Yao N. Focused Ion Beam Systems: Basics and Applications

Подождите немного. Документ загружается.

angle dependence as backscattered ions, because the recoil cascade approa-

ches the surface as the incidence is inclined. Figure 4.11 shows the ﬁtting

parameter f as a function of the atomic number, Z

, of materials under 10 keV

electron impact and 30 keV Ga ion impact, where the ﬁt is applied to the

calculated results for the angles of less than 60

[60]. For electron impact, the

backscattering effect changes the f value from above 1 to below 1 as

the backscattering coefﬁcient increases with increasing Z

. On the other hand,

due to the small contribution of the backscattered ions and the recoiled target

atoms to kinetic emission, the general dependence approximately obeys the

inverse cosine law ( f ¼1), except for low-Z

materials ( f > 1) where the

contribution of recoiled target atoms largely increases.

Further study of topographic contrast is required to resolve differences in

topographic contrast of SE imaging in the FIB and SEM systems. The

topographic contrast is convolved with detector geometry and conﬁguration

to produce an image with many features familiar to the casual observer. It

should be noted that topographic contrast is frequently overwhelmed by

channeling contrast in certain material systems and geometries, pointed out

by Prewett and Mair [64].

4.3.3 Channeling contrast

Channeling (orientation or grain) contrast is perhaps the most striking fea-

ture of FIB microscopy of metallic samples, and arises from channeling of the

30 40

Target atomic number, Z

2.0

1.5

1.0

0.5

10 keV e

30 keV Ga

Fitting parameter, f

Figure 4.11 Calculated ﬁtting parameter f in ()/(0) ¼(cos )

f

and

–()/–(0) ¼(cos )

f

as a function of Z

for 30 keV Ga ion impact and

10 keV electron impact, respectively [60].

Focused ion beam systems108

incident ions between lattice planes of the sample. Depth of penetration of

the channeled incident ion varies with the relative angle between the ion beam

and the lattice plane and the interplanar spacing of the lattice. A typical SIM

image showing the channeling contrast from the aluminum (Al) surface on a

silicon (Si) specimen is shown in Figure 4.12. The dark grains with low SE

yield correspond to the channeled grains.

Through channeling contrast of the SEM images, grains in polycrystalline

samples are seen with BSE (backscattered electron) contrast of the SEM

images that depends sensitively on the sample tilt [15]. This sensitivity results

from the dependence of the BSE yield on the orientation of the primary beam

relative to the lattice planes and is caused by the primary Bloch wave ﬁeld.

(a)

(b)

5 mm

Figure 4.12 Channeling contrast SIM images of Al layer on Si substrate. (a)

Image observed at an incident angle of ﬁ ¼15



and (b) observed at ﬁ ¼30



after a few minutes bombardment at ﬁ ¼15



. Note: the bright grains in (a)

are more etched than the dark ones.

Imaging using electrons and ion beams 109

Due to the sensitivity of the variation on the surface, the channeling contrast

can decrease with an increasing electron-probe focusing angle which should

not exceed 1–10 mrad. The depth of average channeling varies with Z

, but

for most elements, the additional contribution to the contrast is much smaller

from depths > 50 nm. The channeling contrast is also very sensitive, either on

an amorphous surface oxide layer or multiple scattering in the crystal.

As to the directional effects of kinetic emission, experimental and theore-

tical work has been reviewed by Brusilovsky [65 ]. Fast particles entering

crystalline materials in low-index lattice directions are steered into the

interior of the target by many small-angle scattering collisions. This causes a

drastic reduction in the SE yield for two reasons: ﬁrst, kinetic emission

depends on inelastic energy loss and channeled ion transfers relatively small

energy in each collision along its path; second, SEs generated deep in the

material have difﬁculty in escaping to the surface. Very recently, an experi-

ment [66] revealed that the ﬁne changes in SE intensities in SIM with the tilt

angle was in reasonable agreement with the incident angle dependence of the

SE yield observed using an argon ion beam by Mashkova et al.[67] (Figures

4.13(a) and (b)).

According to the continuum model of Lindhard [68], the critical angle for

ion channeling is approximately estimated to be

/ðZ

1=2

: ð4:12Þ

(a)

(b)

–20 –10 0 10 20 30 40 50 60

0.4

0.8

1.6

2.4

2.8

3.2

3.6

Tilt an

le (de

rees)

Relative intensity

0˚

(001)

20˚

(114)

36˚

(112)

54˚

(111)

Figure 4.13 (a) Relative SE intensity for the FIB-SIM as a function of the

Ga ion beam angle tilted on a Cu crystalline surface [66]. (b) Incident angle

dependence of SE yield  of a Cu crystalline due to the impacts of Ar ions

with the energies of 30 keV (1) and 20 keV (2) [67]. (Adapted from [66].)

