JP3371754B2 - Ion implanter - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [0001] The present invention relates to a semiconductor wafer,
Large current ions for large workpieces such as glass substrates
The present invention relates to an ion implantation apparatus for realizing implantation. Especially
Multiple useful ion species with similar masses, except for ions of interest
Refers to the ion implanter that is to be implanted into the sample.
You. For example, in the case of Si wafer, boron, phosphorus, arsenic
Ion implantation to dope p-type and n-type impurities
Often. To increase throughput, use a uniform
An apparatus for doping with a high distribution and a high speed is desired. [0002] 2. Description of the Related Art Conventional high current ion implanters are roughly classified.
There are two types. One is thin (0-dimensional spread
High quality ion beam with fan magnet
Rotation that translates and rotates only the desired ion species by mass analysis
Irradiates the workpiece (wafer) placed on the target
Is what you do. Fan magnet because it is a thin ion beam
Is easy to separate by mass. But the beam itself
Irradiates the entire surface of the wafer as the body does not have much spread
To do so, the wafer must be translated and rotated.
Two-dimensionally at the end station supporting the wafer
You need to move the wafer. [0003] Another is a large-diameter ion source to a large-diameter ion source.
Extract ion beam and illuminate wafer without mass analysis
To shoot. The diameter of the wafer is W and the ion source
Let D be the diameter of the ion beam extracted from. D>
Since a large diameter W ion beam is generated, the wafer
Need not be scanned. [0004] [Problems to be solved by the invention] 1. In the method using a rotating target, a thin 0-dimensional beam
The beam optics are simple because they generate a beam. However
End stations that must perform rotational translation
The structure of is complicated. The disk with the wafer is high
Fast rotation and proportional to the ion beam current.
Translational speed control must be performed in proportion.
Also, as the beam current density at the wafer increases,
The charge-up phenomenon becomes remarkable. Charge up is positive
Or charge does not escape when negative ion beam is implanted,
This is a phenomenon in which electric charges are accumulated on a surface such as a wafer. This
Damage to devices on the surface of the wafer
Is also undesirable. Ion implantation into large diameter wafers
This rotation is actually used as a method
Although the target method is used, there is such a disadvantage. [0005] 2. Large-diameter ion source non-mass separation method
Since the system optical system is only an ion extraction system,
The result is even simpler. End station structure
Simple. However, large-diameter ion beams are mass analyzed
Can not distinguish between ion species other than desired ion
Instead, it is injected into the wafer. Note on impurity ions
In some cases, the desired characteristics may not be obtained due to the insertion. In addition
The injection depth distribution is
It changes depending on the state of the plasma. Constant injection depth
Another drawback is that the device characteristics are difficult to maintain
You. For the latter, for example, if boron
If diborane (B_Two H₆ ) Is used as a source gas
You. At this time, hydrogen ions are injected into the wafer along with boron.
Is entered. Since the implanted hydrogen is fast, gate oxidation
May penetrate the membrane and be injected into the channel layer.
You. This causes lattice defects in the channel below the gate electrode.
May cause. All ion energy is converted to heat
But ions other than the desired ion species (eg, hydrogen
Ion), which heats the sample
Would. Device on sample is destroyed by temperature rise
There is a fear. [0006] 3. The disadvantage of the large-diameter ion source system is that much
is not. 8 inch wafer and 12 inch wafer
When trying to implant ions into a large area substrate
Injection uniformity within several% is required everywhere on the face
It is. An ion beam with such a large area and uniform beam density
Realizing the source is an extremely difficult technique. Large diameter ion source
At present, small area wafers with less demanding injection uniformity
Can only be used for Fah. [0007] 4. Ion implantation without mass spectrometry
In addition, the PIII method (Plasma ImmersionIon Implantatio)
n) has been proposed. Shu Qin and Chung C
han, "Plasma immersion ion implantaion doping exper
iments for microelectronics ", J.Vac.Sci.Technol. B
12 (2), (1994) p962. This is the Si wafer in plasma
By applying a negative voltage to the wafer
By accelerating ions in the sheath region and implanting them into the wafer
is there. Using the sheath voltage to extract the ion beam
Therefore, no extraction electrode is required. Often impurities are
The electrode was sputtered. in this way
There is no possibility that impurities generated from the electrodes are mixed. So
This means that mass spectrometry does not have to be performed. However, unnecessary arising from the elements contained in the raw material gas
ON cannot be excluded. The above-mentioned boron doping field
In this case, diborane is used as the source gas, but hydrogen ions
Injected into wafer. Same for hydrogen ion and boron ion
With the same acceleration energy. The hydrogen ion implantation
Fa is overheated. During injection to increase throughput
When the time is shortened, the temperature rise due to hydrogen implantation becomes remarkable.
