The Heidelberg MLA150, our direct-write laser lithography system, is known for its outstanding performance when aligning a new design to a patterned wafer. Nominally the system is capable of better than ±500 nm global alignment precision in both X- and Y-directions (for topside, ±1 μm for backside), typically surpassing that and producing results in the ±300 nm range. However, depending on factors such as stress build-up on wafers due to deposited materials, this precision may become compromised and produce both worse alignment overall as well as non-uniform precision (i.e., varying die-to-die alignment precision).

To overcome the limitations caused by using a single, global alignment for a whole wafer, we now have the option to perform local alignment on each die. This new method, as evidenced by the image below, greatly improves alignment precision on each die, reducing both the overall offset and the die-to-die variation. In this demo, standard alignment (left) shows offsets in the ±300 nm range. While all dies are within specification, the offset still varies by up to 500 nm die-to-die. On the other hand, field alignment (right) greatly improves the results, such that all dies exhibit offsets in the ±50 nm range (i.e., the resolution of the verniers used for this test).

This improved precision does come at the cost of longer exposure times, however, since the system now has to scan and measure alignment marks for each die being patterned. However, this is a reasonable cost to pay when poor alignment is detrimental to device performance. Please note that this new technique is only available for topside alignment—it cannot be used for backside alignment due to the requirements for this procedure.

If you are interested in using this new field alignment feature, a new document detailing the procedure is available by the tool, and in our Knowledge Base. Also, please do not hesitate in submitting a training request via LMACS if you wish to get hands-on training. For more information, please contact Gustavo de Oliveira.

In semiconductor processing, packaging is a critical step in finishing a final product. After completing the dicing process to singulate dies from a larger substrate, the cut dies generally remain affixed to dicing tape. It can be difficult to manually remove the dies without scratching or rubbing against one another, causing chipping or other damage. A die expander allows for the tape to be stretched on expander rings, separating the dies for easier removal or shipping.

We are happy to announce the addition of a Hugle Model-1810 Die Matrix Expander to the nanoFAB's tool lineup. This expander is capable of accommodating diced specimens up to 150 mm in diameter. It is equipped with a heated chuck, allowing for easier separation of the dies without delamination from the tape, as well as a preheat timer, which helps with process repeatability. The stage expansion distance is adjustable, allowing for customized spacing between the dies. The system also includes an automated cutter tape separation system for ease of use. Previously, without the use of this tool, users in our facility needed to mount diced 150 mm wafers onto the expander ring by hand, a task that often required two people to do!

We hope that the addition of this new system in our toolset will facilitate the final processing and packaging of the many devices fabricated in our open-access facility. Please submit an equipment training request via LMACS if you wish to get trained on the die expander. For more information, please contact Breanna Cherkawski.

Please see below for some videos of the tool in action:

Overview

In plasma-enhanced chemical vapour deposition (PECVD) tools, the deposition rate of deposited films can often fluctuate with relative gas concentrations, material build-up on the chamber walls, and other mechanisms that make timed depositions inconsistent. To combat this problem, a simple technique called Optical Emission Interferometry (OEI) can be used. OEI provides a more robust form of deposition control based on the interference of light reflected off of the growing film to precisely hit target film thicknesses.

OEI uses similar principles to other interferometric tools like the Filmetrics F50-UV; the intensity of light at a single wavelength reflected off of a thin film will vary depending on the wavelength of light and the optical thickness of the film. This is due to the relative phase difference between light reflected off the top surface of the film and at the film-substrate interface, i.e., conditions of constructive and destructive interference (shown in Fig. 1). Across a broad spectrum of wavelengths, this can be used to fit to both refractive index and film thickness robustly (as Filmetrics does). However, if the intensity of light reflected off of a growing film is recorded continuously, the intensity at a "single" wavelength will oscillate approximately sinusoidally as the thickness increases (satisfying constructive and destructive interference conditions at specific thicknesses)—this is the fundamental premise of OEI. By noting when the extrema occur during film growth, the period of this oscillation (T) can be used to measure the deposition rate of the film, since T is set by the phase conditions for constructive/destructive interference. T = λ/(2n); where λ is the wavelength of interest and n is the refractive index of the material at this wavelength.

To use OEI in the Plasma-Therm Versaline PECVD, light produced by the plasma is reflected off of the growing film and collected by a spectrometer centred above the wafer for plasma monitoring. The Plasma-Therm EndpointWorks software package is then used to monitor the evolution in time of the spectral intensity emitted from the plasma within a narrow range of wavelengths. Choosing a range that encompasses an emission line of a typical gas/radical species within the plasma produces the oscillatory data shown in Fig. 2. Values of T are calculated based on known values of n and λ to give an estimate of the deposition rate based on the spacing of the extrema. Thus, in order to function correctly, each OEI recipe requires the user to have knowledge of both the growing film (single wavelength refractive index) and of typical optical emission lines (usable wavelengths) within the plasma.

