked

through the relation:

Ra/Rc = (Ra/Rb) /((Ra/Rb) –1) [Eq. 1-5]

where Ra is the (V/I) ratio for configuration A and similar for configurations B and C.

Figure 2-3. Configurations for current drive and voltage measurements for a linear 4PP.

For a linear 4PP array, measurements using any two of the three configurations can be combined,

linked by a transcendental relation which can be approximated as a series expansion in terms of

the measurement ratios [Perloff76]. In practice, Ra and Rb are usually chosen for dual-

9-5

configuration measurements since they have comparable magnitudes and have higher voltage

signals than Rc. A good approximation for Ka is:

Ka = -14.696 + 25.173(Ra/Rb) – 7.872 ((Ra/Rb)2 [Eq. 1-6]

where Rs = Ka*Ra. The value of ka ranges on both sides of the value of 4.532 for the ideal case

of an equally spaced 4PP array (Fig. 2-4).

Figure 2-4. Proportionality factor linking Ra, the (V/I) ratio measured using configuration A, to

the junction sheet resistance, Rs, when using measurements from configurations A and B (see

Fig. 2-3). The factor of /ln2 ≈ 4.532 is for an ideal 4PP, with equally spaced probe locations.

The use of dual-configuration 4PP measurements not only reduces the errors due to variable

probe spacing to <0.1% but also relaxes the requirements for probe orientation (usually solved for

the case of a linear probe array aligned along the wafer radius) and reduces the distance over

which wafer edge corrections become important (usually ≈5x the probe spacing for a 0.1% error).

Probe penetration for hard material 4PP tips A serious limitation of the use of 4PP measurements for shallow junctions is the probe

penetration depths associated with the pressure applied by hard (such as WC) and sharp probe

tips. The pressure usually exceeds the fracture strength of Si, resulting a region of crushed Si

chips, a “footprint”, at the probe locations (Fig. 2-5). For a typical probe loading of 80 to 100 g,

the depth of the crushed Si region exceeds 100 nm for hard probe tips.

Figure 2-5. 5x5 um AFM images (left) of probe landings for a spreading resistance probe (depth

40 nm) and a 4PP probe (depth 100 nm) and probe depths vs. loading (right) [Clarysse02].

9-6

Micro-4PP arrays

The penetration of hard materials probe tips for shallow junctions has led to the use of “soft”

probe tip materials (such as Hg and elastically-deformable metals) and the development of probes

with significantly lower probe contact force. The application of semiconductor-MEMS based

fabrication procedures has resulted in probe tip spacing on the order of 10s of um (see Figs. 2-2

and 2.6) [Petersen08]. The probe tip materials are Si or SiO2 coated with thin metal (Ni or Au)

skins and contacted to junctions with pressures on the order of 10-5

N (≈0.3 mg load on contact

size of 50-100 nm).

Figure 2-6. SEM images of a micro-4PP with spacing of 20 um.

The use of micro-4PPs results in minimal probe penetration (allowing accurate measurements for

≈10 nm junctions), reduced sensitivity to leakage current errors (both effects discussed later) and

minimized edge correction distances of much less than 1 mm. The principal advantage of the use

of micron-scale micro-4PPs is the high spatial resolution of local sheet resistance variations due

to local dopant (due to implant) or activation (due to annealing) non-uniformities (Fig. 2-7).

Figure 2-7. Rs values for line scans (left) with various spacing micro-4PPs measured (in 25 um

steps) perpendicular to a multi-pass laser anneal scan direction and a 45x101 point map (right) of

a laser annealed region using a 10 um 4PP spacing with step sizes of 50 um (x-direction) and 250

um (y-direction) [Petersen08]. Note the larger measured Rs variations in the line scans for the

smaller probe spacings and the complexity of the local Rs variation pattern in the map (laser scan

direction is in the x-direction).

9-7

Leakage current issues with 4PPs

Shallow junctions, which are often formed by combinations of highly-damaging implants,

including intentional amorphization implants, and annealing cycles with strongly constrained

thermal budgets, often contain high densities of residual defects near the junction location which

results in high leakage currents through combinations of recombination/generation and tunneling

mechanisms (discussed in more detail in Sec 2.6 of Chapter 13). Leakage current effects are

particularly strong when the local doping at the junction is high and depletion layers are thin,

such as for shallow junctions in halo or “pocket” profiles. The effects of local probe bias from

the current source probe of a 4PP include generation of leakage currents that penetrate below the

junction along a parallel path to the current flow in the junction though carrier generation and

tunneling effects (which increase with the density of dislocations located in the depletion layer of

the junction) and collapse of the depletion layer near the probe (Fig. 2-8) [Clarysse02]. The

effect of substrate leakage currents in a 4PP measurement is to add an additional current

component to the voltage differences along the current flow directions, usually resulting in a

lower measured Rs value than the sheet resistance of the junction [Faifer06].

An example of junction leakage current effect on macroscopic, hard-tip 4PP measurements is the

dependence of the measured Rs value on the probe loading for shallow junctions formed in halo

profiles (Fig. 2-8). For a 0.5 keV B high-dose implant annealed with a 1050 C “spike” RTP

cycle in a lightly-doped (≈1015

dopants/cm3, 20 Ohm-cm) wafer, the 4PP Rs values do not depend

on probe loading up to 100 g and agree well with a non-contact, carrier spreading measurement

(RsL). However, for the same junction formed in an As-doped halo profile with a junction

doping level of ≈1019

dopants/cm3 (and depletion layer of ≈20 nm), the 4PP measured Rs

decreases strongly as the probe loading (and probe penetration depth) is increased. In this case,

the As implant was annealed with an RTP cycle (1050 C/ 10s) prior to the B doping cycle so the

net recombination leakage current for the SDE/halo junction was relatively low, ≈4x10-7

A/cm2.

And yet, the 4PP induced leakage current for a 100 g loading was strong enough so that nearly all

of the probe current was flowing in the wafer (resulting in an Rs value of ≈wafer resistivity/

wafer thickness), giving a measured Rs value that was less than half of the RsL value for the

SDE/halo junction.

Figure 2-8. Sketch of junction and substrate current flows for a leaky shallow junction under the

reverse bias conditions near a 4PP current source (left) and effects of 4PP loading on a shallow

junction in a 20 Ohm-cm wafer and in a As-doped halo profile [Current07].

200

300

400

500

600

700

0 20 40 60 80 100 120

Probe loading (g)

Sh

ee

t R

es

ista

nc

e (

Oh

m/s

qu

are

)

20 Ohm-cm wafer

RsL

4PP

40 keV As halo

Halo anneal

0.5 keV B

1050 C / 1 s anneal

4PP, no Halo

Probe Loading (g)

9-8

Sheet resistance errors for leaky junctions with 4PPs in direct contact with the junction surface

extends to all probe tip types and include measurements made with 4-terminal “van der Pauw” or

“Hall” resistor test structures (Fig. 2-9). For shallow junctions formed in heavily-doped

substrates, with low resistivities, measured 4PP values can be more than an order of magnitude

lower than the junction Rs value (measured by non-contact carrier spreading, RsL, probes). For

micro-4PPs, with probe pacing of <100 um, the shifts in measured Rs values are minimal even

for junction leakage in the range of ≈10-2

A/cm2 (Fig. 2-9). However the accurate measurements

of such highly-leakage junctions, although technically possible with micro-4PP and RsL probes,

is certainly questionable and of limited practical value.

Figure 2-9. Measured ratio of 4PP and non-contact (RsL) Rs values for shallow junctions with

various carrier recombination leakage currents (left) and calculated effect of junction leakage on

Rs measurements with 4PPs for various probe spacings (right) [ Peterson06].

Comparison of Rs measurements methods with CVD junctions A detailed comparison of Rs measurements for shallow Boron-doped junctions was carried out

for various forms of 4PPs, including a micro-4PP, variable spacing probe (VPS) and a non-

contact, carrier-spreading probe (RsL) [Clarysse06]. The Boron-doped junctions were formed in

mid-doped (7x1017

As/cm3) n-epi layers by epi grown at 700 C to a doping level of 2x10

19 B/cm

3

and junction depths ranging from 2 to 132 nm (Fig. 2-10).

The probes with significant probe penetration into the Si surface, a WC-tipped 4PP (RS75) and a

lighter-loaded (≈3 g) variable spacing, 2-point, probe (VPS) showed significantly lower Rs values

for junctions thinner than 100 nm, with the heavier-loaded 4PP giving Rs values that approached

the14 Ohm/square value of the underlying As-doped epi layer (Fig. 2-11). These types of probes

were heavily compromised by a combination of substrate leakage currents and probe penetration

effects. Contact probes with much lower probe loading, an “elastic metal” tipped 4PP (EMP)

with 25 g loading and a 0.3 mg loaded micro-4PP, gave values in close agreement with the non-

contact RsL for junctions as thin as 10 nm. For the junctions thinner than 10 nm, the leakage

current (measured by the RsL probe) increased from ≈10-4

A/cm2 measured for the thicker

junctions to >10-2

A/cm2, indicating a very poor activation for the sub-10 nm layers.

9-9

Figure 2-10. SIMS profiles of Boron-doped epi layers used for comparisons of Rs measurements

[Clarysse06].

