JP2002261029A - Growth method and apparatus of epiwafer - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a method for examining characteristics in an intermediate layer being subjected to epitaxial growth. SOLUTION: In a suscepter where one wafer or pluralities of wafers of a thin-film growth apparatus can be placed, a shutter that can cover partial wafers or one portion of a wafer is provided, multilayer thin-film is grown simultaneously, the shutter is closed to the partial wafers or one portion of a wafer as required for prohibiting the thin-film growth, and the thin-film growth is allowed to continue to remaining product wafers. The characteristics in a test wafer are measured to determine the characteristics in the product wafer.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a HEMT (High Ele
ctron Mobility Transistor or HBT (Heterojunction Bipolar Transistor)
Epiwafer (Epitaxial Wafe) with a structure such as istor
r) The method and apparatus for growing. In particular, the present invention provides a growth method and apparatus in which a number of thin films are stacked on top of each other and the characteristics of the thin film hidden inside can be measured. Si-transistors are also used as output transistors for mobile phones, but GaAs FETs (Fi
In many cases, an HEMT (a kind of FET) having a two-dimensional electron gas (2DEG) is often used. GaAs
Since s has a high electron mobility, an n-channel FET is advantageous.

[0002] The distance between a mobile phone and a station that receives a signal generated by the mobile phone is constantly changing. In a current mobile phone, the output of a transmission signal on the mobile phone side is constant, and the signal is received by a station and the amplification factor is changed to obtain a signal of a desired strength. In that case, an FET may be used. However, mobile phone standards are expected to be unified, such as changing the transmission signal strength on the mobile side by knowing the distance to the station. Then, it is necessary to increase or decrease the transmission signal power. In that case, GaAs FETs and HEMTs have insufficient linearity.

Attention has been paid to GaAs bipolar transistors. GaAs bipolar transistors have not yet been proven and are under development. Unlike a Si bipolar transistor, a base and an emitter cannot be formed by diffusing impurities into a collector layer. Ga
In the case of As, the collector, base, emitter and the like are all formed by epitaxial growth. G in the 1960s
Bipolar transistors with an AlGaAs layer on an aAs substrate have been energetically studied. However, AlGaAs is easily oxidized and cannot be put to practical use, and research on GaAs-bipolar transistors has not been successful. But the study continued without abandonment.

[0004] It revived in the 1990s. A
An lGaAs-based AlGaAs bipolar transistor has finally been put to practical use as a transistor for a mobile phone or an optical communication. Further, instead of the AlGaAs system, InGa
P-type GaAs transistors are being vigorously developed. Since it does not contain Al and is hardly oxidized, the characteristics should be further improved. In the future, large market expansion for CDMA is expected. This is called a heterojunction bipolar transistor because the materials on both sides of the pn junction are not the same,
Abbreviated as HBT.

Here, HEMTs and HBTs of GaAs are taken as output transistors for mobile phones, but the present invention can also be used for manufacturing and testing epitaxially grown films of other materials and other structures.

In the case of HBT, the collector, base, and emitter are stacked by epitaxial growth. Therefore, even if it is desired to examine the characteristics of the base layer, the intermediate base layer cannot be exposed. Therefore, the characteristics of the base layer cannot be inspected. H
In the case of EMT, the characteristics cannot be examined because the channel layer is covered by the contact layer thereon. It is an object of the present invention to provide a method by which the properties of an intermediate layer during epitaxial growth can be investigated.

[0007]

2. Description of the Related Art First, some terms will be defined. The substrate (wafer) is S, and the epitaxially grown layers are R1, R2,..., Rm in order from the bottom. m is the number of epitaxial layers. The m-th layer is the outermost layer. Assume that the layer to be inspected is the k-th layer (k <m). On the wafer S,
The epitaxial wafer on which the thin film layers R1, R2,..., Rm are epitaxially grown is represented by (S; R1; R2;... Rm). The k-th thin film from the bottom is called a k-layer.

It is desired to examine the characteristics of a layer that is buried in the middle in a series of epitaxial growth. In such a case, a conventional method is to epitaxially grow the same layer under the same conditions, stop the epitaxial growth, take out the substrate to the outside, and inspect the target layer exposed on the uppermost layer. That is, it is grown to the k-th layer, taken out of the apparatus, and the characteristics are examined by another inspection apparatus.

Although this may be a natural method, the target layer (k
); On the epitaxial growth layer (Rk + 1;... R
Since m) is not included, the state is different from the state of the k layer in the epitaxial wafer that exactly exists up to the m layer. In addition, since a wafer on which only 1 to k layers have been epitaxially grown is practically useless, the epitaxial growth process cannot be used for device fabrication at all. This means that it was epitaxial growth for inspection. Then materials,
Electricity costs are also wasted. Other conventional techniques will be described.

[0010] Yukio Sasaki, Shoichi Nagao, Ken Meguro, Yohei Otogi "Production technology of MOVPE wafer for high frequency devices", Hitachi Cable No. 16 (1997-1) p75-8
0 This is because the MOVPE method replaces the MBE method with GaAs.
It is a report that an s-FET was manufactured. Conventionally, GaAs-FETs and HEMTs for mobile phones have been manufactured by the MBE method, but it is said that a comparable device can also be manufactured by the MOVPE method. An important parameter that characterizes an FET is a pinch-off voltage (Vp). This must be a predetermined value, but it is H
It is a value that can be measured only after manufacturing EMT or FET. It would be very convenient if Vp could be known at the epi-wafer stage.

In this paper, an n-GaAs channel layer is deposited halfway (up to the height of the gate electrode) on an SI-GaAs substrate and a buffer layer, and the carrier concentration is measured by the CV method with the electrodes attached. When the concentration is n = 1 × 10 ¹⁵ c
The applied voltage Vs when the voltage becomes m ⁻³ is a pinch-off voltage Vp.
(Vs = Vp). Using such an empirical rule, the final HEMT can be measured by measuring an in-progress epi-wafer up to the channel layer.
Can be predicted. This is to predict the characteristics of the final product by taking out the wafer from the apparatus during the epitaxial growth, exposing the channel layer, and performing CV measurement. It may be the opposite of the object of the present invention.

