Handbook of Materials for Product Design Part 9 pdf

Although the conductive nature ofthese materials prevents them from being used as a conventional sub-strate, they have a high thermal conductivity and may be used in ap-plications where

of polyisoprene.23 NR’s principal uses are automotive tires, tire tread,and mechanical goods Automotive applications are always com-pounded with carbon black to impart UV resistance and to increasemechanical properties.10d Latex concentrate is used for dipped goods,adhesives, and latex thread.23 Latex concentrate is produced by cen-trifuge-concentrating field latex tapped from rubber trees The dryrubber content is subsequently increased from 30 to 40 to 60% mini-

Vulcanization is the most important NR chemical reaction.23 Mostapplications require cross-linking via vulcanization to increase resil-iency and strength Exceptions are crepe rubber shoe soles and rubbercements.23 There are a number of methods for sulfur vulcanization,with certain methods producing polysulfidic cross-linking and othermethods producing more monosulfidic cross-links.10d

NR is imported from areas such as Southeast Asia to the world’smost industrial regions, North America, Europe, and Japan, since it isnot indigenous to these regions The huge rubber trees require about

80 to 100 in/y (200 to 250 cm/y) rainfall, and they flourish at an tude of about 1000 ft (300 m).23 As long as NH is needed for tires, in-dustrial regions will be import dependent

alti-NR has good resilience; high tensile strength; low compression set;resistance to wear and tear, cut-through and cold flow; and good elec-tricalproperties.10a Resilience is the principal property advantagecompared with synthetic rubbers.10a For this reason, NH is usuallyused for engine mounts, because NR isolates vibrations caused when

an engine is running NH is an effective decoupler, isolating vibrationssuch as engine vibration from being transmitted to another locationsuch as the passenger compartment.10d With decoupling, vibration isreturned to its source instead of being transmitted through the rub-ber.10d Polychloroprene is used for higher under-hood temperaturesabove NR service limits; butyl rubber is used for body mounts and forroad vibration frequencies, which occur less frequently than engine vi-brations or have low energy; EPDM is often used for molded rubberbumpers and fillers throughout the vehicle, such as deck-lid over-slambumpers

Degree of crystallinity (DC) can affect NH properties, and millingreduces MW MW is reduced by mastication, typically with a Banburymill, adding a peptizing agent during milling to further reduce MW,which improves NR solubility after milling.23 NR latex grades are pro-vided to customers in low (0.20 wt %) and high (0.75 wt %), with am-

eliminates the need for deammoniation.23

Properties of polymers are improved by compounding with ing agents (additives), and NR is not an exception Compounding NRwith property enhancers improves resistance to UV oxygen, andozone, but formulated TPEs and synthetic rubbers overall have betterresistance than compounded NR to UV, oxygen, and ozone.10a NR doesnot have satisfactory resistance to fuels, vegetable, and animal oils,while TPEs and synthetic rubbers can possess good resistance tothem.10a NR has good resistance to acids and alkalis.10a It is soluble inaliphatic, aromatic, and chlorinated solvents, but it does not dissolveeasily because of its high MW Synthetic rubbers have better agingproperties; they harden over time, while NR softens over time (see Ta-ble 6.28).10a

enhanc-There are several visually graded latex NRs, including ribbedsmoked sheets (RSS) and crepes such as white and pale, thin and thickbrown latex, etc.23 Two types of raw NR are field latex and raw coagu-lum, and these two types comprise all NR (“downstream”) grades.23

TABLE 6.28 Typical Thermal and Electrical Property Profile of NR 23

Property Value Specific gravity

@ 32°F (0°C)

@ 68°F (20°C)

T g, °F (°C)

Specific heat

Heat of combustion, cal/g (J/g)

Thermal conductivity, (BTU-in) (h-ft 2 -°F)

64 (266.5) 1.5192 1.5218

Depolymerized NR is used as a base for asphalt modifiers, pottingcompound, and cold-molding compounds for arts and crafts.10b

6.17 Conclusion

Producers can engineer polymers and copolymers, and compounderscan formulate recipes for a range of products that challenges the de-signers’ imaginations Computer variable-controlled machinery, tools,and dies can meet the designers’ demands Processing elastomeric ma-terials is not as established as the more traditional thermoplastic andthermosetting polymers Melt rheology, more than just viscosity, is thecentral differentiating characteristic for processing elastomeric mate-rials Processing temperature and pressure settings are not fixedranges; they are dynamic, changing values from the hopper to thedemolded product Operators and management of future elastomericmaterials processing plants will be educated to the finesse of melt pro-cessing these materials Elastomeric materials industries, welcome tothe twenty-first century

References

1 James M Margolis, “Elastomeric Polymers 2000 to 2010: Properties, Processes and Products” Report, 2000.

2 K RATON Polymers and Compounds, Typical Properties Guide, Shell Chemical

Com-pany, Houston, Texas, 1997.

3 Products, Properties and Processing for PELLETHANE Thermoplastic

Polyure-thane Elastomers, Dow Plastics, The Dow Chemical Company Midland, Michigan,

ca 1997.

4 Modern Plastics Encyclopedia ’99, McGraw-Hill, New York, 1999, pp B-51, B-52.

5 Engage, A Product of DuPont Dow Elastomers, Wilmington, Delaware, December 1998.

6 Product Guide, Goodyear Chemical, Goodyear Tire & Rubber Company, Akron,

Ohio, October 1996.

7 Injection Molding Guide for Thermoplastic Rubber–Processing, Mold Design,

Equipment, Advanced Elastomer Systems LP, Akron, Ohio, 1997.

8 Santoprene Rubber Physical Properties Guide, Advanced Elastomer Systems LP,

Akron, Ohio, ca 1998.

9 Hifax MXL 55A01 (1998), FXL 75A01 (1997) and MXL 42D01 Developmental Data Sheets secured during product development and subject to change before final com- mercialization Montell Polyolefins Montell North America Inc., Wilmington, Dela- ware.

10 Charles B Rader, “Thermoplastic Elastomers,” in Handbook of Plastics,

Elas-tomers, and Composites, 3d ed., Charles A Harper, ed., McGraw-Hill, New York,

1996.

10a Joseph F Meier, “Fundamentals of Plastics and Elastomers,” in Handbook of

Plas-tics, Elastomers, and Composites, 3d ed., Charles A Harper, ed., McGraw-Hill,

New York, 1996.

10b Leonard S Buchoff, “Liquid and Low-Pressure Resin Systems,” in Handbook of

Plastics, Elastomers, and Composites, 3d ed., Charles A Harper, ed., McGraw-Hill,

New York, 1996.

10c Edward M Petrie “Joining of Plastics, Elastomers, and Composites,” in Handbook

of Plastics, Elastomers, and Composites, 3d ed., Charles A Harper, ed.,

McGraw-Hill, New York, 1996.

10d Ronald Toth, “Elastomers and Engineering Thermoplastics for Automotive

Appli-cations,” in Handbook of Plastics, Elastomers, and Composites, 3d ed., Charles

Harper, ed., McGraw-Hill, New York, 1996.

