In contrast to the other electron beam lithography systems, the hardware of the smile2 contains no dedicated pattern generation but IO-devices, so it’s IO-hardware only. The smile2 hardware is responsible for data dispatching, timing, triggering and the digital-to-analog and analog-to-digital conversion.
With extensibility in mind, the hardware is divided into a digital part that can drive one or more analog “output pairs”. Each output pair can drive one beam, the beam blanker, and the signals from two detectors. In this way, existing systems can be extended to real multi-beam systems.
A novel and unique feature of the IO-harware is its ability to change the sampling rate (in terms of e-beam lithography: dwell time) at every sampled point without any time loss. We named this feature elementary time partitioning (etp). It is the the driving factor not only for high-resolution 3D Lithography, but also for high-quality proximity effect correction, since no fracturing of the geometry for correction is required.
In many cases (e.g. for small structures, lenses), it is important on which path the beam moves during exposure. Alos for a large amount of repeatable elements such photonic crystals, the exposure time can be reduced by minimizing jumps between two elements. Since the pattern generation is performed in software, the beam movement strategy can be adapted to the current task.
The main features for pattern generation can be characterized as:
Minimization of jumps between shapes
Control of the order in which the structures will be exposed
Control of the filling pattern strategy (the beam movement)
The most common pattern filling strategies (in smile2, called pattern generators) are included in smile2:
Meander (horizontal, vertical)
Single line (inside to outside, outside to inside)
Concentric (horizontal, vertical)
Smile2 also include some filling strategies for special cases:
Hatching (in any direction)
Single Point array
The following images illustrate the meander and the concentric filling strategies:
Pattern generators can be created by loading maps. The concentric pattern generators (e.g. lens) can be created by loading a profile curve. Maps and profiles represent scalar data which are recalculated to the dwell time. The original data can be provided as an image, a data file or a python script. The image below shows an example of the rectangle shape with a dose map created from the loaded image and exposed using the meander filling pattern strategy:
Structure after development (right) exposed with the meander filling strategy. Applied dose map and setup (left). Click for a full-size image.
The image below shows the exposed hexagon, which was made with an applied “Dose profile” using the concentric filling pattern strategy.
Hexagon structure (in the middle) generated with the help of a profile (on the left) and the structure exposed with the concentric filling strategy after development (on the right). Click for full-size image.
Proximity Effect Correction
The interaction of the electron beam with the resist layer and the substrate leads to an undesired “blurring” so called proximity effect. Depending on the structure geometry, this effect can lead to a significant size reduction of the fine features and destroy the concept of the entire structure.
There are tools and packages on the market, that aim to modify the structure (cut it into smaller pieces) such that the exposure of this partitioned (or fractured) structure minimizes the proximity effect. Since they target for traditional e-beam systems, these methods are limited by the amount of constant dose shapes. That leads to a suboptimal solution, because of the accuracy in the long-range proximity effect correction (PEC) and the short-range proximity effect is not precisely taken into account at all.
neomicra systems does not face this issue. Thanks to elementary time partitioning (etp), the dwell time of each point can be addressed and adjusted individually. This gives the following advantages:
No partitioning of the original geometry is required (what reduce artifacts)
The PEC can be performed as precise as the e-beam can be controlled (up to full 16 bit resolution)
The following image shows the working principle of the PEC in smile2.
Dose map calculation is based on the desired structure and the scattering parameters. The map is being applied during the exposure. Click to view the full-size version.
A real-world example of a structure with and without PEC is shown below.
Click to view the full-size version
The PEC package is available as an extension module of smile2. Based on the inputs, it can either produce a dose map (dose PEC map) or partitioned geometry (comparable to traditional PEC systems). The partitioned geometry can be exposed with any other system, the exposure of the dose PEC map requires the etp smile2 hardware.
The computation of the PEC can be performed at call just before the exposure starts, or in advance on the local computer in order to save time on the lithography system.
The PEC package comes with two integrated algorithms:
Local: Based on a “self-consistent” approach; used in other packages
Deconvolution: Corrects the shape and the dose simultaneously.
The framework is open for users to implement their own algorithms for proximity effect correction.
smile2 provides a state-of-the-art CAD editor (called itself also smile2). In the following, an overview of its main features is given.
All elements in smile2 are represented through nodes. Attributes of these nodes can be “connected”, forming a “node graph”. The complex relationships between nodes can be easily built and managed in the smile2 framework.
Almost everything that has been created in smile2 can be edited later. All actions and modifications in the document are recorded and placed into the so called Construction History. With the help of the Construction History some modifications could be called back or changed. If the History prevents the application of the new actions (shows wrong results due to connections between elements) there is a possibility to clean it.
The following example illustrates the concept of the Construction History for a rectangle, a text element and a boolean operation:
smile2 provides support for complex polygons with multiple outlines and holes. It also has a set of tools and boolean operations with the help of which any desired structure could be done.
Sophisticated user interface
smile2 comes with a modern and customizable user interface which has simple outline and is easy to use. Features like drag-and-drop or numeric inputs that can evaluate mathematical expressionsmakes the creation and editing of the structures easy and more efficient. Snapping tools such as grid, polar, point and guide snappingcan be used to place object points and objects at specified points.
The alignment in smile2 can be done either with automatic alignment tools or manually. The last one can be unavoidable in some cases such as bad contrast or destroyed marks. In contrast to other e-beam lithography softwares where alignment is often complicated and not very reliable, smile2 software has interactive easy-to-use graphical alignment tools which with its simplicity can increase the outcome of your lithography.
More information on this topic can be found in the smile2 user manual.
Automation & Scripting
smile2 was designed always keeping automation in mind. Its capabilities to run long-standing or complicated task/tasks automatically goes beyond the concept of simple task lists (nevertheless, task lists are also available). Event handlers can be used to react on all kind of exposure related events (e.g. after the stage has been moved). Within this framework, many tasks can be realized:
Automatic detection of proper locations based on scans
Fully automated alignment
Automatic current measurements in regular time frames
Interactive creation of wafers with the possibility to save positions of exposed structures
smile2 offers a set of python modules that can be used as the smile2 application or from external applications. Since the graphical user interface is written in Qt5, PyQt can be used to extend the user interface of smile2. It also comes with a full Python 3.5 installation including common modules like numpy, scipy or matplotlib. In fact, many high-level functions smile2 are implemented using the python framework and can act as an example.
The integrated Python editor, features autocompletition and synthax hilighting
Example of a structure, genereted with a python script