Japanese Patent Application 2000-090881 “InG
aP / GaAs-HBT Manufacturing Method and Evaluation Method "is a prior application of the present applicant. This is because, after epitaxially growing all the layer structures on a GaAs substrate, the wafer is taken out, the uppermost emitter layer is removed, and the p-
The GaAs base layer is exposed, and the minority carrier lifetime τ, thickness, and carrier concentration p of the base layer are measured.
The target base layer is usually 70 nm, but in order to enable measurement, a thick base layer having a thickness of 300 nm to 1000 nm is used. After the emitter layer is once grown, it is taken out to remove the emitter layer and expose the base layer thereunder. Since the growth conditions are the same, the characteristics of the base layer should be the same as those of the base layer in the epitaxial layers including the first to m-th layers. This is a more sophisticated method, but again the epitaxial growth process cannot be used for device fabrication. The materials and power costs used for it are wasted.

[0013]

With the substrate (wafer) as S, the epitaxially grown layers are R1, R2,
.., Rk,. One method would be to make a test wafer of up to k layers, such as (S; R1; R2; ...; Rk). According to the above-mentioned prior art, an epiwafer (S; R1; R2;...; Rk ′;...; Rm) having a structure different from that of an epi product is once prepared, and upper layers (Rm,. To expose (S; R1; R2; ...; R
k ') states that the characteristics of Rk' appearing on the top as a wafer are tested.

However, these methods require extra operation of the growth apparatus to grow a test wafer having a different structure, and consume extra raw materials. In addition, since it grows separately from a product, it cannot necessarily be said that it has the same characteristics as a product. This is because the condition is not the same because the wafer is grown on another occasion than the wafer to be a product.

[0015]

A plurality of (n) thin film growth apparatuses are provided.
A susceptor that can hold a number of wafers is provided with a shutter that can cover some of the wafers. At the same time, a thin film is grown on a plurality of wafers. Growth is prohibited and thin film growth is continued for the remaining product wafers. The properties of the thin film layer exposed on the top surface of the test wafer can be measured.

Alternatively, a shutter capable of covering a partial area of one wafer is provided on a susceptor of the thin film growth apparatus on which one or a plurality of wafers can be mounted, and a thin film is grown on the wafer on the susceptor, and a test wafer is formed. On the other hand, the shutter is closed as needed to inhibit the thin film growth in the covered area, and the thin film growth is continued on the other exposed areas of the wafer and the test wafer. The characteristics of the thin film layer exposed on the uppermost surface of the covered region of the test wafer can be measured.

[0017]

DESCRIPTION OF THE PREFERRED EMBODIMENTS [1. When a Test Wafer and a Product Wafer Exist] A specific whole wafer is used for the test, and the other wafers are products. For convenience of explanation, a distinction is made between the concept of test wafer and product wafer. The test wafer is represented by Wj. The product wafer is represented by Vi.

S test wafers W1, W2, W3,.
Assuming that there are Ws and q product wafers V1, V2,..., Vq, a thin film is grown on all the wafers of n = s + q. An openable and closable shutter capable of covering the entire test wafer is provided. The s shutters open and close simultaneously.

Since there are n wafers on the same susceptor of the same apparatus, thin films are grown simultaneously under the same conditions. At a certain time, the test wafers W1, W2, W3,..., Ws are covered with a shutter to prevent thin film growth. At a certain time, the shutter is opened, and the subsequent thin film growth is restarted.

There are several aspects to the layer structure of the test wafer. (A) The test wafer Wj is grown from one layer to k layers, and stops growing at k layers. The product wafer Vi grows up to all layers 1 to m. Wj = (S; R1; R2; ...; Rk) Vi = (S; R1; R2; ...; Rk; Rk + 1; ...; R
m)

(B) The test wafer Wj is grown from the (k + 1) th layer to the mth layer without growing up to the kth layer. The product wafer Vi grows up to all layers 1 to m. Wj = (S; Rk + 1; Rk + 2; ...; Rm) Vi = (S; R1; R2; ...; Rk; Rk + 1; ...; R
m)

(C) The test wafer Wj is grown from the first layer to the k-th layer, the growth is stopped here, and the growth is resumed with the h-th layer (k <h). The product wafer Vi has all the layers 1 to
Growing up to m layers. Rk; Rh;... Rm) Vi = (S; R1; R2;... Rk; Rk + 1;.
h-1; Rh; ...; Rm)

(D) The test wafer Wj is grown from k layers to h layers (k <h). The product wafer Vi grows up to all layers 1 to m. Wj = (S; Rk; Rk + 1; ...; Rh-1; Rh) Vi = (S; R1; R2; ...; Rk; Rk + 1; ...; R
h-1; Rh;...; Rm) There are also a number of other combinations, with a total of 2 ^m -2. The present invention can realize all of those combinations.

[2.1 Case Where Test Area and Product Area Coexist in One Wafer] A part of one wafer is used for a test, and the other area is a product. . For convenience of explanation, a distinction is made between the concept of test wafer and product wafer. The test wafer is represented by Wj.
The product wafer is represented by Vi.

S test wafers W1, W2, W3,.
Assuming that there are Ws and q product wafers V1, V2,..., Vq, a thin film is grown on all the wafers of n = s + q.

There are provided s shutters capable of covering a part of the s test wafers. The s shutters open and close simultaneously. A portion of the surface area of the test wafer that is not covered by the shutter is called a product region Wv, and a portion that may be covered by the shutter is called a test region Ww. Since there are n wafers on the same susceptor of the same apparatus, thin films are grown simultaneously under the same conditions. The test area Ww of the test wafers W1, W2, W3,.
j is covered with a shutter to prevent thin film growth. At a certain time, the shutter is opened and the subsequent thin film growth is restarted. This also has several aspects.

