S e s s s s i i o n 2 : : A s s s s e s s s s m e n t t s s o f f C h a n g e

Changing to Pb-free Profoundly
Impacts the Manufacturing Production
Process
By
Richard Brooks & David Day
Indium Corporation of America
Vahid Goudarzi
Abstract — The paper will outline the issues related to the implementation of a Pb-free solder paste into a standard Sn/Pb manufacturing facility and product. The Pb-free study includes the compatibility and impact to the various manufacturing processes that include, screen printing, component placement, reflow process, and solder joint quality. These parameters must be fully characterized to ensure that the lead-free solder paste meets the manufacturing and product quality requirements. In addition, a lower peak reflow temperature is critical to reduce the thermal stress on components, since the lead-free alloy composition (Sn/Ag/Cu) liquifies at about 34 ° C higher than the current leaded material (Sn/Pb).
Several implementation issues were discovered during the pilot phases and resolved prior to fullscale production manufacturing. Some of the problems revealed were BGA voiding, chip component tombstoning, and component integrity as it relates to termination plating and moisture sensitivity.
I NTRODUCTION
Motorola Corporation concern. Silver is a known toxin. Silver can kill up to 650 different viral, bacterial and fungal organisms, as well as most marine life. The SAC alloy also fails the EPA testing for leaching in standard landfills. Therefore, the elimination of Pb may not be the answer to the problem, but only creating additional problems to ones that already exist. Recycling should be the long-term solution to the problem. So, expect additional legislations and higher landfill costs to come in the near future.
The issue is, Pb-free manufacturing will become a reality in the near future, so we better be ready for it.
Recent legislation in Europe has further led OEMs to consider the implementation of Pb-free products into the industry. In the last year, the elimination of lead from the consumer electronics has mainly been pursued in Japan. As a result, major Japanese
OEMs are removing lead from their products and making them environmentally friendly. There are a number of lead-free alloy compositions available for consumer electronic products. Electronic
Industry groups (NEMI, IDEALS, JEITA) have evaluated several of these lead-free materials and they have chosen the Sn/Ag/Cu alloy for the midterm solution. The Sn/Ag/Cu alloy (is being called
SAC, for S n – A g - C u) reflows at a temperature of
217 o C, which is approximately 34 o C higher than the standard Sn/Pb alloys. There are several issues besides the higher reflow temperatures that are of
Pb-Free Implementation and Concerns
What are the concerns when implementing a Pbfree soldering process (or Pb-free reduce by using a SAC alloy)? The first and most critical concern is the component integrity. The components will have to withstand higher processing temperatures, of which most are not specified. Therefore, in order to implement a Pb-free solder paste into production, the components must first be evaluated and qualified to survive higher process temperatures. Most recently, IPC is attempting to develop new standards for the Pb-free soldering process (020B specification). As an assembly manufacturer, it is required to work with the component supplier on all new product designs in order to implement a Pb-free soldering process.
A second concern with implementing a Pb-free soldering process is the qualification of a Pb-free solder paste that is capable of handling a very large process window. Ideally, the manufacturer would like to have a solder paste that is capable of reflowing at very low temperatures (230 o very high temperatures (260 o
C) and at
C). The former is potentially more important because if a very low processing temperature is capable then a low peak reflow temperature is possible and the component integrity may not be an issue. The following section details one possible Pb-free solder paste
31

evaluation method for determining the reflow process window.
Reflow Profile Process Window
Phase 1 – Bare board wetting
To evaluate the reflow process window of a solder paste, various peak temperatures and liquidus times should be experimented. A minimum of three reflow profiles can be used with a low peak temperature of 230 of 245 o o C, a medium temperature peak
C and a high temperature of 260 profile is shown in Figure 1.
o C. An example of a standard Pb-free reflow soldering and is considered to be a failure. The lack of coalescence in Figure 2 can be attributed to the oxidation of the solder particles through the reflow process at the higher process temperatures or the inability to provide adequate fluxing at the lower process temperatures.
Figure 3 shows full coalescent solder particles after the reflow process. This should be considered the minimum requirement for an acceptable solder joint after the reflow process.
Reflow Temp 217 ° C
0 1 2 3 4 5
Time (min)
6 7 8
Figure 1 – Typical Pb-free reflow profile
The criteria to compare the reflow performance of the Pb-free solder paste materials should be based on the wetting performance of the solder paste in air atmosphere using various PCB finishes at the different reflow profiles.
Figure 2 - Non-coalescent Solder
The wetting performance of the lead-free solder pastes can be determined based on the coalescent of the solder particles after the reflow process.
Figure 2 shows poor performance of a solder paste, as the solder powder did not completely coalesce
Figure 3 - Fully coalescent solder
Phase 2 – Component wetting
The reflow performance of a Pb-free solder paste must also be evaluated in soldering performance on standard components. A typical production or test board can be used that contains standard components. The test boards should be reflowed again using the various reflow profiles (low, medium and high peak temperatures).
The following cross-sections exhibit very good wetting results of component soldering when using a Pb-free solder paste with a wide process window.
Figure 4 shows the cross-section of a component at the low reflow peak of 230 o C.
Figure 4 – Component soldered at peak of 230 o C
32

Figure 5 – Component soldered at peak of 245 o
Figure 5 provides a sample of a component soldered at a peak temperature of 245 temperature of 260 o C.
o C. And
Figure 6 shows a component soldered at a peak
Figure 6 – Component soldered at 260 o C
C
Figure 7 shows a cross section of a component that was soldered at a peak temperature of 210 o C using a typical lead containing solder paste (63Sn/37Pb).
In this case, it was observed from the crosssections, that the component solder joint fillets for both the Pb-free and the Sn/Pb alloys are comparable.
Figure 7 – Sn/Pb soldered component at 210 o C
Though comparable solder joints can be observed in this case, however, this was not always true with
Pb-free alloys. The solder paste must wet to a variety of component and PCB terminations.
Component terminations changed drastically as the
Pb-free initiative began to snowball, but it was also believed that the components would lag in the implementation of a Pb-free termination. During the process of component terminations migrating toward Pb-free solutions, the solder paste must be able to wet to all types of exotic metallurgy. Also, during this migration, we must also be aware of the forward and backward compatibility of the various
PCB and component terminations.
The following table (Table 1) provides some idea as to the various component and PCB terminations available today.
PCB finishes
HASL
Comp finishes
Various Sn/Pb comp
Various OSP
Immersion Ag
Sn
Au
ENIG Ag
Pd/Ni Pd/Ni
Immersion Sn
Pb-free HAL
Ni/Pd
Ni/Au
Ag/Pd
Ni/Au/Cu
Pd/Ni
Table 1 – PCB & Component terminations
Determining the Product Readiness
After evaluation of the various solder pastes and the ability to solder components at the required higher temperatures, a product was designed to undergo the Pb-free transition. The preferred product should have a minimal thermal gradient across the entire circuit board, so that a lower process temperature can be implemented. The most critical stage of the Pb-free implementation was the selection of the components. The components must be capable to handle the higher reflow temperatures without issues. The required reflow temperatures must be implemented in the component specification. If a component cannot function at the increased process temperatures, then either a new supplier must be chosen or component analysis completed to verify the capability of the component.
Production Evaluation and Implementation
During the pilot stages and product qualification of the Pb-free soldering process, several manufacturing issues were revealed that had to be resolved prior to full-scale production.
Some of the typical problems observed during the product pilot stages are given below:
33

o Reduce wetting or spreading o Increase in tombstone components o Increase in BGA voiding on Sn/Pb bumps o Component integrity during soldering o Component MSL rating o Soldering to Gold plated contacts or connectors
The following sections will detail the resolutions to the manufacturing product issues.
Reduced wetting or spreading
The implementation of a Pb-free soldering process is different than the standard Sn/Pb process. The most obvious observed difference in the soldering is the visual wetting difference. The Pb-free solder will not wet completely out to the solder pads.
This may only be a cosmetic difference but the operators must be trained to inspect for the difference. If the reduced wetting is believed to cause a problem, for example exposed copper on the PCB, then an increase in stencil apertures may be required to resolve the issue.
Tombstone defects
One of the largest problems found in the initial pilot stages was the increase in small chip component tombstone performance. Initially, it was thought that the reduced wetting of the Pb-free
(SAC) alloy would enable an improvement in the tombstone performance, but this was not the case.
Compared to the Sn/Pb alloys, the Pb-free SAC alloys have higher surface tension and thus a stronger pulling force on the chip component.
Figure 8 shows an example of a typical tombstone failure.
Figure 8 – 0402 tombstone failure
Therefore, the chip component pad design is critical for optimum soldering performance. It was noted that the reduction of overall pad length assisted in improving the tombstone performance.
Also a reduction of paste deposition may also be considered in order to optimize the soldering performance.
From Figure 8, it was observed that a via-in-pad design was utilized. This was also noted to increase the tombstone occurrence and was eliminated in future designs. The use of HDI technology can be considered for designs that require this application.
Additionally, the reflow profile was also found to have an impact on the tombstone performance.
The critical part of profile was observed to be the ramp or soak portion of the heating cycle.
Therefore, the optimization of the profile is also necessary to improve the soldering performance of small chip capacitors and resistors.
BGA Solder Voids
The next major problem found in implementing a
Pb-free solder paste was the compatibility with
Sn/Pb solder bumps on BGAs or CSPs. The problem discovered was very large solder voids in the solder joint. See Figure 9 below as an example of the typical BGA solder joint formation using a Pb-free paste and Sn/Pb (lower melting) ball.
Figure 9 – BGA solder void
The problem is simple and easy to understand.
The Sn/Pb ball on the BGA component reflows prior to the higher temperature (Pb-free) solder paste wetting and this is a problem. The typical solder paste makeup for the Pb-free solder paste contains solvents and activators that are active and volatilize at temperatures higher than 180 o C. Thus, as the BGA solder ball begins to reflow, the solder paste is still actively wetting and volatizing within the solder ball.
Several experiments were completed to evaluate the main effects of the BGA voids. It was found that modifying the reflow profile had the greatest
34

