Part 2: Adhesion evaluation

Author: Donald M. Mattox

Ideally adhesion tests should duplicate the stresses that appear at or in the interface and near-interface region. These stresses may differ through the life cycle of the coating (during deposition, immediately after deposition, subsequent processing, storage, and service). Adhesion measurement techniques should subject the coatings to reproducible stresses that correlate to satisfactory performance of the coating/substrate structure. The testing procedure(s) may include several tests to evaluate the adhesion under different stress conditions. Accelerated stressing (example: elevated temperature) may be used to simulate long-term conditions, though this must be done with care. Good adhesion may be measured quantitatively such with a pull-test and/or by observation of the behavior of the coating under stress such as with the scratch test. Often adhesion tests are used as a part of the verification that the process is producing a reproducible product (i.e. used as a comparative test). The adhesion tests may be done on a “dummy” part or a “witness” substrate that is representative of the production part (same material, processing, etc.). Most tests are “destructive” since they introduce damage to the coating or contamination on the coating surface.

The evaluation of adhesion may begin with studying the nucleation of the vacuum deposited material on the substrate of interest. This may be done in an electron microscope, by electrical conduction, or by optical transmission. Suffice to say that the higher the nucleation density the better the adhesion that may be attained since it shows chemical reactivity between the depositing atoms (adatoms) and the substrate material (see later blog – “Interface formation and its effects on adhesion”). Nucleation studies do not provide a practical means of evaluating adhesion but may be useful in understanding the adatom/surface interaction and in evaluating surface contamination effects on film growth (see later blog – “Contamination effects on adhesion”).

Observation of such deadhesion patterns while the parts are still in the fixturing may show high and low regions of stress and stress anisotropy due to fixturing (non-equivalency of positions), variation of energetic particle (ions or high energy reflected neutrals) bombardment during deposition, or the depositing flux angular distribution (texturing).

After the parts have been removed from the deposition chamber the adhesion may be evaluated by adhesion tests. One test that can be done easily is the “Mattox bad breath test.” In this test a person breathes on the coating to give moisture condensation. If the coating has a high intrinsic stress level the moisture may accelerate fracture propagation and deadhesion occurs giving blisters or flakes. If the coating cannot pass this test it will fail on more stringent adhesion testing. Obviously the uninformed observer attributes the failure to the bad breath of the tester.

There are literally hundreds of adhesion tests that have been described in the literature [1-5]. Figure 1 shows some of the most common tests.

The tape (peel) test uses an adhesive tape applied to the film and then peeled from the surface [6]. Usually the coating is scratched before the tape is applied to provide a fracture initiation site(s). This test has many variables (type of tape, angle of peel, rate of pull, etc.) and reveals poor adhesion but does not quantify better adhesion (i.e. it is mostly qualitative). It has the advantage of revealing spots of poor adhesion (“pullouts”) due to contamination such as dust particles. The pullout material is left on the tape.

The pull (tensile) test uses a wire or golf-tee-shaped stud bonded (glued, soldered, ultrasonic bonded) to the coating surface and pulled to failure under tensile stress. This test has the problem that the bonding process may introduce unknown stresses particularly near the edges. A variation of this test is the shear test that uses a blade to push on the side of a bump bonded to the surface. The shear test may also be performed in a lap-shear configuration. The wire bonded pull test maybe used in a non-destructive mode by just pulling to a specified value.

The scratch test is performed by moving a progressively loaded stylus over the surface until fracturing or deadhesion is detected optically [7-9]. The test is very dependent on the stylus shape and material and the physical/mechanical properties of the coating material. The scrape test is a more crude form of the scratch test and might be called a wear (abrasion) test. To further characterize the behavior of a scratch the acoustic emission during scratching may be monitored.

The indentation test uses a progressively loaded indenter and detecting the fracturing and spalling of the coating around the indenter impression. The indentation test is used most frequently for hard coatings. The indentation test is dependent on the physical and mechanical properties of both the coating and the substrate material though the use of microindentation techniques may minimize the effects of the substrate material.

The adhesion of a film or coating to a substrate may increase or decrease with subsequent processing, environments or with time. This change in adhesion may be caused by a number of effects such as diffusion to or away from the interface, stress relief, void formation in the interfacial region, deformation by repeated stressing, chemical effects, etc. Accelerated aging may be used to try to duplicate some or all of these effects but consideration must be given to the fact that interactions between mechanisms may provide misleading adhesion results. For example, heating may increase diffusion rates but may also give stress relief.

Conclusion

 A reproducible adhesion test program can give a useful comparative measure of the practical adhesion* to help ensure that the deposition process is reproducible and the product is adequate for the function and service life of the product. Adhesion measurement values should never be presented as an absolute or “basic” [2,3] property of the system.

* The term practical adhesion is used to include failure in the interfacial region or in “near-by” substrate or coating material [10].

References

1. D.M. Mattox, “Adhesion and deadhesion,” Ch. 12 in Handbook of Physical Vapor Deposition (PVD) Processing, 2^nd edition, Donald M. Mattox, Elsevier (2010)
2. K.L. Mittal, “Adhesion measurement of films and coatings: a commentary,” pp. 1-15 in Adhesion Measurement of Films and Coatings, K. L. Mittal editor, VPS (1995)
3. ASTM Proc. of Conf. on Adhesion Measurement of Thin Films, Thick Films and Bulk Coatings (STP 640), edited by K.L. Mittal, ASTM, (1978)
4. K.L. Mittal, “Adhesion measurement of thin films” Electrocomponent Sci. Technol., 3, 21-42 (1976); References to adhesion testing history and prior review articles
5. Adhesion Measurement Methods – Theory and Practice, Robert Lacombe, CRC, Taylor and Francis (2006)
6. ASTM D3359-17 “Standard methods for rating adhesion by tape test”
7. P.A. Steinmann and H.E. Hinterman, “A review of the mechanical tests for assessment of thin-film adhesion,” J. Vac. Sci. Technol. 7(3) 2267 (1989)
8. G. Aldrich-Smith, N.M. Jennett and J. Housden, Mechanical Property Measurement of Thin Films and Coatings: A Round Robin to Measure the Adhesion of Thin Coatings, VAMAS TWA 22, NPL Report DEPC-MPE 001 (Sept. 2004)
9. H.E. Hinterman, “Characterization of surface coatings by the scratch adhesion test and by indentation measurements,” ‘Fresenius’ J. Analytical Chem. 346(1-3) 45-52 (1993)
10. Donald M. Mattox, “Thin-film adhesion and adhesive failure,” pp. 54-62 in [3]

 

 

Part 1: Introduction

Author: Donald M. Mattox

Vacuum deposition may be defined as the deposition of a film and coating by atom-by-atom or molecule-by-molecule on a suitable substrate surface. The vacuum is necessary for the transport of the atom/molecule from the source to the substrate by having a long mean path for collision in the vapor state and for controlling the partial pressure(s) of gaseous reactive species in the deposition environment.