(A) Test area Wwj of test wafer Wj
Is grown from one layer to k layers, and the growth is stopped at k layers. Product area Wvj of test wafer Wj and product wafer Vi
Grows from all layers 1 to m. Wwj = (S; R1; R2; ...; Rk) Vi, Wvj = (S; R1; R2; ...; Rk; Rk +
1; ...; Rm)

(B) Test area Wwj of test wafer Wj
Is grown from the k + 1 layer to the m layer without growing to the k layer. The product region Wvj and the product wafer Vi of the test wafer Wj grow to all layers 1 to m. Wwj = (S; Rk + 1; Rk + 2; ...; Rm) Vi, Wvj = (S; R1; R2; ...; Rk; Rk +
1; ...; Rm)

(C) Test area Wwj of test wafer Wj
Is grown from the first layer to the k-th layer, the growth is stopped here, and the growth is resumed in the h-th layer (k <h). Test wafer Wj
The product region Wvj and the product wafer Vi grow to all the layers 1 to m. Rw; Rm) Vi, Wvj = (S; R1; R2; ...; Rk; Rk +
1; ...; Rh-1; Rh; ...; Rm)

(D) Test area Wwj of test wafer Wj
Is grown from the k-th layer to the h-th layer (k <h). The product region Wvj and the product wafer Vi of the test wafer Wj grow to all layers 1 to m. Wwj = (S; Rk; Rk + 1; ...; Rh-1; Rh) Vi, Wvj = (S; R1; R2; ...; Rk; Rk +
1; ...; Rh-1; Rh; ...; Rm)

There are many other combinations, and there are cases where the total is 2 ^m -2. The present invention can realize all of those combinations.

[3. Shutter Structure] A shutter structure for selectively shielding only some wafers or only a part of a certain wafer will be described. Vapor phase growth equipment (MB
E, OMVPE), it is difficult to make a shutter with a complicated structure. In the case of MBE, the susceptor faces downward. The susceptor rotates with the wafer facing down. The susceptor can be kept stationary. OMVP
In the case of E, the susceptor is facing upward, and the wafer is rotated while holding the wafer upward. The shutter differs depending on the structure of the device. There are several candidates for shutters.

(1) Shutter Cantilevered by a Rotating Shaft Standing on the Susceptor Surface A disk-shaped shutter cantilevered by a rotating shaft is provided on the susceptor. Since the susceptor turns clockwise and counterclockwise, the shutter opens and closes according to the direction of left and right rotation of the susceptor. This is simple because it is opened and closed by rotating the susceptor left and right. The shutter can be opened and closed by improving only the susceptor structure without changing the entire device. However, there is a point that attention must be paid to operating conditions.

(2) Shutter Operated by Externally Operable Linear Introducer A shutter is attached to the tip of the linear introducer that is externally operated, and only a specific wafer is shielded and opened by linear reciprocating motion. In that case, the susceptor needs to be stationary.

(3) The inside of the rotation axis of the susceptor is doubled,
With a triple cylindrical structure, a shutter is attached to the upper end of the central axis. By rotating the central axis, a shutter opening / closing operation is performed on a specific wafer.

[(1) Mechanism for Opening / Closing Shutter by Rotation of Susceptor] FIG. 6 illustrates that the rotation axis K is set on the susceptor O, the arm is extended from the rotation axis K, and the shutter J is opened / closed. FIG. The center of the susceptor is O, the center of the rotation axis is K, and the center of the shutter is J. Let p be the distance between the susceptor center O and the rotation axis K. The distance between the center K of the rotation axis and the center J of the shutter is defined as q. p and q are constants (p> q). The shutter arm becomes KJ, which swings about the rotation center K. The angle between the radius OK and the shutter arm KJ is defined as a shutter swing angle θ. This is a variable. The range of the swing angle θ is mechanically limited. It is assumed that there is no resistance to the rotation, and that it can swing freely within the range. Then, the shutter moves due to the rotation of the susceptor.

Let J be the center of swing of the shutter, and let the distance from the center (center of rotation) O of the susceptor be OJ = r.
r is the radius from the rotation center O of the shutter, which is a variable. Let the rotational angular velocity of the susceptor be Ω. If the susceptor is constant rotation (Omega constant) to the force generated on the shutter only centrifugal force, which is equal to MrΩ ^2. M
Is the mass of the shutter. Centrifugal force is always a radially outward force. No force other than centrifugal force is generated. Of course, there is a reaction force that the shutter rotation axis exerts on the shutter, but this is due to the centrifugal force. The only force exerted on the shutter by the rotation of the susceptor in the steady state is the centrifugal force. Centrifugal force is the force that pushes an object in the radial direction. This means that it is a force that works to maximize the absolute value of θ. In the steady state, the absolute value of θ takes the maximum value. If the range of θ is only positive or only negative, the shutter can take only one state.

Centrifugal force stably holds two states of shutter
You do it. So the centrifugal force MrΩ ²Is two stable states
(Open and close) must push the arm in different directions
I have to. That means the rotating arm of the shutter
As shown in FIG. 7, the domain has a negative constant angle (−
Φ₂) To plus fixed angle (+ Φ₁)
That is. The shutter only rotates in the direction of rotation of the susceptor
To open and close,

(A) The susceptor turns left (counterclockwise)
When over the (Omega positive when) the shutter rotation angle theta to take a negative boundary value theta = - [Phi] _2.

[0040] (b) the susceptor when over to the right around (clockwise) (Ω is when negative) the shutter rotation angle θ to take a positive boundary value θ = + Φ _1. It must be like that.

This is possible because the shaft is on the outer periphery of the susceptor, the arm is inward, and the shutter J is inside the shaft K. However, it is a necessary condition, not a sufficient condition.

The problem is, the shutter from the - [Phi] ₂ in a short transient time direction of rotation of the susceptor is switched + [Phi
To _1, or it is whether switched to -Φ ₂ from + Φ _1. If the susceptor flips too slowly,
The desired shutter movement will not occur. The swing angle of the shutter is θ, and the angle formed by a semi-line OJ connecting the center J of the shutter and the center O of the susceptor with the shaft radius OK is ψ.