effect. Experimentation and modification of the ramp rate or soak time and temperatures of the reflow profile can minimize BGA voids.
Additional notes for soldering to a Sn/Pb BGA component, are that if via-in-pad applications were used in this process then even more voids may be observed. Also, a reduction in paste deposition may be attempted to reduce the void formation.
The solution to this problem is to implement Pbfree BGA solder balls for the Pb-free soldering process.
Component Integrity
The higher reflow process temperatures will have an effect on the component integrity. The following are a few observations noted during the pilot builds.
1) The higher process temperatures can discolor or degrade the appearance of the component.
This may be purely cosmetic, but needs to be understood and evaluated for the product.
2) Some components have sensitivity to absorbing moisture. These components are presently controlled in the standard production environments by placing them in dry boxes or baking them if they are exposed to long time humidity conditions. The process temperature will also have effect on the component performance. The higher the peak reflow temperature, the increased possibility of damaging the component. Therefore, experimental analysis must be completed to evaluate the effect on the components ability to perform. Thus, greater care must be taken on ALL components that are sensitive to moisture when evaluating the ability to solder at Pb-free temperatures.
Gold Connector Soldering
This was an issue that did not reveal itself until late in the product pilot stages. It is known that the Pbfree (SAC) solder alloys exhibit less wetting and spreading of the solder on the pad or component, but it was found not to be the case with Gold plated components, such as contacts or connectors.
The high temperature Pb-free solder has an affinity to wet with gold, since gold goes into solution faster than most other metals. It was found that in some cases the Pb-free solder would wet all the way up a component lead termination and intrude into the package. Figure 10 shows the Pb-free solder extending all the way through the package and on the lead. This example had caused the failure to mate with the corresponding female connector and thus a defect.
Pb-free
Figure 10 – Pb-free solder on a Gold plated connector
The resolution to the problem was to modify the paste printing apertures in order to minimize the wetting of the Pb-free solder paste on the gold plated lead.
CONCLUSIONS
A Pb-free soldering process is possible to implement on the typical electronic products today, but it is not a drop-in fit compared to Sn/Pb process. The following
1) A robust Pb-free solder paste must be first qualified for the Pb-free process. The solder paste must have a large process window and capable of soldering at a wide range of temperatures, preferably lower temperatures, but also at mid and high temperatures.
2) The components must be carefully chosen and qualified to perform at the higher process temperatures.
3) The PCB finishes and component terminations should be considered prior to initial product assembly and also the possible compatibility issues.
4) The product design must be well thought out and evaluated for a Pb-free process.
Modifications may be required to the standard pad geometries to optimize the Pbfree soldering process.
5) The Product must undergo Accelerated Life
Testing to evaluate the solder joint strength and reliability of the Pb-free solder joints, as they compare to the standard Sn/Pb solder joints.
The product design and component selections have an extreme effect on the ability to implement a Pbfree soldering process. There are several manufacturers today that have implemented Pb-
35

free products into the standard manufacturing production environment. A majority of these products are only Pb-reduced, since not all components terminations are 100% Pb-free.
Additionally, in order to implement a Pb-free manufacturing solution into production, ALL of the manufacturing processes must be characterized and optimized for peak performance. This is because the Pb-free manufacturing process window is much tighter than the standard Sn/Pb process that is required today.
REFERENCES
1. A.Butterfield, V.Visintainer, V.Goudarzi,
“Lead Free Solder Flux Vehicle Selection
Process”, SMTA International 2000 proceedings.
2. N.C.Lee, Benhlih Huang, “Prospect of Lead
Free Alternatives for Reflow Soldering”
3. E.Bradley, “Overview of No-Lead Solder
Issues, NEMI meeting, Anaheim, February 23,
1999.
36

rd
www.
indium .com
©

Slide #2
•Japan / Europe / US
•Is Pb-free a good thing?
•NEMI alloy investigation
•Manufacturing Issues
•Product Performance and Reliability www.
indium .com
©

Slide #3
www.
indium .com
©

Slide #4
– Dec 2002 EU parliament passed legislation to ban use of:
• Pb, Hg, Cd, Cr VI, PBB, PBDE
– Affects ALL products that are sold after July 2006.
• Some exceptions e.g. lead free solder in telecom equipment, high lead solder applications
– Recycling program enforced June 2005.
www.
indium .com
©

Slide #5
– Japan home electronics recycling law requires OEMs to collect and recycle four major products since April
2001.
www.
indium .com
©

Slide #6
– “The IPC… recognizes there are no data indicating environmental health hazards posed by Pb in
PWB manufacturing and electronics assembly” –
Dennis McGuirk, IPC president www.
indium .com
©

•Pb solder < 0.5% of all Pb used.
•Batteries comprising > 80%.
Slide #7
Ref: M. Buetow, “The Latest on the Lead-Free Issue”, Technical Source, IPC 1999 Spring/Summer Catalog.
www.
indium .com
©

Slide #8
ALL
www.
indium .com
©

Slide #9
www.
indium .com
©

Slide #10
• Despite perceptions, Pb can and does leach into groundwater at landfill sites, particularly where water is acidic
• So too does Pb-free solder!
• Simply changing from SnPb to SnAgCu may not fix the problem
• Recycling must be the long term answer
• Expect more legislation and higher landfill costs in the years to come!
www.
indium .com
©

Slide #11
– 95.5Sn/3.9Ag/0.6Cu
– 217 C liquidus
• Pb-Free is now driven by both market factors and now legislation
• SnAgCu preferred short/medium term solution
– NEMI, JEITA, IDEALS all agree on Sn/Ag/Cu
• When components become completely Pb-Free,
SnAgBi may become preferred solution
– Lower process temp
– Excellent wettability www.
indium .com
©

Slide #12
– PCB Design
– Component integrity
– Reflow Thermal Gradient reduction
– Pb-Free soldering compatibility
www.
indium .com
©

Slide #13
– Majority of components are not spec’d to withstand reflow temperatures >240 C
– Melting temperature of 217C
– Different wetting characteristics compared to Sn/Pb
– Large boards with poor delta T in the oven may need as much as 255C www.
indium .com
©

Wetting Times as a Function of Temperature With a Range Of Alloys With
Un-Activated Flux: (a) Air, (b) Nitrogen
Slide #14 www.
indium .com
©

Slide #15
www.
indium .com
©

Slide #16
•
•
•
•
•
•
•
www.
indium .com
©

Slide #17
www.
indium .com
©

Slide #18
•Air voids cannot be eliminated
www.
indium .com
©

Slide #19
– Reduced wetting
– Chip component tombstoning
– BGA / CSP Voiding
– Gold plated component soldering
– Temperature rating
• Rework and Repair concerns
– MSL rating www.
indium .com
©

Slide #20 www.
indium .com
©

BGA/CSPs are more prone to voids due to leaded bumps on package & increased oxidation of powder
Slide #21
0.5 mm CSP
DIME www.
indium .com
©

Slide #22
Lead-free Solder paste is more prone to tombstone failures due to higher wetting forces.
T3 T3
T1 T2
T4
T5
Before Reflow After Reflow
T1 & T2 : Tack Force
T3 : Weight
T4 : Surface Tension (outside)
T5 : Surface Tension (underneath)
T4 is significantly higher using lead-free solder paste www.
indium .com
©

Slide #23
Modify Pad designs to improve tombstone defects, as well as stencil modifications
0402 Stencil Aperture Openings
.015" .011" .018"
Stencil design to minimize tombstone failures on pads with blind vias
C .022"
A*
.041"
.007"
.008"
.047"
Tombstone failures are eliminated on products with
HDI technology www.
indium .com
©

Solder wicking
Board to Board Connector
Slide #24
Lead-free wets more than leaded solder to the Gold Contact www.
indium .com
©

Solder wicking on the Gold Contact was eliminated by shifting solder paste
Slide #25 www.
indium .com
©

Slide #26
•Observed several problems with moisture sensitive components in production
– Related to moisture and higher process temperatures www.
indium .com
©

Slide #27
- Solder Paste (Pb free & Pb/Sn/Ag)
- Component Type ( 0402, 0603, 0805, BGAs, CSPs,
VCO, Transformer)
- Electrical test at every 75 cycles for 450 cycles
- Red dye analysis at 150, 300, and 450 cycles www.
indium .com
©

Slide #28
125
100
75
Crack data @ 450 cycles
50
25
0
-25
Pb-free Sn-Pb
All Pairs
Tukey-Kramer
0.05
Solder
All components passed 300 cycles of LLTs www.
indium .com
No statistically significant difference
©

Slide #29
• 10
• 6
• 5
• 9
• 8
• 7
• 4
• 3
• 2
• 1
• SnPb
• SnAgCu
• 0
• Ceramic
• Inductors
• Tantalum
• Capacitors
• Small
• Capacitors
• Ferrite Bead • Mid-size
• capacitors
No significant difference in shear force after LLTS.
www.
indium .com
©

Slide #30
6 mechanical planar drops at 5 feet, 2 hrs vibe (vert, horiz.),
24 hours Temp. shock (-40 to +80C), then repeated 5 times.
Figure 1: % FRACTURED Shield Solder Joints: Lead and No-Lead Process
25
20
15
10
5
0
Re g
Ca ps
Sw itch er
Gc ap
ODC
T
Mu lti
/AD
D
Au dio
PA
PA
NL
R
No - Lead Process eg.
..
NL
S wi t..
.
NL
G ca p
NL
OD
CT
NL
M ult
...
NL
A ud
...
NL
P
A
Shield solder joint cracking is significantly reduced using lead free solder www.
indium .com
©

Slide #31
– Sn/3.9Ag/0.6Cu (NEMI, JEITA, IDEALS)
– Component temperature will be main concern
• MSL level is adversely affected
• Rework & Repair is difficult
www.
indium .com
©

