Yao N. Focused Ion Beam Systems: Basics and Applications

Подождите немного. Документ загружается.

such defects are observed (Figure 9.7(b)). From the TEM image quality of

the sample thinned at 10 kV, it is again obvious that the sample can be cleanly

and smoothly thinned using a lower energy ion beam. These experimental

results demonstrate well the importance of choosing a low energy ion beam

for ﬁnal milling in TEM sample preparation.

But these data does not tell us how deep samples will get damaged. When

studying the surface region of a sample, it is worth knowing the depth of

damage on your sample. A measurement of the thickness of the damage that

occurred on a single crystal Si is shown in Figure 9.8 . The samples were ﬁnal

milled at 30 kV and 10 kV, and the thickness of the amorphous layers formed

as the result of ion beam irradiation damage are 21 nm and 8 nm, respec-

tively. The results of the measurement reveal that the damage depth is almost

proportional to the energy of the ion beam employed in the ﬁnal milling. In

case you still have a damage problem, even after reducing the ion beam

energy to the practical limit of the FIB system, additional thinning by a

conventional method, e.g., Ar ion milling, will work.

9.3 High-resolution TEM image observation of

FIB prepared samples

Although a certain rate of damage occurs on the cross-sectioned surfaces of

an FIB milled thin sample, high-resolution TEM image observation is pos-

sible in most cases if the sample is prepared properly. Figure 9.9 shows a

high-resolution TEM image of a semiconductor device. The sample was

21nm

30 kV

50nm

8nm

10 kV

50nm

21nm

30 kV

50nm

8nm

10 kV

50nm

21 nm

30 kV

50 nm

(a)

8 nm

10 kV

(b)

50 nm

Figure 9.8 Cross-sectional dark-ﬁeld STEM images of amorphous layers

formed during FIB milling. 21 nm and 8 nm thick amorphous layers were

formed at (a) 30 kV and (b) 10 kV, respectively. C and W in the photographs

indicate carbon and tungsten deposition layers.

Focused ion beam systems258

thinned to a thickness of 80 nm at 10 kV. Although it has a 16 nm thick

amorphous layer on both surfaces, as demonstrated in the previous section,

crystal lattice fringes obtained with the electron beam incident along the (110)

plane of the silicon and individual Si atoms at the interface between the Si

substrate and the silicon oxide above it are clearly observed. This kind of

image quality, however, only became possible after the FIB milling technique

at low energy, e.g., 10 kV, was routinely adopted for TEM sample prepara-

tion. Figure 9.10 shows a high-resolution TEM image of hydroxyapatite,

which is one of the most sensitive materials to electron beam irradiation, and

5 nm

Figure 9.9 Cross-sectional high-resolution TEM image of an Si substrate

and silicon oxide layer of a semiconductor device. The sample was thinned

at 10 kV to the thickness of 80 nm. The atomic array at the top of the

substrate and the ﬁne structure of the amorphous silicon oxide layer can be

clearly observed.

(100)

0.817 nm

Figure 9.10 High-resolution TEM image of a hydroxyapatite crystal of

human tooth. Lattice fringes of (100) crystal plane with a separation

distance of 0.817 nm can be clearly observed.

In-situ sample manipulation and imaging 259

therefore considered to be difﬁcult to thin using the FIB technique without

causing damage. However, crystal lattice fringes of the (100) plane with a

distance of 0.817 nm and other ﬁne structures can be clearly observed.

From the image, it is evident that the quality of the high-resolution TEM

image does not suffer from the damage that occurred during the FIB milling.

9.4 FIB capability in TEM sample preparation

Recent developments and improvements of the FIB techniques for TEM

sample preparation have enabled us to observe ﬁne structures of many

composite materials [12–13]. Figure 9.11 shows a low-magniﬁcation cross-

sectional TEM image of a tungsten (W) ﬁber embedded in zirconia (ZrO

)

(Figure 9. 11 (a)) and a high-resolution TEM image with a nano-probe electron

diffraction pattern of the interface (Figure 9.11(b)). Since the material is a very

hard metal and ceramic composite material, it was impossible to prepare an

adequate TEM sample for the study of the interface between the two

materials. As is shown here in the sample prepared by the FIB technique,

however, we can observe the whole structure of the cross section of the W

ﬁber and the interface structure of the W–ZrO

interface at an atomic

resolution.

