JP2008182070A - Method for forming silicon oxide layer - Google Patents

Abstract

To reduce the density of interface states formed at the interface between a silicon carbide semiconductor layer and a silicon oxide layer.
A silicon oxide layer 4 is formed by heating the surfaces of silicon carbide semiconductor layers 18, 6 and 8 in a mixed gas atmosphere containing a gas containing at least an oxygen element, an inert gas, and an ammonia gas. Silicon carbide semiconductor layers 18, 6, 8 are removed while removing carbon that is not bonded to silicon present in silicon carbide semiconductor layers 18, 6, 8 and repairing crystal defects in silicon carbide semiconductor layers 18, 6, 8. The silicon oxide layer 4 is formed by oxidizing the surface.
[Selection] Figure 1

Description

  The present invention relates to a method for oxidizing a surface of a silicon carbide semiconductor layer to form an insulating silicon oxide layer.

In a semiconductor device using silicon carbide (SiC), a gate electrode opposed to the silicon carbide semiconductor layer may be formed through an insulating silicon oxide (SiO ₂ ) layer. In the semiconductor device, the surface of the silicon carbide semiconductor layer is oxidized to form a silicon oxide layer. Since the silicon oxide layer is insulative, the gate electrode and the silicon carbide semiconductor layer are insulated from each other, and when a voltage equal to or higher than a predetermined value is applied to the gate electrode, A channel can be formed.
However, an interface state is easily formed between the silicon carbide semiconductor layer and the silicon oxide layer, and when the density of the interface state is large, resistance when carriers move through the channel is increased. That is, the mobility of carriers in the channel of the semiconductor device is reduced.
In Patent Document 1, after the surface of a silicon carbide semiconductor layer is oxidized to form a silicon oxide layer, a gas containing at least one of nitrogen oxide gas and ammonia gas is diluted with an inert gas having a molecular weight smaller than that of argon. A technique for heating to 1300 ° C. or higher in a gas atmosphere is disclosed. When the latter step is added, excess carbon (C) present at the interface between the silicon carbide semiconductor layer and the silicon oxide layer is removed, or nitrogen (N) is bonded to silicon generated by removing carbon from silicon carbide. be able to. In the technique of Patent Document 1, the density of interface states is reduced by removing impurities (carbon) present at the interface between the silicon carbide semiconductor layer and the silicon oxide layer or by bonding nitrogen to unpaired electrons of silicon. ing. That is, the on-resistance of the semiconductor device is reduced by increasing the carrier mobility in the channel of the semiconductor device.

JP 2006-210818 A

As disclosed in Patent Document 1, after the silicon oxide layer is formed, the silicon oxide layer is thermally treated in a gas atmosphere containing at least one of a nitrogen oxide gas and an ammonia gas. The density of interface states between the silicon oxide layers can be reduced. However, in the technique of Patent Document 1, since the interface between the two after the silicon oxide layer is formed on the surface of the silicon carbide semiconductor layer, the effect of reducing the interface state density is low. Therefore, the density of interface states formed between the two cannot be sufficiently reduced.
The present invention provides a technique for sufficiently reducing the interface state density between a silicon carbide semiconductor layer and a silicon oxide layer while forming a silicon oxide layer on the surface of the silicon carbide semiconductor layer.

In the present invention, the surface of the silicon carbide semiconductor layer is oxidized while removing the cause of increasing the interface state density between the silicon carbide semiconductor layer and the silicon oxide layer.
In the method of the present invention, the surface of the silicon carbide semiconductor layer is heated in a mixed gas atmosphere containing a gas containing at least an oxygen element, ammonia gas, and an inert gas.

  According to the above method, silicon and unbonded carbon (hereinafter sometimes referred to as unbonded carbon) in the silicon carbide semiconductor layer react with ammonia to form methane gas. Can be removed. Further, nitrogen in ammonia is bonded to unpaired electrons of silicon in silicon carbide, and unpaired electrons of silicon can be terminated with nitrogen. That is, the cause of increasing the density of the interface state between the silicon carbide semiconductor layer and the silicon oxide layer can be eliminated. Further, carbon constituting the silicon carbide semiconductor layer reacts with ammonia gas, and carbon can be desorbed from the silicon carbide semiconductor layer. Silicon generated by desorption of carbon from the silicon carbide semiconductor layer reacts with a gas containing an oxygen element, and a silicon oxide layer is formed on the surface of the silicon carbide semiconductor layer. According to the method of the present invention, the cause of increasing the density of the interface state between the silicon carbide semiconductor layer and the silicon oxide layer is removed, and the surface of the silicon carbide semiconductor layer is oxidized to form the silicon oxide layer at the same time. Therefore, the cause of increasing the density of the interface state between the silicon carbide semiconductor layer and the silicon oxide layer hardly remains. A silicon oxide layer can be formed while reducing the density of the interface state between the silicon carbide semiconductor layer and the silicon oxide layer. The on-resistance of the semiconductor device can be reduced.

  In the method of the present invention, a first step of heating the surface of the silicon carbide semiconductor layer in a mixed gas atmosphere containing at least an oxygen element, ammonia gas, and an inert gas, and a gas containing at least an oxygen element; The second step of heating at a higher temperature than the first step may be performed in a mixed gas atmosphere made of an inert gas.

