JP2003114400A - Laser optical system and laser machining method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an apparatus which converts a laser beam having a Gaussian distribution and other nonuniform power distributions into a uniform power distribution beam of a top hat type. SOLUTION: A diffraction point where the power is equally distributed in a specified range is formed and is defocused, by using a diffraction type optical element and a condenser lens, to obtain the optical quantity distribution of the top hat type within this range. The phase pattern of the diffraction type optical element (DOE) is previously determined in such a manner that the beam power at the lattice point within a certain range is made the same. The laser beam having the nonuniform distribution is cast to the diffraction optical element and the transmitted light thereof is condensed by the condenser lens. The beam emitted from the condenser lens is made into the uniform optical quantity distribution on an objective surface by defocusing to slightly shift the objective surface forward and backward from the focal plane.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a technical field of using a laser beam as a heat source to locally heat an object to be processed by the heat of the beam to perform laser processing such as drilling, welding and heat treatment. Since a carbon dioxide gas laser, a YAG laser and the like have a strong power, it is possible to apply a beam to the object to be processed and perform hole drilling, welding and the like. When the beam power is reduced, it can be used for heat treatment. When it comes to drilling, it can be said that productivity is high because a large number of holes can be drilled in a short time rather than mechanical drilling. The problem here is the spatial power distribution of the laser beam, and it is intended to obtain a top-hat type laser beam having a constant power in a certain range.

[0002]

2. Description of the Related Art As a laser for melting and processing a target material by heat, a carbon dioxide gas laser having a large power or a YAG laser is used.
There is a laser. A mirror or lens is required to guide the laser beam to the target. Carbon dioxide laser has a wavelength of 9μ
Since it is m to 11 μm and is infrared light, it is condensed using a ZnSe lens or the like that can transmit it. Since the YAG laser is 1.06 μm, a lens made of a normal glass material may be used.

Regarding laser light emission, there are a case of continuous light emission (CW) and a case of pulsed light emission.
Depending on the purpose, continuous light and pulsed light can be used separately. In many cases, laser processing is performed by generating pulsed light with large energy by the Q switch.

[0004]

The laser beam usually has a beam power distribution represented by a Gaussian distribution exp (-2r ² / w ² ) (1). That is, the intensity is often strong at the center of the beam and weak at the outer periphery of the beam, and often has a spatially non-uniform power distribution. Here, w is the beam waist. The diameter or radius of the beam is defined by the distance to the point where the intensity drops to the central e ⁻² . That is, the waist w is the beam radius. In many cases, such a Gaussian distribution does not matter.

However, in the field of laser processing, such non-uniform power distribution may not be desirable.
For example, when making a hole by applying a laser beam to a printed circuit board, there is a problem that the center of the bottom of the hole is scorched.

In the case of laser drilling, it is desirable to use a beam having the same power density in the central portion and the peripheral portion in order to make uniform holes. Then, the object is burned off with the same power in the center and the periphery, so that uniform holes can be drilled.

A uniform laser beam intensity (power) distribution in such a fixed range is referred to as a flat top (Fl
Let's call it the "Top" type or the Top Hat type. This means that the laser beam has almost the same power density in a desired range.
When expressing the power density at r from the center by P (r), The top-hat distribution is.

This shows a beam of circular cross section. Although the above notation is easy to understand, it is inconvenient over two lines, so the sgn function is defined. It is sgn (x) = 1 (x
> 0), sgn (x) = 0 (x ≦ 0). Alternatively, for two variables x and y, sgn (x, y) =
1 (x> 0 and y> 0), sgn (x, y) = 0 (x ≦
0 or y ≦ 0)). The outer shape of the laser beam is often circular. The Top Hat distribution is given by equation (2).

P (r) = Wsgn (w−r) (3) However, since the shape of the hole required is various, the shape is not limited to a circle but has a pattern such as a square beam, a linear beam, a ring beam, or the like. It may be a beam.

[0010] 2a laser power distribution for top hat square to one side of the can be represented by ^{P (x, y) = Wsgn} (a 2 -x 2, a 2 -y 2) (4).

[0011] 2a, 2b the laser power distribution for top hat square and two sides is ^{P (x, y) = Wsgn} (a 2 -x 2, b 2 -y 2) (5). Since the linear beam means that a / b is considerably larger than 1, it can be expressed by equation (5).

[0012] outer diameter 2a, and say top hat ring-shaped beam having an inner diameter 2b, can be represented by ^{P (r) = Wsgn (a} 2 -r 2, r 2 -b 2) (6).

FIG. 8 shows the Gaussian beam, and FIG. 9 shows the ideal shape of the top hat beam. Several proposals have been made in the past to create such a top hat power density distribution.

John A. Hoffnagle and C. Michael Je
fferson, "Design and performance of a refractive o
ptical system that converts a Gaussian to a flatto
p beam ", APPLIED OPTICS, Vol.39, No.30, 20 Octobe
r, 2000, p5488-5499

Uses two aspherical lenses to convert the laser Gaussian beam into a parallel top hat distribution. The first aspherical lens is a concave lens that expands the dense beam in the central part to the outside and makes the sparse beams in the peripheral part closer to each other to increase the density. The dark areas are made thin and the thin areas are made thick to make them uniform. The following aspherical lenses convert those tilted beams into parallel beams. Therefore, the light passing through the second aspherical lens is parallel, and the top hat has a uniform power density. This is to convert a Gaussian beam into a beam having a uniform distribution with only two lenses, and the idea is easy to understand.

Moreover, since the converted beam is a parallel beam, the distance to the object may be changed. It becomes a top hat beam that is easy to handle. Further, since the lens optical system is used, there is an advantage that the wavelength (λ) dependency is small. The lens refractive index has a wavelength dependence, but it is small. Since refraction is utilized, such a thing is possible even if the light source has no wavelength dependence and has a spectrum close to white.

However, the two aspherical lenses are difficult to process. It is possible to calculate the surface shape of an aspherical lens, but it is difficult to actually manufacture a lens having a curved surface as calculated by machining. The result is an expensive mechanism.

