Yao N. Focused Ion Beam Systems: Basics and Applications

Подождите немного. Документ загружается.

the required TEM observation resolution. In order to realize such micro-

processing, the process must be performed by a well-focused and compara-

tively low current ion beam. For this reason, processing cannot be completed

within a short time, and alteration of the ion beam irradiation position

caused by environment changes, such as the instrument room temperature,

during processing time cannot be ignored. Accordingly, technology to enable

the beam irradiation position to be corrected is required.

As shown in Figure 14.33, a mark is formed for position correction

within the same visual ﬁeld as the process region. The position of the

position correction mark is periodically veriﬁed with the observation ﬁeld

during processing. Moving the mark position in the observation ﬁeld

means moving the beam irradiation position according to the amount of move-

ment, allowing the beam irradiation position to be set at a ﬁxed position. As

a result, precision processing can be done even when the process time is long.

Automation

With the design rule miniaturization of integrated circuits, device structures

have become miniaturized and complex. As a result, high-resolution obser-

vation of device cross-sectional structures by TEM or STEM must increase

and multiple productions of TEM samples for observation are required.

In the past, the operator searched for problems and defective spots at the

site of device development or production processes, and the main trend was

in specimen production.

From this, however, utilization by production control is assumed, and ﬁxed

point observation is called for within the routine work of speciﬁed points on a

Figure 14.32 Example of TEM specimen production using the lift-out

method.

Focused ion beam systems378

wafer. In this case, a system is required in which the tool automatically detects

a targeted pattern on a wafer, and automatically produces a TEM specimen at

a set location. The remaining thickness of the specimen is 100 nm.

Figure 14.34 shows an example of evaluating STEM specimen auto pro-

duction for 200 mm wafers using the SMI3200 FIB tool, which has the same

performance as the SMI3050.

The following evaluation method was used:

1. Process targe t pos ition settings : the pro cess posit ion is set a t a dist ance of 1.7 mm

below the wire as shown in Figure 14.34(a). (Not e that since the deviat ion in the

x direction is sufﬁci ently small for proce ss wi dth, it is not a target for evaluat ion.)

2. Sample size: wid th of 10 mm, dep th of 7 mm, remai ning thickne ss of 160 nm. The

process ing tim e is app roximatel y 20 min for making one sampl e.

3. Thirty dies on top of the same wafer were selected and automa tically pro cessed.

The total process ing tim e was app roximatel y 10 hours.

Figure 14.34(b) shows the measured thickness of 30 lamellae. The fol-

lowing results were veriﬁed:

 standar d deviat ion of specimen ’s remaining thickne ss (3 ) ¼ 15.2 nm;

 standar d deviat ion of specimen pro duction pos ition (3 ) ¼ 43.6 nm.

We veriﬁed that no large error exists in the image if the specimen thickness

has a deviation range of 80 to 150 nm when doing STEM observation of the

Si that is used mainly by semiconductor integrated circuits, as shown in

Figure 14.35.

Consequently, the specimen thickness deviation obtained by this experi-

ment is permitted to get reasonable STEM images. Technology that auto-

matically produces nanometer level specimens using the FIB tool is now

being established.

Mark is a circle. Center of a

circle is not changed by FIB

observation.

Figure 14.33 Drift correction.

Focused ion beam systems as a multifunctional tool for nanotechnology 379

Low acceleration voltage process

TEM observation is a useful tool for observing the internal structure of

specimens and in particular the lattice structure. Nonetheless, the crystal

structure may be damaged by FIB processing. Figure 14.36 shows that the

amount of damage becomes smaller in proportion to the acceleration voltage

of the ion beam used in sputter etching.

A: Thickness

160 nm

B: Distance

1.7 µm

Measured thickness of samples

(a)

(b)

100

120

140

160

180

200

1 6 11 16 21 26

Number of samples

Thickness (nm)

Figure 14.34 TEM specimen auto production. (a) Process condition, (b) result

of evaluation.

STEM observation

Thickness=80 nm Thickness=150 nm

ACC

= 200 kV

Figure 14.35 Specimen thickness and STEM image observation by

JEM2500SE (JEOL).

Focused ion beam systems380

For this reason, the ﬁnishing process in producing a specimen must be

performed at a low acceleration voltage when observing lattice images at a

high magniﬁcation.

The latest research has reported that the acceleration voltages for this

purpose can be as low as 1 kV [15].

14.4.2 Probe preparation for the scanning probe microscope

The scanning probe microscope is a tool that scans the surface of specimens

with the tip of a needle, detects movement between the needle tip and

specimen surface, and observes and measures the shape and physical properties

of the specimen surface. Because of its ability to measure various properties, the

scanning probe microscope is a powerful tool for nanotechnology.

The performance of the scanning probe microscope cannot be overstated

although it is controlled by the properties of the needle itself. Figure 14.37

ACC

= 5 kV

Damage layer = 6.6 nm

Si surface

ACC

= 15 kV

ACC

= 30 kV

Damage layer = 14.4 nm

Damage layer = 23.3 nm

Figure 14.36 Three examples of the damage caused to similar specimens at

different acceleration voltages.

