A novel on-axis interferometric alignment scheme, especially applicable to x-ray lithography, is described which combines the position sensitivity of interferometry and the robustness of imaging. It employs broadband illumination, and hence should be relatively immune to many of the effects that tend to corrupt alignment signals in conventional interferometric systems. In its initial demonstration a standard deviation (σ) of 6 nm was achieved in both X and Y. Ultimate limits are calculated to be below 1 nm. On an Apple Quadra 800 computer, the spatial-phase information that measures misalignment is fully analyzed in 200 ms.
The implementation of x-ray lithography requires that in-plane distortion caused by absorber stress be eliminated (i.e., be made less than ∼10% of minimum feature sizes). As a rule of thumb, for large-area membranes (i.e., ≥30 mm diam) stress-induced distortion is negligible if absorber stress is kept below 108 dynes/cm2 (10 MPa). The most common way of determining absorber stress on membranes is to measure the out-of-plane deflection in a Linnik or Mireau interferometer. Out-of-plane deflection is caused by absorber bending moment. It is possible, in principle, for an absorber to have a stress that varies with depth in such a way that bending moment is zero but net in-plane stress is not. To determine if there is a one-to-one correspondence between out-of-plane and in-plane distortion, and in order to check the validity of Yanof’s model, we have developed two methods of measuring in-plane distortion: a moiré method and a holographic interferometry (HI) method. Both require that one etch or lift-off a grid on one surface of the x-ray mask membrane using holographic lithography and reactive-ion etching. On the opposite surface an x-ray absorber is deposited and patterned. In the moiré method another holographic exposure is performed, at the same spatial period as the grid, but at a small azimuthal rotation. After development, a clear moiré pattern is formed, from which one can calculate the in-plane distortion and corresponding absorber stress. In the HI method, the mask is placed in the standing wave produced by two intersecting laser beams. A fringe pattern can then be seen on a charge coupled device camera. The HI method provides a higher sensitivity to in-plane distortion than the moiré method and should enable one to measure in-plane strain at a level corresponding to in-plane distortion below 1 nm for typical pattern geometries.
We report on the fabrication of AlGaAs/GaAs split-gate electron waveguide devices of lengths between 0.1 and 2 μm using x-ray lithography, and the measurements of these devices at liquid-helium temperatures and up to 15 K. An x-ray mask (parent mask) was fabricated using e-beam lithography and replicated using proximity x-ray lithography (λ=1.32 nm) to generate a replica (daughter) mask. The daughter mask was then aligned to patterns on a high-mobility AlGaAs/GaAs sample and x ray exposed using a conformable mask fixture. The conductance of the electron waveguides was measured as a function of the split-gate bias. Sharp 2e2/h conductance steps were observed in devices up to 0.75 μm long at T=2 K. The features in the conductance remain visible up to 15 K.
To realize a technology for x-ray nanolithography (<100 nm features), which is compatible with manufacturing, a number of mask design requirements must be met that are unrelated to patterning, repair, and alignment. These include high-flatness membranes and support structures so that mask-wafer gaps less than 10 μm can be achieved without risk of damage, and a rigid mask frame to avoid problems of distortion during handling. The membrane material should be compatible with semiconductor-processing, possess high strength, be radiation hard, and be transparent to light for alignment purposes. Details of a mask architecture that meets these requirements will be described.
The Submicron Structures Laboratory at MIT develops techniques for fabricating surface structures with feature sizes in the range from nanometers to micrometers and uses these structures in a variety of research projects. These projects, described briefly below, fall into four major categories: (1) development of submicron and nanometer fabrication technology; (2) nanometer and quantumeffect electronics; (3) crystalline films on non-latticematching substrates; and (4) periodic structures for x-ray optics, spectroscopy and atomic interferometry.
We have developed a compact, low-cost system for sub-100-nm x-ray lithography. This system, which provides mask alignment and exposure at atmospheric pressure, was designed for sub-100-nm electronics research where the process latitude and high throughput of x-ray nanolithography have proven to be especially valuable. The system includes a CuL electron bombardment x-ray source (λ=1.34 nm) and an exposure chamber. The exposure chamber is filled with 1 atm of helium and is separated from the vacuum chamber housing the electron bombardment source by a 2-cm-diam window of 1–2-μm-thick SiNx. The exposure chamber contains a pin chuck to keep the wafer flat, a mask holder, and piezoelectrically driven X, Y, and θ stages. In this initial design, mask-to-wafer alignment is performed using a high NA (0.95), 1000× optical microscope. Future designs will incorporate computer-controlled interferometric alignment, which is compatible with sub-10-nm precision. For sub-100-nm lithography, the mask and substrate are separated by less than ∼5 μm to avoid the deleterious consequences of diffraction at the CuL wavelength. The mask-to-wafer gap is adjusted using a piezoelectrically controlled mask holder. The system is modular, easily reproduced, and suitable for low-budget nanolithography.
At linewidths of 0.1μm and below, the time required to expose large-area patterns (∼1cm2) by electron-beam lithography becomes impractically long. Thus, there is a critical need for reliable replication techniques in this linewidth domain. Replication of features as fine as 18nm by x-ray lithography has been demonstrated. However, all sub-0.2μm-linewidth x-ray lithography to date has been done in intimate contact using polyimide-membrane masks. Polyimide is subject to distortion, and contact can cause mask damage. For these reasons we have investigated x-ray nanolithography with inorganic membranes, and mask-sample gaps of ∼4μm. We discuss the resolution-limiting factors of penumbra, diffraction, and photo-electron range, and consider the tradeoffs of gap and wavelength that need to be made to achieve high resolution. With a CuL source (λ=1.3nm), a linewidth of 0.1μm requires a gap of ∼4μm. We have used stress-controlled SiON0.5 membranes of 6cm2 area attached to a narrow rim etched into a Si substrate. The gap was set by small spaces studs. Gaps of 4–6μm were achieved with uniformities of ∼0.5μm. They were measured using an angular fringe counting technique. Microgap nanolithography was used to replicate 100nm lines with the CuL x-ray. High aspect-ratio PMMA structures were achieved with steep sidewalls, suggesting that 50nm lines could be faithfully printed with the microgap technique.
Contains table of contents for Part I, table of contents for Section 1, description of Submicron Structures Laboratory research, reports on twelve research project and a list of publications.