11 Aflas TFE Elastomers Technical Information and Performance Profile Data Sheets, Dyneon LLC, A 3M-Hoechst Enterprise, Oakdale, Minnesota, 1997.

12 Vistalon User’s Guide, Properties of Ethylene-Propylene Rubber, Exxon Chemical Company, Houston, Texas, Division of Exxon Corporation, ca 1996.

13 K RATON Liquid L-2203 Polymer, Shell Chemical Company, Houston, Texas, 1997.

14 Affinity Polyolefin Plastomers, Dow Plastics, The Dow Chemical Company, land, Michigan, 1997.

Mid-14a Affinity HF-1030 Data Sheet, Dow Plastics, The Dow Chemical Company Midland, Michigan, 1997.

14b Affinity PF 1140 Data Sheet, Dow Plastics, The Dow Chemical Company Midland, Michigan, 1997.

15 Bayer Engineering Polymers Properties Guide, Thermoplastics and thanes, Bayer Corporation, Pittsburgh, Pennsylvania, 1998.

Polyure-16 Rubber World Magazine, monthly, 1999.

17 Jim Ahnemiller, “PU Rubber Outsoles for Athletic Footwear,” Rubber World,

De-cember 1998.

18 Charles D Shedd, “Thermoplastic Polyolefin Elastomers,” in Handbook of

Thermo-plastic Elastomers, 2d ed., Benjamin M Walker and Charles P Rader, eds., Van

Nostrand Reinhold, New York, 1988.

18a Thomas W Sheridan, “Copolyester Thermoplastic Elastomers,” Handbook of

Ther-moplastic Elastomers, 2d ed., Benjamin M Walker and Charles P Rader, eds., Van

Nostrand Reinhold, New York, 1988.

18b William J Farrisey “Polyamide Thermoplastic Elastomers,” in Handbook of

Ther-moplastic Elastomers, 2d ed., Benjamin M Walker and Charles P Rader, eds., Van

Nostrand Reinhold, New York, 1988.

18c Eric C Ma, “Thermoplastic Polyurethane Elastomers, in Handbook of

Thermoplas-tic Elastomers, 2d ed., Benjamin M Walker and Charles P Rader, eds., Van

Nos-trand Reinhold, New York, 1988.

19 N R Legge, G Holden, and H E Schroeder, eds., Thermoplastic Elastomers, A

Comprehensive Review, Hanser Publishers, Munich, Germany, 1987.

20 P S Ravisbanker, “Advanced EPDM for W & C Applications,” Rubber World,

De-cember 1998.

21 Junling Zbao, G N Chebremeskel, and J Peasley “SBR/PVC Blends With NBR As

Compatibilizer,” Rubber World, December 1998.

22 John E Rogers and Walter H Waddell, “A Review of Isobutylene-Based

Elas-tomers Used in Automotive Applications,” Rubber World, February 1999.

23 Kirk-Othmer Concise Encyclopedia of Chemical Technology, John Wiley & Sons,

27 C P J van der Aar, et al., “Adhesion of EPDMs and Fluorocarbons to Metals by

Using Water-Soluble Polymers,” Rubber World, November 1998.

28 Larry R Evans and William C Fultz, “Tread Compounds with Highly Dispersible

Silica,” Rubber World, December 1998.

29 Vector Styrene Block Copolymers, Dexco Polymers, A Dow/Exxon Partnership, Houston, Texas, 1997.

30 Fluoroelastomers Product Information Manual (1997), Product Comparison Guide

(1999), Dyneon LLC, A 3M-Hoechst Enterprise, Oakdale, Minnesota, 1997.

31 Engel data sheets and brochures, Guelph, Ontario, 1998.

32 Catalloy Process Resins, Montell Polyolefins, Wilmington, Delaware.

32a Catalloy Process Resins, Montell Polyolefins, Wilmington, Delaware, p.7.

33 EniChem Europrene SOL T Thermoplastic Rubber, styrene butadiene types, rene isoprene types, EniChem Elastomers Americas Inc., Technical Assistance Laboratory, Baytown, Texas.

sty-34 “Arnitel Guidelines for the Injection Molding of Thermoplastic Elastomer TPE-E,” DSM Engineering Plastics, Evansville, Ind., ca, 1998.

35 Correspondence from DuPont Engineering Polymers, July 1999.

36 Correspondence from DuPont Dow Elastomers, Wilmington, Delaware, August 1999.

in this context.

The primary bonding mechanism in ceramics is ionic bonding Anionic bond is formed by the electrostatic attraction between positiveand negative ions Atoms are most stable when they have eight elec-trons in the outer shell Metals have a surplus of electrons in the outershell, which are loosely bound to the nucleus and readily become free,creating positive ions Similarly, nonmetals have a deficit of electrons

in the outer shell and readily accept free electrons, creating negativeions Figure 7.1 illustrates an ionic bond between a magnesium ionwith a charge of +2 and an oxygen ion with a charge of –2, formingmagnesium oxide (MgO) Ionically bonded materials are crystalline innature and have both a high electrical resistance and a high relativedielectric constant Due to the strong nature of the bond, they have ahigh melting point and do not readily break down at elevated temper-atures By the same token, they are very stable chemically and are notattacked by ordinary solvents and most acids

7.2 Chapter 7

A degree of covalent bonding may also be present, particularly insome of the silicon and carbon-based ceramics The sharing of elec-trons in the outer shell forms a covalent bond A covalent bond is de-picted in Fig 7.2, illustrating the bond between oxygen and hydrogen

to form water A covalent bond is also a very strong bond and may bepresent in liquids, solids, or gases

Figure 7.1 Magnesium oxide ionic bond.

Covalent bond between oxygen and hydrogen to form water.

Ceramics and Ceramic Composites 7.3

A composite is a mixture of two or more materials that retain theiroriginal properties but, in concert, offer parameters that are superior

to either Composites in various forms have been used for centuries.Ancient peoples, for example, used straw and rocks in bricks to in-crease their strength Modern day structures use steel rods to rein-force concrete The resulting composite structure combines thestrength of steel with the lower cost and weight of concrete

Ceramics are commonly used in conjunction with metals to formcomposites for electronic applications, especially thermal manage-ment Ceramic-metal (cermet) composites typically have a lower TCEthan metals, possess a higher thermal conductivity than ceramics,and are more ductile and more resistant to stress than ceramics.These properties combine to make cermet composites ideal for use inhigh-power applications

This chapter considers the properties of ceramics used in tronic applications, including aluminum oxide (alumina, Al2O3), beryl-lium oxide (beryllia, BeO), aluminum nitride (AlN), boron nitride(BN), diamond (C), and silicon carbide (SiC) Several composite mate-rials, aluminum silicon carbide (AlSiC) and Dymalloy, a diamond/cop-per structure, are also described Although the conductive nature ofthese materials prevents them from being used as a conventional sub-strate, they have a high thermal conductivity and may be used in ap-plications where the relatively low electrical resistance is not aconsideration

microelec-7.2 Ceramic Fabrication

It is difficult to manufacture ceramic substrates in the pure form Themelting point of most ceramics is very high, as shown in Table 7.1, andmost are also very hard, limiting the ability to machine the ceramics.For these reasons, ceramic substrates are typically mixed with fluxingand binding glasses, which melt at a lower temperature and make thefinished product easier to machine