Focused ion beam systems110

The experimental 9

values mostly range from 5 to about 20

, which

are sufﬁciently larger than the FIB focusing angles of a few mrad. A clear

contrast, as shown in Figure 4.12, has been obtained after removing the

native oxide surface layer with a high SE yield. On SIM imaging, channeling

contrast has been mostly observed for metals such as Al, Cu, Fe, Ni, and Au,

but not for semiconductors of Si, GaAs, etc. At ﬁrst sight one might expect

severe disruption of the crystalline caused by the ion beam to amorphize the

surface. A plausible explanation by R. Shimizu (Osaka Institute of Tech-

nology, Pers. Comm.) is that many metals remain crystalline as a result of

fast recrystallization under ion bombardment at room temperature.

4.3.4 Spatial resolution and information depth

The SEs have information along their own trajectories in addition to the

trajectories of the projectiles that excite the SEs of interest. Most of the SEs

are produced in the surface layer, as they are independent of the incident

probe. Under electron impacts, the SEs consist of two components which

correspond to the SEs excited by PEs (primary electrons) and BSEs (back-

scattered electrons), respectively. The SEs excited by PEs have a small spatial

spread of < 10 nm, but the SEs excited by BSEs have a large one of several

tens to several hundreds of nanometers, decreasing PE energy. Under the Ga

ion impacts, the spatial spread of SE information is roughly as small as

10 nm, decreasing with an increasing Z

Figures 4.14(a) and (b) show birthplaces of the SEs that have escaped from

the surface at various beam incident positions near the Al–Au boundary

[4 ,55]. Incident probes for SIM and SEM imaging are assumed to be the

point beams of 30 keV Ga ions and 10 keV electrons, respectively. The SEs

for the Ga ion impact are classiﬁed into three types because of the collision

partners: projectile ion (black), recoiled target atom (white), and cascade

electron (gray). It can be observed that the three components contribute

equally to the SE yield for the low-Z

sample (Al). For the high- Z

sample

(Au), however, the electron excitation by the projectile ions dominates those

by both the recoiled target atom and the cascade electrons. A large number of

SE birthplaces for both ion and electron impacts are < 5 nm in depth. For the

ion impacts, a laterally narrow spread of the SE birthplace shown in Figure

4.14(a) is caused by the short ranges of the projectile ions, recoiled target

atoms, and cascade electrons. Under electron impacts, on the other hand, the

SEs excited by PEs have a small lateral spread of < 10 nm, but the SEs

excited by BSEs have a large one of several hundreds of nm, as mentioned

above [34].

Imaging using electrons and ion beams 111

Assuming the SIM and SEM observations of an FIB milled cross section,

we consider a layered sample to have an Al–Au–Al structure, the cross sec-

tion of which is tilted by /4 radians with regard to their probe incident

directions, as shown in an inset in Figure 4.15 [4]. All the yields of SEs for

Depth (nm)

Lateral (nm)

(a)

(b)

Figure 4.14 Birthplaces of SEs that have escaped from the surface (made of

Au and Al by halves) due to the beam impacts of (a) 30 keV Ga ions and (b)

10 keV electrons [4,55]. The arrows indicate the impact points on the surface.

The number of projectiles is 10

. In (a), the SEs are excited by projectile ion

(black), recoiled target atom (white), and cascade electron (gray).

Focused ion beam systems112

π/4

Scan

30 keV Ga ions or 10 keV electrons

0.8

–30 –20 –10 0 10 20 30

Au Al

(b2)

0.2

0.4

0.6

0.8

–30 –20 –10 0 10 20 30

Beam

osition (nm)

Al Au

(b1)

–150 –100 –50 0 50 100 150

Beam position (nm)

Au Al

(a2)

0.2

0.4

0.6

0.8

–150 –100 –50 0 50 100 150

Beam position (nm)

Al Au

(a1)

γ (30 keV Ga)

δ (10 keV e)

η (10 keV e)

γ (30 keV Ga)

δ (10 keV e)

η (10 keV e)

,  and 

Figure 4.15 Line proﬁles of SE yield  for 30 keV Ga ions and SE yield – and

BSE yield · for 10 keV electrons as a function of the beam scanning position

for /4-tilted sample revealing a cross-sectional Al-Au-Al structure [4]; (a1)

near the boundary (P

) between Al and Au and (a2) near the boundary (P

)

between Au and Al; (b1) and (b2) magniﬁed proﬁles at the boundaries P

and P

corresponding to the proﬁles of (a1) and (a2), respectively.

