IPC-D-279 EN.pdf - 第61页

Appendix B Design for Reliability (DfR) of Plated-Through Via (PTV) Structures B-1.0 PLATED-THROUGH VIA (PTV) RELIABILITY ISSUES Plated-through-holes (PTHs) serve to electrically connect dif ferent conductor layers in mu…

41. MIL-HDBK-217F, Military Handbook—Reliability

Prediction of Electronic Equipment, January 15, 1982,

Departmental of Defense, Washington, DC, 20301.

42. Orsten, G. S. F.,W. New, Y. Wang, and M. L. Peloquin

‘‘SMT Considerations in Spaceﬂight and Critical Mili-

tary Applications,’’ Proc. 18th Ann. Electronics Manu-

facturing Seminar, China Lake, CA, February 1994,

pp. 75-89.

43. Orsten, G. S. F., ‘‘The Problems Associated with

Adhesive Bonding of Components on Surface Mount

Assemblies,’’ Proc. 19th Ann. Electronics Manufactur-

ing Seminar, China Lake, CA, February 1995, pp. 153-

163.

44. Engelmaier, W., ‘‘Reliability Figures of Merit for Sur-

face Mount Solder Attachments of Components: 2nd

Generation Generic Design Tools,’’ Proc. Surface

Mount International Conf., San Jose, CA, August

1991, pp. 1239-1243.

45. Katchmar, R., ‘‘Position Dependence of CTE in Plas-

tic Ball Grid Arrays,’’ Proc. Int. Electronic Packaging

Conf. (IEPS). Atlanta, September 1994, pp. 271-283.

46. Engelmaier, W., ‘‘Solder Joint Reliability for BGAs

and Other Advanced Electronic Components,’’ Work-

shop Notes, Engelmaier Associates, Inc., Mendham,

NJ, 1995.

47. Freyman, B., ‘‘Trends in Plastic BGA Packaging,’’

Proc. Ball Grid Array Nat. Symp., Dallas, TX, March

1995.

48. Mawer, A. J., S.C. Bolton, and E. Mammo, ‘‘Plastic

BGA Solder Joint Reliablity Considerations,’’ Proc.

Surface Mount International Conf.. San Jose, CA,

August-September 1994, pp. 239-251.

49. Attarwala, A. I., and R. Stierman, ‘‘Failure Mode

Analysis of a 540 Pin Plastic Ball Grid Array,’’ Proc.

Surface Mount International Conf., San Jose, CA,

August-September 1994, pp. 252-257.

50. Phelan, G., and S. Wang, ‘‘Solder Ball Connection

Reliability Model and Critical Parameter Optimiza-

tion,’’ Proc. 43rd Electronic Components and Technol-

ogy Conf., Orlando, FL, June 2-4, 1993, pp. 858-862.

51. Rukavina, J., ‘‘Ball Grid Array Attachment Methodolo-

gies,’’ Proc. Ball Grid Array Nat. Symp.. Dallas, TX,

March 1995.

52. Bogatin. E.. ‘‘BGAs for Workstation Application,’’

Proc. Ball Grid Array Nat. Symp., Dallas, TX, March

1995.

53. ANSI/IPC-MF-150F, ‘‘Metal Foil for Printed Wiring

Applications,’’ The Institute for Interconnecting and

Packaging Electronic Circuits, Northbrook, IL, Octo-

ber 1991.

54. Reichelt, G., DeTeWe, Germany. Report on Results of

German Joint Project on SMT-Reliability to Working

Group 2 of IEC Technical Committee TC91, Surface

Mounting Technology, Anaheim, CA, February 27,

1995.

55. Ehrich, M., M. Kampfert, D. Naujokat and G. Reich-

elt, ‘‘Lotstellensichtprtufung: Eine Kritische Stellung-

nahmeausder Sichtvon Dauerbelastbarkeitsuntersu-

chungen an Oberﬂachenmontierten

Keramikvielschichtkondensatoren mit Systematischen

Montageabweichungen, ’’Verbindungstechnik in der

Elektronik, Vol. 7, Heft 1, March 1995, pp. 44-51.

July 1996 IPC-D-279

Appendix B

Design for Reliability (DfR) of Plated-Through Via (PTV) Structures

B-1.0 PLATED-THROUGH VIA (PTV) RELIABILITY

ISSUES

Plated-through-holes (PTHs) serve to electrically connect

different conductor layers in multilayer printed circuit

boards (MLBs). In conventional interconnection technol-

ogy employing through-mounted components, the PTHs

also serve the function of providing a structure that accepts

the component leads and to which these leads can securely

be solder attached. This structure consisting of the copper

PTH barrel containing a component lead and ﬁlled with

solder provides a very robust, multiple-redundant electrical

and mechanical connection between the component and the

MLB.