Since ∠JKO = θ and ∠KOJ = ψ,
{KJO = π−θ−}. From the sine theorem

[0044]

(Equation 1)

Is as follows. The equation of motion of the shutter is derived. Suppose the susceptor is not stationary but in transition. The angular velocity Ω of the susceptor is not constant, and its time derivative α = dΩ / dt
Exists. The acceleration at point K with the axis is pα in the circumferential direction. Since the axis K and the shutter center J are connected by an arm, an acceleration -pα is generated at the shutter center J in the opposite direction. As a result, a force of −Mpα is applied.

Another force (centrifugal force) is generated at the shutter center J in the radial direction. This is a MrΩ ^2, a force directed radially outward. Shutter center J
Can move only the arc FHE. The cosine component of the transitional force -Mpα for the arc is cos θ. The cosine component of the centrifugal force arc is cos (π / 2 − （−
θ). The equation of motion of the shutter along the arc FHE is

[0047]

(Equation 2)

Is as follows. The first term on the right side is simplified by (1) as follows.

Rcos (π / 2−ψ−θ) = rΩsin
(Ψ + θ) = psinθ (3)

[0050]

(Equation 3)

This is a transitional equation. When the steady state, dΩ / dt = 0, a Ω = Ω _0, θ is +
When Φ ₁ and −Φ ₂ , a moment −MpΩ ² sin θ that cancels the first term on the right side is given from the shutter axis, so that the right side becomes 0, and θ takes a constant value (+ Φ ₁ , −Φ ₂ ). . Ω is the angular velocity of the susceptor.

What happens when the susceptor flips? Although this is not known, the movement is uniquely determined by the motor driving the susceptor, the speed reducer, the inertia of the susceptor, and the like. Once the transient motion of Ω is determined, the equation of motion should be able to be solved. Since it is not linear, it cannot be solved analytically.
An exact solution can also be obtained by a numerical analysis method. However, here I will try to find an approximate solution and go further.

For example, from negative Ω (clockwise) to positive Ω
Since it captures the moment of reversal that changes in (counterclockwise), sinθ is in the sinΦ _1. So sin θ is s
(Constant), and approximates cos θ = 1. Assuming that Ω rises in direct proportion to time (acceleration is constant), dΩ / dt = b, where the rotational angular acceleration is b (constant).

[0054]

(Equation 4)

This can be integrated. The initial condition is t = 0,
dθ / dt = 0, and the θ = θ _0,

[0056]

(Equation 5)

When Ω turns from negative to positive, θ ₀ is positive, s is positive, and b is positive. When obtaining the t ₂ which takes a minimum value by differentiating the θ at t,

T ₂ = (3 / bs) ^1/2 (7)

The minimum value θ _mim of θ with respect to _this is:

[0060]

(Equation 6)

Is obtained. If this is true, the shutter will return to its original position and no state change will occur. So this must be negative.

[0062]

(Equation 7)

Since s = sin θ ₀ , in the first approximation,

Sin θ ₀ <(3p / 4q) ^1/2 (10)

Is obtained. If the angle exceeds this, the shutter does not move even if the susceptor is inverted. Therefore, two states θ ₀ = + Φ ₁ , −Φ ₂
Together

SinΦ ₁ <(3p / 4q) ^1/2 (11)

SinΦ ₂ <(3p / 4q) ^1/2 (12)

Must be If the initial angle is within this range, the shutter arm may be inverted when the susceptor is inverted. Whether or not the inversion is actually performed cannot be determined only from the equations (10) to (12). There are further requirements. When θ becomes 0, the angular velocity must be opposite to θ ₀ .

[0069] When the second term of (6) the right-hand side is ignored because the contribution is small, the time t ₁ that θ becomes zero (the arm KJ comes on radius OK is)

T ₁ = (2qθ ₀ / pb) ^1/2 (13)

Is obtained. At this time, the angular velocity Ω of the susceptor is

Ω = bt ₁ = (2qbθ ₀ / p) ^1/2 (14)

However, this must be lower than the steady state angular velocity Ω ₀ . Otherwise, the susceptor will rotate normally before θ = 0, and forces other than centrifugal force will be lost and the rotational force of the arm will be lost.

(2qbθ ₀ / p) ^1/2 <Ω ₀ (15)

[0075]

(Equation 8)

This is an equation for determining the upper limit of the angular acceleration b at the time of reversal. If the steady rotation of the susceptor is fast (Ω ₀ is large) or the initial angle θ ₀ is small, the range that b can take is wide. However, when the steady rotation of the susceptor is slow (Ω ₀ is small) or the initial angle θ ₀ is large, the range that b can take is narrow. This means that the susceptor must be slowly inverted. If such conditions are satisfied,

(A) When the susceptor is rotated clockwise (clockwise) (Ω = −Ω ₀ ), the shutter position θ = + Φ
₁

(A) When the susceptor is rotated to the left (counterclockwise) (Ω = + Ω ₀ ), the shutter position θ = −
It becomes Φ _2.

[0079]

EXAMPLES Example 1: HEMT Structure Epitaxial Wafer (FIGS. 1 and 4) (Three HEMT product wafers (2DEG structure) and one test wafer (2DEG structure) are epitaxially grown simultaneously by OMVPE). HEMT FIG. 4 shows the structure of the product wafer V.
(A). FIG. 4 shows the structure of the HEMT test wafer W.
(B). These are in order from the top,

Product Epi Wafer Test Epi Wafer n-GaAs: 100 nm non-AlGaAs: 50 nm non-AlGaAs: 50 nm n-AlGaAs: 10 nm n-AlGaAs: 10 nm non-AlGaAs: 3 nm non-AlGaAs: 3 nm non-InGaAs: 10 nm Non-InGaAs : 10 nm non-AlGaAs: 3 nm non-AlGaAs: 3 nm n-AlGaAs: 10 nm n-AlGaAs: 10 nm non-AlGaAs: 150 nm non-AlGaAs: 150 nm GaAs buffer layer: 500 nm GaAs buffer layer: 500 nm semi-insulating GaAs substrate semi-insulating GaAs substrate GaAs substrate

It has a layer structure consisting of: The product wafer Vi (i = 1, 2, 3) is obtained by growing nine thin films on a GaAs substrate. The test wafer (W1) is an eight-layer thin film grown on a GaAs substrate. The test wafer excludes the uppermost layer of n-GaAs 100 nm of the product wafer.