The manufacturing process for Al2O3, BeO, and AlN substrates isvery similar The base material is ground into a fine powder, severalmicrons in diameter, and mixed with various fluxing and bindingglasses, including magnesia and calcia, also in the form of powders Anorganic binder, along with various plasticizers, is added to the mix-ture, and the resultant slurry is ball-milled to remove agglomeratesand to make the composition uniform

The slurry is formed into a sheet, the so-called green state, by one ofseveral processes as shown in Fig 7.31 and sintered at an elevatedtemperature to remove the organics and to form a solid structure

7.4 Chapter 7

par-tially dried to form a sheet with the consistency of putty The sheet isfed through a pair of large parallel rollers to form a sheet of uniformthickness

under a knife-edge to form the sheet This is a relatively low-pressureprocess compared to the others

subjected to very high pressure (up to 20,000 psi) throughout the

sin-TABLE 7.1 Melting Points of Selected

Ceramics and Ceramic Composites 7.5

tering process This produces a very dense part with tighter as-firedtolerances than other methods, although pressure variations may pro-duce excessive warpage

sur-rounded with water or glycerin and compressed with up to 10,000 psi.The pressure is more uniform and produces a part with less warpage

forced through a die Tight tolerances are hard to obtain, but the cess is very economical and produces a thinner part than is attainable

pro-by other methods

In the green state, the substrate is approximately the consistency ofputty and may be punched to the desired size Holes and other geome-tries may also be punched at this time

Once the part is formed and punched, it is sintered at a temperatureabove the glass melting point to produce a continuous structure Thetemperature profile is very critical, and the process may actually beperformed in two stages: one stage to remove the volatile organic ma-terials and a second stage to remove the remaining organics and tosinter the glass/ceramic structure The peak temperature may be ashigh as several thousand degrees celsius and may be held for severalhours, depending on the material and the type and amount of bindingglasses For example, pure alumina substrates formed by powder pro-cessing with no glasses are sintered at 1930°C

It is essential that all the organic material be removed prior to tering Otherwise, the gases formed by the organic decomposition mayleave serious voids in the ceramic structure and cause serious weaken-ing The oxide ceramics may be sintered in air In fact, it is desirable tohave an oxidizing atmosphere to aid in removing the organic materi-als by allowing them to react with the oxygen to form CO2 The nitrideceramics must be sintered in the presence of nitrogen to prevent ox-ides of the metal from being formed In this case, no reaction of the or-ganics takes place; they are evaporated and carried away by thenitrogen flow

sin-During sintering, a degree of shrinkage takes place as the organic isremoved and the fluxing glasses activate Shrinkage may range from

as low as 10% for powder processing to as high as 22% for sheet ing The degree of shrinkage is highly predictable and may be consid-ered during design

cast-Powder pressing generally forms boron nitride substrates Varioussilica and/or calcium compounds may be added to lower the processingtemperature and improve machinability Diamond substrates are typi-cally formed by chemical vapor deposition (CVD) Composite sub-strates, such as AlSiC, are fabricated by creating a spongy structure ofSiC and forcing molten aluminum into the crevices

7.6 Chapter 7

7.3 Surface Properties of Ceramics

The surface properties of interest, surface roughness and camber, arehighly dependent on the particle size and method of processing Sur-face roughness is a measure of the surface microstructure, and camber

is a measure of the deviation from flatness In general, the smaller theparticle size, the smoother will be the surface

Surface roughness may be measured by electrical or optical means.Electrically, surface roughness is measured by moving a fine-tippedstylus across the surface The stylus may be attached to a piezoelectriccrystal or to a small magnet that moves inside a coil, inducing a volt-age proportional to the magnitude of the substrate variations The sty-lus must have a resolution of 25.4 nm (1 µin) to read accurately in themost common ranges Optically, a coherent light beam from a laser di-ode or other source is directed onto the surface The deviations in thesubstrate surface create interference patterns that are used to calcu-late the roughness Optical profilometers have a higher resolutionthan the electrical versions and are used primarily for very smoothsurfaces For ordinary use, the electrical profilometer is adequate and

is widely used to characterize substrates in both manufacturing andlaboratory environments

The output of an electrical profilometer is plotted as shown in matic form in Fig 7.4 and in actual form in Fig 7.5 A quantitative in-terpretation of surface roughness can be obtained from this plot in one

sche-of two ways: by the rms value and by the arithmetic average

The rms value is obtained by dividing the plot into n small, even crements of distance and measuring the height, m, at each point, asshown in Fig 7.4 The rms value is calculated by

Ceramics and Ceramic Composites 7.7

and the average value (usually referred to as the center line average, CLA) is calculated by

(7.2)

wherea1, a2, a3, … = areas under the trace segments (Fig 7.4)

L = length of travelFor systems where the trace is magnified by a factor M, Eq (7.2) must

be divided by the same factor

value is 0.707 × peak, which is 11.2% larger than the average The filometer trace is not quite sinusoidal in nature The rms value may begreater than the CLA value from 10 to 30%

pro-Of the two methods, the CLA is the preferred method of use, cause the calculation is more directly related to the surface roughness.However, it also has several shortcomings

in Fig 7.6.2

ampli-tudes yield the same results, although the effect in use may besomewhat different

■ The value obtained is a function of the tip radius

Surface roughness has a significant effect on the adhesion and formance of thick and thin film depositions For adhesion purposes, it

per-is desirable to have a high surface roughness to increase the effectiveinterface area between the film and the substrate For stability andrepeatability, the thickness of the deposited film should be muchgreater than the variations in the surface For thick films, which have

Figure 7.5 Surface trace of three substrate surfaces.

CLA a1+a2+a3+… a+ n

L

-=

7.8 Chapter 7

a typical thickness of 10–12 µ, surface roughness is not a ation, and a value of 25 µin (625 nm) is desirable For thin films, how-ever, which may have a thickness measured in angstroms, a muchsmoother surface is required Figure 7.7 illustrates the difference in athin film of tantalum nitride (TaN) deposited on both a 1 µin surfaceand a 5 µin surface Tantalum nitride is commonly used to fabricateresistors in thin film circuits and is stabilized by growing a layer oftantalum oxide, which is nonconductive, over the surface by bakingthe resistors in air Note that the oxide layer in the rougher surfacerepresents a more significant percentage of the overall thickness ofthe film in areas where the surface deviation is the greatest The re-sult is a wider variation in both the initial and post-stabilization resis-tor values and a larger drift in value with time

consider-Camber and waviness are similar in form in that they are variations

in flatness over the substrate surface Referring to Fig 7.6, cambercan be considered as an overall warpage of the substrate, while wavi-

Figure 7.6 Surface characteristics.

Figure 7.7 TaN resistor with TaO passivation on substrates

with different surface roughness.

