DETAILED DESCRIPTION OF THE INVENTION

[0045] Various embodiments of the present invention will be described with reference to the accompanying drawings. It is to be noted that the same or similar reference numerals are applied to the same or similar parts and elements throughout the drawings, and the description of the same or similar parts and elements will be omitted or simplified.

[0046] First Embodiment

[0047] As shown in FIG. 4 , an electron beam exposure apparatus according to a first embodiment of the present invention performs exposure while variably controlling a size of an electron beam EB by use of first and second shaping apertures 105 and 108 . The electron beam EB radiated from an electron gun 101 is adjusted on its current density and Koehler illumination condition by first and second condenser lenses 103 and 104 and evenly illuminates the first shaping aperture 105 . An image of this first shaping aperture 105 is formed on the second shaping aperture 108 by first and second projection lenses 106 and 107 . In the second shaping aperture 108 , a plurality of openings for shaping the electron beam EB are provided. In accordance with the size defined in processing pattern data, the electron beam EB irradiates to a position passing through a portion of the openings.

[0048] An irradiation position of the electron beam EB is controlled in such a manner that a shaping deflection system deflects the electron beam EB and thus the beam irradiation position on the second shaping aperture 108 is controlled. The shaping deflection system includes: a shaping deflector 109 ; a shaping deflection amplifier 120 ; and a pattern data decoder 119 for sending deflection data to the shaping deflection amplifier 120 .

[0049] The electron beam EB passing through the second shaping aperture 108 is reduced and projected by a reducing lens 110 and an objective lens 111 and an image thereof is formed on a semiconductor substrate 112 . Accordingly, the irradiation position of the electron beam EB is set on the semiconductor substrate 112 by an objective deflector 113 . The objective deflector 113 is controlled by an objective deflection amplifier 121 for applying a voltage to the objective deflector 113 based on position data sent from the pattern data decoder 119 .

[0050] The semiconductor substrate 112 is provided on a movable stage 116 together with a Faraday cup 114 and a marking table 115 for electron beam measurement. By moving the movable stage 116 , the semiconductor substrate 112 , the Faraday cup 114 or the marking table 115 can be selected.

[0051] In the case of moving the position of the electron beam EB on the semiconductor substrate 112 , so as to avoid exposure of an unnecessary portion on the semiconductor substrate 112 , the electron beam EB is deflected by a blanking electrode 130 and is cut by a blanking aperture 131 . Thus, the electron beam ED is prevented from reaching the semiconductor substrate 112 . A deflecting voltage applied to the blanking electrode 130 is controlled by the blanking amplifier 122 based on the position data sent from the pattern data decoder 119 . Between the pattern data decoder 119 and the blanking amplifier 122 , a proximity correction module 124 for correcting a proximity effect is provided. Delineating control data such as processing pattern data and underlying pattern data are stored in a pattern data memory 118 .

[0052] A proximity correction method according to the first embodiment of the present invention will be described by use of FIGS. 5 to 11 . Herein, the case where an underlying pattern of a partially different material is formed in an exposure region is taken as an example. As shown in FIG. 5 , the proximity correction module 124 includes an area density calculation unit 140 , an area density map memory 141 and a dose correction calculation unit 142 . The proximity correction method will be described below in accordance with a flowchart of FIG. 6 .

[0053] (a) First, in Step S 50 , the pattern data decoder 119 reads stored processing pattern data and underlying pattern data from the pattern data memory 118 and outputs the data to the area density calculation unit 140 of the proximity correction module 124 by breaking the data down by a unit region into pattern dimensions and coordinates. For example, as shown in FIG. 7, a processing pattern 30 is positioned in the center of a unit region and is a rectangle long in a horizontal direction of the page. In addition, as shown in FIG. 8 , underlying patterns 32 a to 32 i are squares disposed in a 3×3 lattice manner at equal intervals.

