Yao N. Focused Ion Beam Systems: Basics and Applications

Подождите немного. Документ загружается.

[92] F. V. Nolﬁ, ed. Phase Transformation During Irradiation (London: Applied

Science Publishers, 1983).

[93] J. S. Williams. Mater. Sci. Eng. A, 253 (1998), 8.

[94] L. Ma. Micron, 35:4 (2004), 273.

[95] C. R. Hutchinson, R. E. Hackenberg and G. J. Shiﬂet. Ultramicrocopy, 94

(2003), 37.

[96] A. De Veirman and L. Weaver. Micron, 30 (1999), 213.

[97] S. Rubanov and P. R. Munroe. Mater. Lett. 4142 (2002), 1.

[98] J. F. Walker and R. F. Broom. Inst. Phys. Conf. Ser., 157 (1997), 473.

[99] B. W. Kempshall, L. A. Giannuzzi, B. I. Prenitzer, F. A. Stevie and S. X. Da.

J. Vac. Sci. Technol. B, 20 (2002), 286.

[100] N. I. Kato, N. Miura and N. Tsutsui. J. Vac. Sci. A, 16 (1998), 1127.

[101] N. Miura, K. Tsujimato, R. Kanehara, N. Tsutsui and S. Tsuji. Proc. 22nd Int.

Symp. Testing and Failure Analysis (1995), 353.

[102] T. Ishitani, H. Koike, T. Yaguchi and T. Kamino. J. Vac. Sci. Technol. B, 1

(1998), 1907.

[103] R. M. Langford and A. K. Petford-Long. Proc. 12th European Microscopy

Congress, Vol. II (2000), 557.

[104] R. L. Martens, D. J. Larson, T. F. Kelly, A. Cerezo, P. H. Clifton and N. Tabat.

Micros. Microanal., 6, (suppl. 2) (2000), 522.

[105] C. A. Ferryman, J. E. Fulghum, L. A. Giannuzzi and F. Stevie. Surf. Interface

Anal., 33 (2002), 907.

[106] M. Tanaka, K. Furuya and T. Saito. Jpn. J. Appl. Phys., 37 (1998), 7010.

[107] J. P. McCaffrey, M. W. Phaneuf and L. D. Madsen. UltraMicroscopy, 87

(2001), 97.

[108] H. Bender and P. Russel. Inst. Phys. Conf. Ser., 157 (1997), 465.

[109] N. I. Kato, N. Miura and N. Tsutsui. J. Vac. Sci. Technol. A, 16 (1998), 1127.

[110] D. Sutton, S. M. Parle and S. B. Newcomb. Inst. Phys. Conf. Ser., 168 (2002), 377.

[111] B. D. Huey and R. M. Langford. Nanotechnology, 14 (2003), 409.

[112] B. Basnor, A. Lugstein, H. Wanzenbeck et al. J. Vac. Sci. Technol. B, 21

(2003), 209.

[113] W. Brezna, H. Wanzenbock, A. Lugstein et al. Physica E, 19 (2003), 178.

[114] R. J. Young, P. D. Carleson, T. Hunt and J. Walker. Proc. From the 24th Int.

Symp. Testing and Failure Analysis, (1998).

[115] R. M. Langford, D. Ozkaya, C. Kaufman et al. Electron Microsc. Anal. Inst.

Phys. Conf. Ser. 168 (2001), 437.

[116] R. M. Langford, D. Ozkaya, B. Huey, A. K. Petford-Long. Microsc. Semicond.

Mater. Inst. Phys. Conf. Ser., 169 (2001), 511.

[117] Z. Wang, I. K. Takeharu, T. Hirayama et al. J. Vac. Sci. Technol. B, 21 (2003),

2155.

[118] A. Yamaguchi, M. Shibata and T. Hasshinga. J. Vac. Sci. Technol., B, 11

(1993), 2016.

[119] A. Leslie, K. L. Pey, K. S. Sim, M. T. F. Beh and G. P. Goh. Proc. 21st Int.

Symp. Testing and Failure Analysis, 353 (1995).

[120] P. E. Russell, T. J. Stark, D. P. Grifﬁs, J. R. Phillips and K. F. Jarausch. J. Vac.

Sci. Technol. B, 16 (1998), 2494.

[121] Y. Sugimoto, M. Taneya, H. Hidaka and K. Akita. J. Appl. Phys., 68 (1990),

2392.

[122] A. Yamaguchi and T. Nishikawa. J. Vac. Sci. Technol. B, 13 (1998), 962.

Focused ion beam systems248

[123] D. Cooper, A. C. Twitchett, R. E. Dunin-Borkowski et al. Proc. of 13th

European Microscopy Congress, Vol. II (2004), 783.

[124] C. L. Collins, M. MacKemzie, A. J. Craven et al. Proc. 13th European

Microscopy Congress, Vol. II (2004), 775.