Ceramics and Ceramic Composites 7.9

ness is more periodic in nature Both of these factors may occur as aresult of uneven shrinkage during the organic removal/sintering pro-cess or as a result of nonuniform composition Waviness may also oc-cur as a result of a “flat spot” in the rollers used to form the greensheets

Camber is measured in units of length/length, interpreted as the viation from flatness per unit length, and is measured with reference

de-to the longest dimension by placing the substrate through parallelplates set a specific distance apart Thus, a rectangular substratewould be measured along the diagonal A typical value of camber is0.003 in/in (also 0.003 mm/mm), which for a 2 × 2 inch substrate repre-sents a total deviation of 0.003 × 2 × 1.414 = 0.0085 in For a substratethat is 0.025 thick, a common value, the total deviation represents athird of the overall thickness!

The nonplanar surface created by camber adversely affects quent metallization and assembly processes In particular, screenprinting is made more difficult due to the variable snap-off distance.Torsion bar printing heads on modern screen printers can compensate

subse-to a certain extent, but not entirely A vacuum hold-down on the screenprinter platen also helps but only flattens the substrate temporarilyduring the actual printing process Camber can also create excessivestresses and a nonuniform temperature coefficient of expansion Attemperature extremes, these factors can cause cracking, breaking, oreven shattering of the substrate

Camber is measured by first measuring the thickness of the strate and then placing the substrate between a series of pairs of par-allel plates set specific distances apart Camber is calculated bysubtracting the substrate thickness from the smallest distance thatthe substrate will pass through and dividing by the longest substratedimension A few generalizations can be made about camber

sub-■ Thicker substrates will have less camber than thinner ones

■ Square shapes will have less camber than rectangular ones

camber than the sheet methods

7.4 Thermal Properties of Ceramic

Materials

7.4.1 Thermal Conductivity

The thermal conductivity of a material is a measure of its ability tocarry heat and is defined as

7.10 Chapter 7

(7.3)

q = heat flux in w/cm2 = temperature gradient in °C/m in steady state

The negative sign denotes that heat flows from areas of higher perature to areas of lower temperature

tem-Two mechanisms contribute to thermal conductivity, (1) the ment of free electrons and (2) lattice vibrations, or phonons When amaterial is locally heated, the kinetic energy of the free electrons inthe vicinity of the heat source increases, causing the electrons to mi-grate to cooler areas These electrons undergo collisions with other at-oms, losing their kinetic energy in the process The net result is thatheat is drawn away from the source toward cooler areas In a similarfashion, an increase in temperature increases the magnitude of thelattice vibrations, which, in turn, generate and transmit phonons, car-rying energy away from the source The thermal conductivity of a ma-terial is the sum of the contributions of these two parameters

move-(7.4)

where k p = contribution due to phonons

k e = contribution due to electrons

In ceramics, the heat flow is primarily due to phonon generation,and the thermal conductivity is generally lower than that of metals.Crystalline structures, such as alumina and beryllia, are more effi-cient heat conductors than amorphous structures such as glass Or-ganic materials used to fabricate printed circuit boards or epoxyattachment materials are electrical insulators and highly amorphous,and they tend to be very poor thermal conductors

Impurities and other structural defects in ceramics tend to lowerthe thermal conductivity by causing the phonons to undergo more col-lisions, lowering the mobility and lessening their ability to transportheat away from the source This is illustrated by Table 7.2, which liststhe thermal conductivity of alumina as a function of the percentage ofglass Although the thermal conductivity of the glass binder is lowerthan that of the alumina, the drop in thermal conductivity is greaterthan expected from the addition of the glass alone If the thermal con-ductivity is a function of the ratio of the materials alone, it follows therule of mixtures

dx

–

Ceramics and Ceramic Composites 7.11

(7.5)

where k T = net thermal conductivity

P1 = volume percentage of material one in decimal form

k1 = thermal conductivity of material one

P2 = volume percentage of material two in decimal form

k2 = thermal conductivity of material two

In pure form, alumina has a thermal conductivity of about 31

°C, and the binding glass has a thermal conductivity of about 1

W/m-°C Equation (7.5) and the parameters from Table 7.2 are plotted in

Fig 7.8

TABLE 7.2 Thermal Conductivity of Alumina

Substrates with Different Concentrations of

Alumina

Volume percentage

of alumina

Thermal conductivity, W/m-°C

Figure 7.8 Thermal conductivity of alumina vs.

concentration, theoretical and actual.

7.12 Chapter 7

By the same token, as the ambient temperature increases, the

num-ber of collisions increases, and the thermal conductivity of most

mate-rials decreases A plot of the thermal conductivity vs temperature for

several materials is shown in Fig 7.9.3 One material not plotted in

this graph is diamond The thermal conductivity of diamond varies

widely with composition and the method of preparation, and it is

much higher than those materials listed Diamond will be discussed in

detail in a later section Selected data from Fig 7.9 was analyzed and

extrapolated into binomial equations that quantitatively describe the

thermal conductivity vs temperature relationship This data is

The specific heat, c, is defined in a similar manner and is the

amount of heat required to raise the temperature of one gram of rial by one degree, with units of watt-s/gm-°C The quantity “specific

mate-heat” in this context refers to the quantity, c V , which is the specific

heat measured with the volume constant, as opposed to c P , which is

measured with the pressure constant At the temperatures of interest,these numbers are nearly the same for most solid materials The spe-cific heat is primarily the result of an increase in the vibrational en-ergy of the atoms when heated, and the specific heat of most materials

increases with temperature up to a temperature called the Debye

tem-perature, at which point it becomes essentially independent of ature The specific heat of several common ceramic materials as afunction of temperature is shown in Fig 7.10

temper-The heat capacity, C, is similar in form, except that it is defined in

terms of the amount of heat required to raise the temperature of amole of material by one degree and has the units of watt-s/mol-°C

7.4.3 Temperature Coefficient of Expansion

The temperature coefficient of expansion (TCE) arises from the metrical increase in the interatomic spacing of atoms as a result of in-creased heat Most metals and ceramics exhibit a linear, isotropicrelationship in the temperature range of interest, while certain plas-tics may be anisotropic in nature The TCE is defined as

asym-TABLE 7.3 Approximate Thermal Conductivity vs Temperature for

Selected Ceramic Materials (Binomial Relationship)

Material Constant T Coefficient T2 Coefficient

T1 = initial temperature

T2 = final temperature

l(T1) = length at initial temperature

l(T2) = length at final temperature

The TCE of most ceramics is isotropic For certain crystalline or gle-crystal ceramics, the TCE may be anisotropic, and some may evencontract in one direction and expand in the other Ceramics used forsubstrates do not generally fall into this category, as most are mixedwith glasses in the preparation stage and do not exhibit anisotropicproperties as a result The temperature coefficient of expansion of sev-eral ceramic materials is shown in Table 7.4

sin-7.5 Mechanical Properties of Ceramic

=

anisms, which create slip mechanisms in softer metals, are relativelyscarce in ceramics, and failure may occur with very little plastic defor-mation Ceramics also tend to fracture with little resistance

7.5.1 Modulus of Elasticity

The temperature coefficient of expansion (TCE) phenomenon has ous implications in the applications of ceramic substrates When asample of material has one end fixed, which may be considered to be aresult of bonding to another material that has a much smaller TCE,

seri-the net elongation of seri-the hotter end per unit length, or strain (E), of

the material is calculated by

(7.8)