Therefore, the injection time cannot be shortened too much. To one wafer,
1.9 × 10 in 10 minutes^Fifteencm^-2Density ion injection
Has been reported to have entered. This will increase throughput
Too low. I want to implant ions within one minute
You. In fact, this method does not appear to be practical. [0009] 5. Until now, a zero-dimensional beam (cross section
Mass spectroscopy of the beam at point 2) and a two-dimensionally scanned wafer
Beam injection into large diameter two-dimensional beam (cross section)
Of a stationary beam without mass analysis
And the method of injecting into the plasma without using an ion source.
The PIII method of implanting ions into a wafer has been described. 2
Mass spectrometry in the second large-aperture ion beam method
If so, the problem should be resolved. But a large beam
Attempts to mass spectrometer have not yet been realized.
Light elements such as hydrogen can be removed by mass separation.
It is a first object of the present invention to provide an ion implantation apparatus
It is a target. Even if the ion species contains the target element,
Distinguish ion species that differ greatly, and
Uniform implantation depth distribution by implanting only ON species
It is a second object of the present invention to provide an ion implanter
It is a target. Provide high-throughput ion implantation equipment
Is a third object of the present invention. The first object is, for example, no need for hydrogen ions.
That is, to eliminate light ions. High speed H_Two ⁺,
H⁺ Channels are defective when ions are injected into the sample
And kinetic energy is released
Overheats the wafer. The second purpose is more subtle
Things. If a certain dopant is implanted,
Suppose there are several ionic species. Ion species with different masses
When implanted, the depth varies significantly. Injection depth varies
And the properties such as bonding deteriorate. A junction is a localized point
Should be formed. Dopant implantation depth
Don't pull it long. [0011] Therefore, the ionic species containing the desired dopant
Even if it is, be sure to inject only those with a mass within the specified range.
It is. For example, when using boron as a dopant
In addition, if diborane is used as a source gas, one boron
Containing ion B⁺ , BH⁺ , BH_Two ⁺And two boron
Containing ion B_Two ⁺, B_Two H⁺ , B_Two H_Two ⁺,… Etc. occur
You. In that case, the ion containing one boron is light,
To go deep inside. This varies the implantation depth
You. Conversely, the ion species containing two borons is B_Two ⁺~ B_Two H₆ ⁺
However, these are similar in mass number and
Since there is no distribution at the depth of inflow, it is safe to inject all
No. Rather, implanting all the ion species including two borons
The raw material gas can be used effectively and the injection process can be shortened.
It is useful. [0012] SUMMARY OF THE INVENTION The apparatus of the present invention comprises
Width smaller than the sample diameter and vertical width larger than the diameter.
An ion source that produces a vertically elongated ion beam
A vertically long yoke having a long ion beam path, and a yoke
The main coils M1 and M2 wound around the center of the
First auxiliary coil A wound side by side at one end of the yoke
1, A2 and wound around the other end of the yoke alongside the main coil.
Vertically long mass consisting of the inserted second auxiliary coils B1 and B2
Analysis magnet and position detection to check the distribution of the ion beam
And the ion species separated by the mass spectrometer magnet
That is, only ions having a mass number in a desired range
Slits to block ion species with mass numbers other than
The sample so that the ion beam that has passed through the
Hold the sample in a direction perpendicular to the longitudinal direction of the ion beam.