Figure 2-11. Rs measurements for various junction thickness epi layers (Fig. 2-10) for a WC 4PP

(RS75), a variable spacing probe (VPS), a micro-4PP (M4PP), am “elastic metal” 4PP (EMP) and

non-contact carrier spreading probes (RsL) at two testing sites (IMEC and FSM) [Clarysse06].

A strong confirmation of the failure of the sub-10 nm doped epi layers can be found in the trends

for the inverse of the sheet resistance with the junction depth (Fig. 2-12). For the ideal case of a

uniformly doped and completely activated layer, the inverse sheet resistance, 1/Rs, should

increase from the origin with a slope equal to the layer conductivity (inverse resistivity). The

actual measured trend line, using non-penetrating or non-contact probes, shows an intercept at ≈8

nm for both batches of epi layers, indicating a “dead”, non-active layer in all films for this epi

doping process.

9-10

Figure 2-12. Inverse sheet resistance vs. junction depth for two batches of epi-doped layers using

non-penetrating or non-contact probes. The “SIMS” points are for complete activation of the

dopant atom profile. Note the intercept at ≈8 nm in the Rs data, indicating a non-active layer,

approximately 8 nm thick, for all epi thicknesses.

CV measurements of surface active dopant concentrations

Single-probe versions of the “elastic material” (EM) probe tips used in non-penetrating 4PPs can

be used as surface capacitance-voltage (CV) probes when a thin (2 to 100 nm) metal oxide is

formed on the tip surface [Hillard07]. EM probe tips also are used for non-damaging CV

probing of dielectric films and have active contact diameters of 40 to 60 um depending on the tip

shape and loading pressure (Fig. 2-13).

Figure 2-13. “Elastic material” probe tip sketch (left) and SEM image (right) [Hillard07].

When the metal oxide thickness on the EM tip surface is ≈2 nm, equilibrium CV scans can be

used to determine the active surface dopant density, Nsurf, for levels up to and above 1020

dopants/cm3 by measurements of the ratio of the capacitance values in accumulation and

depletion, with corrections for band gap narrowing at high dopant concentrations (Fig. 2-14)

[Hillard07].

Si

Implant

9-11

Figure 2-14. CV measurements with an oxidized EM probe (left) and comparisons of 4PP sheet

resistance and inverse of the surface dopant concentration for a variety of heavily-doped junctions

(right) [Hillard07].

Nsurf measurements of active dopant concentrations sample the charge in the equilibrium space-

charge region, between 1 and 3 nm thick at the Si surface. These measurements are direct

discriminators for the relative efficiency of various anneal procedures for dopant activation. In

the example shown in Fig. 2-15, 0.5 keV B implants at 1015

B/cm2, with and without 5 keV Ge

PAI, were annealed with 5 different “limited diffusion” anneals ranging from a 650 C / 5 s SPE,

two peak temperature RTP “spikes”, a 1300 C “Flash” and a laser anneal. The Nsurf

measurements clearly showed the beneficial impact of the PAI step in creating conditions that

result in high surface concentration doping.

Figure 2-15. Nsurf measurements of 0.5 keV B implanted Si, with andwithout a Ge PAI step,

annealed by various methods [Borland06].

2-D resistivity profiling with scanning contact probes (SSRM)

The critical importance of the details of the dopant distributions near and under a CMOS gate

structure has encouraged the development of a number of scanning methods based on atomic

force microscopy (AFM) techniques. The most successful method to date is scanning spreading

resistance microscopy (SSRM). SSRM provides 2-D maps of local surface resistivity with high

spatial resolution (1-3 nm) with a reproducibility of 5-10% and the capability to measure dopant

profile abruptness to the level of 1-2 nm/decade [Eyben03].

EM-gate Equilibrium CV: S/D USJ Structures

0.4

0.5

0.6

0.7

0.8

0.9

1

-1.7 -1.2 -0.7 -0.2 0.3 0.8 1.3

Vg(V)

C(p

F) s19

s20

s21

s22

s23

s24

s19

s24

s23

0.5 keV B-doped SDE

0.0E+00

2.0E+19

4.0E+19

6.0E+19

8.0E+19

1.0E+20

1.2E+20

1.4E+20

Laser

1300

Fla

sh

1080

Spik

e

1000

Spik

e

650/

SPE 5

sec

1300

Fla

sh

1080

Spik

e

1000

Spik

e

650/

SPE

Nsu

rf (

cm

-3)

No Ge PAI

With Ge PAI

0

2,000

4,000

6,000

8,000

10,000

12,000

14,000

16,000

18,000

20,000

19 20 21 22 23 24

Sample#

Rs (

Oh

ms

/sq

.)

0.0E+00

5.0E-20

1.0E-19

1.5E-19

2.0E-19

2.5E-19

1/N

su

rf (

cm

3)

DC 4pp

FG CV

9-12

The sample surfaces are prepared by polishing or cleaving, with back-surface contacts formed by

focused ion beam sputtering and deposition (Fig. 2-16). The probe tip force is 1-5 uN as the tip is

scanned in contact with the device cross-section surface with a back bias of ≈5 mV. Early

implementations of the SSRM probes were diamond-coated AFM tips (Fig. 2-16). However the

contact surface area was not well defined due to the complexity of the diamond micro-crystals on

the tip surface. Diamond-coated tips were also prone to multi-point contacts and the Si base

material of the tip was prone to fracture under the scanning stress as it was drawn across the

sample surface. These problems were eliminated by the use of molded pyramidal diamond tips

which had well-controlled and flat contact surfaces (Fig. 2-16 (right)).

Figure 2-16. Sketch of SSRM measurement method (left) and SEMS images (right) of SSRM

probes, a diamond-coated AFM tip (left) and a pyramidal solid diamond tip (right) [Eybem06].

SSRM maps produce high-resolution 2-D images of dopant distributions and junction locations,

such as the laser annealed gate-area doping shown in Fig. 2-17 for 0.5 and 1 keV B implanted

SDE junctions with n-type halo profiles. SSRM maps provide the location of vertical and lateral

junctions as well as a precise measure of the gate overlap distances for SDE dopants.

14-16

16-18

33-37 20-25

39-4140-42

27-29

8-11

Laser - 0.5keV Laser - 1keV

14-16

16-18

33-37 20-25

39-4140-42

27-29

8-11

Laser - 0.5keV Laser - 1keV

14-16

16-18

33-37 20-25

39-4140-42

27-29

8-11

Laser - 0.5keV Laser - 1keV

14-16

16-18

33-37 20-25

39-4140-42

27-29

8-11

Laser - 0.5keV Laser - 1keV

Figure 2-17. SSRM maps of laser-annealed doped gate regions with SDE junctions formed with

0.5 keV B (left) and 1 keV B (right) implants. The vertical and lateral junctions for the 1 keV B

case show more extensive diffusion during the laser anneal than the 0.5 keV B case. The n-halo

doping (outlined with dotted lines) is more localized in the 0.5 keV B case. The junction depths

under the meal contacts (≈40 nm) are similar for both cases [Eyben06].

SSRM studies C cocktail implants

The addition of non-dopant “cocktail” ions, such as C and F, to B or P-doped pre-amorphized Si

can result in significant reductions in dopant diffusion during spike-RTP anneals, resulting in

9-13

shallower and more abrupt junction profiles (discussed in more detail in Chapter 13). The power

of SSRM analysis of 2-D dopant distributions is shown in the Fig. 2-18, with a comparison of

gate regions with SDE junctions doped with BF2+ and a combination of Ge

+ PAI, C

+ and B

+

implants annealed with spike-RTP process. In the BF2 case, which includes the B-diffusion

retardation effects of the F co-implant, the spike-RTP anneal results in a 38 nm SDE junction

depth with a 20-24 nm under gate diffusion. In comparison, the SDE region doped with a

sequence of Ge PAI, C and B implants results in a more than 2x reduction in junction depth (13-

15 nm) and under-gate diffusion (10 nm).

Spike

BF2

Cocktail

PAI+C+B

9-11 20-24

4-6

32-36

60-62

92-96

37-3913-15

Spike

BF2

Cocktail

PAI+C+B

9-11 20-24

4-6

32-36

60-62

92-96

37-3913-15

1.00E+18

1.00E+19

1.00E+20

1.00E+21

0 5 10 15 20 25 30 35 40

Depth (nm)

Net

carr

ier

co

ncen

trati

on

(cm

-3)

BF2

PAI+C+B

VERT.

1.E+17

1.E+18

1.E+19

1.E+20

1.E+21

-50 -40 -30 -20 -10 0 10 20 30

Depth (nm)

Ne

t c

arr

ier

co

nc

en

tra

tio

n (

cm

-3)

BF2

PAI+C+B

LAT.

Gate edgeSpacer edge

Figure 2-18. SSRM images (upper) and vertical and lateral dopant distribution profiles (lower)

for gate regions with SDE junctions doped with BF2+ (right) and a combination of Ge

+ PAI, C

+

and B+ (left) implants annealed with spike-RTP process. The halo doped regions are highlighted

in dotted line enclosures. The lateral profile was taken at a depth of 5 nm below the SDE surface.

