KLA-Tencor – a revolution in multi-beam electron lithography

This is the 3rd article in a series.

1st article – Technology of electron-multi-beam lithography from IMS Nanofabrication

2nd article – Mapper – ours, or another technology for electron-multi-beam lithography

KLA-Tencor Corporation is a major manufacturer of equipment for all stages of microelectronics production: wafers, matrices, chips and packaging. From research and development to final series production. The corporation perfectly understands the needs and trends in the industry, continues to develop electronica lot ofbeam technology with financial support from DARPA as part of the Maskless Nanolithography Program (Maskless Nanowriter Program).

Published information about 2 generations of technology from KLA-Tencor, called REBL (Rreflective Eelectronic Beam Lithography).

Technology from KLA-Tencor is the most sophisticated representative of electron beam lithography. But … yet, not out of the walls of the laboratory.

Researchers from KLA-Tencor decided to take a radical approach to one of the main problems of electron beam lithography. This problem is the source of electrons. The fact is that in order to get a spot of the minimum size on the target (silicon wafer), it is desirable to have a source with a radiation surface of the minimum size (ideally from one point), with a minimum energy spread (ideally the same). On the other hand, the source must emit as many electrons as possible per unit time, the illumination speed depends on this. Thus, we have some compromise dilemma: if a large radiation area – then a large (blurred) spot, if a small radiation area – then a small current (low illumination speed).

Might be worth looking elsewhere?

Indeed, such a valuable resource as emitted electrons is spent completely irrationally. After radiation and creation of a unidirectional flow (collimation), the electrons enter a plate with an array of holes (aperture plate). Moreover, the area occupied by the holes is much less than the total area of ​​the plate, and the electrons are simply extinguished on it. For example, in IMS, the plate has holes 4×4 µm with a step of 32 µm, i.e. out of 1024 sq. microns (32×32) only 16 square meters are useful. microns (4×4). The consumption of electrons is simply wasteful, for one emitted electron we have 64 ballast ones! This is explained by the fact that deflecting electrodes and a control system for these electrodes must be placed between the holes.

Researchers from KLA proposed an original solution – to use not a plate with holes, but a plate with electronic micro-mirrors. The control of micro-mirrors can be transferred to the rear side and thus save on the gaps between them. KLA-Tencor has succeeded in creating an array of one million micro-mirrors with a diameter of 1.4 µm each and a pitch of 1.6 µm. The savings in area (and valuable electrons) are obvious.

The creation of such a matrix of micro-mirrors turned out to be a rather difficult task and took about 3 years, since electron mirrors did not have a wide practical application and their theory was developed very poorly. I had to solve a number of problems with the interaction of rays in neighboring mirrors when turning on and off, the weak contrast of mirrors due to the spread of electrons in energy, poorly studied microlysis effects.

As a result, a matrix with 1’015’808 micro-mirrors (4096×248) was used on the experimental setup. The chip size was about 25×27 mm in the center of which an array of mirrors occupied an area with dimensions of 6.5×0.4 mm.

Moreover, separate control was carried out only by 248 columns of 4096 micro-mirrors. Such control is the result of simplifying the task so as not to create a control CMOS structure, but to get by with only a metallization layer.

Micro-mirrors were a stack of 5 conductive layers separated by insulating layers. Thus, each micro-mirror was actually a micro-lens in the form of a “cup”, the bottom of which was a reflective surface. If a positive voltage was applied to this surface, the mirror absorbed the electron flow. If negative, the electrons changed the direction of movement to the opposite.

And where to attach such a wonderful find?

In order to use mirrors, the electron-optical system must differ quite significantly from the classical one.

In the 1st generation, the flow of electrons (blue in the diagram) from the Electron Gun (Electron Gun) using a magnetic prism (Magnetic Prism) is directed towards the Digital Pattern Generator (Digital Pattern Generator). Using the Objective Optics system, the electron energy is reduced to 1 eV, which allows a negative potential of 1-2 V on the mirror array to reflect the incident electron flux in the opposite direction. If a small positive potential is applied to the mirror, the electrons from the stream will be absorbed by the mirror. The mirrors are an array of isolated MEMS elements with a diameter of 1.4 µm with a step of 1.6 µm. Thus, a beam of reflected rays (orange in the diagram) is formed from the total stream, which are rotated by a magnetic prism towards the projection optics (Projection Optics). Projection optics scales (reduces) the beam of rays by 50 or more times and projects onto a rotating table (Multiple Wafer Rotary Stage). 6 plates are placed on the table at the same time.

