JP5043445B2 - Manufacturing method of semiconductor device - Google Patents

Description

  The present invention relates to a method for manufacturing a semiconductor device, and more particularly to a method for controlling the lifetime of minority carriers in a bipolar power device such as an IGBT (Insulated Gate Bipolar Transistor).

  In a pn junction type bipolar semiconductor device, in order to reduce energy loss, it is necessary to control the lifetime (recombination lifetime) of minority carriers in a silicon substrate. Conventionally, techniques for controlling the lifetime of minority carriers include techniques for diffusing heavy metals that form recombination centers in a silicon substrate, such as Au and Pt, into the silicon substrate, and high energy on the entire surface of the silicon substrate. A technique is used in which a lattice defect serving as a recombination center is formed in a silicon substrate by irradiation with a charged particle beam.

  For example, in the following Patent Document 1, in a semiconductor device manufacturing method in which a silicon substrate is irradiated with an electron beam to generate lattice defects and then annealed, defect distribution is performed by scanning a laser beam during annealing. Has been proposed to arbitrarily control the lifetime in the wafer thickness direction and arbitrarily control the minority carrier lifetime in the wafer thickness direction.

  The following Patent Document 2 discloses a buffer having a corrugated cross-sectional shape including a plurality of regions having a carrier concentration higher than that of a silicon substrate on the anode electrode side (back surface side) and having a predetermined pitch in a semiconductor power device. There has been proposed a method of reducing energy loss during turn-off by forming a layer and forming a region having a long lifetime and a region having a short lifetime at a predetermined pitch.

  Patent Document 3 below discloses a defect area that reaches the inside of the intermediate region through the pn junction in the anode-side emitter region that is p-type doped so as to have a higher concentration than the n-type region in the power semiconductor device. There has been proposed an IGBT manufacturing method in which lifetime is controlled by forming a stripe pattern by laser light irradiation and power loss is reduced.

  The following Patent Documents 4 and 5 disclose a method of activating an impurity layer using laser annealing.

JP 7-226405 A JP-A-9-116131 Japanese Patent Laid-Open No. 1-149481 JP 2006-59876 A JP 2005-223301 A

  In Patent Document 1, a manufacturing process for realizing lifetime control includes four processes of an impurity implantation process, an impurity annealing process, an electron beam irradiation process, and an electron beam irradiation damage annealing process.

  In Patent Document 2, a manufacturing process for realizing lifetime control includes six processes including a film forming process, a photolithography process, an etching process, an impurity implantation process, a mask film removing process, and an impurity annealing process.

  Accordingly, in both Patent Documents 1 and 2, the number of steps for performing desired lifetime control increases, and the number of manufacturing apparatuses to be used increases, resulting in an increase in manufacturing cost.

  In Patent Literature 3, a defect area is formed by laser light irradiation in the anode-side emitter region. Although the tail current flow time can be shortened when the device is shut down, if the area of the defect area increases, the on-resistance when the device is on increases, the steady loss of the device increases, and the total energy loss when using the device Will increase. If the area of the defect area is reduced, the effect of shortening the time during which the tail current flows when the device is cut off is lost, and the energy loss during switching increases.

  An object of the present invention is to provide a semiconductor device manufacturing method capable of performing lifetime control with a small number of man-hours and manufacturing costs in order to comprehensively reduce steady-state loss and switching loss during on-state.

In order to achieve the above object, a semiconductor device manufacturing method according to the present invention performs laser irradiation toward a semiconductor layer to reduce lattice defects in the semiconductor layer in the laser irradiation region, thereby reducing the lattice defects. Including a step of controlling the lifetime of minority carriers in the selected area to be long,
By scanning the laser beam at a predetermined scanning pitch, the laser irradiation is performed so that the laser irradiation regions partially overlap, and the width of the beam profile at which the energy density of the laser beam becomes the first ratio of the maximum value, and the first Of the two regions (33, 34) sandwiched between the beam profile widths having the second ratio smaller than the ratio, a region having a short lifetime is formed in the region (33) behind the scan, and the regions other than the region (33) are formed. A region having a long lifetime is formed in the region .