Vacuum deposition may be accomplished by: 1) deposition of vapors from the source material(s) with no reaction on the substrate, 2) reaction of co-deposited materials on the substrate or 3) reaction of the deposited material with the gaseous environment (“reactive deposition”). Vapor sources include: 1) thermal evaporation or sublimation from solid or liquid, 2) sputtering from a solid cooled surface, 3) vaporization from an arc moving over a solid surface, or 4) decomposition of a chemical vapor precursor. In some cases a low pressure plasma may be used to “activate” reactive species and/or provide ions for concurrent energetic particle bombardment of the depositing species (sometimes called “atomic peening”).

The terms “vacuum deposited thin film” and “vacuum deposited coating” are used in the literature without any definition of the thickness or mass per unit area. Early use of vacuum deposition was to deposit films for reflecting and antireflecting films both of which require film thicknesses on the order of a quarter of the wavelength of visible light or about 150 nanometers. This seems to be a logical upper limit for the thickness of a vacuum deposited thin film. At this thickness the physical and mechanical properties of the film often are not important for adhesion. As the vacuum deposit becomes thicker it would be logical to call them thick films but this terminology has a specific meaning in hybrid microelectronics. Therefore the term coating seems to be a good compromise. The properties of the coating may have a significant effect on the adhesion of the coating to the substrate.

Adhesion is the mechanical strength joining two different materials (designated A and B) and is a fundamental requirement of most vacuum deposited films and coatings. In a few cases such as fabrication of free-standing structures, the adhesion may be made deliberately low. The loss of adhesion is sometimes called deadhesion in adhesion technology.

In practical (“apparent”) adhesion the loss of adhesion is by fracture and may be at the actual interface of A&B or may be in A or B (“near interfacial region”) or in an interfacial material A/B that is developed between A and B. This interfacial material is sometimes called the interphase material. The A/B interphase material may be an alloy, a mixture, or a compound of the A&B material or may be a reaction product of the A or B material with the ambient gas/vapor or codeposited material such as an oxide or carbide (see later blog – Types of interfaces)

Fractures are caused by mechanically stressing the interfacial region thus providing energy for fracture propagation. Fractures may be classed, as brittle fracture where there is little plastic deformation and comparatively little energy is needed to propagate a crack in the “brittle” material. If the fracture involves significant plastic deformation to separate the surfaces the failure is called ductile fracture and requires significantly more energy to propagate the failure. In brittle fracture often very little energy must be added to propagate the crack while in ductile fracture more and continuous energy must be supplied to propagate the crack. A material that is both strong and ductile is often called a “tough” material.

Corrosion may enable fracturing by embrittling the material at a crack tip such as is the case with moisture at the crack tip in glass, (a glass-cutter will scribe the glass, then wet the scratch, before applying the stress needed to break the glass) [1] or by “wedging” of the crack by corrosion products.

Fractures may be “blunted” by having the propagating fracture encounter a “tougher” material or by having to change the direction of propagation because of a change in the stress vector.

In vacuum coating, generally fractures are caused by mechanical stresses applied to the interfacial region. If the stress is normal (perpendicular) to the plane of the interface it is call a pure tensile stress. If the stress is parallel to the plane of the interface the stress is call pure shear stress. Of course the applied stress usually will have both a tensile and a shear component due to the direction of the applied stress and/or the roughness of the interfacial region.

The stress that appears at the interface is composed of intrinsic stress and extrinsic stress. The intrinsic stress is the “grown-in” (growth) stress due to the deposition conditions and may vary through the thickness of the deposit. The extrinsic stress is due to external factors such as differences in the coefficient of expansion of the substrate and coating material. The total value of these stresses depends on the thicknesses of the stressed regions. Very thin films may have high stress but still be adherent whereas thick deposits may have such a high total stress that there will be spontaneous deadhesion. For example chromium is often used as a “glue” layer for metallization on glass. Vacuum deposition by thermal evaporation will give a high tensile stress and up to several tens of nanometer thickness will exhibit good adhesion while hundreds of nanometers will cause fracturing in the near-surface region of the glass.

Fractures in brittle materials often have point of fracture initiation where there is a defect or where stresses are concentrated. Microcracks or scratches in the surface of a brittle material, such as glass, are an example of fracture initiation sites.

Deadhesion may be local, creating “pull-outs” or “pinholes.” Common sources of pinholes are dust particles and “spits” or “macros” from arc vapor deposition. The pinholes may act as initiation sites for further fracturing.

Other types of time-dependent stresses such as interfacial corrosion may cause deadhesion. Deadhesion by weakening of the adhesion bonding may also result from diffusion to or from the interfacial region by chemical species or structural defects such as voids. In some cases diffusion may result in strengthening of the adhesive bond.

Designing for adhesion in vacuum deposition requires an understanding of the substrate and coating materials involved, their interaction, and how the deposition process influences their properties. This in turn requires an understanding of the condensation and nucleation of the depositing material, the interface formation and the nature of the film growth. Even monolayers of contamination may influence nucleation and growth so control of contamination prior to and during vacuum deposition is important in having a reproducible vacuum deposition process.

Reference

1. M. Tomozawa, “Fracture of glasses,” Annual. Rev. Mat. Sci. 26, 43-74 (1996)

 

 

Author: Donald M. Mattox

It all began with vacuum cadmium plating of high strength steel compressed gas bottles.

After serving as a meteorologist and Air Weather Officer in the USAF during and after the Korean War, I obtained a M.S. degree in solid state physics from the University of Kentucky in 1961 (see more in Endnote A).

I reported to Sandia Corporation in April of 1961 as a Technical Staff Member and began work as a solid-state physicist in the metallurgical engineering group. The Sandia “metallurgy group” was composed of a mix of material science people with specialties ranging from metallurgy, organic adhesives, surface cleaning, and ceramic fabrication, to process “troubleshooters” and technology transfer. In order to learn more about material science I began teaching a noon-hour course on material science for interested laboratory personnel – I was teaching the course when President John F. Kennedy was assassinated on Nov. 22, 1963.