Slide #32 www.
indium .com
©

NiPdAu – Better than Matte Sn for Pb-Free Leaded
Components?
Donald C. Abbott
Texas Instruments
Attleboro, MA 02703
508-236-1569 d-abbott1@ti.com
Abstract – The requirement of lead (Pb)-free industry. Note that the Pd thickness is much lower in electronics challenges integrated circuit (IC) manufacturers in two areas: first, elimination of Pb from the IC package and second, development of the NiPdAu system. packages to withstand the rigors of reflow at 260°C.
This paper discusses the elimination of Pb from IC’s by using nickel-palladium (NiPd) based lead finishes.
NiPd based lead finishes have been in commercial use since 1989 with more than 50 billion NiPd finished devices in the field from one semiconductor house alone. The NiPd finish with 10+ years of
Palladium (0.075µ min.)
Nickel (1.0µ min.)
Palladium/Nickel Strike
Nickel Strike history was recently modified to a nickel-palladiumgold (NiPdAu) finish. The structure of the current
NiPdAu finish and the reasons for the change are discussed.
Pre-plated (PPF) leadframe finishes eliminate the need for plating after plastic encapsulation of the device. That NiPd based finishes have not become
Copper Base
Figure 1 . 4-layer NiPd
Gold (0.003µ min.) more widely used by the industry is puzzling. There are obvious advantages to any type of pre-plated leadframe. There are some unique advantages to Pd based finishes that are less obvious. The disadvantages and risks, real and perceived, associated with NiPd based finishes are also covered.
Palladium (0.02µ min.)
Nickel (0.5µ min.)
I NTRODUCTION
Structure - A Ni/Pd finish for IC leads was introduced in the late 1980s. [1,2] To date (March
2003) an estimated 50 billion Ni/Pd finished IC packages are in the field. A typical four-layer nickel palladium (NiPd) structure is shown in Figure 1. A variation on the NiPd lead finish, using nickel palladium gold (NiPdAu) was introduced in Japan in the early ‘90s. The prototypical three-layer NiPdAu finish is shown in figure 2. Upon its introduction, many Japanese IC users opted for the NiPdAu finish because of its key technical attribute of faster wetting time in solderability tests. Faster wetting times in solderability tests can be taken as a bellwether of improved wetting with the variety of Pb-free solder alloys under consideration by the electronics
Copper Base
Figure 2. 3-Layer NiPdAu
Pb-free Implications - With the interest in Pb-free processing that developed in the mid-‘90s the need for Pb-free package terminations became critical.
[3,4,5,6,7,8,9] Because NiPd and NiPdAu are Pbfree finishes, use of either on components in conjunction with a Pb-free solder alloy and OSP
PWB pad finish yields a Pb-free solder joint.
Studies have shown that NiPd and NiPdAu finishes achieve equivalent or better lead pull and
37

W I R E B O N D A B L E
H IG H Y IE L D
P R O C E S S
• S IM P L IC IT Y
• C Y C L E T IM E
C O S T
E F F E C T IV E
T O T A L C O S T O F
O W N E R S H I P
S O L D E R A B L E
L E A D F R A M E
F I N I S H
R E Q U I R E M E N T S
S H E L F L IF E
A P P L IC A B L E F O R
F IN E P IT C H
2 6 0 ° C
M O U N T C O M P A T IB L E
A L L T Y P E S
O F P A S T E
E N V IR O N M E N T A L L Y
G R E E N
A S S E M B L Y
P R O C E S S
C O M P A T IB L E
T E M P E R A T U R E
M O L D C O M P O U N D
N O P b
L e a d fr a m e F in is h A ttr ib u te s performance difference of the different lead finishes
(Sn/Pb, Ni/Pd, Ni/Pd/Au) was merely visual.
Lead Finish Attributes - The two primary functions of a lead frame finish are to provide a surface that is wire bondable and solderable after device assembly.
There are other attributes or requirements that make a finish desirable and commercially viable. These are shown in Figure 3.
Lead Frame Finish History – Lead frames started out with a gold (Au) finish, 35 years ago. Cost and Au embrittlement led to a move to full silver (Ag) finish on what was the typical base metal at that time, alloy
42 or Kovar. This system was prevalent in the mid-
‘70s. The base metals with high Ni content were used because the IC’s were relatively large and fragile – matching the coefficient of thermal expansion (Cte) of the leadframe to the silicon was important. The Ag was relatively thick, on the order of 3 – 4 microns. Ag did have the advantage of not only being wire bondable, but also preserving solderability, though at this time IC’s were dual inline packaged and the molded devices were solder dipped after assembly.
Several companies noticed a propensity for Ag, particularly Ag outside the plastic, to grow Ag dendrites that could short adjacent leads. The
Figure 3 temperature cycle results than Sn/Pb plated component leads (control). [10,11,12] Any
This observation led to major players in the IC industry mandating there be no Ag outside the plastic. The leadframe makers’ solution was spot Ag plating, where the Ag is selectively plated in the wire bond area that after assembly will be inside the plastic. About the same time as Ag spot plating took hold, there was a shift to copper (Cu) alloy base metals. The silicon had shrunk so matching Cte was not as critical. The Cu also offered lower cost and better thermal and electrical performance. The problem with Ag spot plating is that while wire bondable, it leaves a non-solderable lead surface after assembly of the IC. The thermal exposure of
Cu base metal by die attach, wire bond and molding/mold cure leaves a thoroughly oxidized Cu surface. To overcome this a major change and investment in assembly technology and equipment occurred.
Post mold solder plating was developed. This took care of the solderability issue with Ag spot plated leadframes, but moved part of the leadframe finishing operation into the assembly/test (A/T) site.
Now the leadframe maker provided the leadframe form and a wire bondable surface as an Ag spot.
But, the A/T site was responsible for plating that part of the leadframe that was outside the plastic. This put cost, cycle time, equipment, and yield issues directly in the A/T house. As will be shown below, this is not as bad as it sounds for some in the industry. dendrites grew under bias and humid conditions.
38

COST OF OWNERSHIP
ELIMINATION OF POST MOLD Sn/Pb PLATING
ELIMINATED COSTS
PLATING
PROCESS W ATER
W ASTE TREATM ENT
QUALITY CONTROL
CHEM ICAL STORAGE
9
9
9
9
9
9
9
9
SP
AC
E
9 9
EQ
UI
PM
EN
PE
RS
T
O
NN
EL
UT
IL
IT
IE
WA
S
9 9 9
TE
R
EX
H
AU
ST
IN
VE
9 9
NT
O
RY
CO
NSU
M
AB
LE
S
YIELD RELATED COSTS:
• No solder flake/burrs @ t/f
9
9
9
9
9
9
9
9
9
9 9
9
9
9
9
9
9
9
•
•
Improved lead planarity
No chemical exposure or thermal shock to assembled device
• Less mechanical damage
DISPOSAL COSTS ELIMINATED
WASTE TREAT SLUDGE
SPENT PLATING LINE BATHS
Pb CONTAMINATED TRIM SCRAP
OTHER COSTS
PERMITTING -- WASTE TREATMENT
CHEMICAL STORAGE
REDUCES CYCLE TIME/LOWERS
INVENTORY OF ASSEMBLED DEVICES
Figure 4.
For the dual in line packages, there was still the option to solder dip the oxidized Cu, using strong fluxes. But, as surface mount packages with finer pitch gull wing lead designs started to gain acceptance, the solder dipping option evaporated because of solder bridging. The need for high speed, reliable solder plating was driven by fine pitch surface mount packages. And since that time, this has been the prevalent lead frame finishing system used – spot Ag plating on Cu alloy leadframes followed by post mold solder, SnPb, plating.
The NiPd system was developed to eliminate post mold solder plating. It is a so called pre-plated lead finish (PPF), meaning the leadframe maker does all of the plating, leaving the A/T site to do die attach, wire bond, molding, trimming and forming and test.
The A/T site does not need any wet chemical processes for plating – or associated equipment, water treatment, exhaust, permitting, etc. See figure
4.
For a brief time in the mid-‘70s there were efforts to develop a pre-plated spot Ag + SnPb spot lead finish that was done at the leadframe maker. The technical catch with this system was the need for assembly temperatures that never exceeded the melting point of eutectic SnPb, ~170°C. At that time, wire bonding and die attach equipment could not function in this temperature range .
T
ECHNICAL
F
EATURES OF
N
I
P
D
A
U
Functions of the Ni, Pd and Au – The Ni is the functional layer in this finishing system. It is a barrier to Cu diffusion, it is the surface to which the wire bond is made and it is also the surface that is soldered. The Pd is a sacrificial layer that protects the Ni from severe oxidation. The Au, in the NiPdAu system, serves a similar function as the Pd and permits lower thickness of Pd.
Die Attach - For NiPd based finishes, die attach is functionally the same as for Ag spot plated finishes, namely attachment typically by Ag filled epoxy.
This attachment is made to the topmost layer of the leadframe finish – either Pd or Au in the cases at hand.
Wire Bond - The mechanism for wire bond is different than for Ag spot plated leadframes. The bond is to the Ni surface. The Au wire scrubs through the thin Pd (up to 0.25µ) and makes a NiAu intermetallic. In Ag spot plated lead frames; the wire bond is to the Ag surface, a relatively thick, soft substrate. The differences in bonding mechanics have implications for capillary design and bonding parameters. Competent bonding technicians can develop parameters for bonding NiPd and NiPdAu finishes in less that 8 hours.
Soldering - Soldering proceeds by a very different mechanism than for solder plated leads. The Pd and
Au dissolve rapidly into the molten solder. The reflow temperature required is substantially below the melting point of Pd. See Table 1 for dissolution rates. The solder joint is made to the Ni and the Pd and Au move into the bulk solder. This has been shown to be the case for SnPb and the Sn-rich Pb free solders. [11,12]. Sn or SnPb finishes are fusible coatings – they melt during reflow. The increased reflow temperatures required for Pb-free solders accelerate the dissolution of the Pd/Au layers.
Table 1 Dissolution Rates of Metals in Eutectic SnPb
Solder [After Bader]
RATE RATE
µ/s (215°C) µ/s (250°C)
Ni
Pd
Cu
Au
< 0.0005
0.0175
0.08
1.675
0.005
0.07
0.1325
4.175
Solder joints to NiPd or NiPdAu finished leads look different than solder joints to solder finished leads.
The only solder brought to the joint is in the paste.
This allows the total volume of solder is controlled absolutely by the board assembly house. When NiPd finished leads were introduced, programs for some automated solder joint inspection machines needed to be adjusted to recognize as good, solder joints with no solder on the top surface of the lead. A cross section of a typical solder joint to a NiPdAu finished lead is shown in Figure 5.
39

