Jackson M.J. Micro and Nanomanufacturing

Подождите немного. Документ загружается.

392 Micro- and Nanomanufacturing

the refractive index of the material. Hence, the material acts as an

optical grating therefore controlling the beam).

The Q-switch itself must be cooled because when it is blocking

the beam it absorbs energy. Q switches can be configured to provide

pulse rates between 0 and 50kHz. The idea is that while the shutter

is closed, energy builds up and a high peak power is released when

the shutter opens, e.g., a 20 W Nd:YAG Q-switched laser can pro-

duce 6 ns pulses of

mJ/pulse, which is 100 kW.

Beam output can be changed by a process known as frequency

doubling. Non-linear optical devices can be swamped with photons,

they can absorb two or more photons, and therefore rise to a higher

energy state. This energy is released in one step; the resulting radia-

tion has half the wavelength and twice the photon energy, e.g., A

1.06 |Lim Nd:YAG beam can be converted to 0.530 |iim green light,

this can be repeated again which would produce ultraviolet light.

A problem with the Nd:YAG laser is the poor efficiency (10-

15%) of converting flash lamp energy to high Nd^"^ energy states,

which produces waste heat that can distort the YAG rod leading to

poor beam quality (M^ approximately 15-100). The flash lamp can

be replaced by using a diode laser that has higher better efficiency

(30-40%).

This produces much better values of

M^,

as low as 1.1.

8.2.3 Diode Lasers

Diode lasers are very similar to Nd:YAG lasers where electrical

energy is translated via the diode into electron excitation and even-

tually light is emitted. Usually they are grouped and stacked together

to form high power lasers. Semiconductor materials have a bandgap

such that if the electrons have enough energy (provided by an elec-

trical field) they convert from a non-conductive state to a conductive

state.

Part of this change of state can release photons, similar to the

laser actions already discussed. GaAs is a common type. It has a

bandgap energy of 1.35 eV corresponding to a wavelength of 905

nm. Diode lasers have the advantage of being small and affordable.

Diodes tend to emit over a frequency range but they can be tuned

by a grating. Low-power diode lasers have low power conversion ef-

ficiencies of around 2% whereas high power diode lasers have high

power efficiencies up to 30%. A 5 |Lim wide strip can produce 100

Laser Micro- and Nanofabrication 393

mW of power, a 50 |j.m wide strip can produce 0.5 W, and a 500 |Lim

wide strip can produce 4 W. Stacking such arrays can produce even

more power although they usually have high divergence values that

make them unsuitable for laser applications.

8.2.4 Excimer Lasers

Excited dimer molecules (hence excimer) decay and release their

laser radiation. The excited dimer

Kr'^F",

lasts around 5-15 ns and the

photons produced are in short pulses of

ns that are spread in a 0.4

nm band, which is large but the power is around 35 MW (0.2

J/Pulse).

8.2.5 ThSapphire Lasers

Ti:sapphire lasers produce pulses in the femtosecond time regime.

Short, low-power nanojoule femtosecond pulses are created by an

erbium doped fibre laser. Amplification of these low-power pulses is

required to produce a useful power output, but there are problems

associated with this. These problems are overcome by using chirped

pulse amplification (CPA). The pulse is stretched, amplified, and

then compressed to create a high-intensity femtosecond pulse.

The problem with directly amplifying short pulses is their ten-

dency to stretch in time, the pulse may be amplified but it will no

longer be on the femtosecond timescale. Femtosecond pulses also

tend to destroy the medium through which they travel. CPA avoids

these problems by the basic principle of stretching short pulses to

reduce the peak intensity. The pulse is then compressed back to the

original time scale. The femtosecond pulse is made up of several

wavelengths. The diffraction gratings or chirping mirrors separate

out the different wavelengths of the pulse. This is done because the

grating causes the individual frequencies to reflect at different an-

gles.

Thus light from the same pulse travels different distances,

which tends to stretch the pulse and reduce the peak intensity. The

stretched pulses are let into the regenerative cavity for amplification.

The pulse is sent along a certain path. At a point in this path it passes

through a

Ti:

sapphire crystal which is the gain medium, and this is

394 Micro- and Nanomanufacturing

pumped by an Nd:YAG laser. Each time the stretched pulse passes

through the cavity it gains a little more energy, receiving a boost

from the interactions in the cavity. Along a part of this path there is

a component that reflects the pulse if it is not of high enough energy.