  According to the above method, a semiconductor device in which the density of the interface state between the silicon carbide semiconductor layer and the silicon oxide layer is small while reducing the time required to form the silicon oxide layer on the surface of the silicon carbide semiconductor layer. Can do. The rate of oxidizing the silicon carbide semiconductor layer is faster as the ambient temperature is higher. However, when the silicon carbide semiconductor layer is oxidized in an atmosphere containing ammonia gas, if it is heated at an excessively high temperature, the ammonia gas damages the silicon carbide semiconductor layer or the silicon oxide layer. That is, since the silicon carbide semiconductor layer cannot be heated at a very high temperature, the time required to form the silicon oxide layer is lengthened. However, in the above method, after removing the cause of increasing the interface state density between the silicon carbide semiconductor layer and the silicon oxide layer (first step), heating is performed in an atmosphere not containing ammonia gas (second step). Therefore, it can be heated at a high temperature. In the second step, the interface state of the silicon carbide semiconductor layer and the silicon oxide layer is reduced in the first step even though the effect of removing the cause of increasing the density of the interface state between the silicon carbide semiconductor layer and the silicon oxide layer is small. Since the cause of increasing the density is removed, it is possible to prevent silicon having unbonded carbon or unpaired electrons from concentrating on the interface between the silicon carbide semiconductor layer and the silicon oxide layer.

In the method of the present invention, the heating temperature in the first step is preferably 900 ° C. or higher and 1100 ° C. or lower, and the heating temperature in the second step is preferably 1100 ° C. or higher and 1300 ° C. or lower.
As described above, in order to oxidize the silicon carbide semiconductor layer, it is preferable to heat the silicon carbide semiconductor layer at a high temperature. In the first step, when the heating temperature is lower than 900 ° C., the speed of forming the silicon oxide layer is remarkably slowed. On the other hand, when the heating temperature is higher than 1100 ° C., the silicon oxide layer is damaged by the ammonia gas.
In the second step, if the heating temperature is lower than 1100 ° C., the formation rate of the silicon oxide layer cannot be made too fast. On the other hand, when the heating temperature is higher than 1300 ° C., an apparatus for processing the silicon carbide semiconductor layer may be damaged.

In the method of the present invention, prior to the formation of the silicon oxide layer on the surface of the silicon carbide semiconductor layer, the pretreatment step of heating the surface of the silicon carbide semiconductor layer in a mixed gas atmosphere composed of ammonia gas and an inert gas. It is preferable to implement.
According to the above manufacturing method, impurities (such as unbonded carbon) existing on the surface of the silicon carbide semiconductor layer can be removed, or crystal defect portions generated in the process of forming the silicon carbide semiconductor layer can be nitrided. Causes of increasing the density of interface states between the silicon carbide semiconductor layer and the silicon oxide layer in the silicon oxide layer because the surface of the silicon carbide semiconductor layer can be stabilized before the silicon carbide semiconductor layer is oxidized. Is difficult to remain. The density of the interface state between the silicon carbide semiconductor layer and the silicon oxide layer can be efficiently reduced.

In the method of the present invention, after a silicon oxide layer is formed on the surface of the silicon carbide semiconductor layer, a post-treatment step is performed in which the surface of the silicon carbide semiconductor layer is heated in a mixed gas atmosphere composed of ammonia gas and an inert gas. It is preferable to do.
According to the above method, functional groups bonded to the surface of the silicon oxide layer can be removed, and impurities present in the silicon oxide layer or at the interface between the silicon carbide semiconductor layer and the silicon oxide layer can be removed. The on-resistance of the semiconductor device can be further reduced.

It is preferable that the heating temperature in a pre-processing process or a post-processing process is 500 degreeC or more and 1100 degrees C or less.
By setting the heating temperature in the pretreatment step within the above range, the surface of the silicon carbide semiconductor layer can be more effectively stabilized. When the heating temperature is lower than 500 ° C., the temperature difference from the previous process or the next process (process for forming a silicon oxide layer) is large, and the manufacturing time of the semiconductor device becomes long. On the other hand, when the heating temperature is higher than 1100 ° C., ammonia gas may damage the silicon carbide semiconductor layer or the silicon oxide layer as described above.

  According to the present invention, the density of the interface state between the silicon carbide semiconductor layer and the silicon oxide layer can be reduced. When an electrode facing the silicon carbide semiconductor layer is formed through the silicon oxide layer and a voltage is applied to the electrode, carriers move through a channel formed at the interface between the silicon nitride semiconductor layer and the silicon oxide layer. , Carriers can be suppressed from being trapped by the interface state.

The features of the present invention are listed.
(First Embodiment) A p-type silicon carbide semiconductor region (base region) 6 is formed on the surface of n-type silicon carbide semiconductor layer (drift region) 18. Silicon carbide semiconductor region 6 is dispersedly formed on the surface of silicon carbide semiconductor layer 18.
(Second Feature) A gate electrode 2 is formed on both the drift region 18 and the base region 6 with the silicon oxide layer (insulating layer) 4 therebetween.
(Third Feature) An n ⁺ type semiconductor region (source region) 8 is formed on the surface of the base region 6. Source region 8 is separated from drift region 18 by base region 6.
(Fourth Feature) In the step of forming silicon oxide layer 4 on the surfaces of silicon carbide semiconductor layers 18, 6, and 8, water vapor is used as a gas containing an oxygen element.
(5th characteristic) In the said process, the density | concentration of ammonia is 5-30 vol%.

