Solid Freeform Fabrication from Gas Precursors Using Laser Processing

Final Report, 1996-2001

________________________________________________________________

Principal/co-Principal Investigator.

* PI/co-PI Name(s): Harris L. Marcus

* PI Institution: University of Connecticut, Storrs

* PI Phone Number: (860) 486-4623

* PI Fax Number: (860) 486-4745

* PI Street Address: 97 North Eagleville Road

* PI City,State,Zip: Storrs, CT 06269-3136

* PI E-mail Address: hmarcus@mail.ims.uconn.edu

* PI URL Home Page: http://www.ims.uconn.edu/metal/faculty/marcus.htm

* Grant/Contract Number: N00014-95-1-0978

* Period of Performance: 07/01/96 - 06/30/01

_________________________________________________________________

Productivity Measures.

* Number of refereed papers submitted not yet published: 1

* Number of refereed papers published: 11

* Number of unrefereed reports and articles: 30

* Number of books or parts thereof submitted but not published:

* Number of books or parts thereof published: 2

* Number of project presentations: 40

* Number of patents filed but not yet granted:

* Number of patents granted and software copyrights: 2

* Number of graduate students supported >= 25% of full time: 6

* Number of post-docs supported >= 25% of full time: 1

* Number of minorities supported: 3

_________________________________________________________________

Objective of Research

The objective of the research was to establish the technical approaches required to use localized laser induced chemical vapor deposition as an approach for solid freeform fabrication (SFF). The three processing approaches to be developed are Selective Area Laser Deposition (SALD), Selective Area Laser Deposition Vapor Infiltration (SALDVI), and Selective Area Laser Deposition Joining (SALD-Joining) of ceramics.

Executive Summary

In this overall final report on the research into the three processing approaches, Selective Area Laser Deposition (SALD), Selective Area Laser Deposition Vapor Infiltration (SALDVI), and Selective Area Laser Deposition Joining (SALD-Joining) of ceramics only limited details are included. The reader can get greater detail in the over forty publications listed below that resulted from this research.

In terms of producing SFF shapes the SALDVI was the most successful. It was used to produce simple geometric parts of SiC or Si₃N₄ into SiC, Si₃N₄, metal and other powders. One of the more successful composites made was SiC deposited into Mo powder. The advantage of the layer by layer infiltration technique was clearly shown. The primary difficulties in the SALDVI processing were prevention of SALD overlay onto each layer, some residual porosity, and the very slow processing rate. The approach was characterized for the different materials in terms of laser power, scanning rate, preheat temperature, precursor gas pressure, precursor gases, raster geometry, line overlap, and laser wavelength, primarily using Nd:YAG and CO₂ lasers. As part of the effort finite element calculations of the processing parameters were made to help define the appropriate processing space.

To demonstrate the use of SALD and SALDVI together devices buried in a bulk material were produced. The most successful were C/SiC embedded thermocouples. This used SALDVI to produce the bulk SiC, SALD to coat the bulk SiC with Si₃N₄ as an insulator followed by SALD deposition of the C and SiC legs of the thermocouple. The thermocouple was then sealed with another SALD layer of Si₃N₄ and finally encapsulated in the multiple layers of SALDVI SiC. Samples with multiple thermocouples were also produced and demonstrated. This showed the wide potential of the combined SALD/SALDVI processing.

The third approach of SALD joining was fully demonstrated. SALD deposition of SiC deposited from the tetramethylsilane gas precursor joined two pieces of SiC. Similar results were obtained for Si₃N₄. This ability to "weld" ceramics with the same material to be joined has great promise. The primary difficulty is associated with the limited throwing power of the approach due to the surface being hotter than the inside of any joint geometry used.

A Solid Freeform Fabrication laboratory was designed and assembled, with a starting point of an empty room. Three unique and fully functional computer controlled gas phase laser-based SFF systems are presently in place. Overall the SALD, SALDVI and SALD Joining were all demonstrated and many of the processing parameters worked out. The potential availability of gas precursors for most materials makes them potentially a very strong possibility for SFF of many complex materials. Overall there will be a great deal of additional effort necessary to make the processes robust.

_________________________________________________________________

Summary of Technical Progress.