Figure 14.37 Scanning probe microscope needle produced using 3D

deposition technology. Length ¼1300 nm, diameter ¼60 nm.

Focused ion beam systems as a multifunctional tool for nanotechnology 381

shows an example of a scanning probe microscope needle being made using

FIB 3D deposition technology.

The tungsten pillar is formed similarly to that of carbon introduced earlier,

and is used as a needle. Depending on the need, the needle can be further

sharpened by sputter etching the tip with the FIB. Here we see the start of 3D

deposition as applied to an actual product.

14.4.3 EB mask repair

To get smaller structures, a short wavelength light source is needed. A

method of making electronic beams a light source electron beam is one of the

candidates for new generation lithography. EPL and LEEPLE are known as

electron beam lithography technologies. Masks used by electron beam

lithography consist of a shading part that does not allow electrons on a

silicon substrate to reach the specimen surface and a transparent part that

allows electrons to pass through and reach the specimen surface. Figure 14.38

shows defects in the photo and EPL masks.

The masking portion of the conventional photo mask is thin. Its thickness

is about 0.1 mm. When white defects and black defects are corrected,

operators do not need to think three-dimensionally to ﬁx them.

On the other hand, the EPL mask repair operator must see a 2 mm thick

silicon substrate in 3D, grasp the defects, and correct them. These corrections

ﬁrst became possible using FIB 3D shape production technology [16,17].

Several technologies used in correction of masks for electron beam expo-

sure are outlined below.

Taper correction

Ion beam density distribution is a normal distribution. Generally, the ‘‘width

of half peak’’ deﬁnes the beam diameter and advances the development into

Clear defect

(a) (b)

Opaque defect Clear defec

~0.1 µm

2 µm

Glass Si

2D structure

3D structure

Rigid glass

Low <1

Thin metal film (~0.1 mm)

Thin membrane

High >10

Si membrane

Substrate

aspect ratio

pattern material

(~2 mm)

Figure 14.38 Defects of (a) photo masks and (b) EPL masks.

Focused ion beam systems382

narrowing this width. When the FIB tool was ﬁrst commercialized, the beam

diameter was about 100 nm. Today tools with a diameter of 4 nm are being

produced. The performance of the FIB has been dramatically improved,

but the beam proﬁle relies on physical phenomena and still has a normal

distribution.

For this reason, when the sputter etching process is performed, a taper that

corresponds to this beam proﬁle in the sidewall of the processed area can be

done. This is shown in Figure 14.39. In the case of the EPL mask, you must

see through the entrance of the hole of the translucent part to the exit and

there must be no obstacles. What is obvious is that the exit slit cannot be

narrower than the entrance slit.

In order to conquer this problem, the specimen stage is tilted. As shown by

Figure 14.39, a taper of 1.5–3



can be used, and the sidewall can be made

vertical to the surface by tilting the specimen to cope with the taper angle.

This technology is used even for TEM specimen production. In order to

deﬁne the thickness of a lamella specimen, the ﬁnishing process tilts and the

process is performed.

Preparation of high aspect holes

As the mask is for forming microscopic shapes on the surface of specimen, a

tool that makes as small a hole as possible is required. Figure 14.40 intro-

duces an example of making a microscopic hole using gas accelerated etching.

GAE processing using accelerated etching and normal processing not using

gas is performed on a silicon substrate that becomes a mask.

Normal processing can only drill extremely shallow holes because there is

no means of transporting the drill cuttings outside the hole. GAE processing

is able to open holes of high aspect ratio because the cuttings react with the

gas and are carried outside the hole.

Cross-sectional image

of FIB-etched trench

FIB

: 48 pA

C depo

~3˚

EPL mask

~50 –100 nm

Figure 14.39 Taper correction. The typical taper angle of an FIB etched

pattern is roughly 1.5–3



; difference of critical dimension between top and

bottom 50–100 nm. I

is the ion beam current at a sample.

Focused ion beam systems as a multifunctional tool for nanotechnology 383

Shape reappearance

Correction of defects is based on normal shape data. The best way to obtain

normal shape data is from CAD data used in production of the original

mask. Nonetheless, in the present state where integration has drastically

improved, CAD data typically consist of information of the entire element

and the amount of information is large and cannot be easily handled.

Therefore, you can select a normal shape near the defect, create correction

information from the image of the normal shape using pattern recognition

technology, and perform the process based on the correction information.

Figure 14.41 introduces G-Copy, which performs defect correction by this

idea. (Note: G-Copy is a function name of the SIR-series.)

GAE

(a) (b) (c)

Normal

50 nm

Within 90±1˚

0.3 µm 0.3 µm

0.5 µm

Figure 14.40 High aspect holes. (a) Surface image, (b) transmission image,

¼3·10

3

Pa, I

¼0.15 pA, time ¼250 s. The taper angle is within

90 ± 1



. P

is the pressure of the halogen gas for GAE.