The capability of the FIB technique in practice is greater than expected.

Figure 9.12 shows a TEM image of a toner particle thinned by the FIB

technique. The sample was coated with an FIB assisted W deposition layer

prior to the thinning process and thinned by FIB milling at an acceleration

voltage of 30 kV. Although the material has a very low melting point and was

therefore considered impossible to thin using the FIB technique, the original

500nm

interface

2 n m

inte rface

ZrO

500 nm

(a)

(b)

Interface

ZrO

2 nm

ZrO

Interface

Figure 9.11 (a) Low-magniﬁcation TEM image of a cross-sectioned W ﬁber

embedded in ZrO

and (b) a high-resolution TEM image of the interface

with the inser tion of electron diffraction patterns from the interface.

Focused ion beam systems260

shape of the particle and its internal structure are well preserved. So far the

thin sectioning of this kind of material with a very low melting point has only

been possible with a cryo-sectioning technique that requires special training.

Figure 9.13 shows a STEM image of a thin section of parafﬁn embedded

human lung tissue which was prepared for characterization of inhaled

mineral ﬁbers. The sample was extracted from the parafﬁn block of tissue

and thinned to a thickness of 5 mm at an acceleration voltage of 30 kV.

Mineral ﬁbers in the tissue (indicated by the arrows) can be clearly observed.

2 m

Figure 9.12 Bright ﬁeld STEM image of a toner particle thinned at 40 kV.

The toner is covered with an FIB assisted W deposition layer. The original

shape of the toner particle and the inside structures are well preserved.

3.00m

HD-2000 200kV ×10.0k TE

Figure 9.13 STEM image of a parafﬁn embedded human lung. A small piece

of the sample was extracted and thinned at 30 kV.

In-situ sample manipulation and imaging 261

From these images it can be said that the FIB technique has made TEM

sample preparation of the materials of low-melting point much easier than

before.

9.5 TEM sample preparation from a speciﬁc site

One of the advantages of the FIB technique compared with other conven-

tional TEM sample preparation techniques is the excellent positional accu-

racy in setting the area to be thinned. However, because structures are

becoming smaller and more complicated, TEM sample preparation from

speciﬁc sites in state of the art materials such as semiconductor devices is no

more easy than it was in the past. Various approaches to improve the posi-

tional accuracy are being developed and some have already been put to

effective use. One of the most reliable systems used for this purpose is a

dedicated FIB-STEM system. The system has a compatible sample holder so

that sample preparation in the FIB system and its observation in the STEM

are possible without reloading the sample [14].

Figure 9.14 shows the procedure for TEM sample preparation from a

speciﬁc site using the FIB-STEM system. In this method, the sample is milled

to a thickness of 3–5 mm and transferred to the STEM unit to observe STEM

and SEM images. The presence of the site in the sample can be conﬁrmed by

the STEM observation and the position of the site can be assured by the

SEM observation three dimensionally. The sample is then transferred back to

(a)

Ga+

3–5m

<0.1m

(b)

–

(d)

(c)

Figure 9.14 Pr ocedure for TEM sample preparation from a speciﬁc site.

(a) The sample is milled to a thickness of 3–5 mm. Then transferred to the

STEM unit for (b) STEM and (c) SEM observation to ﬁnd the site to be

thinned. (d) Finally, transferred to the FIB system to make a <0.1 mm thick

sample from the speciﬁc site.

Focused ion beam systems262

the FIB system to make a TEM sample exactly from the site to be char-

acterized. In this method, we use STEM for the observation of interior

structures so that we can minimize unwanted sample damage caused by Ga

ion beam irradiation. This is an important advantage when the structural

analysis of a speciﬁc site at atomic level resolution is required.

9.5.1 FIB lift-out technique

The FIB lift-out technique was developed for lifting out a thinned sample

directly from a bulky sample [ 15–16]. The general ﬂow chart of the lift-out

technique is shown in Figure 9.15. First, the sample is inserted into the FIB

system and the surrounding area of the site to be characterized is trench

milled (Figure 9.15(a)) and the site is thinned to electron-transparent thick-

ness (100 nm or less) (Figure 9.15(b)). The sample is then transferred to the

stage of a light microscope and the thinned sample is lifted out using a glass

rod (Figure 9.15(c)), and deposited onto a carbon coated TEM grid (Figure

9.15(d)). The technique may be applied to most of the substances used in

industrial materials. Although the procedure is simple, the samples prepared

by the technique have a clean surface and uniform thickness so that even

atomic level high-resolution TEM images can be observed. Since the tech-

nique does not require any mechanical polishing or dicing prior to the FIB

milling, the time required for the whole process is much shorter than that

required for conventional FIB techniques.