[125] Z. Huang. J. Microsc., 215 (2004) 219.

[126] S. Rajsiri, B. W. Kempsh all, S. M. Schwarz and L. A. Giannuzzi. Proc.

Microsc. Microanal., 99 (2002), 50.

[127] T. L. M atteson, S. W. Schwarz, E. C. Houge, B. W. Kempshall and

L. A. Giannuzzi. J. Electron Mater., 31 (2002), 31.

[128] J. M. Cairney and P. R. Munroe. Micron, 34 (2003), 97.

Preparation for physico-chemical analysis 249

In-situ sample manipulation and imaging

t. kamino and t. yaguchi

Hitachi Science Systems

t. ohnishi and t. ishitani

Hitachi High-technologies Corporation

9.1 Process of TEM sample preparation

9.1.1 Pre-thinning

Since a focused Ga ion beam with a maximum diameter at the sub-

micrometer level is used in the FIB technique, the sputtering speed is not so

high as in other ion milling techniques [1,2]. To reduce the FIB milling time,

pre-thinning of the sample using a polisher or cutter is required. There is no

deﬁnite sample target thickness to be achieved by pre-thinning, but a typical

thickness used for FIB milling is in the range of 50 to 100 mm in most cases.

Preparing thinner samples promises shorter FIB milling times but we should

be careful not to cause any structural and chemical changes in the sample at

this stage [3]. Note that the thinner samples may become more stressed

during the pre-thinning process than the thicker ones. There are various

techniques and tools to be used for the pre-thinning process. A micro-

cleaving system is widely used for pre-thinning of the devices with single

crystal Si or GaAs substrates. The system allows cleaving of the sample

including the circuit of interest quite easily and precisely and the cleaved

sample may be directly FIB milled or ﬁrst sliced to 10 to 20 mm using a ﬁne

blade prior to the FIB milling. In the case of polycrystalline materials such

as metals, ceramics, and their composites, mechanical-polishing or dicing

can be done. Relatively soft materials such as polymers, rubbers, and papers

can be sliced with a razor blade. No special tool is required for these

materials. Nonconductive materials such as glasses, polymer, and ceramics

should be metal coated prior to FIB milling. Carbon or metals such as Pt,

Focused Ion Beam Systems: Basics and Applications, ed. N. Yao.

Published by Cambridge University Press. ª Cambridge University Press 2007.

250

Au, or Pd are used for the coating. A coating thickness of 20–30 nm is

sufﬁcient.

9.1.2 Mount on a sample support

The pre-thinned sample is then sectioned into a plate about 1–2 mm wide and

2.5–3 mm long and directly mounted onto the FIB sample holder. If the

sample is smaller than that, it should be sectioned into small plates and glued

to a sample support made from a semicircular metal disk [4 –5]. Conductive

paste such as carbon paste or silver paste can be used to glue the sample to

the support disk. Mo, Cu, and Al plates with a thickness of 30–50 mm are

available as the support material. The choice of the materials should be made

from the analytical point of view because the system peak from the support

material may cause trouble in elemental analysis using the EDX system. In

the case of particles or ﬁbers, the sample may be embedded in epoxy. The

embedded sample is sectioned into the foil with a thickness of 100 nm or less,

and then glued to an edge of the support. Don’t forget to coat the whole

glued sample with evaporated carbon or metal layers. This is very important

to avoid any instability of the ion beam position caused by the charged

sample during the FIB milling.

9.1.3 Surface treatment

In TEM sample preparation, having a smooth surface is very important to

get enough uniformity in the thickness. Always remember that the smoother

the surface the better the uniformity in the thickness of the sample. There-

fore, it is strongly recommended to keep the sample surface smooth before

FIB milling. Methods such as polishing or slicing can be applied to make

the sample surface smooth but be careful to avoid introducing any

mechanical damage with this process. It is recommended that if you want

to observe the cross-sectional structure of the top surface, omit the

mechanical polish or metal deposition, but instead add a carbon coating

using a vacuum evaporator. The recommended thickness of the carbon

layer is 20–40 nm. FIB assisted metal deposition will follow as the ﬁnal

surface treatment. Deposited metal smoothes the rough surface of the

sample. If you don’t see any sharp protuberance on the surface in an SIM

image after the metal deposition, it means the deposition is sufﬁcient to

obtain a uniformly thin sample. A deposition thickness of 0.5–2 mmis

usually sufﬁcient.