Embodiments will be described in detail below with reference to the drawings. In addition, although an Example demonstrates MOSFET, the technique of this invention is applicable also to IGBT etc.
(Example 1)
FIG. 1 schematically shows a cross-sectional view of a semiconductor device 10 of this embodiment. FIG. 1 shows a unit structure of the semiconductor device 10. In practice, this semiconductor unit structure is repeatedly formed in the left-right direction on the paper. Note that the configuration of each part does not accurately represent the actual size scale. In order to clarify the drawing, the scale of the drawing is appropriately changed.
A drain electrode 14 in which aluminum (Al) and nickel (Ni) are stacked is formed on the back surface of the semiconductor device 10. An n ⁺ -type silicon carbide semiconductor layer (drain region) 16 mainly composed of silicon carbide is formed on the surface of drain electrode 14. Nitrogen (N) is used as an impurity of the silicon carbide semiconductor layer 16 and its impurity concentration is adjusted to about 5 × 10 ¹⁸ cm ⁻³ .
An n ⁻ -type silicon carbide semiconductor layer (drift region) 18 mainly composed of silicon carbide is formed on the surface of silicon carbide semiconductor layer 16. Nitrogen is used as an impurity of the silicon carbide semiconductor layer 18, and the impurity concentration is adjusted to approximately 5 × 10 ¹⁵ cm ⁻³ .
A p-type silicon carbide semiconductor region (base region) 6 containing silicon carbide as a main material is formed on the surface of silicon carbide semiconductor layer 18. Silicon carbide semiconductor region 6 is not formed over the entire surface of silicon carbide semiconductor layer 18, but is formed discontinuously. Adjacent silicon carbide semiconductor regions 6 are separated by a silicon carbide semiconductor layer 18. Aluminum is used as an impurity of silicon carbide semiconductor region 6 and its impurity concentration is adjusted to 0.8 to 1 × 10 ¹⁷ cm ⁻³ . As will be described later, silicon carbide semiconductor region 6 is formed by ion implantation of aluminum into a predetermined portion of the surface of silicon carbide semiconductor region 18. Therefore, the value of the impurity concentration of silicon carbide semiconductor region 6 has a width.
An n ⁺ type silicon carbide semiconductor region (source region) 8 is formed on the surface of silicon carbide semiconductor region 6. Source region 8 is separated from drift region 18 by base region 6. Phosphorus (P) is used as an impurity of the source region 8 and its impurity concentration is adjusted to 1 to 3 × 10 ¹⁹ cm ⁻³ . The source region 8 is formed by ion implantation of phosphorus into a predetermined part of the base region 6. Therefore, the impurity concentration value of the source region 8 has a width.

  A silicon oxide layer (gate insulating film) 4 is formed on the surface of the base region 6. The silicon oxide layer 4 is also formed on the surface of the drift region 18 that separates the adjacent base regions 6 and the surface of a part of the source region 8. On the surface of the silicon oxide layer 4, a gate electrode 2 made mainly of polysilicon is formed. A source electrode 12 in which aluminum and nickel are laminated is formed on the surface of the source region 8. The source electrode 12 and the gate electrode 2 are electrically insulated.

Next, the operation of the semiconductor device 10 will be described.
The n ⁺ -type source region 8 connected to the source electrode 12, the n ⁺ -type drain region 16 connected to the drain electrode 14, and the n ⁻ -type drift region 18 are electrically connected by the p-type base region 6. Separated. In a state where no voltage is applied to the gate electrode 2, the travel of electrons (carriers) between the source region 8 and the drift region 18 is stopped, so that the semiconductor device 10 is turned off. When a voltage is applied to the gate electrode 2, the conductivity type of the base region 6 in the range facing the gate electrode 2 is reversed, a channel for electrons to travel is formed in the base region 6, and the semiconductor device 10 is turned on. . That is, the semiconductor device 10 performs a normally-off operation.
The electrons that have traveled from the source region 8 through the channel formed in the base region 6 flow in the drift region 18 and the drain region 16 in the vertical direction and reach the drain electrode 14. The source electrode 12 and the drain electrode 14 are electrically connected.
In the semiconductor device 10, the density of interface states between the gate insulating film 4 and the base region 6 is small. When electrons travel through the channel, the number of electrons trapped in the interface state is reduced, and the on-resistance of the semiconductor device 10 can be reduced.