It is then sent back into the regenerative amplifier to gain more en-

ergy until it reaches some critical value. When this value is reached,

the pulse encounters the device and it now has enough energy to es-

cape from. Thus, the pulses have been amplified in terms of power

but are stretched in terms of time. An exact reversal of the process

that stretched the pulse is now applied to recapture the original pulse

length. The pulses are now at full power and are on the femtosecond

timescale. The pockel cell and polarizer perform the tasks of timing

entry of pulses into the regenerative amplifier and determine how

the pulses are let out of the cavity and how the gain is received.

8.2.6 Beam Characteristics

Ideally the beam is Gaussian in shape. The cross section shows

the intensity of the beam. The beam is defined when 86.5% of the

energy is across its diameter. Not all beams are ideal, e.g. the hot

spot may be shifted slightly to the right (Fig. 8.3). This means that

the laser cavity needs adjusting to correct the beam profile. The pro-

file of

the

beam can be obtained by making an image in a material or

by using three-dimensional beam profiling cameras that are avail-

able that describe the beam profile and intensity variations within

that profile (Fig. 8.4).

The beam many be affected by the optics and apertures it travels

through; for example if it passes through a small hole it could be-

come diffracted (Fig. 8.5). Eventually the propagating beam will

naturally spread out but stays parallel better than any other light

source. The laser light spreads out owing to diffraction; the waist

being the minimum diameter of the beam denoted

WQ.

It is at this

point that it becomes sensible to focus the beam on the surface.

However, all is not lost if this is not achieved. Either side of

the Rayleigh length, this is a length where the beam is still highly

focused and though not ideal, laser ablation can still be achieved.

Laser Micro- and Nanofabrication 395

Fig. 8.3. Eccentric beam profile

Fig. 8.4. Three-dimensional laser beam profile

396 Micro- and Nanomanufacturing

Fig. 8.5. Effects of diffraction on the laser beam

8.2.7 Laser Optics

Optics are used to change the diameter of the beam (increases the

power, suppHes the beam to a device with a specific inlet diameter,

and changes the focal distance to the substrate), the direction of the

beam, or the intensity of the beam. The conventional laws of optics

apply but special care must be taken in terms of coatings and the ma-

terial the optic is made from. For an ideal beam,

ew=xi7r

(8.1)

The realistic laser beam equation is:

ew ^M^XIn

(8.2)

Laser Micro- and Nanofabrication 397

Where, 9 = half the divergence angle, o = beam waist size, and X =

wavelength. As everything is ideal here this expression is true over

the whole length of the beam. However, in reality perfect conditions

never exist, is a beam quality number, which is one for a perfect

beam. The minimum spot size can be found in the following way;

d^=D/2f

(8.3)

Resulting in a minimum spot size,

-M'A^ (8.4)

Where, / is the focal distance, D is the lens diameter, and 8 is the

minimum spot size diameter. Minimum spot size governs the mini-

mum feature size. Remember that in conventional machining small

features can be created with relatively large cutting tools. Clearly

is an important parameter to control as it one of the major influ-

ences of the minimum spot size. Hence, a high-quality beam is de-

sirable for micromachining. For micromachining, a small spot size

is required. This is obtained when M^ =1, with a short wavelength,

and a short focal lens length. Examples of beam properties are

shown in Table 8.1.

8.2.8 Optical Quality

The laser cavity geometry affects the quality of the beam. Cavity

length and mirror configuration cause different types of laser spot to

be produced. Oscillating light between the mirrors is not forced to

travel along the optical axis. It is these oscillations that create trans-

verse electromagnetic modes (TEM). The number of off-axis modes

is described by the Fresnel number, N = a^/X,L, where

is the Fres-

nel number, a is the radius of output aperture, and X is the wave-

length, and L is the cavity length. Off-axis oscillations would be

visible on a screen placed at mirror B when differences in wave-

398 Micro- and Nanomanufacturing

lengths between on-and-off axis oscillations differs by a whole

number of wavelengths n when using Pythagoras' theorem.

Table 8.1. Properties of laser beams

Laser Wavelength, Power, P w.0 Beam

A,(^in) (W) (mm.mrad) quality.