Equipment

A Solid Freeform Fabrication laboratory was designed and assembled, with three SFF systems presently in place. Two systems have Powder Delivery System (PDS) units for spreading out a powder substrate, necessary for the SALDVI process. One system offers a four inch by four and a half inch target area for large part production, while the other system has a one inch diameter target area. The third system is used for SALD/SALD joining and contains a SALD joining machine, incorporating a variable speed rotational device designed to work with tube structures. The large PDS sits in a 28 inch diameter, 15 inch tall vacuum system with a double-walled cooling jacket. The two other systems are eight inch diameter, eight inch tall vacuum chambers that have been nickel-plated to protect against the corrosive environments of some precursors and reaction by-products. Two in-situ chemical analysis tools have been implemented into the SALD/SALDVI systems. The first is a residual gas analyzer, and the second is a four point sampling emission spectrometer

Five lasers are operational, 25 and 50 watt continuous wave CO₂ (10.6 micron wavelength), 50 and 150 watt Nd:YAG (1.064 micron wavelength), and a 6 watt harmonic generating pulsed Nd:YAG (1.064, .532, .355, .266 micron wavelengths at the fundamental and 2nd, 3rd, and 4th harmonic wavelengths, respectively). X-Y translational equipment, optics, electronic components, three computers and associated software are all utilized in laser scanning. Ancillary laboratory equipment needed for proper laboratory procedures and safety has also been acquired and put to use, including a glove box for liquid precursor handling, a micron filter vacuum for powder cleanup, and compressed gas cylinder equipment.

An integrated temperature closed loop and scanning motion control computer program has been designed and refined in-house for the SFF systems. The program was written in Visual Basic. The closed loop uses the output from an emissivity-measuring infrared pyrometer to monitor temperature at the reaction zone of the laser beam. The program compares this readout to a user-defined temperature and makes adjustments in laser power by altering the input voltage to the laser controller with an analog voltage output card. Concurrent with this operation, the Visual Basic program directs the motion control hardware. Custom motion programs can be designed and inputted into the program. Data acquisition of the processing parameters such as laser power, pyrometer temperature and emissivity, and other signals is included. These processing control programs also have the capability to generate 2-D scan patterns automatically from 3-D computer geometrical models.

Technical Results

The specific research programs at the UCONN SFF lab focus on ceramic fabrication from the gas-phase reactions. Fundamental understanding of the SALD and SALDVI processes are essential to applying the technology to real world applications. The results below are summaries of technical progress in the three SALD research areas. For greater details, the reader is referred to the relevant publications.

SALDVI

The SALDVI work is focused on fabricating SiC bulk shapes by infiltration of SiC powder particles with vapor deposited SiC derived from tetramethylsilane (TMS) gas, although other starting powders (carbides, nitrides, oxides, metals) were also investigated. Powders of different size were mixed to vary the packing fraction and surface area. Laser scanning was performed using a range of gas precursor pressures using different temperature distributions and local heating times. Quantitative metallography was used to measure the variations in the amount of infiltration with position in the samples for each powder and processing condition. A comparison of the measured infiltration profiles shows that the properties of the starting powder influence the vapor infiltration in the powder layer. The variations in the measured infiltration profiles were analyzed and correlated with the variations in the physical properties of the starting powders to obtain a general predictive model of the process. The results show that solid density of more than 90 percent can be obtained in a region within 250 microns of the free surface of the powder layer under certain processing conditions. Process control was enhanced to allow automatic control of the heating and cooling rate whenever the laser beam is turned on or off. The capability to preheat the powder bed and SALDVI workpiece was also added.

The feasibility of fabricating SiC cermets by SALDVI was investigated, particularly by combining vapor deposited SiC with copper, molybdenum, and nickel powder layers. The microstructures of the three SALDVI cermets reveal varying degrees of reactions between the metal powders and the vapor deposited SiC matrix. Both the Cu and Ni powders react with vapor deposited SiC during the SALDVI process at all of the processing temperatures (1000-1200 C) and heating times (200-800 s) examined here. For the Cu powder, particles at least 20 mm in diameter had reacted to form a Cu silicide phase even at the shortest heating time of 200 s and lowest processing temperature of 1000 C. Further evidence of Cu silicide formation was observed in x-ray diffraction measurements at 1000 and 1200 C target temperatures. These results indicate that Si is a fast diffuser in Cu. In the Ni case, the extent of the reactions progressed more notably with increasing processing temperature and heating duration. X-ray diffraction (XRD) and energy dispersive x-ray spectroscopy (EDS) measurements indicate the formation of the Ni₂Si phase, as well as additional phases not yet identified. For the Mo/SiC cermet, no Mo-Si interdiffusion was detected by XRD or EDS for the 1000 C target temperature. This is not surprising as the melting temperature of Mo is 2610 C, so diffusion would be expected to occur slowly at 1000 C. The Mo/SiC cermet shows the highest bending strength of 50 MPa of the three cermets. This behavior is attributed to the close thermal expansion match between the two materials resulting in few particle debonds compared to the Cu and Ni cases. Figure 1 shows a typical as-fabricated single line, three layer, 20 mm long SALDVI Mo/SiC cermet (a) and a SEM image of a polished cross-section through its centerline region (b). The Mo particles appear well bonded to the SiC matrix and the solid density exceeds 95 % in this region, as measured by quantitative metallography.

