JP5661009B2 - Manufacturing method of semiconductor device - Google Patents

Description

  The present invention relates to a method for manufacturing a semiconductor device in which a silicon substrate is irradiated with a pulsed laser beam to perform impurity activation annealing.

  A technique using a high-power pulsed laser such as a second harmonic of a solid-state laser such as a YAG laser or an excimer laser is known for activation annealing of impurities implanted into a silicon substrate. The pulse width of the pulse laser output from these laser oscillators is several hundred ns at most. If a pulse laser beam having such a short pulse width is used to secure the pulse energy necessary for activation, the peak energy must be increased. When the peak energy increases, the temperature in the vicinity of the surface rises in a short time, so that it is difficult to activate impurities in a deep region of 1 μm or more. Moreover, the wavelength of these pulsed laser beams is an ultraviolet region or a visible region shorter than 530 nm. Since the laser beam in this wavelength band is easily absorbed by silicon, the depth of penetration into the silicon substrate is 1 μm or less. For this reason, it is disadvantageous for activating impurities in a region deeper than 1 μm.

  A technique is known in which a deep region is activated by using a laser beam having a wavelength of 690 nm to 900 nm and setting the irradiation time of the silicon substrate to 10 μs to 100 μs (Patent Document 1). A technique is known in which impurities of a deep region of 2 μm or more are activated by using a second harmonic of a YAG laser, a pulse width of 250 ns to 1200 ns, and using a laser pulse with a slow pulse rise. (Patent Document 2). Furthermore, a technique for activating impurities in a region having a depth of 2 μm or more by using a near infrared laser that assists annealing is known (Patent Document 3).

JP 2006-351659 A JP 2011-60868 A JP 2011-119297 A

  As described in Patent Document 1, in order to use a continuous wave laser with a beam width of 50 μm and an irradiation time of 10 μs, the moving speed of the stage holding the silicon substrate must be 5 m / s. It is difficult to move the stage at such a high speed.

  Although the laser annealing technique described in Patent Documents 2 and 3 can activate a deep region of 2 μm or more, a technique for further increasing the activation rate is desired.

According to one aspect of the invention,
While a pulse laser beam having a wavelength of 690 nm to 950 nm, a pulse width of 10 μs to 30 μs, and a pulse repetition frequency of 0.5 kHz to 2.5 kHz is incident on the surface of the silicon substrate into which impurities are implanted, If the cross section of the pulse laser beam is moved at an overlap rate of 50% to 80% and the silicon substrate surface is irradiated with a power density that does not melt the surface of the silicon substrate and the thickness is 525 μm, the opposite is true. Provided is a method of manufacturing a semiconductor device that performs activation annealing for activating impurities up to a depth of at least 2.5 μm under the condition that the highest temperature on the side surface is 115 ° C. or lower.

  By performing laser annealing under these conditions, the activation rate of impurities in a deep region of 2 μm or more can be increased.

FIG. 1 is a schematic diagram of a laser annealing apparatus used in a semiconductor device manufacturing method according to an embodiment. FIG. 2 is a cross-sectional view of a semiconductor device (IGBT) manufactured by the method according to the embodiment. FIG. 3A is a timing chart of laser annealing employed in the method according to the embodiment, and FIG. 3B is a plan view showing a beam cross section on the silicon substrate. 4A, FIG. 4B, and FIG. 4C show the maximum ultimate temperature and activity of the non-irradiated surface when annealing is performed under various laser irradiation conditions with the frequency f set to 0.5 kHz, 1.0 kHz, and 2.5 kHz, respectively. It is a chart which shows conversion rate. FIG. 5 is a graph showing the distribution of impurity concentration and carrier concentration in the depth direction. FIG. 6 is a graph showing the distribution of impurity concentration and carrier concentration in the depth direction. It is a graph which shows the relationship between a pulse width when frequency f is set to 0.5 kHz, and the upper limit of the power density which satisfy | fills non-melting conditions. It is a graph which shows the relationship between a pulse width when the frequency f is set to 1.0 kHz, and the upper limit of the power density which satisfies non-melting conditions. It is a graph which shows the relationship between the frequency f and the upper limit of the power density which satisfy | fills non-melting conditions.