The metallurgy department had an old CVC vacuum system that wasn’t being used. It had a glass bell jar with a cable hoist, a manually cranked high vacuum gate valve, and an oil diffusion pump. I had done some vacuum work at University of Kentucky so I decided to get it working.

My “bible” was Leslie Holland’s book Vacuum Deposition of Thin Films (1956) (Chapman Hall). I got the system into operation with a filament evaporator then added a DC sputtering capability. I controlled the pressure for sputtering by manually “cracking” the gate valve. My mass flow controller was a manual needle valve. I had a CEC LC-031 DC high voltage power supply that used mercury thyratron rectifier tubes. It could get to 5000 volts DC though it tended to arc internally at the high voltages. When the sputtering cathode arced the voltage would go down, the current would go up, the tubes would light up and the plates in the high voltage transformer would buzz. The only way to turn off a “hard arc” was to manually turn off the power supply – there was no arc suppression/quenching circuit in that power supply!

One of my first troubleshooting trips was to a contractor’s processing line for applying vacuum deposited cadmium (“VacCad”) plating to high-pressure gas bottles. They were having problems with the adhesion of the cadmium coating to the steel. The problem was simple – they were blowing off the part with “house air” and the air contained oil from the compressor. That got me thinking about the adhesion of vacuum deposited coatings [1]. When I discussed this with the metallurgists they maintained that to get good adhesion you needed solid solubility. Being a solid state physicist I thought just getting good atom-to-atom contact should give you good adhesion.

In the library I found that the field of high vacuum surface science studies on atomically clean surfaces really began in the mid-1950s when H. Farnsworth produced atomically clean surfaces by repeated sputtering and annealing in high vacuum [2,3]. I thought “What if you started depositing using thermal evaporation while you are sputtering the substrate surface. Might that not give a “clean” interface and good atom-to-atom contact?” So in one afternoon I tried vacuum evaporation while the sputter cleaning was still going on. It worked great! When I evaporated copper into the plasma there was a nice green glow indicating excitation/de-excitation of the copper atoms and probably ionization.

I decided that the definitive experiments should involve immiscible materials so I chose silver and iron, which are immiscible even in the molten state. I coated a series of iron tensile bars with silver and elongated them to test the adhesion. My “ion plating” technique gave adhesion greater than that detectable by the test.

I presented the work at a Gordon Research Conference on Adhesion in the summer of 1963. My paper (“Deposition by Exploding Wire and Ion Plating”) was scheduled for presentation on Thursday evening. On Wednesday I had yet to received permission to present the paper because a patent application was being made. On Thursday I received permission to present the paper. After the Thursday evening all-you-can-eat lobster dinner I was completely full of lobster and gave the paper. The paper and the subsequent discussion went on for 3 hours! The concept of ion plating was first published in the technical literature in 1964 [4] and the patent was issued in 1967 [5]. Of course in hindsight, there was very little ionization of the evaporated material in the plasma that I used – bombardment was mainly by the inert gas ions used for sputtering.

The 1967 (priority 1964) ion plating patent covered the use of both thermal evaporation and chemical vapor precursors as sources for the depositing material [6]. In 1969 (priority 1964) H.F. Sterling and R.C.G. Swann patented the use of chemical vapor precursors in an RF plasma with no substrate bias as a means of depositing coatings by what was later called plasma enhanced chemical vapor deposition (PECVD) [7,8].

Bob Culbertson of EIMAC/Varian expressed an interest in using the ion plating process for metallized ceramics at a low temperature and we metallized ceramics using a WCl₄ precursor [9]. Bob later patented ion plating for depositing low-electron-emission coatings (C and TiC) on electron tube grids [10]. One of the claims in his 1971 patent (filed 1968) used both a filament evaporator and a chemical vapor precursor to deposit coatings such as TiC. This was the first use of ion plating as a hybrid deposition process of PVD/PECVD. His patent was probably the first report on what came to be known as “reactive ion plating” [11]. Y. Murayama was the first to use the term “Reactive Ion Plating” in the peer-reviewed literature in 1975 [12]. The Japanese seemed to pick up the use of the term “ion plating” much more quickly than did others.

The term “ion plating” was used in the 1967 paper published in Electrochemical Technology and in the patent. The term came under some criticism and was discussed in the Journal of the Electrochemical Society in 1968 [13]. Over the years others have used different terminology for much the same process. These include: Ion Vapor Deposition (IVD), Ionized Physical Vapor Deposition (iPVD), Ion Assisted Deposition (IAD), Ion Beam Assisted Deposition (IBAD), Biased Activated Reactive Deposition (BARE), and Energetic Condensation.

The Sandia reactor group was developing nuclear reactors for providing pulses of neutrons; the SPR series of pulsed neutron reactors [14]. The SPR II reactor contained 105 kg of U-Mo alloy of which 93% of the uranium was ²³⁵U (more than the 64 kg that was in the gun-type “Little Boy” atomic bomb of WW II). When pulsed by bringing the components together, the uranium parts attained very high temperatures very quickly. In the SPR I reactor the uranium was protected by electroplated nickel and the nickel had to be periodically replaced. The old nickel had to be chemically stripped and the surface “pickled” all of which was undesirable since they lost some ²³⁵U in the process and the surface was roughened. They wanted a substitute for the electroplated nickel coating.

We ion plated samples of depleted uranium with aluminum and subjected them to corrosion tests, which it passed with flying colors [15]. We then coated some mock-ups of the reactor components. When it came to coating the real parts we had to strip the nickel and then pump down overnight – all the while with an armed guard by our side. The coated parts worked very well [16].

In analyzing the coating substrate-coating interface we found that there was an appreciable layer of UAl₂ formed at the interface due to the clean interface, the heating from the sputter cleaning, and thermal diffusion during the deposition.

While we were working on the uranium project we were also developing other fixtures for ion plating, several of which were described in the patent. These included a “cold finger” fixture for keeping the back-side of the sample cooled to liquid nitrogen temperature during deposition. We used that technique to deposit metallization electrodes on semiconductor materials to avoid “pipe diffusion”.

We were also using a rotating cage fixture (patterned after the barrel plating technique used in electroplating) for coating small parts [17]. Ion plating using the cage fixture was used to vacuum coat ultra-high vacuum components with gold to reduce hydrogen outgassing.