HEMT (High Electron Mobility Trans)
istor is a two-dimensional electron gas (Two Dimensional E)
This is an FET utilizing an electron gas (2 DEG). n
"on-" means non-doped. non
-InGaAs is a channel layer, and the channel width changes according to the gate voltage at which electrons move therethrough, and the drain current changes like the FET. Normal FET
The difference is that the channel layer itself does not have impurities and provides carriers, but the channel layer has no impurities and obtains carriers from other adjacent layers.

N-Al on InGaAs channel layer
GaAs (10 nm) and underlying n-AlGaAs (10 n
m) to receive carriers (electrons). This carrier is I
It flows through the nGaAs layer. Since InGaAs is non-doped, it has high crystallinity and low scattering, and thus has high electron mobility. A
Electrons accumulate in InGaAs because InGaAs has a narrower band gap than 1GaAs. It can move at a high speed with little scattering in InGaAs and can take advantage of the high speed of the electron mobility. That's why High Electron Mobility. Conventionally, n-AlGaAs for donating electrons is only on the channel layer, but in FIG. 4, the electron-donating n-AlGaAs is above and below the channel layer, and is called a double-doped HEMT. Since there are many electrons, a large current can flow. Suitable as a high-speed power transistor.

Non-AlG above and below n-AlGaAs
aAs is a layer interposed to enhance crystallinity. Since the crystallinity of the layer containing the dopant is inevitably deteriorated, a non-doped layer is formed to restore the crystallinity.

The uppermost n-GaAs contact layer has high conductivity, and the drain electrode and the source electrode are in ohmic contact. The thickness is as thick as 100 nm for electrode bonding. Source by alloying process at the time of ohmic joining,
The drain electrode and the non-InGaAs layer are connected with low resistance. Etching is removed halfway through the non-AlGaAs, and a gate electrode is attached thereto.

When the uppermost n-GaAs layer of the product wafer is 10
Since it is as thick as 0 nm, the n-AlGa
The carrier concentration, mobility, and the like of an active layer composed of As, non-InGaAs, n-AlGaAs, or the like cannot be measured.

Therefore, the uppermost n-GaAs 100n
Those excluding m are used as test wafers. Since there is not this, an electrode is mounted on non-AlGaAs to form an active layer (non-InGaAs, n-AlGaAs).
The carrier concentration and mobility in s) can be measured. Since it is an object of the present invention to measure electronic parameters in the active layer, it is particularly advantageous to be able to make such a test wafer.

A product wafer having such a layer structure (Vi
(I = 1, 2, 3) Three test wafers and one test wafer (W1) were placed on the susceptor of FIG. 1 and grown simultaneously inside the vapor phase growth apparatus.

As shown in FIG. 1, the susceptor has four depressions. The susceptor can rotate around the center (O). A shutter is mounted around the susceptor. A shutter shaft is set up around the susceptor, the arm is extended from the shaft, and a disk-shaped shutter is attached to the tip of the arm. The shutter and the arm are integrated, and can freely swing around the axis. The range of the swing angle θ is mechanically determined, and as shown in FIG. 7, the swing angle θ can swing up to + Φ _{1 in} the positive direction and up to −Φ _{2 in the} negative direction. The swing angle θ takes one of + Φ ₁ and −Φ ₂ depending on the rotation direction of the susceptor, and is stabilized at that position. Either is “closed” and “open”. The state where the wafer is exposed is “open”, and the state where the wafer is hidden is “closed”.

At the initial stage of growth, the tray is rotating in the direction B (clockwise; clockwise) in FIG. 1, and the shutter maintains the "open" state. Each layer is sequentially grown from the substrate side, and non-AlGaAs immediately below the contact layer: 50 n
When the growth of the m-layer is completed, the rotation of the tray is changed to B in FIG.
"Close" the shutter by reversing from
Switch to state.

Then, the remaining contact layer (n-GaAs)
s; 100 nm) to complete the growth. OMVP
The test structure epi taken out from the E-growth furnace, Ns (sheet carrier concentration) and μ (mobility) were further measured by HALL measurement and Vp was measured by CV (voltage-capacity) characteristics measurement.
Measurements required for confirming the state of the active layer such as (pinch-off voltage) can be freely performed. In addition, since the product structure epi and the test structure epi grow simultaneously, the characteristics measured in the test structure completely match the characteristics of the epi in the product structure.

Example 2: HBT Structure Epitaxial Wafer (FIGS. 2, 5 (a), 5 (c))] (Epitaxial growth of three HBT product wafers and one test wafer simultaneously by OMVPE method) In configuration 2, the susceptor has four wafer holes and three equivalent shutters. A shutter shaft for rotatably holding the arm is set up on the susceptor, and the shutter is fixed at the end of the arm. The HBT product wafer 3 shown in FIG.
The wafer and one HBT test wafer shown in FIG. 5C were simultaneously grown. In the second embodiment, unlike the previous example, the test wafer is not covered by the shutter, and the product wafer is covered by the shutter on the way.

HBT product wafer (a) HBT test wafer (c) n-InGaAs: 50 nm n-InGaAs: 50 nm n-GaAs: 100 nm n-GaAs: 100 nm n-InGaP: 50 nm n-InGaP: 50 nm p-GaAs: 70 nm p-GaAs: 500 nm n-GaAs: 500 nm non-AlGaAs: 100 nm n-GaAs: 200 nm n-GaAs: 500 nm Semi-insulating GaAs substrate n-GaAs: 200 nm Semi-insulating GaAs substrate

The product wafer (a) and the test wafer (c) are different in two points. One is the difference in the thickness of the p-GaAs base layer (test wafer 500 nm, product wafer 70 nm)
That's what it means. The other is that in the case of a test wafer, non-AlGaAs: 100 nm is added below the base layer.