  FIG. 1 is a schematic diagram of a laser annealing apparatus used in a method for manufacturing a semiconductor device according to an embodiment. The laser light source 11 is driven by the driver 10. For the laser light source 11, for example, a laser diode having an oscillation wavelength of 690 to 950 nm is used. In this embodiment, a laser diode having an oscillation wavelength of 800 nm is used as the laser light source 11. During the annealing, the laser light source 11 is continuously oscillated.

  A laser beam emitted from the laser light source 11 enters the half-wave plate 12. The half-wave plate 12 changes the polarization direction of the laser beam by changing the direction of the slow axis. The laser beam that has passed through the half-wave plate 12 enters the acoustooptic device 13. The acoustooptic device 13 is controlled by the control device 25 to move the laser beam straight or change the traveling direction. The laser beam whose traveling direction has changed enters the beam damper 14.

  The laser beam that has traveled straight through the acoustooptic device 13 passes through the homogenizer 15 and then enters the beam splitter 16. The beam splitter 16 reflects some components of the incident laser beam toward the beam damper 17 and advances the remaining components straight. The ratio of the component that travels straight through the beam splitter 16 can be controlled by changing the direction of the slow axis of the half-wave plate 12 to change the polarization direction.

  The laser beam that has passed through the beam splitter 16 passes through the quarter-wave plate 19 and the condenser lens 20 and enters the silicon substrate 30 that is to be annealed. The silicon substrate 30 is held on the movable stage 21. The homogenizer 15 and the condensing lens 20 make the beam cross section on the surface of the silicon substrate 30 into a long shape, and uniformize the light intensity with respect to the long axis and the direction (width direction) perpendicular thereto. The movable stage 21 moves the silicon substrate 30 in the width direction of the beam cross section.

  The reflected light reflected by the surface of the silicon substrate 30 passes through the condenser lens 20 and the quarter wavelength plate 19 and enters the beam splitter 16. By passing through the quarter-wave plate 19 twice, the polarization direction changes by 90 °. Therefore, the reflected light is reflected by the beam splitter 16 and enters the beam damper 18.

  The acoustooptic device 13 can generate a pulse laser beam from a continuous oscillation laser beam by alternately repeating a period in which the laser beam goes straight and a period in which the laser beam is incident on the beam damper 14. By controlling the acoustooptic device 13, the pulse width of the pulse laser beam and the pulse repetition frequency (hereinafter simply referred to as “frequency”) can be controlled.

  FIG. 2 shows a cross-sectional view of an insulated gate bipolar transistor (IGBT) as an example of a semiconductor device manufactured by the method according to the embodiment. The IGBT is manufactured by forming an emitter and a gate on one surface of an n-type silicon substrate 30 and forming a collector on the other surface. The structure of the surface on which the emitter and the gate are formed is manufactured in the same process as a general MOSFET manufacturing process. For example, as shown in FIG. 2, a p-type base region 33, an n-type emitter region 34, a gate electrode 35, a gate insulating film 36, and an emitter electrode 37 are disposed on the surface of the silicon substrate 30. Current on / off control can be performed by the voltage between the gate and the emitter.

  A p-type collector layer 39 is formed on the opposite surface of the silicon substrate 30. An n-type buffer layer 38 may be formed between the collector layer 39 and the silicon substrate 30 as necessary. The collector layer 39 and the buffer layer 38 are formed by implanting boron and phosphorus as impurities, respectively, by ion implantation and performing activation annealing. The laser annealing apparatus shown in FIG. 1 is applied to this activation annealing. A collector electrode 40 is formed on the surface of the collector layer 39 after the activation annealing.

  FIG. 3A shows a timing chart of laser annealing employed in the semiconductor device manufacturing method according to the embodiment. A laser pulse having a pulse width PW is applied to the silicon substrate 30 at a frequency f, that is, a period 1 / f. Since this laser pulse is cut out from the continuously oscillating laser beam by the acoustooptic device 13 (FIG. 1), the waveform thereof is close to a rectangle. The laser light source 11 may be pulse-oscillated by the driver 10 and the acoustooptic device 13 and the beam damper 14 may be omitted.

