JPH04194738A - Method for determining haze and continuous foreign object group - Google Patents

Abstract

PURPOSE:To determine a haze and a continuous foreign object group by calculating an average value of haze signal within a unit region which is obtained by classifying a surface of a wafer in square, detecting a signal protruding from this average value, and then comparing it with a reference value. CONSTITUTION:A light-receiving signal Vr which is output from an amplifier 3a of a signal processing part 3 is fed to plus terminal of a comparator 3b. This output foreign object signal Vp is sampled and digitized by a sampling digital circuit 3d-1 and is converted to the signal Vp. Then, it is taken into a microprocessor 3f through an interface 3e, a coordinate value is added to it, and then it is stored in a memory 3f. On the other hand, frequency components of the signal Vr which is fed to a low-pass filter 3c are cut off, it is converted to a digital haze signal Vh by a sampling digital circuit 3d-2, is taken into a microprocessor 3f through an interface 3e, the coordinates value is added to it, and it is stored in a memory 3g.

Description

DETAILED DESCRIPTION OF THE INVENTION [Industrial Application Field] The present invention relates to a method for determining haze (cloudiness) and a continuous group of foreign objects in a foreign object inspection apparatus for wafers and the like.

[Prior Art] When foreign matter adheres to the surface of silicon wafers, which are the raw material for semiconductor ICs, the quality of the product deteriorates! When the surface is polished to a mirror surface, it is inspected by a foreign matter inspection device. However, in addition to foreign matter, the mirror surface has cloudiness (or mist) called haze attached to a rather wide area, and since haze is harmful to the formation of wiring patterns, it is inspected at the same time.

The concept of the inspection device for foreign matter and haze and the outline of the inspection method will be explained with reference to FIGS. 4(a), (b), and (c).

In Figure (a), there is a foreign particle p1'+ on the surface of the wafer 1.
p2 is isolated and attached, and haze b1. -h2 is attached in four distributions in a certain range. However, the foreign substance p2 exists within the range of the haze h2. Also, haze hl+
The concentration of tlz changes depending on the position, so to speak, there is a concentration distribution. In Figure (b), a laser beam is projected and scanned from the light source 2a of the detection optical system 2 onto the surface of the wafer 1, and the light scattered by foreign objects or haze is focused by the condensing lens 2b to form a photoelectric converter 2C. Receives light. Note that on a mirror surface free of foreign matter or haze, the laser beam is specularly reflected and is not received by the photoelectric converter 2c. The output of the photoelectric converter 2C is sent to the amplifier 3 of the signal processing section 3.
The foreign substance pt shown in Figure (a) is amplified to an appropriate level by
A light reception signal vr shown in the upper column of FIG. 3(e) is obtained corresponding to tI)2 and haze htyh2. The light reception signal vr is input to the child terminal of the comparator 3b, and
The signal is input to a low-pass filter 3c, high frequency components are cut off, and low frequency components are added to one terminal of a comparator 3b. Haze ht, signal for h2 (ht
) and (h2) change slowly, so they pass through the low-pass filter 3C, and haze signals vhl and vh2 with peak values corresponding to the shading of the haze are output from the output terminal 3d-2. On the other hand, the pulses (ps) with short time widths (1) and 2) for the foreign objects pl and p2 are passed through the comparator 3b as they are, and the foreign object signal vP1. vP2 is output.

Nao, output capacitor 3b of low-pass filter 3C
Therefore, the foreign object signal vP2 can be obtained by subtracting the height of the haze h2.

[Problem to be Solved] Although the foreign matter 1) t t p2 in the above is an isolated object, it may be continuous and form a group of foreign objects.

It is difficult to detect such a continuous group of foreign substances using the above-mentioned inspection method. This will be explained below using figures.

FIG. 5(a) shows a continuous foreign object group consisting of a plurality of n foreign objects p1'=pn attached to a haze-free mirror surface in contact with or close to each other. The light reception signal vr for this is
As shown in Figure (b), the light reception signals for individual foreign objects overlap and the peak value changes slightly, resulting in a waveform that continues for the length of the foreign object group. The foreign object signal vP with respect to the received light signal vr has a very small peak value as shown in Figure (e), so the S/
Detection is difficult due to poor N ratio. On the other hand, the high frequency component of the received light signal vr is cut by the low-pass filter 3c, and this is output as a haze signal vh'.

Figures (d) to (f) show that the foreign object group p1-1)n has a haze h
When the foreign object signal vP is too weak to be detected, an erroneous haze signal v is placed on top of the correct haze signal vh.
h' is output in a superimposed manner. As a result of the above, in either case, the continuous foreign object group is erroneously detected as haze. Note that when a foreign object is separated by a certain distance or more, it is of course detected as the above-mentioned isolated foreign object.

Now, in contrast to haze, which is distributed over a relatively wide range as described above, the length of a continuous foreign object group is actually very short. Furthermore, even if the foreign object group is long, there is often a break in the middle and the foreign object group is divided into several short continuous foreign object groups.