∆T = temperature differential across the sample

Elongation develops a stress (S) per unit length in the sample as

given by Hooke’s Law

(7.9)where S = stress in psi/in (N/m2/m)

Y = modulus of elasticity in lb/in2 (N/m2)

TABLE 7.4 Temperature Coefficient of

Expansion of Selected Ceramic

AlSiC (70% SiC loading) 6.3

When the total stress, as calculated by multiplying the stress/unitlength by the maximum dimension of the sample, exceeds the strength

of the material, mechanical cracks will form in the sample that mayeven propagate to the point of separation The small elongation that

occurs before failure is referred to as plastic deformation This

analy-sis is somewhat simplistic in nature but serves as a basic ing of the mechanical considerations The modulus of elasticity ofselected ceramics is summarized in Table 7.5, along with other me-chanical properties

(7.10)

M = maximum bending moment in N-m

x = distance from center to outer surface in m

TABLE 7.5 Mechanical Properties of Selected Ceramics

Material

Modulus of elasticity, GPa

Tensile strength, MPa

Compressive strength, MPa

Modulus of rupture, MPa

Flexural strength, MPa Density, g/cm 3

Beryllia (99.5%) 345 138 1550 233 235 2.87 Boron nitride (normal) 43 2410 6525 800 53.1 1.92 Aluminum nitride 300 310 2000 300 269 3.27 Silicon carbide 407 197 4400 470 518 3.10 Diamond (type IIA) 1000 1200 11000 940 1000 3.52

I

-=

The expressions for σ, M, x, and I are summarized in Table 7.6.

When these are inserted into Equation (7.10), the result is

(7.11)

(7.12)

x = long dimension of rectangular cross section in m

y = short dimension of rectangular cross section in m

L = length of sample in m

R = radius of circular cross-section in m

The modulus of rupture is the stress required to produce fractureand is given by

where σr = modulus of rupture in n/m2

F r = force at rupture

The modulus of rupture for selected ceramics is shown in Table 7.5

7.5.3 Tensile and Compressive Strength

A force applied to a ceramic substrate in a tangential direction mayproduct tensile or compressive forces If the force is tensile, in a direc-tion such that the material is pulled apart, the stress produces plasticdeformation as defined in Equation (7.9) As the force increases past a

value referred to as the tensile strength, breakage occurs Conversely,

a force applied in the opposite direction creates compressive forces

until a value referred to as the compressive strength is reached, at

which point breakage also occurs The compressive strength of ramics is, in general, much larger than the tensile strength The ten-sile and compressive strength of selected ceramic materials is shown

ce-in Table 7.5

In practice, the force required to fracture a ceramic substrate ismuch lower than predicted by theory The discrepancy is due to smallflaws or cracks residing within these materials as a result of process-ing For example, when a substrate is sawed, small edge cracks may

be created Similarly, when a substrate is fired, trapped organic rial may outgas during firing, leaving a microscopic void in the bulk.The result is an amplification of the applied stress in the vicinity ofthe void that may exceed the tensile strength of the material and cre-ate a fracture If the microcrack is assumed to be elliptical with themajor axis perpendicular to the applied stress, the maximum stress atthe tip of the crack may be approximated by4

mate-TABLE 7.6 Parameters of Stress in

Modulus of Rupture Test (from Ref 5)

3

12 -

FL

4 - πR2

4 -

where S M = maximum stress at the tip of the crack

S O = nominal applied stress

a = length of the crack as defined in Fig 7.11

ρt = radius of the crack tipThe ratio of the maximum stress to the applied stress may be de-fined as

(7.16)

where K t = stress concentration factor

For certain geometries, such as a long crack with a small tip radius,

K t may be much larger than 1, and the force at the tip may be tially larger than the applied force

substan-Based on this analysis, a material parameter called the plain strain

fracture toughness, a measure of the ability of the material to resist

fracture, can be defined as

(7.17)

where K IC = plain strain fracture toughness in psi-in1/2 or Mpa-m1/2

Z = dimensionless constant, typically 1.2 (Ref 4)

S c = critical force required to cause breakageFrom Eq (7.17), the expression for the critical force can be definedas

(7.18)

When the applied force on the die due to TCE or thermal differencesexceeds this figure, fracture is likely The plain strain fracture tough-ness for selected materials is presented in Table 7.7 It should be notedthat Eq (7.13) is a function of thickness up to a point but is approxi-mately constant for the area to thickness ratio normally found in sub-strates

ρt -

 

 

1 2

 

 

1 2

7.5.4 Hardness

Ceramics are among the hardest substances known, and the hardness

is correspondingly difficult to measure Most methods rely on the ity of one material to scratch another, and the measurement is pre-sented on a relative scale Of the available methods, the Knoopmethod, is the most frequently used In this approach, the surface ishighly polished, and a pointed diamond stylus under a light load is al-lowed to impact on the material The depth of the indentation formed

abil-by the stylus is measured and converted to a qualitative scale called

the Knoop or HK scale The Knoop hardness of selected ceramics is

given in Table 7.8

7.5.5 Thermal Shock

Thermal shock occurs when a substrate is exposed to temperature tremes in a short period of time Under these conditions, the substrate

ex-is not in thermal equilibrium, and internal stresses may be sufficient

to cause fracture Thermal shock can be liquid to liquid or air to air,

TABLE 7.7 Fracture Toughness for Selected Materials

Material Fracture toughness, MPA-m 1/2

TABLE 7.8 Knoop Hardness for Selected Ceramics

Material Knoop hardness, 100 g

with the most extreme exposure occurring when the substrate istransferred directly from one liquid bath to another The heat is morerapidly absorbed or transmitted, depending on the relative tempera-ture of the bath, due to the higher specific of the liquid as opposed toair

The ability of a substrate to withstand thermal shock is a function ofseveral variables, including the thermal conductivity, the coefficient ofthermal expansion, and the specific heat Winkleman and Schott5 de-

veloped a parameter called the coefficient of thermal endurance that

qualitatively measures the ability of a substrate to withstand thermalstress

(7.19)

P = tensile strength in MPa

α = thermal coefficient of expansion in 1/K

Y = modulus of elasticity in MPa

k = thermal conductivity in W/m-K

ρ = density in kg/m3

c = specific heat in W-s/kg-K

The coefficient of thermal endurance for selected materials is shown

in Table 7.9 The phenomenally high coefficient of thermal endurancefor BN is primarily a result of the high tensile strength to modulus ofelasticity ratio as compared to other materials Diamond is also high,primarily due to the high tensile strength, the high thermal conductiv-ity, and the low TCE

TABLE 7.9 Thermal Endurance Factor for Selected Materials at 25°C

Material Thermal endurance factor

The thermal endurance factor is a function of temperature in thatseveral of the variables, particularly the thermal conductivity and thespecific heat, are functions of temperature From Table 7.9, it is alsonoted that the thermal endurance factor may drop rapidly as the alu-mina to glass ratio drops This is due to the difference in the thermalconductivity and TCE of the alumina and glass constituents that in-crease the internal stresses This is true of other materials as well.