  The element structure on the front side shown in FIG. 2 is already formed in the silicon substrate 30 at the time of annealing, and impurities of the buffer layer 38 and the collector layer 39 are implanted on the back side. At this stage, the implanted impurities are not activated.

  FIG. 3B shows a beam cross section 23 on the laser irradiation surface of the silicon substrate 30 (FIG. 1). The beam cross section 23 has an elongated shape that is long in one direction. Since the light intensity distribution is made uniform by the homogenizer 15 (FIG. 1), the light intensity distribution in the longitudinal direction and the width direction has a substantially top flat shape.

  The length in the longitudinal direction is L, and the beam width is Wt. Irradiation with the pulse laser beam is performed while moving the silicon substrate 30 in the width direction of the beam cross section 23. By moving the silicon substrate 30, the pulse laser beam scans the surface of the silicon substrate 30. By scanning the pulse laser beam in the beam width direction, a rectangular region in which the length of one side is equal to the length L of the beam cross section can be annealed.

  The width of the overlapping region between the beam cross section of one laser pulse (shown by a solid line in FIG. 3B) and the beam cross section of the next incident laser pulse (shown by a broken line in FIG. 3B). Let Wo. The overlap rate is defined as Wo / Wt. When the moving speed of the silicon substrate 30 is V, Wt−Wo = V / f. Therefore, the overlap rate is determined by the pulse repetition frequency f, the beam width Wt, and the stage moving speed V. In the embodiment, the beam width Wt is 240 μm.

  Boron (B) and phosphorus (P) were implanted into the silicon substrate 30 (FIG. 1), and activation annealing was performed under various laser irradiation conditions. Hereinafter, this evaluation experiment will be described.

  Generally, the thickness of the silicon substrate 30 (FIG. 2) used for the IGBT is about 100 μm. When the laser beam is incident on the back surface of the silicon substrate 30 (the surface on which the collector layer 39 is formed), if the temperature on the front surface (the surface on which the emitter region 34 or the like is formed) becomes too high, it is already formed. The device structure is damaged. The power density on the back surface of the silicon substrate 30 was gradually increased, and the irradiation conditions immediately before the element structure on the front side began to be damaged were determined. When laser irradiation was performed on the back surface of a silicon substrate having a thickness of 525 μm under the same conditions as those immediately before the start of damage, the maximum temperature reached on the front surface was 115 ° C. From this result, when the activation annealing of the IGBT is performed, the condition that the highest temperature of the opposite surface when the laser irradiation is performed on the silicon substrate having a thickness of 525 μm is 115 ° C. or lower (this condition is expressed as “non- It is considered preferable to perform laser irradiation under “damage conditions”).

  Laser irradiation was performed on a silicon substrate having a thickness of 525 μm under various conditions, and the highest temperature reached on the opposite surface and the activation rate of impurities were measured. For measurement of the maximum temperature reached on the surface of the silicon substrate, WAX temperature ink manufactured by Asei Industry Co., Ltd. was used. The WAX temperature indicating ink is applied to the surface opposite to the laser irradiation surface of the silicon substrate, and the maximum temperature reached can be known from the change in the transparency of the ink after the laser irradiation.

  In FIG. 4A, the surface of the silicon substrate opposite to the laser irradiation surface when the frequency f is 0.5 kHz and the laser irradiation is performed with the pulse width PW, the power density PD, and the overlapping ratio being different (hereinafter, referred to as the following) It shows the result of measuring the maximum temperature reached and the activation rate of impurities. The pulse width PW was selected from the conditions of 10 μs, 15 μs, 20 μs, 25 μs, and 30 μs. The overlap rate was selected from the conditions of 50%, 67%, and 80%. The power density was set to a value close to the upper limit of the power density that satisfies the condition that the laser irradiation surface of the silicon substrate does not melt (hereinafter referred to as “non-melting condition”). When laser irradiation is performed under this irradiation condition, the maximum temperature reached on the laser irradiation surface becomes substantially equal to the melting point of silicon. However, since the energy corresponding to the heat of fusion is not applied to the silicon substrate, the substrate surface does not melt.