The ion plating patent also covered the use of a grid in front of an insulator or irregular surface to provide bombardment. We used this arrangement to coat ceramics and polymers [18,19]. Ion plating was also used to coat metals with chromium for glass-to-metal sealing [20].

The use of ion plating for aluminum coating spread quickly through the AEC nuclear weapons complex for coating uranium and plutonium as well as silver on beryllium for diffusion bonding [21]. R.T. Bell and J.C. Thompson at the Oak Ridge Y-12 Plant used ion plating as a “strike” for electroplating in 1973 [22] as did others [23].

In January 1965 Time Magazine ran a short news item called “Plating with a permanence” about the ion plating process and the article caught the attention of McDonnell-Douglas. They sent a group to look at what we were doing. They saw the rotating cage fixture and they took the idea back and developed their “IVADIZING” machine for coating fasteners for the aerospace industry [24-26]. This became the basis for the IVD (“ion vapor deposition”) process, which is widely specified for military and aerospace applications [27].

T. Spalvins and Donald Buckley of NASA (Cleveland) picked up the ion plating process for their studies of lubrication in space using low-shear metal films [28]. Based on the NASA work General Electric began using ion plating for coating their X-ray tube bearings with low shear metals for lubrication in vacuum. The dissemination of the ion plating process was aided by the fact that the patent was in the public domain.

In the sixties, RF sputtering began to be used for sputtering dielectric materials. I converted an old diathermy machine for RF sputtering. We began to sputter dielectric materials [29]. Several months after we began rf sputtering FCC inspectors showed up! We had been broadcasting all over the radio spectrum and my laboratory was right by the runway at Kirkland AFB. We were interfering with the aircraft-to-ground communication. Since we were only doing it sporadically they had a hard time finding us. I was shut down – but the “powers that be” at Sandia then purchased an RF system for my group with proper shielding.

The mid-sixties was also the time that scanning electron microscopy (SEM) was coming onto the scene as a commercially available analytical tool [30] (1965 – Cambridge Scientific Instruments – “Stereoscan”). I began to look at the fracture cross-section morphology of thick ion plated coatings. It was immediately obvious that ion bombardment during deposition was disrupting the columnar growth and densifying the deposit as well as affecting the surface morphology [31,32]. For a period of time the densification by bombardment was called “atomic peening” [33] though this descriptive terminology has fallen out of favor.

At about the same time as I was doing the ion plating work using thermal evaporation, several others were doing similar work using “bias sputtering” [34-37]. However, they did not seem to do much analytical work on the films other than that related to the function they were studying. Some authors later called this process “sputter ion plating [38].

In the early work on ion plating it was found that ultrafine particles were deposited on the interior walls of the chamber but not on the negatively biased substrate [39]. The particles that formed in the plasma attained a negative charge and were repelled by the negative potential on the substrate. Later the same phenomenon was observed in high rate sputtering [40,41].

In the case of reactive materials such as titanium, the deposited ultrafine particles were stable when exposed to air, apparently because of a very thin surface oxide layer. When disturbed the fine particles would “burn.” To quote from Handbook of Physical Vapor Deposition Processing, 2^nd edition (2010) (p. 322, section 9.9.2), Donald M. Mattox, Elsevier.

“In the early work on ion plating, the particles formed in the plasma and deposited on the walls were called ‘black sooty crap’ (BSC) and could be very combustible. One game was to ask an observer to wipe the particles off a window with a paper towel. When the window was wiped, the towel caught fire and a flame front moved over the interior surface of the chamber, which was covered with BSC.”

In hindsight I wish that I had done more studies on the charged ultrafine particles since 20 years later ultrafine (nano-) particles became an important field of study.

By the mid 1970s several papers had been published on how ion plating could be used to tailor the properties of thick deposits using both continuous and periodic bombardment during deposition. It was well established that forming the interface by beginning the deposition during sputtering cleaning gave good adhesion, bombardment with reactive ions allowed reactive deposition of dense compound materials, and bombardment during deposition could be used to control intrinsic stress, morphology, and crystallographic orientation of the deposited coating materials.

In the book chapter “Ion Plating Technology” in the first edition of Deposition Technologies for Films and Coatings, Development and Applications edited by R.F. Bunshah et al (1982) a rather extensive bibliography of papers on ion plating and related subjects was provided for the period 1963 to 1980 [42].

With the advent of arc vaporization in the early 1980s [43], and HIPIMS in the early 2000s [44] where a large percentage of the vaporized material is ionized to condensable “self-ions,” bombardment effects on the properties of coatings entered a new realm [45].

Some time in the early 1970s someone called attention to the fact that Bernard Berghaus, of plasma nitriding fame, had patented a process similar to “ion plating” in Europe in the mid-1930s [46]. His work in this area seems to have been an extension of his earlier work on ion nitriding and sputter deposition [47-49]. Berghaus apparently did not pursue any application of the process nor did he publish anything in the technical literature about the process that I could find. Leslie Holland didn’t discuss Berghaus’ coating work in his book Vacuum Deposition of Thin Films so it must not have been well known at that time (1956).

By the mid-70s my group became involved in solar energy (depositing selective solar thermal absorbers), fusion reactor technology (coating TOKAMAK limiters) and in developing a lithium source for the Sandia fusion PBFA fusion reactor experiments. Ion plating just became another “tool.”

Using ion plating for tailoring the coating stress was carried over into stress control for the deposition of thick Mo coatings for the BOLVAPS lithium ion source for Sandia fusion energy experiments. At low pressure sputtering the high energy reflected neutrals from the sputtering target may bombard the deposit and affect the stress in the deposit [50]. The technique of “pressure pulsing” for stress control in the sputter deposition of thick deposits was developed to tailor the stress in thick sputter deposited coatings [51]. Pressure pulsing relies on bombardment by high-energy reflected neutrals from the sputtering target at a low pressure environment rather than accelerating ions from a plasma.

In a 2013 publication Professor Allan Mathews, University of Sheffield, Sheffield, UK, credited Ion Plating with being the “First Wave” of “Plasma Assisted PVD Processing” [45].

Conclusion:

The quest for better adhesion by having a clean interface led to a serendipitous way to modify film/coating properties (optical, metallurgical, electrical, chemical, etc.) by energetic particle bombardment during deposition of PVD and PECVD films/coatings.