The base layer made of carbon-doped GaAs is
The product wafer (a) is as thin as 70 nm. The thickness of the base (p-GaAs) strongly affects the response speed of the device. Since the base should be thinner, the product wafer is as thin as 70 nm. However, the thickness of the test wafer is as thick as 500 nm. This is for measuring the hole density p and the electron lifetime τ of the base layer. The base layer is p-type by carbon doping. However, when hydrogen enters, the crystallinity deteriorates, the hole density p decreases, and the electron lifetime τ also shortens. It degrades the response speed and linearity of the transistor. Therefore, it is necessary to evaluate the quality by actually measuring p and τ.

However, if the thickness is as thin as 70 nm like a product wafer, it may be difficult to measure the hole concentration p by hole measurement. The Hall measurement is difficult to measure for a thin layer because the carrier concentration is known by applying a magnetic field, applying a current, and determining the electromotive force generated thereby.

However, what is more problematic for thin base layers is that it is not possible to measure the lifetime τ (lifetime) of electrons (minority carriers). The lifetime of electrons is measured by photoluminescence. A laser beam which emits light having an energy higher than the band gap of GaAs of 1.4 eV is applied. Holes and electrons are paired by the laser light, but the electrons eventually emit luminescence (fluorescence) and recombine with the holes to disappear. The decay of the fluorescence corresponds to the lifetime. The speed of change in the fluorescence intensity indicates the electron lifetime. Since the base layer of a product wafer is as thin as 70 nm, applying a laser will reach the layer below and also pick up photoluminescence from the layer below. Therefore, the electron lifetime of the base layer cannot be measured. 50 to avoid it
Grow test wafers with base layers as thick as 0 nm. At this thickness, there is no photoluminescence from the underlying layers. The reason why the base layer of the test wafer is thick is described above.

In the test wafer (c), 100 nm of non-AlGaAs under the base layer is an etching stop layer. When the wafer is etched with an etching solution having an etching rate of GaAs / AlGaAs close to 0, only the upper layer GaAs is removed. When a part of the base layer is covered with the resist and etched, a portion of the base layer not covered with the resist exposes the underlying AlGaAs. In the portion covered with the resist, p-GaAs remains and a step is generated. The thickness of the base layer is measured by observing the step with a microscope. AlGaAs for accurate measurement of base layer thickness
Put in.

Such a situation is described in Japanese Patent Application No.
00-090881. The prior art test wafer (b) shown here is

HBT test wafer (b) n-InGaAs: 50 nm n-GaAs: 100 nm n-InGaP: 50 nm p-GaAs: 500 nm non-AlGaAs: 100 nm non-GaAs: 100 nm Semi-insulating GaAs substrate

Has the following structure. The test wafer (c) of the present invention has n-GaAs 500 nm and n under the etching stopper layer non-AlGaAs, like the product wafer.
The test wafer having 200 nm of GaAs is no
It has n-GaAs 100 nm. In that respect, the test wafer (c) of the present invention is different from the test wafer (b) of the prior art.

The prior art (Japanese Patent Application No. 2000-090) cited by the HBT test structure of FIG.
881) is different from the HBT test structure (FIG. 5B).

Prior art (Japanese Patent Application No. 2000-09088)
The test wafer of 1) is too thin (7
0 nm) and a time-resolved PL (photoluminescence) measurement cannot be performed, and the p-GaAs base layer is thick (500 nm).
In order to measure the thickness of the p-GaAs base layer by step measurement, a non-AlGaAs: 1
A 00 nm layer is added. In the prior art, since the test structure epi is grown separately from the HBT product structure epi, the layers (n-GaAs 500 nm and n
(GaAs 200 nm) is omitted to save growth time and raw materials.

However, in the present invention, since the HBT product structure (a) epi and the HBT test structure (c) epi grow simultaneously, the measurement required in the structure (c) in which the features of the prior art test wafer are added to the HBT product structure. Can all be performed.

Further, since the crystal is grown at the same time as the product structure epi, the obtained measurement result agrees with the characteristics of the product structure epi except for items where the thickness of the epi layer is changed.

Now, the process of epitaxial growth will be described. At the initial stage of growth, the susceptor (tray)
The shutters are rotating in the directions (clockwise and clockwise), and all three shutters are maintained in the “open” state.

Each layer is sequentially grown from the semi-insulating GaAs substrate side. n-GaAs: 200 nm, n-GaAs
A 500 nm thin film is grown. When the growth of the second layer (n-GaAs: 500 nm) from the substrate side is completed, the tray is rotated in the direction B (clockwise, clockwise) in FIG.
In the direction A (counterclockwise, counterclockwise). The inertia causes the three shutters to rotate counterclockwise to cover the product wafer. That is, the three shutters were switched to the “closed” state.

One test wafer of the test structure is not covered by a shutter since there is no shutter. The thin film can be grown. Therefore, a non-AlGaAs 100 nm layer serving as an etching stop layer is grown on the test wafer. Further, the growth of a 430 nm base layer corresponding to the thickness difference between the base layer 70 nm of the HBT product structure and the thick base layer 500 nm of the HBT test structure is promoted.

When the thickness of the p-GaAs base layer of the test wafer reaches 430 nm, the rotation of the tray is again reversed in the B direction (clockwise, clockwise). The shutter opens due to inertia. Three product wafers are again exposed. A 70 nm base layer is newly grown on the product wafer, and a 70 nm base layer is additionally grown on the test wafer. After that, the shutter is kept open, and the remaining three emitter layers (n-InGaP: 5) are formed on both the test wafer and the product wafer.
0 nm, n-GaAs: 100 nm, n-InGaAs
s: 50 nm) to complete the thin film formation.