Sectional profiles of the dopant levels through the SDE region show the strong increases in

junction profile abruptness in the vertical (11 nm/dec for BF2, 4 nm/dec for Ge/C/B) and lateral (8

nm/dec for BF2, 6 nm for Ge/C/B) directions. The vertical SDE abruptness improves the sheet

resistivity of the junction and the lateral abruptness increases carrier injection efficiency into the

channel (improving Ion) and reduces short-channel effects such as threshold voltage “roll-off” for

small gate dimensions. The C co-implant cocktail also improved the localization of the halo

profile doping, shown in the SSRM images enclosed in dotted lines, closer to the SDE junction,

which also contributes to the shallow and abrupt junction characteristics.

9-14

SSRM mapping of doping in 3-dimensional finFET arrays

SSRM methods can map 3D distribution of active dopant levels when used with astutely designed

test structures. In the example shown in Fig. 2-19, when a cross-section is exposed in a test array

with a stepped pattern of gate sizes so that a single SSRM analysis can sample resistance levels

both under and outside of the gate area, dopant levels can be measured along the whole length of

the fin [Mody11a]. By combining multiple-fin scans, composite images of doping levels in a

finFET structure can be constructed (Fig. 2-20).

Figure 2-19. Schematic and analysis planes for a finFET (upper left), top view of a test area

design with multiple fins capped by a gate with a stepped size (upper right) and example SSRM

images along the measurement plane (lower) [Mody11a].

Figure 2-20. Re-constructed SSRM images based on multiple-fin scans of the gate and

source/drain regions of a bulk finFET showing well, channel halo, extension and contact

(heavily-doped drain, HDD) doping regions [Mody11a].

9-15

3. Photo-voltage methods: The Surface Photo-Voltage (SPV) effect in semiconductors has been known for more than 50

years, [Brattain47, Brattain53]. In 1961 A. M. Goodman demonstrated that SPV dependence on

the intensity of irradiating light allows determination of minority carrier diffusion length in a

semiconductor, now widely used to monitor metal contamination in Si wafers [Goodman61]. The

basic grounds for the practical implementation of SPV metrology for dopants was developed by

R. Nakhmanson in a linear model of small signal ac-SPV [Nakhmanson75]. AC-SPV methods

were developed for measurement of doping density and resistivity of bulk semiconductors and

epi-layers [Kamieniecki97] and for analysis of ion implantation [Tsidilkovski04]. Analysis of

Junction Photo-Voltage (JPV) signals from light illuminated p-n junctions coupled with carrier

spreading and recombination effects gives a method for non-contact measurements of electrical

properties, junction sheet resistance and leakage current, known as RsL [Faifer94, Faifer04].

SPV basics The SPV effect is a change of a surface potential in a semiconductor resulting from illumination

of the surface with photons with energy greater than the Si bandgap. A surface potential in

semiconductor is due to a presence of different types of charges, such as fixed charge, trapped

and mobile charge, in a surface oxide layer and at the interface. Charge on the wafer surface

induces a space charge layer, with width W, in the semiconductor, that is a function of doping

concentration. Light can modify the surface potential by creating charge carriers in the bulk that

drift to the surface (Fig. 3-1). To establish a surface photo-voltage effect in semiconductor

materials, two components are needed: light with photon energy sufficient to generate electron-

hole pairs and an electric field, E. at the surface to separate those carrier pairs.

Figure 3-1. Schematic diagram of a Surface Photo Voltage (Vspv) measurement.

For small signal ac-SPV, where the modulated, super-band-gap light only slightly changes the

surface potential VSPV << kT, the photo-voltage response is proportional to the light intensity.

The signals yield either the space-charge capacitance (at high-frequencies) or minority carrier

lifetime (at low frequencies). The semiconductor resistivity can be determined from the space-

charge capacitance in inversion, and minority carrier lifetime can be used as an indicator of near-

surface metals contamination. Ion implantation introduces lattice defects, which act as recombination centers for charge carriers,

and changes the charge distribution in the implanted region. When light illuminates an implanted

sample, some of photo-generated carriers recombine at the defects before reaching the surface

(Fig. 3-2), reducing the measured photo-voltage. The higher the implant dose, the more

crystalline defects are created, the shorter is the photo-carrier lifetime and the smaller the surface

photo-voltage signal.

LIGHT SOURCE

Vspv

COUNTER ELECTRODE

Transparentelectrode

W - SurfaceDepletion Layer

-+

- - -+ + +

Beam of super band-gap light h > Eg

_

Semiconductor

+ + + + + + + + + +

+ + ++E

9-16

Figure 3-2. Band diagram of an implanted semiconductor in depletion; showing the SPV effect

and recombination through surface/space-charge states and through bulk implant defects.

A small signal frequency modulated ac-SPV can be described by:

Vspv() = I()Z() [Eq. 3-1]

where I is a light-generated current, Z = (Gtot +iCtot)-1

is a total impedance of the combined

system, is a light modulation frequency. Gtot and Ctot are total conductance and capacitance,

respectively [Nakhmanson75]. For an implanted semiconductor, the total capacitance consists of

a space-charge capacitance, Csc, capacitance of implanted region, CI, and interface states

capacitance, Cit:

itIsc

Isctot C

CC

CCC

)( [Eq. 3-2].

Similarly, the total conductance of implanted semiconductor can be presented as a function of a

space-charge conductance Gsc, implanted region conductance GI and a conductance of the neutral

bulk region Gb:

bIsc

bIsctot

GGG

GGGG

)( [Eq. 3-3].

The analysis of the typical conditions for ac-SPV measurement of implanted silicon allows

simplification of the above relationships, yielding: Ctot ~ Csc ~ /W and Gtot ~ GI ~ 1/0, where is

material dielectric permittivity and 0 is a carrier lifetime in implanted region [Steeples 06].

Approximating the photo-current as I ~ q, where q is the elementary charge and is the light

flux density, and combining all the above formulas, we arrive at a simple equation for the surface

photo-voltage:

)1()( 0

i

WqVspv

[Eq. 3-4],

with a lifetime 0 dominated by implant-induced carrier recombination.

h

h

VSPV Wd

illuminated

dark

h

Rh

Re

Wd

Energy band

modulated by

the charge from

implant

Implant

defects

EF

e

9-17

Another measured parameter, dynamic charge, Qd, makes correlation between photo-response

and material/implant parameters more lucid. Essentially, Qd is a function of Vspv and represents

the charge in a depletion layer modulated by the photo-carrier response:

Qd = kT/q(0W)2. [Eq. 3-5].

Calculating photo-carrier lifetime 0 allows correlation of dynamic charge Qd to the effective total

number of recombination centers induced by the implant process, Neff :

2

2

2

effi

id Nm

nqqnQ

[Eq. 3-6],

where Neff = (R)ND, is a charge carrier capture cross-section, R is an average width of an

implanted region and ND is a concentration of defects-recombination centers. Therefore, dynamic

charge, Qd , is proportional to the square of the implant-induced defect density and the width of

the implanted region.

AC-SPV sensitivity to implant parameters

Measurements based on ac-SPV are sensitive to implant dose, implant energy and beam angle.

Sensitivity to implant dose, the ratio of Qd change to the dose change, typically varies in the range

from 1 to 3. This high sensitivity is due to the quadratic dependence of measured dynamic

charge, Qd , to the implant induced defect density ND. Higher dose produces more defects that

shortens the lifetime and increases the dynamic charge. Sensitivity of Qd to implant damage is

illustrated in Fig. 3-3, where a ≈7x variation in the ion AMU results in about five orders of

magnitude change in the dynamic charge QD value.

Figure 3-3. Sensitivity of dynamic charge, Qd, to implant defect density at a constant dose of

6x1013

ions/cm2, with a common projected range of ~130nm [Tsidilkovski04].

The effect of the implant energy on the SPV measurement depends on the relationship of the

damage profile and the probing light absorption. For all heavy ion implants or low-energy light

ion implants, the energy dependence has a monotonic behavior with a positive sign, determined

by the functional dependence of Qd on the implant region width R. For lighter implant species,

dependence of Qd on implant energy can be non-monotonic demonstrating first positive and later

negative sign, due to the influence of the charge layer associated with the implant region. In both

cases, the SPV method sensitivity factor to implant energy varies from 1 to 4.

1.0E+01

1.0E+02

1.0E+03

1.0E+04

1.0E+05

1.0E+06

1.0E+07

0 10 20 30 40 50 60 70 80

Atomic Mass Units (AMU)

Qd

(t)

(C/m

3)

31P

+, 92keV

40Ar

+, 116keV

49BF2

+, 156keV

75As

++, 211keV

11B

+, 35keV

9-18

Variation of beam angle changes the depth distribution of defects, which modifies the

recombination times in the implanted region and, hence, alters the measured Qd signal (Fig. 3-4

and 3-5). The sensitivity of the method to the tilt/beam angle is approximately 5% to 10% of Qd

value per degree of angle, depending on implant energy.

Figure 3-4. SPV signal with various implant beam incidence angles for a 70 keV B implant at a

dose of 5x1013

B/cm2.

Figure 3-5. Qd wafer maps (contour interval 2%) of P31

implanted wafers for an energy of 1400

keV and dose of 6x1012

P/cm2 for average tilt angles of +/- 7

o (left, right) and 0

o (center).

Other studies have shown good success for the ac-SPV method in detecting beam glitches, micro-

uniformity problems and other failures in beam-wafer scanning [Krueger06, Pennella07].