Acknowledgements:

From the mid-1960s and through the following years I had a number of technicians and several staff members who worked with me on ion plating projects. Notable among the technicians were: Frank Rebarchik, Ray Bland (who later became a Laboratory {shop} Supervisor), George Kominiak (who later became a Technical Staff Member), and Chuck Peeples. Notable among the technical staff members were: Bob Cuthrell, Ray Berg, Janda Panitz, Don Sharp, Art Mullendore, Jack Houston, and John Whitley. Special thanks to my supervisors who gave me the leeway to pursue ion plating projects. These include, in the early years: John McDonald, Charlie Bild, Lou Berry, Bill O’Neal and Dick Schwoebel. Thanks also to Lew Jones who took me under his wing when I first arrived at Sandia and taught me some of the nuances of troubleshooting at production facilities.

Endnote A: I graduated from Eastern Kentucky State University during the Korean War. I joined the USAF for four years to become an officer and be trained as a meteorologist. I attended two semesters of meteorology at Massachusetts Institute of Technology (MIT) (1953-54) before being deployed to Eglin AFB, FL as a weather forecaster.

After a year I was assigned to the 58^th Weather Recon Sq. at Eielson AFB, Alaska as an Air Weather Officer. My job was as an in-flight weather observer on synoptic weather fights that also included air sampling for detecting Russian nuclear tests. The flight routes were over the Bering Sea to Attu at the western end of the Aleutian Island chain and north to near the North Pole. I flew about 1500 hours in B-29 and then B-50 aircraft.

After leaving the USAF I went to graduate school at the University of Kentucky (Lexington, KY) under the GI Bill. I was in the Physics Department and was a graduate assistant in the Solid State Physics laboratory. My advisor was Professor Lee Gildart. In 1960 Dr. Gildart left unexpectedly and I was offered a position in the High Energy Physics lab counting cloud chamber tracks. No way was I interested in doing that! That sent me looking for a job.

I wanted to move to the Southwest so I traveled around CO, AZ, NM interviewing. I had noted that Sandia Corporation, a US Atomic Energy Commission (AEC) laboratory run by AT&T in Albuquerque, NM used bicycles to get around their Technical Area but I didn’t know much about what they did (about 6 months after I went to work for Sandia they quit using bicycles in the Tech Area). At their Personnel Office I said that I wanted to interview for a staff member position. It turned out that this was pretty unusual in that Sandia usually went to universities to recruit and had very few “walk-ins” for a technical staff member position.

Personnel had several people from the Metallurgy Group come over for an interview. Up to that time the metallurgy group had been “handbook engineers” but they wanted to expand into R&D. They asked “Can you do research?” Of course I said yes since I had co-authored several peer- reviewed papers while at the U of K. They offered me $750 per month salary – a magnificent sum compared to what I was making at U of K!

I had the option of going to work immediately outside the technical (“tech”) area in the “leper colony” and wait for my security clearance or return in about 6 months after I had a clearance. I chose to wait since I wanted to finish my Masters Degree [A-1]. This was the beginning of my work and sojourn at the Sandia National Laboratories until I retired in 1989 to become active as a consultant and course instructor in Management Plus, Inc. (MPI) with my wife Vivienne. MPI is a consulting and education training organization in the field of vacuum coating.

A-1. D.M. Mattox and L. Gildart, “Energy gaps in bismuth trioxide,” J. Phys. Chem. Solids 18(2-3) 215 (1961)

References:

1. D.M. Mattox and J.E. McDonald, “Interface formation during thin film deposition,” J. Appl. Physics 34, 2493 (1963)
2. H.E. Farnsworth, R.E. Schlier, T.H. George and R.M. Burger, “Ion bombardment cleaning of germanium and titanium as determined by low-energy electron diffraction,” J. Appl. Phys, 26, 252 (1955)
3. H.E. Farnsworth, “Preparation, structural characterization, and properties of atomically clean surfaces,” (AVS Welch Award presentation) J. Vac. Sci. Technol., 20, 271 (1982): doi: 10.1116/1.571282
4. D.M. Mattox, ”Film deposition using accelerated ions,” Electrochem. Technol. 2, 295 (1964); Sandia Corp. Development Report SC-DR-281-63 (1963)
5. Donald M. Mattox, “Apparatus for coating a cathodically biased substrate from plasma of ionized coating material,” USP 3329601 (priority date, Sept. 5, 1964; filed, Sept. 30, 1966; publication, July 4, 1967) (assigned USAEC)
6. D.M. Mattox, “Tungsten CVD in a gas discharge,” Sandia Corp. Development report, SC-DC-67-2026A (1967); presented at the “Conf. on Chemical Vapor Deposition of refractory metals, alloys, and compounds,” Gatlinburg, TN (Sept. 12-14, 1967)
7. Henry Frank Sterling and Richard Charles George Swann, “Method of forming silicon nitride coating,” USP 3485666 (priority date, 8 May 1964; filed, 3 May 1965; published, 23 Dec. 1969) (assigned International Standard Electric Corp.); British pat. 1104935; French pat. 1442502
8. R.C.G. Swann: https://ethw.org/w/index.php?title=Special:Search&search=The+Birth+of+Glow+Discharge+Chemistry+%28aka+PECVD%29+-+extensive+references&searchToken=chbbl6060oeonkahx2lto80hu – extensive references
9. R. Culbertson and D.M. Mattox, “High strength ceramic-metal seals metallized at room temperature,” 8^th Conf. Tube Technol. IEEE Conf. Record TK 6563, 3, U52, p. 101 (1966)
10. Robert D. Culbertson, Russell C. McRae, and Harold P. Meyn, “Nonemissive electrode structure utilizing ion-plated nonemissive coatings,” USP 3604970 (filed 14 Oct. 1968; published 14 Sept. 1971) (assigned Varian Assoc.)
11. Donald M. Mattox, “Short history of reactive evaporation,” pp. 50-51, SVC Bulletin, Society of Vacuum Coaters (Spring 2014)
12. Y. Murayama, “Thin film formation of InO₂, TiN and TaN by RF reactive ion plating,” J. Vac. Sci. Technol. 12, 818 (1975)
13. Discussion on “Adherence and porosity of ion plated gold,” C.F. Schroeder and J.E. McDonald, J. Electrochem. Soc., 115(12) 1255 (1968) – Reply by Donald M. Mattox
14. T.R. Schmidt and J.A. Reuscher, “Overview of Sandia National Laboratories pulse nuclear reactors”, SAND-94-2466C (1994); also CONF-941102-28 (Conference: Winter meeting of the American Nuclear Society (ANS), Washington, DC, 13-18 Nov 1994); OSTI ID: 10190030
15. R.D. Bland, J.E. McDonald, and D.M. Mattox, “Ion plated coatings for the corrosion protection of uranium,” Sandia Development Report SC-DR-65-519 (1965)
16. D.M. Mattox and R.D. Bland, “Aluminum coating of uranium reactor parts for corrosion protection,” J. Nucl. Mater., 21, 349 (1967)
17. D.M. Mattox and F.N. Rebarchik, “Sputter cleaning and plating small parts,” Electrochem. Technol., 6, 74 (1968)
18. D.M. Mattox, “Metallizing ceramics in a gas discharge,” J. Am. Ceramic Soc. 48, 385 (1965)
19. “High strength ceramic-metal seals metallized at room temperature,” R. Culbertson and D.M. Mattox, 8^th Conf. Tube Technol., IEEE Conf. Record TK 6563, 3, U52, p. 101 (1966)
20. D. M. Mattox, “Chromizing molybdenum for glass sealing,” Rev. Sci. Instrum. 37, 1609 (1966)
21. G. Mah, P.S. Mcleod, and D.G. Williams, “Characterization of silver coatings deposited from a hollow cathode source,” J. Vac. Sci. Technol., 14, 152 (1977)
22. R.T. Bell and J.C. Thompson, “Applications of ion plating in metals fabrication,” Oak Ridge Y-12 Plant, Y-DA-5011 (1973)
23. J.W. Dini, “Ion plating can improve coating adhesion,” Metal Finishing, 91(9) 15-20 (Sept. 1993)
24. K.E. Steube and L.E. McCrary, “Thick ion vapor deposited aluminum coatings for irregular shaped aircraft and spacecraft parts,” J. Vac. Sci. Technol., 11, 362 (1974)
25.  J. Carpenter, G. Kesler, A. Klein, L. McCrary, and K. Steube, USP “Glow discharge coating apparatus,” 3750623 (filed 11 Feb 1972; published 7 Aug. 1973) (assigned to McDonnell Douglas Corp.)
26. Kenneth E. Steube, “Glow discharge-tumbling vapor deposition apparatus,” USP 2926147 (filed 15 Nov 1974; published 16 Dec. 1975) (assigned to McDonnell Douglas Corp.)
27. MIL-DTL-83488D; “Coating; High purity aluminum,” (01 April 1999); superseding MIL-C-83488C, “Coating; Aluminum ion vapor deposited,” (01 Dec. 1978)
28. T. Spalvins, J.S. Przbyszewski, and D.H. Buckley, ”Deposition of thin films by ion plating on surfaces having various configurations,” NASA Lewis TN-D3707 (1966)
29. G.J. Kominiak and D.M. Mattox, ”Physical properties of thick sputter deposited glass films,” J. Electrochem. Soc. 120, 1535 (1973)
30. C.W. Oatley, W.C. Nixon and R.F.W Pease, “Scanning electron microscopy,” Adv. Electron Phys., 21, 181 (1965)
31. D.M. Mattox and G.J. Kominiak, ”Structure modification by ion bombardment during deposition,” J. Vac. Sci. Technol. 9, 528 (1972)
32. R.D. Bland, G.J. Kominiak, and D.M. Mattox, ”Effects of ion bombardment during deposition on thick metal and ceramic deposits,” J. Vac. Sci. Technol., 11, 671 (1974)
33. A.G. Blackman, Metall. Trans. 2, 699 (1971)
34. G.K. Wehner, ”Growth of solid layers on substrates which are kept under ion bombardment before and during deposition,” USP 3,021,271 (filed April 27, 1959, publication, Feb, 1962)
35. R. Frerichs,”Superconductive films by protected sputtering of tantalum or niobium,” J. Appl. Phys., 33, 1898 (1962)
36. L.I. Maissel and P.M.Schaible, “Thin films formed by bias sputtering,” J. Appl. Phys., 36, 237 (1965)
37. Orla Christensen, “Characteristics and applications of bias sputtering,” Solid State Technology, p. 39 (Dec. 1970)
38. N.M. Renevier, V.C. Fox, D.G. Teer, and J. Hampshire, “Coating characteristics and tribological properties of sputter-deposited MoS₂/metal composite coatings deposited by closed-field unbalanced magnetron sputter ion plating,” Surf. Coat. Technol., 127(1) 24 (2000)
39. “A short history of ultrafine (nano-) particles formed in vacuum,” Donald M. Mattox. pp. 54-56, SVC Bulletin, Society of Vacuum Coaters (Summer 2014)
40. G. Selwyn, J. McKillop, K. Haller and J. Wu, J. Vac. Sci. Technol., A8, 1726 (1990)
41. G. Jellum and D. Graves J. Appl Phys., 67, 6490 (1990)
42. “Ion Plating Technology,” Donald M. Mattox, Ch. 6, pp. 244 – 268, in Deposition Technologies for Films and Coatings: Development and Applications, edited by R. F. Bunshah et al, Noyes Publications (1982)
43. I.I. Aksenov, V.A. Belous, and V.E. Strel’nitskij, “Vacuum-arc surface modification and coating deposition methods in KIPT, Ukraine (Historical Review)” Proceedings 55^th Annual Technical Conference, pp. 3-8 Society of Vacuum Coaters (2012); also pp. 48-52, SVC Bulletin, Society of Vacuum Coaters (Summer 2012)
44. A.P. Ehiasarian, R. New, W-D. Munz, L. Hultman, U. Helmersson, V. Kouznetsov, “Influence of high power densities on the composition of pulsed magnetron plasmas,” Vacuum 65 (2) 147–154 (2002) doi:10.1016/S0042-207X(01)00475-4
45. Allan Matthews, “Plasma Assisted PVD: The Past and Present,” p. 24-27, SVC Bulletin, Society of Vacuum Coaters, (Fall 2013); also Keynote paper ISSP 2013 (12th International Symposium on Sputtering and Plasma Processes) July, 2013, Kyoto, Japan
46. “Improvements in and Relating to the Coating of Articles by Means of Thermally Evaporated Materials,” B. Berghaus, UK Patent# 510, 993 (priority date Aug. 26, 1937, filed Aug. 23, 1938, issued Aug.11, 1939)
47. B. Berghaus, “Improvements in and relating to the coating of articles by means of thermally vaporized material,” UK patent #510,993 (filed; publication 1938); also “An improved process for metallizing metallic articles by means of cathode disintegration in vacuum,” UK patent #520592-A (priority date, Nov. 19, 1937, publication date, Oct. 26, 1938).
48. Wilhelm Burkhardt and Rudolf Reinecke, “Method of Coating Articles by Vaporized Coating Material,” USP# 2,157,478 (priority date, June 17; 1936; filing date, June 15; 1937, publication May 9, 1939) (assigned to Bernard Berghaus) – thermal evaporation source
49. Wehnelt Eilhart, Hermann Ramert, Wilhelm Burkhart, “Coating of articles by means of cathode disintegration,” USP 2,239,642 (priority date , May 27, 1936; filing date, May 21, 1937; publication date, April 22, 194) (assigned to Bernard Berghaus) – sputtering source
50. J.A. Thornton and D.W. Hoffman, “Internal stresses in titanium, nickel, molybdenum and tantalum films deposited by cylindrical magnetron sputtering,” J. Vac. Sci. Technol. 14, 165 (1977)
51. D.M. Mattox, R.E. Cuthrell, C.R. Peeples and P.L. Dreike, “Preparation of thick stress-free Mo films for a resistively heated ion source,” Surf. Coat. Technol., 36, 117-124, 1988