A method for manufacturing the semiconductor device 10 will be described with reference to FIGS.
First, as shown in FIG. 2, silicon carbide semiconductor substrate 16 containing n ⁺ -type impurities is prepared, and n ⁻ -type silicon carbide semiconductor layer 18 is epitaxially grown on the surface of silicon carbide semiconductor substrate 16. Silicon carbide semiconductor substrate 16 is 4H—SiC, and its main surface is a plane orientation (0001) silicon surface. Next, mask layer 20 is formed on a predetermined portion of the surface of silicon carbide semiconductor layer 18, and p-type impurities are ion-implanted into the exposed surface of silicon carbide semiconductor layer 18. On the surface of n ⁻ type silicon carbide semiconductor layer 18, p type silicon carbide semiconductor region 6 is formed discontinuously dispersed. The arrows in the figure indicate the range in which impurities are ion-implanted.
Next, as shown in FIG. 3, after removing mask layer 20, mask layer 22 is formed on the exposed surface of silicon carbide semiconductor layer 18 and a predetermined portion of silicon carbide semiconductor region 6. Next, n-type impurities are ion-implanted into the exposed surface of silicon carbide semiconductor region 6. N ⁺ type silicon carbide semiconductor region 8 is formed on the surface of p type silicon carbide semiconductor region 6. The arrows in the figure indicate the range in which impurities are ion-implanted.
Next, as shown in FIG. 4, after removing mask layer 22, silicon carbide semiconductor layer 18, silicon carbide semiconductor region 6, and silicon carbide semiconductor region 8 (hereinafter referred to as silicon carbide semiconductor layers 18, 8, 6). A mask layer 24 is formed on the surface, and heat treatment is performed in an inert gas atmosphere. By performing the heat treatment, the ion-implanted impurities (aluminum and phosphorus) are activated.

Next, as shown in FIG. 5, after removing mask layer 24, the surfaces of silicon carbide semiconductor layers 18, 8, 6 are oxidized to form silicon oxide layer 4. At this stage, the drift region 18, the body region 6 and the source region 8 shown in FIG. 1 are completed. A detailed method for forming the silicon oxide layer 4 will be described later.
Next, the silicon oxide layer 4 in the portion (see FIG. 1) where the source electrode 2 is formed on the surface of the source region 8 is removed. Next, the gate electrode 2 is formed on the surface of the gate insulating film 4, the source electrode 2 is formed on the exposed surface of the source region 8, and the drain electrode 14 is formed on the back surface of the drain region 16. The semiconductor device 10 shown in FIG.

A method for forming the silicon oxide layer 4 will be described in detail.
FIG. 8 shows the temperatures of silicon carbide semiconductor layers 18, 6, 8 when the surfaces of silicon carbide semiconductor layers 18, 8, 6 are oxidized to form silicon oxide layer 4 (see also FIG. 5), and the silicon oxide layers. The relationship of each process for forming 4 is shown. The vertical axis of the graph represents the temperature (° C.) of the silicon carbide semiconductor layers 18, 6 and 8, and the horizontal axis represents each step (A1 to A4). Time changes from the left side to the right side in FIG. Step A1 is a pretreatment step and is a step of stabilizing the surfaces of silicon carbide semiconductor layers 18, 6, and 8. Steps A2 and A3 are oxidation steps in which the surfaces of the silicon carbide semiconductor layers 18, 6, and 8 are oxidized to form the silicon oxide layer 4. Step A4 is a post-processing step, and is a step of removing impurities remaining at the interface between the silicon oxide layer 4 and the silicon carbide semiconductor layers 18, 6, 8 and the silicon oxide layer 4. In this embodiment, the silicon oxide layer 4 is formed through steps A1, A2, A3, and A4. However, in the present invention, steps A1, A3, and A4 are not necessarily essential steps. What is important is to carry out step A2. The reason will be described later. Further, there are portions where the temperature is flat on the left side of the process A1 and on the right side of the process A4. These flat portions indicate that the temperature of the silicon carbide semiconductor layers 18, 6, and 8 has increased from room temperature to the temperature in the device. In this embodiment, the temperature of the apparatus for forming the silicon oxide layer 4 is not set to 450 ° C. or lower. By not excessively reducing the temperature of the device, the time for forming the silicon oxide layer 4 can be shortened when the plurality of semiconductor devices 10 are continuously manufactured.

First, the process A1 will be described.
In step A1, silicon carbide semiconductor layers 18, 6, and 8 are heated to 500 ° C. or higher and 1100 ° C. or lower in a mixed gas atmosphere composed of ammonia gas and argon (Ar). As described with reference to FIG. 1, silicon carbide semiconductor layer 18 is formed by epitaxial growth from the surface of silicon carbide semiconductor substrate 16. Fine crystal defects and impurities (for example, unbonded carbon) are present on the surface of silicon carbide semiconductor layer 18. As described with reference to FIG. 2, silicon carbide semiconductor region 6 is formed by ion-implanting p-type impurities into a predetermined portion of the surface of silicon carbide semiconductor region 18. As described with reference to FIG. 3, silicon carbide semiconductor region 8 is formed by ion-implanting n-type impurities into the surface of silicon carbide semiconductor region 6. Also by performing ion implantation, crystal defects and unbonded carbon are generated on the surfaces of the silicon carbide semiconductor regions 6 and 8. When a crystal defect occurs, a dangling bond (unpaired electron) is generated in the silicon atom, and a functional group may be bonded to the unpaired electron.
By performing the step A1, it is possible to prevent other functional groups from being bonded due to nitrogen bonding to unpaired electrons of silicon, or to remove unbonded carbon. The surfaces of silicon carbide semiconductor layers 18, 6, 8 can be stabilized, and when the silicon oxide layer 4 is formed, the interface state between the silicon carbide semiconductor layers 18, 6, 8 and the silicon oxide layer 4 can be reduced. The density can be reduced. In addition, it is preferable to implement process A1 in 500 degrees C or more and 1100 degrees C or less. When step A1 is performed at a temperature higher than 1100 ° C., silicon carbide semiconductor layers 18, 6, and 8 may be damaged by ammonia, and silicon may escape from the crystal. If the step A1 is performed at a temperature lower than 500 ° C., the temperature difference from the step A2 described later becomes too large, and the time for raising the temperature from the step A1 to the step A2 becomes too long. In order to shorten the manufacturing process of the semiconductor device 10, the process A1 is preferably performed in the above temperature range. In this embodiment, step A1 is performed at 700 ° C. Further, as described above, the step A1 can be omitted.