Probe

TEM

rid

(a)

(b)

(c)

(d)

Figure 9.15 General ﬂow chart of a lift-out technique. (a) Trench mill the

surrounding area, (b) thin the site to be characterized, (c) lift the thinned

sample out, (d) deposit the sample on a carbon coated TEM grid.

In-situ sample manipulation and imaging 263

This is an advantage when many samples need to be prepared in a short

time. However, since a glass rod is used for lifting and mounting the sample,

some experience is required to achieve a good success rate, and a better way

to mount and align the sample is desired. Figure 9.16(a) shows a trench

milled and cross-sectioned area, and the sample mounted on a carbon coated

TEM grid is shown in Figure 9.16(b).

9.5.2 FIB micro-sampling technique

This technique has also been developed for TEM sample preparation of a

speciﬁc site. The general ﬂow chart of the FIB micro-sampling technique is

shown in Figure 9.17. First, the area surrounding the site to be characterized is

trench milled (Figure 9. 17 ( a )) . S ec on d, c r o s s s ec ti on t h e s it e a n d p i ck u p a

sample piece (Figure 9.1 7(b)). Next the sample is mounted on a carrier (Figure

9.17(c)), and thinned using FIB milling (Figure 9.17(d)). All of the steps above

are accomplished under vacuum in the FIB system. Since the sample is

manipulated by an elaborate mechanical probe and bonded by FIB assisted

deposition, the success rate is very high and therefore it is used as one of the

standard TEM sampling techniques in a wide range of materials [17–18].

The technique has the following unique features:

1. No mech anical pre-th inning such as dicing or fracturi ng is necessa ry.

2. It is capab le of both cross- sectional and plan-v iew TEM micro -sampli ng.

3. Plan-vi ew TEM micr o-sampling from identical areas in cross- sectional TEM view

are pos sible, an d vice versa.

4. It is well suited for TEM obs ervation of magn etic mate rials because the sampl e is

so small and the inﬂuence of the magnetic ﬁe ld from the sample is negligibly

small.

5. Good results in quantitative EDX analysis can be expected because the sample is

so small and any material may be used as the sample carrier.

10um10um

10 m

(a)

(b)

Figure 9.16 Trench milled area and a thinned sample (a) and the sample

mounted on a carbon coated TEM grid (b).

Focused ion beam systems264

Fi gu r e 9.18(a) s hows a micro-sample mounted on a carrier and Fi gure 9.18(b)

shows a sample after thinning. A typical size of the micro-sample is 10–20 mm

wide, 10–15 m m high, and 2–5 m m thick. The technique can be applied to a

wide range of materials including semiconductor materials, metals, ceramics,

polymers, and various kinds of composite materials.

Figure 9.19(a) shows a micro-sample of a coating of a beverage can and

Figure 9.19(b) shows a magniﬁed STEM image of a framed area in (a). The

layers composed of an anti-bacterial coating, the printed layer and a pro-

tective layer of the can are thinned uniformly. Although the sample was

extracted upward by a mechanical probe bonded to the surface of the micro-

sample, no damage such as micro-cracking occurred at the interface between

Al and coating. As is demonstrated here, the adhesive force of the FIB

assisted metal deposition layer is strong enough to extract and carry the

micro-sample, even if the surface is covered with a nonconductive and plastic

polymer.

9.6 Summary and future prospects

Nanotechnology is based on well-designed and manufactured nanomaterials.

In the characterization of the nanomaterials, the structural and composi-

tional analyses of nanometer-scaled areas is essential. TEM is one of the

carrier

probe

Probe

(a)

(b)

(c)

(d)

carrier

Carrier

Figure 9.17 General ﬂow chart of an FIB micro-sampling technique.