In-situ sample manipulation and imaging 251

9.1.4 Thinning

Generally, thinning of the sample is carried out in three steps. First, we use a

large beam to mill the sample as quickly as possible (rough milling). When

the sample is thinned to a thickness of 2–5 mm, the ion beam current is

decreased to reduce the sample damage caused by the FIB milling (medium

milling). Finally, a small beam is used to ﬁnish the thin sample to minimize

ion beam irradiation damage (ﬁnal milling). When a Ga ion beam with an

acceleration voltage of 30 kV is employed, the average ion beam diameters/

ion beam currents for rough, medium, and ﬁnal millings are 60 nm/10 nA,

100 nm/2 nA, and 40 nm/30 pA, respectively. It is usual for samples for TEM

observation to be thinned to a thickness of 50 nm to 0.2 mm but the best

thickness depends on what you want from the sample. A thickness of less

than 0.1 mm is needed for high-resolution TEM image observation and 0.1–

0.3 mm for EDX or EELS analysis. Note that the required sample prepara-

tion time is proportional to the size of the milling area and the sample

damage is also nearly proportional to the total milling time. Don’t forget to

minimize unnecessary damage introduced to the sample. Also don’t observe

the cross-sectional SIM image during preparation. If you do, the sample may

be damaged and what you observe in the TEM may not be the original one

but the damaged one. Ga ion beams can easily change or damage the original

structure of the sample.

9.2 Sample damage

Since FIB involves bombarding a sample with energetic Ga ions, the sample

gets the ion implantation and the damage caused by the bombardment [6–7].

The Ga implantation may reduce the accuracy of the quantitative elemental

analysis and irradiation damage may confuse the structural analysis. Ga

concentration in the sample and structural change caused by the irradiation

damage depend on the substances to be milled and the ion milling condition.

At the ﬁnal stage of the FIB milling, the cross section is approximated to be

in a steady state, where as many implanted atoms are removed by sputtering

as are replenished by implanted Ga. This can be expressed as

¼ Y

imp

; ð9:1Þ

and

¼ R

GaM



; ð9:2Þ

Focused ion beam systems252

where S

and S

are the partial sputtering yields of components Ga and M,

respectively; S is the total sputtering yield (¼S

þS

); Y

imp

is the implan-

tation yield; C

and C

(¼1 C

) are the atomic concentrations of their

components near the target surface; and R

Ga M

is the enhancement factor in

the Ga preferential sputtering for the host matrix M (i.e., the ratio of the

probabilities of a Ga atom near the surface to be sputtered to that of an M

atom. The R

Ga M

value may not be unity because of differences in surface

binding energies, sputter escape depths, and energy transfers within the

collision cascade.

From (9.1) and (9.2), C

is derived as

GaM

S=Y

imp

 1



þ 1

: ð9:3Þ

At R

Ga – M

¼1 it simply yields to

imp

: ð9:4Þ

Figure 9.1 shows the basic parameters of Ga FIB cross-sectioning of the

sample M, sputtering yield S (¼S

þS

), implantation yield (Y

imp

), and

backscattering yield (1Y

imp

In the FIB prepared TEM sample, two trenches are milled, leaving a

narrow strip of about 100 nm. The specimen thickness is usually larger than

the transverse depth of incident Ga ions, typically about 20 nm for Si and

10 nm for W when a high voltage of 30 kV is applied to accelerate the Ga ion

Ga ion

Sputtering yield:

S (= S

+ S

)

Backscattering yield:

1 – Y

imp

Implantation

yield: Y

imp

Figure 9.1 Basic parameters of Ga FIB cross-sectioning of a sample M,

sputtering yield S (¼S

þS

), implantation yield (Y

imp

), and back-

scattering yield (1Y

imp

In-situ sample manipulation and imaging 253

beam. The average Ga concentration (C

) in a sample of thickness t is

expressed as

t ; av

þ N

;ð 9:5 Þ

and

when C



;ð 9:6 Þ

where N is the atomic density of the sample, and C

is the Ga concentration in

the implanted area.

9.2.1 Implantation of Ga ions in the thinned sample

Figure 9. 2 shows the calculated Ga concentration as a function of sample

thickness in W and Si samples cross-sectioned at 30 kV. In the 100 nm thick W

and Si samples, Ga concentrations of 1.5% and 0.5% are expected. Since the

implanted Ga ions in the cross-sectioned sample are concentrated near the

surface [ 8–10 ], the Ga concentration decreases with an increase in sample

thickness. In fact, the Ga concentrations in 200 nm thick W and Si samples are

0.5% and 0.2%, respectively. If you want to minimize the relative intensity of

the Ga peak in EDX analysis, do not thin your samples too much.

The concentration of Ga ions in the samples was measured experimentally.

Figure 9.3(a) shows an STEM image of a 100 nm thick W/Si multi-layer

sample and Figures 9.3(b and c) shows the results of EDX analysis of the two

layers. The sample was prepared at 30 kV. In the ﬁgures, besides the W-M

0 50 100 150 200

1.0

2.0

3.0

4.0

Sam

le thickness (nm)

Ga concentration (at.%)

Figure 9.2 Calculated Ga concentration in a 100 nm thick W and Si sample

milled at 30 kV as a functi on of sample thickness.