7.6 Electrical Properties of Ceramics

The electrical properties of ceramic substrates perform an importanttask in the operation of electronic circuits Depending on the applica-tions, the electrical parameters may be advantageous or detrimental

to circuit function Of most interest are the resistivity, the breakdownvoltage or dielectric strength, and the dielectric properties, includingthe dielectric constant and the loss tangent

7.6.1 Resistivity

The electrical resistivity of a material is a measure of the ability ofthat material to transport charge under the influence of an appliedelectric field More often, this ability is presented in the form of theelectrical conductivity, which is the reciprocal of the resistivity as de-fined in Eq (7.20)

(7.20)

ρ = resistivity in ohm-unit lengthThe conductivity is a function primarily of two variables: the con-centration of charge and the mobility—the ability of that charge to betransported through the material The current density and the ap-plied field are related by the expression defined in Eq (7.21)

(7.21)

E = electric field in volts/unit length

It should be noted that both the current density and the electric fieldare vectors, since the current is in the direction of the electric field.The current density may also be defined as

σ = 1ρ

v d = drift velocity of electrons in unit length/second

The drift velocity is related to the electric field by

(7.23)where µ = mobility in length2/volt-second

In terms of the free carrier concentration and the mobility, the rent density is

cur-(7.24)Comparing Eq (7.19) with Eq (7.23), the conductivity can be de-fined as

(7.25)The free carrier concentration may be expressed as

(7.26)

where n t = free carrier concentration due to thermal activity

n i = free carrier concentration due to field injectionThe thermal charge density, nt, in insulators is a result of free elec-trons obtaining sufficient thermal energy to break the interatomicbonds, allowing them to move freely within the atomic lattice Ce-ramic materials characteristically have few thermal electrons as a re-sult of the strong ionic bonds between atoms The injected chargedensity, nI, occurs when a potential is applied and is a result of the in-herent capacity of the material The injected charge density is given

by

(7.27)where ε = dielectric constant of the material in farads/unit lengthInserting Eq (7.27) and Eq (7.26) into Eq (7.24), the result is

For conductors, n t >> n i and Ohm’s law applies For insulators, n i >>

n t, and the result is a square law relationship between the voltage andthe current.6

(7.29)The conductivity of ceramic substrates is extremely low In practice,

it is primarily due to impurities and lattice defects and may varywidely from batch to batch The conductivity is also a strong function

of temperature As the temperature increases, the ratio of thermal toinjected carriers increases As a result, the conductivity increases andthe V-I relationship follows Ohm’s law more closely Typical values ofthe resistivity of selected ceramic materials are presented in Table7.10

7.6.2 Breakdown Voltage

The term breakdown voltage is very descriptive While ceramics are

normally very good insulators, the application of excessively high

po-TABLE 7.10 Electrical Properties of Selected Ceramic Substrates

Property Electrical

resistivity ( Ω-cm)

Breakdown voltage (ac kV/mm)

Dielectric constant

Loss tangent (@ 1 MHz) Alumina (96%)

tentials can dislodge electrons from orbit with sufficient energy to

al-low them to dislodge other electrons from orbit, creating an avalanche

effect The result is a breakdown of the insulation properties of the

material, allowing current to flow This phenomenon is accelerated byelevated temperature, particularly when mobile ionic impurities arepresent

The breakdown voltage is a function of numerous variables, ing the concentration of mobile ionic impurities, grain boundaries, andthe degree of stoichiometry In most applications, the breakdown volt-age is sufficiently high as to not be an issue However, there are twocases in which it must be a consideration:

includ-1 At elevated temperatures created by localized power dissipation orhigh ambient temperature, the breakdown voltage may drop by or-ders of magnitude Combined with a high potential gradient, thiscondition may be susceptible to breakdown

2 The surface of most ceramics is highly wettable, in that moisture

tends to spread rapidly Under conditions of high humidity, pled with surface contamination, the effective breakdown voltage

cou-is much lower than the intrinsic value

7.6.3 Dielectric Properties

Two conductors in proximity with a difference in potential have theability to attract and store electric charge Placing a material with di-electric properties between them enhances this effect A dielectric ma-terial has the capability of forming electric dipoles (displacements ofelectric charge) internally At the surface of the dielectric, the dipolesattract more electric charge, thus enhancing the charge storage capa-bility, or capacitance, of the system The relative ability of a material

to attract electric charge in this manner is called the relative dielectric

constant, or relative permittivity, and is usually given the symbol K.

The relative permittivity of free space is 1.0 by definition, and the solute permittivity is

ab-(7.30)where εo = permittivity of free space

The relationship between the polarization and the electric field is

(7.31)

36π -×10 9 farads/meter

where P = polarization, coulombs/m2

E = electric field, V/m

Four basic mechanisms contribute to polarization

1 Electronic polarization. In the presence of an applied field, thecloud of electrons is displaced relative to the positive nucleus of theatom or molecule, creating an induced dipole moment Electronicpolarization is essentially independent of temperature and may oc-cur very rapidly The dielectric constant may therefore exist atvery high frequencies, up to 1017 Hz

per-manent dipoles that exist even in the absence of an electric field.These may be rotated by an applied electric field, generating a de-gree of polarization by orientation Molecular polarization is in-versely proportional to temperature and occurs only at low tomoderate frequencies Molecular polarization does not occur to agreat extent in ceramics and is more prevalent in organic materi-als and liquids such as water

3 Ionic polarization. Ionic polarization occurs in ionically bondedmaterials when the positive and negative ions undergo a relativedisplacement to each other in the presence of an applied electricfield Ionic polarization is somewhat insensitive to temperatureand occurs at high frequencies, up to 1013 Hz

4 Space charge polarization. Space charge polarization exists as aresult of charges derived from contaminants or irregularities thatexist within the dielectric These charges exist to a greater orlesser degree in all crystal lattices and are partly mobile Conse-quently, they will migrate in the presence of an applied electricfield Space charge polarization occurs only at very low frequen-cies

In a given material, more than one type of polarization can exist,and the net polarization is given by

Normally, the dipoles are randomly oriented in the material, and theresulting internal electric field is zero In the presence of an externalapplied electric field, the dipoles become oriented as shown in Fig 7.12 There are two common ways to categorize dielectric materials: aspolar or nonpolar, and as paraelectric or ferroelectric Polar materialsinclude those that are primarily molecular in nature, such as water,and nonpolar materials include both electronically and ionically polar-ized materials Paraelectric materials are polarized only in the pres-ence of an applied electric field and lose their polarization when thefield is removed Ferroelectric materials retain a degree of polarizationafter the field is removed Materials used as ceramic substrates areusually nonpolar and paraelectric in nature An exception is siliconcarbide, which has a degree of molecular polarization.