Even if it is Gaussian, the beam distribution actually differs from laser to laser. Even if the early talk is a carbon dioxide laser of the same scale at Beam West w, it is scattered. In designing an aspherical lens, w needs to be determined, but since w varies from laser to laser, calculation must be performed for each laser, and a different set of aspherical lenses must be created for each laser. Such out-of-control diversity adds to the cost. Since it is troublesome, when w is fixed and an aspherical lens is manufactured, variations occur depending on the laser and the beam after molding does not necessarily become the top hat.

Since the aspherical surface is formed by cutting, only a cylindrically symmetric lens can be formed. A circular beam like the above formula (2) may have a top hat distribution by combining two lenses, but it is difficult to make a square top hat or a rectangular top hat distribution. Furthermore, the donut type top hat has a cylindrical symmetry, but it cannot be easily done.

Fred M. Dickey, Burton D. O'Neil, "
Multifaceted laser beam integrators: general formu
relation and design concepts ”, OPTICAL ENGINEERING,
Vol.27, No.11, November, 1988, p999 is a laser beam is divided into M ² pieces of independent beams against a mirror having a non-continuous surface also ^two M having a cross section like saw teeth, It is intended to obtain a uniform distribution on the image plane by applying this to one concave mirror and reflecting it. Assuming that the laser beam has a Gaussian distribution and a diameter of D, the entire mirror surface using a circular mirror or a square mirror with one side of about D is divided into M × M.

Although the directions of the respective mirrors are slightly different, since the light is reflected by the concave mirror, the boundaries are stepped with substantially the same inclination. Originally, the intensity largely changes from 1 to 1 / e ² in the spread of D with a Gaussian beam, but the intensity distribution variation in a small area of D / M × D / M obtained by dividing it into M pieces is extremely small. . When this is reflected to the concave mirror, M ² beams are superposed, so that the total beam intensity should be equal in the whole concave mirror. When the light reflected by the concave mirror is applied to the image plane, it means that light with a uniform distribution can be obtained. In other words, since the laser beams are once collected and distributed, the optical power should be uniform over the entire surface.

However, since the M ² split mirror is used, the phase relationship is broken, and when the concave mirror is a clean concave surface, the phase relationship on the image plane is disturbed. Therefore, not all the divided beams satisfy the image forming condition on the image plane. Therefore, even if the beam trajectory can be obtained by geometrical optics, the beam intensity distribution does not always become the desired one. Therefore, the top hat has a jagged distribution.

Not only that. Since the phases of the light from all partial mirrors do not match at the image plane, a slight movement of the image plane back and forth changes the power distribution again. In other words, it is like dealing with noise. The initial laser beam is analytic, but if it is divided, the analyticity is no longer present, so the two-dimensional light amount distribution is non-uniform and unstable.

Although this is a theoretical difficulty, there is a practical problem that the cutting of the M ² split mirror having various steps and inclined surfaces is delicate and difficult. In addition, since it is a reflective system, it has a drawback that the system becomes large.

Other techniques for homogenization using an aspherical lens are based on the assumption that the laser beam is a Gaussian beam and that the spread σ is also known. However, the laser beam does not always have a clean cylindrically symmetric Gaussian distribution, and the beam spread σ is not constant. Even if they are carbon dioxide lasers of the same model and made by the same manufacturer, the beam spread and distribution will be slightly different. Therefore, it is necessary to make different lenses depending on the distribution and spread of the laser beam. Therefore, the one using an aspherical lens has a strong dependence on the laser profile and is not general.

It is desirable that the structure be applicable regardless of whether the laser beam is Gaussian or Gaussian even if the spread σ varies. It is also important that it is easy to design and easy to manufacture.

A laser beam having various cross-sectional shapes can be converted into a top-hat type laser beam having a constant power density in a desired region (circle, square, line, ring), which is easy to manufacture and designed. It is a first object of the present invention to provide an easy optical device. It is a second object of the present invention to provide a laser processing apparatus that can convert a laser beam into a top hat type laser beam and can process an object with the top hat type beam.

[0028]

According to the method of the present invention, a laser beam (not limited to a Gaussian beam) is passed through a diffractive optical element to be diffracted and branched into a plurality of partial beams having different emission angles and having substantially the same power. Then, the partial beam is condensed by a condensing lens, and a portion slightly shifted from the image plane is set as a target surface to form a beam having a top hat distribution. It is also possible to use a diffractive optical element with a lens in which the diffractive optical element and the lens are integrated.

A combination of a diffractive optical element (DOE) having a function of branching one incident beam into a plurality of beams having different exit angles by diffraction and a condensing lens are assumed to be on the object surface. The divergent beam can be focused on a lattice point of a lattice having a certain period. For example, Japanese Patent Laid-Open No. 2001-62578 (laser drilling device;
2001.13))).
The purpose of this is to drill many thin holes at the same time, and the beam is required to be thin and highly parallel.

The present invention instead seeks rather the opposite purpose. The challenge is to obtain a beam with constant power in a continuous area. The present invention provides a diffractive optical element (D) so that the beam power is the same at lattice points within a certain range.
The phase pattern of OE) is determined, a laser beam having a non-uniform distribution such as Gaussian distribution is applied to the diffractive optical element, and the transmitted light (reflected light in the case of a reflection type) is condensed by a condenser lens. If the image plane is placed on the focal plane, an equal amount of optical power is distributed to the lattice points and a power distribution localized only at the lattice points can be obtained. To shift. That is called defocus. By defocusing, a uniform light amount distribution is obtained on the target surface.

When the position of the object (object surface) is slightly shifted from the focal plane in the axial direction, the beam is not focused on the object surface and spreads. If the amount of deviation increases, adjacent beams will overlap each other. If the gap is large, the overlap will increase further. By making the overlap of the partial beams appropriate, a uniform intensity distribution can be obtained as a whole. As described above, according to the present invention, a branched beam is generated using a diffractive optical element having a pattern capable of distributing an equal amount of energy to lattice points to be included in a top hat, and defocused and superposed to obtain a uniform intensity distribution. Is.