Junction Photo-Voltage (JPV) measurements of sheet resistance (RsL)

Collection and analysis of SPV signals from p-n junctions, identified as junction photo-voltage

(JPV), can be used to measure, with a non-contact probe, the junction electrical properties of

sheet resistance (Rs) and recombination leakage current (L), collectively called RsL methods.

The basis of RsL measurements is to use photo-excitation of carriers in a p-n junction and wafer

substrate and to monitor, in a spatially resolved manner, the JPV signals inside and outside the

illumination area, when absorption of modulated light flux, (t)= 0(1-cos(2ft)), creates

electron–hole pairs in the semiconductor material. Two electrodes, a circular transparent

electrode with diameter 2r0 at the center of the probe and second round arc conducting electrode

subtending an angle and coaxial with the first electrode a small distance away, are used to

measure JPV voltages V1 and V2 (Fig. 3-6) [Faifer06].

B70k 5e13 X Tilt Angle Sensitivity

X=Various, Y=0

R2 = 0.99

1.5E+06

1.7E+06

1.9E+06

2.1E+06

2.3E+06

2.5E+06

-1.5 -1 -0.5 0 0.5 1 1.5

X Tilt Angle [degrees]

Qd

[C

/cm

^3]

9-19

Figure 3-6. (a) Photo-excitation and carrier drift with a modulated light source and two capacitor

electrodes for monitoring the induced junction photo-voltage in a spatially resolved manner; (b)

top view of the electrode configuration.

The JPV voltages V1 and V2 under low-level excitation are given by

)()(21)1(

01012

01 krKkrI

k

RRqV S

[Eq. 3-7]

)()()()1(

21211101

0

2

0

2 krKrkrKrkrIrk

RRqV S

[Eq. 3-8]

npSnpS CfRiGRk 2 [Eq. 3-9]

where I0, I1 , K0 and K1 are modified Bessel functions, 0 the incident photon flux density

modulated at frequency f, R the reflectivity, the quantum efficiency, RS, Gp-n , Cp-n the p-n

junction sheet resistance, capacitance and conductance [Faifer06, Clarysse08]. Under low-level

illumination, the p-n junction is forward biased (V>0) and the p-n junction voltage is V<< VT

=kT/q, where q is the electron charge, k Boltzmann’s constant, and T the wafer temperature (K).

The junction conductance, Gp-n, and leakage current density measured using RsL method, JRsL, are

defined as:

TRsL

V

np VJdV

dJG

0

[Eq. 3-10]

where J(V) is the p-n junction current density. For an ideal diode,

]1)/[exp()( 0 TVVJVJ [Eq. 3-11]

and the RsL leakage current density, JRsL , simply equals the pre-factor J0 (A/cm2).

r2

Junction

Depletion

Substrate

Modulated Light Beam

V V2

P+

Depletion

N

Spreading

Modulated

V1 V

h

e

r0

r11

2

1 2

(a)

(b)

r2

Junction

Depletion

Substrate

Modulated Light Beam

V V2

P+

Depletion

N

Spreading

Modulated

V1 V

h

e

r0

r11

2

1 2

r2

Junction

Depletion

Substrate

Modulated Light Beam

V V2

P+

Depletion

N

Spreading

Modulated

V1 V

h

e

r0

r11

2

1 2

r2

Junction

Depletion

Substrate

Modulated Light Beam

V V2

P+

Depletion

N

Spreading

Modulated

V1 V

h

e

r0

r11

2

1 2

Junction

Depletion

Substrate

Modulated Light Beam

V V2

P+

Depletion

N

Spreading

Modulated

V1 V

h

e

r0

r11

2

1 2

(a)

(b)

9-20

V1 and V2 versus light modulation frequency for high and low leakage current densities (Fig. 3-7

(left)) show the higher sensitivity of JPV voltages to leakage effects at lower frequencies, when

the contribution of Gp-n*Rs in Eq. 3-9 becomes dominant.

1 5 10 50 100

Frequency, kHz

0.01

0.1

1

10

V2 ; I0=10-3 A/cm2V2 ; I0=10-7 A/cm2

V1 ;I0=10-3 A/cm2V1; I0=10-7 A/cm2

Ph

oto

vo

lta

ge

, a

rb.

un

its

1 5 10 50 100

Frequency, kHz

0.01

0.1

1

10

V2 ; I0=10-3 A/cm2V2 ; I0=10-7 A/cm2

V1 ;I0=10-3 A/cm2V1; I0=10-7 A/cm2

Ph

oto

vo

lta

ge

, a

rb.

un

its

1.E-04

1.E-03

1.E-02

1.E-01

1.E+00

1.E+01

1.E+02 1.E+03 1.E+04 1.E+05

Modulation frequency (Hz)

JP

V s

ign

al (m

V)

V1 Experimental

V2 Experimental

V1 Simulation

V2 Simulation

V1 Theory

V2 Theory

V1

V2

Figure 3-7. JPV signals as a function of light modulation frequency for p-n junctions with low

(10-7

A/cm2) and high (10

-3 A/cm

2) recombination leakage currents (left) and simulated,

theoretical and experimental frequency curves for V1 and V2 (right). The dopant structure for the

right-hand figure was a p-type box profile doped to 1.5x1019

/cm3 with thickness of 41 nm on top

of a medium doped n-type 7x1017

/cm3 layer. Light source was red light (650 nm) with an intensity

of 10-4

W/cm2 [Faifer07a, Clarysse08].

By measuring the JPV at the two electrodes at different frequencies, combined with reference

JPV measurements on a wafer with a deep p-n junction with known sheet resistance, the sheet

resistance, Rs, conductance Gp-n and capacitance of the p-n junction, Cp-n can be simultaneously

determined using measured voltages V1 and V2 (Eqs. 3-7 and 3-8). The measurement sequence is

to first analyze V1 and V2 under high light modulation frequencies, where Rs*Cp-n, in Eq. 3-9 is the

dominant term in the solution. This allows for a direct determination of the USJ sheet resistance,

Rs, independent of junction depth or leakage current effects. The analysis is then repeated for a

lower light modulation frequency, where Gp-n*Rs, is a more important factor. Since Rs is already

determined at this point, Gp-n (and JRsL) can be determined directly. RsL measurements have been

coupled directly to Synopsis/Medici process and CMOS device simulations and the

characteristics of the JPV signals have been verified and calibrated by SIMS and SSRM

measurements (Fig. 3-7 (right)) [Faifer07, Clarysse08].

Implant process applications of JPV measurements of sheet resistance

The non-contact aspect of JPV measurements makes them well suited for evaluation of shallow

junctions to below 10 nm (see Fig. 2-11 and related discussion). The ability of the modulated

light beam to penetrate and create carrier pairs at the depths of several microns, when long

wavelength light sources are used, also makes JPV measurements uniquely well suited for direct

evaluation of the sheet resistance of deep MeV implants (Fig. 3-8). The range of Rs

measurements available to JPV techniques far exceeds contract probes and extends the dose range

coverage to 1011

ions/cm2 and below (when high-resistivity wafers are used).

9-21

Figure 3-8. Sheet resistance measurements with an RsL JPV probe ranging over 5 orders of

magnitude for dose and 2 orders of magnitude for energy. The energy range for JPV

measurements extends down to at least 0.2 keV for B ions.

The speed of non-contact JPV measurements allows for continuous motion of the test junction

under the RsL probe and results in efficient collection of data for high-resolution (sub-mm2 pixel

size) Rs mapping (Fig. 3-9).

Figure 3-9. A high-resolution (30,000 pts/150 mmD, ≈0.6 mm2/pixel) Rs map of a 90 keV B

+,

4x1011

B/cm2 implanted wafer with an x-y scanned ion beam. The average Rs is 57,671

Ohms/square with a uniformity of 2.4%

Leakage current measurements with RsL

Junctions are forward biased by ≈kT = 26 meV under low-intensity light illumination used for

JPV measurements. The current mechanisms in forward-bias include the diffusion current,

carrier recombination at the surface and at defects in the depletion layer and trap-assisted

tunneling to and from depletion layer defects. For reverse-bias junctions, such as in transistors

and diodes, the recombination mechanism is replaced by carrier generation at depletion layer

defects and band-to-band tunneling (BTBT) effects are added to the mix. Leakage currents in p-n

junctions and CMOS transistors are discussed in more detail in Chapter 13.

For heavily-doped shallow junctions, surface recombination rates are small so the relevant

forward bias leakage current effects are directly related to the size, number and location of defects

in the junction depletion layer, through carrier recombination and trap-assisted tunneling. The

nature and location of defects relative to the junction depends on the damage accumulation during

the various implantation steps and the recovery of the lattice structure and dopant diffusion and

9-22

activation during the anneal cycle. Leakage current measurements by RsL methods are therefore

a very sensitive monitor of the success (or failure) of the doping process design and operation.

