Systems and methods for semiconductor chip surface topography metrology
Systems and methods for measuring a surface topography of a semiconductor chip are disclosed. A disclosed system comprises a light source configured to provide low coherent light to a first beam splitter, a scanner configured to use the low coherent light reflected from the first beam splitter to scan positions on a surface of a semiconductor chip, a second beam splitter configured to receive reflected signals from the positions on the surface of the semiconductor chip, a detector configured to detect interference signals from a first output of the second beam splitter, wherein each of the interference signals corresponds to a respective one of the positions, and a spectrometer configured to detect spectrum signals from a second output of the second beam splitter, wherein each of the spectrum signals corresponds to the respective one of the positions.
1 . A system, comprising:
a light source configured to provide low coherent light to a first beam splitter;
a scanner configured to use the low coherent light reflected from the first beam splitter to scan a plurality of positions on a surface of a semiconductor chip;
a second beam splitter configured to receive a plurality of reflected signals from the plurality of positions on the surface of the semiconductor chip;
a detector configured to detect a plurality of interference signals from a first output of the second beam splitter, wherein each of the plurality of interference signals corresponds to a respective one of the plurality of positions;
a spectrometer configured to detect a plurality of spectrum signals from a second output of the second beam splitter, wherein each of the plurality of spectrum signals corresponds to the respective one of the plurality of positions; and
a processor configured to determine, using machine learning, surface topography of the semiconductor chip based at least on the plurality of interference signals and the plurality of spectrum signals,
wherein the processor is further configured to:
classify the plurality of spectrum signals into a plurality of categories using a machine learning model, wherein each of the plurality of categories corresponds to a region having a same material on the surface of the semiconductor chip,
determine a surface height offset between a surface baseline and at least one of the categories based at least on a calibration signal associated with the region corresponding to the at least one of the categories, and
characterize the surface topography of the semiconductor chip based at least on the surface height offset and the interference signals.
2 . The system of claim 1 , wherein:
a spectrum of the light source has a range from about 400 nm to about 900 nm.
3 . The system of claim 2 , wherein the detector comprises:
a white light charge-coupled device or a white light charge complementary metal-oxide-semiconductor image sensor.
4 . The system of claim 1 , wherein:
a spectrum of the light source has a range from about 190 nm to about 1100 nm.
5 . The system of claim 1 , further comprising:
a scanning probe microscopy configured to provide a plurality of calibration signals each associated with a topographic height.
6 . The system of claim 5 , wherein the scanning probe microscopy comprises:
an atomic force microscopy or a scanning tunneling microscope.
7 . The system of claim 1 , further comprising:
an electron microscope configured to provide a plurality of calibration signals each associated with a topographic height.
8 . The system of claim 7 , wherein the electron microscope comprises:
a scanning electron microscope or a transmission electron microscope.
9 . The system of claim 1 , further comprising:
a database configured to store the plurality of interference signals and the plurality of spectrum signals.
10 . A system, comprising:
an interferometer configured to generate a plurality of interference signals each corresponding to a respective one of a plurality of positions on a surface of a semiconductor chip;
a spectrometer configured to generate a plurality of spectrum signals each corresponding to a respective one of the plurality of positions on the surface of the semiconductor chip; and
at least one processor configured to determine, using machine learning, surface topography of the semiconductor chip based at least on the plurality of interference signals and the plurality of spectrum signals,
wherein to determine the surface topography, the at least one processor is further configured to:
classify the plurality of spectrum signals into a plurality of categories using a model including machine learning features, wherein each of the categories corresponds to a region having a same material on the surface of the semiconductor chip,
determine a surface height offset between a surface baseline and at least one of the categories based, at least in part, on at least one of a plurality of calibration signals associated with the region corresponding to the at least one of the categories, and
characterize the surface topography of the semiconductor chip based, at least in part, on the surface height offset and the interference signals.
11 . The system of claim 10 , wherein:
the interferometer comprises a low coherent interferometer, and each of the interference signals comprises an interference fringe of low coherent interferometry; and
the spectrometer comprises a reflectance spectrometer, and each of the plurality of spectrum signals comprises a reflectance spectrum.
12 . The system of claim 10 , wherein each of the interference signals is associated with an original surface height at the respective position.
13 . The system of claim 10 , wherein:
each of the plurality of calibration signals is received from a scanning probe microscopy (SPM) or an electron microscope, and is associated with a topographic height.
14 . The system of claim 13 , wherein the scanning probe microscopy comprises an atomic force microscopy.
15 . The system of claim 13 , wherein the scanning probe microscopy comprises a scanning tunneling microscope.
16 . The system of claim 10 , wherein to characterize the surface topography of the semiconductor chip, the at least one processor is further configured to calculate a calibrated surface height at one of the positions on the surface of the semiconductor chip based, at least in part, on the corresponding original surface height and the surface height offset.
17 . The system of claim 10 , wherein the model comprises a classification model including a supervised learning process, and the plurality of categories are predetermined.
18 . The system of claim 10 , wherein:
the model comprises a clustering model including an unsupervised learning process, and the plurality of categories are non-predetermined; and
to classify the plurality of interference signals, the at least one processor is further configured to:
cluster the plurality of spectrum signals into a plurality of clusters, and
adjust the plurality of clusters to become the plurality of categories based, at least in part, on a design of the semiconductor chip, such that each of the categories corresponds to the respective region having the same material.
19 . The system of claim 10 , wherein to determine the surface height offset, the at least one processor is further configured to:
designate the surface baseline to a baseline region corresponding to one of the categories;
receive the calibration signal and a baseline signal of the calibration signals corresponding to the baseline region; and
determine the surface height offset based, at least in part, on the calibration signal, the baseline signal, and the interference signals.
20 . The system of claim 10 , wherein the at least one processor is further configured to:
generate an original height map based, at least in part, on the interference signals;
generate a spectrum map based, at least in part, on the plurality of spectrum signals spectrum signals;
transform the original height map into a classification map based, at least in part, on the classified spectrum signals and the categories; and
generate a calibrated height map from the original height map based, at least in part, on the surface height offset and the classification map.