  According to the present invention, by performing laser irradiation so that the laser irradiation regions partially overlap, regions having different minority carrier lifetimes can be formed with high accuracy. Therefore, the steady loss and switching loss at the time of ON can be reduced comprehensively with less man-hours and manufacturing costs.

Embodiment 1 FIG.
FIG. 1 is a cross-sectional view showing an example of a semiconductor device to which the present invention can be applied. Here, IGBT is exemplified as the semiconductor device, but the present invention is applicable to a general bipolar semiconductor device.

  In the semiconductor device shown in FIG. 1, a transistor 11 and a cathode electrode 12 are formed on the surface side of a substrate 10 made of n-type silicon. The transistor 11 includes an n-type charge storage layer 3, a p-type base layer 4, a gate insulating film 5, a gate electrode 6, an n-type emitter layer 7, an interlayer oxide film 8, and the like, and has a so-called trench gate structure. It is composed.

  On the other hand, an n-type buffer layer 13, a p-type collector layer 14, and an anode electrode 15 are formed on the back side of the substrate 10. Furthermore, a region 21 having a short lifetime is provided so as to pass through the p-type collector layer 14 and the n-type buffer layer 13 and reach a part of the substrate 10. Between the areas 21, areas 22 having a long lifetime are arranged.

  FIG. 2 is a plan view seen from the back side of the substrate 10 without the anode electrode 15. The short lifetime region 21 and the long lifetime region 22 are arranged in stripes in the p-type collector layer 14 and the n-type buffer layer 13.

  Here, the operation of this semiconductor device will be briefly described. When a positive bias voltage is applied to the gate electrode 6, a surface inversion layer is formed in the vicinity of the gate insulating film 5, and electrons pass from the n-type emitter layer 7 through the p-type base layer 4 and the n-type charge storage layer 3. Then, it flows into the substrate 10 and then reaches the anode electrode 15 via the n-type buffer layer 13 and the p-type collector layer 14. On the other hand, holes flow from the anode electrode 15 into the substrate 10 via the p-type collector layer 14 and the n-type buffer layer 13. At this time, conductivity modulation occurs due to double injection of electrons and holes, and the on-resistance of the device is lowered. At the turn-off time, the surface inversion layer disappears, so that electrons and holes do not flow.

  When the device is conductive, it is preferable that the minority carrier has a long lifetime in order to utilize conductivity modulation. On the other hand, when the device switches from on to off, it is preferable to quickly eliminate minority carriers to reduce tail current associated with switching losses.

  Next, a method for manufacturing this semiconductor device will be described. First, the transistor 11 is formed on the surface side of an n-type silicon substrate 10 such as an FZ (Floating Zone) wafer. That is, the n-type charge storage layer 3 is formed by introducing an n-type impurity into the substrate 10, and the p-type base layer 4 is formed by introducing the p-type impurity so as to be shallower than the n-type charge storage layer 3. Then, an n-type emitter layer 7 is formed by locally introducing an n-type impurity using a mask. Subsequently, using a mask and dry etching, a groove is formed so as to reach the substrate 10 inside the n-type emitter layer 7, and subsequently, a gate insulating film 5 is formed on the inner surface of the groove using CVD or the like. The gate electrode 6 is formed so as to fill the trench. Next, an interlayer oxide film 8 is formed so as to cover the gate electrode 6 by using CVD or the like. Subsequently, a cathode electrode 12 is formed on the interlayer oxide film 8 so as to be in contact with the n-type emitter layer 7 and the p-type base layer 4.

  Next, in order to reduce the on-resistance of the device, the back surface side of the substrate 10 is polished and thinned to a desired thickness. Next, damage (defects) generated on the back side of the substrate 10 by polishing is removed using CMP (Chemical Mechanical Planarization), wet etching, dry etching, or the like.