The magnitude of the leakage currents also depends on the width of the depletion layer, which

decreases as the sub-junction doping level is increases, as in the case of a SDE/halo/well

structure. The recombination (generation) rate profiles for substrate and halo doping

concentrations of ND=1015

cm-3

and ND=1018

cm-3

and a model damage distribution centered 20

nm below the surface junction are shown in Fig. 3-10 (left). With increasing reverse bias the

width and magnitude of the recombination rate profiles increase and the leakage current

increases. In the case of low doping concentration, ND=1015

cm-3

, for 10 ohm-cm resistivity Si test

wafers, the recombination rate profile does not overlap the residual damage profile, Nt(z), and

hence the leakage current is low (Fig. 3-10 (right)). With increasing sub-junction doping

concentration, “halo”, the depletion region width decreases and the recombination profile

overlaps the implant residual damage profile, causing the leakage current to rise by orders of

magnitude. Trap-assisted and band-to-band tunneling for ND>1018

cm-3

increase the leakage

current further when the depletion region width is W10-20 nm.

Figure. 3-10 Carrier recombination-generation rates (left) versus depth for USJ with Xj=10 nm

formed in a c-Si substrate, ND=1015

cm-3

, and halo, ND=10

18cm

-3, for RsL and a reverse-biased

diode with Vr=-1V; dashed line shows the residual EOR damage profile, Nt(z), and ,,,forward

(RsL) and reverse-biased diode leakage current (right) versus substrate doping concentration for

implant (Ntmax=1017

cm-3

) and CVD EPI (Ntmax=1021

cm-3

) USJs. Dashed curves show the leakage

current without the BTBT contribution [Faifer07b].

Leakage current measurements with JPV signals

Leakage current measurements are based on an analysis of the effects of carrier recombination

and tunneling on the frequency dependence of JPV signals from p-n junctions [Faifer07c]. A

basic electrical circuit description of a leaky p-n junction includes the induced photo-current, Iph,

the capacitance of the depletion layer and conductance of the junction, Cscr and Gscr (Fig. 3-11).

At high light modulation frequencies, the JPV signals decrease as ≈1/frequency, with a slope

proportional to the junction capacitance (substrate doping). At lower frequencies the junction

conductance (leakage) dominates and the JPV signals become constant with changes in frequency

below a cutoff frequency, fc, which increases with higher leakage. As the leakage current

increases, fewer carriers survive in the junction region and the overall JPV signal at low

frequencies drops sharply.

VJPV = Iph*[1/(Gscr + iCscr)] [Eq. 3-12].

9-23

Figure 3-11. Circuit model (left) and frequency dependence of the JPV signals for a model

leaky p-n junction [Park01].

Measured JPV signals follow the general characteristics of the circuit model, with higher

frequency dependence slopes for heavier doped sub-junction regions (compare n-Si and n+(halo)

in Fig. 3-12) and the saturation and drop in JPV signals at low frequencies for increasingly leaky

junctions.

Figure 3-12. Measured JPV signals from shallow Boron-doped junctions in lightly-doped n-Si

test wafers and n-halo profiles annealed with RTP (low leakage) and laser (higher leakage)

anneals methods.

The measured RsL leakage, JRsL, is related to the junction conductance by (compare to Eq. 3-11):

JRsL = (q/kT)*Gscr = Jo(A/cm2) [Eq. 3-13].

The practical range of Jo measurements is ≈10-7

to ≈10-2

A/cm2.

CscrGscrIph

CscrGscrIph

.

fc

0.01

0.1

0.1 1 10

Gscr=10-4 Ohm-1*cm-2

Gscr=5*10-5

Gscr =10-5

JP

V,

arb

.un

it

.

fc

0.01

0.1

0.1 1 10

Gscr=10-4 Ohm-1*cm-2

Gscr=5*10-5

Gscr =10-5

Modulation frequency (Hz)

JP

V,

arb

.un

it

0.01

0.1

1

10

100

1000

100 1,000 10,000 100,000

Modulation Frequency (Hz)

Ju

ncti

on

Ph

oto

-vo

ltag

e, JP

Vo

ut

(mV

)

p+/n-Si

p+/n

+(halo)

leakage <10-7

A/cm2

leakage >10-3

A/cm2

RsL Data Analysis

9-24

Comparisons of Rs and leakage current maps based on JPV measurements is particularly

instructive for the case of limited-thermal budget annealing using flash lamp or laser methods.

For a “step-and-pulse” laser anneal, the shot-to-shot power variations in the laser anneal resulted

in local regions of the wafer with higher Rs and much higher leakage distributed over the wafer

(Fig. 3-13). The range of Rs values was ≈25% and the leakage current ranged over 2 orders of

magnitude.

Figure 3-13. 973-point Rs (left) and Jo (right) maps for a 0.5 keV B implant annealed with a

step-and-pulse laser on a 200 mm wafer. Regions of high Rs and Io are indicated with the darker

shading. Laser pulse-to-pulse power variations resulted in a global ≈25% variation in Rs around

an average of 996 Ohm/square and Io values from <10-7

to 2x10-5

A/cm2 [Current06].

4. Optical methods: Many types of optical probes have been used to characterize ion implanted surfaces. Modulated

optical reflectance, MOR, or “Therma-Wave”, has been widely used since the middle 80’s for

measurements of low-dose implants, with hundreds of tools in use. MOR is now employed to

track a wider range of implants, including shallow junctions. Recently, analysis of

photoluminescence (PL) signals arising from carrier recombination effects has been developed

for mapping of dose and anneal uniformity effects and for defect characterization with micron-

scale resolution images.

Modulated optical reflectance (MOR) or “Therma-Wave”

MOR measurements are made of the changes in reflectivity of an implanted semiconductor

material driven by absorption of an intensity-modulated “pump” light beam probed with a CW

beam. The absorption of the pump beam light with photon energy, E > Egap, leads to excitation

of valance electrons into the conduction band (Process 1 in Fig. 3-14). The photo-excited

electrons decay to the conduction band edge with emission of phonons (Process 2) in the time

scale of a few ps. If the light beam is modulated (usually in the MHz range), a periodic “thermal

wave” of phonons propagates into the semiconductor with a speed given by the thermal diffusion

rate, DT. The electrons at the conduction edge also diffuses into the semiconductor, (Process 3),

coupled with a hole carrier, creating a carrier density or “plasma” wave moving with a speed set

by the carrier diffusion coefficients. If the carrier pair encounters a lattice defect (or the free

surface) they recombine, releasing more phonons (Process 4 and 5).

9-25

Figure 4-1. Energy of carriers following the absorption of light with photon energy, E, with

excitation of electrons into the conduction band (1), decay to the conduction band edge, releasing

phonons (2), diffusion of an electron-hole “plasma wave” (3) and carrier recombination, releasing

more phonons at defects (4) or the surface (5).

The spatial extent of the propagation of the phonon, “thermal”, and carrier, “plasma”, waves is

strongly sensitive to the presence of defects in the semiconductor [Mandelis87, Salnik02]. For

modulated light incident on un-damaged Si, the carrier wave extends much deeper into the bulk

than the phonon wave due to the higher carrier diffusion rate (Fig. 4-2 (left)). For lightly

implanted Si, defects created by the stopping of the ions begin to increase the recombination of

carriers and the plasma wave penetration is decreased. For high-dose implanted Si, often

containing a band of amorphous material, the plasma wave is quickly recombined and only the

thermal wave of phonons reaches the heavily damaged layer.

Figure 4-2. Spatial extent of carrier “plasma” waves (lighter oval, PW) and phonon “thermal”

waves (darker oval, TW) illuminated by a modulated “pump” light beam (“pulsed” light beam)

for undamaged Si (left), moderated dose implanted Si (middle) and high-dose implanted Si

(right). Changes in surface reflectivity are measured with a CW “probe” light beam (solid beam).

E Eg

E-Eg

Valence band

Conduction band

1

5

4

2

3

9-26

The reflectivity of the Si surface changes with the degree of accumulated damage as the dose

increases. At very low doses (<1010

ions/cm2) the MOR signal is dominated by the plasma wave

penetration into the bulk Si (Region I, Fig. 4-3). For higher doses, the dominant effect changes

from the carrier wave to the interaction of the phonon, “thermal”, wave with the accumulated

damage (Region II). For doses from ≈1011

to ≈1014

ions/cm2, the MOR signal rises with

increasing dose, dominated by the thermal wave effects (Region III). At doses where a buried

amorphous layer begins to grow, ≈1014

ions/cm2, and higher (Region IV), the MOR signal gets

more complex, with effects from optical interference effects by reflections from the surface and

amorphous regions as well as beam annealing by the pump beam.

Figure 4-3. MOR (“Therma-Wave”) signals as a function of dose for 100 keV As implanted Si

showing the amplitude and phase of the reflected signal (a) and close up of the low dose plasma

and thermal wave signals (b).

MOR measurements are made with co-linear (or nearly so) laser beams; a modulated Ar laser

“pump” beam and a CW He-Ne laser “probe” beam to monitor the reflectivity (4.4).

Figure 4-4. Block diagram of a MOR measurement, “Therma-probe”, system.

10

102

103

104

105

9 10 11 12 13 14 15 16

Log(Implantation dose),(ions/cm 2)

TW

am

plitu

de, a.u

.

