Patent Application
20230215689
The invention is related to a method of automatic detection of a required peak for sample machining by a focused ion beam. It is particularly related to automatic detection of required milled sample layer of multilayer materials, particularly of multilayer semiconductor chips or printed circuit boards.
Nowadays, focused ion beam (FIB) has been increasingly used for machining of various multilayer materials, particularly in semiconductor industry. The aim of such machining can be, for example, an analysis of defects on a printed circuit board (PCB), production of PCB prototypes, PCB repair, analysis of defects of multilayer semiconductor chips, production of multilayer semiconductor chip prototypes, or repair of multilayer semiconductor chips. In order to allow such machining, it is usually necessary to know the depth of machining, and particularly it is necessary to detect an endpoint at transition between individual layers during machining of complex structures. Various methods of such detection have already been introduced with regards to this requirement, such as a detection of a current received by sample or detection of various signal particles, such as secondary particles, e.g., secondary ions or secondary electrons. A drawback of the secondary particle detection is their wide range of energy values which depends in particular on the material from which they are emitted.
One of patents disclosing a use of secondary particles is U.S. Pat. No. 5,952,658. This patent discloses detecting secondary particles generated by impinging of a beam of charged particles, using the effect of different production of these particles depending on the layer material. Subsequently, based on the detected signal, peaks which correspond to given materials which are impinged by the beam of charged particles in the given moment are determined. Afterwards, based on these peaks, the required endpoint is determined. When detecting a signal, noise is often detected, which hinders the correct identification of peaks in the signal, and therefore it is essential to suppress the noise in a suitable way as much as possible. The patent thus further discloses a method for filtering the detected signal. To suppress the noise, the detected signal is first cumulated and then averaged by means of an averaging filter with a floating window, and then approximated by a polynomial. A drawback of the described signal filtration method is a necessity to modify the said polynomial depending on the amount of signal, which reduces robustness of the whole process. Another drawback is the use of the averaging filter with floating window as the first step of signal processing since the use of this type of filter in this point of signal processing is insufficiently effective and the noise is not effectively suppressed.
Another document that discloses the use of secondary electrons is a patent application JP2000036278. This patent application discloses detection of secondary electrons generated by a focused ion beam impinging the sample. In contrast to the above-mentioned American patent, this application discloses a comparison of detected signal without further modifications thereof with a reference signal. Based on the result of this comparison, a command is then issued to stop the sample machining by the focused ion beam. However, this type of machining is not suitable, since it is necessary to know the reference signal in advance, which is very complicated due to possible ranges of secondary electron energies.
All the above-mentioned documents thus represent solutions which are not robust and their use in an arbitrary case is very complicated or even impossible.
Therefore, it would be desirable to introduce a robust solution that would allow for processing of secondary particles which would not depend on previous knowledge of a reference signal or would not require modifications of filters and would allow for detecting the machining endpoint and significantly reducing noise under any conditions.
The above-mentioned goal is achieved by a method of automatic detection of required peak for sample machining by a focused ion beam by a system comprising an ion column with an ion source arranged for irradiating the sample with a focused ion beam, a working chamber to which the ion column is connected, a detector of secondary particles which is located in the working chamber or in the ion column, a sample holder located in the working chamber and arranged for accommodating the sample, a sample located in the sample holder, and an evaluation unit comprising a memory which stores at least information on the required number of peaks, comprising a first group of steps comprising a step of:
In another variant of the step of transforming the stored discrete values, the stored discrete values are transformed at least to a part with high frequencies, to a part with medium high frequencies, to a part with medium low frequencies, and to a part with low frequencies by performing four-level discrete wavelet transformation of the stored discrete values based on decomposition filters of the mother wavelet. This variant of the step of transforming the stored discrete values helps to achieve the above-mentioned goals by, in case of a noisy signal, allowing to filter away a large amount of noisy signal in comparison to just one-level discrete wavelet transformation. A noisy signal is considered a signal which contains higher number of peaks than the required number of peaks or a signal with ratio noise to signal higher than 15%.