Cross Section of Solder Joint
Figure 5.
E
CONOMICS OF THE
N
I
P
D
A
U
F
INISH
There are potentially three areas of economic concern with any NiPd based leadframe finish that may have led to its less than enthusiastic adoption as a standard leadframe finish by the industry.
First, there is the cost of post mold plating that is part of the total cost of ownership for a silver spot plated leadframe. This is often not understood, disregarded or discounted. The initial cost of the Ag spot plated leadframe may be lower, but the cost of post mold plating must be added to the total cost of ownership.
Often, this is not done. Refer to Figure 4.
A second potential concern is the intrinsic cost and supply of Pd and Au. These are precious metals, with Au being widely traded and Pd less so. Au is mined in many countries while the major sources of
Pd are Russia and South Africa. Anxiety over reliable Pd supply has been know to occur. As with any precious metals, the key is to minimize content.
The conversion from NiPd to NiPdAu did this dramatically. The cost of Pd versus Au must take into account the density difference between the two.
For a given thickness, the mass of Pd required is about 40% less than for Au. Managing the purchasing of the metals through various precious metal market trading instruments available is prudent. Of the three concerns, the intrinsic costs, and to a lesser extent supply, of the metals are the most valid.
The last concern is related to the first. Much IC assembly is done at subcontract A/T houses that have a vested interest in maintaining a post-mold solder plating business. The two prime areas for subcontract A/T’s to add value are plating and final testing. Die attach, wire bond, and molding are materials intensive so there is little margin to add value with the highly automated equipment used for these functions. The leadframe is the most expensive part of the IC package after the silicon. With a PPF leadframe, such as NiPdAu, the plating function and associated value added is transferred to the leadframe maker. Through efficiencies in plating and minimizing precious metal content and the drop in
Pd price, the cost for NiPdAu finished leadframes is approaching that of Ag spot plated leadframes.
Adoption of any PPF, even Ag spot + Sn or SnPb spot, leaves the A/T with all the investment for solder plating and much loser revenue. It is not surprising that many subcontract A/T houses have quoted higher prices for package assembly using a
PPF than for post-mold solder plated packages. The end users, not the subcontract A/T houses, will ultimately drive the wider acceptance of NiPdAu or any PPF.
A
DVANTAGES OF
N
I
P
D
A
U
Leadframe makers see many advantages to using the
NiPdAu finish: no Pb, no free cyanide, fast cycle time, no selective plating required, low capital investment for plating equipment, better yields, and very high throughput. No Pb is key. NiPdAu offers a
“green” solution that starts with the leadframe manufacture and runs through to the consumer. The elimination of a selective plating requirement simplifies the plating process and leads to the yield and productivity improvements. Whole categories of defects are eliminated – spot Ag location for example.
For the A/T house, NiPdAu also improves yields by eliminating solder burrs and flakes at the lead trim/form operation. It also lowers the total cost of ownership by eliminating the post-mold plating operation. This frees factory space for additional die attach, wire bond and molding machines. NiPdAu effectively takes out the wet processes from the A/T house. NiPdAu is also compatible with the existing
A/T process. There is no need for lower assembly temperatures as there is for an Ag/SnPb PPF as noted above. Finally, the process options for the A/T house are simplified, no decision need be made among
SnPb, matte Sn, SnBi or any other solder plating system.
For the IC user, there are very controlled solder filets/joints because the amount of solder brought to the joint is that screened onto the board. As lead pitch is reduced, this advantage increases in importance. Lead planarity is improved, as there is less handling of the part so the risk of mechanical damage is reduced. The absolute variation of the
NiPdAu finish is less than that for either Sn or SnPb that are typically plated in the range of 2 – 10 µ versus an aggregate thickness of less than 1µ for
40

NiPdAu. This too leads to improved lead planarity.
There is no Pb in the system and unlike matte Sn finished leads; there is no risk of Sn whiskers.
Conclusions
As noted in the abstract, it is a puzzle that the
[11] D. Romm, B. Lange, D. Abbott, Evaluation of
Nickel/Palladium-Finished ICs With Lead-Free
Solder Alloys, TI SZZA024, January 2001.
[12] D. Romm, B. Lange, D. Abbott, Evaluation of
Nickel/Palladium/Gold Finished Surface-Mount
Integrated Circuits, TI SZZA026 July 2001.
NiPdAu lead frame finish has not gained wider acceptance in the semiconductor industry. When fully understood it offers cost advantages – through yield improvements, throughput, faster cycle time, lower capital investment, and streamlined factory flow in all stages of IC assembly and use. It uses no
Pb, is compatible with the leading Pb-free solder pastes under consideration, and require no modification to the device assembly process, save
Donald Abbott is a TI Fellow and Manger of New
Technology Development at Texas Instruments,
Attleboro, MA. He has more than 20 years experience in lead frame manufacturing; metal finishing research, waste minimization and recycling. He is a member of the American
Chemical Society, the American Electroplaters and Surface Finishers Society and the Minerals, elimination of post-mold soldering and no modification for the board mounting operation. The risks for using the NiPdAu finish are associated with the cost and supply of primarily Pd and to a lesser
Chemistry from Northeastern University, Boston,
MA. extent Au. The leading alternate candidate for Pbfree lead finishing is post-mold plated matte Sn. The primary risk with matte Sn is the well-known and poorly understood Sn whisker phenomenon. NiPdAu
Metals and Materials Society. Dr. Abbott received his BA degree in biology from Bowdoin College,
Brunswick, ME and the Ph.D. degree in Analytical carries no such risk. As industry consortia (NEMI,
JEITA, and ITRI) struggle with defining Sn whisker appearance and inspection protocol and also continue to investigate the mechanism driving Sn whisker growth, more companies are starting to look favorably on a NiPdAu based lead finish.
R EFERENCES
[1] D. C. Abbott, R. M. Brook, N. McLelland, and J.
S. Wiley, IEEE Trans. CHMT , 14, 567 (1991).
[2] Murata and D. C. Abbott, Technical Proceedings,
Semicon Japan, 415 (1990).
[3] M. Kurihara, M. Mori, T. Uno, T. Tani, T.
Morikawa, SEMI Packaging Seminar, Taiwan, 59
(1997).
[4] Yanada, IPC Printed Circuits Expo 1998.
[5] M. Jordan, Trans IMF, 75(4), 149, (1997).
[6] T. Kondo, K. Obata, T. Takeuch & Masaki,
Plating and Surface Finishing, 51, Feb. (1998).
[7] R. Schetty, IPC Works 99 Proceedings, Oct.
(1999).
[8] Y. Zhang, J. A. Abys, C. H. Chen, & T. Siegrist,
SUR/FIN 96 (1996).
[9] Ji-Cheng Yang; Kian-Chai Lee; Ah-Chin Tan,
Electronic Components and Technology Conference
Proceedings, 49 th , 842-847 (1999).
[10] D. W. Romm and D. C. Abbott, Lead Free
Solder Joint Evaluation, Surface Mount Technology,
March, (1998).
41

Structure of NiPd and NiPdAu
History of Lead Finishes
Leadframe Finish Attributes
Technical Features of NiPdAu
Economic Concerns
Advantages
Conclusions

[1/3 Pd content of NiPd system]

WIRE BONDABLE
HIGH YIELD
PROCESS
• SIMPLICITY
• CYCLE TIME
COST
EFFECTIVE
TOTAL COST OF
OWNERSHIP
SOLDERABLE
LEADFRAME
FINISH
REQUIREMENTS
SHELF LIFE
APPLICABLE FOR
FINE PITCH
260°C
MOUNT COMPATIBLE
ALL TYPES
OF PASTE
ENVIRONMENTALLY
GREEN
ASSEM BLY
PROCESS
COMPATIBLE
TEMPERATURE
MOLD COMPOUND
NO Pb

GOLD PLATE
COST
FULL SILVER PLATE (70s)
DENDRITIC
SILVER
SPOT SILVER PLATE WITH SOLDER DIP
POST MOLD (LATE 70s)
MID 70s
FINER PITCH
SPOT SILVER + POST
MOLD Sn/Pb PLATE
SPOT SILVER WITH
PREPLATED Sn/Pb
1988/1989
2001-2002
Pd/ Ni INTRODUCTION
NiPdAu CONVERSION

ELIMINATION OF POST MOLD Sn/Pb PLATING
-
-
-
ELIMINATED COSTS
SP
A
C
E
EQ
U
IP
M
EN
T
PE
R
SO
N
N
EL
U
TI
LI
TI
ES
W
A
TE
R
EX
H
A
U
IN
ST
V
EN
C
R
Y
TO
O
N
SU
M
A
BL
ES
PLATING
PROCESS W ATER
W ASTE TREATM ENT
QUALITY CONTROL
CHEM ICAL STORAGE
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
YIELD RELATED COSTS:
• No solder flake/burrs @ t/f
• Improved lead planarity
• No chemical exposure or thermal shock to assembled device
• Less mechanical damage
DISPOSAL COSTS ELIMINATED
WASTE TREAT SLUDGE
SPENT PLATING LINE BATHS
Pb CONTAMINATED TRIM SCRAP
OTHER COSTS
PERMITTING -- WASTE TREATMENT
CHEMICAL STORAGE
REDUCES CYCLE TIME/LOWERS
INVENTORY OF ASSEMBLED DEVICES

Wire Bonding - is done to the Ni surface. The wire goes through the thin Pd (up to 0.25µ) and makes a NiAu intermetallic. Whereas in Ag spot plated leadframes, wire bonding is done to the Ag surface, a relatively thick, soft substrate.
The differences in wire bonding mechanics have implications for capillary design and bonding parameters.

Soldering - the Au dissolves rapidly into the solder paste, as does the Pd. The solder joint is made to the Ni.
METAL RATE µ/SEC
(215C)
RATE µ/SEC
(250C)
Ni
Pd
Cu
Au
< 0.0005
0.0175
0.08
1.675
0.005
0.07
0.1325
4.175
Sn or SnPb finishes are fusible coatings, they melt during reflow.
This explains some anomalous results seen with steam aging.

• Au is a widely traded metal. Cost is ~ $300/t.oz with a density of 19.32 g/cm3.
• Pd is less widely traded. Cost is ~ $230/t.oz. with a density of 12.16 g/cm3.
• Au is widely used in industry, jewelry, dentistry, bullion.
• Pd - major user is/was automotive catalysts.
Major producers of Pd: Russia, S. Africa
• Electronics is <10% of annual usage; Semiconductors is <5% of the electronics usage.

Courtesy Kitco - http://www.kitco.com/market/

Courtesy Kitco - http://www.kitco.com/market/

'000 oz
Supply
South Africa
Russia
North America
Others
Total Supply
Pd Supply & Demand
1992 1993 1994 1995 1996 1997 1998 1999 2000 2001
1,260 1,395 1,500 1,600 1,690 1,810 1,820 1,870 1,860 2,010
2,100 2,400 3,300 4,200 5,600 4,800 5,800 5,400 5,200 4,340
450 415 410 470 455 545 660 630 635 850
70 70 70 70 95 95 120 160 105 120
3,880 4,280 5,280 6,340 7,840 7,250 8,400 8,060 7,800 7,320
Demand by Application
Autocatalyst gross 490 705 975 1,800 2,360 3,200 4,890 5,880 5,640 5,110 recovery -95 -100 -105 -110 -145 -160 -175 -195 -230 -290
Chemical * 205 190 185 210 240 240 230 240 255 235
Dental
Electrical
Jewelry
1,195 1,210 1,265 1,290 1,320 1,350 1,230 1,110 820 670
1,830 2,015 2,230 2,620 2,020 2,550 2,075 1,990 2,160 700
205 210 205 200 215 260 235 235 255 240
Other 60 35 115 110 140 140 115 110 60 65
Total Demand 3,890 4,265 4,870 6,120 6,150 7,580 8,600 9,370 8,960 6,730
Movements in stocks -10 -15 410 220 1,690 -330 -200 -1,310-1,160 590
3,880 4,280 5,280 6,340 7,840 7,250 8,400 8,060 7,800 7,320
After Johnson Matthey http://www.platinum.matthey.com/data/pd_92-02.pdf

the entire semiconductor industry converted to Pd based leadframe finishes?
• Our calculations show that the usage would be in the range of 5-10% of the Pd supply .
• And as the shift from PDIPs to SO’s to Leadless packages continues, this will offset volume increases in units.
• And the “entire semiconductor industry” won’t convert to NiPdAu or NiPd.