 

 

 

 

 

 

Author: Donald M. Mattox

Many materials used for cleaning have hazardous properties such as the ability to irritate or chemically “burn” skin; be toxic, carcinogenic, or mutagenic, cause degeneration of body organs, such as the liver (many solvents), or lungs (acids, chlorine, ozone); or be flammable or explosive. The Material Safety Data Sheets (MSDSs) for hazardous materials (HazMats) should be readily available because they contain information about the hazards, recommendations for use and storage, and first-aid procedures (where applicable). USA law requires that MSDSs for all chemicals used in the workplace be made available to the workers using those materials.

Gases and vapors are often the most hazardous form of materials encountered in the workplace. The hazards may be due to flammability or they may be toxic if inhaled. Generally there is a level, given in parts per million by volume (ppmbv) or weight per volume (mg/m³), below which, the material is not considered hazardous.

The Threshold Limit Value (TLV) or Permissible Exposure Limit (PEL) is the concentration of gas or vapor to which an average worker can be exposed on a continuous eight-hour-shift basis, day after day, without suffering ill effects. For example, for isopropyl alcohol the TLV is 400 ppmbv or 980 mg/m³, while for ozone (O₃) it is 0.1 ppmbv. Because of the difference in susceptibility of different people to different chemicals, it should be recognized that there should be exceptions to these limits. These values are continually changing so reference should be made to the latest values from OSHA (www.osha.gov).

Frequent exposure is measured as a Time-Weighted Average (TWA) value over the term of the exposure. The exposure can exceed the TLV value for some period of time but still keep the average value within the TLV limits. The maximum exposure, under these circumstances is given by the Short-Term Exposure Limits (STEL). For example, the STEL for trichloroethylene (TCE) is 200 ppmbv while the TLV is 50 ppmbv.

In some cases there is a ceiling value (“C” limit) or IDLH (Immediately Dangerous to Life or Health) limit to exposure that should not be exceeded at any time. If this value is exceeded, the area should be evacuated immediately. For example, hydrochloric acid (HCl) has a C-value of 5 ppmbv. There are detectors available for measuring the concentration of hazardous gases and vapors and usually they have an adjustable alarm level.

Fires are classified as: Class A—ordinary solid combustible material, Class B—combustible fluids such as petroleum product, Class C—involve energized electrical equipment, and Class D—involve combustible and reactive metals such as magnesium. Fire extinguishers appropriate for the class of the fire to be expected should be available. Some materials, such as vinyl, produce toxic vapors when they burn. More people are killed by inhalation of smoke and vapors than are killed by burns. Great care should be exercised in fighting a fire. Each work area should have a list of the chemicals in the area posted on the door. This alerts firefighters to the possibility of toxic fumes or explosive hazards in the area if there is a fire.

Flammable liquids, such as alcohol, should be stored in appropriate fireproof containers in limited amounts. Materials not being used should be stored away from the use area in fireproof cabinets that prevent overheating of the containers in the event of a fire. Oxidizing and reducing chemical should not be stored in the same cabinet. Chemical containers should be compatible with the chemical they contain and should be clearly labeled as to their content. Flammable liquids can fuel fires and make extinguishing a fire difficult.

The flash point is the temperature at which a liquid or solid will give off enough vapor to form an ignitable mixture with air just above the surface. For example, pure isopropyl alcohol (C₃H₈O) has a flash point of 12°C, which makes it a very flammable material. The flammable limits give the percent by volume of the material that supports a flame in air. Above or below those limits a flame will not burn by itself. For pure isopropyl alcohol, the flammable limits are 2.3 to 12.7%, while for hydrogen it is 4 to 75%. Acid cleaning can cause the evolution of hydrogen, and acid cleaning should be done in a well-ventilated area with no ignition (spark) source present.

Often the plant Fire Marshal will set the limits on the type and amount of a material that can be used based on the flash point of the material. As a general rule, use of large amounts of materials with a flash point of 140°F in an open area would be banned, while the use of large amounts of materials with a flash point of 165°F or higher is all right. The ignition temperature is the temperature at which there will be self-sustained combustion independent of an ignition source. The ignition temperature of a flammable mixture of isopropyl alcohol is 399°C. Use of low-flash-point materials requires the use of good area ventilation or some type of vapor containment.

Areas where chemicals are used can vary from poorly ventilated open areas to ventilated “chemical hoods,” to fully enclosed processing volumes. Ventilation of an open area is often the first thing to be considered. For example, trichloroethylene (TCE) is a solvent that is nonflammable but is toxic if inhaled. OSHA health standards allow 50 ppmbv of TCE in the air. This standard can generally be met by appropriate area ventilation.

More controlled ventilation can be obtained in a chemical hood that is enclosed except for an opening (face) through which operations are performed in the enclosed work area. Generally this opening may be closed by a sliding window. Air is swept from the outside area, through the face opening, into the work area and out the exhaust. Ventilation is measured by the “face velocity” of the air. A minimum velocity of 100 fpm (feet per minute) is generally used, with higher face velocities used for more toxic materials. In some applications the chemical hood should be fireproof and explosion proof. The exhaust-duct system should be engineered, located, and verified so that the exhaust gases and vapors do not re-enter the building through another ventilation system or pose a threat to people working near the exhaust outlet. Exhaust vents should be clearly marked at the outside location.