Therefore, it is considered possible to distinguish between the two by focusing on the fact that the erroneously detected haze signal vh' has a somewhat shorter length and a peak value larger than the original haze signal Vh.

Based on the above concept, it is an object of the present invention to provide a method for discriminating between haze and continuous foreign matter groups in a foreign matter inspection apparatus.

[Means for Solving the Problems] The present invention projects a laser beam onto the surface of a wafer to be inspected, receives scattered light due to foreign matter or haze attached to the surface using a photoelectric converter, and converts the received light signal vr. are respectively input to the child terminal of the comparator and the low-pass filter, and the high-frequency components included in the received light signal vr are cut by the low-pass filter and input to one terminal of the comparator, and the output signal of the comparator is used as the foreign object signal vP for the foreign object, and the low-pass filter This is a method for distinguishing between haze and a continuous foreign object group in a foreign object inspection apparatus in which the output signal of a filter is used as a haze signal vh for haze, and the haze signal vh is sampled at minute intervals Δl to generate a digital haze signal [
Convert to vhl. Through microprocessor processing,
The average value [vhl-] of the haze signal [vhl] in the unit cell S obtained by dividing the surface of the wafer into squares is calculated. The average value [vh of the haze signal [vhl] in each unit cell S and the prominent protruding signal [:VO3] are detected, and the number N of consecutive haze signals is counted. When the number N is smaller than the reference value M, these N protruding signals [VO3
is the foreign object signal for a continuous foreign object group [:v,
p'].

In the above discrimination method, the haze signal vh is sampled and digitized at minute intervals Δnu, and the haze signal [
vhl, and its average value [v h1w+ within each unit area S is calculated by processing by the microphone processor. For each unit area S, haze signal [v, h]
The average value [VHL of the protruding signal [V
O3 is detected and the consecutive number N is counted, and when this is smaller than the reference value M, the N projecting signals [Vtl
] is determined to be a foreign object signal [vp' ] for a continuous foreign object group. In the above, the reason why a rectangular unit area S is set and a salient signal [Vtl] is detected by comparing the haze signal [vhl] with the average value [vh] within the unit cell S is that the haze density varies depending on the location. This is because the haze signal [vhl is also thicker (varies), making it difficult to compare the two.
is set to a range where the change in haze density is relatively small, and by taking the average value [v h]m of that range, the prominent signal [
VO3 is to be detected reliably, and therefore it is necessary to determine one side of the rectangle of the unit area and the reference value M in consideration of the haze density distribution and the maximum length of the continuous foreign object group.

In this case, one side of the rectangle is sufficiently thicker than the reference value M (setting is a necessary condition, otherwise the average value [v h]m will be influenced by the peak value of the foreign object group.

[Embodiment] FIG. 1 shows a schematic block diagram of an embodiment of the haze and continuous foreign object group discrimination method according to the present invention. The detection optical system 2 has the same structure as the conventional structure shown in FIG. 4(b), and a description thereof will be omitted. The light reception signal vr output from the amplifier 3a of the signal processing section 3 is input to the child terminal of the comparator 3b,
The output foreign object signal vp is sampled and digitized by sampling and digitizing circuits (SP/A/D) 3a-t and converted into a foreign object signal [vp], which is then sent to the microprocessor ( MPU
)3r, coordinate values are added, and the memory (MEM
) 3b. On the other hand, Low! Dimension filter (LF
) 3c, the high frequency component of the received light signal vr is cut and converted into a digital haze signal [vhl] by the SP・A/D 3 d-2, which is sent to the M via I/F 3e.
It is taken into the PU 3r, added with coordinate values, and stored in the MEM 3g.

The unit area S set on the wafer and the operation of each part in FIG. 1 will be explained with reference to FIGS. 2(a) to 2(d).

In Figure (a), the orientation flat OF of the wafer 1 is taken as the X axis, and the xy coordinate system is set by the y axis perpendicular to this, and the laser beam of the detection optical system 2 is swept in the X direction, with an interval Δl' The surface is scanned. Considering the haze density distribution and the maximum length of the continuous foreign object group mentioned above,
On the surface of the wafer 1, a rectangular unit area S with lengths of two sides Lx and Ly is set in a matrix. Figure (b) shows one unit area S and a scanning line. The haze signal vh for each scanning line is sampled and digitized at intervals of Δl,
The haze signal shown in Figures (C) and (d) [vhl is ME
Stored in M3g. Figure (c) shows the case where only the haze h adheres within the unit area S, and each haze signal [vhl changes slowly according to the density of the haze h, and the MPU3
Their average value [Vh]a+ is calculated by f and ME
Stored in M3g. In this case, the average value [v hl+
Since the deviation of each haze signal [vhl for * is small,
It is determined that there are no foreign bodies. In contrast, figure (d)
is a prominent signal [vt+] that is larger than the average value [vt+] layer.
If there are V olds, detect this and calculate the number of consecutive pieces N
is counted, and when the number N is smaller than the reference value M, this is determined to be a foreign object signal [vp'] for a continuous foreign object group. The foreign object signal [vp'] is combined with the foreign object signal Cvp for the isolated foreign object described above, edited, and output to the printer (PRT) 3h. However, when the number N is larger than the reference value M, it is not necessarily regarded as a continuous foreign object group, and the haze concentration may change rapidly in that part, and this is output as a suspected foreign object signal [vp#]. In addition,
The haze signal [Vh] is determined regardless of the presence or absence of a continuous foreign object group.
Data of the average value [v h]m is output to PRT3h.