Furthermore, by giving intensity distributions to the respective spots, the intensity distribution when defocused and superposed is not completely uniform but has a periodic intensity distribution. You can also

[0033]

BEST MODE FOR CARRYING OUT THE INVENTION Since the present invention obtains a uniform top hat power distribution by defocusing a diffracted image using a diffractive optical component and a condenser lens, it is necessary to explain the operation of the diffractive optical component. . Before that, diffraction will be briefly described. The most proven component that uses diffraction is the diffraction grating. This is a two-dimensional structure in which convex lines and concave lines are provided in parallel on the surface of the component at equal intervals. There are a reflection type diffraction grating and a transmission type diffraction grating.

The reflection type diffraction grating is used in a spectroscope.
The period of the unit of the arrangement of the convex stripes and the concave stripes is d. Let Θ be the angle between the reflected light and the diffraction grating normal. It is assumed that incident light enters parallel to the normal (normal incidence). In order for the reflections from the adjacent protrusions to be mutually strengthened, the phases of the reflected lights must be aligned, so that dsin Θ = λm. This is called the Bragg diffraction condition. This is because the difference in the optical paths of the diffracted light from the adjacent ridges is dsin Θ. m is the order of diffraction and takes positive and negative integers. Increasing the diffraction order increases the tilt of the beam with respect to the normal.

Although there are few transmission type diffraction gratings, absorption, dispersion, refraction, etc. must be taken into consideration, and the diffraction grating becomes more complicated. However, since the present invention uses a transmissive DOE, a transmissive diffraction grating will be described as its preparation. It is different in that it allows light to pass through, but it also has ridges and grooves formed at equal intervals. The period of the unit in which the convex stripes and the concave stripes are lined up is d, and the angle formed by the transmitted diffracted light and the normal is θ. In order for the transmitted lights from adjacent protrusions to strengthen each other, dsin Θ
Must be = λm. m is the order of diffraction and is a positive or negative integer. If m = + 1, first-order diffraction; if m = -1,-
The order of diffraction is called by the value of m, such as first-order diffraction. m
When the diffraction angle theta _m order diffracted light it theta _{m =} sin
⁻¹ (λm / d). The diffraction angle increases as the order m increases. However, the diffraction order m and the diffraction angle Θ are not directly proportional.

This is diffraction in which linear ridges and ridges are parallel.
It is a lattice, and the irregularities are one-dimensional. Further surface
In the x direction_x, D in the y direction_yCut into squares
S × T basic rectangles d_x× d_yIf divided into
Diffraction occurs not only in the direction but also in the y direction. d in the x direction
_xsin Θ_x= Λm, d in the y direction_ysin Θ _y
= Λn. Two-dimensional due to unevenness distributed in a plane
Diffraction can occur, but it is
Can be thought of as the chief. m order in x direction, y direction
The diffraction angle of the diffracted light of the nth order is Θ_x= Sin^-1(Λm /
d), Θ_y= Sin ^-1(Λn / d) parallel beam
It This is two-dimensional diffraction. Two-dimensional beam separation
However, the principle is the same as one-dimensional.

In this case, the m-th order and the n-th order diffracted lights (parallel lights) are separated vertically and horizontally. Nothing is decided about the intensity of the diffracted light. The intensity distribution of the diffracted light depends on the fine structure of each protrusion or recess. If the protrusions and recesses are simply uniform, they are strong at the center and weaker toward the periphery (higher-order diffraction). In one-dimensional diffraction and two-dimensional diffraction due to simple unevenness, the power distribution is sin
It becomes something like Θ / Θ. Although this is a continuous function, since the diffraction angles Θ _m and Θ _n take discrete values, the power distribution has several discrete sharp peaks.

Since the light is parallel light, it forms an image at infinity. Since this is inconvenient, an image may be formed at a finite distance.
A condenser lens is used to form an image at a finite distance. Since it is parallel light, all diffracted light can be imaged on the same plane (image plane) orthogonal to the axis. Although it can be condensed by an ordinary convex lens, the ordinary lens is an ftan θ lens that forms an image at a position of ftan θ from the center on the focal plane when the beam tilt angle is θ. When diffracted light is condensed by such a lens, the position of the m-th diffraction point (distance from the center) on the focal plane is ftansin ⁻¹ (mλ /
d). Images at diffraction points do not become equidistant.

Therefore, a special lens was devised to image a beam having an inclination angle of θ from the center to a position of fsin θ on the focal plane. It is called fsin θ lens. With the fsin θ lens, the distance from the center of the focal plane of the m-th order diffracted light is fsinsin ⁻¹ (mλ / d)
= Fmλ / d. That is, the image points of the diffracted light are arranged at equal intervals on the focal plane. However, in the case of the present invention, a regular ftan θ lens is sufficient because strict equal spacing is not required.

With only the diffraction grating, the lens, and the image plane of the focal plane, the diffraction points are arranged vertically and horizontally. The intensity of the diffracted light only follows the sin Θ / Θ rule described above. A diffractive optical element is one in which the intensity of each diffracted light is freely given by giving a fine structure inside one concave portion and one convex portion. By providing a finer structure to each of the convex portions, concave portions, and convex stripes described above, a diffraction image having an arbitrary light amount can be obtained. Discrete diffraction spots (points) occur on the image plane, but the amount of light can be arbitrarily given. The present invention uses a simple diffractive optical element (DOE) that equalizes the amounts of diffracted light in a certain range.

The above-mentioned convex portions and concave portions will be referred to as patterns. The patterns are for causing diffraction and all have the same structure. The fine structure of the pattern is again a vertical and horizontal lattice structure, and the fine lattice is called a cell. A set of cells is a pattern.

The patterns are all the same. Diffraction occurs because they are the same. It is the repetition of the same pattern that causes diffraction. It is the size of the pattern that determines the repeat pitch. There are S × T patterns in the DOE. Let the sizes of one pattern be d _x and d _y . d _x and d _y determine the position of the diffraction point. This is the default. The dimensions of the DOE are therefore Sd _x × Td _y . It is assumed that M × N cells exist in one pattern.
The cell size delta _x, and delta _y. d _x = MΔ _x , d _y =
N Δ _y . That is, the area of the cell, which is the minimum unit, is Δ _x
Δ _y , and the area of one pattern is MN Δ _x Δ _y . Area of the DOE will be referred STMNΔ _x Δ _y. In other words, a pattern is a collection of MN cells, and a DOE is a collection of ST patterns.

