Abschlussarbeiten - Master
SEM Based Overlay - development of dedicated workflow using new imaging capability on SD-SEM for inline process control
Art der Abschlussarbeit
Master
Autoren
- Millat, Shawn Shakahwat
Betreuer
- Prof. Dr. rer. nat. Johann Wolfgang Bartha
- Dr. rer. nat. Christian Wenzel
Weitere Betreuer
Ph.D. Jon Tobias Hoeft (Globalfoundries)
Abstract
Optical instruments are used for accurate overlay control in inline metrology. This
methodology requires quite large test patterns (min. 10 10m2), which can be placed
in the scribe line area only and therefore no precise information of local effects within the
active chiplet area are available. By use of Critical Dimension Scanning Electron Microscope
(CD-SEM), which is state of the art instrument for dimensional inline control, this
limitation can be overcome. because here small test patterns can be used (max. 44m2),
which can be placed within the chiplet area as well. Newest CD-SEM generation is able
to operate with higher landing energies (< 20keV ) and comes with optimized detection
schemes, so that “see through” overlay measurements can be executed. This means that
e.g. features of a layer beneath developed resist can be imaged at the same time as the
structured features patterned within the resist. For each layer and process step different
stacks are present, which requires different and optimized imaging schemes (working
points) on the CD-SEM. Setup and maintenance of the working points is very time consuming
and a limited set of working points is desired to minimize the cost of ownership
within the production environment. This is the reason why SEM imaging simulations of
various combinations of stacks and layers are required upfront to determine the best and
most suitable working points.
We have established a workflow to setup this measurements with dedicated working points
which incorporates a very sophisticated SEM simulation package, developed by SEMATECH
based on JMONSEL, and its final verification. There were two important tasks
carried out in the entire thesis work. The first task included the Simulation and evaluation
of optimized SEM beam conditions (< 20KeV ) for overlay measurements using simulation
package JMONSEL (Java Monte Carlo Scanning Electron) modeling for 28nm CA-PC-RX
and MX-VX layers at mask step and after etch. Five different simulations were made for
vii
three different stacks throughout the work. These structures in the process line were selected
for a study of CD-SEM based overlay metrology using marks covered with insulating
materials, PMMA resist and hardmask. The simulations were directed by optimizing the
acceleration voltage, number of pixels etc. The second task was the study and investigation
of the behaviors of simulated stacks in the real semiconductor manufacturing environment
(subsets of 2-3 beam settings). In that case the beam conditions which are analyzed by
the simulation package proven to be optimal are applied to verify in real CDSEM tool and
for a final modification.
For the first planned stack we have shown that it is possible to measure a trialed mark
through several materials layers under optimized measurement conditions. The second
mark (trialed mark) consists of a metal line with very low thickness and width in reference
with the first mark (reference mark). In addition, we have also conducted the simulation
with other beam energies lower than optimum beam energy to show that they are not
enough to spot the line edge topographies of the second mark. This was also proven by
experimental imaging.
The simulations and experiments for the second proposed structure were setup in three
variations. For all of these variations, the second mark was counted for a gate poly line
embedded into the active area. A via has been created through the ILD Oxide between
two second mark (trialed mark) which works now as the 1st mark. We have presented that
same beam energy is the optimum measurement condition for all of these variations to
detect the line edge topography of the second mark. It has also shown that beam energies
lower than optimum beam energy are necessary for the recognition of the contrast variation
of second mark with other materials.
For the third proposed structure we have seen that if there is available edge topographic
difference between materials interfaces than detection of edge topography of the 2nd mark
is satisfactory. If there is no edge topography difference in their interfaces then the contrast
variation between these materials is required. This contrast variation will separate
the second mark from the other materials. We have shown that an optimum beam energy
is required for the edge topography detection and contrast variation of second mark proven
by simulation as well as experiment.
Finally we concluded that optimum beam energy for the overlay measurement of different
structures depends on the materials properties, structural properties, charging effect etc.
For these reasons, simulation is performed upfront to find the optimum beam energy for
different structures.
methodology requires quite large test patterns (min. 10 10m2), which can be placed
in the scribe line area only and therefore no precise information of local effects within the
active chiplet area are available. By use of Critical Dimension Scanning Electron Microscope
(CD-SEM), which is state of the art instrument for dimensional inline control, this
limitation can be overcome. because here small test patterns can be used (max. 44m2),
which can be placed within the chiplet area as well. Newest CD-SEM generation is able
to operate with higher landing energies (< 20keV ) and comes with optimized detection
schemes, so that “see through” overlay measurements can be executed. This means that
e.g. features of a layer beneath developed resist can be imaged at the same time as the
structured features patterned within the resist. For each layer and process step different
stacks are present, which requires different and optimized imaging schemes (working
points) on the CD-SEM. Setup and maintenance of the working points is very time consuming
and a limited set of working points is desired to minimize the cost of ownership
within the production environment. This is the reason why SEM imaging simulations of
various combinations of stacks and layers are required upfront to determine the best and
most suitable working points.
We have established a workflow to setup this measurements with dedicated working points
which incorporates a very sophisticated SEM simulation package, developed by SEMATECH
based on JMONSEL, and its final verification. There were two important tasks
carried out in the entire thesis work. The first task included the Simulation and evaluation
of optimized SEM beam conditions (< 20KeV ) for overlay measurements using simulation
package JMONSEL (Java Monte Carlo Scanning Electron) modeling for 28nm CA-PC-RX
and MX-VX layers at mask step and after etch. Five different simulations were made for
vii
three different stacks throughout the work. These structures in the process line were selected
for a study of CD-SEM based overlay metrology using marks covered with insulating
materials, PMMA resist and hardmask. The simulations were directed by optimizing the
acceleration voltage, number of pixels etc. The second task was the study and investigation
of the behaviors of simulated stacks in the real semiconductor manufacturing environment
(subsets of 2-3 beam settings). In that case the beam conditions which are analyzed by
the simulation package proven to be optimal are applied to verify in real CDSEM tool and
for a final modification.
For the first planned stack we have shown that it is possible to measure a trialed mark
through several materials layers under optimized measurement conditions. The second
mark (trialed mark) consists of a metal line with very low thickness and width in reference
with the first mark (reference mark). In addition, we have also conducted the simulation
with other beam energies lower than optimum beam energy to show that they are not
enough to spot the line edge topographies of the second mark. This was also proven by
experimental imaging.
The simulations and experiments for the second proposed structure were setup in three
variations. For all of these variations, the second mark was counted for a gate poly line
embedded into the active area. A via has been created through the ILD Oxide between
two second mark (trialed mark) which works now as the 1st mark. We have presented that
same beam energy is the optimum measurement condition for all of these variations to
detect the line edge topography of the second mark. It has also shown that beam energies
lower than optimum beam energy are necessary for the recognition of the contrast variation
of second mark with other materials.
For the third proposed structure we have seen that if there is available edge topographic
difference between materials interfaces than detection of edge topography of the 2nd mark
is satisfactory. If there is no edge topography difference in their interfaces then the contrast
variation between these materials is required. This contrast variation will separate
the second mark from the other materials. We have shown that an optimum beam energy
is required for the edge topography detection and contrast variation of second mark proven
by simulation as well as experiment.
Finally we concluded that optimum beam energy for the overlay measurement of different
structures depends on the materials properties, structural properties, charging effect etc.
For these reasons, simulation is performed upfront to find the optimum beam energy for
different structures.
Schlagwörter
-
Berichtsjahr
2016