[0054] (b) In Step S 51 , the area density calculation unit 140 obtains a pattern area for each unit region from the pattern dimension and coordinates of the processing pattern 30 , which are obtained from the pattern data decoder 119 . Thus, the area density calculation unit 140 calculates and creates a pattern area density map of the processing pattern 30 .

[0055] (c) In Step S 52 , the area density map memory 141 of the proximity correction module 124 obtains the pattern area density map of the processing pattern 30 by the unit region from the area density calculation unit 140 and stores the coordinates of the unit region as an index.

[0056] (d) Next, in Step S 53 , the area density calculation unit 140 extracts overlap processing patterns which overlap the underlying patterns 32 a to 32 i. As shown in FIG. 9 , among the underlying patterns 32 a to 32 i, the processing pattern 30 overlaps the underlying patterns 32 d to 32 f. Thus, as shown in FIG. 10 , the underlying patterns 32 d to 32 f are set as overlap processing patterns (first patterns) 33 d to 33 f. A region of the processing pattern 30 , which does not overlap the underlying patterns 32 d to 32 f, is separated as a separate processing pattern (a second pattern) 31 , as shown in FIG. 11 .

[0057] (e) In Step S 54 , the area density calculation unit 140 obtains a pattern area of the extracted first patterns 33 d to 33 f from the pattern dimensions and the coordinates, and calculates a pattern area density of the first patterns 33 d to 33 f. The processing is performed in each unit region and thus pattern area density maps of the first patterns 33 d to 33 f and the second pattern 31 , are created.

[0058] (f) In Step S 55 , the area density map memory 141 obtains the pattern area density map of the first patterns 33 d to 33 f and the second pattern 31 from the area density calculation unit 140 , and stores the coordinates of the unit region as an index.

[0059] (g) In Step S 56 , the dose correction calculation unit 142 of the proximity correction module 124 reads the respective pattern area density maps of the first patterns 33 d to 33 f and the second pattern 31 from the area density map memory 141 by use of the coordinates of the unit region as the index, performs a correction calculation and thus calculates a corrected dose D. Thereafter, the dose correction calculation unit 142 calculates a corrected irradiation time based on the calculated corrected dose and outputs a blanking control signal.

[0060] (h) In Step S 57 , the blanking control signal outputted from the dose correction calculation unit 142 is subjected to digital-analog conversion (DAC) by the blanking amplifier 122 and is converted into a voltage applied to the blanking electrode 130 . The electron beam EB is deflected by the voltage applied to the blanking electrode 130 and is cut by the blanking aperture 131 . Thus, an irradiation time of the electron beam EB reaching the semiconductor substrate 112 is controlled. As a result, an exposure of the electron beam is controlled.

[0061] Note that the pattern area density map creation processing described in (a) to (f) may be performed prior to the delineating processing of the electron beam exposure.

[0062] In the proximity correction method according to the first embodiment of the present invention, in the underlying patterns 32 a to 32 i, the underlying patterns 32 d to 32 f which overlap the processing pattern 30 are extracted to be the first patterns 33 d to 33 f. The pattern area density α ₁ of the first patterns 33 d to 33 f is calculated and the corrected dose D is obtained by the following equation.

D=C/{ ½+η*[α+(η ₁ /η−1)*α ₁ ]} (5)

[0063] Here, α is a pattern area density of a processing pattern and C is a constant. Moreover, as to a ratio η and η ₁ of a backscattered energy of a substrate and of an underlying pattern material to an irradiation energy respectively, a value previously obtained empirically or by calculation for each underlying pattern is used.

[0064] As shown in FIGS. 10 and 11 , the electron beam exposure is performed for the two kinds of processing patterns, which include: the first patterns 33 d to 33 f in which the underlying patterns 32 d to 32 f exist; and the second pattern 31 without the underlying pattern. The dose correction equation (5) is provided in consideration of a difference between these two kinds of processing patterns. Specifically, the dose correction equation (5) can be rewritten as below.