Next, an n-type buffer layer 13 having a desired depth and a desired impurity concentration is formed by ion-implanting an n-type impurity such as phosphorus or arsenic on the back side of the substrate 10. Subsequently, a p-type collector layer 14 is formed in a region close to the back surface of the substrate so as to be shallower than the n-type buffer layer 13 by ion implantation of p-type impurities such as boron and BF ₂ into the back surface side of the substrate 10. Form.

  Next, by irradiating the back surface of the substrate with a laser beam, the implanted impurities are activated, and at the same time, lattice defects caused by the impurity implantation are reduced by laser annealing. At this time, as the amount of laser irradiation increases, the number of lattice defects that serve as recombination centers of minority carriers decreases, and the carrier lifetime increases. Therefore, the region 21 with a short lifetime and the region 22 with a long lifetime can be formed by spatially modulating the laser irradiation amount. Finally, the anode electrode 15 is formed on the p-type collector layer 14 by sputtering or vapor deposition.

  Hereinafter, the laser irradiation process according to the present invention will be described in detail. As the laser light source to be used, it is preferable to use a Q-switch pulse laser light source capable of generating a stable laser beam with high output.

  Further, the beam shape on the substrate can be circular or rectangular, but using a rectangular pulse laser capable of realizing a high annealing temperature with relatively high production efficiency, for example, scanning in the short side direction of the rectangle, It is preferable to form the region 21 having a short lifetime and the region 22 having a long lifetime in a stripe shape.

  FIG. 3 is a graph showing an example of a beam profile during laser irradiation. The vertical axis represents the energy density of the laser beam, and the horizontal axis represents the position along the scanning direction.

The beam profile LB _n by the n-th pulse irradiation indicated by a solid line shows a beam profile LB n _{+ 1} by the next (n + 1) th pulse irradiation by a dotted line. Here, the irradiation area and the beam profile LB n _{+ 1} of the irradiation region beam profile LB _n are partially overlapping, to both of the shift width (scanning pitch) and L. The maximum value of the energy density of these beam profiles is set so that, for example, 90% or more of the impurity amount can be activated so that the impurities introduced by the ion implantation can be activated.

  Next, the area | region 21 with a short lifetime is demonstrated. The width of the beam profile at which the energy density of the laser beam is 90% of the maximum value (peak value) is referred to as 90% peak width and is represented by W (90). Further, the width of the beam profile at which the energy density is 50% of the maximum value is referred to as a half width, and is represented by W (50).

The regions 33 and 34 sandwiched between W (50) and W (90) are regions where minority carrier recombination centers remain at a high density due to halfway heating, and have a shorter lifetime than other regions. It becomes. When the laser scanning, since at least a portion of the 90% peak width and the beam profile LB n _{+ 1} of the 90% peak width of the beam profile LB _n overlap, region 34 will be annealed again, recombination of minority carriers The center almost disappears. As a result, in the pitch of the shift width L, the region 33 having the width Ws = 0.5 × {W (50) −W (90)} is a region having a shorter lifetime than the other regions.

  In this way, as shown in FIGS. 1 and 2, it is possible to realize a semiconductor device in which the short lifetime region 21 and the long lifetime region 22 are alternately formed on the back side of the substrate 10.

  FIG. 4 is a graph showing the relationship between the energy loss of the semiconductor device and the ratio (Ws / L) of the width Ws to the shift width L. The vertical axis represents energy loss (J), and the horizontal axis represents Ws / L.

  As the ratio (Ws / L) of the width of the region 21 with a short lifetime increases, the number of recombination centers per unit area increases. Therefore, conductivity modulation due to hole injection from the p-type collector layer 14 to the n-type buffer layer 13 and the substrate 10 is less likely to occur, the on-resistance between the anode electrode 15 and the cathode electrode 12 is increased, and as a result, the device is turned on. E (steady loss) 41, which is a steady loss during operation, increases. Further, since the number of recombination centers increases, holes which are excess minority carriers existing in the substrate 10 and the n-type buffer layer 13 are easily recombined with electrons, which is a switching loss when switching the device E ( (Switching loss) 42 becomes smaller.

