Using ultrasonics to measure the strength of porous ULK dielectrics
11/01/2006
Low-k dielectric materials permit faster IC operation and improve electrical performance, but these materials are generally weaker and prone to mechanical failure. The mechanical strength is particularly low for porous dielectrics, so they are subject to failure during stressful manufacturing processes. Picosecond ultrasonics tools can rapidly measure the mechanical strength of porous dielectrics with k values approaching 2, and can resolve within-wafer uniformity. The technique can be used to develop more robust deposition processes and to control those processes in high-volume production.
Advanced IC designs incorporate copper conductors and low-k dielectric materials to improve conductivity and reduce parasitic capacitance between the more closely spaced interconnect lines. Both copper and low-k dielectrics contribute to faster operating speeds and improved performance by reducing RC delays. Although copper integration has proceeded quickly, a number of issues remain to be resolved in the integration of low-k dielectrics. Compared to the tougher silicon dioxide and fluorinated silicate glasses used in previous-generation ICs, these newer materials exhibit relative mechanical fragility. The stresses associated with CMP processing and advanced packaging techniques can lead to new failure modes for low-k dielectrics.
The International Technology Roadmap for Semiconductors calls for reducing the effective k value of interlayer dielectrics from 2.7-3.0 at the 65nm node, to 2.5-2.8 at 45nm, and 2.1-2.4 at 32nm [1]. Efforts to achieve these goals focus on introducing porosity into the material. While this does reduce the dielectric constant, it also reduces the films’ mechanical strength. Reduced strength leads to failure modes such as interfacial delamination and cohesive cracking, particularly during mechanically aggressive processes.
Young’s modulus, the ratio of force per unit-cross-sectional-area to deformation, is a measure of elastic strength. It has been shown to be the primary determinant of crack initiation and propagation behavior in low-k films [2, 3], with lower modulus materials being weaker and more prone to mechanical failure. Higher Young’s modulus materials have also been correlated with improved resistance to electromigration of copper [4]. Thus, the modulus can be an important predictor of device reliability, and its measurement and control become significant in process development and volume manufacturing.
Conventional mechanical testing techniques such as nano-indentation [5] and bend/bulge tests [6] are slow and destructive. Nano-indentation has limitations on porous, low-k materials: effects of the harder underlying substrate, densification of the film under the nano-indenter tip, and viscoelasticity [7]. Surface acoustic wave techniques have also been used to measure modulus [8].
Picosecond ultrasonics (PULSE Technology from Rudolph Technologies) is already used to measure modulus in a production environment on low-k dielectrics for the 90nm node. Its capabilities were recently evaluated by STMicroelectronics on ultra-low-k (ULK) films formed by PECVD of MSQ precursors in the k value range of 2.0 to 2.65 with porosity ranging from 0 to 60% [9]. These ULK films may be considered for use at sub-65nm technology nodes.
Picosecond ultrasonics
The picosecond ultrasonics method [10] was developed primarily to measure the thickness of opaque films, such as those used for metal interconnect layers. A very brief laser pulse (0.1 psec) heats the film’s surface, causing it to expand rapidly, and the expansion creates a sound wave that travels down through the film stack, partially reflecting at each interface. Film thickness and various other properties can be calculated by analyzing these echoed sound waves as they return to the surface.
The technique has been extended to the measurement of transparent films such as interlayer dielectrics. In this case, most of the laser energy is transmitted through the overlying transparent film to the opaque metal or silicon substrate, so that the sound wave is generated in the opaque layer. The wave then travels up through the transparent layer, causing a local variation in the refractive index that partially reflects the probe laser. The light reflected from the traveling wave front interferes with light reflected from the surface of the transparent film, resulting in constructive and destructive interference that causes the detected signal to oscillate in intensity (Fig. 1).
As a result of this interference, the measured signal oscillates with a period, τ, from which the sound velocity (V) in the material can be determined:
 |
where n is the index of refraction that can be measured independently with an ellipsometer, λ is the wavelength, and ϕ is the angle of refraction. Combining velocity with density, a known quantity for the material, gives elastic stiffness. Young’s modulus, which is elastic stiffness in a material that is free to deform laterally [12] (not the case in this analysis), can be related to elastic stiffness through Poisson’s ratio, assumed to be constant for the films examined, as shown:
 |
where ρ is the density and ν is Poisson’s ratio. Poisson’s ratio is assumed to be constant for the MSQ films used in this study.
Figure 2 shows an example of the picosecond ultrasonic signal detected from a 5000Å thick a-SiOC:H dielectric film, typically used at the 90nm technology node. Using the speed of sound, derived from the period of the oscillations, and the elapsed time between the pulse initiation and the arrival of the sound wave at the film’s surface marked by a phase reversal, thickness and modulus can be determined independently.
ULK measurement results
Film uniformity across the wafer is an important parameter, especially during process development when optimal deposition conditions are not yet known. Picosecond ultrasonic measurements, which take only a few seconds each, can rapidly survey the wafer surface to accumulate profiles or contour maps from multiple site measurements.
In the current study, 25 point modulus maps were used to compare the uniformity of two porous ULK films. The results gave a standard deviation within the wafer of 5.6% for one film and 10% for the other. Results such as these can be used during process development and after a process has been transferred to production.