D/ 2+η*(α−α ₁ )* D+η ₁ *α ₁ *D=C (6)

[0065] The first term in the left side of the equation (6) means that, when a proximity effect is not considered, a half of the irradiation energy is set to an exposure threshold C. In the second term in the left side, the pattern area density α ₁ of the first patterns 33 d to 33 f is subtracted from the pattern area density a of the processing pattern 30 . In other words, the second term expresses the backscattered energy generated by subjecting the second pattern 31 to the electron beam exposure. The third term expresses the backscattered energy generated by subjecting the first patterns 33 d to 33 f to the electron beam exposure. Specifically, the dose correction equation (5) or (6) is an equation for obtaining the corrected dose D which sets the sum of the energies expressed in the first to the third terms in the equation (6) to a constant value, that is, the exposure threshold C. Therefore, in the case of delineating by use of the dose D corrected by the dose correction equation (5), energies accumulated in a resist can be kept constant without depending on the pattern area densities of the processing pattern and the underlying pattern. As a result, a desired resist size is obtained.

[0066] By extracting the underlying pattern directly below the processing pattern prior to the correction calculation, it is possible to provide the accurate correction calculation at a position irradiated by the electron beam EB with a difference in a structure of the underlying pattern. The extraction processing can be easily executed in a short time by a logical production “AND” processing and a logical subtraction “MASK” processing in the graphic logic operation. Moreover, in the correction calculation equation, the pattern area density α of the processing pattern and the pattern area density α ₁ of the first patterns are expressed by a linear connection. Consequently, the overlap of the backscattered energies can be accurately expressed and thus the proximity correction accuracy can be dramatically improved compared to the earlier case.

[0067] Note that more accurate proximity correction may be possible when, in Steps S 51 and S 54 , the area density map is smoothed by use of a unit region determined by a backscattered distance β _b value corresponding to each structure of the underlying pattern. For example, as the unit region, a half of the backscattered distance β _b value or less is desirable. In the case of performing exposure at an accelerating voltage 50 kV, the backscattered distance β _b of an Si substrate, a silicon oxide (SiO ₂ ) film, an aluminum (Al) metal film or the like is about 10 μm and thus a unit region of 5 μm square or less may be suitable. On the other hand, the backscattered distance β _b of a heavy metal film such as tungsten (W) is about 5 μm and thus a unit region of 2.5 μm square or less may be suitable.

[0068] Next, description will be given of a semiconductor device manufacturing process by the electron beam exposure using the proximity correction method according to the first embodiment of the present invention.

[0069] (a) As shown in FIG. 12 , on a surface of a semiconductor substrate 1 , an interlayer insulating film 2 of a SiO ₂ film as an underlying layer is deposited by a chemical vapor deposition (CVD) method. In a part of the interlayer insulating film 2 , via holes are provided by a reactive ion etching (RIE) method and the like. In the via holes, plugs 3 made of W as underlying patterns are buried by a sputtering method and the like. After planarizing a surface of the interlayer insulating film 2 by chemical-mechanical polishing (CMP), a conducting film 4 made of Al is formed thereon by sputtering, vacuum deposition or the like. On a surface of the conducting film 4 , a resist film 5 is coated using a spinner and the like. Then, the semiconductor substrate 1 is loaded on the movable stage 116 of the electron beam exposure apparatus.

[0070] (b) Based on the processing pattern data and underlying pattern data which are read by the pattern data decoder 119 from the pattern data memory 118 , the proximity correction module 124 obtains the respective pattern area densities α and α ₁ . Here, the underlying pattern is the plug 3 . Therefore, the unit region of the exposure area including the plugs 3 , which is the first pattern, is 2.5 μm square and the unit region of the area not including the plug 3 , which is the second pattern, is 5 μm square. Compared with the case of calculating the entire region as the unit region of a 2.5 μm square, computational complexity is reduced and thus throughput is improved. Moreover, as to the ratios η and η ₁ of the backscattered energy to the irradiation energy, values of the Si substrate and W, which are empirically obtained for each underlying pattern in advance, are used. By the proximity correction using these values, the electron beam exposure is performed. Thereafter, by development processing, as shown in FIG. 13 , first and second resist patterns 6 a and 6 b as processing patterns, are formed, respectively, in a region not including the plugs 3 and in a region including the plugs 3 in a lower part of the processing layer.