  Conversely, when the ratio of the width of the region 21 with a short lifetime (Ws / L) becomes small, conductivity modulation due to hole injection is likely to occur, the on-resistance of the device also becomes low, and E (steady loss) 41 becomes small. Become. Further, holes existing in the substrate 10 and the n-type buffer layer 13 are difficult to recombine with electrons, and E (switching loss) 42 increases.

  The total energy loss of the device is represented by the sum of the steady loss and the switching loss at the time of ON, that is, E (total loss) = E (steady loss) + E (switching loss).

  The present inventor has actually measured E (steady loss) and E (switching loss) for devices manufactured by changing the 90% peak width W (90), half width W (50), and shift width L of the beam profile. did. The result is the graph of FIG.

  As can be seen from the graph, when the ratio of the width Ws to the shift width L (Ws / L) is about 0.7 or more, E (steady loss) increases rapidly, while Ws / L is about 0.15 or less. It turns out that E (switching loss) increases rapidly. Accordingly, Ws / L is preferably in the range of 0.15 to 0.7 in order to comprehensively reduce steady-state loss and switching loss during on-state, and in order to more reliably reduce total energy loss. , Ws / L is more preferably in the range of 0.2 to 0.6.

It is sectional drawing which shows an example of the semiconductor device which can apply this invention. FIG. 3 is a plan view seen from the back side of the substrate 10 without the anode electrode 15. It is a graph which shows an example of the beam profile at the time of laser irradiation. It is a graph which shows the relationship between the energy loss of a semiconductor device, and ratio (Ws / L) of width Ws and shift width L.

Explanation of symbols

3 n-type charge storage layer, 4 p-type base layer, 5 gate insulating film, 6 gate electrode,
7 n-type emitter layer, 8 interlayer oxide film, 10 substrate, 11 transistor,
12 cathode electrode, 13 n-type buffer layer, 14 p-type collector layer,
15 anode electrode, 21 short lifetime region,
22 An area with a long lifetime.

Claims (6)

A semiconductor including a step of performing laser irradiation toward a semiconductor layer, reducing lattice defects in the semiconductor layer in the laser irradiation region, and controlling the lifetime of minority carriers in the region where the lattice defect is reduced to be increased. A device manufacturing method comprising:
By scanning the laser beam at a predetermined scanning pitch, the laser irradiation is performed so that the laser irradiation regions partially overlap, and the width of the beam profile at which the energy density of the laser beam becomes the first ratio of the maximum value, and the first Of the two regions (33, 34) sandwiched between the beam profile widths having the second ratio smaller than the ratio, a region having a short lifetime is formed in the region (33) behind the scan, and the regions other than the region (33) are formed. A method for manufacturing a semiconductor device, wherein a region having a long lifetime is formed in the region. A step of injecting impurities into the semiconductor layer before the laser irradiation step;
2. The method of manufacturing a semiconductor device according to claim 1, wherein in the laser irradiation step, lattice defects caused by impurity implantation are reduced by laser irradiation. The width of the beam profile having the first ratio of 90% is W (90), the width of the beam profile having the second ratio of 50% is W (50), and the shift between the two overlapping laser irradiation regions is performed. Let the width be L,
Ws = 0.5 × {W (50) −W (90)},
0.2 ≦ Ws / L ≦ 0.6
The method of manufacturing a semiconductor device according to claim 2, wherein: 4. The method of manufacturing a semiconductor device according to claim 3, wherein the laser irradiation is performed so that the width W (90) of the beam profile at which the energy density of the laser beam is 90% of the maximum value overlap.   The method for manufacturing a semiconductor device according to claim 1, wherein a region having a long lifetime and a region having a short lifetime are formed in a stripe shape.   6. The method of manufacturing a semiconductor device according to claim 1, wherein laser irradiation is performed using a pulsed laser light source.

Images