In the presence of an electric field that is changing at a high quency, the polarity of the dipoles must change at the same rate as thepolarity of the signal to maintain the dielectric constant at the samelevel Some materials are excellent dielectrics at low frequencies, butthe dielectric qualities drop off rapidly as the frequency increases.Electronic polarization, which involves only displacement of freecharge and not ions, responds more rapidly to the changes in the di-rection of the electric field and remains viable up to about 1017 Hz.The polarization effect of ionic displacement begins to fall off at about

fre-1013 Hz, and molecular and space charge polarizations fall off at stilllower frequencies The frequency response of the different types isshown in Fig 7.13, which also illustrates that the dielectric constantdecreases with frequency

Figure 7.12 Orientation of dipoles in an electric

field.1

Changing the polarity of the dipoles requires a finite amount of ergy and time The energy is dissipated as internal heat, quantified by

en-a pen-aren-ameter cen-alled the loss ten-angent or dissipen-ation fen-actor Furthermore,

dielectric materials are not perfect insulators These phenomena may

be modeled as a resistor in parallel with a capacitor The loss tangent,

as expected, is a strong function of the applied frequency, increasing

as the frequency increases

In alternating current applications, the current and voltage across

an ideal capacitor are exactly 90° out of phase, with the current ing the voltage In actuality, the resistive component causes the cur-rent to lead the voltage by an angle less than 90° The loss tangent is ameasure of the real or resistive component of the capacitor and is thetangent of the difference between 90° and the actual phase angle

lead-(7.33)

The loss tangent is also referred to as the dissipation factor (DF).

Figure 7.13 Frequency effects on dielectric materials.

Loss tangent = tan(90°–δ)

The loss tangent may also be considered as a measure of the timerequired for polarization It requires a finite amount of time to changethe polarity of the dipole after an alternating field is applied The re-sulting phase retardation is equivalent to the time indicated by thedifference in phase angles.

7.7 Metallization of Ceramic Substrates

There are three fundamental methods of metallizing ceramic strates; thick film, thin film, and copper, which includes direct bondcopper (DBC), plated copper, and active metal braze (AMB) Not all ofthese processes are compatible with all substrates The selection of ametallization system depends on both the application and the compat-ibility with the substrate material

sub-7.7.1 Thick Film

The thick film process is an additive procedure by which conductive,resistive, and dielectric (insulating) patterns in the form of a viscouspaste are screen printed, dried, and fired onto a ceramic substrate at

an elevated temperature to promote the adhesion of the film In thismanner, by depositing successive layers as shown in Fig 7.14, multi-layer interconnection structures can be formed that may contain inte-grated resistors, capacitors, or inductors

The initial step is to generate 1:1 artworks corresponding to eachlayer of the circuit The screen is a stainless steel mesh with a meshcount of 80–400 wires/inch The mesh is stretched to the proper ten-sion and mounted to a cast aluminum frame with epoxy It is coatedwith a photosensitive material and exposed to light through one of theartworks The unexposed portion is rinsed away, leaving openings inthe screen mesh corresponding to the pattern to be printed

Thick film materials in the fired state are a combination of glass

ce-ramic and metal, referred to as cermet thick films, and are designed to

be fired in the range 850–1000°C A standard cermet thick film pastehas four major ingredients

1 An active element, which establishes the function of the film

2 An adhesion element, which provides the adhesion to the strate and a matrix that holds the active particles in suspension

sub-3 An organic binder, which provides the proper fluid properties forscreen printing

4 A solvent or thinner, which establishes the viscosity of the vehiclephase

7.7.1.1 The active element. The active element within the paste tates the electrical properties of the fired film If the active element is

dic-a metdic-al, the fired film will be dic-a conductor; if it is dic-a conductive metdic-aloxide, a resistor; and, if it is an insulator, a dielectric The active ele-ment is most commonly found in powder form ranging from 1 to 10 µ

in size, with a mean diameter of about 5 µ

7.7.1.2 The adhesion element. There are two primary constituentsused to bond the film to the substrate: glass and metal oxides, which

may be used singly or in combination Films that use a glass, or frit, are referred to as fritted materials and have a relatively low melting

Figure 7.14 Screen printing process for material deposition onto a substrate.

point (500–600°C) There are two adhesion mechanisms associatedwith the fritted materials: a chemical reaction and a physical reaction.

In the chemical reaction, the molten glass chemically reacts with theglass in the substrate to a degree In the physical reaction, the glassflows into and around the irregularities in the substrate surface Thetotal adhesion is the sum of the two factors The physical bonds aremore susceptible to degradation by thermal cycling or thermal storagethan the chemical bonds and are generally the first to fracture understress The glass also creates a matrix for the active particles, holdingthem in contact with each other to promote sintering and to provide aseries of three-dimensional continuous paths from one end of the film

to the other Principal thick film glasses are based on B2O3-SiO2 work formers with modifiers such as PbO, Al2O3, Bi2O3, ZnO, BaO,and CdO added to change the physical characteristics of the film, such

net-as melting point, viscosity, and coefficient of thermal expansion Bi2O3also has excellent wetting properties, both to the active element and tothe substrate, and is frequently used as a flux The glass phase may be

introduced as a pre-reacted particle or formed in-situ by using glass

precursors such as boric oxide, lead oxide, and silicon Fritted tor materials tend to have glass on the surface, making subsequentcomponent assembly processes more difficult

conduc-A second class of materials utilizes metal oxides to provide the hesion to the substrate In this case, a pure metal, such as copper orcadmium, is mixed with the paste and reacts with oxygen atoms onthe surface of the substrate to form an oxide The conductor adheres tothe oxide and to itself by sintering, which takes place during firing.During firing, the oxides react with broken oxygen bonds on the sur-face of the substrate, forming a Cu or Cd spinel structure, such asCuAl2O4 Pastes of this type offer improved adhesion over fritted ma-

ad-terials and are referred to as fritless, oxide-bonded, or

molecular-bonded materials Fritless materials typically fire at 900–1000°C,

which is undesirable from a manufacturing aspect Ovens used forthick film firing degrade more rapidly and need more maintenancewhen operated at these temperatures for long periods of time

A third class of materials utilizes both reactive oxides and glasses.The oxides in these materials react at lower temperatures but are not

as strong as copper A lesser concentration of glass than found in ted materials is added to supplement the adhesion These materials,

frit-referred to as mixed bonded systems, incorporate the advantages of

both technologies and fire at a lower temperature

The selection of a binding material is strongly dependent on thesubstrate material For example, the most common glass compositionused with alumina is a lead/bismuth borosilicate composition Whenthis glass is used in conjunction with aluminum nitride, however, it is

rapidly reduced at firing temperatures.1 Alkaline earth borosilicatesmust be used with AlN to promote adhesion

7.7.1.3 Organic binder. The organic binder is generally a thixotropicfluid and serves two purposes: it holds the active and adhesion ele-ments in suspension until the film is fired, and it gives the paste theproper fluid characteristics for screen printing The organic binder is

usually referred to as the nonvolatile organic, since it does not

evapo-rate but begins to burn off at about 350°C The binder must oxidizecleanly during firing, with no residual carbon that could contaminatethe film Typical materials used in this application are ethyl celluloseand various acrylics