  By heating the silicon substrate 30 to substantially the melting point without melting the laser irradiation surface, the activation rate of impurities in the surface layer portion of the laser irradiation surface can be increased. When laser irradiation is performed under conditions where the pulse width PW is 10 μs or longer and the laser irradiation surface starts to melt, the entire surface does not melt uniformly, and the melted and unmelted regions are distributed in a patch pattern. End up. When the power density is increased to such an extent that the entire surface is uniformly melted, the temperature rise of the non-irradiated surface becomes remarkable. That is, it is difficult to find a condition that the irradiated surface is melted and the element structure of the non-irradiated surface is not damaged under a long condition that the pulse width PW is 10 μs or more. Therefore, it is preferable to perform laser irradiation under non-melting conditions.

  The highest level of the non-irradiated surface of the silicon substrate is shown in the upper part of each column of the table shown in FIG. 4A, and the activation rate of phosphorus is shown in the lower part. The notation of the activation rate in parentheses is not the value obtained by actually measuring the sample, but the activation rate is greater than the parenthesized value based on the measurement result of the sample under irradiation conditions where the maximum reached temperature of the non-irradiated surface is lower It is expected to become. For example, in a sample manufactured under conditions of an overlap rate of 50% and a pulse width of 15 μs, the maximum temperature reached on the non-irradiated surface is 80 ° C., the activation rate is 100%, and the sample manufactured with a pulse width of 20 μs to 30 μs is not irradiated. Since the maximum surface temperature is 80 ° C. or higher, the activation rate of these samples is expected to be almost 100%.

  4B and 4C show the measurement results of the maximum temperature reached and the activation rate of the non-irradiated surface when the frequency f is 1.0 kHz and 2.5 kHz, respectively.

FIG. 5 shows the measurement results of the impurity concentration distribution and the carrier concentration distribution in the depth direction of the sample used for the evaluation. The horizontal axis represents depth in units of “μm”, and the vertical axis represents concentration in units of “cm ⁻³ ”. Boron is implanted under the conditions of an acceleration energy of 40 keV and a dose of 1 × 10 ¹⁵ cm ⁻² . The implantation of phosphorus was performed under the conditions of an acceleration energy of 700 keV and a dose of 1 × 10 ¹³ cm ⁻² . A broken line B0 and a broken line P0 indicate the boron concentration and the phosphorus concentration immediately after the injection, respectively. The thin solid line p1 is a sample S1 irradiated with laser under the conditions of a pulse width of 15 μs, an overlap rate of 50%, and a frequency of 0.5 kHz. The irradiated sample S2 and the thick solid line p3 indicate the hole concentration of the sample S3 irradiated with laser under the conditions of a pulse width of 25 μs, an overlap rate of 67%, and a frequency of 0.5 kHz. A thin solid line n indicates the electron concentration of the samples S1, S2, and S3. Since the electron concentration distributions of the three samples are almost overlapped, they are indicated by one thin solid line.

  Since the surface of the silicon substrate is not melted, the electron concentration distribution in the surface layer portion has a shape reflecting the boron concentration distribution immediately after implantation. It can also be seen that phosphorus up to a depth of 2.5 μm is almost 100% activated.

  Returning to FIG. 4A, the description will be continued. When the overlap rate is 80%, annealing can be performed under non-damage conditions within a range of pulse width PW of 10 μs to 15 μs. When the pulse width PW is within a range of 20 μs to 30 μs, when the irradiated surface is heated to the vicinity of the melting point, the maximum temperature reached on the non-irradiated surface becomes 115 ° C. or higher. For this reason, the non-damage condition cannot be satisfied.

  When the overlap rate is set to 67%, it is possible to heat the irradiated surface to near the melting point within the range of the pulse width PW of 10 μs to 25 μs and to make the maximum temperature of the non-irradiated surface less than 115 ° C. That is, annealing can be performed under non-damage conditions.

  When the overlap rate is set to 50%, the non-damage condition can be satisfied when the pulse width PW is in the range of 10 μs to 30 μs.

  As shown in FIG. 4B, when the frequency f is 1.0 kHz, annealing cannot be performed under non-damage conditions when the overlap rate is 80%. When the overlap rate is set to 67%, the non-damage condition can be satisfied by setting the pulse width to 10 μs. When the overlap rate is set to 50%, the non-damage condition can be satisfied within the range of the pulse width of 10 μs to 20 μs.