Next, step A2 will be described.
By performing step A2, silicon carbide semiconductor layers 18, 6, and 8 are oxidized to form silicon oxide layer 4. In step A2, silicon carbide semiconductor layers 18, 6, and 8 are heated to 900 ° C. or higher and 1100 ° C. or lower in a mixed gas atmosphere containing at least an oxygen element-containing gas, ammonia gas, and argon. The concentration of ammonia gas in the mixed gas atmosphere is not particularly limited, but the concentration of ammonia gas is preferably in the range of 1 to 50%. In particular, it is preferably in the range of 5-30 vol%. In this example, the ammonia gas concentration was about 15 vol%. Note that examples of the gas containing an oxygen element include oxygen gas and water vapor. In this example, water vapor was used as the gas containing oxygen element.
Here, the reason why ammonia gas is used in step A2 will be described. In order to remove unbonded carbon that could not be removed in step A1, it is necessary to heat silicon carbide semiconductor layers 18, 6, and 8 in a mixed gas atmosphere containing a gas having high reactivity with carbon. . Examples of the gas having high reactivity with carbon include hydrogen gas and ammonia gas. When hydrogen gas is used, silicon unpaired electrons are terminated with hydrogen. Even if unpaired electrons of silicon are terminated with hydrogen, hydrogen is easily desorbed, and as a result, unpaired electrons of silicon cannot be terminated. In addition, ammonia gas is easier to react with carbon than hydrogen gas. For these reasons, it is effective to select ammonia gas over hydrogen gas.
FIG. 6 shows the result of calculating the carbon removal ability using hydrogen gas or ammonia gas using thermochemical equilibrium state calculation. The vertical axis of the graph indicates the remaining amount (%) of carbon, and the horizontal axis of the graph indicates the reaction temperature (° C.). The smaller the amount of carbon, the higher the ability to remove carbon. A curve C1 in the figure indicates the remaining amount of carbon when hydrogen gas is used, and a curve C2 indicates the remaining amount of carbon when ammonia gas is used.
As is apparent from FIG. 6, the amount of carbon remaining is smaller when ammonia gas is used than when hydrogen gas is used. That is, it is shown that unbound carbon can be efficiently removed by using ammonia gas rather than hydrogen gas. In both hydrogen gas and ammonia gas, the remaining amount of carbon increases as the temperature rises. This phenomenon is because the generated methane gas is decomposed to generate carbon again. However, since the generated methane gas is exhausted outside the apparatus in the implementation process, there is little influence due to redeposition of unbonded carbon on the silicon carbide semiconductor layers 18, 6, 8 and the silicon oxide layer 4.
By using ammonia gas in step A2, a silicon oxide layer can be formed while removing unbonded carbon or terminating unpaired electrons of silicon with nitrogen. The interface state density between the silicon carbide semiconductor layer and the silicon oxide layer can be reduced.

Next, the reason why the process A2 is heated at a temperature of 900 ° C. or higher and 1100 ° C. or lower will be described.
FIG. 7 shows the result of thermodynamic equilibrium calculation for the reaction state of silicon carbide and ammonia. The vertical axis of the graph indicates the molar amount (mol) of the substance, and the horizontal axis indicates the temperature (° C.). Curve C3 represents silicon carbide, curve C4 represents methane (CH ₄ ), curve C5 represents hydrogen (H ₂ ), curve C6 represents silicon nitride (Si ₃ N ₄ ), and curve C7 represents carbon. Curve C8 represents silicon and curve C9 represents nitrogen.
As is apparent from FIG. 7, silicon in silicon carbide reacts with ammonia to produce silicon nitride of about 1/100 of silicon carbide (see C3 and C6). The generation of silicon nitride is almost constant even when the temperature changes. Also, carbon in silicon carbide reacts with ammonia to produce methane (see C4). Methane production decreases with increasing temperature. As the production of methane decreases, that is, as the temperature increases, carbon begins to be generated (see C7). Carbon is produced by the decomposition of methane. This phenomenon occurs as a result of thermodynamic equilibrium calculation in a closed system. When the silicon carbide semiconductor layers 18, 8, 6 are actually oxidized to form the silicon oxide layer 4, since methane is excluded from the outside of the apparatus, the silicon carbide semiconductor layers 18, 8, 6 and the silicon oxide layer 4 The unbonded carbon will not be redeposited. As shown by curve C5, the amount of hydrogen generated increases as the temperature increases. Hydrogen is generated by the reaction of silicon and ammonia nitrogen, or by the decomposition of methane.
As shown by curve C8, silicon is generated when the temperature is 600 ° C. or higher. The production of silicon increases as the temperature increases. That is, the higher the temperature, the easier it is to oxidize silicon carbide to form a silicon oxide layer. Further, as shown by a curve C9, when the temperature exceeds 850 ° C., nitrogen starts to be generated. Nitrogen is generated by desorption of nitrogen from silicon nitride. That is, when the temperature is 850 ° C. or higher, it becomes easier to form the silicon oxide layer. Further, when the temperature is lower than 900 ° C., the reaction between silicon and oxygen is slow, and it takes time to form the silicon oxide layer. Therefore, Step A2 is performed at a temperature of 900 ° C. or higher. Further, when heat treatment is performed at a temperature higher than 1100 ° C., the generated silicon oxide layer is damaged by ammonia. Therefore, step A2 is performed at a temperature of 900 ° C. or higher and 1100 ° C. or lower. In this example, the temperature in step A2 was 1030 ° C.