Defective region

Repaired pattern

1 Before repair

3 After repair

2 Geometry-copy process

Correct pattern

Binarization of image

Processing Setup of processed region

Corner round: corres

ondin

to correct

attern

Copy of correct pattern

Figure 14.41 G-Copy.

Focused ion beam systems384

Nanotechnology is expected to be applied for biotechnology. Cells and

genetics are subjects in the ﬁeld of biotechnology, but when making micro-

scopic structures related to these, the shape information of these must be

taken in and cases of production based on the shape information anticipated.

From the technology introduced here we can expect a growth in production

of structures that correspond to unknown samples.

14.4.4 Micro-mold preparation

Professor Choi proposed nanoimprint technology as a way to mass produce

microscopic shapes by pushing a mold in thermoplastic organic matter and

transcribing shapes [18]. Figure 14.42 outlines the process of producing

nanoimprints.

In this example, the micro-mold was realized by processing a silicon oxide

ﬁlm on a silicon substrate. Thermoplastic organic matter is placed on the

silicon substrate then heated at 200



C. After the thermoplastic organic

matter is sufﬁciently heated, the mold is pushed against the workpiece then

cooled. The mold is removed after sufﬁcient cooling, enabling the shape of

the mold to be transcribed onto the surface of the workpiece.

This technology is expected to be a technology that complements the

current trend of photo lithography. Generally, only 2D shapes can be tran-

scribed with photo lithography but nanoimprinting lets you tilt sidewalls, for

example, by making a 3D mold. We believe that making and correcting this

mold can be done by using FIB sputter etching and deposition [19]. Making

shapes is possible by using technology introduced in Section 14.4.3.

Incidentally, the evaluation of temperature characteristics of microscopic

structures made by FIB are not being sufﬁciently implemented [20]. This is a

subject for the future.

Figure 14.43 introduces preparatory experimental results for temperature

characteristics of microstructures made by FIB. A carbon wall by carbon

Step 1

Heating (200 ˚C)

Step 2

Push and cooling

Step 3

Release

Substrate (Si)

Mold (SiO

)

Thermoplastic

resin

Si substrate

Figure 14.42 Nanoimprint technology.

Focused ion beam systems as a multifunctional tool for nanotechnology 385

deposition was made on a heater wire. Changes in the carbon wall are

observed when the heater temperature increases.

Because this experiment was preparatory, measurement of temperature was

not correctly performed. The heater temperature exceeds 700



What is clear from the experiment is that gallium is extracted from the

surface of the specimen if high-temperature conditions continue, becomes

shaped like a water drop, and ﬁnally evaporates. Extremely small particles

can be seen when the surface of the specimen is observed after high-

temperature treatment; however, they can be removed by washing with water.

This experiment shows that heat treatment must be performed in advance

and the gallium removed when using a workpiece made by FIB, as a mold,

for example, at high temperatures. Since the shape does not change before

and after the heat treatment, the possibility that it can be used as a tool for

Carbon wall

Heater

Current source

In the vacuum chamber

Before heating 2.5A 3min

(a)

(b)

2.5A 8min

+2.6A 5min

+2.7A 5min

2.5A 8min

+2.6A 5min

After heating

Figure 14.43 Basic experiment with micr o molds. (a) Experiment, (b) results.

Focused ion beam systems386

making and correcting molds used in nanoimprinting of FIB tools can be

veriﬁed by gathering more detailed data.

14.4.5 Unfolding of biotechnology

In the latest research, while verifying the behavior of cells one by one,

advanced experiments are desired. For this reason, it is desirable that speciﬁc

conditions be given to cells one by one, and experimental technology that

separately veriﬁes how the effects are received is established.

A cell is generally several micrometers across, which is about the size of the

nano wine glass introduced in Figure 14.27. In terms of size, production of a

container that controls cells one by one is made possible using FIB. If

information about the shape of the specimen is precisely obtained in advance,

a container of optimum shape can be made from those data.

In addition, a nanoscale manipulator was fabricated by FIB deposition

[21]. Cells will be picked up by nanomanipulator, and brought to containers

made by FIB deposition.

Today, evaluation has begun on what effect a container made by FIB will

have on the organic specimen, such as a cell. These are only the ﬁrst steps and

we look forward to the results of later research.

14.4.6 Ultra-high speed device development

The performance improvement of integrated circuits in which Si is used is

remarkable. However, a device made of GaAs is able to get better perfor-

mance than a device made of Si.

Nevertheless, an ultra-high speed element produced on a semiconductor

board made from a chemical compound other than silicon, for example,

achieves a high performance not realized by silicon. Accordingly, as shown in

Figure 14.44, a hybrid integrated circuit is one of the solutions to achieve a

high-speed device.

Device produced using conventional processes

h s

eed device

Wire produced by FIB 3D-deposition

Figure 14.44 Ultra-high speed hybrid device.

Focused ion beam systems as a multifunctional tool for nanotechnology 387