However, it is the arrangement of cells that determines the light quantity distribution of the diffracted light. Therefore, the layout of cells in one pattern is a design matter. A cell can take a binary value such as a binary value, a quaternary value, or an octetary value. In the simplest binary case, there are 2 ^MN degrees of freedom. Therefore, there is no one-to-one correspondence between the distribution of cell values of one pattern and the distribution of diffracted light. It can be created by the same diffracted light amount distribution and various different patterns. Next, a method for determining the arrangement of cells on the pattern will be described. Even with the value distribution of cells, the dimensions (Δ _x , Δ _y ) and positions of the cells themselves are fixed. For binary,
The problem is to decide whether each is 0 or π.

FIG. 10 shows a schematic view of an optical mechanism including the diffractive optical element (DOE) 1. The condensing lens 2 is not necessary if only diffraction is performed by the diffractive optical component. However, since the diffracted light is a collection of parallel beams, it does not form an image no matter how far it goes. Although there is an application in which such a thing can be used, here, the parallel beam is formed into an image with a finite focus by the condenser lens 2. As the condenser lens, it is preferable to use the fsin θ lens instead of the normal ftan θ lens as described above. However, if it is not necessary to strictly set the diffraction points at equal intervals, an ftan θ lens can be used.

The parallel incident beam is diffractive optics (DO
E) The light passes through 1 and is diffracted on the back surface to form a finite number of diffracted beams. The beam is condensed by the condenser lens 2. The image plane 3 is assumed to be at the focal length f of the condenser lens 2. Since it deals with the diffraction phenomenon, the geometrical optics method can only discuss qualitatively. Wave optics must be discussed.

The extension direction of the optical axis is the z axis. The DOE 1 and the condenser lens 2 are installed parallel to the xy plane.
The amplitude distribution of the incident beam on the xy plane is a (x, y). The incident beam is assumed to be a plane wave. It is a reasonable assumption to use a plane wave because a thick parallel beam such as a carbon dioxide laser is a problem. If it is a plane wave, the variables x, y and z can be separated. Therefore, the incident beam is multiplied by a function in the z direction and the whole wave function can be written as a (x, y) exp (jkz-jωt) (7).

K is the wave number (2π / λ), and ω is the angular frequency (2πf). The fact that the carbon dioxide laser beam is Gaussian means that a (x, y) is Gaussian. It can be written as a (x, y) = exp {-(x ² + y ² ) / 2σ ² } (8) That's what it means. However, actually, the diffractive optical component can be designed without depending on the power distribution a (x, y) of the laser beam. This can be seen in the discussion below. Actually, the profile of a (x, y) does not matter.

Since the parallel beam reaches the DOE, the power distribution immediately before the DOE is a (x, y). How to represent DOE1 is a problem, but since parallel plane waves have been used up to this point, it is sufficient to consider one in which the power distribution immediately after DOE is multiplied by a (x, y). What happens in DOE is
This means that the phase change may differ by π depending on (x, y). Of course, there is absorption, but not phase absorption, but phase change only. Even if it is called a phase change, the change itself does not matter, and only the difference (π) between changes is a problem.

Therefore, 1 at some point is -1 (= e at some point)
It suffices to consider only the binary distribution such that xp (−jπ)). The parameter of such DOE is called complex transmittance t (x, y). This is a discontinuous function (referred to as a characteristic function) that associates -1 or +1 at every point of the DOE. Then, the wave function of light immediately after leaving the DOE is a (x, y) t (x, y) exp (jkz-jωt) (9).

Let f be the focal length of the condenser lens and k (= 2π / λ) be the wave number of light. The lens has a large thickness at the center and a small thickness at the periphery. Therefore, the phase changes so that it advances in the center of the lens and lags in the periphery. The lens center is x = 0, y = 0
Is. Since the parallel beam is incident on the lens and is focused on the focal point f behind the lens and the phase there is the same, the phase at the point (x, y) of the two-dimensional coordinates taken on the lens surface is Multiplying the distance of by the wave number k should be negative. Further, since f is much larger than x and y, it is approximated as follows: phase shift = −k {f ² + x ² + y ² } ^1/2 = −k {f + (x ² + y ² ) / 2 f} (10) , The phase difference from the center is a problem, so the constant term is omitted, and the imaginary unit j is multiplied into the exp () to determine the complex amplitude at (x, y) behind the lens.

Therefore, the amplitude distribution U of the light behind the lens
(X, y) is U (x, y) = a (x, y) t (x, y) exp {-jk (x 2 + y 2) / 2f} (11).

It is assumed that the image plane is at the focal length f of the lens.
In that case, an arbitrary point (x ', y') on the image plane from the center of the image plane
The complex amplitude W (x ′, y ′) at 1 changes the phase by the distance of the lens back point from (x, y) times k. Since the image surface and the lens surface are displaced by f in the z direction, this is also within the approximate range: phase shift = k {(x'-x) ² + (y'-y) ² + f ² } ^1/2 = k [f + {(x -x ') 2 + (y-y') 2} / 2f] because made (12),

The complex amplitude at the focal plane is the superposition of all the waves from the lens surface and W (x ', y') = A∫∫U (x, y) exp [jk {(xx-x ' ^{) 2 + (y- y ')} 2} / 2f] becomes dxdy (13).

Here, A is a normalization constant, and A = (jλf) ⁻¹ (14).

Substituting (11) into U (x, y) in (13), the term of (x ² + y ² ) drops, so W (x ', y') = Aexp {jk (x ' ² + y' ² ) / 2f} ∫∫a (x, y) t (x, y) exp {-jk (xx '+ yy') / f} dxdy (15). This double integral is a Fourier transform formula.

