JPH05182887A - Method and apparatus for inspection of pattern - Google Patents

Abstract

PURPOSE:To provide an outward-appearance inspection method and its apparatus wherein an inspection operation can be performed by discriminating a detected defect from a foreign body. CONSTITUTION:A pattern inspection apparatus which detects the defect of a pattern to be inspected is provided with a defect detection means 60, a defect- information storage means and a foreign-body detection means 62. In the apparatus, after a defect has been detected, a foreign body is detected in a position where the defect has been detected, and the foreign body is sorted from others. Thereby, the detected defect can be inspected while it is discriminated from the foreign body, and more appropriate measures can be taken in order to prevent the defect from being caused.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a visual inspection apparatus for detecting defects in a pattern to be inspected, and more particularly to a pattern detection and comparison method and apparatus suitable for visual inspection of patterns such as semiconductor wafers and liquid crystal displays.

[0002]

2. Description of the Related Art Conventionally, an inspection device of this kind has been disclosed in Japanese Patent Laid-Open No. 59-59.
As described in No. 192943, while moving the pattern to be inspected at a constant speed, an image of the pattern to be inspected is detected by an image sensor such as a line sensor, and the position of the detected image signal and the image signal delayed for a certain period of time. The misalignment is recognized as a defect by correcting the deviation at predetermined time intervals and comparing them. As a line sensor,
One-dimensional CCD line sensor and recently a time delay integration type
A (Time Delay Integration) CCD image sensor is used to detect an image of a target pattern through an objective lens having a relatively high magnification. The TDI image sensor has a structure in which a plurality of one-dimensional image sensors are two-dimensionally arrayed, and the output of each one-dimensional image sensor is delayed by a predetermined time and adjacent ones of the same position of the target imaged.
By adding the output of the 3D image sensor,
This is intended to increase the amount of detected light.

[0003]

However, with such a configuration, it is not possible to determine whether the defect detected as a mismatch is a foreign substance or a defect other than a foreign substance such as a shape defect. If the detected defect is a foreign substance, appropriate measures can be taken to cope with the occurrence of the foreign substance, such as cleaning the device, but if the judgment cannot be made, appropriate measures cannot be taken.

When the target pattern is a multi-layered pattern, the layers are overlapped with each other and have high steps or irregularities. As a result, the objective lens of high magnification lacks the depth of focus, and the narrow range of image formation on the image sensor. Only the pattern is detected as an image, and most of the others are blurred. Therefore, according to the conventional method, only a part of the pattern that is in focus can be accurately inspected, and most of the other patterns that are not in focus have low defect detection sensitivity, and a highly reliable inspection cannot be realized. It was

An object of the present invention is to provide a visual inspection method and apparatus capable of inspecting a detected defect by identifying it as a foreign substance.

Another object of the present invention is to perform a highly reliable inspection for a single-layer or multi-layer pattern, so that even a pattern with a high level difference can be accurately focused, highly accurate image detection, and high sensitivity. An object of the present invention is to provide a visual inspection method and apparatus that realize various comparisons.

[0007]

Therefore, the present invention has achieved the above object by realizing the following idea.

After the defect is detected, the detected defect position is inspected by the foreign matter detecting means. Based on this detection result, it is identified whether the defect is a foreign matter.

Defect detection has a structure in which a plurality of one-dimensional image sensors are arranged two-dimensionally, and the output of each one-dimensional image sensor is delayed for a predetermined time, and adjacent one-dimensional images of the same position of an object are picked up. An image sensor called a TDI image sensor having a function of adding the output of the image sensor is used. The image sensor is obliquely arranged, a pattern is formed on the image sensor by a confocal imaging optical system, and this output signal is compared with a reference signal.

As described above and / or, a highly accurate image is detected to detect a defect, and it is discriminated whether or not it is a foreign matter at the detected position.

[0011]

According to the above-mentioned means, not only the defect detection but also the foreign matter detection is performed, so that only the foreign matter can be extracted and detected.
Therefore, it can be determined whether the defect is a foreign matter. Further, since the foreign matter detection is performed only at the detected defect position, the detection conditions and the like can be optimized and only the foreign matter can be detected with higher accuracy than in the conventional foreign matter inspection performed on the entire surface of the sample. Therefore,
It is also possible to more reliably determine whether or not a foreign object.