FIG. 5B shows the test structure epi of FIG. 5C taken out of the OMVPE growth furnace according to the present invention.
Patent Application (Japanese Patent Application No. 2000-09088)
The following measurements were performed in the same manner as in 1).

First, n-InGaAs: 5 from the epi surface side
The 0 nm layer and the n-GaAs: 100 nm layer were removed by etching. At this time, it is preferable to use an etching solution and etching conditions in which the InGaAs and the GaAs layer are dissolved but the InGaP is not dissolved.

With the n-InGaP: 50 nm layer exposed on the surface, the crystallinity of the thick base (p-GaAs: 500 nm layer) was evaluated by the time-resolved PL (photoluminescence) method.

In this example, it was confirmed that the lifetime of the base layer was 300 psec (picosecond = 10 ⁻¹² sec) and the crystallinity was good.

Subsequently, n-InGaAs exposed on the surface
The s: 50 nm layer is removed, a HALL measurement is performed, the p-GaAs: 500 nm layer is further partially removed, and the remaining portion is measured for a step using a surface shape measuring device to obtain a thick base layer (p-GaAs). (GaAs: 500 nm layer) was measured.

When each layer is etched, n-InG
To remove the aP: 500 nm layer, use an etchant that dissolves InGaP but does not dissolve GaAs, and to remove the thick base layer (p-GaAs: 500 nm), an etchant that dissolves GaAs but does not dissolve AlGaAs. It is desirable to use etching conditions.

As a result of the measurement, thick p-GaAs: 500 n
It was confirmed that the thickness of the m-layer and the carrier concentration grew as designed.

The HBT product structure epi of FIG. 5A grown at the same time was shipped as a product to a device manufacturer (or advanced to a lower process in the company). Note that InGaP may be used as the etching stop layer.

Example 3 HEMT Structure Epitaxial Wafer (FIG. 3) (One HEMT epi was grown by OMVPE method, half of the epi was a HEMT product structure, and the other half was a test structure (2D
EG structure) epi)

According to the present invention, a part of the wafer is covered by the shutter, a part of the process of thin film growth is prohibited, and the growth of the remaining wafer is continued, so that the entire wafer is classified into a test wafer and a product wafer. Not only do
There are other possibilities.

That is, a part of one wafer is used for testing.
Some can be used for products. Example 3 is such. FIG. 3 shows the structure of the susceptor in that case. There is one wafer recess at the center of the susceptor, where one test / product wafer is loaded. The shutter is of a half-moon type and covers the right half of the wafer when closed. In the open state, the entire wafer is exposed. The right half of the wafer is the test wafer and the left half is the product wafer.

This is an epitaxial wafer having the same HEMT structure as that of the first embodiment. The epi structure is shown in FIG. 4, the test part is shown in FIG. 4 (b) and the product part is shown in FIG. 4 (a).
Shown in

As in the first embodiment shown in FIG. 1, at the initial stage of growth, the susceptor is rotated in the direction B (clockwise or clockwise) in FIG. 3 to open the shutter. In that state, the buffer layer: 500 nm, non-AlGaAs:
150 nm, n-AlGaAs: 10 nm, non-A
lGaAs: 3 nm, non-InGaAs: 10 n
m, non-AlGaAs: 3 nm, n-AlGaAs
A thin film of s: 10 nm and non-AlGaAs: 50 nm is grown.

Contact layer (n-GaAs: 100n)
m) Immediately after the growth of the non-AlGaAs: 50 nm layer immediately below, the rotation of the susceptor is switched from B in FIG. 3 to direction A (counterclockwise, counterclockwise) to close the shutter. In this state, a contact layer (n-GaAs: 100 nm layer) is grown only on the left half (in FIG. 3) of the wafer.

The same measurement as in Example 1 was performed on the test structure portion in the right half of the epiwafer. The same results as in Example 1 were obtained. The remaining left half of the product structure was used for product manufacture.

Instead, the same measurement as in Example 1 is performed on the test structure half of the epi-wafer, and the other half of the product structure is subjected to measurement including another destructive inspection. May be grown on another wafer placed at another location.

[0126]

According to the present invention, a product wafer and a test wafer are arranged in the same thin film growth apparatus to grow a thin film, the test wafer or the product wafer is covered with a shutter, and a part of the thin film growth step is omitted. By measuring the characteristics of test wafers by measuring the characteristics of test wafers, the characteristics of product wafers manufactured under the same conditions can be examined.
It is possible to examine a test wafer on which a thin film has been grown under the same conditions as a product, and it is also possible to measure characteristics that are difficult to measure with a product wafer itself.

The gist of the present invention is that by covering a part of the epi growing at once with a shutter,

(1) To remove a layer which becomes an obstacle to measurement or makes measurement difficult

(2) or reducing the thickness;

(3) or conversely, add a layer that helps measurement.

(4) Or by increasing the thickness,

A test epi-wafer (test wafer) whose characteristics are partially matched with the product is simultaneously grown, the characteristics of the test wafer are measured, and the characteristics of the product wafer can be closely examined. .

In the embodiment, GaAs is formed using the OMVPE method.
Although epi growth using a substrate is described as an example, the present invention can be applied to other fields as follows.

The present invention is also applicable to semiconductor lasers and light receiving elements using an InP substrate, and of course, to semiconductor lasers and light receiving elements using a GaAs substrate or a GaN substrate, and to epitaxial growth of transistors. In the embodiment, the HEMT is also DHE.
Although the MT (double-doped HEMT) structure is taken as an example, it is needless to say that the structure can be applied to a normal single-doped structure or another structure.

In addition to the OMVPE method, the present invention can be applied to a manufacturing method capable of interrupting epi growth by opening and closing a shutter, such as an MBE method and a VPE method. The mechanism for opening and closing the shutter is not limited to the method of FIG. 1, FIG. 2, and FIG.