The Fourier transform of g (x, y) is g
_F (x ', y') if in writing _{and, g F (x ', y} ') = A∫∫g (x, y) exp {-jk (xx '+ yy') / f} dxdy (16 ) W (x ′, y ′) = exp {jk (x ′ ² + y ′ ² ) / 2f} (at) _F (x ′, y ′) (17)

This is at the focused image plane (x ',
y ') is the complex amplitude at point. Since the power at the diffraction point is the square of its absolute value, the first oscillating portion becomes 1 and need not be considered. What we need to find out is the Fourier transform of at. a is the power distribution of the laser beam, and t is the distribution of the surface height of the diffractive optical element.
Strictly speaking, a should be taken into consideration, but a (x, y)
Has less dependence on (x, y), and t (x, y) has (x, y)
y) has a strong dependency. This is because the cell dimensions Δ _x and Δ _y are much smaller than the beam spread σ. Therefore, it can be said that the complex amplitude on the image plane is almost determined by t (x, y).

In reality, however, the opposite is true, and the desired distribution of diffracted light is given first, and what should be done is t (x, y) of the diffractive optical element (DOE) for realizing it. ? That's what it means. DOE consists of ST identical patterns with MN cells. Since the pattern dimensions are fixed, what about the cell phase distribution? Becomes a problem.

FIG. 11 shows the phase distribution of cells in one pattern of DOE. The pattern is a set of M × N cells. The size of one cell is Δ _x × Δ _y . So pattern dimensions transverse to MΔ _x (= _{d x),} vertically NΔ _y (=
d _y ). The center of the pattern is taken as the origin to give the coordinate system.

When the value is binary, the DOE thickness takes one of two values. Since the phase difference of transmitted light is a problem, the thickness is λ /
It means that it is different by 2 (n-1). The m-th cell in the x-direction and the n-th cell in the y-direction are called mn cells, and the phase thereof is φ _mn = 0 or π. The black background shows the case where the phase difference is 0, and the white background shows the case where the phase difference is π.

The thickness in the case where the phase difference is two is drawn on the horizontal side of FIG. A little strange, but if π is more than 0, λ
I intend to symbolically express that it is only 1/2 (n-1) thick. The phase φ _{mn of the} mn cell is either 0 or π. The transmission amplitude of the mn cell is represented by t _mn . It takes either 1 or -1.

T _mn = exp (jφ _mn ) = 1, φ _mn = 0 −1, φ _mn = π (18)

Since the complex amplitude t (x, y) of the entire DOE is a set of these, {t _mn } for MN cells.
The power of the diffraction image can be completely determined by determining

Using the rectangular function rect (u), t
complex amplitude p (x of one of the pattern by a set of _mn,
y) can be expressed. It is defined as follows. rect (u) = 1, | u | ≦ 0.5 0, | u |> 0.5 (19)

[0065]

[Equation 1]

Since one pattern has a size of d _x × d _y , t (x, y) is a set of these,

[Equation 2] Becomes

Substituting (20),

[Equation 3] Becomes

Then the Fourier transform of t (x, y) is

[Equation 4]

The sum of s and q can be calculated as

[Equation 5]

It becomes This gives the character that it is a diffractive element. The quotient in the division of the sin function has a finite value only when the sin function in the denominator converges to 0, but the condition gives discrete diffraction points.

Further, by integrating this in the pattern,

[Equation 6] Becomes

Here, the sinc function sinc (w) is defined as follows. sinc (w) = sin (πw) / πw (26)

If this is used for notation,

[Equation 7] become that way.

Up to this point, this is a strict story that does not include much approximation. The first division such as sin function sin (Sb) / sinb is performed only when b is an integer multiple of π.
And b is 0 at other values. That kx'MΔ _x / 2f = eπ (the e _{integer) (28) ky'NΔ y / 2f} = hπ (h is an integer) (29) only has a value.

In other words, it is the diffraction spot on the image plane.
It is a condition that gives (x ′, y ′), and MΔ _x= D_x, N
Δ_y= D_y(Pattern size), since k = 2π / λ,           x '= 2πef / kMΔ_x= Efλ / d_x      (30)           y ′ = 2πhf / kNΔ_y= Hfλ / d_y      (31)

It is This is the same as the Bragg diffraction condition dsin θ = hλ described above. The lattice points on the image plane are represented by (e, h). The diffracted light exists at the lattice points, but the diffracted light is 0 at the non-lattice points. Therefore, the Fourier transform t _F at the lattice point (x ′, y ′) on the image plane should be obtained from the beginning.

[0077]

[Equation 8]

The last side is (e, h) of the Fourier transform of t
Because it is an ingredient

[Equation 9] And write

T _F (e, h) = π ⁻² STΔ _x Δ _y f ² sinc (ef / M) sinc (h f / N) t _Feh (34) can be written.

It should be noted that this equation does not include the wavelength of light. Wavelength is an expression that determines diffraction point (Bragg condition)
, But is not included in the equation that determines the power at the diffraction point. This means that the power of the obtained diffraction pattern remains unchanged even if the laser wavelength is changed. Such wavelength invariance is by no means trivial.

Here, S is the pattern x included in the DOE.
The number of directions, T is the number of patterns in the y direction included in the DOE, and Δ _x and Δ _y are the side lengths of the cells. M × N cells gather to form a pattern, and S × T cells gather to form a DO
It becomes E. If the area of the DOE is B, then B = STMNΔ _x
Since Δ _y , the above formula is expressed as t _F (e, h) = π ⁻² M ⁻¹ N ⁻¹ Bf ² sinc (ef / M) sinc (hf / N) t _Feh (35). can do.

In one-dimensional diffraction, one sinc function appears,
In two-dimensional diffraction, two sinc functions appear, so D
Only the last t _Feh is newly generated by the presence of the OE.

The fact that the powers are the same at diffraction points within a certain finite range means that | t _F (e, h) | is the same within a certain range (e, h). That is, t _Feh = const × {sinc (ef / M) sinc (hf / N)} ⁻¹ (36).

Then, the inverse Fourier transform is used to obtain t _mn = ΣΣ exp (2πjem / M) exp (2πjhn / N) × {sin c (ef / M) sinc (hf / N)} ⁻¹ (37) Should be. This is analytically true, but since the sinc function has many zeros and its reciprocal has many divergence points, it cannot always be accurately calculated by computer approximation.

Therefore, in actual calculation, t _mn is variously assumed, and this is Fourier-transformed to obtain t _{F eh} , and its absolute value is determined by trial and error so as to take a constant value within a certain range (e, h). A method of gradually making t _mn accurate is adopted. Such calculations are suitable for computers because they do not include divergence.