Figure 1. As-fabricated three layer, single line macrostructure (a) and SEM image of the microstructure of Mo/SiC SALDVI cermet.

The role of the starting powder on the development of the final SALDVI structure was also investigated. Silicon carbide was deposited by SALDVI from the gas precursor tetramethylsilane, Si(CH₃)₄, into loosely packed powder layers of SiC, ZrO₂, WC or Mo. Layered samples were fabricated for each powder material using both single line (bar) and multiple line (rectangle) laser scan patterns and 10 Torr Si(CH₃)₄, 2.5 mm/s scan speed, 1000°C target temperature, and 120 mm layer thickness. Samples of SiC and ZrO₂ are prone to surface cracking in the bar geometry, and cracking and delamination of layers in the rectangle geometry. Samples fabricated with Mo powder have no cracks or delamination defects in either bar or rectangle geometry as well as a better surface appearance. Figure 2 shows a Mo powder/SiC matrix rectangle with 6 layers (a) as-fabricated and (b) cross-section across layers. Thermal stress plays a major role in cracking in bar and rectangle samples of SiC matrix/SiC powder and SiC matrix/ ZrO₂ powder made by SALDVI. The effective thermal conductivity (k_eff )of the SALDVI workpiece correlates well with the severity of thermal cracking. The lower k_eff the more severe the cracking. SiC matrix/Mo powder samples have the highest k_eff, and show no cracking. The irregular shape of the SiC and ZrO₂ powders, as well as a higher surface roughness, could provide stress risers in the matrix during processing, increasing their susceptibility to thermal cracking compared to the spherical Mo powder. The type of powder was found to affect the surface appearance and internal structure of SiC matrix multiple layer SALDVI rectangles. Samples with SiC and ZrO₂ powder show a porous surface appearance due to displacement of the powder during processing. Mo and WC powder samples have a dense surface and continuous solid material across adjacent layers. The lower density of the SiC and ZrO₂ powders and convective effects in the gas are the likely cause of their poor structure and rough surface appearance.

Figure 2. Mo powder/SiC matrix rectangle with 6 layers (a) as-fabricated and (b) cross-section across layers.

A 3D finite element model was also developed to simulate the SALDVI of silicon carbide. The model predicts the laser input power and the distribution of vapor deposited SiC within the powder bed as well as on the surface of the powder bed (SALD). The model includes closed-loop control of the laser power to achieve a desired target processing temperature on the top surface of the power bed. This model considers a moving Gaussian distribution laser beam, temperature- and porous-dependent thermal conductivity, specific heat and temperature-dependent deposition rate. A schematic of the model is shown in Figure 3. The simulation results compare well with experimental data as shown in Figure 4.

Figure 3. Finite element model of the SALDVI process.

Figure 4. Comparison between the experimental and simulated solid fraction distribution in depth direction under the center of the laser beam after the scanning process at x=7 mm.

The simulation results of the incident laser power history and the distribution of vapor infiltrated SiC in the powder bed agree fairly well with the experiments. As the laser scanning rate decreases, the incident laser power increases. As the laser scanning rate decreases, the deposition of SiC both within the powder bed as well as on the surface of the powder bed (SALD) increases (Figure 4). Uniform temperature on the top surface of powder bed during laser scanning process needs non-uniform laser power at the initial scanning period and nearly uniform laser power elsewhere. These simulation results offer guidelines for further experimental studies of the SALDVI process.