And a moving sample holding mechanism. The apparatus of the present invention first emits a vertically elongated ion beam.
There is a first feature in that it is generated. Sample
Since a vertically long beam larger than the diameter W is used, the sample
It only needs to be moved in the lateral direction of the system. For injection time
Can be shortened and the throughput is improved. However
When trying to bend a vertical beam with a magnet,
Since the magnetic pole interval is extremely long, a uniform magnetic field is unlikely to occur. this
Is the problem. Bending a narrow beam requires a short distance between magnetic poles
It's easy. Even if the beam is vertically long,
Mounting is relatively simple because the distance between the magnetic poles is short. I
Do not bend at a considerably large angle when bending in the vertical direction
Mass spectrometry magnets
It will be. The present invention attempts to bend a long beam horizontally.
I do. That is, it is bent in the direction of the short side. Bend angle
Is not so necessary, but because the distance between the magnetic poles is long,
difficult. Increase the yoke cross-section to increase the number of coil turns and current
If the product AT is increased, a certain amount of magnetic flux
Degrees can be formed in space. But even in that case
It is extremely difficult to produce a uniform magnetic flux density in the vertical direction.
No. Why should a uniform magnetic field be created in the vertical direction
Can't you? If not a uniform magnetic field, the bend angle depends on the location
Because the degree is different, the cross section of the beam will be distorted and correctly
Mass spectrometry cannot be performed. Whole vertical beam
If a uniform magnetic field can be obtained, a beam with a rectangular cross section
If so, it is a rectangular cross section even after mass analysis. Accordingly, the present invention provides a method for long magnetic pole intervals.
It is devised so that such a magnetic field can be formed. Yo
Main coil, first auxiliary coil, second auxiliary coil
A uniform magnetic field between long magnetic poles
I let them live. [0016] A uniform longitudinal magnetic field causes the longitudinal beam to
Bend to the side and mass separate the ions. Has a vertically elongated opening
Slits allow only ion species in the desired range of mass numbers
Inject into the selected object (wafer). The object (sample)
The beam moves in a direction perpendicular to the long side of the beam
Thus, the ion beam can be implanted over the entire surface of the sample. [0017] FIG. 1 shows an ion implantation apparatus according to the present invention.
It is a schematic longitudinal cross-sectional view. FIG. 2 is a cross-sectional plan view of the same.
You. If the vertically elongated ion source 1 is the vertically elongated ion outlet 2
Thus, a band-like (vertical) beam 3 is generated. Beam length
The width is 2d. Sample (wafer)
Assuming that the diameter is W, the vertical length T of the beam is larger than W,
The width 2d of the beam is smaller than W (T> W> 2d). Bi
In the path of the beam 3, there is a longitudinal magnetic field By parallel to the long side T of the beam.
Is formed. Here, the coordinate system is defined. Beam
The traveling direction is the z direction, the direction of the short side of the longitudinal beam is the x direction,
The direction of the beam long side is defined as the y direction. Strip beam 3 is longitudinal
Under the Lorentz force for the field By, it turns in the x direction. Song
If the rate radius is R, [0018] ByR = (2MV / q)^1/2       (1) ## EQU1 ## q is the charge, V is the extraction voltage, M is the quality
Quantity. Heavy ions have a large R. In other words, it is hard to bend
No. Light ions have a small R. That is, it is easy to bend. light
The ions bend like a beam 5. Heavy ions
It hardly bends like beam 4. Light ions are hydrogen
Ion (H⁺ , H_Two ⁺, H_Three ⁺). Heavy io
In the example of B doping described above, one boron
⁺ , BH⁺ , BH_Two ⁺And boron two B_Two ⁺, B_Two H
⁺ , ~, B_Two H₆ ⁺And so on. [0020] A slide is provided so as to be able to move back and forth during the path of the ion beam.
And a position detector 11. Slit 7 is vertical
A plate having a long hole 8 in the short side direction (x direction) of the beam.
Direction). A passed through hole 8
Only ON reaches the wafer. The width ε of the hole 8 is ion injected
What is necessary is just to determine according to the width of the mass distribution of the ion to be input. The position detector 11 first detects the ion beam.