Now, the conditions for generating an arbitrary diffraction pattern on the image plane placed on the focal plane by the DOE have been sought up to now. If we add the condition that the power is constant within the radius ρ of an image plane, we can write x ′ ² + y ′ ² <ρ ² (38), but since the diffraction points (e, h) are discrete, (e / D _x ) ² + (h / d _y ) ² <(ρ / λf) ² (39).

Especially when the pattern is isotropic, d _x = d _y = d
Therefore, e ² + h ² <(ρd / λf) ² (40). But the story doesn't end here. What has been described above is only the relationship between the light amount distribution at discrete diffraction points and the structure of the DOE cell, which is obtained when the image plane is matched with the focal plane of the lens.

The present invention does not, however, deviates the image plane back and forth from the focal plane of the lens. The purpose is to create a top-hat type beam profile by deliberately defocusing. The defocus δ is defined by δ = g−f (41) where g is the distance of the image plane from the lens.

Looking back, the equation (11) holds true for defocus. However, the equation (12) for the phase shift between the lens surface and the image plane no longer holds, and the equation in which f is replaced by g holds. Phase shift ^{= k {(x'-x)} 2 + (y'-y) 2 + g 2} 1/2 = k [f + {(x-x ') 2 + (y-y') 2} / 2g] (42)

Therefore, the complex amplitude on the defocus surface is a superposition of all the waves from the lens surface, and W (x ′, y ′) = A∫∫U (x, y) exp [jk { (X−x ′) ² + (y−y ′) ² } / 2g] dxdy (43).

When (11) is substituted for U (x, y) in (43), the term of (x ² + y ² ) no longer drops. W (x ', y') = Aexp {jk (x '2 + y' 2) / 2g} ∫∫a (x, y) t (x, y) exp {-jk (xx '+ yy') / g} become ^{exp {jk (x 2 + y} 2) / 2g-jk (x 2 + y 2) / 2f} dxdy (44).

The last term can be written as exp {-jk (x ² + y ² ) δ / 2f ² } (45). Since this is a term having a larger change than the Fourier transform term, it can be adiabatically approximated.

Letting the radius of the lens be R, this can be replaced by (πR ² ) ⁻¹ ∫∫exp {-jk (x ² + y ² ) δ / 2f ² } dxdy (46). The integration range is the entire lens surface.

When converted into cylindrical coordinates, this becomes (πR ² ) ^-1 ∫∫exp {-jkr ² δ / 2f ² } 2πrdr (47), so integration is possible.

It means that sinc (δkR ² / 2πf ² ) = sinc (δR ² / λf ² ) (48). That is, the light quantity at the center of the diffraction point on the defocus surface is obtained by multiplying the diffraction peak light quantity when the image plane is on the focal plane by the square of (48). Defocus δ
Is increased back and forth, the light quantity at the center of the diffraction point becomes sinc
² (δR ² / λf ² ). This gives only a light amount reduction due to defocus at the center of the diffraction point. The spread of the light beam in the vicinity of the diffraction point must be considered by replacing f in equation (12) with g and returning to equation (15).

That is, ∫∫exp [{jk (x-x ')^Two+ (Y-y ')^Two} / 2g-jk (x^Two+ Y ^Two ) / 2f] dxdy (49) Must be integrated over the lens aperture.

[0097] When taking out only the portion of the variable, ∫∫exp {-jk (xx '+ yy') / g-jkδ (x 2 + y 2) / 2f 2} dxdy (50) which can not be calculated analytically.

The term x'is approximately

[Equation 10] It becomes such a value. The term of y'is also the same. However,
x'and y'include the deviation amounts from those satisfying (35) and (36).

[0099]

[Equation 11] Therefore, the defocus surface (x ', y')
The power distribution at

[0100]

[Equation 12] Can be expressed as

Σ means the integration of e and h. Although this is an approximate expression, it can be estimated how much defocus is required to obtain a desired top hat. X ′ is λf / d _x at the defocus image plane,
y ′ has peak points arranged at intervals of λf / d _y (diffraction points)
However, if the power in the middle of adjacent diffraction points should be half the peak (it becomes 1 when superposed),

Since λf / 2d _x = δR / 2f (54) with d _x = d _y = d, it is given by δ = λf ² / Rd (55).

[0103]

EXAMPLE An optical system for a CO ₂ laser was designed according to the principles of the present invention. The overall layout is shown in FIG.
There is a diffractive optical element (DOE) 1 at the left end, and light from a CO ₂ laser travels from the left side. The light of the CO ₂ laser is a parallel beam but has a Gaussian distribution. That is, the power density is high in the central part and low in the peripheral part.
The DOE converts a parallel beam into a total of 20 diffracted beams in five rows and five rows. The beam from the DOE is drawn according to the power density.

Although 11 beams are drawn, each is a set of 20 diffracted beams. Since the diffraction angle is small, it cannot be seen as dispersed. It passes through the condenser lens 2 with the focal length f = 63.5 mm. Since the beams are substantially parallel, they converge at the convergence point F of the focal length f. In the present invention, the object to be processed is provided not at the convergence point F but at the front position K or the rear position M thereof to perform laser beam irradiation.

The light source is CO with a wavelength of 9.3 μm._TwoWith a laser
It Beam diameter (intensity is 1 / e ^TwoΦ)
At 15 mm, the intensity distribution profile is Gaussian.
The diffractive optical element (DOE) 1 was manufactured by the CVD method.
It is made of polycrystalline ZnSe. It is transparent to laser light
Is. As described above, the diffractive optical element 1 has five horizontal lines and five vertical lines.
The light is diffracted into 20 beams. In all respects
Similar diffraction occurs.

The beams of the same diffraction order are parallel at any point. In other words, it does not converge to infinity as it is. So put the condenser lens 2 behind the DOE 1. DO
The distance h between E1 and the condenser lens 2 is 60 mm here. This condenser lens is made of ZnSe and has an effective focal length of f =
It is a 63.5 mm plano-convex lens.