Imaging using electrons and ion beams 113

30 keV Ga ion impact, and SEs and BSEs for 10 keV electron impact are

plotted as a function of the ion or electron scanning position in Figure 4.15.

The  proﬁle sharply changes at both the boundaries of Al–Au (P

) and

Au–Al (P

) and is roughly symmetric between the boundaries due to the

substructure effects. It is predicted that SEM images using the SE and BSE

signals show a combination of an indistinct or dull change of the image

brightness at the Al–Au boundary (P

) with a distinct or sharp change at the

Au–Al boundary (P

). Another interesting point is that the beam positions

corresponding to the half contrast of the line proﬁles at the boundaries are

shifted from the exact points of P

and P

. Therefore, SIM imaging is better

in spatial resolution for the structure/material measurements than SEM

imaging. Decreasing the electron impact energy is a direct method for

reducing the depth proﬁles of SEs.

4.3.5 Relevant subjects

This section discusses the relevant subjects on the SIM imaging, i.e., charging

up for insulator samples and image deterioration due to sputtering.

Charging up

For insulator samples (including electrically ﬂoated conductors), charge will

build up on the surfaces while the primary ion beam (i.e., Ga

FIB) is being

scanned across them. The charging up causes unwanted deﬂections of the

primary beam as well as SEs emitted from the sample. The former deﬂection

reduces the positional accuracy of the SIM imaging and FIB milling. The

latter deﬂection makes the SE detection yield ﬂuctuate with the ion bom-

barding time and/or the bombarding position. Then, the image intensity is

locally or wholly modulated to become too dark or too bright. There are two

ways of solving this problem: (i) coat the samples with a conductor (e.g., Au,

Pt, C), or (ii) an electron beam ﬂood to compensate for the positive charging

up. The other method is ultra-violet (UV) irradiation, where the light pene-

tration induces a conductive modulation in the semiconductor surfaces.

Coating is usually done using a thin ﬁlm coater (or evaporator). As a

variation of the external coating, an FIB assisted deposition (FIB-AD) of

conductor ﬁlms is used to do the selective coating and to connect the char-

ging site to a known ground site. In the case of the use of electron ﬂooding of

low energies to neutralize the incident positive charge on the sample, SEs are

no longer used as the SIM imaging signal because no discrimination can be

made between the SEs and the ﬂood electrons. Positive secondary ions (SIs)

Focused ion beam systems114

are now used instead of the SEs, but the SI intensity is rather smaller than the

SE one to deteriorate the SIM image quality. In the FIB cross-sectioning of

nonconductive materials, for example, the rough milling using a high-current

FIB of up to 30 nA requires the built-up charge to be dissipated using the

methods mentioned above. In the ﬁnal milling using a low-current FIB of less

than 100 pA, on the other hand, little charging up is observed if the milled

surface connects to somewhere at ground level. The reason is that the

implanted Ga atoms make the surface conductive. Even a high-resistivity

surface is generous with the scanning beam of low current.

In the SEM imaging for the insulator samples, on the other hand, the

primary electron energy is popularly adjusted to some speciﬁed energy in the

range of 200 eV to 2 keV (depending on the sample) so as to balance the SE

yield. This method is not applied to the SIM imaging using the positive ion

probe. The SE emission does not decrease the positive charges building up on

the surface, but does increase them.

Deterioration in the SIM image quality due to sputtering

To form a good image, there must be resolution, contrast, and adequate

signal-to-noise ratio (S/N). Good images are essentially a system issue, i.e.,

the instrument plus sample. Unnecessary ion bombardment must be reduced

to minimize the unavoidable inﬂuence of ion sputtering and redeposition of

sputtered atoms.