It is for examining the spatial distribution. Vertical spread in xy directions
To observe the long beam profile,
It must be a position detector having a resolution. Two-dimensional case
Use a Faraday cup with ion detectors arranged like a child
be able to. In addition, the one-dimensional position detector
Measure the two-dimensional distribution by displacing in the direction perpendicular to
Can also be. The position is detected when the beam rises.
And the beam cross section when the beam is bent by the magnet
It's only time to see if the shape is appropriate. Real
When ion implantation is performed, pull in the position detector 11
Instead, put the slit 7 on the front. Vertical of slit 7
Only the ion species that have passed through the long hole 8 reach the sample 6 and
Is injected. Now, mass spectrometry will be described. The present invention
Creates a uniform magnetic field By in the long side (y direction) of the beam
However, this attempts to mass analyze the beam. Lid
One problem must be pointed out. One is between the magnetic poles
Because the distance is too long (for example, 40 cm)
Due to the large resistance, it is necessary to give a large magnetomotive force (AT)
It must be. Just give the magnetomotive force
A large current should flow, but then the yoke saturates
There is a fear. The other is non-uniformity of magnetic flux density between magnetic poles
It is. Simply put the coil in a yoke consisting of an elongated closed loop
In the case of rolled material, uniform due to magnetic flux leakage from the yoke surface
The magnetic field cannot be formed. Strong at the center, at both ends
It becomes a weak magnetic field. It bends strongly near y = 0 and y = 0
When you move away from it, you will bend weakly. Then it's an ion source
When passing out, even though it has a rectangular cross section, it passes through the magnet
And it will be distorted in the bow section. The beam cross section simply curves
If you only need to bend the hole 8 of the slit together
is there. However, it is not only curved, but also a part away from y = 0
Of the ion species with different masses overlap
I will. Here, the mass separation becomes incomplete. Beam cross section
It must not be distorted. First, the problem of yoke saturation is described first.
Bell. Let g be the magnetic pole interval, and let the magnetic flux density in a vacuum between the magnetic poles be
B, the magnetomotive force NI applied to the yoke is NI = Bg / 4π × 10^-7        (2) Can be roughly evaluated. Yoke permeability
Is sufficiently large compared to the magnetic permeability of vacuum,
It consumes almost no magnetomotive force, and most of the magnetomotive force is applied between the magnetic poles.
Call 4π × 10^-7That is the vacuum permeability. g
Is 40 cm, the magnetic flux density in vacuum is 600
NI = 19100A as Gauss (0.06 Tesla)
It becomes T. In the case of such a magnetomotive force, make the yoke thick enough.
Then, the magnetic flux density in the yoke can be reduced to about 5 kG to 6 kG.
You. Ferromagnetic material used for ordinary yoke if this degree
Does not saturate. The next problem is that the magnetic flux density in the vertical direction (y direction)
Non-uniformity. Inhomogeneous field formation can lead to beam
Since the bending becomes uneven in the vertical direction, mass separation is
Be complete. Therefore, the present invention is specially designed to form a uniform longitudinal magnetic field.
We have another ingenuity. As can be seen in FIG.
Six independent coils such as main coil and auxiliary coil along the longitudinal direction
Provide a controllable coil to prevent magnetic flux density leakage inside the yoke.
I try to prevent that. FIG. 3 shows the yoke 10 and the six coils.
I have. The yoke constitutes a ferromagnetic closed magnetic path. Rectangular
It is long in the y direction and short in the x direction. X frame in x direction
The frame and the frame in the y direction are called Y frames. Right in the Y frame
Six coils around, A1, M1, B1, B2, M2,
A2 is provided. The center main coils M1 and M2 are
Is also a coil with many turns. The winding direction is the same
In each case, an upward magnetic flux is generated. Vertical circulation
It does not create a mold flux. Below the main coil
Have first auxiliary coils A1 and A2. This is also the winding direction
They are the same and generate an upward magnetic flux. From the main coil
Above there are second auxiliary coils B1, B2. This is also upward
To generate magnetic flux. That is, this yoke has a closed curved magnetic path.