SALD and integrated SALD/SALDVI

SALD research looked at multiple material deposition. Specifically, silicon nitride, from TMS and ammonia, and silicon carbide and carbon, from TMS and acetylene respectively, were examined. The silicon nitride was deposited as a thin film, and analyzed with respect to its insulative quality. The deposition was performed at low ammonia pressures, approximately 10 to 20 torr, using a carbon dioxide laser. The successful formation of the nitride layer countered previous notions of not being able to use ammonia gas precursors with a CO₂ laser. Silicon carbide and carbon were deposited in line formations, with the resulting electrical properties coming under scrutiny. The line resistances and emf response to temperature changes were tested with respect to varied line widths (a function of scan speed and laser beam size) and deposition temperature field. These SALD/SALDVI investigations were applied to an embedded sensor project sponsored by DARPA. The DARPA program studied the feasibility to fabricate a ceramic matrix with an operational, in-situ thermocouple formed inside the matrix structure in one continuous process. An example of a SALD/SALDVI silicon carbide/carbon thermocouple in a silicon carbide matrix, along with its temperature and emf response, can be found at the UCONN website.

It was found that the experimental results are in excellent agreement with the theoretical thermodynamic predictions. With the use of acetylene (C₂H₂), tetramethylsilane (TMS), and a mixture of TMS and ammonia, graphite, SiC, and Si₃N₄ were successfully deposited, respectively. Strong temperature dependency of the SALD products in morphology, composition, crystal structure and size, growth kinetics and relevant properties were revealed. The predicted carbon co-deposition and the role of hydrogen gas in eliminating this co-deposition in SiC or Si₃N₄ were experimentally confirmed by a Raman scattering study. The functional test on the fabricated device showed that the embedded SiC/C thermocouples exhibited stable and repeatable response to temperature variations. Thus, the overall results indicated that it is feasible to embed the in-situ sensors within a ceramic matrix using the combined SALD/SALDVI techniques.

SALD-Joining

Another area of investigation in the SFF laboratory is ceramic joining by SALD, a patented process for substrate-involved SALD deposition. The joining of ceramic parts is accomplished by using a filler material deposited from the gas-phase reaction in SALD. In this manner, the ceramic joint can be tailored to match the material composition of the constituent ceramic part to be joined. The initial efforts focused on joining clay-bonded silicon carbide (approximate 75 to 80% density) tubes with silicon carbide deposited from TMS precursor. The adhesion of the SALD material to the tube is excellent. In fact, the interface between the SALD material and the tube is inconspicuous. The success of joints in this initial phase has been inconsistent. While the adhesion is outstanding, successful connective bonding of the SALD material across the joint has been challenging. An example of two clay-bonded silicon carbide tubes connected by the SALD joining method can be found at the UCONN website.

Investigation on this project has yielded several successfully joined silicon carbide tube structures using silicon carbide deposited filler from the gas-phase pyrolosis of tetramethylsilane. Adhesion of the deposit to the tube surface remains excellent. Bend testing of joined samples produced poor strength values, likely due to the butt-end nature of the joint geometry. This leaves an existing 'crack' at the joint seam the length of the tube wall. Beveling of the tube ends has become a central focus to achieve stronger joints, not only to reduce the initial crack length but to also increase the surface area of tube/deposit adhesion interface. The hermetic quality of the several joints was examined, compared to a monolithic standard of the base tube used. The SALD joints held a steady-state vacuum (in Torr) within a factor of 10 of the vacuum held by the standard. Improvements in the deposit morphology, from a billowy, porous nature to a more laminar deposition, are expected to improve these hermeticity values.

Preheating the silicon carbide substrate during the SALD deposition process is a controlling parameter for producing high density, high purity, defect-reduced silicon carbide filler. The preheating reduces the thermal gradient occurring during the joining process. SALD tube joint structures show greatly reduced mechanical bend strengths compared to monolithic tube standards. The origins for the lower strength are the initial crack found at the joint seam of the butt-end tube configuration and the reduction in the cross-sectional area of the SALD filler deposit compared to the full tube cross-sectional area due to the poor ‘throwing power’ of the laser associated with localized CVD. The SALD gas-phase decomposition of tetramethylsilane, with and without hydrogen, deposits nano-crystalline silicon carbide. Characterization of the SALD silicon carbide filler material shows deviations from the expected beta-polytype in the XRD, NMR and TEM patterns that are attributed to twinning faults. This is the first evidence of twins in SALD silicon carbide but is consistent with the very low stacking fault energy for silicon carbide.

_________________________________________________________________

Transitions and DOD Interactions.