According to the above, each one-dimensional image sensor inside the TDI image sensor forms an image at a slightly different position (Z position) in the direction perpendicular to the optical axis. Therefore, even if the object has a high step or unevenness, the object is imaged on any one-dimensional image sensor surface, and as a result, a sharp signal output with a large amplitude can be obtained. The presence or absence of a defect is reflected in the value of any one-dimensional image sensor output. Due to the signal addition function of the TDI image sensor, this clear large-amplitude signal and other blurred small-amplitude signals are added. Although these added signals are compared, any one-dimensional image sensor output signal that has captured the defect has a large amplitude without blurring and has a large difference from the normal part. It is possible to detect the presence. As a result of these, even in a multilayer pattern in which each layer is overlapped to form irregularities, highly accurate pattern detection focused on each layer and high sensitivity comparison can be performed in a short time. Therefore, it is possible to dramatically improve the defect detection performance as compared with the related art.

[0013]

EXAMPLES Examples of the present invention will be described below. FIG. 10 shows a pattern inspection apparatus showing one embodiment of the present invention. Figure 10
In the above, the LSI wafer 1, which is the pattern to be inspected, can be detected by the image sensors 5 and 54 via the objective lens 4, and the XY table 6 allows the LSI wafer 1 to be in a direction orthogonal to the scanning of the image sensors 5 and 54, that is, By moving in the X direction, the pattern to be inspected can be detected as a two-dimensional image. Note that L
The SI wafer 1 is illuminated by the illumination lamp 3.
The output image signal of the image sensor 5 is input to the defect detection means 60, which will be described in detail later, and the defect on the wafer is detected. The output of the defect detection means 60 is defect information such as a defect position. When a defect is detected, the XY table 6 is positioned at the defect position based on the defect position information.

At the defect position, foreign matter detection means 62, which will be described in detail later, detects foreign matter at the detected defect position. When the foreign matter is detected, it is judged that the previously detected defect is the foreign matter. If no foreign matter is detected, it is determined that the previously detected defect is a shape defect or discoloration. In the above example, the defect detection and the foreign matter detection are performed using the common XY table 6, but they may be performed separately using different XY tables.

FIG. 11 is a diagram for explaining the configuration and operation of the defect detection of the present invention in detail. In the figure, the S-polarized laser beams 52a and 52b irradiate the wafer 1 with S-polarized laser light at an angle ψ. ψ is about 1 degree. Here, the polarized light that vibrates perpendicularly to the plane formed by the irradiation laser beam and the wafer normal is called S-polarized light, and the polarized light that vibrates in parallel is called P-polarized light. At this time, in the circuit pattern on the wafer having a low step, the scattered light does not change its polarization direction and advances toward the objective lens 4 as it is in the S-polarized light shown in the actual battle. Since the polarization direction of the laser light hitting the light changes, it contains a large amount of P-polarized light component indicated by the dotted line. Therefore, a polarizing plate 53 that blocks S-polarized light is provided behind the objective lens 4, and light passing through the polarizing plate 53 is detected by the image sensor 54 to detect scattered light from a foreign substance and a circuit pattern edge with a high step. This scattered light signal is converted into a digital signal by the A / D converter 55.
The detected signal and the signal at the corresponding portion of the adjacent chip are detected, stored in the image memory 56, and compared with this signal. The alignment circuit 57 aligns these signals, and the difference signal detection circuit 58 detects the difference signal. Since the scattered light signal from the circuit pattern edge having the high step is included in both signals in common, the difference signal includes only the scattered light signal from the foreign matter. By binarizing this difference signal by the binarizing circuit 59, a foreign substance can be detected. Therefore, not only the defect detection but also the foreign matter detection is performed, so that only the foreign matter can be extracted and detected. This makes it possible to determine whether the defect is a foreign substance. Further, since the foreign matter is detected only at the detected defect position, it is possible to optimize the detection condition, for example, the angle ψ and the like, as compared with the conventional foreign matter inspection performed on the entire surface of the sample, and detect only the foreign matter with higher accuracy. it can. In the above example, since the position where the defect is found is compared with the position where the position is found to be normal, it is possible to more reliably determine whether or not there is a foreign substance.