(a) Trench mill the surrounding area, (b) cross-section and pick up using

a mechanical probe, (c) mount on a carrier, (d) Thin the site to be

characterized.

In-situ sample manipulation and imaging 265

most used instruments in this characterization, and sample preparation for

TEM observation is the ﬁrst and most important thing to do. Among several

TEM sample preparation methods in use, the FIB technique is distinguished

as the best method to prepare a TEM sample from a speciﬁc site. In fact, the

success rate in failure analysis of semiconductor devices using TEM

improved signiﬁcantly after the FIB technique was applied. In materials

characterization, however, new results bring new requirements, and the

requirements of the materials scientist are growing steadily more challenging.

Actually, quite a few of the latest requirements coming from high-tech

industry concern the capability of the single FIB technique, especially in

positional accuracy. The combination of FIB and STEM systems, however,

could improve the positional accuracy in TEM sample preparation drasti-

cally. In this method, electrons instead of Ga ions, are used for observation

of the sample; consequent damage during searching the site or positioning

the milling area is much less than that of conventional FIB technique.

(a)

(b)

10 m 10 m

Figure 9.18 FIB micro-sample of semiconductor device mounted on a

carrier (a) and a thinned micro-sample (b).

Protective W

layer

Print layer

Anti-bacterial

coating

(a)

(b)

Figure 9.19 (a) SEM image of a micro-sample of a coating of a beverage can

and (b) a dark ﬁeld STEM image of the framed area in (a).

Focused ion beam systems266

In materials science, the importance of TEM investigation is constantly

increasing and most investigations require site-speciﬁc sample preparation.

The future will throw up many challenges in this world, and FIB will

continue to be the best solution in the future as well.

References

[1] E. C. G. Kirk, D. A. Williams and H. Ahmed. Microsc. Semicond. Mater. Sci.,

Inst. Phys. Conf. Ser., 100 (1989) 501–6.

[2] T. Ishitani, H. Tsuboi, T. Yaguchi and H. Koike. J. Electron Microsc., 4 (1994),

322–6.

[3] A. R. Neureuther, C. Y. Liu and C. H. Ting. J. Vac. Sci. Technol., 16 (1979) ,

1767–71.

[4] Y. Kitano, Y. Fujikawa, T. Kamino, et al. J. Electron Microsc., 44 (1995),

410–13.

[5] Y. Kitano, Y. Fujikawa, T. Kamino, et al. J. Electron Microsc., 44 (1995),

376–83.

[6] T. Ishitani and T. Ohnishi. J. Vac. Sci. Technol. A, 9 (1991) , 3084–9.

[7] D. W. Susnitzky and K. D. Johnson. Proc. Microsc. Microanal., 4 (1998), 636–7.

[8] J. Melngailis. Electron Beam, X-ray and Ion-beam Submicrometer Lithographies

for Manufacturing, Proc. SPIE, 1465 (1991), 36–49.

[9] T. Ishitani. Jpn. J. Appl. Phys., 34 (1995), 3303–6.

[10] T. Ishitani, H. Koike, T. Yaguchi and T. Kamino. J. Vac. Sci. Technol. B, 16

(1998), 1907–13.

[11] W. Hauffe. Top. Appl. Phys., 64 (1991), 305–38.

[12] H. Saka, T. Kato, M. H. Hong, et al. Proc. Galvatech., (1995).

[13] H. Saka. The 10th Int. Symp. on Advanced Materials, (Tsukuba, Japan, 2003),

57–8.

[14] T. Ishitani, Y. Taniguchi, S. Isakozawa, et al. J. Vac. Sci. Technol. B, 16 (1998),

2532–7.

[15] F. A. Stevie, R. B. Irwin, T. L. Shofner et al. Characterization and Metrology:

International Conference (1998), 868–72.

[16] L. A. Giannuzzi, J. L. Drown, S. R. Brown, R. B. Irwin and F. A. Stevie.

Microsc. Res. and Tech., 41 (1998), 285–90.

[17] T. Yaguchi, T. Kamino, M. Sasaki, G. Barbezat and R. Urao. Microsc.

Microanal., 6 (2000), 218–23.

[18] T. Ohnishi, H. Koike, T. Ishitani, et al. Proc. 25th Int. Symp. Testing and Failure

Analysis, (1999), 449–53.

In-situ sample manipulation and imaging 267