Next, step A3 will be described.
As described above, by performing step A2, the silicon oxide layer can be formed on the surface of the silicon carbide semiconductor layer while reducing the density of the interface state between the silicon carbide semiconductor layer and the silicon oxide layer. However, the production rate of silicon oxide increases as the temperature increases. That is, as the silicon carbide semiconductor layer is heat-treated at a higher temperature, the manufacturing time of the silicon oxide film can be shortened. In step A3, the temperature can be set to 1100 ° C. or higher in order to perform the heat treatment in a mixed gas atmosphere composed of a gas containing at least an oxygen element and an argon gas (not containing ammonia gas). In addition, when heat-treating the silicon carbide semiconductor layer, the heat treatment is usually performed in a space formed of quartz. If the temperature exceeds 1300 ° C., quartz is deformed, and therefore, the heat treatment temperature in step A3 is preferably 1300 ° C. or lower. In this example, the heat treatment temperature in step A3 was 1180 ° C., and water vapor was used as the gas containing oxygen element.
In step A3, since ammonia gas is not included, the effect of terminating unpaired carbon or silicon unpaired electrons with nitrogen cannot be obtained. However, by performing the step A2, the surface layer portion of the silicon oxide layer 4 has very few unbonded carbon and silicon unpaired electrons. Unbonded carbon and unpaired electrons of silicon present in the portion oxidized in step A3 are diffused into a region where there are few of them (silicon oxide layer formed in step A2). Therefore, the density of the interface states between the silicon carbide semiconductor layers 18, 8, 6 and the silicon oxide layer 4 can be reduced even if heat treatment is performed with a mixed gas not containing ammonia in the step A 3. As described above, step A3 can be omitted. For example, when the thickness of the silicon oxide layer 4 is thin, a necessary thickness can be obtained by performing only the step A2.

Next, step A4 will be described.
In step A4, silicon carbide semiconductor layers 18, 8, 6 and silicon oxide layer 4 are heated to 500 ° C. or higher and 1100 ° C. or lower in a mixed gas atmosphere composed of ammonia gas and argon gas. By performing step A4, functional groups such as (—H) and (—OH) bonded to silicon can be removed from the silicon oxide layer 4 and terminated with nitrogen. Further, the silicon dioxide layer 4 and the unbonded carbon existing at the interface between the silicon carbide semiconductor layers 18, 6, 8 and the silicon oxide layer 4 are removed, or unpaired electrons of silicon in those portions are terminated with nitrogen. Can do. The density of the interface state between silicon carbide semiconductor layers 18, 6, 8 and silicon oxide layer 4 can be further reduced. In this embodiment, step A4 is performed at 700 ° C. Further, as described above, step A4 can be omitted.
Note that an inert gas such as nitrogen (N ₂ ) may be used in place of the argon in steps A1 to A4.

(Experimental example)
The semiconductor device 10 of Example 1 was manufactured by changing the method of forming the silicon oxide layer 4, and then the channel mobility (unit: cm ² / Vs) of the semiconductor device 10 was measured. The semiconductor devices 10 of Experimental Examples 1 to 5 described below are the same except for the method of forming the oxygen silicon layer 4.
(Experiment 1)
Without performing steps A1, A3, and A4, silicon oxide layer 4 was formed only in step A2. In this experiment, the silicon oxide layer 4 was formed by changing the gas component of the mixed gas to the following four conditions. The results are shown in Table 1.
Experiment A: Ammonia gas, water vapor (H ₂ O), and argon gas.
Experiment B: Water vapor and argon gas.
Experiment C: Ammonia gas, oxygen (O ₂ ), and argon gas.
Experiment D: Oxygen and argon gas.

  From the results in Table 1, it was confirmed that the channel mobility was increased when the mixed gas in Step A2 contained ammonia. This shows that the density of interface states between the silicon carbide semiconductor layers 18, 6, 8 and the silicon oxide layer 4 is reduced. In addition, channel mobility is increased by using water vapor as a gas containing an oxygen element. Carbon generated when the silicon carbide semiconductor layers 18, 6, 8 are oxidized reacts with hydrogen in the water vapor to form the inside of the silicon oxide layer 4 or the interface between the silicon carbide semiconductor layers 18, 6, 8 and the silicon oxide layer 4. It is removed from.