Briefing on SFF applications to future naval operations to Fellows of the Chief of Naval Operations' Strategic Studies Group, Capt. T.D. Glass USN, Capt. D.R. Ellison USN, and LCDR S.L. Kruppa USN, at Newport, RI April, 1996

Presentation on "Laser Solid Freeform Fabrication: SLS, SALD, SALDVI" at the ONR review meeting in Woods Hole, MA, May 29th-30th, 1996

Presentation on "Solid Freeform Fabrication at the University of Connecticut" at the ONR annual review meeting in Woods Hole, MA, June 16th , 1997

Presentation of research programs and tour of the SFF laboratory for Dr. William Coblenz of DARPA, August 8th, 1997

Dr. Marcus took a 4 day trip on the USS Tunney submarine from San Diego, CA to Seattle, WA, September, 1997

Breifing on SFF at Navy War College, Newport, RI, 1998.

Sustained contacts and interactions with Bettis and Knowles Atomic Power Laboratory on SALD, SALDVI, and SALD joining of ceramics.

Presentations on SALD SFF techniques to Carderock Division of NSWC, 2000.

_________________________________________________________________

Software and Hardware Prototypes.

1. Prototype Name: SALD and SALDVI System Prototypes

+ URL:

+ Availability: Available in SFF laboratory, Institute of Materials Science, UCONN

+ Description: Hardware and software of performing gas phase SFF of structures and infiltration of powder layers.

Invited Presentations

Presentations on "Selective Area Laser Deposition: A Gas Phase Solid Freeform Fabrication Approach" and "Solid Freeform Fabrication Processing Using Gas Phase Approaches" at the ASM/TMS Materials Week conference, Cleveland, OH, October 30th-November 2nd, 1995

Presentation on "Solid Freeform Fabrication Using Powder and Gas Precursors" at the Dept. of Mechanical Engineering, Aeronautical Engineering and Mechanics Colloquium series, Renssalaer Polytechnic Institute, November 17th, 1995

Presentation on "Solid Freeform Fabrication" at the Institute of Materials Science Associate Program Annual Meeting, University of Connecticut, May 23rd, 1996

Presentation on "Solid Freeform Fabrication at The University of Connecticut" at the Solid Freeform Fabrication Symposium at The University of Texas at Austin, August 12th-14th, 1996

Presentation on "Solid Freeform Fabrication of Powders Using Laser Processing" at the PM2TEC conference, Washington, D.C., June 19th, 1996

Presentation on "Solid Freeform Fabrication of Ceramics" at the Society of Manufacturing Engineers conference, Newton, MA, October 23rd, 1996

Presentation on "Solid Freeform Fabrication: An Overview" at the ASME/MED Symposium on Rapid Response Manufacturing, Atlanta, GA, November, 1996

"Recent Advances in SALD and SALDVI," Seventh International Conference on Rapid Prototyping 1997, San Francisco, March 31-April 3, 1997

"Gas Phase Solid Freeform Fabrication at the University of Connecticut," MRS Annual Meeting, Solid Freeform Fabrication Session, San Francisco, March 31 - April 4, 1997

"Selective Area Laser Deposition (SALD) of Titanium Oxide," 6th European Conference on Rapid Prototyping and Manufacturing 1997, Nottingham, UK, July 1-3, 1997

Solid Freeform Fabrication Symposium 1997, Austin, TX , August 11-13, 1997, 4 presentations on the following topics: a) The Use of VRML to Integrate Design and Solid Freeform Fabrication, b) Gas Phase SFF Control System for Silicon Nitride Deposition by SALD/SALDVI, c) Fabrication of In-Situ SiC/C Thermocouples by Selective Area Laser Deposition, d) Net Shape Functional Parts Using Diode Laser

"Selective Area Laser Deposition Joining of Silicon Carbide," American Ceramic Society 1997 Fall Meeting, San Francisco, October 12-15, 1997

Workshop on Layered Manufacturing, Oxford, England, June, 1998, Invited Speaker

"Gas-Phase Selective Area Laser Deposition (SALD) Joining of SiC Tubes with SiC Filler Material", 1998 Solid Freeform Fabrication Symposium, August 10th 12th, 1998, Austin, TX

"Preparation and Properites of In-Situ Devices Using the SALD and SALDVI Techniques", 1998 Solid Freeform Fabrication Symposium, August 10th-12th, 1998, Austin, TX

"Properties of SiC/C Thermocouple Device Made with SALD and SALDVI Techniques", 1998 American Ceramic Society Meeting, May 3rd -May 6th, 1998, Cincinnati, OH

"Selective Area Laser Deposition of Silicon Nitride", 1998 American Ceramic Society Meeting, May 3rd -May 6th, 1998, Cincinnati, OH

_________________________________________________________________