As shown in FIG. 2, the arrangement of the focused spots on the focal plane is designed so that 20 spots are arranged in a substantially circular shape on a grid having a pitch of 0.1 mm. The intended intensity distribution is a circular flat top with a diameter of about 0.5 mm. Therefore, the spread of the target spot distribution in FIG. 2 is set to 5 mm in diameter. When a spot is expressed by coordinates having an origin O at the center and 0.1 mm as one section, (± 1, 0), (0,
± 1), (± 1, ± 1), (± 2,0), (0, ±
2), (± 2, ± 1), and (± 1, ± 2), 20 points. These 20 points are points uniformly distributed in a circle having a diameter of about 5 mm.

However, the origin (0,0) is excluded. Also, (± 2, ± 2) does not exist. This is to make the shape of the collective beam closer to a circle. When a square beam is used, the points (± 2, ± 2) are also defined as the diffraction direction.

The phase distribution of the unit pattern of the diffractive optical device designed at this time is as shown in FIG. The minimum pixel size of this phase distribution (which is called a cell) is 46.137 μm square. The pixels are arranged in 128 × 128 in FIG. 3 (unit pattern). Four
Since 6.137 μm × 128 = 5.9055 mm, the size (of the phase distribution) of the unit pattern of FIG. 3 is 5.
It is 9055 mm square.

The same pattern as such a unit is arranged vertically and horizontally. The number of laser beams determines the number. Since a CO 2 laser having a large beam diameter of 15 mmφ is targeted here, a DOE consisting of 8 patterns × 8 patterns = 64 patterns which can sufficiently cover the laser was manufactured. That is, DOE1 is a diffractive element including 64 identical unit patterns (FIG. 3).

In the unit diffraction pattern, either phase 0 or phase π is given to one cell. The white background is in phase 0. The black part is the part given the phase π. Focusing on the black background, it has a distribution like "λ '". The target diffraction pattern is x as shown in FIG.
The DOE thickness pattern distribution has no symmetry in the xy directions, although it is symmetric in the y and y directions. It has an irregular distribution. The DOE thickness pattern (phase distribution) does not mean that one solution is determined. There can be many possible patterns. There can be many solutions, but one of them is the optimal pattern. Therefore, the DOE thickness pattern (phase distribution: FIG. 3) cannot be predicted from the target diffraction pattern (FIG. 2). It is not possible to intuitively understand the diffracted beam distribution from the DOE thickness pattern.

This is the simplest binary value for the phase (binary; binary value of φ = 0, π). Of course four values (φ
= 0, π / 2, π, 3π / 2), 8 values (φ = 0, π /
4, π / 2, 3π / 4, π, 5π / 4, 3π / 2, 7π
It is also possible to make the phase difference finer, such as / 4). Then, the strength of the top hat may be more uniform. However, doing so would complicate the calculation, and DO
The number of masks for E fabrication and the number of etchings also increase.

The pattern of FIG. 3 is arranged vertically and horizontally so as to cover the entire incident beam diameter φ15 in an actual optical element with this as one unit. Here, the incident beam diameter (1 / e2 diameter) is φ15. In order to receive it without omission, a pattern of 3 × 3 or more is sufficient. Here, we want to set the side length of the DOE to about 400 mm to allow more room. To do so, 5.
A 9055 mm square unit diffraction pattern of 8 × 8 units must be finely processed on the DOE element.

The material of the diffractive optical element 1 is a ZnSe polycrystalline plate having a diameter of 2 inches (50 mm) and a thickness of 5 mm. A ZnSe disk is coated with a resist, and light from a mercury lamp is irradiated through the mask to print the mask pattern on the resist. In ZnSe-DOE fabrication, mask fabrication accounts for a significant portion of the cost.

ZnSe itself was prepared by using a large reactor.
This is done by supplying a source gas and forming a ZnSe plate having a large area on the substrate by the CVD method.
A polycrystal is made by the CVD method. Although the growth rate is slow, it takes time, but a ZnSe plate having a large area can be produced. This board is hollowed out and used as the material.

After development, reactive ion etching (RI
In step E), a fine step structure corresponding to the phase distribution is provided on the ZnSe substrate. The amount of step difference is λ / 2 corresponding to the phase difference π of light.
(N-1) = 3.78 μm.

In FIGS. 4 to 7, the focus position F (from the lens 6
The defocused state when the image plane distance from (1.0 mm) is changed is shown. Each intensity distribution is represented by the density of light rays. The focal length of the lens is 63.5 mm, but since the lens is thick, the plane of the plano-convex lens (rear surface)
To 61.0 mm is the position F of the focal length. It is shown in FIG. Twenty light spots are separated on the image plane and exist at correct positions. A focus beam in which the optical power is concentrated is formed in a narrow portion. At each point, the power distribution has a sharp conical peak. The broken line shows the outline of the part where the power of each light spot disappears.

In FIG. 5, the distance from the rear surface of the lens is longer than the focal point F and M = 61.5 mm. That is 0.5
Defocused to the rear of mm. The light spots have become blurred and the peaks have become lower, so that the lights at adjacent points overlap. In FIG. 6, the distance from the rear surface of the lens is further increased, and M = 62.0 mm.
It is defocused 1.0 mm backward. The boundary between the adjacent light spots becomes unclear, and the light amount becomes nearly constant at the light spot set.

In FIG. 7, the distance from the rear surface of the lens is made longer and M = 62.5 mm. That is 1.5m
It is defocused to the rear of m (ΔM = 1.5 mm).
There is no trace of the separated light spots, and the power is almost uniform in the circular portion with a diameter of 0.5 mm. From the results of FIGS. 5 to 7, it can be seen that a uniform top hat type power distribution can be obtained by defocusing about 1.5 mm.

Here, the image plane is shifted rearward, but the image plane may be shifted forward (closer to the lens side) than the focal plane F. The amount of deviation at the front position K is the same as before,
A uniform power distribution is obtained at about ΔK = −1.5 mm. It can be seen that the distance 61 mm from the laser emission surface of the condenser lens is the focal plane, and when defocusing by 1.5 mm, a substantially uniform light density is obtained within a region of 0.5 mm.

[0121]

As described above, according to the present invention, a branched beam is generated using a diffractive optical element having a pattern capable of distributing an equal amount of energy to lattice points to be included in a top hat, and defocused and superposed. A uniform intensity distribution is obtained. Since the diffractive optical element only has a surface with two steps (or four steps) in height, it can be easily manufactured by photolithography if the pattern is determined. It does not require complicated machining such as an aspherical lens, and the same one can be easily created.