Picosecond ultrasonic results for the MSQ films were also compared to those of a nano-indentor. As previously noted, nano-indentors suffer some limitations when applied to thin films. While there was a strong correlation between results, the nano-indentation measurements were consistently higher by a factor of greater than three. The most plausible explanation is that nano-indentation measurements of soft thin films are strongly affected by the presence of the stiff substrate, resulting in the modulus values being over estimated.
Porosity, mechanical strength, and k values
Picosecond ultrasonic modulus measurements provide a reliable method to characterize mechanical strength in porous low-k films. However, their use for routine process control requires a more thorough understanding of the relationship between material strength and porosity, and between porosity and k value. Porosity is difficult to measure directly; we used ellipsometric porosimetry, a laboratory technique that relies on observations of the rate at which an organic solvent is adsorbed and desorbed, filling and emptying the pore spaces within the film.
Figure 3 compares porosity measured by ellipsometric porosimetry with Young’s modulus as measured by picosecond ultrasonics. At 60% porosity, the k value is approaching 2. Increasing the porosity from 30% to 60% results in a decrease in modulus from 5 to ~1 GPa.
Figure 4 directly plots modulus as measured by picosecond ultrasonics versus k value for a range of film types. Conventional SiO2 is relatively robust with a modulus of about 70GPa but a k value above 4. Dense a-SiOC:H is weaker at 10 GPa and has a k value of 3. Finally, the porous ULK films from this study have moduli between 1 and 5 and k values in the 2.0-2.7 range.
Conclusion
Currently, porous low-k interlayer dielectric films can achieve the k values needed to reduce RC delays in advanced semiconductor devices, but only at the sacrifice of mechanical strength. Young’s modulus has been shown to be the primary governing property for crack initiation and propagation in low-k films.
Picosecond ultrasonic measurements of Young’s modulus provide a reliable method for monitoring mechanical strength; they are nondestructive and have sufficient spatial resolution to be used on test structures on production wafers. The measurements require only a few seconds at each site, fast enough to offer timely feedback for process control or to provide multi-site full-wafer surveys in process development applications.
Acknowledgments
This work was performed by the Crolles2 Alliance in collaboration with Rudolph Technologies. The authors would like to thank J. Vitiello of Philips Semiconductors, JC. Royer of CEA-Leti, and D. Barbier of LPM INSA for fruitful discussions about porous dielectrics and metrology methodology; the metrology group of Crolles2 Alliance for their technical support; and G. Tas and P. Mukundhan for their assistance on modeling with PULSE Technology, a trademark of Rudolph Technologies.
References
1. International Technology Roadmap for Semiconductors, 2005 Edition, Semiconductor Industry Association, 2005.
2. G. Xu, J. He, E. Andideh, J. Bielenfeld, T. Scherban, “Cohesive Strength Characterization of Brittle Low-k Films,” Proceedings of the IEEE, 2002 IITC, pp. 57-59, 2002.
3. R.F. Cook, E.G. Liniger, “Stress-corrosion Cracking of Low-dielectric-constant Spin-on-glass Thin Films,” J. Electrochem. Soc, Vol. 146, No. 12, pp. 4439-4448, 1999.
4. K. Lee, E.T. Ogawa, S. Yoon, X. Lu, P.S. Ho, “Electromigration Reliability of Dual-damascene Cu/Porous Methylsilsesquioxane Low-k Interconnects,” Appl. Phys. Letters, Vol. 82, No. 13, March 2003, pp. 2032-2034.
5. G.M. Pharr, W.C. Oliver, “Measurement of Thin-film Mechanical Properties Using Nano-indentation,” MRS Bulletin, pp. 28-33, July 1992.
6. Xiang, X. Chen, J.J. Vlassak ,”The Mechanical Properties of Electroplated Cu Thin Films Measured by Means of the Bulge Test Technique,” Mat. Res. Soc. Symp. Proc., Vol. 695, pp. L4.9.1-6, 2002.
7. C.M. Flannery, D.C. Hurley, “Laser Ultrasound: An Inspection Tool of Soft Porous Low-dielectric Constant Films for Microelectronic Interconnect,” Review of Nondestructive Evaluation, Vol. 22, 2000.
8. M. Gostein, A. Mazurenko,, A. A. Maznev, “Measuring Young’s Modulus of Low-k Dielectrics Using Surface Acoustic Waves,” MICRO, June 2004.
9. L.L. Chapelon, V. Arnal, M. Broekaart, L.G. Gosset, J. Vitiello, J. Torres, Microelectronic Engineering 76 (MAM 2004) 1-7.
10. C. Thomsen, H.T. Grahn, H.J. Maris, J. Tauc, “Surface Generation and Detection of Phonons by Picosecond Light Pulse,” Physical Review B, Vol. 34, No. 6, pp. 4129-4138, 1986.
11. W.P. Mason, Physical Acoustics and the Properties of Solids, D. Van Nostrand Co. Inc., New York, 1958.
Laurent-Luc Chapelon is an engineer working on ultra low-k/advanced dielectric barriers at STMicroelectronics, 850 rue Jean Monnet, F38926 Crolles Cedex, France; ph 33/476-925-765, fax 33/476-925-071, e-mail [email protected].
Daniele Neira is an engineer working on metrology applications for STMicroelectronics.
Joaquin Torres is an engineer in advanced BEOL for STMicroelectronics.
F. Noudin is an application scientist at Rudolph Technologies, Flanders, NJ.
Jana Clerico is the opaque film product manager for Rudolph Technologies.