In another variant of the step of resetting the part of transformed discrete values, the parts of transformed discrete values with high frequencies, with medium high frequencies, and with medium low frequencies are reset. This variant of the step of resetting the part of transformed discrete values in connection with the variant of the step of transforming the stored discrete values described in the previous paragraph helps to achieve the above-mentioned goals, since in case of a noisy signal, it filters away a large amount of noisy signal from the signal in comparison to the state when just a part of transformed discrete values with high frequencies is filtered away from the signal.
In another variant of the second group of steps, after the step of creating a filtered signal and before the step of detecting the number of peaks, a step of averaging the filtered signal with the use of a floating window and by averaging the magnitude of values of the filtered signal located in this floating window is further performed by the evaluating unit. Using the step of averaging the filtered signal after the previous steps allows to smooth the already filtered signal even more, thus subsequently facilitating peak detection and achieving the above-mentioned goals. The floating window length corresponds to 3 to 15% of the current number of detected discrete values. However, the maximum floating window length corresponds to the maximum number of 100 discrete values, or equivalent length of a time period based on the sampling frequency.
In another variant of the second group of steps, after the step of detecting the number of peaks and before the step of issuing a command to stop sample machining, a step of skipping close peaks, wherein close peaks are the peaks with the distance from the closest peak smaller than 50% of the average distance value between the individual consecutive peaks, is further performed by the evaluating unit. Using the step of skipping close peaks helps to achieve the above-mentioned goals by removing the peaks which are too close to the preceding and following peaks, and it is therefore likely that this is a false peak not identifying a different layer, but just a measurement error. Such false peak is, for example, a small excess of a subsequent value of the filtered signal during otherwise decreasing trend.
In another variant of the second group of steps, after the step of detecting the number of peaks of the filtered signal and before the step of issuing a command to stop sample machining by a focused ion beam, a step of skipping the last peak is performed by the evaluating unit. Using the step of skipping the last peak helps to reach the above-mentioned goals by removing the last false peak generated due to the borderline phenomenon of the noisy signal. The borderline phenomenon is generated during averaging of the filtered signal with the use of a floating window.
Peaks are local maxima of the filtered signal or local minima of the filtered signal.
According one of the variants, the mother wavelet is Daubechies-4. Using this mother wavelet allows for reliable detection of changes of the gradient of stored discrete values in comparison with other types of mother wavelets, e.g., Haar wavelet.
The subject matter of the invention is described by way of exemplary embodiments thereof, which are described by means of accompanying drawings, in which:
The described embodiments represent exemplary embodiments of the invention, which, however, have no limiting effect in terms of scope of protection.
An exemplary embodiment of the invention is a method of automatic detection of a required peak for sample machining by a focused ion beam by means of a system. The system comprises an ion column with an ion source. The ion column with the ion source is arranged for irradiating the sample by a focused ion beam. The ion column is arranged for irradiating the sample by the focused ion beam in such way that it comprises an ion source, an extractor, a condenser lens, and a deflector. These are located in the ion column such that the extractor is located behind the ion source along the ion column optical axis in the direction of the ion beam propagation. The condenser lens is located behind the extractor along the ion column optical axis in the direction of the ion beam propagation. The deflector is located behind the condenser lens along the ion column optical axis in the direction of the ion beam propagation. Behind the deflector along the ion column optical axis in the direction of the ion beam propagation the opening of the ion column is located, through which the focused ion beam emerges from the ion column. The ion beam is focused during its passage through the condenser lens. The deflector deflects the ion beam in two mutually perpendicular directions perpendicular to the direction of the ion beam propagation. In one of the exemplary embodiments the deflector can be composed of two levels of scanning elements, wherein they are arranged for applying a force field on the ion beam, which is, based on this effect, deflected relative to the ion column optical axis. The deflector can be composed of electromagnetic coils or electrostatic electrodes.
The system further comprises a working chamber to which an ion column is connected. The ion column is connected to the working chamber in such way that the focused ion beam emerging from the opening of the ion column enters the working chamber. The system further comprises a sample holder and a sample. The sample holder is arranged for accommodating the sample. The sample is located in the sample holder. The sample holder is located in the working chamber. In the first exemplary embodiment of the sample holder, the sample holder is arranged for tilting around three mutually perpendicular axes and arranged for movement along three mutually perpendicular axes. In the second exemplary embodiment of the sample holder, the sample holder is arranged for tilting around at least a single axis.