Unlike the case of an aspherical lens, a laser having any power distribution can be converted into a top hat type power distribution without depending on the sectional shape of the laser beam.

[Brief description of drawings]

FIG. 1 is a configuration diagram showing an optical system including a DOE, a condenser lens, and an image plane according to an embodiment of the present invention.

FIG. 2 is a photograph of a focused diffracted beam spot produced when an image plane is placed exactly on the focal plane in the optical system including the DOE and the condenser lens according to the embodiment of the present invention.

FIG. 3 is a view of a diffractive optical element 2 for generating a diffraction pattern on a focal plane that gives a power distribution equal to that of a diffraction point as shown in FIG.
A surface pattern having a step thickness, a white background indicates a phase 0 point, and a black character indicates a phase π point.

FIG. 4 is a diagram showing a diffraction pattern generated when the image plane is exactly aligned with the focal plane (F = 61.0 mm) in the optical system including the diffractive optical element and the lens described above. Power density is represented by a black background. The spots at the diffraction points are separated. It seems to be different because there is a difference in negative and positive, but it is equivalent to the photograph in FIG.

FIG. 5 is a diagram showing a diffraction pattern generated when the image plane is defocused 0.5 mm behind the focal plane (M = 61.5 mm) in the optical system including the diffractive optical element and the lens. Power density is expressed by black letters. The black areas overlap each other, and the power density in the circular area is uniform.

FIG. 6 is a diagram showing a diffraction pattern generated when the image plane is defocused 1.0 mm behind the focal plane (M = 62.0 mm) in the optical system including the diffractive optical element and the lens. Power density is expressed by black letters. The black areas overlap each other so that they cannot be distinguished, and the power density is uniform in the circular area.

FIG. 7 is a diagram showing a diffraction pattern generated when the image surface is defocused 1.5 mm behind the focal plane (M = 62.5 mm) in the optical system including the diffractive optical element and the lens. Power density is expressed by black letters. The black area is indistinguishable, and the outline of the area with light approaches a circle.

FIG. 8 is a power distribution diagram on the beam diameter of a Gaussian beam seen in a carbon dioxide laser or the like.

FIG. 9 is a power distribution diagram on the diameter of the top-hat type beam which is the object of the present invention.

FIG. 10: A plane wave is input to a diffractive optical component to generate a plurality of parallel diffracted beams, and a condensing lens is used to form a diffracted image spot focused on an image plane on a finite focal plane. Of the diffractive optical system shown in FIG.

FIG. 11 is a diagram showing a height distribution of a pattern which is a basis of the diffractive optical component. The black background represents the phase 0, and the white background represents the phase π. This means that there are two layers of different thickness. The step is λ / 2 (n-1) so that the phase of the transmitted light differs by π.

[Explanation of symbols]

1 DOE (diffractive optical element) 2 condenser lens 3 image plane

Claims (12)

[Claims] 1. A diffractive optical element having a function of branching a laser beam into a plurality of beams within a certain range, and a condenser lens provided subsequent to the diffractive optical element, which is out of focus of the condenser lens. The image plane is installed on the defocused surface,
A laser optical system characterized in that beams branched by a diffractive optical element are combined on a defocusing surface to obtain a desired intensity distribution. 2. A diffractive optical element with a lens, which has both a function of branching a laser beam into a plurality of beams within a certain range and a function of condensing the beam, and an image plane is provided on a defocused surface out of focus. Then, the laser optical system is characterized in that the beams branched by the diffractive optical element with a lens are combined on the defocusing surface to obtain a desired intensity distribution. 3. A circular intensity distribution having a uniform power within a circle assumed on the defocus image plane and a power decreasing to 0 outside the circle is created. Laser optical system according to. 4. A ring-shaped intensity distribution having uniform power within a certain ring assumed on the defocus image plane and reducing to 0 outside the ring. Laser optical system according to. 5. A quadrangle-shaped intensity distribution having uniform power within a quadrangle assumed on the defocus image plane, and having a power decreasing to 0 outside the quadrangle. Laser optical system according to. 6. A linear region-shaped intensity distribution having uniform power within a certain linear region assumed on the defocus image plane and decreasing to 0 outside the linear region. The laser optical system according to claim 1 or 2. 7. A diffractive optical element having a function of branching a laser beam into a plurality of beams within a certain range, and a condenser lens provided subsequent to the diffractive optical element, which is out of focus of the condenser lens. The image plane is installed on the defocused surface,
The beams branched by the diffractive optical element are combined on the defocusing surface to obtain a desired light intensity distribution, and the processing target is processed by placing the processing target at the defocus image plane position. A laser processing method characterized by the above. 8. A diffractive optical element with a lens having both a function of branching a laser beam into a plurality of beams within a certain range and a function of condensing the same is included, and an image plane is provided on a defocused surface out of focus. Then, the beams split by the diffractive optical element with a lens are combined on the defocus surface to obtain the desired light intensity distribution, and the object is processed at the defocus image plane position. A laser processing method characterized in that. 9. An intensity distribution having uniform power within a circle assumed on the defocus image plane and decreasing to 0 outside the circle is created, and an object to be processed is placed at the defocus image plane position. The laser processing method according to claim 7 or 8, wherein the object to be processed is processed. 10. An intensity distribution that creates uniform power within a certain ring assumed on the defocus image plane and decreases to 0 outside the ring, and places the object to be processed at the defocus image plane position. The laser processing method according to claim 7 or 8, wherein the object to be processed is processed. 11. An intensity distribution is created such that the power is uniform within a certain quadrangle assumed on the defocus image plane and the power is reduced to 0 outside the quadrangle, and the object to be processed is placed at the defocus image plane position. The laser processing method according to claim 7 or 8, wherein the object to be processed is processed. 12. An intensity distribution having uniform power within a certain linear region assumed on the defocus image plane and decreasing to 0 outside the linear region is created, and the power is processed at the defocus image plane position. 9. The laser processing method according to claim 7, wherein the object to be processed is processed by leaving the object.