However, the magnetic flux created by the coil
It is not something that goes around. Three coils in the left Y frame
B1, M1, and A1 all provide an upward magnetic flux. Right Y
The three coils B2, M2, and A2 in the frame also generate an upward magnetic flux.
create. The magnetic flux collides at the center U of the X frame, and from the yoke up and down
Jump out. When the yoke is projected upward, the yoke is enlarged.
The outer frame returns to the center d of the X frame on the opposite side. This is a magnetic
The loss of the bundle. Those that have jumped downwards
It becomes the magnetic flux density By. This penetrates the mass analysis space 12 vertically.
Then, it returns into the yoke at the center D of the lower X frame. Like this
In the workpiece 10, the magnetic poles are formed at the midpoints U and D of the upper and lower X frames. right
The magnetic flux on the side passes through A2, M2, and B2 from bottom to top, and
Descends from the middle point U of the upper frame of the
Return to the middle point D of the lower frame and pass through A2, M2 and B2
Draw a closed loop. The magnetic flux on the left is below A1, M1, B1
From above, descend from the middle point U of the upper frame of the yoke 10
After passing through beam 3, return to the middle point D of the lower frame,
Draw a closed loop passing through M1 and B1. As described above, the magnetic flux is strong at the middle points U and D of the X frame.
Collision and the magnetic field energy increases here,
Leakage of magnetic flux from the yoke nearby is large. This is the problem
is there. If there is only a pair of main coils M1 and M2,
By is uniform in y direction due to magnetic flux leakage in the middle of the yoke
No. FIG. 5 shows that only the main coils M1 and M2
A current of 0 AT flows, and the auxiliary coils A1, A2, B1, B
2, the y direction of the magnetic flux density By (y) when no current is applied
It is a distribution of directions. The length of the yoke opening in the y direction is 40c.
Assuming that m is 700 gauss at the center y = 0. Only
And at both ends y = ± 15cm, it becomes only 570 gauss
No. By becomes non-uniform in the y direction. pass near y = 0
The beam bends strongly, but around y = ± 15-20cm
The passing beam has a small bend. Therefore, auxiliary coils A1, B1, A
2 and B2 are provided and a large current is passed through them to
To increase the magnetic field. FIG. 6 shows the main coils M1 and M2
A current of 10000AT is applied to each of the auxiliary coils A1, A
A current of 5000AT is applied to each of 2, B1 and B2.
7 shows the magnetic flux density By (y) in the y-direction at the time of turning on. Conversely, central
In the part, By is weak and 570 gauss. y = ± 15c
At m, it is 680 Gauss. Main coil current and auxiliary coil
Current in a well-balanced manner,
, A uniform By (y) can be generated. On purpose
Therefore, an inverted distribution as shown in FIG. 6 is also useful. First, with no current flowing through the magnet
Position detector with two-dimensional resolution that generates an ion beam
11 (Faraday cup etc.)
Find the distribution. The ion beam goes straight and moves to the position detector 11.
to go into. This helps to determine the shape of the rectangular (strip) beam.
You. Next, a current is applied to the main coils M1 and M2 to distribute the beam.
Look at the deformation of. Since the vicinity of y = 0 bends particularly strongly, it is rectangular
The beam deforms. This also eliminates mass separation
It is. But the beam is distorted. A current is also applied to the auxiliary coil, and the beam cross section is
Monitor the deformation of. Beam of light ions such as hydrogen
And an ion beam containing one boron and boron
The beam splits into three heavier ion beams. So
Each beam becomes a beam with the same cross section as the first beam
Adjust the currents of the auxiliary coil and the main coil so as to achieve. Sa
Only ions containing two borons can pass through hole 8
Thus, the position of the slit 7 is determined. After that, slit
7. Fix the coil current and move the wafer in the x direction.
While implanting ions containing two borons into the wafer surface
I do. The wafer focuses the beam over the entire surface with only translation in the x direction.