[0071] (c) By using the first and second resist patterns 6 a and 6 b as a mask, selective etching of the conducting film 4 is performed by the RIE method and the like. Thus, as shown in FIG. 14 , wiring layers 7 a and 7 b are formed.

[0072] As described above, according to the electron beam exposure method using the proximity correction method of the first embodiment of the present invention, the proximity correction can be performed in accordance with the underlying patterns of the processing pattern region. Thus, the resist pattern can be formed uniformly with good reproducibility.

[0073] Second Embodiment

[0074] A proximity correction method according to a second embodiment of the present invention is applied to the case where a processing pattern has first and second underlying patterns and is characterized in a proximity correction equation expressed by a linear connection between three kinds of pattern area densities. Points other than the above are similar to those of the first embodiment and thus repetitive description will be omitted.

[0075] The proximity correction method according to the second embodiment of the present invention will be described with reference to FIGS. 15 to 22 .

[0076] (a) First, in Step S 70 of FIG. 15 , the pattern data decoder 119 reads stored processing pattern data and first and second underlying pattern data from the pattern data memory 118 and outputs the data to the area density calculation unit 140 of the proximity correction module 124 by breaking the data down into pattern dimensions and coordinates by unit region. For example, as shown in FIG. 16, a processing pattern 40 is positioned in the center of a unit region and is a rectangle long in a horizontal direction of a page space. As shown in FIG. 17 , the first underlying patterns 42 a and 42 b are squares and positioned in the center of a vertical direction of the page space, respectively, at left and right sides of a unit region. Moreover, as shown in FIG. 18 , the second underlying patterns 44 a and 44 b are rectangles long in the vertical direction of the page space at left and right sides of a unit region.

[0077] (b) In Step S 71 , the area density calculation unit 140 obtains a pattern area for each unit region from the pattern dimension and coordinates of the processing pattern 40 , which are obtained from the pattern data decoder 119 . Thus, the area density calculation unit 140 calculates and creates a pattern area density map of the processing pattern 40 . In Step S 72 , the area density map memory 141 of the proximity correction module 124 obtains a pattern area density of the processing pattern 40 from the area density calculation unit 140 . Thereafter, the area density map memory 141 obtains the pattern area density map of the processing pattern 40 for each unit region and stores the coordinates of the unit region as an index.

[0078] (c) Next, in Step S 73 , as shown in FIG. 20 , the area density calculation unit 140 extracts first overlap processing patterns (first patterns) 43 a and 43 b in which the processing pattern 40 and the first underlying patterns 42 a and 42 b overlap with each other. As shown in FIG. 19 , th first underlying patterns 42 a and 42 b also overlap with the second underlying patterns 44 a and 44 b. In Step S 74 , the area density calculation unit 140 obtains a pattern area for each unit region from a pattern dimension and coordinates of the extracted first patterns 43 a and 43 b and calculates and creates a pattern area density map. In Step S 75 , the area density map memory 141 obtains the pattern area density map of the first patterns 43 a and 43 b for each unit region from the area density calculation unit 140 and stores the coordinates of the unit region as an index.

[0079] (d) Similarly, in Step S 76 , as shown in FIG. 21 , the area density calculation unit 140 extracts second overlap processing patterns (another first patterns) 45 a and 45 b which are the overlapped area of the processing pattern 40 and the second underlying patterns 44 a and 44 b with each other from which regions of the first patterns 43 a and 43 b are removed. In Step S 77 , the area density calculation unit 140 obtains a pattern area for each unit region from a pattern dimension and coordinates of the extracted another first patterns 45 a and 45 b and calculates and creates a pattern area density map. In Step S 78 , the area density map memory 141 obtains a pattern area density of the another first patterns 45 a and 45 b from the area density calculation unit 140 . The area density map memory 141 obtains the pattern area density map of the another first patterns 45 a and 45 b for each unit region and stores the coordinates of the unit region as an index. A region which overlaps with none of the first and second underlying patterns 42 a, 42 b, 44 a and 44 b is separated as a separate processing pattern (a second pattern) 41 , as shown in FIG. 22 .