Focused ion beam systems254

and Si-K peaks, expanded Ga-K peaks are shown. Ga concentrations in the

W and Si layers calculated from the intensities of these peaks were 1.6% and

0.4%, respectively. The measured values are in good agreement with the

theoretical value in this case. The depth of Ga implantation is proportional

to the energy of the incident Ga ion. It suggests that if you have too high a

Ga concentration in the sample, one of the effective remedies against this

problem is to use a low energy ion beam for ﬁnal milling.

Monte Carlo simulation of 100 Ga ion trajectories in Mg for an accelera-

tion energy of 40 kV (a), 30 kV (b), and 10 kV (c) are shown in Figure 9.4.

The depth of area with high Ga concentration for the samples milled at

40 kV, 30 kV, and 10 kV are roughly 20 nm, 15 nm, and 10 nm, respectively. A

practical example is shown in Figure 9.5. The sample is Mg-Al alloy and ﬁnal

milling was carried out at 40 kV and 10 kV. Mg alloy is one of the most

sensitive materials to Ga ion beam irradiation, so the ion beam employed in

the milling was relatively small and the milling time was longer. The calcu-

lated Ga concentrations for 40 kV and 10 kV are 4% and 1.3%, respectively.

9.2.2 Structural changes caused by the Ga ion beam irradiation

We should not forget that Ga ion beam irradiation can substantially alter the

structure [11] and composition of the surfaces of a sample through dis-

placement damage. The structural changes of the surfaces depend on the

milling conditions and the nature or composition of the material. Details will

W la

1.0 1.02.0 2.09.2 9.2 K

Si la

(a)

(b)

0.1 µm

W-M W-Lα W-Lβ Si-K

Ga-K

(×70)

(×40)

(c)

Figure 9.3 STEM image of a 100 nm thick W/Si layer and EDX spectra from

the layers. (b) W-M peak and 40 times enlarged Ga-K peak, (c) Si-K peak

and 70 times enlarged Ga-K peak.

In-situ sample manipulation and imaging 255

not be given here but you should be aware that choosing the correct glancing

angle is very important to minimize the damage caused to a sample. In

general, the lower the glancing angle the less the damage; on the other hand,

the lower the glancing angle the lower the milling rate. In practice, the

glancing angle is set from both the damage and milling speed points of view.

We usually set lower angles of around 2



for lighter elements and around 6



for heavier elements. An example of the damage caused by Ga ion beam

irradiation and a result of the remedy is shown in Figure 9.6. The sample is

an InP single crystal which was cross-sectioned at the acceleration voltages of

40 kV and 10 kV. Bright dots on the surface of the sample milled at 40 kV

(a)

(b)

(c)

20 nm

Figure 9.4 Monte Carlo simulation of 100 Ga ion trajectories in Mg at

(a) 40 kV, (b) 30 kV, and (c) 10 kV at a glancing an gle of 5



40kV

10kV

(a) (b)

Figure 9.5 EDX analysis of a 100 nm thick TEM sample of Mg-Al alloy

prepared by Ga FIB milling at (a) 40 kV and (b) 10 kV.

Focused ion beam systems256

(Figure 9.6(a)) are the particles of redeposited indium. On the sample cross-

sectioned at 10 kV, however, no redeposited particles are observed (Figure

9.6(b)). The surface looks clean and smooth. Comparing the differences

between these images it can be said that the employment of a low energy ion

beam for ﬁnal milling is quite effective in reducing redeposition.

In the case of metals, a different type of damage happens. A typical

example of the damage to a metal sample caused by high energy Ga ion beam

irradiation and its countermeasure are shown in Figure 9.7. The sample is

Mg-Al alloy thinned at both 40 kV and 10 kV. In the TEM image of the alloy

thinned at 40 kV (Figure 9.7 (a)), many crystal lattice defects are observed as

dark dots. In the TEM image of the sample thinned at 10 kV, however, no

500nm

40kV

10kV

500nm

500 nm

(a)

(b)

40 kV

10 kV

500 nm

Figu re 9 .6 SEM imag es of InP cross- sectioned at (a) 40 kV and (b) 10 kV.

Bright dots appeari ng on the sampl e cross-sect ioned at 40 kV are the re-

deposited In particles. No redeposition of In occurred at 10 kV.

200nm

40kV

10kV

200nm

200 nm

(a)

(b)

40 kV

10 kV

Figure 9.7 TEM images of Mg-Al alloy thinned at (a) 40 kV and (b) 10 kV.

The dark dots in (a) are the point defects caused by Ga irradiation. No

defects can be seen in (b).

In-situ sample manipulation and imaging 257