-100

-80

-60

-40

-20

0

20

40

60

80

100

120

TW

ph

ase

, d

eg

ree

s

I II III IV

As-Si, 100 keV

10

102

103

104

9 10 11 12

TW

am

plitu

de, a.u

. PLASMA

THERMAL

TW SIGNAL

As-Si, 100 keV

Log(Implantation dose),(ions/cm 2)

9-27

MOR measurements are widely used for dose tracking of threshold implants in the range of 1011

to 1013

ions/cm2, as well as well, halo and SDE implants at higher doses. Because the pump and

probe beams, which are co-linear on the surface, can be focused to a spot of ≈1 um, MOR

techniques can be used to probe test pad areas on product wafers for in-line process controls.

An interesting new application of MOR measurements is the extraction of junction depth and

abruptness for shallow junctions. In the example shown in Fig. 4.5, a Boron-doped, RTA

annealed 200 mm wafer was mapped with 21 sample points. The amplitude and phase of the

MOR signal was analyzed to determine the junction depths and the MOR signal with the pump

beam offset from the probe beam by 1 um was used to measure the abruptness (nm/decade); with

both parameters calibrated by a SIMS profile at the center of the wafer. Note the presence of a

local “hot spot” from the RTP anneal, seen in the deeper junctions at the bottom part of the wafer.

Figure 4-5. Contour maps from 21-point samples of junction depth (left, scale in Angstroms)

and abruptness (right, scale in nm/decade) for a Boron-doped, RTA-annealed junction [Salnik04].

Photoluminescence (PL)

The basis of photoluminescence (PL) measurements is the detection and analysis of light emitted

from a semiconductor material which is illuminated by a light beam with a photon energy greater

than the band gap. The PL light emission process is associated with a carrier recombination

process and is sensitive to doping levels and presence of defects in the semiconductor and, thus,

the PL signal can be used as a monitor of implant and annealing process. The optical system used

in PL measurements, sketched in Fig. 4-6, includes detection of both the PL signal and the

surface reflection of the incident beam. When the detected PL signal is scaled by the incident

photon flux, the resulting unit-less ratio, PLi, ranges from ≈104 for undamaged and lightly-doped

Si to ≈0 for heavily damaged material with very high carrier recombination rates. One should

note that the present system of PLi units is approximately an inverse of the earlier scale, where

highly damaged Si had a higher PLi value [Buczkowski06].

The wavelength of the excitation laser is chosen to excite the near-surface, “channel”, region or,

using a longer wavelength light, a deeper, “bulk”, volume. The probing laser light can be

focused to a sub-micron spot so test pads on product wafers can be tested in addition to mapping

of test wafers. The recombination process for electron-hole carrier pairs created by the

absorption of the incident laser light includes the PL emitting component and completing non-

radiative processes, Shockley-Read-Hall carrier recombination, emitting phonons, and Auger

process, involving internal excitation of electrons. All of these processes are influenced by the

local doping levels and the type and density of lattice defects in the semiconductor. Whole

wafer mapping of the PL signal can reveal subtle variations in implant doping uniformity (Fig. 4-

7) and the effects of local thermal variations during heat cycles on the efficiency of damage

annealing (Fig. 4-8).

9-28

Figure 4-6. Basics of a PLi measurement system showing a map of the photoluminescence (PL)

signal and the surface reflection [Buczkowski06].

Figure 4-7. PLi maps for implanted wafers with (a) a scanned spot beam system, (b) a ribbon

beam with mechanical scan and (c) a quad-mode implant with the ribbon beam-mechanical scan

implanter.

Figure 4-8. PLi wafer maps of 0.5 keV Boron implanted Si (10

15 B/cm

2) with anneals by a (a)

spike-RTP, (b) Flash lamp and (c) step-and-pulse laser scan. Note the imprints from the 3

support pins in the spike-RTP anneal and 6 support pins in the flash lamp apparatus. Compare

the pulsed-laser anneal Pli map with the Rs and leakage maps (Fig. 3-13) for a similar wafer.

9-29

In heavily-damaged Si, where the carrier diffusion lengths before recombination is short, the use

of a sub-micron probe beam spot can reveal local fluctuations in recombination rates which

reveals something of the character of the residual defect distributions after implant and annealing.

In the examples shown in Fig. 4-9, a Ge PAI and low-energy Boron doping implant is annealed

by a variety of flash lamp anneals. In sample (a), the 500 um scale local regon showed a uniform

and very high recombination rate (low PLi number) after flash annealing with a combination of

low base and peak temperatures. In (b), annealed with the same low base temperature but a

higher peak temperature, the recombination rate is lower (higher PLi value) indicating a more

effective damage anneal with the hint of a granularity to the local defect distribution. In (c),

annealed at higher base and peak temperatures, the PLi value is higher, indicating more effective

annealing and the local variations in the PLi signal hint at a distributed defect distribution,

perhaps an array of dislocation loops. In (d) the flash anneal conditions resulted in strong wafer

distortion, warping, and a dense array of local PLi features which are suggestive of local slip lines

with a very high carrier recombination rate (low PLi value). These results hint that local PLi

mapping could be highly useful as a quick-turn supplement to TEM analysis of defect annealing.

Figure 4-9. Local 500x500 um PLi maps (1 um steps) of implanted (Ge PAI and low-energy

Boron) and flash annealed Si. The flash anneals are (a) low temperature base and low peak

temperature, (b) low temperature base and high peak temperature, (c) higher temperature base

and high peak temperature, (d) very high peak temperature resulting strong wafer warping.

5. Physical methods: Physical methods for analysis of implanted materials, such as secondary ion mass spectroscopy

(SIMS), low-energy x-ray spectroscopy (LEXES) and transmission electron microscopy (TEM),

continue to provide vital information on the nature of implanted and contaminant species and

accumulated damage after implant and residual defects after annealing. We discuss here two

imaging techniques, holographic-TEM and atom-probe tomography, which have been developed

recently to the level where junctions can be located to nm resolution as well as dopant atoms on

an atom-by-atom basis.

5.1. Electron holography in TEM

In off-axis electron holography (Fig. 5-1) an electron biprism, which consists of a thin wire

electrode, is used to overlap the object wave, through the imaged material, with the reference

wave, which passes nearby in vacuum. When a positive voltage is applied to the biprism, the

electron waves on either side of the biprism bend toward the center to overlap. The overlapping

waves form an interference pattern, and this interferogram, or hologram, changes depending on

how the object wave is affected by the specimen [Volkl98].

9-30

Figure 5-1. Electron beam optics (left) for collection of holographic-TEM images and an

example image of a MOSFET structure showing interference fringes.

The phase of an electron wave is changed by the positive Coulomb potential, usually 10 to 15 eV,

in a specimen, which accelerates the incident electron, making the electron wavelength shorter in

the specimen than in vacuum. In the absence of dynamical diffraction effects and magnetic fields,

the phase shift of the object wave, (x, y), is given by [Reimer82],

(x,y) = CE*V(x,y)*t [Eq. 5-1]

where V is the potential, CE is an interaction constant (7.28106 rad/Vm for a 200k-V electron),

and t is the specimen thickness. We assume that the potential V(x,y) does not vary along the

incident beam direction, and that the specimen thickness t is uniform over the observed area.

In cross-sectional observation of a p-n junction, the inner potential V(x, y) can be divided

into two parts: the mean inner potential, V0, and a built-in potential generated at the junction,

Vbi(x, y). V0 is uniform throughout the specimen, while Vbi(x, y) varies across the p-n junction,

with a phase shift of

(x,y,z) = CE*[Vo + Vbi(x,y)]*t [Eq. 5-2].

Although holo-TEM imaging is possible on the atomic-scale, the typical spatial resolution is 4 to

10 nm with an energy resolution of 0.1V with fields of view from 200 nm to ≈1 um, adequate for

imaging of MOSFET structures [Lehmann02]. Holo-TEM spatial resolution is also limited by

the Debye length (~4 nm in Si with 1018

cm-3

dopant concentration).

9-31

Alternating type doping regions, not visible in conventional TEM images (Fig. 5-2), are seen with

holo-TEM methods and can be directly compared to process simulations. For a state-of-the-art

10 nm junction, formed with a combination of B18

H22

and multiple-halo implantation and high-

temperature millisecond annealing, the solid curve in the holo-TEM image (Fig. 5-2, right, upper)

shows the 0.0-V equi-potential line, and the dashed curve show the equi-potential line obtained

by a Monte Carlo doping-process simulation [Uejima07, Ikarashi08].

Figure 5-2. Conventional (upper, left) and holo-TEM (lower, left) images of the junctions under

a 30 nm gate structure showing the delineation of n and p-type regions for a SDE and SDC

structures and holo-TEM (upper, right) and process simulation (lower, right) for a B18H22-doped

10 nm SDE junctions with multiple-angle halo profiles and a ms timescale anneal [Uejima07].

The use of holo-TEM in process improvements is discussed in detail in Section 2.7 in Chapter 12.

To investigate the lateral potentials and depletion layer thickness under a 30 nm gate electrode,

holo-TEM images and lateral scans of the internal potential at a depth of 10 nm (Fig. 5-3) for a

Be+C+B “cocktail” co-implantation is compared to a BF2 process after annealing [Mineji06].

The potentials of the darkest areas (p+-type) were set as -0.5 V and those of the brightest areas (n-

type) as +0.4 V. The solid curves indicate the 0.0-V equi-potential lines, which show the

approximate position of the p-n junction. The equi-potential lines in the co-implantation sample

were deeply curved at a depth of 20 nm, and the SD and SDE regions were clearly

distinguishable. The depth of the curve indicates that the SDE junction depth was about 20 nm.