The best ventilation control can be obtained in a completely enclosed processing chamber(s). This may be a batch-type system where the parts are placed in a chamber, the chamber is sealed before processing begins, and the chamber is evacuated before the chamber is opened to the ambient. The processing chambers can also be arranged in an air-to-air in-line configuration where parts enter the enclosed processing volume through an entrance chamber and exit through another chamber in a periodic or continuous manner. The inlet and exit chambers may be periodically evacuated or continuously ventilated. A gas “curtain” can also be used to prevent escape or mixing of gases and vapors.

Ozone (O3) is used for cleaning (UV/O3 cleaning) and for water purification. Ozone may be generated by short-wavelength ultraviolet radiation (UVC preferably) or by an oxygen-arc or gas-discharge plasma. Exposure to short-wavelength ultraviolet radiation can be harmful to eyes and promote skin cancer. You can smell ozone (it is what you smell after a lightning storm), but above about 10 ppmbv ozone will kill the olfactory nerves and you will cease to smell it. At concentrations above 10 ppmbv, ozone becomes toxic. As long as you can smell the ozone it is safe for short-term exposure. Ozone should be used in a well-ventilated area and when using UV/O₃ cleaning cabinets the ultraviolet light should be turned off when ever the operator opens the door of the cabinet. If you must look at short-wavelength ultraviolet radiation you should be well covered with clothing and wear UV-adsorbing safety glasses (a UV 400 rating provides 100% protection from UVA and UVB radiation but not UVC ^a).

UVA – 320-400 nm – may be generated with LEDs

UVB – 290-320 nm

UVC – 200-290 nm from mercury vapor glow discharges, quartz envelopes, band pass filtered

EUV – <200 nm – Extreme or vacuum UV

^a Do not just use regular sunglasses – the dark glasses cause the iris of the eye to open wide and if the sunglasses are not UV absorbing this increases the harm done by the UV exposure.

Concentrated acids should be added slowly to water because heat is generated during the dilution. The heating can cause splashing so face shields, goggles, and appropriate clothing and gloves should be worn. Never add water to concentrated acid. Generally one of the main first-aid procedures for any chemical accident is flushing with copious amounts of water. The work area should have an emergency shower facility, an eyewash stand, and/or other sources of water available. Eye rinse cups and solutions should be available.

Personnel in the processing area should be trained in basic first aid. This includes cardiopulmonary resuscitation (CPR), stopping blood loss by local pressure, treatment for shock, and first aid relating to specific hazards. It is particularly important that personnel be familiar with in-house and external emergency communication, alarms, and procedures.

Hydrofluoric acid is used for etching and cleaning oxide materials and is a particularly egregious material for causing chemical burns. These burns are progressive unless the residual acid is removed or neutralized. First-aid procedure is to immediately remove contaminated clothing, rinse the area with copious amounts of water, and apply a chemical neutralizing agent. Medical assistance should be obtained immediately. The burn area should be soaked in an iced aqueous or alcoholic solution of benzalkonium chloride (USP 0.1 to 0.133%) for 1 to 4 hours. Soaking may be accomplished using an immersion bath or by a wet compress. Hydrofluoric acid should not be used without appropriate first-aid chemicals available and personnel trained in their use.

Proper clothing for the workplace is important to safety. Eyes and face should be protected from liquids and flying objects by safety glasses, safety goggles, and/or face shields where appropriate. Goggles, and to a lesser extent safety glasses and face shields, also prevent a person from inadvertently rubbing their eyes with contaminated fingers or gloves. Street clothing should be covered by a lightweight coat (smock) that contains loose clothing, such as neckties, and prevents chemical spills from reaching adsorbent materials such as cotton or silk. The smock can be made of an impermeable material such as Tyvek™, a closely woven fabric such as Nylon™, or a breathable material such as GoreTex™. Cotton is not recommended because it is very adsorbing. When using acids, rubber-coated fabric may be used for smocks or aprons.

When handling chemicals, hands can be protected by gloves— polyethylene or vinyl gloves being the most commonly used. When handling acids, rubber gloves are appropriate. When handling hot surfaces, Nomex™ polyamide gloves are a substitute for the asbestos gloves widely used in the past. Nylon™ and Dacron™ are not good materials to wear in a hot environment because they can melt, adhere to the skin, and cause very severe burns. Shoes should be such that chemicals and liquids are not trapped in the shoes if they are spilled. This may mean using high-topped shoes or shoe covers with top closures.

Removal of some types of film deposits from PVD deposition chambers and fixtures can present safety hazards. In particular, heavy metals such as lead, or particles of beryllium, tellurium or selenium can pose a health hazard if ingested. It is best to remove these materials by wet techniques to avoid dust. Refer to Sax’s book for specific safety hazards associated with materials. Often grit blasting is used to strip film buildup from surfaces. If silica sand is used the workers should avoid breathing the dust because it can cause silicosis, a lung disease. Silica sand should be washed and used in a well-ventilated area with workers protected by appropriate respiratory equipment. This potential health hazard can be avoided by using cast-iron grit or alumina grit instead of silica grit.

General References:

1. Sax’s Dangerous Properties of Industrial Materials, 10th edition, Richard J. Lewis, and N. Irving, John Wiley and Sons, 1999.
2. Hazardous Chemicals Desk Reference, 4th edition, Richard J. Lewis, John Wiley and Sons, 1996.
3. CRC Handbook of Laboratory Safety, A.K. Furr, 5th edition, CRC Press, 2000.
4. Safe Storage of Laboratory Chemicals, 2nd edition, edited by David A. Pipitone, John Wiley and Sons, 1991.
5. Code Compliance for Advanced Technology Facilities: A Comprehensive Guide for Semiconductor and Other Hazardous Occupancies,    William R. Acorn, William Andrew Publishing/Noyes Publications, 1993.
6. Quick Selection Guide to Chemical Protective Clothing, 3rd edition, Krister Forsberg and S.Z. Mansdorf, John Wiley and Sons, 1997.

If you have a safety-related comment, picture or story please send it to me and it will be posted on this blog.

© Donald Mattox – Not to be reproduced without permission.