by Thomas Werner, Johann Steinmetz, Michael Kiene, Björn Eggenstein, Frank Richter, Frank Kahlenberg, GLOBALFOUNDRIES; Simon Heghoyan, Darron Jurajda, Daniel Sullivan, Brewer Science
EXECUTIVE OVERVIEW
As features sizes continue to scale according to the ITRS, Cu wiring based on dual damascene is beginning to hit fundamental limits of current via-first, trench-last (VFTL) integration. To accommodate feature-dependent variations, the thicknesses of the various films comprising the lithography stack for trench patterning are not able to scale as rapidly. This article outlines a novel approach for eliminating these variations using existing processes and equipment, in turn enabling potential benefits to each of the process units involved.
Planarity is increasingly a limiting factor in printing structures in the 32nm technology node and beyond. In addition to non-planarity caused by CMP variations, further loss of planarization arises during the via-fill process with the use of an organic planarization layer (OPL), and subsequently leads to a severe reduction of the lithography focus budget. Since the degree of planarization is compromised by the range of via size and layout within the die, the OPL is therefore coated thickly to compensate for this. The trade-off is to increase the aspect ratio when migrating to smaller dimensions, leading to OPL pattern collapse. Furthermore, feature-dependent variation of the plug height after OPL etch causes a wide distribution in the extent of profile chamfering at the trench-via interface during the ensuing dielectric etch. Such profiles negatively impact the requirements for the subsequent liner fill, which must be correspondingly optimised for a range of possible geometries.
A developer-soluble gap-fill material that uses a wet etch-back process to remove the gap-fill material back to the substrate surface has been developed by Brewer Science. A thick coating of this material, Brewer Science WGF 300, is applied to the substrate, with excess material above the vias removed using a standard photoresist developer, leaves the vias fully filled (steps 2 and 3 of the “WGF approach” in Figure 1). Using this method, it is possible to reduce the initial bias from the coating and provide a more planar surface for the ensuing lithographic and etch processes [1]. Wet etch-back processes have an advantage over dry etch-back gap-fill materials as they eliminate the need to transfer wafers between the etch and photo bays, since all the processing for a wet etch-back material can be performed with standard lithography equipment [2]. The WGF approach is compared to the process of record (POR) in Fig. 1.
Figure 1. Process flows, POR versus WGF approach. (Image courtesy of GLOBALFOUNDRIES Dresden Module One LLC & Co. KG)
CLICK HERE to view larger image
Impact on planarity
At the initial step of the investigation, the impact of WGF 300 material upon OPL planarity was measured on a 32nm test chip. WGF 300 material was applied after via etch, after which the gap-fill layer was developed-back until reaching the top of the vias. This step was followed by a second deposition of the POR OPL. Figure 2 shows a comparison of the planarization using the POR and WGF 300 approaches. Using the WGF 300 approach, it was possible to improve the feature-dependent variation in OPL height from 60nm to nearly zero. Depending on the product layout, this planarity improvement is expected to translate into improvements in the lithography process window across various via densities. The exact impact on the lithography process will be subject of a later investigation.
Due to the variations in planarity, the OPL etch time is limited by the lowest OPL plug-height found in any feature. This limitation can cause unfavourable trench profiles, especially in the transition area between trench and via. When WGF 300 material was used, it was possible to use longer OPL etch times, which leads to more chamfer in this region and consequently allows better liner coverage. This effect was found to be beneficial for the subsequent copper fill process.
Figure 2. OPL thicknesses over various features. (Image courtesy of Thomas Werner, GLOBALFOUNDRIES Dresden Module One LLC & Co. KG)
CLICK HERE to view larger image
Furthermore, the improvements in OPL planarity with the WGF approach allow thinning of the OPL layer, which could not be achieved with the POR approach. Thinner OPL allows for better CD control over various features. Additionally, the amount of OPL which needs to be stripped after the trench etch is reduced. This aspect can also help to reduce low-k/ultralow-k damage effects, such as increased capacitance and undercut below the hardmask layer.