(Experiment 2)
Without performing steps A1, A3, and A4, silicon oxide layer 4 was formed only in step A2. In this experiment, the silicon oxide layer 4 was formed by changing the ambient temperature in the step A2 to the following three conditions. The results are shown in Table 2. Further, in this experiment, the channel mobility was measured, and the time required for the silicon oxide layer 4 to reach the target thickness was compared with the time required for reaching the target thickness earliest.
Experiment E: Atmospheric temperature 850 ° C.
Experiment F: Atmospheric temperature 1030 ° C.
Experiment G: Atmospheric temperature 1180 ° C.

  From the results in Table 2, it can be seen that the silicon oxide layer 4 quickly reaches the target thickness as the temperature in the step A2 increases. However, it was confirmed that the channel mobility decreases when the temperature is too high (1180 ° C.). That is, the formed silicon oxide layer 4 is damaged by ammonia, and as a result, the density of the interface state between the silicon carbide semiconductor layers 18, 6, 8 and the silicon oxide layer 4 is increased. Yes. It was confirmed that the temperature in step A2 is preferably 1100 ° C. or lower. In this experiment, when the temperature in step A2 was too low (850 ° C.), not only did the silicon oxide layer 4 reach the target thickness, but also the channel mobility decreased. As described above, in the thermochemical equilibrium calculation of silicon carbide and ammonia, more silicon is generated as the temperature increases. In other words, carbon is easily desorbed from silicon carbide as the temperature increases. That is, as the temperature in step A2 increases, carbon is less likely to remain in silicon oxide layer 4 when silicon carbide semiconductor layers 18, 8, and 6 are oxidized to form silicon oxide layer 4. It was confirmed that the temperature in step A2 is preferably 900 ° C. or higher.

(Experiment 3)
Without performing steps A1 and A4, steps A2 and A3 were performed to form silicon oxide layer 4. Also in this experiment, the silicon oxide layer 4 was formed under the following three conditions by changing the atmosphere temperature in the step A2 as in the experiment 2. In this experiment, the channel mobility was measured, and the time until the silicon oxide layer 4 reached the target thickness was compared with the time required for the experiment G to reach the target thickness. The results are shown in Table 2.
Experiment H: Step A2 was performed at an ambient temperature of 850 ° C, and then Step A3 was performed at an ambient temperature of 1180 ° C.
Experiment I: Step A2 was performed at an ambient temperature of 1030 ° C, and then step A3 was performed at an ambient temperature of 1180 ° C.
Experiment J: Step A2 was performed at an atmospheric temperature of 1180 ° C., and then step A3 was performed at an atmospheric temperature of 1180 ° C.
As is clear from Table 2, no phenomenon was observed in which the channel mobility decreased even after the step A3 was performed after the step A2. When the temperature of the step A2 was carried out at 1030 ° C., the time until the silicon oxide layer 4 reached the target thickness was shortened to 1/5 by combining the step A2 and the step A3 (Experiment F and Experiment I). See). In this experiment, channel mobility was also improved. It is confirmed that by performing step A3 after performing step A2, the interface state between silicon carbide semiconductor layers 18, 6, 8 and silicon oxide layer 4 can be reduced and the formation rate of silicon oxide layer 4 can be increased. It was.

(Experiment 4)
In this experiment, the silicon oxide layer 4 was formed under the following conditions.
Experiment K: Without performing step A4, steps A1, A2, and A3 were performed to form a silicon oxide layer 4.
Experiment L: Without performing step A1, steps A2, A3, and A4 were performed to form silicon oxide layer 4.
Experiment M: Without performing steps A1 and A4, steps A2 and A3 were performed to form silicon oxide layer 4 (same as Experiment I).
The channel mobility was also measured in this experiment. The results are shown in Table 3.

  As is clear from Table 3, it can be seen that the channel mobility is increased by performing the step A1 prior to the steps A2 and A3. That is, by removing unbonded carbon existing on the surfaces of the silicon carbide semiconductor layers 18, 6, 8 or reducing crystal defects in the silicon carbide semiconductor layers 18, 6, 8, It was confirmed that the density of interface states is reduced by forming the silicon oxide film 4 after stabilizing the surface. Similarly, it can be seen that the channel mobility is increased by performing the step A4 after the silicon oxide film 4 is formed. Unbonded carbon remaining in the silicon oxide film 4 and at the interface between the silicon carbide semiconductor layers 18, 6, 8 and the silicon oxide film 4 is removed or bonded to silicon in the silicon oxide layer 4 (− It was confirmed that the density of the interface states between the silicon carbide semiconductor layers 18, 6, 8 and the silicon oxide film 4 can be reduced by removing functional groups such as (H) and (—OH).

(Experiment 5)
Without performing steps A1, A3, and A4, silicon oxide layer 4 was formed only in step A2. In this experiment, the concentration (vol%) of ammonia gas in the mixed gas was changed, and the silicon oxide layer 4 was formed under the following five conditions. The results are shown in Table 4.
Experiment O: Ammonia gas concentration 1 vol%.
Experiment P: Ammonia gas concentration 5 vol%.
Experiment Q: Ammonia gas concentration of 15 vol%.
Experiment R: Ammonia gas concentration of 30 vol%.
Experiment S: Ammonia gas concentration 50 vol%.