FIG. 1 is a diagram for explaining the configuration and operation of the defect detection of the present invention in detail. In the figure, 5
Is a time delay integration type CCD
This TDI image sensor has a structure in which a plurality of one-dimensional image sensors are arranged two-dimensionally, and the outputs of the respective one-dimensional image sensors are delayed by a predetermined time and the same position of the target is imaged adjacent to each other. The amount of detected light is increased by adding the output of the one-dimensional image sensor. As this kind of TDI image sensor, there are those manufactured by Darusa of Canada and those of Reticon of the United States. The TDI image sensor 5 is arranged at an angle of θ with respect to a plane perpendicular to the optical axis as shown in FIG.
The tilt direction is the longitudinal direction (Y direction) of the multiple one-dimensional image sensors inside with the center of the TDI image sensor as the fulcrum.
The direction is orthogonal to. The amount of tilt is an amount corresponding to the unevenness of the target. The scanning of, for example, the internal one-dimensional image sensor of the TDI image sensor arranged so as to be inclined is made to coincide with the scanning in the Y direction, whereby the LSI wafer 1 as the pattern to be inspected is made one-dimensional through the objective lens 4. Detectable and LSI by XY table 6
By moving the wafer 1 in the direction orthogonal to the main scanning of the TDI image sensor 5, that is, in the X direction, the pattern to be inspected can be detected as a two-dimensional image. X
A linear scale 8 is mounted on the Y table 6,
The actual position of the wafer can be accurately detected. The timing generation circuit 9 generates a start timing signal indicating a pixel from the output signal of the linear scale 8, and the TDI image sensor 5 is driven by this start timing signal every time it moves a certain distance. For example, if the pixel size is 0.1
In the case of 5 μm, a start timing signal is generated every time the wafer moves in the X direction by 0.15 μm to drive the image sensor.

In the above structure, as shown in the positional relationship in FIG. 2, the one-dimensional image sensors 5-1 to 5-m inside the TDI image sensor are located at slightly different positions (Z positions) in the direction perpendicular to the optical axis. The TDI image sensor is arranged so as to be tilted so as to form an image. At this time, the rotary disk 35 having a pinhole and the optical path length conversion element 36 are arranged in front of the TDI image sensor. The rotating disk 35 has a large number of pinholes, for example, 200,000, so that the disk can be rotated at high speed. The diameter of the pinhole is, for example, 20 μm. This pinhole acts as a confocal point. That is, the illumination light that has passed through the pinhole forms an image on the pattern, and the reflected light from the pattern focuses on the pinhole. The high speed rotation of the disk images the pattern at all positions on the TDI image sensor. The optical path length conversion element 36 is provided for each one inside the TDI image sensor.
It provides different optical path lengths to the two-dimensional image sensor. The one-dimensional image sensor 5-k forms an image on the upper surface of the layer A at time Ti by the rotating disk 35 and the optical path length conversion element 36. When the LSI wafer 1 moves in the X direction,
At time Tj, an image is formed on the adjacent one-dimensional image sensor at a position deviated from the upper surface of the layer A in the Z direction, that is, out of focus. In this case, the amount of light detected by the effect of confocal does not focus on the pinhole but is interrupted by the pinhole and becomes small. In this way, the same position (X position) of the object is detected in each one-dimensional image sensor at a position slightly shifted in the Z direction, and the pattern without the focus does not contribute to the detection signal. If the target has high steps or unevenness, either 1
Since the object is imaged on the surface of the one-dimensional image sensor, a sharp signal output having a large amplitude is obtained, and the presence or absence of a defect is reflected in the value of any one-dimensional image sensor output. All of these signals are added by the signal addition function of the TDI image sensor. As shown in FIG. 3 (a), when the target is a pattern of three layers, if there is a pattern defect in the lowermost layer, conventionally, for example, only the intermediate layer is focused, and as shown in FIG. The contrast of the detection signal waveform was large only in the layer pattern, but in the present invention, all three layers show the same contrast. Therefore,
When there is a defect in the lower layer, the contrast is small and the difference from the normal part is not clear as in the prior art as shown in FIG. 3B, but in the present invention, the difference from the normal part is clear as shown in FIG. 3C. Can be