The continuing drive towards higher functionality, higher

density, and lower weight was brought about and made

possible by the development of surface mounted technol-

ogy. This reduced the purpose of the previously multi-

functional PTHs to providing only the electrical intercon-

nections between the MLB layers; the diameters of PTHs

could also be reduced, since they no longer needed to be

large enough for the component lead insertion operation.

To distinguish the two types of PTHs, the PTHs without

component leads are frequently referred to as ‘PTH-vias’ or

PTVs. At the same time, the increased functionality and

density of the components often brought about a need to

increase the number of layers in MLBs, and thus to

increase the MLB thickness.

The decreasing PTV diameters, particularly in combination

with the increasing MLB thicknesses, make copper plating

into the PTVs more difficult. This problem was ﬁrst recog-

nized during an IPC round robin study [Ref. B-7: 1], which

led to a more detailed round robin study speciﬁcally

focused on this issue [Ref. B-7: 2]. PTVs having a small

diameter—less than ~0.5 mm—and/or a high aspect ratio

of MLB thickness to drilled PTV diameter—more than

4—were found to require special treatment for adequate

reliability.

During this study the material properties, processing

parameters, and environmental test and use conditions

important for the reliability of PTVs were identiﬁed. The

results of these studies together with prior and subsequent

work [Refs. B-7: 3-9] has been utilized to develop a prac-

tical methodology to aid in the DfR of PTVs, as well as to

permit the assessment of the reliability of PTVs given the

assembly and test procedures and the use environments in

the ﬁeld [Ref. B-7: 10].

In the IPC round robin test program [Ref. B-7: 1] the IEC

(Hot Oil Test, IEC 362-2) test—designed to simulate solder

reﬂow thermal shock cycles—was used to evaluate the reli-

ability of small-diameter PTVs. It was found that assembly

processes involving large temperature excursions constitute

a signiﬁcant reliability threat due to low-cycle fatigue for

PTV copper barrels with low ductility or large stress con-

centrations. It was concluded that failures, to the extent

they occur at all, occur typically in the ﬁrst 10 cycles due

to overstress crack initiation followed by crack propaga-

tion. This conclusion was reinforced by the ﬁndings of

Oien [Refs. B-7: 3, 4], which showed that crack initiation

typically occurs during the ﬁrst or second cycle of over-

stress. Unless failure occurs within 10 cycles of the over-

stress loading typical of solder reﬂow, solder reﬂow over-

stressing is not a problem. Additional cycles will

eventually lead to fatigue-induced failures. The failures

that occurred in the IPC study were observed in product

from vendors rated ‘poor’ to ‘good’ on an arbitrarily sub-

jective scale with some consistency in differentiation. No

superior-rated vendor product failed, but it has to be noted

that the testing had an arbitrary cut-off. This led to the

deﬁnition of a numerical quality index that is now utilized

in an improved more detailed form in the reliability mod-

eling shown in Reference B-7:2 [Ref. B-7: 10].

During the product use, the severity of the thermal use

environment has a great impact on the reliability of the

PTVs. The ‘‘MIL-T’’ and ‘‘COM-T’’ thermal cycling tests

[Ref. B-7: 2] were designed to simulate severe and rela-

tively benign use environments, respectively. While fail-

ures in the ‘‘MIL-T’’ test occurred depending on the con-

struction of the MLBs and the quality of the PTVs, failures

did not occur as the result of the ‘‘COM-T’’ test. It needs

to be noted however, that the ‘‘MIL-T’’ and ‘‘COM-T’’

thermal cycling tests unfortunately had arbitrary test cut-

offs at 400 and 1000 cycles, respectively.

B-1.1 Copper Plating Process

B-1.1.1 Acid Copper Plating

One ﬁnding of the IPC

round robin study [Ref. B-7: 2] was that PTVs with aspect

ratios larger than three and plated with standard electrolytic

acid copper show decreasing thermal cyclic fatigue life. It

was found that the copper plating process window narrows

as the PTV aspect ratio increases and that the standard

electrolytic plating processes become inadequate even with

optimum process controls. These ﬁndings agree with ear-

lier studies investigating the effects of plating current den-

sity and agitation level on copper deposit quality in PTVs

[Ref. B-7: 5]. In this study it was shown that inadequate

electrolyte replacement rates, which can clearly occur in

high-aspect-ratio PTVs, will lead to mass-transport limited

IPC-D-279 July 1996

plating conditions. Under these conditions, in combination

with the non-uniform plating current densities that also get

worse with increasing aspect ratios, the copper deposit

quality rapidly deteriorates with increasing aspect ratios.

