How FTIR modeling delivers metrics based on materials’ bond types for compositional process control.
In the leading high-volume manufacturing (HVM) process flows, materials-enabled scaling has increased inline applications for compositional metrology.
A previous blog discussed how Fourier transform infrared (FTIR) spectroscopy was used for inline composition measurements. These measurements informed advanced process control for the wafer-level processing of selectively etched 3D NAND wordlines and DRAM capacitor profiles.
FTIR metrology has further HVM applications, including incoming substrate quality assurance, hardmask selectivity qualifications in the middle of the line, and verification of Low-K porogen evolution during interlayer dielectric (ILD) depositions on the back end of the line. These examples illustrate how FTIR modeling delivers metrics based on materials’ bond types for compositional process control.
The infrared-active bonds of dielectric and semiconducting materials produce unique absorbance signatures. The signatures are analyzed to determine the concentrations and associations of a material’s component elements. The FTIR optical modeling engine uses a physical model of the material structure being measured and its optical interaction.
Thin film layers are described by thicknesses and dispersions. Dispersions can be fixed, tabulated functions or built from a set of oscillators with customized line-shapes. These line-shapes describe specific bonds, such as NH amine, or bond polytypes, such as carbon’s sp1, sp2 and sp3.
The completed model describes the measured structure’s parametric inputs for calculations of reflectance and transmittance as a function of wavenumber. To extract information from the experimental FTIR data, the parameters of the optical model are floated while searching for the best fit to the data. One must carefully select the floating parameters to maximize sensitivity to the process variation of interest and avoid over fitting.
The FTIR optical modeling engine produces a set of optimal values for all floating parameters. This allows for interpretation of the results at various levels of granularity. At the most detailed level, deconvolution of a complex absorption peak into contributions from component oscillators is useful when the contribution of a particular bond is of interest. In other cases, the combined absorbance of a set of neighboring oscillators can be reported as an integrated group response.
Interstitial oxygen (Oi) converted to SiO2 precipitates during the thermal annealing processes. Oi acts as an internal getter and, at low concentrations, strengthens the substrate. At higher concentrations, however, Oi introduces slippage and warping of the substrate. As such, process control is critical.
Because the absorbance band of interstitial oxygen overlaps with the silicon phonon bands of the substrate, the default methodology to eliminate overlapping bands is to perform accurate measurements of the interstitial oxygen. This is done by performing a differential measurement, with a reference acquired on an ultra-pure Float Zone reference sample (FZ).
However, with the FTIR optical modeling engine, it is possible to directly measure Oi concentrations. The amplitude of the oscillator modeling Oi absorbance can be used to extract the Oi concentration of the substrate. This approach offers excellent repeatability, with a standard deviation of (1σ) <5ppb.
Fig. 1: (a) Absorbance of a Si bare Si substrate (red) and fitted spectrum from modeling (blue). The arrow indicates the position of the [Oi] absorbance peak. (b) Correlation between [Oi] concentration from the standard method and normalized oscillator amplitude from modeling.
Hydrogenated amorphous carbon (aC:H) is deposited by hydrocarbon precursors to form ashable hardmasks, with bonding and densities controlled by PECVD processing. The hardmask must resist pattern degradation during long dielectric etches but also be easily removable by oxygen stripping or CMP. This performance span depends on networked carbon-carbon bonding, with graphite-like and diamond-like bond types related to the hydrocarbons’ bonding fractions.
Using a modeling approach, it is possible to perform a deconvolution of the absorbance peaks of C-Hx bonds in the 3,200nm-3,600nm range on the absorbance spectra using the FTIR system. The amplitude of the peaks from sp2 CH2 and sp3 CH2 bonds can be extracted from the deconvolution. This enables the characterization of the sp2/sp3 ratio from the deconvolution.
Fig. 2: (a) Deconvolution of the CHx peaks on a thin amorphous carbon layer. (b) sp3/sp2 ratio mapping obtained on a 300mm automated Element FTIR system.
The baseline process for low-K ILDs is co-deposition of a backbone material, with mechanical integrity and modulus, along with a labile organic porogen. The porogen creates nanoscale voids when it is removed by ultraviolet thermal processing (UVTP).
UVTP strongly affects the CHx bonds in the layer. As such, being able to perform a deconvolution of those peaks allows for advanced vibrational spectroscopy studies and monitoring of the degree of cure and variations of K in the layers. The FTIR modeling engine enables this capability for automated measurements in HVM fabs.
Fig. 3: (a) Deconvolution of the CHx peaks on a thin SiOC low-K layer. (b) Example of deconvolution on samples with different post processing. The variations of the CHx bonds could be captured by the modeling approach.
As shown, FTIR modeling demonstrates advantages when characterizing the infrared-active bonds in an HVM semiconductor material structure. The FTIR optical modeling engine has been constructed from first principles to provide a flexible environment for modeling. This engine has been shown to provide key absorption peak deconvolutions in hydrogenated amorphous carbon and low-K interlayer dielectrics; it also offers direct measurements of interstitial oxygen in silicon substrates. Advanced modeling is critical to interpreting FTIR data when measuring devices for inline HVM applications.
Name* (Note: This name will be displayed publicly)
Email* (This will not be displayed publicly)
Threats are growing and widening, but what is considered sufficient can vary greatly by application or by user. Even then, it may not be enough.
Real money for VR tech, plus big chips and self-driving vehicles; 100 startups raise over $4B.
Technologies are being designed into more systems as defectivity drops and reliability increases.
Mature, low-cost manufacturing and proven reliability spur use in EVs, smart phones, and consumer electronics.
Optimizing energy and power efficiency in server chips and software is a multifaceted challenge with many moving parts.
As SiC moves to higher voltages, BEV users get faster charging, extended range, and lower system costs
Some segments are normalizing, others may be impacted through 2022.
Imec’s SVP drills down into GAA FETs, interconnects, chiplets, and 3D packaging.
Threats are growing and widening, but what is considered sufficient can vary greatly by application or by user. Even then, it may not be enough.
Abstraction is the key to custom processor design and verification, but defining the right language and tool flow is a work in progress.
Massive expansion campaign targets wide variety of chips, but export controls limit growth at leading edge.
Sharing resources can significantly improve utilization and lower costs, but it’s not a simple shift.
New and faster memory designs are being developed, but their future is uncertain.