  As shown in FIG. 4C, when the frequency f is 2.5 kHz, the non-damage condition cannot be satisfied when the overlap rate is in the range of 67% to 80%. When the overlap rate is set to 50%, the non-damage condition can be satisfied within the pulse width of 10 μs to 15 μs. When the frequency f was 3 kHz, the maximum temperature reached on the non-irradiated surface exceeded 115 ° C. even when the overlap rate was 50%. From this experimental result, it is considered that the upper limit of the frequency f for satisfying the non-damage condition is 3 kHz.

  In any of the samples prepared under the non-melting and non-damaging conditions shown in FIGS. 4A to 4C, a high activation rate of 94% or more is obtained.

  Even under non-melting and non-damaging conditions, if the power density is too low, the activation rate is lowered. Next, conditions for maintaining a high activation rate will be described.

FIG. 6 shows the measurement results of the electron concentration distribution and hole concentration distribution of two samples irradiated with laser at different power densities. The horizontal axis represents depth in units of “μm”, and the vertical axis represents concentration in units of “cm ⁻³ ”. The pulse width PW was 15 μs, the frequency f was 0.5 kHz, and the overlap rate was 67%. A thin solid line n1 and a thick solid line p1 in FIG. 6 indicate an electron concentration distribution and a hole concentration distribution of a sample manufactured under the condition of a power density of 445 kW / cm ² , respectively. A thin broken line n2 and a thick broken line p2 indicate an electron concentration distribution and a hole concentration distribution of a sample manufactured under the condition of a power density of 410 kW / cm ² , respectively.

In the sample manufactured under the condition of a power density of 445 kW / cm ² , an activation rate of almost 100% is obtained. On the other hand, it can be seen that the activation rate is insufficient in the sample manufactured under the condition of a power density of 410 kW / cm ² . According to the inventors' various evaluation experiments, it is considered that a sufficient activation rate can be obtained if the power density is 92% or more of the upper limit value of the non-melting condition.

FIG. 7 shows the upper limit value of the power density that satisfies the non-melting condition when the frequency f is set to 0.5 kHz. The horizontal axis represents the pulse width in the unit “μs”, and the vertical axis represents the power density in the unit “kW / cm ² ”. Both the horizontal and vertical axes are logarithmic scales. A circle symbol plotted in a region having a pulse width of 10 μs or more is a value obtained by actually performing laser beam irradiation, and a circle symbol plotted in a pulse width of 5 μs is a value obtained by simulation.

The power density PD and the pulse width PW in the lower left region from the solid line connecting the circle symbols satisfy the non-melting condition. When the non-melting condition is expressed by a mathematical formula,
PD ≦ −45 + 1950 × PW− ^{1 / 2} (1)
It becomes. Here, PW is a value representing the pulse width in the unit “μs”, and PD is a value representing the power density in the unit “kW / cm ² ”.

When the pulse width PW and the power density PD are set to values that satisfy the inequality (1), the irradiated surface does not melt. However, if the power density becomes too low, the activation rate decreases. From the result of the evaluation experiment shown in FIG. 6, the power density is preferably 92% or more of the value on the right side of inequality (1). Expressing this condition as an inequality,
PD ≧ 0.92 × (−45 + 1950 × PW ^−1/2 ) (2)
It becomes.

  In FIG. 7, a value of 92% of the upper limit value of the power density is indicated by a broken line. In order to achieve a sufficient activation rate without melting the irradiated surface, it is preferable to select the power density PD and the pulse width PW from the region between the solid line and the broken line in the graph shown in FIG.

FIG. 8 shows the upper limit value of the power density that satisfies the non-melting condition when the frequency f is set to 1.0 kHz. The horizontal axis represents the pulse width in the unit “μs”, and the vertical axis represents the power density in the unit “kW / cm ² ”. Both the horizontal and vertical axes are logarithmic scales. A circle symbol plotted in a region having a pulse width of 10 μs or more is a value obtained by actually performing laser beam irradiation, and a circle symbol plotted in a pulse width of 5 μs is a value obtained by simulation.