You can enter. Conventional large mouth that simultaneously performs high-speed rotation and translational movement
In the end station of the ion implanter for the diameter wafer
It is greatly simplified in comparison. [0037] [Example] B ions on a silicon wafer 30 cm in diameter
Will be described. The yoke opening is vertical
Direction length (distance between magnetic poles) is 40 cm,
The length is 30 cm. Thickness of the magnetic pole in the flow direction (z direction)
The size is 20 cm. The yoke has two main coils, 4
There are two auxiliary coils wound. Coil section is auxiliary carp
Is 5cm x 10cm and the main coil is 5cm x 20cm
It is. As shown in Fig. 2, the axial direction of the magnet
The axial direction of the source is shifted by 6.15 ° from the beginning. Magne
From the beginning, the axial direction of the cut and the axial direction of the wafer are 6.
15 ° off. Magnetic pole of yoke (near point U and point D)
The opposing surface is not flat so as to have a sector pole component
A radius of curvature of 67.47 cm is provided. by this
Assuming that By is 1, 0.000449 (x^Two / R^Two 2)
The following components can be realized. The thickness of the magnetic pole is 20 cm
The effective magnetic field spread in the direction is 61.6 cm. The case of doping with boron will be described. material
The gas is diborane B_Two H₆ Gas. This is an ion source
Into the plasma, and by the action of the extraction electrode
Pull out as a beam. Acceleration energy is 20 keV
It is. Height 40 cm, width from vertical ion source 1
A 5 cm band ion beam is generated. Initially a magnetic field
The beam travels straight without hanging and two-dimensional position detection in front of the beam
Two-dimensional components of the ion beam in the x and y directions
Examine the cloth. One peak is present because mass spectrometry has not been performed.
It is. You can see the initial shape of the vertical beam profile. The two-dimensional signals of these beam intensities are parallelized.
Read serially or read
Performs appropriate mathematical processing on a personal computer to produce a two-dimensional beam distribution.
Can be displayed on the CRT. In this state, hydrogen ions and
Ron ions make overlapping peaks. Beam cross section
Is 40cm x 5c, similar to when exiting the ion source
m. Next, the main coils M1 and M2 are energized to
A vertical magnetic field By is generated from the magnetic field. The magnetic field distribution is as shown in Fig. 5.
become. The ion beam bends in the x direction and the peak is almost
Separate into three. Light ion beam 5 containing only hydrogen
(Fig. 2) is the most bent ion containing one boron
Beam 13 then bends well. Bee containing two boron
The arm 4 is hard to bend and is separated outward. In this state
In each case, the beams 5, 13, and 4 are distorted. Further, auxiliary coils A1, A2, B1, B2
To excite it. Current of main coil M1, M2
While increasing the auxiliary coil current.
The distribution of each ion beam 5, 13, 4 is the original rectangle
What is necessary is that it is in a shape (band) and can be completely separated. In that state
Main coil and auxiliary coil currents are fixed. Vertical magnetic flux at this time
The density was 567.46 gauss at y = 0, x = 0. For the magnet, the ion source is
The beam is also approximately 6.15 ° in the x direction, so the beam is roughly
It can pass through the center of the magnet. The beam is a wafer
At almost right angles. Beam cross section is long in the y direction
The substrate 6 is translated in the x direction and
Injection. The separation of ions containing only hydrogen is shown in FIG.
No problem occurs in the magnet path. Here's the question
Are two ions containing one boron (BHn) and two
Containing ions (B_Two Hn). Both are considered monovalent ions
I have decided. The + sign on the right shoulder is omitted for simplicity. (1) Ion containing one boron (monomer) B, B
H, BH_Two , BH_Three (2) Ion containing two borons (dimer) B_Two ,
B_Two H, B_Two H_Two , B_TwoH_Three , B_Two H_Four , B_Two H_Five Since both groups contain boron,
It becomes a p-type impurity, but has the same energy but different mass
And the monomer penetrates deeper. Both are mixed
Large variations occur in the depth distribution. Still good
Of course, there are cases. However, there is also a demand for making a shallow junction.
You. In that case, inject only the dimer with large mass
There is a need. For this purpose, the monomer is blocked by slits.