In contrast, no such curves of the equi-potential lines were observed in the BF2 implantation

sample. The potential distribution at 10 nm beneath the wafer surface showed that the depletion

width, the width of the potential transition region from an n-type to a p-type region, in the co-

implantation sample was clearly narrower and more abrupt than that in the BF2 implantation

sample.

9-32

Figure 5-3. Potential maps and the lateral distribution (10 nm beneath the surface), (a) and (b),

for a Ge+C+B co-implantation sample, and, in (c) and (d), a conventional BF2 implantation

sample. In (a) and (c), the solid curves are the 0.0-V equi-potential lines, and the dashed lines

represent the outlines of the equi-potential lines. In (b) and (d), the thick arrows indicate the

depletion regions at the p-n junctions. The gate size is 30 nm. [Mineji06, Ikarashi08].

5.2 Atom probe tomography (APT)

Building on the development of field-ion microscopy from the 1950’s and field-evaporation time-

of-flight “atom probing” in the ‘70’s and onward, atom probe tomography (APT) is now capable

of efficient analysis of 3-D distributions of dopants and other atoms in selected CMOS devices

structures [Tsong93, Kelly07a]. Sample preparation for APT analysis uses many of the

specialized micro-etching and manipulation tools developed for efficient failure analysis of chip

structures (Fig. 5-4, left). The goal of the sample preparation is to place the materials to be

analyzed on the tips of an array of etched Si pillars. The samples are then shaped by ion

bombardment into sharply pointed tips. The samples are then placed in the APT tool and located

under a negatively-biased local electrode (Fig. 5-4, right). When the local electrode is pulsed to

a high potential, or the tip is heated with a local laser pulse, with the local electrode at a potential

close to the field evaporation level, single atoms and small clusters are removed from the sample

and accelerated toward a 2-D detector plate. The mass and charge state of the field-evaporated

ions are resolved by tracking the time of flight (TOF) of the ions to the detector array following

the evaporation pulse. By collecting the 2-D location of the individual ion hits, a much

magnified (≈106) image of the evaporating surface is obtained with atomic resolution. By

collecting TOF and location data from all of the atoms removed from the sample by repeated

pulsing of the local electrode (or laser beam), a 3-D image of the chemical distributions in the

sample can be reconstructed and analyzed with atomic resolution.

9-33

Figure 5- 4. Sample preparation (left) showing an array of etched Si pillars to hold APT samples

(a and b), removal of a focused-ion beam sputtered sample slice (c), placement of a sample

segment on a Si pillar (d) and after ion beam sharpening of the sample prior to APT analysis (e)

and APT field evaporation (right) with a local electrode placed over a sample and Si pillar and a

2-D imaging detector with capabilities to collect the time of flight for each evaporated atom.

APT analysis of B atoms in poly-Si (Fig. 5-5) shows that many of the dopant atoms are located in

and near the grain boundaries, and are not likely to be electrically active, with a much smaller

fraction of B contained within the individual grains. Similar analysis of the results of various

methods for poly-Si deposition, implantation and annealing for very-high (>1016

ions/cm2) dose

doping would likely produce valuable insights towards improving the electrical properties of

DRAM gate electrodes (discussed in more detail in Section 3 of Chapter 12).

Figure 5-5. APT image (left) of Si and B atoms in poly-crystalline Si showing B atoms clustered

near the grain boundaries and line scan B distributions (right) across a grain boundary and within

a grain [Kelley07a].

APT imaging of an As-doped SDE junction (Fig. 5-6) yielded lateral atomic doping distributions

at and under the gate electrode, with good agreement with process simulations.

9-34

Figure 5-6. Dopant (As), Si and O distributions near the edge of a Si gate electrode by APT

imaging obtained using a Local Electrode Atom Probe (LEAP) (a), a corresponding TEM image

of a similar region (b), a FLOOPS process simulation for the As distribution under the gate (c)

and lateral As concentration profiles at a depth 2 nm below the gate edge from FLOOPS and APT

data [Kelly07b].

The combination of chemical APT analyses with high-resolution TEM imaging of residual

defects after doping processing is particularly powerful. In the case shown in Fig. 5-7, As-doped

Si annealed at 600 and then 1000 C contains dislocation loops near the peak of the as-implanted

As profile, as seem in TEM images. APT analysis of similar areas shows that a high

concentration (>2 %) of As forms near and at the edges of the dislocation loops. The presence of

As in these “Cottrell atmospheres” near the dislocation edges relaxes the strain fields in these

areas and contributes to the stability of the loops against further annealing at high temperatures.

The size of these dislocation loops, ≈35 nm, is larger than the gate length for advanced CMOS

transistors, ≈20 nm, which indicates that similar defects and dopant clustering effects could have

seriously detrimental effects on the operation of advanced process transistors.

9-35

Figure 5-7. TEM (left) and APT (right) images after As doping of Si and annealing at 600 C for

30 min and then 1000 C for 30 s. The APT image highlights As clustering, at concentrations

>2 %, at the edges of dislocation loops. A 2 nm native oxide film is seen in both images

[Thompson07].

APT analysis of doping in finFETs

APT provides atom-by-atom distributions of dopants and other atoms in 3D devices structures,

such as finFET channels. In the example shown in Fig. 5-8, the impact of beam incidence angle

on lateral B profiles in 40 nm thick finFET test structures is shown for sidewall incidence angles

of 10o and 45

o [Mody11b]. The APT data was collected after the implant anneal and contact

silicide formation, so the measured dopant distributions include the effect of implant and anneal

conditions.

Figure 5-8. Lateral and vertical profiles for B implants into ≈40 nm wide finFET test structures

with a grazing angle (10o) incidence on the fin sidewalls (left) and lateral profiles for SiO2, SiF

from fin sidewall oxides and B implants with a 45o beam incidence on the fin sidewall (right)

[Mody11b].

By comparing SSRM (see Fig. 2-19 and 2-20 and related text) and APT results (Fig. 5-9) one can

see that a large fraction of the B dopants near the finFET sidewalls were not electrically activated

(the sidewall atomic APT concentrations are much higher than the carrier concentrations from

SSRM analysis) by the process conditions used in this study.

9-36

Figure 5-9. Comparison of lateral profiles for B-doped 40 nm wide finFET test structures

measured by SSRM and APT for implant beam incidence angles of 10o and 45

o [Mody11b].

6. Summary: It is indeed fortunate that the variety and capabilities of metrology tools available for process

development and in-line controls continues to expand in this time of many changes and

challenges in doping technology, including new forms of ion implantation materials and tools, the

integration of ms-timescale annealing methods into production processes and the proliferation of

3D device architectures in memory, logic and power IC devices.

Even the venerable 4PP technique has seen many useful improvements in the last few years, with

the development of non-damaging “soft” probe tips and recently the availability of MEMS-style

micro-4PPs with micron-scale probe spacing. Coupling of single-probe resistivity analysis with

AFM-style probe scanning techniques now provides 2-D maps of junctions, gate electrodes and

metal contacts (SSRM) with nm-scale resolution.

Physical analysis of defects, junctions (holo-TEM) and atomic distributions (APT) have been

developed into methods which can be implemented for efficient characterization of full 3D

CMOS transistor structures as well as 1-D analysis of complex materials.

New classes of non-contact probes are now available based on local measurements of surface

charge (Qd) and junction photo-voltage (RsL) signals with illumination with light beams as well

as improved “pure” optical signal probes based on analysis of surface reflectivity changes under

pulsed light illumination (MOR) and photoluminescence (PL). All of these non-contact

metrologies are linked by a common sensitivity to carrier recombination effects, which in turn

reflect in sensitive fashions the local dopant and residual defect distributions.

The integration of these new and improved metrology techniques into process and device

simulation procedures, already well underway for SSRM, RsL, holo-TEM and APT, will add to

the power and general utilization of these methods.

The task for young students (process and manufacturing engineers) is to understand and apply

these metrology tools to the formidable challenges at hand. We (the authors) hope that this

chapter will aid in that process.

9-37

Acknowledgements

The discussion of the specific metrologies this chapter was based on the work, advice and

publications of Pierre Eyben at IMEC, Peter Nielsen at CAPRES, Mark Benjamin at Solid State

Measurements, Edward Tsidilkovski at QC Solutions, Alex Salnik at KLA-Tencor, Andrzej

Buckowski at Nanometrics, Tom Kelly at Cameca/Imago, and Noboyuki Ikarashi at Renesas. In

addition, many co-workers contributed to the development, understanding and application of

these metrologies, including but certainly not limited to, Vladimir Faifer, Trudo Clarysse,

Wilfried Vandervorst, John Borland, Ken Steeples, Walt Johnson and Bob Hillard.

References Brattain 47 W.H. Brattain, Phys. Rev. 72 (1947), p.345

Brattain 53 W.H. Brattain, J. Bardeen, Bell System Tech. J. 32 (1953), p.1

Borland06 J. Borland, S. Shishiguchi, A. Mineji, W. Krull, D. Jacobson, M. Tanjyo, W. Lerch,

S. Paul, J. Gelpy, S. McCoy, J. Veturini, M.I. Current, V. Faifer, R. Hillard, M. Benjamin,

T. Walker,A. Buczkowski,Z.Li,J. Chen,“45nm node p+ USJ formation with high dopant

activation and low damage”, AIP Proc. 866 (2006) 96-100.