In FIG. 1, the TDI image sensor 5 outputs, for example, eight plural image signals in parallel. These plural output signals are transferred to the wafer 1 by the delay memory 7.
Delay by the time to move the chip. This makes TD
The output signal of the I image sensor 5 and the output signal of the delay memory 7 correspond to the image signals of the adjacent chips 2a and 2b. These multiple output signals are compared in parallel for each channel to detect defects at high speed. Next, the processing content for one channel will be described. Image signal edge detection circuit 11
Input to a and 11b to detect the pattern edge of the pattern to be inspected. The edge detection circuits 11a and 11b are equipped with, for example, a differential operator that detects a dark pattern edge. Then, the detected pattern edges are binarized by the binarization circuits 12a and 12b. Next, 22 is a mismatch detection circuit,
It has a structure as shown in FIG. That is, the mismatch detection circuit
Reference numeral 22 denotes a TD that outputs an image signal from the output of one of the binarization circuits 12a.
Shift register for delaying one scan of I image sensor
26a to 26f and serial-in / parallel-out shift registers 27a to 27g provide 7 × 7 pixel (range is arbitrary) 2
Cut out a three-dimensional local image. The output of the other binarization circuit 12b is the same as the above shift registers 28a to 28c and 2
The output is delayed by 9 and its output is synchronized with the center position of the two-dimensional local image. Then, the shift register
The EXOR circuits 30a to 30n take an exclusive OR between the output of 29 and each bit output of the local image to detect a mismatched pixel, and the counters 31a to 30n calculate the number of the mismatched pixels. The counters 31a to 31n are cleared to zero every N scans of the TDI image sensor 5, and the value is read immediately before that to obtain an area M × N of the product of the number of pixels M of the TDI image sensor 5 and the number of scans N. The inconsistent pixels in the area are identified. Each bit output of the local image is ± 3 in the X and Y directions with respect to the output of the shift register 29.
Since each pixel is shifted in the pixel range, the counters 31a to 31n count the number of mismatched pixels in each shift amount when the ± 3 pixel input pattern is shifted in the X and Y directions. .. Therefore, by checking which of the counters has a characteristic value such as the minimum value or the minimum value for the number of mismatched pixels, it is possible to perform optimum alignment with a small mismatch of images. Of course, it is possible to obtain a counter in which the number of mismatched pixels is smaller than a predetermined value and set the satisfied one as the optimum position.

As described above, the counters 31a to 31n are set to X,
When counting the number of mismatched pixels in each shift amount when the ± 3 pixel input pattern is shifted in the Y direction, the value counted by the minimum value detection circuit 32 is read, and the counter having the minimum value or the minimum value is selected, The TDI image sensor 5 outputs the shift amounts 34a and 34b for scanning in the Y direction and the shift amounts 33a and 33b for scanning in the X direction at right angles thereto.

Reference numerals 23a and 23b denote delay circuits, which have the number M of pixels of the internal one-dimensional image sensor of the TDI image sensor 5 described above.
And the number of scans N of the TDI image sensor 5 required for alignment, which is constituted by the registers in the area M × N corresponding to the product, and delays the outputs from the binarization circuits 12a and 12b. There is.

Reference numeral 24 is an alignment circuit, and as shown in FIG. 5, shift amounts 33a and 33b when the TDI image sensor 5 is scanned in the X direction are input to the selection circuit 35 from the minimum value detection circuit 32. Then, the selection circuit 35 selects an optimum shift position from the outputs of the one delay circuit 23a and the shift registers 36a to 36f which delays by one scan of the TDI image sensor 5 and inputs it to the registers 37a and 37b. To be done. The optimum shift position in the Y direction of the TDI image sensor 5 is selected by the shift amounts 34a and 34b when the selection circuits 38a and 38b scan from the minimum detection circuit 32 in the Y direction of the TDI image sensor 5. .. Therefore, the local images for the shift positions where the mismatch amount is the minimum or the minimum are extracted from the outputs of the selection circuits 38a and 38b. Also from the output of the other delay circuit 23b, the shift register 3
Using the outputs of 9a-39c and 40, the shift register 29
The image is synchronously extracted at a position delayed by the same amount as the output (see FIG. 4). In this state, the local images output from the selection circuits 38a and 38b are at optimal shift positions with no positional deviation with respect to the local images output from the shift register 40.