Excerpts from – Reuters Article
Implatsseeks to stimulate palladium demand
Wednesday March 12, 10:12 am ET
By Sue Thomas
JOHANNESBURG, March 12 (Reuters) - The world's second-largest platinum producer is anxious to boost palladium use as a platinum boom in South Africa brings with it an increase in production of its sister metal.
…less concerned about reports that forecast soaring palladium supplies from the secondary -- or recycling -- market, than he was about primary supply growth .

Standard Bank London… sees supplies of Pd from the secondary market rocketing to 3.0 million ounces by 2008, from around 300,000 ounces currently as tough European recycling laws kick in.
… there's going to be growth in palladium (from primary sources) and more supply with static demand means lower prices. Unless demand grows it's going to have an impact.
The price of palladium, used to make autocatalysts in the auto industry and components in the electronics industry, crashed to $220 an ounce by
December 2002 after reaching a peak of $1,094 in January 2001.
South African producers, the largest in the world, are expanding at full speed in a bid to meet soaring platinum demand. But the area in the Bushveld complex that they are expanding into produces a higher ratio of palladium than the area currently being mined.

Matthey forecast palladium demand to plunge 28 percent to 4.9 million ounces in 2002 -- its lowest since 1994 -- as auto companies and the electronics industry worked off their inventories of the metal.
"The platinum story looks good going forward, but palladium doesn't. If we don't expand with palladium, we'll put more pressure on platinum. We have to find ways to encourage new palladium users," Engelbrecht said.
He said the car-making industry should be enticed to use more palladium, while he had noted a small increase in interest from the electronics industry recently.

1. Subcons major value added is in post mold plating.
2. Large investment in machines, people, space, support equipment.
3. Theirs is a material intensive business - leadframes, gold wire, mold compound, die attach material.
4. Sn or Sn/Pb plating is one area where they add considerable value.
THE DILEMMA
If a subcon goes to any PPF, then the value added by plating is transferred to the leadframe maker.
The subcon then has lower revenue over which to spread the cost of the expensive equipment he has bought to do PMF.

IC USER
• CONTROLLED SOLDER FILLETS
• NO LEAD (Pb)
• IMPROVED LEAD PLANARITY
• NO Sn WHISKERS
ASSEMBLY SITE
• COMPATIBLE WITH EXISTING
PROCESS
• ELIMINATE POST MOLD Sn/Pb or Sn
• NO LEAD (Pb)
• IMPROVE TRIM/FORM OPERATION
• CYCLE TIME
• LOWER COST OF OWNERSHIP
L/F MANUFACTURER
• NO LEAD (Pb)
• NO FREE CYANIDE
• FAST CYCLE TIME
• NO TOOLING
• NO SPOT LOCATION
• BETTER YIELDS
• PRODUCTIVITY
• LOW CAPITAL
INVESTMENT

•

Stumbling Around the Loop:
Complex Goods, Recovery Strategies, and Assessment of Cycle Closure in Reverse Manufacturing
Charles David White
310 Barrows Hall, MD-3050
University of California, Berkeley
Berkeley, CA 94720 cdwhite@socrates.berkeley.edu
Abstract - The closed-cycle concept in industry ecology literature provides a normative basis for advancing recycling and reuse as means for diverting products from harmful disposal alternatives. However, although this orienting framework sounds straightforward in the abstract, bringing it into practice is much more difficult.
Drawing from research in the computer industry, this study explores the meaning of closing material cycles for technologically complex products. A primary finding is that recovery of computer equipment creates output streams that generally do not return goods to their original manufacturers for reuse. As a result, without significant information infrastructure development, even ambitious endof-life management is unlikely to produce asset flows that can be traced through the manufacturing loop. This finding raises concern about how to assess the extent of recovery and the effectiveness of policies designed to close material cycles.
Introduction
In 1984 my Apple IIe computer cost roughly
$1400, a pretty penny in those days. Long past its prime, this 128 kB (RAM) machine now retains market value only among collectors mesmerized by its green monochrome monitor or 5 ¼-inch floppy drives. Such fate has befallen computer products at an increasing rate, and many owners, shocked by the rapid depreciation, have found it hard to trash their once-costly products. As a result, a multitude of obsolete machines like my old Apple have lain idly in attics and storerooms – awaiting uncertain but inevitable end-of-life disposition. Today, the estimated population of spent computers reaches upward of two hundred million and, combined with rapid turnover in technology, grows by forty million per year [1, 2].
Oddly enough, consumer attachment and consequent reluctance to toss out obsolete machines may have helped reduce environmental damage. While dust collected on the keyboards, studies brought to light toxic hazards from conventional disposal in landfills and incinerators,
42 such as the release of lead from circuit board solder and computer screens.
Dealing safely with the mounting backlog of electronic waste and planning for future obsolescence requires more than simply overcoming emotional attachments and alarm at rapid product devaluation; end-of-life disposition alternatives require institutional changes in industrial production and consumption. In this article, I focus on the emergence of one type of change: the development of reverse manufacturing (i.e., product reuse and recycling of industrial goods) as a means for reclaiming industrial assets and diverting them from disposal.
Of note are efforts to institute reverse manufacturing already underway on both sides of the pond. The European Union has passed a
Directive on Waste Electrical and Electronic
Equipment (WEEE) that assigns responsibility to manufacturers for disposition of spent products
[3], and this regulation is spurring development of reuse and recycling infrastructure. Far less bold, the United States government has taken little concerted action; the onus of responsibility for end-of-life management remains with consumers and municipalities, and the development of reverse manufacturing has largely been left to the market.
Unfortunately, emerging infrastructure has already run into serious problems. Suspected for years, the global trade of e-waste is exporting harm from the industrialized world to less economically advantaged countries [4]. In addition to violating international accords banning shipment of hazardous wastes [5], this cross-boundary transport of industrial trash raises questions about the benefits and drawbacks of reverse manufacturing in practice. Setting aside the assumption that material loops close benignly, this paper follows products, parts, and materials through reverse manufacturing to describe loop dynamics in practice. By examining waste computer processing in the United States, this research reveals challenges inherent in reuse and

recycling for complex products like electronics.
This case further reveals why aspirations for loop closure may be insurmountable, or at least seem unrealistic, in some instances.
My intention in describing emerging infrastructure
(i.e., assets flow, amount of recovery, underlying process choices, etc.) and its challenges is to lay a foundation for asking the bigger questions: how successful is product recovery at reducing environmental impacts, and how might responsibility best be assigned to stimulate it? My argument is that we can only answer these questions by first understanding the processes of product recovery and the consequent shape of the manufacturing loop(s) they create. The purpose of this paper is not to argue for the superiority of one form of loop closure over another; rather the point is to examine the process of reverse manufacturing for complex goods to describe the unfolding infrastructure and to uncover hidden assumptions that affects its form and function. In this vein, this paper does three things: (i) describes closed-loop models and underlying assumptions, (ii) points out challenges for the reverse manufacturing of complex goods like computers, and (iii) suggests next steps that can help improve the development of product recovery for electronics.
Product Recovery in Theory
Product recovery as an environmental imperative has roots in industrial ecology’s application of ecological metaphor to industrial planning and practice [6, 7]. Akin to life-cycle analysis of design and materials flow accounting, product recovery is a tool to help manage the biogeochemistry of industrial systems (i.e., the ecologically acceptable cycling of substances between and through industrial sub sectors). In concept bridging the grave of one product with the cradle of another, product recovery not only proposes to mitigate anthropogenic impacts by diverting goods from harmful disposal; it also reclaims industrial assets – whole products, component parts, and constituent materials – for another generation of use. That is, through reuse and recycling
1 product recovery recirculates assets
back into the industrial economy when they would otherwise dissipate out.
The normative principles of industrial ecology are powerful lenses for considering industrial transformation. However, theoretical critiques of
1 Here “reuse” describes redeployment of industrial assets geared toward preserving their design content. In contrast,
“recycling” is the reprocessing of assets to facilitate another generation of use without concerted attention to design preservation.
43 the core ideas justify closer study of the largely untested derivative concepts. In the case of product recovery, the shortcomings of the closedloop ideas are particularly important. Scholars have noted that descriptions of industrial ecology largely ignore the thermodynamic tension between the entropy in systems and the energetic demands of maintaining order [8, 9]. This limitation translates into trade-off choices between tracking down and redirecting errant molecules and the energy requirements of doing so. Accordingly, the degree of loop closure depends on the dynamism of closure efforts. Implementation studies provide an additional justification for empirical scrutiny of product recovery [10]: simply put, ideas change between the drawing board and the workbench.
Product Recovery in Action
Much of the literature on reverse manufacturing focuses on the technology and logistics for relatively simple products, like beverage containers, toner cartridges, and tires. As this section explains, the end-of-life processing for such goods provides limited insight into the challenges involved in the reuse and recycling of complex products like photocopying machines, automobiles, and computers. Based on field interviews conducted with eight firms involved in the reverse manufacturing of computers, here I offer an empirically based model of recovery as a series of stages.
Commonly labeled simply “recycling” 2 , product
recovery is a complement to production – analogous to the way that reverses manufacturing mirrors “forward” manufacturing. In this sense, the term “recovery” describes a broad set of activities that reclaim value by refurbishing whole products, salvaging components, or simply reprocessing constituent materials. In this paper I focus on the industrial processes involved in product recovery and, for this reason, use the term interchangeably with “reverse manufacturing.” It is important to note, though, that reuse and recycling of product assets need not proceed through industrial processing and, therefore, recovery and reverse manufacturing do not truly have a one-to-one mapping.
In addition to this caveat, I must note a simplification in my model. In practice the blend of products recovered can vary across companies, and computers are not atypical in this regard. One should expect to see a variety of machine types, peripheral equipment, and even other electronics
2
For a more complete treatment of terminology in product recovery, White et al. (2003) provide a glossary of terms [11]