  As is clear from Table 4, it was confirmed that the channel mobility was highest when the concentration of ammonia gas in the mixed gas was 15 vol%. Even if the concentration of ammonia gas in the mixed gas is too low (1 vol%) or too high (50 vol%), the channel mobility becomes small. That is, in step A2, the concentration of ammonia gas is preferably in the range of 5 to 30 vol%. If the ammonia gas concentration is too low, it is presumed that the effect of removing unbonded carbon and termination of unpaired electrons in silicon do not occur sufficiently. Further, it is presumed that when the concentration of ammonia gas becomes too high, oxygen for forming the silicon oxide layer becomes insufficient, and oxygen bonded to silicon decreases. However, in all of the experiments O to S, the channel mobility is higher than the condition in which the mixed gas does not contain ammonia gas (see experiment B). That is, it was confirmed that the channel mobility is increased by setting the ammonia gas concentration to 1 to 50%. As a result of this experiment, it was confirmed that the channel mobility was further increased by setting the ammonia gas concentration to 5 to 30%. However, the concentration of ammonia gas in the mixed gas is not limited to the above range. What is important is that ammonia gas is used as a component of the mixed gas.

Specific examples of the present invention have been described above in detail, but these are merely examples and do not limit the scope of the claims. The technology described in the claims includes various modifications and changes of the specific examples illustrated above.
For example, in the above embodiment, 4H—SiC is used as the silicon carbide semiconductor substrate containing n ⁺ -type impurities, and the main surface uses a (0001) silicon surface. However, other silicon carbide semiconductor substrates can also be used. For example, it is possible to use a silicon carbide substrate having 4H—SiC and a principal surface having a plane orientation (000-1) carbon surface, or 4H—SiC and having a principal surface having a plane orientation (11-20) carbon surface.
In the above-described embodiment, aluminum is ion-implanted as a p-type impurity, but is not limited to aluminum, and boron (B) or the like can also be used. That is, any p-type impurity may be used.
In the above embodiment, phosphorus is ion-implanted as an n-type impurity, but is not limited to phosphorus, and nitrogen or the like can be used. That is, any n-type impurity may be used.
In the above embodiment, argon is used as the inert gas, but the type of the inert gas is not limited to argon, and nitrogen, helium (He), or the like can also be used.
Further, the technical elements described in the present specification or drawings exhibit technical usefulness alone or in various combinations, and are not limited to the combinations described in the claims at the time of filing. In addition, the technology exemplified in the present specification or the drawings can achieve a plurality of objects at the same time, and has technical utility by achieving one of the objects.

Example 1 A semiconductor device is shown. 1 shows a manufacturing process of a semiconductor device of Example 1. 1 shows a manufacturing process of a semiconductor device of Example 1. 1 shows a manufacturing process of a semiconductor device of Example 1. 1 shows a manufacturing process of a semiconductor device of Example 1. The result of the thermochemical equilibrium state calculation regarding the relationship between the amount of carbon removal and temperature is shown. The result of the thermodynamic parallel state calculation about the relationship between the reaction state of silicon carbide and ammonia and the temperature is shown. The relationship between the step of forming the silicon oxide layer and the temperature in the manufacturing process of the semiconductor device of Example 1 will be described.

Explanation of symbols

4: Silicon oxide layer 6: Base region 8: Source region 10: Semiconductor device 16: Drain region 18: Drift region

Claims (6)

A method of forming a silicon oxide layer on the surface of the silicon carbide semiconductor layer;
A method for forming a silicon oxide layer, comprising: heating a surface of a silicon carbide semiconductor layer in a mixed gas atmosphere containing a gas containing at least an oxygen element, ammonia gas, and an inert gas. A method of forming a silicon oxide layer on the surface of the silicon carbide semiconductor layer;
A first step of heating the surface of the silicon carbide semiconductor layer in a mixed gas atmosphere containing a gas containing at least an oxygen element, ammonia gas, and an inert gas;
A method for forming a silicon oxide layer, comprising: a second step of heating at a higher temperature than the first step in a mixed gas atmosphere including a gas containing at least an oxygen element and an inert gas. The heating temperature in the first step is 900 ° C. or higher and 1100 ° C. or lower,
The method according to claim 2, wherein the heating temperature in the second step is 1100 ° C or higher and 1300 ° C or lower. Prior to the step of forming the silicon oxide layer on the surface of the silicon carbide semiconductor layer,
4. The method according to claim 1, wherein a pretreatment step of heating the surface of the silicon carbide semiconductor layer in a mixed gas atmosphere composed of ammonia gas and an inert gas is performed. After the step of forming the silicon oxide layer on the surface of the silicon carbide semiconductor layer,
5. The method according to claim 1, wherein a post-treatment step is performed in which the surface of the silicon carbide semiconductor layer is heated in a mixed gas atmosphere composed of ammonia gas and an inert gas.   The method according to claim 5 or 6, wherein the heating temperature in the pretreatment step or the posttreatment step is 500 ° C or higher and 1100 ° C or lower.

Images