The power density PD and the pulse width PW in the lower left region from the solid line connecting the circle symbols satisfy the non-melting condition. When the non-melting condition is expressed by a mathematical formula,
PD ≦ −65 + 1950 × PW− ^{1 / 2} (3)
It becomes. Here, PW is a value representing the pulse width in the unit “μs”, and PD is a value representing the power density in the unit “kW / cm ² ”.

In order to realize a sufficient activation rate, the power density is preferably 92% or more of the upper limit value of the power density that satisfies the non-melting condition. Expressing this condition as an inequality,
PD ≧ 0.92 × (−65 + 1950 × PW ^−1/2 ) (4)
It becomes.

  FIG. 9 shows the relationship between the frequency f and the upper limit value of the power density that satisfies the non-melting condition. When the frequency f is between 1.0 kHz and 3.0 kHz, the upper limit value of the power density PD that satisfies the non-melting condition is substantially constant. Note that the upper limit value of the power density PD satisfying the non-melting condition is substantially constant even when the pulse width is in the range of 5 μs to 30 μs. Therefore, when the frequency f is between 1.0 kHz and 3.0 kHz, the inequality (3) is a non-melting condition, and the inequality (4) is a condition for realizing a sufficient activation rate.

When the frequency f is 0.5 kHz, the upper limit of the non-molten condition is satisfied power density, the frequency f of about 20kW ^/ cm ² ~40kW ^/ cm 2 degrees higher than the upper limit value when the 1.0kHz~3.0kHz . Specifically, the difference between the upper limit values of the power density satisfying the non-melting condition is about 40 kW / cm ² when the pulse width is 15 μs, and is about 20 kW / cm ² when the pulse width is 20 μs to 30 μs. Under the conditions where the pulse width was 10 μs, the upper limit value of the non-melting condition could not be measured due to the apparatus configuration. Considering that the upper limit value of the power density satisfying the non-melting condition changes almost linearly in the frequency f range of 0.5 kHz to 1.0 kHz, the above inequality (1) is generalized by the frequency f.
When 20 μs ≦ PW ≦ 30 μs,
PD ≦ −45 + 1950 × PW ^−1/2 −40 × (f−0.5) (5a),
When 10 μs ≦ PW <20 μs,
PD ≦ −25 + 1950 × PW− ^{1 /} 2−80 × (f−0.5) (5b)
Is derived. Here, f is a value indicating the frequency in the unit “kHz”.

Similarly, from inequality (2) above,
When 20 μs ≦ PW ≦ 30 μs,
PD ≧ 0.92 × (−45 + 1950 × PW− ^{1 /} 2−40 × (f−0.5))
... (6a),
When 10 μs ≦ PW <20 μs,
PD ≧ 0.92 × (−25 + 1950 × PW− ^{1 /} 2−80 × (f−0.5))
... (6b),
Is derived.

  From the experimental results shown in FIG. 4A, when the frequency f is 0.5 kHz, laser irradiation with the upper limit of the power density satisfying the non-melting condition and satisfying the non-damaging condition can be performed in a region having a pulse width of 30 μs or less. it can.

  From the experimental results shown in FIG. 4B, when the frequency f is 1.0 kHz and the pulse width is set to 25 μs or more, the non-damage condition cannot be satisfied with the upper limit value of the power density that satisfies the non-melting condition. Therefore, it is preferable to set the pulse width within the range of 5 μs to 20 μs.

  From the experimental results shown in FIG. 4C, when the pulse width is set to 20 μs or more when the frequency f is 2.5 kHz, the non-damage condition cannot be satisfied with the upper limit value of the power density that satisfies the non-melting condition. Therefore, it is preferable to set the pulse width within the range of 5 μs to 15 μs.

  If the frequency f is lowered, the time required for annealing one wafer becomes longer. From the viewpoint of shortening the time required for annealing (in other words, increasing the throughput), it is preferable to increase the frequency f. However, when the frequency f is increased, the range of the pulse width PW and the overlapping rate that can satisfy the non-damage condition is narrowed in the upper limit value of the power density that satisfies the non-melting condition.