[0080] ( ) In Step S 79 , the dose correction calculation unit 142 reads the respective pattern, area density maps of the second pattern 41 , the first patterns 43 a and 43 b and the another first patterns 45 a and 45 b from the area density map memory 141 by use of the coordinates of the unit region as the index, performs correction calculations thereof and thus calculates a corrected dose D. Thereafter, the dose correction calculation unit 142 calculates a corrected irradiation time from the calculated corrected dose and outputs a blanking control signal.

[0081] (f) In Step S 80 , the blanking control signal outputted from the dose correction calculation unit 142 is subjected to DAC by the blanking amplifier 122 , and is converted into a voltage applied to the blanking electrode 130 . The electron beam EB is deflected by the voltage applied to the blanking electrode 130 and is cut by the blanking aperture 131 . Thus, an irradiation time of the electron beam EB reaching the semiconductor substrate 112 is controlled. As a result, an exposure of the electron beam is controlled.

[0082] In the proximity correction method according to the second embodiment of the present invention, respective pattern area densities α ₁ and α ₂ are calculated from the extracted first and another first patterns 43 a, 43 b, 45 a and 45 b and thus the corrected dose D is obtained by the following equation.

D=C/{ ½+η*[α+(η ₁ η−1)*α ₁ +(η ₂ /η−1)*α ₂ ]} (7)

[0083] This dose correction equation (7) can be rewritten as below.

D/ 2+η*(α−α ₁ α ₂ )* D+η ₁ *α ₁ *D+η ₂ *α ₂ D=C (8)

[0084] The first term in the left side of the equation (8) means that, when a proximity effect is not considered, a half of the irradiation energy is set to an exposure threshold C. In the second term in the left side, the pattern area density α ₁ of the first patterns 43 a and 43 b and the pattern area density α ₂ of the another first patterns 45 a and 45 b are subtracted from the pattern area density α of the processing pattern 40 . In other words, the second term expresses the backscattered energy generated by subjecting the second pattern 41 caused by the exposure. The third term expresses the backscattered energy generated by subjecting the first patterns 43 a and 43 b caused by the exposure. The fourth term expresses the backscattered energy generated by subjecting the another first patterns 45 a and 45 b caused by the exposure. Specifically, the dose correction equation (7) is an equation for obtaining the corrected dose D which sets the sum of these energies to a constant value, that is, the exposure threshold C.

[0085] Therefore, in the case of delineating by use of the corrected dose D according to the proximity correction method of the second embodiment of the present invention, energies accumulated in a resist can be kept constant without depending on the pattern area densities of the processing pattern and the underlying patterns. As a result, a desired resist size is obtained. The pattern breaking processing can be easily executed in a short time by AND processing and MASK processing of the graphic logic operation. Moreover, in the correction calculation equation (7), the respective pattern area densities α, α ₁ and α ₂ are expressed by a linear connection. Consequently, the overlap of the backscattered energies can be accurately expressed and thus the proximity correction accuracy can be dramatically improved compared to the earlier case.

[0086] Next, description will be given of a semiconductor device manufacturing process by the electron beam exposure using the proximity correction method according to the second embodiment of the present invention.