In order to obtain an image of a given S/N, the beam must dwell at points

on the sample for a time t

to collect enough SE quanta. The number of

detected SE quanta per pixel is given by

¼ I

·=e; ð4:13Þ

where  is the SE yield, · is the detection efﬁciency of SEs (taken to be 1 to

simplify the model), and e is the electron charge. The average human eye

requires an S/N value of at least 5. Considering a two-step process where by

primary ions strike a sample and generate SEs under Poisson statistics, the

number of incident ions per pixel N

ion

forms the image at S/N given by [6,10]

ðS=NÞ¼f=ð1 þ Þg

1=2

ion

1=2

: ð4:14Þ

Assuming · ¼1 and  ¼2, the image at S/N ¼10 requires N

ion

¼150, i.e.,

¼24 msatI

¼1 pA, or an image acquisition time of 6.3 s for a 512 · 512

pixel image, and subjects the surface to erosion by 150Y atoms. Here, Y is the

sputtering yield, i.e., about 1–10 for 30 keV Ga ion bombardment, depending

on the target.

Imaging using electrons and ion beams 115

Looking closer at the sputtering spot, it can be seen that a scanning FIB

forms a local slope on the bombarding spot due to the difference in ion dose

between the beam head and tail positions in the scanning direction [69]. The

slope angle strongly varies with the scanning speed, the beam density, the

beam size, sputtering yield, and redeposition. Since a change in the slope

angle corresponds to that in the beam incident angle, the beam scanning

conditions also inﬂuence the parameters  and Y. In high-resolution SIM

imaging, a single slow scan is usually superior to averaging frames to enhance

the S/N ratio of the image.

In SEM imaging, we don’t usually need to care about the sputtering, but it

does meet a contamination problem. Hydrocarbon molecules on the sample

surface are cracked and polymerized by the knock-on and/or ionization

processes. A carbon-rich contamination ﬁlm grows over the electron scan-

ning area. At a thickness of the order of 10 nm, the SE emission will be of

carbon and the irradiated areas will appear darker in the SEM image.

4.4 Some applications of FIB linked with modern SEM/STEM/TEM

4.4.1 FIB micro-sampling and micro-fabrication

In FIB applications of micro-machining and device transplantation, FIB

sputtering plays the role of ‘‘cutter’’ and FIB induced deposition does the role

of ‘‘attacher’’ [ 70]. One of their successful uses is an FIB micro(m)-sampling for

site-speciﬁc SEM/STEM/TEM inspection. The micro-samples are dug out

by FIB milling without cleaving a wafer or a chip [71,72]. Figure 4.16 shows a

schematic ﬂow of the m-sampling for transmission electron microscope

(TEM) / scanning TEM (STEM) inspection. First, the m-sample (or m-wedge)

of interest (typically 3 mm wide · 13 mm long · 7 mm deep) is FIB dug-out,

leaving a micro-bridge (or a microbeam) to support the m-sample. Second, a

built-in metal needle (or m-probe) is manipulated into position for lifting the

m-sample. FIB assist-deposition (FIB-AD) is used to bond (or weld) the

needle to the m-sample. Further FIB milling of the micro-bridge separates the

m-sample from the wafer or chip. The separated m-sample is transferred and

then ﬁxed onto the side of a partially cut TEM-grid using FIB-AD. Finally,

the m-sample is FIB thinned to a strip of about 0.1 mm thick and 7 mm long

and is ready for the TEM/STEM inspection. All the steps are accomplished

under vacuum in the FIB system. SIM images allow monitoring of each step

in live time thereby assisting the operator. Advantages of the FIB m-sampling

for TEM/STEM are as follows: (i) the wafer is left intact, (ii) FIB rethinning

is possible after TEM/STEM observation, (iii) it is possible to prepare the

Focused ion beam systems116

(a)

(d)

(e)

(f)

(b) (c)

(

) (h)

µ-bridge

probe

W-deposition

10 mm

15 mm

10 mm

20 mm

10 mm

m-sample

Figure 4.16 Schematic ﬂow of the m-sampling for TEM/STEM inspection.

(a) Tungsten (W) deposition as a pro tection layer on the site-speciﬁc region

of interest, (b) FIB trench-milling to dig out the m-sample, leaving a m-bridge

(or a microbeam) to support the m-sample, (c) FIB cutting the m-sample

bottom, (d) bonding the m-sample to the m-probe, (e) FIB cutting the

m-bridge to release the m-sample from the wafer or chip, and (f) lifting the

m-sample to transfer onto TEM grid. All pictures in (a)–(f) are SIM images

to assist the operator. Pictures of (g) and (h) show 200 kV SEM images of

the m-sample attached on the TEM half-grid edge before and after the FIB

rethinning, respectively.

Imaging using electrons and ion beams 117