Spot di-

ameter

with f/4

lens,

HeNe

Fine drill-

ing

Nd:YAG

Q-switched

Nd:YAG

Q-switched

Nd.YAG

Q-switched

Nd:YAG

CO2

Copper va-

por

Ti:

Sapphire

Excimer

0.63

1.06

10.6

0.51

0.78

0.193-0.351

0.002

100

1000

100

1000

100

0.2

0.5

0.98

1.5

200

500

The laser cavity geometry affects the quality of

the

beam. Cavity

length and mirror configuration cause different types of laser spot to

be produced. These are known as transverse electromagnetic modes

or TEM. Hence,

+L^

+ nXf

(8.5)

2'^2,

Simplifying and ignoring the n X term because it is very small

= 2LnX

(8.6)

Laser Micro- and Nanofabrication 399

a^/(2LX)

(Fresnel number)/2

(8.7)

Low Fresnel numbers yield low order modes; the off-axis oscil-

lations are lost by diffraction and cannot aid in the amplifying proc-

ess.

Table 8.2 shows examples of laser types, Fresnel number, and

modes. Transverse electromagnetic modes of the laser beam are

shown in Table 8.3.

Table 8.2. Types of industrial laser [1]

Model

1 Laser Ecosse

MF400

Electrox

PRC1500

Control 2kW

PRC3000

Laser Ecosse

AF5

1 Dimensions of Typical Industrial Lasers

Cavity

radius mm

Cavity

length m

Fresnel

number

Mode

3.5

14.4 0.8

TEM01*/reM00

3,4 2.2

TEMOl*

3.4 2.2

TEMOl*

17.5

6 4.8 Low

5.17 2.2

TBMOlVLow

TEMOO

^°

^^ °-^

contronable

8.2.9 Laser Material Interactions

The principle of laser material removal is that laser light is fo-

cused on the material surface where energy is absorbed. This energy

is converted to heat. The maximum depth where absorption occurs is

called the penetration depth and leads to the conduction of heat into

the material. The energy absorbed for a CO2 laser is about 20%, for

a Nd: YAG, or for a femtosecond laser is approximately 40 - 80; and

400 Micro- and Nanomanufacturing

the rest is reflected. Special care must be taken with optics, special

anti reflective coatings must be applied to prevent beams in un-

wanted areas of the system. Laser energy is converted to heat only

as far as the penetration depth. At the penetration depth the power

density is approximately 1/e of the original density at the surface.

For a CO2 laser this is about 15 nm, for an Nd:YAG laser this is

about 5 nm. Conduction of heat into the bulk depends on the time-

scale of the pulses.

Table 8.3. Transverse electromagnetic modes [1]

Angular zero fields -1

•

TEMOO

TKMlO

.TEM20 J

TEMOl

TEMU

""TlMail

2 1

^ #

TEMO2

TEM12

'i'

.'TsMy

TEMOl*

The mechanism that lasers use to remove material is called abla-

tion. Ablation is achieved by melting and vaporizing the material,

which is then ejected from the vicinity of the surface and is gov-

erned by power density. However, the ejected material can be depos-

ited near to the melt region where it freezes and is known as the "re-

cast" layer (Fig. 8.6).

Laser Micro- and Nanofabrication 401

Fig. 8.6. Re-cast layer created close to the melt region

9 2

For a power density of 10" W/cm the melting point is reached

in 300 ns, if the power density is increased 10-fold this time is re-

duced 10-fold to 3 ns. The high vaporization rate then causes a

Shockwave that can reach speeds of 3 km/s. Expulsion of material

occurs because of high pressure created in the melt and explosive-

like boiling of the superheated liquid after the end of the laser pulse.

This ejected material can re-form; sometimes it is not always totally

vaporized, especially if it is from a deep trench that may cool

quickly and reform. Nanosecond lasers produce heat flow whereas

femtosecond lasers do not because no melt pool has been produced.

Femtosecond laser pulses machine with minimal heat generation

where the heat affected zone (HAZ) is given by the equation, HAZ ~

(D.t) , where D is the thermal diffusion coefficient and t is the

pulse duration. Heat flow produces the melt pool, around which the

bulk of the material heats up. If rapid heating occurs then the dam-

age to the material is incurred at a greater depth than is desired, this