Wavelength choice

The choice of what wavelength to use for OEI depends on the chemistry of the plasma in question, since each gas/radical species emits light at different wavelengths (depending on its electronic transitions), and thus the available wavelengths will differ for each recipe. Potential emission lines are further limited if the gas species characteristic of an emission line is consumed during the reaction. The intensity of these reactant emission lines will fluctuate as the film grows and as the local gas concentration changes. In contrast, the intensity of an emission line of an inert gas (He, Ar, or N₂ are often used to dilute the reactants) will be much more stable, and will provide a reduction in noise in the reflected light. This leads to a more clearly defined periodic signal, where the extrema can be easily identified in software. Additionally, since variation in T is linear with the wavelength of light used, the choice of shorter wavelengths will reduce the period of oscillation, lending the OEI method more sensitivity and flexibility to accurately measure thinner films. Note, for films that are optically absorbing (e.g. a-Si and Si-rich nitrides), the film will absorb more strongly at shorter wavelengths, decreasing the amplitude of the oscillation as the film grows. Ultimately, this decay reaches a point where extrema cannot be identified past a certain thickness. A general rule of thumb is to choose the shortest wavelength inert gas emission line that is not strongly absorbed by the depositing film. A typical spectrum collected during deposition is shown in Fig. 3, where a N₂ emission line at ~333 nm is highlighted, being suitable for the gas chemistry involved. For more details on choosing an emission line appropriate for your deposition, please see the Plasma-Therm Versaline PECVD OEI recipe development How-To on the nanoFAB's Confluence Knowledge Base.

OEI controlled deposition recipes are currently qualified for SiO2 Dep ("OEI SiO2 Dep") and Stoichiometric Nitride ("OEI Stoichiometric Nitride") recipes, with an OEI recipe in progress for the Si-Rich Nitride. For user-developed films that differ from these established recipes, users are welcome to qualify their own OEI recipes following the Step-by-step instructions in the linked OEI recipe development How-To. If you encounter problems setting up your OEI recipe on EndpointWorks, please contact Tim Harrison (tr1@ualberta.ca).

We are happy to announce a new addition to our fabrication tool lineup: a Sono-Tek ExactaCoat Spray Coater. This tool is located in our recently renovated 10k Lithography area (ECERF W1-031); please stay tuned over the coming months as we add more photolithography capabilities to this labspace.

The ExactaCoat, an ultrasonic spray coating system, can accommodate specimens up to 300 mm in diameter. This tool is primarily used for photoresist coating in situations where conventional spin coating is impractical or yields poor results: specimens that are too fragile, exhibit large topographical variations, or possess irregular shapes.

The spray coater offers a particular advantage for specimens where topography presents a challenge to achieving good film uniformity during spin coating: examples include MEMS or microfluidic devices with deep (tens to hundreds of µm) features etched via DRIE or KOH/TMAH. Process development is currently underway to formulate recipes suitable for coating deep and/or high aspect ratio topography wafers. Additionally, spray coating is a valuable technique for those looking to apply photoresist as a dicing protect layer on extremely fragile or through-etched substrates that are not compatible with spin coating.

The spray coating process uses high-frequency (ultrasonic) mechanical vibration that is transferred to the liquid in the nozzle. The standing waves created in the liquid are broken up as the liquid exits the nozzle—this aerosolised liquid is then dispensed onto the surface of the specimen as a fine mist.

 
To coat an extended specimen such as a wafer, the spray of atomised resist is rastered across the work area in overlapping stripes. This is illustrated in the following videos, where a cleanroom wipe is used to show the sprayed resist more clearly.

After spray coating a wafer with resist and softbaking, it may be exposed and developed to complete a photolithography process. The following optical microscope images illustrate some preliminary results of photolithography on Si wafers, showcasing excellent film uniformity and achieving lateral feature sizes as small as 5 µm.

Photolithography using spray coating: Si wafers spray coated with a ~4.5 µm thick film of AZ 1505, exposed in a contact mask aligner, and developed. The smallest feature size pictured in each image is 5 µm.

We are thrilled to introduce this new capability, along with the array of associated applications it makes possible. If spray coating is a technique you feel would be beneficial to your work, please submit a request through LMACS to receive training. If you have any questions, please do not hesitate to contact Breanna Cherkawski or Aaron Hryciw.

Resources

Application note: Spray coating (MicroChemicals GmbH)