During recording, the table rotates and moves linearly so that the beam of rays forms a spiral path on the table from the edge to the center, as on a gramophone record.

Such an electron-optical system has a fairly extended electron trajectory, approximately 1540 mm. On such a long trajectory, it is difficult to keep the electron flow from “spreading” due to the forces of the Coulomb repulsion.

Therefore, the 2nd generation of the system was proposed. With an electron-optical path length of approximately 506 mm.

The essential difference of the 2nd generation of the system is that the Electron Gun has been moved from the top of the column to the side. And the Digital Pattern Generator is placed on top of the column. The main rotation of the beams is 110 degrees. is carried out by an electrostatic deflector (Beam Bender), and the splitting of the trajectories towards the Digital Pattern Generator and towards the table is carried out by the Wien filter (EXB filter). The Wien filter turns the flow from the electron gun by 15 degrees, and the beam of rays reflected from the mirrors does not change direction, due to the compensation of the deflecting forces of the magnetic and electrostatic fields.

Table.

Such a strange decision in the form of a rotating table is due to two reasons.

The first reason is forced. Having received a gain in the rational use of the emitted electrons (the distance between the beams is reduced to a minimum), the designers came to the conclusion that it was no longer possible to scan the target with the electron beams in the transverse direction relative to the main direction of recording. For example, in IMS, the distance between adjacent beams is about 160 nm and the system, after illuminating a 20×20 nm spot, deflects the beam to an adjacent field of 20×20 nm, until all the gaps between the beams are illuminated. In Mapper, the illumination distance between adjacent beams is 2000 nm. This allows the use of a rather slow moving XY table. In KLA-Tencor technology, the distance between the beams is minimal (no more than the beam diameter) – because of this, there is no time for illuminating the space between the beams, and the total area of ​​illumination, by the flux of all beams, on the plate is about 100×6.2 µm (100 µm side is perpendicular to table movement). Beam deflection by 100 µm will cause significant defocusing, so the table must move at a linear speed that is significant for mechanical systems (halftone illumination is realized over a width of 6.2 µm). For KLA-Tencor technology, this speed is in the order of 3-10 m/s. Reciprocating tables will experience significant accelerations (up to 20G).

The second reason is a bonus of solving the first problem. Having created a rotating table for 6 plates, it becomes possible to place several cathode-ray columns above the table leading to simultaneous recording, which greatly increases the productivity of a single installation.

The disk (Platter Assembly) with a diameter of just over 1 meter is mounted on a hub with a magnetic suspension. The weight of the disc is approximately 120 kg.

The metrology system consists of two main parts.

The first part is a system of a 10-axis laser interferometer, consisting of a laser fixed relative to the column (Laser Assembly) and an umbrella-shaped bracket (Umbrella Assembly) with optics attached to it and distributing the laser beam along 10 axes (Distribution Optics ). This system determines the coordinates of a very precise ring mirror attached to the center of the disk (Ring Mirror).

The second part is a rotary encoder that measures the angular position of the disk relative to the “system”.

The magnetic suspension of the table made it possible to realize the self-balancing of the system.

KLA-Tencor’s multibeam electron lithography technology overcomes a fairly significant performance barrier that can be achieved from a single electron source, and therefore from a single installation. However, overcoming this barrier entailed a rather significant complication of the technology.

And, as mentioned above, the technology has not stepped outside the walls of the laboratory. A rather noticeable characteristic of the level of technology development is the absence of any information about the authors’ struggle with vacuum pollution from photoresist fumes. This indicates that the number of operating hours of the experimental setups is very small (the vacuum does not have time to become contaminated), and the authors have not yet encountered this problem.

This is the third article in a series.

1st article – Technology of electron-multi-beam lithography from IMS Nanofabrication

2nd article – Mapper – ours, or another technology for electron-multi-beam lithography