  When the overlap rate is low, the activation rate particularly in a deep region may not be uniform in the plane. Even in a deep region of about 2 μm to 3 μm, in order to make the activation rate uniform in the plane, the overlap rate is preferably 40% or more. If the overlap rate is increased, the in-plane uniformity of the activation rate increases. As shown in FIGS. 4A to 4C, if the frequency f is 0.5 kHz, the overlapping rate can be set within a range of 50% to 80%. On the other hand, when the frequency f is set to 2.5 kHz, the non-damage condition cannot be satisfied when the upper limit value of the power density PD satisfying the non-melting condition and the overlapping ratio is 67% or more.

  When the activation depth is shallower than 2 μm to 3 μm, the overlap rate or the pulse width can be set to a smaller value. As shown in FIG. 4, when the overlapping rate or the pulse width is small, the maximum temperature reached on the non-irradiated surface tends to be low. Therefore, in this case, the frequency f can be made higher than 2.5 kHz. It will be obvious to those skilled in the art that a relational expression between the power density PD and the pulse width PW can be obtained based on this result.

  The frequency f, the overlapping rate, the power density PD, and the pulse width PW are preferably selected from a range in which a further sufficient activation rate can be realized under non-melting and non-damaging conditions. Furthermore, a more preferable value may be extracted from the viewpoint of annealing time, in-plane variation in activation rate, and the like.

  Although the present invention has been described with reference to the embodiments, the present invention is not limited thereto. It will be apparent to those skilled in the art that various modifications, improvements, combinations, and the like can be made.

DESCRIPTION OF SYMBOLS 10 Driver 11 Laser light source 12 Half-wave plate 13 Acousto-optic device 14 Beam damper 15 Homogenizer 16 Beam splitter 17, 18 Beam damper 19 1/4 wavelength plate 20 Condensing lens 21 Movable stage 23 Beam cross section 25 Controller 30 Silicon substrate 33 Base region 34 Emitter region 35 Gate electrode 36 Gate insulating film 37 Emitter electrode 38 Buffer layer 39 Collector layer 40 Collector electrode

Claims (3)

While a pulse laser beam having a wavelength of 690 nm to 950 nm, a pulse width of 10 μs to 30 μs, and a pulse repetition frequency of 0.5 kHz to 2.5 kHz is incident on the surface of the silicon substrate into which impurities are implanted, If the cross section of the pulse laser beam is moved at an overlap rate of 50% to 80% and the silicon substrate surface is irradiated with a power density that does not melt the surface of the silicon substrate and the thickness is 525 μm, the opposite is true. A method for manufacturing a semiconductor device, wherein activation annealing is performed to activate an impurity having a depth of at least 2.5 μm under a condition that the highest temperature on the side surface is 115 ° C. or lower. When the pulse width is PW (μs), the power density threshold upper limit value is PD (kW / cm ² ), and the pulse repetition frequency is f (kHz), it is within the range of 0.5 ≦ f ≦ 1.0. ,
When 20 ≦ PW ≦ 30,
PD ≦ −45 + 1950 × PW ^−1/2 −40 × (f−0.5), and
PD ≧ 0.92 × (−45 + 1950 × PW− ^{1 /} 2−40 × (f−0.5)),
When 10 ≦ PW ≦ 20,
PD ≦ −25 + 1950 × PW ^−1/2 −80 × (f−0.5), and
PD ≧ 0.92 × (−25 + 1950 × PW− ^{1 /} 2−80 × (f−0.5))
2. The method of manufacturing a semiconductor device according to claim 1, wherein the surface of the silicon substrate is irradiated with the pulse laser beam under a condition that satisfies the following conditions. When the pulse width is PW (μs), the power density threshold upper limit value is PD (kW / cm ² ), and the pulse repetition frequency is f (kHz), the range is 1.0 ≦ f ≦ 2.5. ,
PD ≦ −65 + 1950 × PW ^−1/2 and PD ≧ 0.92 × (−65 + 1950 × PW ^−1/2 )
2. The method of manufacturing a semiconductor device according to claim 1, wherein the surface of the silicon substrate is irradiated with the pulse laser beam under a condition that satisfies the following conditions.

Images