Must be refused. BH with one boron_Three But
Although heaviest, it has a mass number of 14. Two boron
Is B_Two Is the lightest and has a mass number of 22. Most heavy
Ino B_Two H_Five And the mass number is 27. Of slit hole
Choose a suitable size and dimer with a mass number of 22-27
It is possible to let everything go through. And the monomer
It is also possible to drop them all. However, the web of ions in a certain mass number range
If the spread in the x direction at the key is Δm, the width of the slit hole
Does not mean that Δm should be Δm. From the ion source
When exiting, the beam has a finite width 2d, so a slit
Hole has a width of Δm + 2d, the entire mass number range
You can pass the ions of the part. But then,
Unnecessary ions with similar mass numbers such as nomers can pass through
Performance comes out. So the width of the slit hole is ｛Δm^Two
+ 4d^Two ｝^1/2 May be made smaller. This is
And all ions within the specified mass range cannot pass
But most can pass. Necessary in such a case
Next, the magnetic field and the range of the magnetic field are calculated. FIG. 4 shows a vertically long bead having a width of JHG (2d).
This indicates that the bend is bent by the Faraday force of the magnetic field. An effective
Let L be the width of the region where the magnetic field exists. Monomer (BH
n) is a small radius R about point C₁ Draw an arc of.
Central angle₁ And The monomer beam is
Because it has an expanse, it spreads to the SQP range even at the exit
You. Intersection of the line drawn in the z direction from the center point H and the line at the end of the area
Let F be the point. The dimer ion is large around point E
Threshold radius R_Two Draw an arc of. Central angle_Two And Die
The mer beam also has a JHG width at the entrance, so the exit is N
Has the spread of MK. The most x value in the exit distribution of the dimer beam is
What is larger is point N. The outlet distribution of the monomer beam
It is at the point P that x is also small. Thus X_N <X_P so
If there is, put the end of the slit hole between PN
You can drop all Nomers. Of course the beam angle
You can actually afford more because it is different. [0049] R₁ = (M₁ V)^1/2 /0.69B (3) [0050] R_Two = (M_Two V)^1/2 /0.69B (4) Here, the unit of the radius R is cm, and the unit of the magnetic flux density is
Tesla, M₁ , M_Two Is the mass number (M₁ Maximum value of
= 14, M_Two = 22). The unit of the acceleration voltage V is 1
0 keV. Central angle of beam bending in magnetic field region L
Θ_Two , Θ₁ Is R₁ sinΘ₁ = L (5) R_Two sinΘ_Two = L (6) Is determined by [0055] FN = R_Two -R_Two cosΘ_Two + D (7) [0056] FP = R₁ -R₁ cosΘ₁ −d (8) Is as follows. Separation condition is FP> FN
From [0058]         R₁ -R₁ cosΘ₁ -D> R_Two -R_Two cosΘ_Two + D (9) It is sufficient if it is. Specifically, the width L of the magnetic field region
50 cm, beam half width d 2.5 cm, beam energy
And the ions to be separated are BH_Three
(14) Ion and B_Two (22) If it is an ion, it is necessary
The magnetic flux density is about 500 gauss. A frame-shaped yoke
The range of magnetism that can be generated by the electromagnet used
The bundle density. Here, the vertical magnetic flux density Br is 567.4.
6 gauss. Separates dimer and monomer
Enough magnetic flux density. In this example, the dimer
The radius of curvature of the beam by the magnet was 168 cm. [0060] According to the present invention, a band-like ion beam is
The ions are mass-analyzed by being bent by a magnet in the direction. material
When gas is ionized, unnecessary light ions and necessary heavy ions
If they contain ON, separate them and remove unnecessary light ions.
Excluding Compared to non-mass spectrometer type ion implanter
As a result, heat generation is small, and adverse effects due to impurity implantation are small. When implanting boron, one boron
On and separate the two ions of boron, only the latter
Can be injected. When forming shallow junctions, the depth distribution varies.
Sticking is a problem. The present invention selects only dimers,
(BHn) can be dropped, so shallow junction
Ideal for forming. Multiple types of heavy ions required (B_Two ~
B_Two H_Five ) In some cases, can all of them be ion implanted?