in a computer recovery stream. However, for illustrative purposes I am reducing my discussion simply to the reverse manufacturing of desktop computers. Overcoming this prevarication requires tackling lots of the messy aspects of recovery, since the handling of myriad products converges and diverges at various points in the process, as well as from firm to firm.
Acknowledging that the lynchpin of predictive power is effective simplification, I modestly leave the complicating effects of product diversity to future discussions.
A Model of Computer Recovery: desktop computers
Generically speaking, three types of recoverable assets form the triumvirate of inputs to reverse manufacturing: the whole product, its individual designed parts, and the materials comprising both.
For a desktop computer three subassemblies, which contain the various recoverable parts, constitute the whole: the central processing unit
(CPU), the monitor, and other input/output peripherals such as keyboards, mice, or speakers.
The CPU contains hard and disk drives, circuit boards with integrated circuit chips, impactresistant plastic casings, an internal metal framework, small batteries, motors, power supplies, and cables. The cathode ray tube-based monitor is housed in an impact-resistant plastic shell and composed of a leaded glass funnel and coated glass front panel, a cathode, and circuitry.
The peripheral equipment is composed of minor circuitry and some movable parts, most of which are not readily salable in their own right. The materials recoverable from this medley of components include ferrous metals, aluminum, copper, precious metals, polystyrene and ABS plastics, thermosetting epoxy plastics, leaded and unleaded glass, and hazardous wastes like batteries
(see Figure 1). Because products are often transported with appreciable amounts of paper, cardboard, and polystyrene foams, these materials also require disposition and present parallel material management challenges [12].
During reverse manufacturing firms must make a variety of choices about assets to recover vis-à-vis logistics, technology, and economics. Some choices limit other options; certain assets are destroyed or “cannibalized” in the process of practically or efficiently extracting others. As a result, the outputs of reverse manufacturing are not simply a reverse bill of materials for the original product but rather an admixture of refurbishable whole products, reusable parts, recyclable materials, and waste.
44
Figure 1. A Desktop Computer and Its Assets
Monitor
CPU
Peripherals
Cathode Ray Tube
Equipment Casings
(external structure)
Structural Components
(internal structure)
Hard Drive, Disk Drives
Circuit Boards
(motherboard and adapter cards)
IC Chips
(microprocessor and memory)
Hazardous Components
(batteries)
Electrical Devices
(cables, motors, power supply, etc.)
Input Devices
(keyboard, mouse, speakers) leaded and unleaded glass, copper-containing circuitry high impact polystrene and ABS plastics ferrous and nonferrous (Al) metals, polystyrene plastic ferrous metals precious and nonferrous (Cu) metals, thermosetting plastics precious and nonferrous (Cu) metals, silicon-based chips hazardous waste nonferrous (Cu) metals, plastic coatings nonferrous (Cu) metals, polystyrene plastics
Product Subassembly Parts Materials
As printed in White et al. (2003)
Drawing on the closed-loop concept, in Figure 2 mimics contemporary models by depicting reverse manufacturing as a multi-stage process analog to forward manufacturing [13, 14, 15]. In this stylized view production and recovery form two halves of a circular industrial process. The left side of this figure demonstrates four generic stages in forward manufacturing: material manufacturing, component manufacturing, product assembly, and distribution and sale. On the right are four activities that constitute reverse manufacturing: acquisition, assessment, disassembly, and reprocessing. The middle of the diagram also contains two ancillary stages: testing and repair (often integrated with warranty and service programs) and redistribution and resale
(which may combine with product distribution and
sale).
3 This discussion focuses on the four
recovery stages on the right.
During the first stage, called acquisition, reverse manufacturers make decisions about which products to recover and how to gather them. That is, acquisition involves selecting product types, locating them, collecting them, and transporting them to the recovery facility. Although the field of reverse logistics offers approaches for controlling the timing, quantity, composition, and quality of products entering the recovery process
[16], computer recovery companies have been able to make little use of it. Because of the spatial dispersion of computers among individual owners and variation in product technologies, firms find it very difficult to reduce recovery input flow
3 In addition, there are two sets of notable activities not depicted as stages in this diagram: product design and contracting strategy, which influence the entire process but exceed the complexity of this basic input-output model.

Figure 2. Supply Loop inv en tor y y
Redistribution and Resale
Acquisition
- collection
- transport inve ntor
- valuation
- reusability inve nto ry
Component inv en
- disaggregation
- dismantling
- demolition in ve nt or y y ry nto ve in ry nto ve in
- recycling
- energy recov.
Adapted from White et al. (2003). Forward manufacturing on the left; reverse manufacturing on the right. uncertainty (quantity and timing) and equipment contingencies (composition and value). Some recovery firms have reduced variability in inputs is by developing contracts with large end-users, such as financial institutions or original equipment manufacturers (OEMs), who produce a steadier and more consistent stream of products from inhouse use. Most of the time, acquisition utterly lacks such sophistication. Instead, firms face a continuing tension between accepting diverse product types, which requires more manufacturing flexibility, and coping with intermittent inputs streams, whose oscillation disrupts the efficient use of labor and shop space. With little ability to control the flow of process inputs, businesses collect what they can find and ship the equipment to their processing facilities
Regardless of its trajectory of arrival, once equipment has arrived at the facility, managers must determine which assets are valuable and, given that the wholes, parts, and materials are mutually exclusive outputs, the optimal mixture to recover. This determination is assessment, the second stage of reverse manufacturing. By assessing products, recoverers determine whether to repair (aka refurbish, renovate, recondition, retrofit, etc.) or just plain resell a product or its parts. For the most part, this process occurs when products first arrive and are sorted and, ideally, would be closely linked with product acquisition. However, the cross-section of computer models in any given year, notwithstanding the longitudinal variability given the backlog in attics and basements, makes it very hard to conduct computer assessment upfront and in one shot. For one thing, reverse manufacturers generally lack product schematics for the computers they receive. For another, many products are modified during their lifetime, and information about this tinkering rarely accompanies waste electronics. As a result, assessing computer products is like playing with Russian dolls. As layers of the product are uncovered, firms discover the constituent parts and materials that are available for recovery. The lack of comprehensive, introductory assessment constitutes one of the greatest risks in recovery. The need for repeat assessment or assessment-on-the-fly during disassembly, a direct result of incomplete information about inputs as well as market immaturity for outputs, takes extra time and opens the process up to errors and oversight.
The stage that follows assessment in the model loop is disassembly, the actual deconstruction of a computer to extract its assets for reuse, recycling, or disposal. For computers, disassembly begins by disentangling reusable subassemblies and by opening the product to locate reusable parts. Once the casings have been penetrated, valuable reusable parts are removed. The remainder of the products
45

is manually or mechanically sorted for material recovery.
There are two primary axes for variation in this process.
The first is the organization of the disassembly activities as continuous or batch processes. Only large reverse manufacturing operations, whose contracts help them to source consistent inputs streams, have been able to establish semi-continuous operating lines along which disassemblers unscrew, pry open, or break apart computer products. Most of the time, disassembly is a batch operation that takes place at a semi-circular workbench, where a lone disassembler sorts assets into bins.
The other axis concerns the mode of disassembly. Here there are three approaches. The first of is disaggregation , the conscientious unfastening of products. This cautious approach protects each part for subsequent reuse or resale and is generally used for large, expensive products like mainframe computers that are repaired for reuse. Most desktop computers are disassembled by dismantling them – disassembling them in a way that destroys the design of some parts to reclaim others. That is, low- or no-value parts, such as plastics panels, are irreparably broken to expedite the extraction of electronic components. The third mode continues the trend by demolishing the whole product and its parts to recover only materials. When the whole and the parts possess no recoverable value and, therefore, retention of their designs is not useful, some firms have found large shredding equipment and mechanical separation to be cost-efficient for achieving these end. In a nutshell, each of the three styles of disassembly represents a different degree of design value retention for recovery and affects the distribution and value of product assets. No matter what combination of approaches is used, the disassembly of products and segregation of their assets greatly multiplies the number of recovery streams: reusable subassemblies and parts go to repair shops, manufacturers, or retail outlets; scrap parts are separated and purified for material recovery; and any residuals are hauled away for reprocessing or waste disposal.
The disassembly outputs become the next stage inputs and, given the limits of discussion here, I focus just on material reprocessing as a final stage in reverse manufacturing. The goal of material reprocessing is to reclaim these materials in salable quantities, and businesses naturally try to maximize profits by optimizing the benefits of purifying materials against the cost of doing so. Initial separation occurs by hand during disassembly, either on a disassembly line or at a workbench, to sort assets into basic categories according to material content. For example, because of its large copper coil, a motor is typically thrown into a bin with cables and wires for copper recovery; bins containing larger internal components (e.g., disk drives, structural pieces) are usually sent to steel mills. As illustrated in Figure 1, computer parts produce seven rough categories of materials:
ferrous metals, aluminum, copper, precious metals (e.g., gold, palladium, and silver), glass, plastics, and hazardous
wastes.
4 The process of recycling is different for each
substance, and the various material streams are shipped to facilities specializing in their reclamation. Depending on the location of the disassembly facility, transportation thousands of miles away to Canada, the UK, or China may be necessary. Some materials, like metals, are easier and more economical to recover than others, like engineering plastics, whose weak recycling infrastructure offers little opportunity for profitability. However, the biggest challenges in material reprocessing are locating markets for materials and determining an efficient target purity for reprocessing. Like disassembly, the reprocessing choices and material output opportunities are highly dependent on choices made upstream. For example, a decision during disassembly to sort motors and cables into the same bin comingles different metals and plastics in ways that may be effectively irreversible downstream.
The primary point of this discussion is to introduce the staged model of computer recovery and, in so doing, introduce the variety of product and material streams
produced by their reverse manufacture.
5 Such description
of computer recovery lays a foundation for discussing how the structural characteristics of computer products, such as their ownership patterns and variable designs, differ from other products and reveal the potential for different types of supply loops.
Form and Function: Remanufacturing & Demanufacturing
Models of product recovery frequently portray reverse manufacturing as an analog to forward manufacturing, as I have done in Figure 2. Given the dynamics of computer recovery, the assumption that reverse manufacturing outputs become inputs to forward manufacturing is at best tenuous and at worst flat wrong. Infrastructure is developing in response to EU mandates and despite little US policy, but markets for older technology cast-offs and scrap materials infrequently channel assets back into another life-cycle of computer manufacturing. More realistically, the assets are only reintegrated through the occasional resale of whole products, through some OEM warranty programs, or through the circumstantial recirculation into the computer supply chain by waste scavengers. Generally speaking, computer asset return to use occurs wherever practical.
This finding conflicts with the general notion that recovery retains, or at least strives to retain, the design content of a product. Here I introduce two terms, “remanufacturing” and “demanufacturing,” to describe a bifurcation in product recovery strategy vis-à-vis reuse. The term remanufacturing describes the tightly linked refurbishment or reconditioning of products and assets to reclaims them for original
46
4 Transport packaging is an additional category of materials that must also be managed during computer end-of-life disposition, but I have omitted it from this discussion.
5 Not to beat a dead horse, but a longer treatment of these topics can be found in White et al 2003.