For nitrogen-fireable films, where the firing atmosphere can containonly a few ppm of oxygen, the organic vehicle must decompose andthermally depolymerize, departing as a highly volatile organic vapor

in the nitrogen blanket provided as the firing atmosphere, since tion into CO2 or H2O is not feasible due to the oxidation of the copperfilm

oxida-7.7.1.4 Solvent or thinner. The organic binder in the natural form istoo thick to permit screen printing, which requires the use of a solvent

or thinner The thinner is somewhat more volatile than the binder,evaporating rapidly above about 100°C Typical materials used forthis application are terpineol, butyl carbitol, or certain of the complexalcohols into which the nonvolatile phase can dissolve The low vaporpressure at room temperature is desirable to minimize drying of thepastes and to maintain a constant viscosity during printing Addition-ally, plasticizers, surfactants, and agents that modify the thixotropicnature of the paste are added to the solvent to improve paste charac-teristics and printing performance

To complete the formulation process, the ingredients of the thickfilm paste are mixed together in proper proportions and milled on athree-roller mill for a sufficient period of time to ensure that they arethoroughly mixed and that no agglomeration exists

Thick film conductor materials may be divided into two broadclasses; air fireable and nitrogen fireable Air fireable materials aremade up of noble metals that do not readily form oxides, gold and sil-ver in the pure form, or alloyed with palladium and/or platinum Ni-trogen fireable materials include copper, nickel, and aluminum, withcopper being the most common

Thick film resistors are formed by adding metal oxide particles toglass particles and firing the mixture at a temperature/time combina-

tion sufficient to melt the glass and to sinter the oxide particles gether The resulting structure consists of a series of three-dimensionalchains of metal oxide particles embedded in a glass matrix The higherthe metal oxide-to-glass ratio, the lower the resistivity, and vice versa.The most common materials used are ruthenium based, such as ruthe-nium dioxide, RuO2, and bismuth ruthenate, BiRu2O7.

to-Thick film dielectric materials are used primarily as insulators tween conductors, either as simple crossovers or in complex multilayer

be-structures Small openings, or vias, may be left in the dielectric layers

so that adjacent conductor layers may interconnect In complex tures, as many as several hundred vias per layer may be required Inthis manner, complex interconnection structures may be created Al-though the majority of thick film circuits can be fabricated with onlythree layers of metallization, others may require several more If morethan three layers are required, the yield begins dropping dramatically,with a corresponding increase in cost

struc-Dielectric materials used in this application must be of the

devitrify-ing or recrystallizable type Dielectrics in the paste form are a mixture

of glasses that melt at a relatively low temperature During firing,when they are in the liquid state, they blend together to form a uni-form composition with a higher melting point than the firing tempera-ture Consequently, on subsequent firings they remain in the solidstate, which maintains a stable foundation for firing sequential layers

By contrast, vitreous glasses always melt at the same temperatureand would be unacceptable for layers to either “sink” and short to con-ductor layers underneath, or “swim” and form an open circuit Addi-tionally, secondary loading of ceramic particles is used to enhancedevitrification and to modify the temperature coefficient of expansion(TCE)

Dielectric materials have two conflicting requirements in that theymust form a continuous film to eliminate short circuits between layersand, at the same time, they must maintain openings as small as 0.010

in In general, dielectric materials must be printed and fired twice perlayer to eliminate pinholes and prevent short circuits between layers.The TCE of thick film dielectric materials must be as close as possi-

ble to that of the substrate to avoid excessive bowing, or warpage, of

the substrate after several layers Excessive bowing can cause severeproblems with subsequent processing, especially where the substratemust be held down with a vacuum or where it must be mounted on aheated stage In addition, the stresses created by the bowing cancause the dielectric material to crack, especially when it is sealedwithin a package Thick film material manufacturers have addressedthis problem by developing dielectric materials that have an almostexact TCE match with alumina substrates Where a serious mismatch

exists, matching layers of dielectric must be printed on the bottom ofthe substrate to minimize bowing, which obviously increases the cost.

7.7.2 Thin Film

The thin film technology is a subtractive technology in that the entiresubstrate is coated with several layers of metallization, and the un-wanted material is etched away in a succession of selective photoetch-ing processes The use of photolithographic processes to form thepatterns enables much finer and better defined lines than can beformed by the thick film process This feature promotes the use thethin film technology for high-density and high-frequency applications.Thin film circuits typically consist of three layers of material depos-ited on a substrate The bottom layer serves two purposes: it is the re-sistor material, and it also provides the adhesion to the substrate Theadhesion mechanism of the film to the substrate is an oxide layer thatforms at the interface between the two The bottom layer must there-fore be a material that oxidizes readily The most common types of re-sistor material are nichrome (NiCr) and tantalum nitride (TaN) Goldand silver, for example, are noble metals and do not adhere well to ce-ramic surfaces

The middle layer acts as an interface between the resistor layer andthe conductor layer, either by improving the adhesion of the conductor

or by preventing diffusion of the resistor material into the conductor.The interface layer is for TaN is usually tungsten (W); for NiCr, a thinlayer of pure Ni is used

Gold is the most common conductor material used in thin film cuits because of the ease of wire and die bonding and the high resis-tance of the gold to tarnish and corrosion Aluminum and copper arealso frequently used in certain applications It should be noted thatcopper and aluminum will adhere directly to ceramic substrates, butgold requires one or more intermediate layers, since it does not formthe necessary oxides for adhesion

cir-The term thin film refers more to the manner in which the film is

deposited onto the substrate as opposed to the actual thickness of thefilm Thin films are typically deposited by one of the vacuum deposi-tion techniques, sputtering or evaporation, or by electroplating.Sputtering is the prime method by which thin films are applied tosubstrates In ordinary dc triode sputtering, as shown in Fig 7.15, acurrent is established in a conducting plasma formed by striking anarc in an inert gas, such as argon, with a partial vacuum of approxi-mately 10 µ pressure A substrate at ground potential and a targetmaterial at high potential are placed in the plasma The potential may

be ac or dc The high potential attracts the gas ions in the plasma to

the point at which they collide with the target with sufficient kineticenergy to dislodge microscopically sized particles with enough resid-ual kinetic energy to travel the distance to the substrate and adhere.The adhesion of the film is enhanced by presputtering the substratesurface by random bombardment of argon ions prior to applying thepotential to the target This process removes several atomic layers ofthe substrate surface, creating a large number of broken oxygen bondsand promoting the formation of the oxide interface layer The oxideformation is further enhanced by the residual heating of the substrate

as a result of the transfer of the kinetic energy of the sputtered cles to the substrate when they collide

parti-Direct current triode sputtering is a very slow process, requiringhours to produce films with a usable thickness By utilizing magnets

at strategic points, the plasma can be concentrated in the vicinity ofthe target, greatly speeding up the deposition process In most appli-cations, an RF potential at a frequency of 13.56 MHz is applied to thetarget The RF energy may be generated by a conventional electronicoscillator or by a magnetron The magnetron is capable of generatingconsiderably more power with a correspondingly higher depositionrate

By adding small amounts of other gases, such as oxygen and gen to the argon, it is possible to form oxides and nitrides of certain

nitro-target materials on the substrate It is this technique, called reactive

sputtering, that is used to form tantalum nitride, a common resistor