In the OMVPE method, the tray accommodating the wafer is often rotated in order to make the characteristics uniform. However, if the manufacturing method does not need to be rotated, the growth furnace such as a rotation introducing machine, a linear introducing machine, and a magnetic coupling mechanism may be used. A shutter can be provided that can be mechanically opened and closed directly from outside the chamber.

As described in the third embodiment, the number of epiwafers to be grown may be plural or one.

The number of epiwafers grown as in Example 3 is 1.
In the case of a sheet, the distribution of the test region area and the product region area may be freely designed in consideration of the area required for inspection.

[Brief description of the drawings]

FIG. 1 shows four thin films simultaneously grown on a wafer.
FIG. 4 is a plan view of a susceptor provided with a shutter capable of partially suspending the growth of only one wafer.

FIG. 2 shows four thin films grown simultaneously on three wafers.
FIG. 4 is a plan view of a susceptor provided with a shutter capable of partially suspending the growth of only one wafer.

FIG. 3 is a plan view of a susceptor provided with a shutter in which a part of one wafer is a test area and a part is a product area, and a shutter is provided so that the thin film growth in the test area can be partially interrupted by a shutter.

FIG. 4 shows GaAs grown in Example 1 of the present invention.
The figure which shows the layer structure of -HEMT. (A) is a layer structure diagram of a HEMT product epiwafer. (B) is a layer structure diagram of a test wafer.

FIG. 5 shows GaAs grown in Example 2 of the present invention.
The figure which shows the layer structure of -HBT. (A) is GaAs-HBT
FIG. 3 is a layer structure diagram of a product wafer. (B) is a layer structure diagram of a GaAs-HBT test wafer shown by the applicant's prior art. (C) shows Ga used in Example 2 of the present invention.
FIG. 3 is a layer structure diagram of an As-HBT test wafer.

FIG. 6 is a view for explaining how the angle of a shutter J having a central axis at one point K on the susceptor changes as the susceptor rotates. O is the center of the susceptor, K is the center of the shutter axis, J is the center of the shutter, and FJE is the arc drawn by the center of the shutter. p is the distance between the shutter axis center and the susceptor center, and q is the effective length of the shutter arm. r is a distance between the susceptor center O and the shutter center J and is a variable. θ is the swing angle of the shutter arm.

FIG. 7 is a diagram showing an allowable range of a swing angle θ of a shutter arm. θ can move in the range of −Φ ₁ ≦ θ ≦ + Φ ₂ .
The range is limited by mechanical means.

[Explanation of symbols]

O Center of the susceptor K Center of the shutter axis J Center of the shutter FHE Arc on which the center J of the shutter moves p Distance between the center O of the susceptor and the center K of the shutter axis (constant) q Distance between the center K of the shutter axis and the center J of the shutter ( Constant) r Distance between shutter center J and susceptor center O (variable) θ Swing angle of shutter arm 線 The line connecting susceptor center O and shutter center J is
The lower limit of the upper - [Phi] ₂ shutter arm swinging angle θ of the radius OK with an angle + [Phi ₁ shutter arm swinging angle θ of the shutter shaft

Continued on the front page (51) Int.Cl. ⁷ Identification symbol FI Theme coat II (Reference) H01L 21/331 H01L 29/80 H 29/778 21/338 29/812 F term (Reference) 4K030 AA11 BA02 BA08 BA11 BA25 BA51 BB02 CA04 FA10 KA12 KA41 LA14 4M106 AA01 AA07 AB17 BA01 BA14 CA12 CB12 5F003 AZ09 BF06 BM03 BP32 BP96 5F045 AA04 AB10 AB17 CA02 CA07 DA52 DP15 DP27 EF18 HA11 5F102 FA09 GB01 GC01 GD01 GJ05 GK05 G01 GJ05 GK05

Claims (8)

[Claims] 1. A method for growing an epi-wafer, comprising attaching a plurality of or one substrate wafer to a susceptor and epitaxially growing a multilayer thin film on the substrate wafer simultaneously.
By covering part of the wafer or part of the wafer with an openable shutter, some layers in the multilayer thin film are not grown, or even if grown, the thickness of the grown layer is reduced, or by opening the shutter A method for growing an epi-wafer, wherein an extra layer is grown or a thickness of a part of the layer is increased. 2. An epi-wafer having a HEMT structure in which an electron supply layer, a channel layer, and a contact layer are grown on a substrate, wherein a plurality of epi-wafers are grown simultaneously.
2. The method for growing an epi-wafer according to claim 1, wherein the contact layer is not grown on a part of the wafer or a part of the wafer. 3. The characteristics of a test wafer in which a contact layer is omitted from a HEMT structure wafer are evaluated by CV (capacitance) measurement and HALL measurement to confirm the performance of a HEMT product wafer which is a product grown simultaneously with the test wafer. 3. The method of growing an epi-wafer according to claim 2, wherein: 4. An epi-wafer having a HBT structure in which a collector layer, a base layer, and an emitter layer are grown on a substrate, wherein a plurality of epi-wafers are grown at the same time, and a part of the wafer or a part of the wafer is formed. The method of claim 1, wherein the base layer is grown thicker than the product structure. 5. The method for growing an epi-wafer according to claim 4, wherein an additional etching stop layer is grown immediately below the base layer. 6. A method for measuring the minority carrier lifetime of a thickened base layer by PL (photoluminescence) method or time-resolved PL (photoluminescence) method, measuring the thickness of the base layer by selective etching, and measuring the carrier concentration by HALL measurement. 6. The method of growing an epi-wafer according to claim 5, wherein the performance of the HBT structure wafer as a product is evaluated. 7. An epi-wafer growing apparatus for mounting a plurality of or one substrate wafer on a susceptor and simultaneously epitaxially growing a multilayer thin film on the substrate wafer,
An epi-wafer growing apparatus comprising a shutter capable of covering a part of a wafer or a part of the wafer and capable of opening and closing during thin film growth. 8. The epiwafer according to claim 7, wherein the shutter is rotatably or removably attached to the susceptor, and the shutter is opened and closed by inertia due to rotation or translation of the susceptor. Growth equipment.