A number of 32nm and 45nm split-lots were processed, where the WGF 300 approach was compared directly to POR. Most electrical parameters, such as line resistance and via chain yield, were comparable or slightly improved when WGF 300 was used. A remarkable improvement was achieved for some via-leakage structures. As shown in Figure 3, these structures are used to measure leakage between neighbouring vias with adjacent trenches. This result can be explained by the improved planarity over areas of dense vias, which provides relatively thicker OPL compared to the POR. This improved planarity results in more OPL remaining in these critical areas, which in turn provides better protection of the narrow isolation during trench etch.
Figure 3. Improved via leakage behaviour in critical test. (Image courtesy of GLOBALFOUNDRIES Dresden Module One LLC & Co. KG)
CLICK HERE to view larger image
The defect performance using 32nm material was slightly improved compared to the reference process. Product yield in the 45nm technology was comparable to the POR manufacturing technology.
Conclusion
Developer-soluble gap-fill materials can offer the following advantages when implemented in the via-first, trench-last (VFTL) OPL integration flow:
- Increased photolithography process window
- Improved dual damascene profiles due to wider OPL etch-process window
- Thinner OPL to allow for vertical scaling of OPL thickness.
With the introduction of developer-soluble gap-fill materials, it becomes possible to extend the current integration approach to technology nodes beyond 32nm. Implementation into an existing process flow requires two additional process steps (OPL coat and develop), which can be performed in a properly configured stand-alone coater-developer track. The material proved compatible with currently used materials and processes in the fab.
Acknowledgment
WGF 300 is a registered trademark of Brewer Science, Inc., Rolla, MO USA.
References
[1] C. Washburn, N. Brakensiek, A. Guerrero, K. Edwards, C. Stroud, N. Chapman, “Wet-recess Process Optimization of a Developer-soluble Gap-fill Material for Planarization of Trenches in Trench-first Dual Damascene Process,” Proc. of SPIE, vol. 6153, 2006, pp. 815-820.
[2] D.M. Sullivan, R. Huang, S. Brown, A. Qin, “New Developer-Soluble Gap-Fill Material,” Proc. of the 6th Inter. Conf. on Semiconductor Tech., vol. 2007-01, 2007, pp. 61-70.
Biographies
Thomas Werner received his masters in electrical engineering at Chemnitz U. (Germany) and is senior member of the technical staff at GLOBALFOUNDRIES Module One LLC & Co. KG, Wilschdorfer Landstr. 101, 01109 Dresden, Deutschland; e-mail [email protected].
Johann Steinmetz received his Dipl.-Ing (FH) in physical chemistry at the U. of Applied Sciences Munich (Germany) and is a member of the technical staff at GLOBALFOUNDRIES Dresden Module One LLC & Co. KG.
Michael Kiene received his PhD at Kiel U. (Germany) and is a member of the technical staff at GLOBALFOUNDRIES Dresden Module One LLC & Co. KG.
Björn Eggenstein received his masters in electrical engineering at the Technical U. of Berlin (Germany) and is a senior process engineer at GLOBALFOUNDRIES Dresden Module Two GmbH & Co. KG.
Frank Richter received his Dr.rer.Nat. at the U. of Regensburg in solid-state chemistry and is a process engineer at GLOBALFOUNDRIES Dresden Module One LLC & Co. KG.
Frank Kahlenberg received his diploma and doctorate degree in chemistry from the U. of Wuerzburg and is a lithography process engineer at GLOBALFOUNDRIES Dresden Module One LLC & Co. KG GLOBALFOUNDRIES, Dresden, Germany.
Simon Heghoyan received his PhD in solid-state physics from the U. of Wales, Cardiff and is an account manager at Brewer Science Ltd, Derby, England UK.
Darron Jurajda received his BS in chemical engineering from the U. of Texas at Austin and is a field applications engineer at Brewer Science, Inc., Rolla, MO USA.