In one of the exemplary embodiments, the system further comprises a gas reservoir and an assembly for supplying gas into the working chamber, connected on one end to the working chamber and on the other end to the gas reservoir. The supplied gas is any gas from the Nanoflat group by TESCAN ORSAY HOLDING a.s., A-Maze by TESCAN ORSAY HOLDING a.s., XeF2 or any other suitable gas supporting acceleration of etching by the focused ion beam, reduction of undesirable doping from the focused ion beam on the sample, reduction of redeposition of the etched material, or reduction of selectiveness for multilayer samples.
The system further comprises a detector of secondary particles. Secondary particles are secondary electrons or secondary ions emitted by the sample after the focused ion beam impinges the sample. In the first exemplary embodiment of the secondary particle detector location, the detector of secondary particles is located in the working chamber. In the second exemplary embodiment of the secondary particle detector location, the detector of secondary particles is located in the ion column. The detector of secondary particles detects the amount of secondary particles emitted from the machined sample area. The detected values of the amount of secondary particles from the detector of secondary particles from the whole machined sample area are, with the use of, for example, an integrated system (also called embedded system) or another computing system, with the sampling frequency in the range of 1 to 3 Hz, averaged such that a single discrete value arises from the whole machined area in regular intervals given by the said sampling frequency. Subsequently, these discrete values are stored in a memory in the form of dependency of magnitude of discrete values on time, based on the sampling frequency or a number of discrete values.
The system further comprises an evaluation unit and a control unit. The evaluation unit comprises a memory and a processor. The evaluation unit and the control unit are any devices from the group of at least a personal computer, a microcomputer, or an integrated system. The evaluation unit is data-connected to the detector of secondary particles and to the control unit. The control unit is arranged for controlling the sample irradiation by the focused ion beam. The control unit is data-connected to the control elements of the ion source, the extractor, the deflector, and the condenser lens. Data connection is an analog or digital connection. The evaluation unit is arranged for issuing a command to stop sample machining by the focused ion beam. The command to stop sample machining by the focused ion beam is sent to the control unit, which, consequently, stops sample machining by the focused ion beam. The memory stores the information on the required number of peaks and the nature of peaks, i.e., whether these peaks should be local maxima, local minima or combinations of both. Local maxima, or local minima, respectively, are considered such signal values which contribute to a change from increasing signal trend to decreasing signal trend, or vice versa.
The method of automatic detection of required peak for sample machining by a focused ion beam comprises a first group of steps and a second group of steps, which are performed simultaneously.
The first group of steps comprises a step of irradiating individual sample spots by the focused ion beam and detecting a quantity of secondary particles emitted from the machined sample area which is impinged by the focused ion beam and storing the discrete values in the memory. The first group of steps further comprises a step of stopping the irradiation of individual machined sample spots by the focused ion beam after the control unit has received a command to stop sample machining by the focused ion beam from the evaluation unit.
In the first exemplary embodiment of the second group of steps, the second group of steps comprises a sequence of steps performed by the evaluation unit comprising steps of: transforming stored discrete values, resetting discrete values, creating filtered signal, detecting number of peaks, issuing command to stop sample machining.
In the second exemplary embodiment of the second group of steps, the second group of steps comprises a sequence of steps performed by the evaluation unit comprising steps of: transforming stored discrete values, resetting part of transformed discrete values, creating filtered signal, averaging of filtered signal, detecting number of peaks, issuing command to stop sample machining.
In the third exemplary embodiment of the second group of steps, the second group of steps comprises a sequence of steps performed by the evaluation unit comprising steps of: transforming stored discrete values, resetting part of transformed discrete values, creating filtered signal, detecting number of peaks, skipping close peaks, issuing command to stop sample machining.
In the fourth exemplary embodiment of the second group of steps, the second group of steps comprises a sequence of steps performed by the evaluation unit comprising steps of: transforming stored discrete values, resetting part of transformed discrete values, creating filtered signal, averaging of filtered signal, detecting number of peaks, skipping close peaks, issuing command to stop sample machining.
In the fifth exemplary embodiment of the second group of steps, the second group of steps comprises a sequence of steps performed by the evaluation unit comprising steps of: transforming stored discrete values, resetting part of transformed discrete values, creating filtered signal, averaging of filtered signal, detecting the number of peaks, skipping close peaks, issuing a command to stop sample machining.