When the optimum shift position is determined in this manner, the non-coincidence which the defect judgment circuit 25 commonly appears is regarded as a defect. FIG. 6 shows the defect determination circuit 25. The figure shows the case of two shift positions, but the EXOR circuit detects a mismatch between the output of the selection circuit 38a and the output of the shift register 40, and at the same time, the output of the output shift register 40 of the selection circuit 38b. The EXOR circuit 43 detects the disagreement with the above, and the AND circuit 44 calculates the logical product of the outputs of the EXOR circuit 42. Then, the determiner 45 determines whether or not the output signal of the AND circuit 44 is defective according to the size and outputs it.

As described above, the edge detection circuit 11, the binarization circuit 12, the mismatch detection circuit 22, the delay circuit 23, the alignment circuit 24, the defect determination circuit 25 and the like can detect a defect for each channel. Is. In this example,
Although it was processed completely independently for each channel, it can be shared. FIG. 7 shows an example of an inspection device in which alignment is standardized.
In the pattern inspection apparatus shown in FIG. 7, in the channels 2 to 8 in each channel, the mismatch pixel number detection circuit 48 calculates the number of mismatch pixels in each channel by the counter 31 as shown in FIG. At 47, these are summed by the adder 46 as shown in FIG. 8, and the output of the adder 46 is processed by the minimum value detection circuit. In this way, a wider range of images can be handled, and the accuracy of positional deviation correction is improved. This is particularly effective for inspecting a place having a low pattern density. Channels 1, 2, 7, and 8 are configured as described above, and other than that, the delay circuit 23, the alignment circuit 24,
The mismatch detection may be omitted with only the defect determination circuit 25.

In the above embodiment, the binarization circuit 12 is used to
I explained defect detection using a binarized image,
In FIGS. 1 and 7, the binarization circuit 12 may be set to thru to detect a defect by using the grayscale image itself. (Figs. 1, 7
Then it is connected like a dotted line. In this case, the binarization circuit 12
Output is connected only to the mismatch detection circuit 22. ) In this case, a region having a large difference in light and shade is regarded as a defect.
With such a configuration, a signal of a defect different from that of a normal portion can be detected more accurately. Further, in reality, each one-dimensional image sensor causes a slight magnification error (Y position) due to the image formation in the direction orthogonal to the moving direction, but this also cancels out due to the comparison, which causes a problem. I won't. in this way,
The TDI image sensor is intended to increase the amount of detected light, but it is extremely effective in increasing the depth of focus. As a result, even if a multilayer pattern in which each layer is overlapped to form unevenness Highly accurate pattern detection that matches and high sensitivity can be compared. In particular, it is possible to detect fine defects in any layer. Therefore, it is possible to dramatically improve the defect detection performance as compared with the related art.

Although the details have been described above using the embodiments, the individual constituent elements can be realized by the existing technology. The foreign matter detecting means may be a method other than the embodiment as long as the foreign matter can be detected. Further, although the method of inspecting by using two TDI image sensors has been described, the method of comparing and inspecting images simultaneously detected by using two TDI image sensors can also be applied.

[0026]

As described above, since not only the defect detection but also the foreign matter detection is performed, only the foreign matter can be extracted and detected from the defects. Therefore, it can be determined whether the defect is a foreign matter. Further, since the foreign matter detection is performed only at the detected defect position, the detection conditions and the like can be optimized and only the foreign matter can be detected with higher accuracy than in the conventional foreign matter inspection performed on the entire surface of the sample.
Therefore, it becomes possible to more reliably determine whether or not the foreign matter.

Further, even in the case of a multilayer pattern in which each layer is overlapped to form irregularities, highly accurate pattern detection in focus and comparison of high sensitivity can be performed, and a visual inspection method and apparatus in which defect detection performance is dramatically improved as compared with the conventional method. It becomes possible to provide. In particular, it is possible to detect fine defects in any layer.

[Brief description of drawings]

FIG. 1 is a diagram for explaining the configuration and operation of defect detection of the present invention.

FIG. 2 is a diagram showing a confocal imaging relationship by a TDI image sensor.

FIG. 3 is a diagram showing an example of a detection signal waveform.

FIG. 4 is a diagram showing a mismatch detection circuit.

FIG. 5 is a diagram showing an alignment circuit.

FIG. 6 is a diagram showing a defect determination circuit.

FIG. 7 is a diagram showing a pattern inspection apparatus showing another embodiment.