FIG. 3 is a flowchart showing the processing procedure for the above-described continuous foreign object group discrimination method. First, the surface of the wafer 1 is scanned by a laser beam, and the obtained foreign object signal [vP] and haze signal [vh] are These coordinate values are added and recorded in MEM3g, and [[
■ The average value of vhl [Vh]ma is calculated. Then,
Detects the protruding signal [VO3 that protrudes [vhlsr] of [vhl in each unit area S, and continues [V11
]Count the number N.■. When the number N is smaller than the reference value M (yes in ■), the N protruding signals [VO3 are determined to be the foreign object signals [vp' 3] for the continuous foreign object group and are stored in the MEM, and the number N is the reference value. When it is equal to or greater than M (nO of ■), it is determined as a suspected foreign object signal [vp'' and stored as ■. Each of the above foreign object signals [vp].

[vp'] and suspected foreign object signal [vp''] data are edited and sent to PRT3h as foreign object data.
■ is output to. Also, the average value of the haze signal [v h]
■ gm is output as haze data.

In the above embodiment, the scanning method of the laser beam is x
Although y constant scanning was used, in the case of rotational scanning, each sampling data is displayed by (R, θ) coordinates and MEM3g
This is recorded in the XY coordinate system and the above processing is performed. Coordinate transformation can be easily performed using common techniques, so detailed explanation will be omitted.

[Effects of the Invention] As is clear from the explanation in F below, in the method for discriminating haze and continuous foreign matter groups according to the present invention, the surface of the wafer is divided into unit regions S in which the change in haze density is relatively small, and each The haze signal [average value of vhlw [vhlw] within the unit area S is calculated, and when the protruding signal [VO3 that is more prominent than the average value [vh] is detected and the consecutive number N is smaller than the reference value M, it is continuously By identifying the foreign object signal [vp'] for a foreign object group, the continuous foreign object group, which was conventionally erroneously determined to be haze, is correctly identified, and this has a great effect in contributing to improving the reliability of foreign object inspection.

[Brief explanation of the drawing]

FIG. 1 is a schematic block diagram of an embodiment of the method for discriminating between haze and continuous foreign matter groups according to the present invention, and FIG.
), (b), (c) and (d) are unit areas S set on the wafer and explanatory diagrams of the operation of each part in Fig. 1, and Fig. 3 shows the procedure of the method for determining continuous foreign matter groups. The flowchart, FIGS. 4(a), (b), and (c) is a block diagram of an inspection apparatus for foreign matter and haze attached to a wafer, and an explanatory diagram of the concept of the inspection method, and FIGS. 5(a), (b).
), (c), (d), (e), and (f) are explanatory diagrams of the signal waveform of a continuous foreign object group and the reason why the continuous foreign object group is erroneously determined to be haze. 1... Wafer, 2... Detection optical system, 2
a... Light source, 2b... Condensing lens, 2c
...Photoelectric converter, 3...Signal processing section, 3a...
・Amplifier, 3b...Comparator, 3C・
・Low pass filter (LF), 3d-1, 3d-2・
...Sampling/digitization circuit (SP@A/D), 3e...Interface (I/F), 3f...Microprocessor (MEM), 3g...Memory (ME
M), 3h...Printer (PRT), ■~■...Step number of flowchart.

Claims (1)

[Claims] (1) A laser beam is projected onto the surface of the wafer to be inspected, and a photoelectric converter receives the scattered light due to foreign matter or haze attached to the surface, and the received light signal v of the photoelectric converter is
r is input to the + terminal of the comparator and the low-pass filter, and the received light signal v is input by the low-pass filter.
A foreign object in which a high frequency component included in r is cut and inputted to one terminal of the comparator, the output signal of the comparator is used as a foreign object signal vP for the foreign object, and the output signal of the low-pass filter is used as a haze signal vh for the haze. In the inspection device, the haze signal vh is sampled at minute intervals Δl to obtain a digital haze signal [v
h], and the microprocessor calculates the average value [vh]m of the haze signal [vh] within a unit area S obtained by dividing the surface of the wafer into squares, and calculates the average value [vh]m of the haze signal [vh] in each unit area Among the haze signals [vh], prominent signals [VH] that are more prominent than the average value [vh]m are detected and the consecutive number N is counted, and the number N is equal to the reference value M.
A method for discriminating between haze and a continuous foreign object group, wherein the N protrusion signals [VH] are determined as foreign object signals [vp'] for a continuous foreign object group in which the foreign objects are continuous.