Buczkowski06 A. Buczkowski, Z. Li, T. Walker, S.G. Hummel, J.O. Borland, “Non-contact, image-based

photoluminescence metrology for ion implantation and annealing process inspection”,

AIP Proc. 866 (2006) 566-569.

Clarysse02 T. Clarysse, D. Vanhaeren, W. Vandervorst, “Impact of probe penetration on the electrical

characterization of sub-50 nm profiles”, J. Vac.Sc. Technol. B20 (1) (2002) 459-466.

Clarysse06 T. Clarysse, A. Moussa, F. Leys, R. Loo, W. Vandervorst, M.C. Benjamin, R.J. Hillard,

V.N. Faifer, M.I. Current, R. Lin, D.H. Petersen, “Accurate sheet resistance measurements

on ultra-shallow profiles”, Mat. Res. Soc . Proc. 912 (2006) 0912-C05-07.

Clarysse08 T. Clarysse, A. Moussa, T. Zangerle, F. Schaus, W. Vandervorst, V. Faifer, M. Current,

“Insights in junction photo-voltage based sheet resistance measurements for advanced

complementary metal-oxide semiconductor”, J. Vac. Sci. Technol. B26(1) (2008) 420-424.

Current06 M.I. Current, V.N. Faifer, T.M.H. Wong, T. Nguyen, A. Koo, “Non-contact sheet

resistance and leakage current monitoring of multi-implant ultra-shallow junctions:

doping and damage effects for ms-anneals” AIP Proc. 866 (2006) 582-585.

Current07 M. Current , V. Faifer, J. Halim, N. Ohno, “Dopant activation and leakage current

aspects of SDE/Halo CMOS junctions measured with non-contact junction photo-voltage

methods”, Proc 7th

Workshop on Junction Technology (2007) 43-46.

Eyben03 P. Eyben et al., AIP Conf. 638 (2003) 685.

Eyben06 P. Eyben, S. Severi, R. Duffy, B. Pawlak, E. Augendre, W. Vandervorst, “Analysis

and optimization of new implantation and activation mechanisms in ultra-shallow

junction implants using scanning spreading resistance microscopy (SSRM), Mat.

Res. Soc. Proc. 912 (2006) 0912-C05-08.

Faifer06 V.N. Faifer, M.I. Current, T.M.H. Wong, V.V. Souchkov, “Non-contact sheet

resistance and leakage current mapping for ultra-shallow junctions”, J. Vac.Sci. Technol.

B24(1) (2006) 414-420.

Faifer07a V.N. Faifer, M.I. Current, D.K. Schroder, “Characterization of ultrashallow junctions using

frequency-dependent junction photovoltage and its lateral attenuation”, App. Phys. Lett. 89

(2006) 151123.

Faifer07b V.N. Faifer, D.K. Schroder, M.I. Current, T. Clarysse, P.J. Timans, T. Zangerle,

W. Vandervorst, T.M.H. Wong, A. Moussa, S. McCoy, J. Gelpy, W. Lerch, S. Paul, D.

Bolze, J. Halim, “Influence of halo implant on leakage current and sheet resistance

of ultra-shallow p-n junctions”, J. Vac.Sci. Technol. B25 (2007) 1588-1592.

Faifer07c V.N. Faifer, M.I. Current, D.K. Schroder, T. Clarysse, W. Vandervorst, “Junction photo-

voltage (JPV) techniques for ultra-shallow junction characterization”, ECS Trans. 11(3)

(2007) 135-147.

Goodman61 A.M. Goodman, J. Appl. Phys. 32 (1961), p. 2550.

Hillard07 R.J. Hillard, C.Win Ye, J.O. Borland, “Dopant activation and profile determination

with an elastic material probe (EM-probe)”, AIP Proc. 931 (2007) 261- 264.

9-38

Ikarashi08 N. Ikarashi, M. Oshida, M. Miyamura, M. Saitoh, A. Mineji, S. Shishiguchi, M.Hane,

“Electron holography characterization of ultra-shallow junctions in 30 nm gate length

metal-oxide-semiconductor field-effect transistors”, Japan J. Appl. Phys. 74 (2008)

Kelly07a T.F. Kelly, D.J. Larson, K. Thompson, R.L. Alvis, J.H. Bunton, J.D. Olson, B.P. Gorman,

“Atom probe tomography of electronic materials” Annu. Rev. Mater. Res. 37 (2007)

681-727.

Kelly07b T.F. Kelly, J. Roberts, K. Thompson, “LEAP tomography of atomic materials”,

7th Workshop on Junction Technology (2007) 25-33.

Krueger06 C. Krueger, C.H. Ng, Z. Zhao, G. Krych, :”Surface charge profiling-an advancement in ion

implant monitoring”, AIP Conf. 866 (2006) 531-533.

Lehmann02 M. Lehmann, K. Brand, H. Lichte, Microscopy and Microanalysis, 8 (2002) 536.

Mandelis87 A. Mandelis, Ed., “Photoacoustic and Photothermal Phenomena in Semiconductors“,

(1987) North-Holland.

Mineji06 A. Mineji, S. Shishiguchi, “Ultra-shallow junction and super-steep halo formation using

carbon co-implantation for 65 nm high performance CMOS devices” 6th

Workshop on

Juncton Technology (2006) 84-87.

Mody11a J. Mody et al., “”3D-Carrier profiling in finFETs using Scanning Spreading Resistance

Microscopy”, IEDM11 (2011) 6.1.1-6.1.4.

Mody11b J. Mody et al., “Dopant and carrier profiling in finFET-based devices with sub-nanometer

resolution”, VLSI11 (2011).

Nakhmanson75 R.S. Nakhmanson, Solid State Electron. 18 (1975), p. 617

Pennella07 F. Pennelle, P. Pianezza, E. Tsidilkovski, G. Krych, K. Steeples, “Non-contact SPV-based

method for advanced ion implant process control”, Adv. Semi. Manu. Conf (2007).

Park01 G.E.Park, D.K. Schroder,et al., J. Electrochem.Soc.148, (2001) G411.

Perloff76 D.S. Perloff, J. Electrochem. Soc. 123 (1976) 1745.

Petersen03 C.I. Petersen, D. Worledge, P.R. Petersen, “Reproducibility of nano-and micro-scale multi-

point probe sheet resistance measurements”, Mat. Res. Soc. Proc. 738 (2003) G4.3.1.

Petersen06 C. Petersen, R. Lin, D.H. Petersen, P.F. Nielsen, “Micro-scale sheet resistance

measurements on ultra-shallow junctions” RTP06 (2006).

Petersen08 D.H. Petersen, R. Lin, T.M. Hansen, E. Rosseel, W. Vandervorst, C. Markvardsen,

D. Kjaer, P.F. Nielsen, “Comparative study of size dependent four-point probe

sheet resistance measurements on laser annealed ultra-shallow junctions”,

J. Vac. Sci. Technol. B26(1) 362-367.

Reimer82 L. Reimer, “Transmission Electron Microscopy”, Springer-Verlag, (1982).

Salnik02 A. Salnik and J. Opsal, “Quantitative photo-thermal characterization of ion-

implanted layers in Si”, J. Appl. Phys. 91, 2874 (2002).

Salnik03 A. Salnik and J. Opsal, “Dynamics of the plasma and thermal waves in surface-

modified semiconductors”, Rev. Sci. Instrum. 74, 545 (2003)

Salnik04 A. Salnik, L. Nicolaidesw, J. Opsal, A. Jain, D. Rogers, L. Robertson, “Simultaneous

determination of ultra-shallow juctions depth and abruptness using thermal wave

technique”, Rev. Sci Instrum. 75(6) (2004) 2144-2148.

Steeples 06 K. Steeples, E. Tsidilkovski, “Photoelectric Measurement Method For Implanted Silicon:

A Phenomenological Approach”, Ion Implant Technology, AIP CP866 (2006), 558-561.

Thomson07 K. Thompson, P.L. Flaitz, P. Ronsheim, D.J. Larson, T.F. Kelly, “Imaging of

Arsenic Cottrell atmospheres around Silicon defects by three-dimensional atom

probe tomography”, Science 317 (2007) 1370-1373.

Tsidilkovski04 E. Tsidilkovski, K. Crocker. K. Steeples, “Ion implant process monitoring with a dynamic

surface photo-charge technique”, Adv. Semi. Manu. Conf. (2004).

Tsong93 T.T. Tsong, “Atom-probe field ion microscopy” PhysicsToday, May 1993, 24-31.

Uejima07 K. Uejima, K. Yako, N. Ikarashi, M. Tanaka, T. Nagumo, A. Mineji,

S. Shishiguchi, M. Hane, “Pushing planar bulk CMOSFET scaling to its limit by

ultimately shallow diffusion-less junctions”, IEDM07 (2007) 151-154.

Van der Pauw58 L.J. van der Pauw, Philips Res. Rep. 13 (1958) 1.

Volkl98 E. Völkl, L.F. Allard and D.C. Joy, Ed., “Introduction to Electron Holography”, Plenum,

New York (1998).