[0087] (a) As shown in FIG. 23 , on the semiconductor substrate 1 , similarly to the manufacturing process shown in FIG. 12 through FIG. 14 , on a surface of first plugs 13 and a first interlayer insulating film 12 in which the first plugs 13 are buried, a first wiring layer 14 is deposited. Then, a second interlayer insulating film 15 in which the second plugs 16 are buried, are formed on the first wiring layer 14 . Furthermore, a conducting film 17 is formed on a surface of the second plugs 16 and the second interlayer insulating film 15 . Thereafter, a resist film 18 is coated on a surface of the conducting film 17 . Consequently, the semiconductor substrate 1 is loaded on the movable stage 116 of the electron beam exposure device. Here, as schematically shown in FIG. 23 , there exist: an area which does not have any plugs as underlying patterns (left side); and an area which has first underlying patterns including the first plugs 13 and the first wiring layer 14 (center); an area which has second underlying patterns including the first and second plugs 13 and 16 and the first wiring layer 14 therebetween (right side). From FIG. 20 through FIG. 22 in sequence, the respective areas correspond to the second pattern, the first and another first patterns.

[0088] (b) Next, based on the processing pattern data and underlying pattern data which are read by the pattern data decoder 119 from the pattern data memory 118 , the proximity correction module 124 obtains the pattern area densities α, α ₁ and α ₂ of the respective pattern regions. Here, the first and second underlying patterns include the first and second plugs 13 and 16 made of W and the like. Therefore, the unit region of the exposure region including the first and another first patterns is 2.5 μm square and the unit region of the second pattern is 5 μm square. Compared with the case of calculating the entire region as the unit region of 2.5 μm square, computational complexity is reduced and thus throughput is improved. Moreover, as to the respective ratios η, η ₁ and η ₂ of the backscattered energy to the irradiation energy corresponding to each area, values empirically obtained for each underlying pattern in advance are used. By the proximity correction using these values, electron beam exposure is performed. Thereafter, by development processing, as shown in FIG. 24 , first, second and third resist patterns 19 a, 19 b and 19 c as processing patterns are formed.

[0089] (c) By using the first, second and third resist patterns 19 a, 19 b and 19 c as a mask, selective etching of the conducting film 17 is performed by the RIE method and the like. Thus, as shown in FIG. 25 , second wiring layers 20 a to 20 c are formed.

[0090] As described above, according to the electron beam exposure method using the proximity correction method of the second embodiment of the present invention, the proximity correction can be performed in accordance with the underlying patterns of the processing pattern areas. Thus, the resist pattern can be formed uniformly with good reproducibility.

[0091] Other Embodiments

[0092] In the first and second embodiments of the present invention, one or two kinds of structures are described as the underlying pattern. However, needless to say, even in the case of further including more kinds of the underlying patterns, the proximity correction method is applicable similarly by extracting a plurality of overlap processing patterns and calculating respective pattern area densities. A proximity correction equation in this case is obtained as below. 1 $\begin{array}{cc}D=\frac{C}{\left\{\frac{1}{2}+\eta \left[\alpha +\sum _{k=1}^n\left(\frac{{\eta }_k}{\eta }-1\right){\alpha }_k\right]\right\}}& \left(9\right)\end{array}$

[0093] Here, η _k is a ratio of the backscattered energy to the irradiation energy with respect to the material of the k-th (k=1 to n) underlying pattern. In addition, α _k is a pattern area density of the k-th overlap processing pattern. Moreover, in the foregoing description, the underlying patterns include the plugs made of W. However, the backscattered distance β _b is approximately in inverse proportion to an atomic mass of an underlying material. Thus, needless to say, the above-described proximity correction method is also applicable to the case of using a heavy metal such as a refractory metal or a compound thereof. Moreover, the backscattered distance β _b changes when an underlying layer differs in a thickness, a depth of a position from the surface or the like. Thus, needless to say, even when such a structural parameter differs, the proximity correction according to the present invention is effective. Moreover, the extraction of the overlap processing patterns can be processed easily in a short time by use of a graphic logic operation processing of a computer-aided design (CAD) software. Moreover, even in the underlying patterns varying in materials, film thicknesses, positions of the inter layer or the like, when an energy distribution of backscattered electrons can be assumed as being the same, the plural underlying patterns are classified as the same structure. Thus, the extraction processing of the overlap processing patterns can be simplified and the time required for graphic processing can be shortened.

[0094] Various modifications will become possible for those skilled in the art after receiving the teachings of the present disclosure without departing from the scope thereof.