FIG. 8 is a diagram showing a mismatch detection circuit.

FIG. 9 is a diagram showing a mismatch pixel number detection circuit.

FIG. 10 is a diagram showing a pattern inspection apparatus showing an embodiment of the present invention.

FIG. 11 is a diagram for explaining the configuration and operation of foreign matter detection of the present invention.

[Explanation of symbols]

 1 ... Wafer, 2 ... Chip, 3 ... Illumination light, 4 ... Objective lens, 5 ... TDI image sensor, 6 ... Stage, 7 ... Delay memory, 8 ... Linear scale, 9 ... Timing generation circuit, 11 ... Edge detection circuit, 12 ... Binarization circuit, 22 ... Mismatch detection circuit, 23 ... Delay circuit, 24 ... Alignment circuit, 25 ... Defect determination circuit, 26 ... 29 ... Shift register, 30 ... EXOR circuit, 31 ... Counter, 32 ... Minimum value Detection circuit, 35 ... Pinhole, 36 ... Optical path length conversion element, 36, 37, 39, 40 ... Shift register, 45 ... Judgment device, 46 ... Adder, 47 ... Mismatch detection circuit, 48 ... Mismatch detection pixel number detection circuit 52 ... S-polarized laser, 53 ... Polarizing plate, 54 ... Image sensor, 55 ... A / D converter, 56 ... Image memory, 57 ... Positioning circuit, 58 ... Difference signal Detection circuit, 59 ... binarization circuit, 60 ... defect detecting means, 62 ... foreign substance detecting means.

Front Page Continuation (72) Inventor Takashi Hiroi 292 Yoshida-cho, Totsuka-ku, Yokohama-shi, Kanagawa Stock Engineering Co., Ltd.

Claims (6)

[Claims] 1. A pattern inspecting method for detecting a defect in a pattern to be inspected, which comprises detecting a foreign substance at a position detected as a defect and then classifying the defect into a foreign substance and a defect other than the defect. 2. A pattern detector having a structure in which a plurality of one-dimensional image sensors are two-dimensionally arranged, and the outputs of the respective one-dimensional image sensors are delayed by a predetermined time, and the same position of an object is adjacently imaged. An image detecting method, wherein an image sensor having a function of adding the output of the one-dimensional image sensor is obliquely arranged, and a pattern is formed on the image sensor by a confocal image forming optical system. 3. A pattern detector having a structure in which a plurality of one-dimensional image sensors are arranged two-dimensionally, and the outputs of the respective one-dimensional image sensors are delayed by a predetermined time, and the same position of an object is imaged adjacent to each other. The image sensor having a function of adding to the output of the one-dimensional image sensor is obliquely arranged, a pattern is formed on the image sensor by a confocal image forming optical system, and the output signal is compared with a reference signal. Accordingly, the pattern inspection method is characterized in that the mismatch of the images of the pattern to be inspected is detected as a defect. 4. A pattern inspection apparatus for detecting defects in a pattern to be inspected, the defect detecting means and defect information (position etc.).
A pattern inspection apparatus having a storage unit and a foreign matter detection unit, wherein after the defect is detected, the foreign matter is detected at the detected defect position to classify the defect into a foreign matter and other types. 5. A means for detecting an image of a pattern to be inspected, comparing the images and determining a mismatch as a defect, and having a structure in which a plurality of one-dimensional image sensors are two-dimensionally arranged as a pattern detector. Then, an image sensor having a function of delaying the output of each one-dimensional image sensor for a predetermined time and adding it to the output of an adjacent one-dimensional image sensor that images the same position of the target is arranged at an angle and confocal. An image detecting apparatus, wherein a pattern is formed on an image sensor by the image forming optical system. 6. A structure having a structure in which a plurality of one-dimensional image sensors are two-dimensionally arranged as a pattern detector has means for detecting an image of a pattern to be inspected and comparing the images to determine a mismatch as a defect. Then, an image sensor having a function of delaying the output of each one-dimensional image sensor for a predetermined time and adding it to the output of an adjacent one-dimensional image sensor that images the same position of the target is arranged at an angle and confocal. A pattern inspection apparatus characterized in that a pattern is formed on an image sensor by the image forming optical system, and a defect is detected by comparing the output signal with a reference signal.