In the sixth exemplary embodiment of the second group of steps the second group of steps comprises a sequence of steps performed by the evaluation unit comprising steps: transforming stored discrete values, resetting part of transformed discrete values, creating a filtered signal, averaging of filtered signal, detecting number of peaks, skipping close peaks, skipping the last peak, issuing command to stop sample machining.
In the first exemplary embodiment of the step of transforming stored discrete values, the stored discrete values are transformed according to frequencies into a part with high frequencies and into a part with the remaining frequencies by performing one-level discrete wavelet transformation of stored discrete values based on decomposition filters of the mother wavelet Daubechies-4. In the second exemplary embodiment of the step of transforming stored discrete values, the stored discrete values are transformed according to frequencies into a part with high frequencies into a part with medium high frequencies, into a part with medium low frequencies, and into a part with low frequencies by performing four-level discrete wavelet transformation of stored discrete values based on decomposition filters of the mother wavelet Daubechies-4.
In the first exemplary embodiment of the step of resetting part of transformed discrete values, the parts of transformed discrete values with high frequencies are reset. In the second exemplary embodiment of the step of resetting part of the transformed discrete values, the parts of the transformed values with high frequencies, with medium high frequencies, and with medium low frequencies are reset. Resetting part of transformed values is always just mere resetting of these values, these values are thus not removed, but the final signal is not affected anymore, in this case in the following step of created filtered signal.
The individual exemplary embodiments of the step of transforming stored discrete values and the step of resetting part of the transformed discrete values can be combined. Particularly, it is possible to combine the first exemplary embodiment of the step of transforming stored discrete values with the first exemplary embodiment of the step of resetting part of the transformed discrete values. Furthermore, it is also possible to combine the second exemplary embodiment of the step of transforming stored discrete values with the second exemplary embodiment of the step of resetting part of transformed discrete values. Furthermore, it is also possible to combine the second exemplary embodiment of the step of transforming stored discrete values with the first exemplary embodiment of the step of resetting part of transformed discrete values.
In the step of creating the filtered signal, the filtered signal is created by performing an inverse discrete wavelet transformation of transformed discrete values based on reconstruction filters of the mother wavelet Daubechies-4. The inputs for individual levels of the inverse discrete wavelet transformation are corresponding outputs of the individual levels of the discrete wavelet transformation.
In the step of detecting the number of peaks, the number of peaks of the filtered signal is detected by the peak nature defined in the memory. The peaks are detected by means of a first derivative approximation. In other words, the peaks are calculated as a difference between two adjacent values of the filtered signal, i.e. d(i)=x(i+1)−x(i), where x(i) stands for the filtered signal value and x(i+1) stands for the following filtered signal value, and if the value d(i) in two consecutive values exceeds zero, in other words, its plus/minus sign changes in relation to the previous value, the value in spot d(i) is marked as a peak.
In the step of issuing a command to stop sample machining, the command to stop sample machining by the focused ion beam is issued after a given number of peaks based on the information of the required number of peaks stored in the memory has been reached.
In the step of averaging the filtered signal, the filtered signal values are averaged with the use of a floating window, wherein the filtered signal values located in this window are averaged. The floating window is a gradually moving section of discrete values of the filtered signal. With growing number of discreet values, the floating window extends so that the maximum length of the floating window is in the range of 3% to 15% of the length of the filtered signal, but no longer than 50 discrete values. Averaging of values in this floating window is based on the formula
where x [ ] are the original values of the filtered signal, y [ ] are averaged values of the filtered signal and Z is the number of discrete values in the floating window.
In the step of skipping close peaks, the skipped peaks are those with distance from the closest peak lower than 50% of the average distance value between consecutive peaks.
In the step of skipping the last peak, the last peak is skipped. The step of skipping the last peak is applied in case of a noisy signal to which the step of averaging of the filtered signal was applied.
| Number | Date | Country | Kind |
|---|---|---|---|
| PV 2020-320 | Jun 2020 | CZ | national |
| Filing Document | Filing Date | Country | Kind |
|---|---|---|---|
| PCT/CZ2021/050058 | 6/2/2021 | WO |