Raw materials can be used effectively and the injection processing capacity is large.
The object (wafer) only needs to be reciprocated in one direction.
Therefore, the structure of the end station is simplified. Since the beam slit is provided,
Particles from ON source reach wafer
Can be prevented. Particle-based wafer
Pollution can be avoided. Conversely, caused by injection
Resist scattered matter that flows back to the beam line
Can be prevented. Small bending angle and short beam line
No. In this example, the beam line must be 100 cm or less.
Can be. Therefore, beam divergence due to the space charge effect is suppressed.
And beam loss is small. Since the vertically elongated ion beam is bent sideways, the beam
Outside (x> 0) and inside the central orbit (x = 0 plane)
(X <0), the traveling distances of the beams are equal. This is a beam
Is narrow (2d). Are the flight distances equal?
When the beam flies, it collides with gas molecules to convert charges
The probability of becoming more neutral is almost the same everywhere in the cross section
You. This can cause poor injection uniformity on the wafer.
It can be removed.

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a schematic longitudinal sectional view of an ion implantation apparatus of the present invention. FIG. 2 is a cross-sectional plan view of the ion implantation apparatus of the present invention shown in FIG. FIG. 3 is a front view showing a relationship between a yoke and a coil of a longitudinal beam mass analysis magnet used in the present invention. FIG. 4 is a diagram showing an example in which a vertically elongated beam is bent in a horizontal direction by a magnet, so that ions containing one boron (BHn:
FIG. 4 is an explanatory diagram showing that mass spectrometry can be performed on a monomer) and an ion containing two boron atoms (B ₂ Hn: dimer). FIG. 5 is a graph showing a y-direction distribution of a magnetic flux density By in a yoke opening when a current of 20000AT is applied only to a main coil. At y = 0, the magnetic flux density becomes excessive, and as the distance increases, the magnetic flux density becomes too small. FIG. 6: 10,000AT for main coil and 50 for auxiliary coil
9 is a graph showing a y-direction distribution of a magnetic flux density By in a yoke opening when a current of 00AT flows. The magnetic flux density is weak near y = 0, and large near y = ± 15 cm. [Description of Signs] 1 Vertical ion source 2 Ion extraction port 3 Strip (vertical) ion beam 4 Beam containing the heaviest ions 5 Beam containing the lightest ions 6 Wafer 7 Slit 8 Slit hole 9 End station 10 Mass analysis magnet Yoke 11 Position detector 12 Beam mass analysis area 13 Beam containing ions with intermediate mass

Continuation of the front page (56) References JP-A-10-302707 (JP, A) JP-A-9-251848 (JP, A) JP-A-60-100351 (JP, A) JP-A-11-288682 (JP) JP-A-5-67448 (JP, A) JP-A-63-48738 (JP, A) JP-A-8-7822 (JP, A) JP-A-6-342639 (JP, A) 5-34653 (JP, U) Table 11-500573 (JP, A) Table 2000-505234 (JP, A) (58) Fields investigated (Int. Cl. ⁷ , DB name) H01J 37/317 C23C 14/48 H01J 37/04 H01J 37/05 H01L 21/265 603

Claims (1)

(57) [Claims 1] A width smaller than a diameter of a target sample,
An ion source for generating a vertically elongated band-shaped ion beam having a vertical width larger than the diameter, a vertically elongated yoke having a passage for the vertically elongated ion beam, and a main coil M wound around the center of the yoke
1, M2, first auxiliary coils A1, A2 wound on one end of the yoke side by side with the main coil, and second auxiliary coils B1, wound on the other end of the yoke side by side with the main coil, B
2, a position detector for examining the distribution of the ion beam, and only the ion species having a mass in a desired range among the ion species separated by the mass analysis magnet. And a sample holding mechanism that holds the sample so that the ion beam passing through the slit is injected and moves the sample in a direction perpendicular to the longitudinal direction of the ion beam. An ion implantation apparatus characterized by the above-mentioned.