applications. In contrast, demanufacturing operates mostly to divert wastes from disposal and loosely recirculates assets. Demanufacturing, the process used to recover computer products, decreases the likelihood of reusing recovered assets to repair existing products or of tracing assets back into next generation manufacturing. Instead, assets are co-modified and dispersed into other manufacturing enterprises (e.g., computer circuits are reused in toys; plastic casing are recycled into telephony).
This difference produces supply loops with vastly different capacities to measure and communicate the extent of recovery and loop closure (see Figure 3).
Figure 3. Manufacturing Strategy and Loop Tension
That computers are complex technological products may partially explain the dominance of demanufacturing as a recycling strategy, but it is unlikely to tell the whole story.
Currently inexplicable, remanufacturing infrastructure has sprung up for some technologically simple products, such as glass milk bottles, while for others, such as glass soda bottles, recycling remains the dominant recovery strategy.
Similarly, more robust remanufacturing exists for some complex products, such as copy machines, than for computers, which are technologically similar. (See Table
1.) Ownership patterns, rates of technological obsolescence, product design, and degrees of supply chain disintegration are other variables likely influencing tension in the supply loop. Explaining how such factors affect strategy is an area ripe for research.
Table 1. Products and Recovery Strategies
Remanufacturin g
Simple
( refurbishment ) milk
Complex bottles copy machines
Demanufacturing
( reuse, recycling soda bottles
) computers
CONCLUSIONS
Industrial ecology literature theorizes that closing manufacturing loops mitigates environmental harm by diverting industrial products from disposal through techniques like product recovery. Operationalizing these
47 theories suggests that the degree of mitigation is dependent on the degree of loop closure that product recovery achieves. To evaluate the extent of loop closure, or the effectiveness of product recovery policies, knowledge of the processes that modify the trajectory of waste products is needed. I argue that existing models of reverse manufacturing often assume that the recovery of complex goods is remanufacturing-dominated. In contrast, empirical research on the reverse manufacturing of computers reveals that recovery is demanufacturing-dominated and disperses product assets into the industrial economy rather than retains them within a product supply loop.
The distinction between demanufacturing and remanufacturing as recovery strategies deepens insights into the form and function of supply loops and the benefits and drawbacks of product recovery for complex goods. For example, on the down side demanufacturing can downgrade assets (e.g., high-value computer plastics become low-value plastic lumber) and make it economically more attractive to export electronic trash. On the up side demanufacturing may stimulate broader and more flexible recovery infrastructure across industrial sectors. For this reason, I remain unconvinced that demanufacturing is an uglier, dumber cousin to remanufacturing and not just an alternative industrial organization for product recovery.
For both demanufacturing or remanufacturing, developments on the horizon may help address continuing challenges: research on process planning may help manage recovery streams more efficiently; design for environment may help improve ease of disassembly; and mandatory recovery might stabilize input flows to reverse manufacturing. However, information poverty is a littlestudied but significant setback warranting concerted attention. Data shortfalls afflict product recovery at every stage but are most noticeable and consequential during assessment. In the face of rapid technology and product architecture changes, recoverers without product information cannot cost-efficiently assess products and, ultimately, recovery options. When information is unaffordable or unobtainable in assessment, downstream recovery wastes still-valuable design content. Disclosing or standardizing product compositions and making market data more available and affordable would help overcome data deficiencies. Similarly, more research on recovery trajectories could improve models of the reverse manufacturing process. Knowledge about the form and function of supply loops is needed to evaluate the extent of loop closure and the effectiveness of disposal bans, stewardship mandates, and recovery incentives at reducing environmental impacts.
These research findings have relevance for both manufacturing loop concepts in industrial ecology as well as practical lessons for policies designed to close them. The theoretically relevant concept is the notion that recovery strategy (i.e., the tendency toward remanufacturing or

demanufacturing) can affect tension in the supply loop.
Understanding how product types, rates of technological change, and industrial organization affect loop tension is certainly an area for further research. For policy, the effects of recovery strategy on supply loop tension have serious implications for loop-closing programs, such as product stewardship mandates. Without appropriate information infrastructure, increased flows of spatially and temporally heterogeneous products is not only likely to overwhelm the recovery process, it also makes it extremely difficult to track asset reuse and recycling to evaluate policy effects.
These consequences provide one more reason to reflect on product recovery models and their limitations.
R EFERENCES
[1] Matthews HS, et al. Disposition of End-of-Life Options for Personal Computers, Carnegie-Mellon University, Green
Design Initiative Technical Report Number 97-10, 1997.
[2] EPR2 Baseline Report: Recycling of Selected Electronic
Products in the United States . National Safety Council, 1999.
[3] Commission of the European Communities, 2000.
Proposal for a Directive on Waste Electrical and Electronic
Equipment.
[4] Silicon Valley Toxics Coalition, 2002. “Exporting Harm:
The High-Tech Trashing of Asia.” (www.svtc.org)
[5] Basel Action Network, 2002. “Comments by the Basel
Action Network to the Environmental Protection Agency on the Proposed Rule on Cathode Ray Tubes and Mercury-
Containing Equipment.” August 12, 2002. (www.ban.org).
Information about the Basel Convention and ban on hazardous wastes is available at www.basel.int.
[6] Frosch, Robert and Nicholas Gallopoulos, 1989.
“Strategies for Manufacturing,” Sci. American , 260: 144-152.
[7] Jelinski, L., T. Graedel, R. Laudise, D. McCall, and C.
Patel, 1992. “Industrial Ecology: concepts and approaches,”
Proc. Natl Acad of Sci , vol 89: 793-797.
[8] O’Rourke D, Connelly L, and Koshland C, “Industrial
Ecology: a critical review.” Int. J. Evnmt. and Pollution
(1996), vol 6 nos. 2/3: 89-112.
[9] Connelly L, Koshland C, “Two aspects of consumption: using an exergy-based measure of degradation to advance the theory and implementation of industrial ecology.” Res, Cons, and Recycling (1997) 19: 199-217.
[10] Mazmanian D, Sabatier P, Implementation and Public
Policy , Scott, Foresman, 1983.
[11] White CD, Masanet E, C.M.Rosen, and S.L. Beckman,
2003. “Product Recovery with Some Byte: an overview of management challenges and environmental consequences in
48 reverse manufacturing in the computer industry,” Journal of
Cleaner Production 11: 445-458.
[12] Das SK, Matthew S. Characterization Of Material
Outputs From An Electronics Demanufacturing Facility.
Proc. IEEE Intl Sym. on Electr. & the Envmt 1999: 251-256.
[13] Ferrer G, Guide VDR. Remanufacturing Cases and State of the Art. Handbook of Industrial Ecology , Robert U. Ayres
& Leslie Ayres eds., 2002, Elgar Academic Publishers.
[14] Ayres R, Ferrer G, Van Leynseele T. Eco-Efficiency,
Asset Recovery, and Remanufacturing. European
Management Journal , Volume 15, No. 5: 557-574.
[15] Thierry M, Salomon M, van Nunen J, van Wassenhove
L. Strategic Issues in Product Recovery Management. Cal
Mgmt Review , 1995, Volume 37, No. 2: 114-135.
[16] Fleischmann M, Bloemhof-Ruwaard, JM, Dekker R, van der Laan E, van Nunen J, van Wassenhove L. Quantitative models for reverse logistics: a review. Eur. J of Operational
Research 1997, 103: 1-17.
Chad White is a doctoral student in the Energy and
Resources Group at UC Berkeley, here he researches environmental protection policy through the lens of organizations and environmental performance. He is a
CITRIS fellow at the University of California and a research fellow at the Belfer Center for Science and
International Affairs at the Kennedy School of Government at Harvard. He wishes to thank Professor Christine Rosen and Dr. Sara Beckman, both of the Haas School of Business at UC Berkeley, for their guidance with this research.

E nergy and
R esources
G roup
Chad White
University of California · Berkeley
C onsortium on
G reen
D esign and
M anufacturing

E-waste as a
• mounting problem
• variable backlog
2

cradle
(input)
?
• Problem : industrial asset dissipation grave
(waste)
• Proposed Solution : product recovery
• Empirical Research : opportunities, challenges, and limitations of the concept
3

4

The Monitor (CRT) panel glass with phosphorescent coating leaded funnel glass
Peripherals
Mouse, Keyboard
(plastic + circuitry) cathode yoke
(copper-containing circuitry) plastic casing
(impact-resistant plastic)
The CPU power supply fan hard drive disk drives microprocessor
(semiconductor chip) memory
(IC chips)
IC chips motherboard ports expansion slots adaptor card
5

Product
Monitor
CPU
Peripherals
Subassembly
Cathode Ray Tube
Equipment Casings
(external structure)
Structural Components
(internal structure)
Hard Drive, Disk Drives
Circuit Boards
(motherboard and adapter cards)
IC Chips
(microprocessor and memory)
Hazardous Components
(batteries)
Electrical Devices
(cables, motors, power supply, etc.)
Input Devices
(keyboard, mouse, speakers)
Parts leaded and unleaded glass, copper-containing circuitry high impact polystrene and
ABS plastics ferrous and nonferrous (Al) metals, polystyrene plastic ferrous metals precious and nonferrous (Cu) metals, thermosetting plastics precious and nonferrous (Cu) metals, silicon-based chips hazardous waste nonferrous (Cu) metals, plastic coatings nonferrous (Cu) metals, polystyrene plastics
Materials 6

•
•
•
•
7

waste assembled products mfd.
parts waste waste end-of-life products salable products salable products
Redistribution
& Resale repaired products
Service reusable products
Repair &
Testing warranteed products repairable products mfd.
parts broken parts reusable parts products with valuable parts reusable parts waste collected products waste products with valuable materials reusable materials waste mfd.
materials extracted materials waste cradle
(input) recyclable materials waste parts with valuable materials waste parts to other uses materials to other uses
8

(spatial and temporal dispersion) various models
- product takeback by OEM
- reverse distribution & sale
- municipal collection
- common carrier implications
- who owns it affects what is made available for reuse waste end-of-life products
Acquisition collected products waste
9