The resulting copper deposits within the PTV can have

signiﬁcantly lower ductility and strength than the copper

deposits plated at the same time external to the PTVs, e.g.

on plating mandrels. The decline in physical properties is

frequently accompanied by increased ‘dog-boning’ and

nodule formation as well [Ref. B-7: 10].

The copper deposits from standard acid copper baths in

high-aspect-ratio PTVs, even with uniform plating in the

PTV barrels and good intrinsic tensile properties, as deter-

mined by testing foil specimens from ﬂat plating mandrels,

perform only marginally in thermal cycling tests. For this

reason, special plating solutions have been developed, that

allow reduced plating current densities at the expense of

increased plating times, but producing signiﬁcantly

improved PTV copper barrel reliability.

B-1.1.2 Pyrophosphate Copper Plating Pyrophosphate

plated copper was unfortunately not part of the IPC round

robin studies [Refs. B-7: 1, 2]. It has however been shown,

that pyrophosphate copper is less susceptible to the effects

of non-uniform plating current densities and higher PTV

aspect ratios.

B-1.2 Material Properties

B-1.2.1 Tensile Properties

The tensile properties of the

PTV copper deposits are very important, both for the per-

formance of the PTVs during subsequent processing and

use, and for the DfR for the PTVs. The properties that are

needed are: (1) the tensile strength, (2) the yield strength,

(3) the modulus of elasticity, (4) the modulus of plasticity,

and (5) the fracture ductility.

It has also been found, that the electrolytically plated cop-

per deposits have a modulus of elasticity signiﬁcantly

below that of cast and rolled copper reported in material

property references. It therefore is necessary to measure the

modulus of elasticity, which can be done during the tests to

determine the yield and tensile strengths of the deposits.

The tensile strength, the yield strength, the modulus of

elasticity and the modulus of plasticity can be determined

from tensile tests. To properly determine the modulus of

elasticity, the ‘interrupted tensile test’ method should be

utilized.

It needs to be noted however, that these tensile properties

come from samples plated onto ﬂat stainless steel mandrels

and only set the upper bounds for the strength and ductil-

ity of the copper deposit inside the PTV. The evidence is

circumstantial, but very strong, that these properties are

signiﬁcantly degraded inside of high-aspect-ratio (board

thickness/PTV-diameter) PTVs [Ref. B-7: 2].

B-1.2.2 Ductility The tensile elongation is a very inaccu-

rate and subjective test for foil samples because of the

specimen geometry and the dependence on the test condi-

tions [Ref. B-7: 10, 11]. Tensile elongation for foil materi-

als is adequate for quality control and comparison pur-

poses, but it signiﬁcantly underestimates the fracture

ductility of the material by about a factor of three (3) and

gives the false indication of a ductility dependence on foil

thickness.

It is for these reasons that the test cited in References

B-7:12 and 13, was developed. This test has a high dis-

crimination power in terms of quality variations of the cop-

per deposit, thus being very valuable as a process control

tool as well as providing direct input for the all-important

ductility of the copper deposit.

In order to assess the quality of the plated copper deposits,

foil samples plated onto mandrels need to be subjected to

fatigue ductility testing [Ref. B-7: 12].

It needs to be noted however, that this ductility comes from

samples plated onto ﬂat stainless steel mandrels and only

set the upper bounds for the ductility of the copper deposit

inside the PTV. The evidence is circumstantial, but very

strong, that this property is signiﬁcantly degraded inside of

high-aspect-ratio (board thickness/PTV-diameter) PTVs

[Ref. B-7: 2].

The ductility of the copper deposit in the PTV barrel can

be determined from the performance of the PTVs in accel-

erated testing resulting in low-cycle fatigue. In Table B-1

the results from two of the tests used in the Reference

B-7:2 applied to coupons from the same sample are given

together with the stresses and strain ranges resulting from

the thermal cycling/shock excursions. Also given is the

minimum ductility resulting from the ﬁrst failures in the

tests using Eq. #1 in Section B-2.0 together with the initial

tensile strength.

The measured mandrel ductility was about 30%. Thus, the

ductility in the PTV barrels as given in Table B-1 indicates

a signiﬁcantly lower deposit ductility as compared to the

ductility of the deposit on the MLB surface.

B-1.2.3 Fatigue Behavior It has been found that

annealed electrolytic acid copper strain-hardens upon the

application of cyclic strain loads during fatigue tests [Ref.

B-7: 11]. This results in high-cycle fatigue lives that are

longer than expected based on the initial material proper-

ties.

Conversely, annealed pyro-phosphate-plated copper strain-

softens as the result of the application of the cyclic strains

during the fatigue testing. Therefore, fatigue life tests in the

July 1996 IPC-D-279