Service
Repair &
Testing repairable products valuation and selection
• pre- or post-collection
• fate of rejects implications
• information demands
• selection = efficiency?
collected products
products with valuable parts products with valuable materials waste
10

three styles
• disaggregation
• dismantling
• demolition implications
• disassembly style affects downstream recovery purity and economics products with valuable parts products with valuable materials parts to mfg parts for repair
parts with valuable materials waste parts to other uses
11

cannibalization
• lower value materials sacrificed to recover others implications
• degree of reprocessing dependent on market values parts with valuable materials
materials to computer mfg waste materials to other industrial uses
12

waste
Use
(spatial and temporal dispersion) waste end-of-life products waste assembled products
Distribution
& Sale salable products salable products
Redistribution
& Resale repaired products
Service reusable products
Repair &
Testing warranteed products
Acquisition repairable products
Assembly broken parts waste collected products
Assessment mfd.
parts mfd.
parts reusable parts reusable parts products with valuable parts waste products with valuable materials
Component
Mfg
Disassembly reusable materials parts to other uses waste waste mfd.
materials
Material
Mfg
Reprocessing parts with valuable materials extracted materials waste recyclable materials waste materials to other uses
13

implications
- how products travel through recovery processes
- the types of assets recovered and markets considered
- general shape of the “manufacturing loop”
14

Desktop Computers
• environmental concerns about e-waste
(heavy metals, toxic coatings, etc.)
• exploding population of spent computers
• useful bits – not necessarily new computers
• different from other complex products
- spatial and temporal dispersion
- variables rates/styles of ownership
- nominal lay culture
• looser recovery infrastructure
15

Remanufacturing-dominated Demanufacturing-dominated
Use
Use
Product Sector
Industrial Economy
Assumption : product sector recidivism
Product Sector
Industrial Economy
Assumption : dispersion within economy
16

17

Remanufacturing-dominated Demanufacturing-dominated
Use
Use
Product Sector
Industrial Economy
Closure: measured in terms of new assets in products
Product Sector
Industrial Economy
Closure : measurement needs data on flows, economy
18

19

Acquisition & Assessment
- collection and network design (OR)
- sourcing and selection (reverse logistics/economics)
Disassembly & Material Reprocessing (engineering)
- organization and efficiency (feedback for design)
Recovery Strategy & Trajectory
- technological complexity, rate of obsolescence, ownership patterns, supply chain structure, lay culture, compositions
Information Availability
data on products and markets, costs and drivers
- traceability of assets along loop: where do they go?
20

- Professor Christine Rosen (Haas School of Business)
- Dr. Sara Beckman (Haas School of Business)
Osborne Vixen
Compaq 8032-SK
- Daniel Guide (Duquesne University) & Closed-Loop
Supply Chain Conference participants
21

Cynthia Folsom Murphy
University of Texas at Austin
(512) 475-6259 cfmurphy@mail.utexas.edu
UCI – April 3, 2003

RIP
IBM 360
1965 - 1985
RIP
IBM 360
1965 - 1985
RIP
IBM 360
1965 - 1985
RIP
IBM 360
1965 - 1985
A cash cow? Or an economic burden?

z z
Typical system being retired had the following characteristics
–
10 years old
–
–
–
–
–
Large units (50 lbs or more), large pieces
Steel, unpainted, mechanical attachments
Gold or aluminum wire bonds, gold backed chips, high base and precious metal content on boards
CRTs a small portion by weight and quantity
Peripherals not common
Market for new electronics
–
–
Unsaturated in US, virtually non-existent in developing countries
High cost per function

z z
Typical system being retired had the following characteristics
–
5 years old
–
–
–
–
–
30-50 lb units, moderately sized pieces
50% steel, some painted, mixture of mechanical attachments and adhesives
Wire-bonded (Al, some Au) and surface mount (Sn/Pb) chips, moderate base and precious metal content on boards
CRTs approaching half by weight and quantity
Peripherals somewhat common
Market for new electronics
–
Partially saturated in US, unsaturated in developing countries
–
Moderate cost per function

z z
Typical system being retired had the following characteristics
–
2-3 years old
–
–
–
–
–
10-30 lb units, numerous small pieces
10% steel, many painted, significant use of permanent attachments and adhesives
Surface mount chips, moderate base and precious metal content on boards
CRTs approaching half by weight and quantity
Peripherals somewhat common
Market for new electronics
–
Highly saturated in US, developing countries prefer new
–
Low cost per function

3
2
1
5
4
0
19
92
19
94
19
96
19
98
20
00
20
02
20
04
20
06
20
08
Years

z z
–
–
Steady state
By 2010
–
–
Older units coming out of storage
Estimate peak between 2005 and 2008

z
Revenues dominated by resale of used and remanufactured systems and components
–
–
PROs z z
Workforce training
Appears more environmentally and socially friendly
CONs
– z z z
High labor costs (remanufacturing and sales)
Plummeting prices
High space and inventory management costs
Examples: z Late 1990’s HP-Noranda, AT&T McDonald Butler, IBM, Dell-RCI

z z z
Charge disposition fee ($.20-$.30/lb)
Sell anything that takes does not require repair
Shred remainder with minimal attempt at optimal material recovery, or only materials with extremely high value (e.g., precious metals)
–
–
PROs z z
Cost structure relatively stable
Material flow management simple
CONs z z
Lowest level of resource conservation
Probably will require legislative and/or financial support

Direct Labor Cost Distributions – 26 million pounds per year
10%
1%
2%
26%
13% 12%
36%
Unload, Triage,
Handle Packaging - 36%
CPU Plug-In Test and
Remanufacture - 12%
Monitor Demanufacture - 13%
CPU Demanufacture - 26%
Misc. Equipment
Disassembly - 2%
Laptop Disassembly - 1%
Shredder Operation - 10%

23%
2%
Disposition Charges – 75%
Unit Sales – 23%
Subcomponent and
Commodity Sales – 2%
75%

z z z z

Material
Plastic (large pieces)
Plastic (small pieces)
Leaded and unleaded glass (CRTs)
Leaded and unleaded glass (small pieces)
Copper (yokes)
Ferrous metal (large pieces)
Ferrous metal (small pieces)
Non-ferrous metal (high-grade)
Non-ferrous metal (low-grade)
Cu and Al (wire and cable)
Cu, Ag, Pt, Pb, Ni, Sn, Au (Connectors and PWBs)
Cu, Ag, Pt, Pb, Ni, Sn, Au (high-grade PWBs)
Components
Hazardous Materials
Clean cardboard
Pallets
Steel baling wire
Paper and low-grade plastic
$/lb (excl. trans) Disposition
$0.10
Plastics recovery
$0.00
Smelter
$0.00
Smelter
$0.00
Smelter
$0.10
Metal recovery
$0.05
Metal recovery
$0.04
Metal recovery
$0.35
Metal recovery
$0.15
Metal recovery
$0.05
Metal recovery
$0.15
Precious metal refiner
$0.80
Precious metal refiner
$0.15
Broker
-$1.00
HazMat
$0.05
Cardboard recycler
$0.10
Wood recycler
$0.05
Metal recovery Fe
-$0.01
Landfill transportation distance
500
500
500
1500
1500
1000
50
50
1000
2000
2000
2000
500
500
500
50
500
50 lbs/ truck
25,000
30,000
30,000
30,000
30,000
30,000
30,000
30,000
30,000
30,000
30,000
30,000
30,000
10,000
10,000
10,000
30,000
10,000

TVs
Monitors
CPUs
Peripherals
Laptops
Total
INCOMING EQUIPMENT lbs/unit units/year lbs/year
60 150,000 9,000,000
30 150,000 4,500,000
23 150,000 3,450,000
8 150,000 1,200,000
8 150,000 1,200,000
19,350,000

z z z z
Disposition of computers and peripherals will occur as
“systems” –
–
1 CPU +
–
–
1 monitor +
1 peripheral (printer, scanner, hard drive)
Move to FPDs will greatly increase disposition of TVs
As laptops become more common, they will be disposed of at same rate as other systems
Operating in 2003 with roughly 20M lbs per year of waste electronics

TVs
Monitors
CPUs
Peripherals
Laptops
Total
INCOMING EQUIPMENT lbs/unit units/year lbs/year
60 100,800 6,048,000
30 408,000 12,240,000
23 288,000 6,624,000
8 144,000 1,152,000
8 24,000 192,000
26,256,000

z z z z z
Disposition of monitors will be greater than CPUs as move to FPDs
Disposition of peripherals will be less than systems because of longer product life
Disposition of laptops is still low due to cascading
Move to flat screen TVs is slow
Operating in 2003 with roughly 26M lbs per year of waste electronics

400,000
300,000
200,000
100,000
0
-100,000
-200,000
-300,000
-400,000
-500,000
Operating Costs plus Material Disposition
Costs/Revenues
Basecase
CEDEIP
Equipment Type

Base/Precious metals
Steel
Aluminum
Glass Thermoplastic

z z z

600,000
400,000
200,000
0
-200,000
-400,000
-600,000
-800,000
Operating Costs plus Material Disposition
Costs/Revenues
Basecase
All Shred
Equipment Type

Operating Costs
0
-500,000
-1,000,000
-1,500,000
-2,000,000
Basecase
All Shred
Equipment Type

z z z
In all cases shown above, disposition fees would still be required
–
$0.15/lb to break even
Operating costs for 100% shred suggests that as the number of CRTs begins to decrease this would be a good option
Material costs/revenues associated with peripherals and laptops also suggests that improved material flow values for non-CRT electronics would be significant

TVs
Monitors
CPUs
Misc Equip
Laptops
Total
INCOMING EQUIPMENT lbs/unit units/year lbs/year
60 10,000 600,000
30 10,000 300,000
23 400,000 9,200,000
8 700,000 5,600,000
8 500,000 4,000,000
19,700,000

Operating Costs plus Material Disposition
Costs/Revenues
200,000
0
-200,000
-400,000
-600,000
-800,000
-1,000,000
-1,200,000
Basecase
All Shred
Equipment Type

z z
US DoE funded project in Parkersburg, WV
US EDA funded project in Austin, TX
–
–
–
Recycle waste electronics in a vertically integrated setting
(Eco-Industrial Park)
Eliminate transportation costs between electronics recycling center and other material recovery systems
Use advantage of co-located metals and plastics recovery operations to improve quality of material streams and provide incentives for improved technologies

z z z z
Eliminate transportation costs except to hazardous waste facility, smelter, and precious metal refiner
Eliminate transportation costs associated with consolidation of end-of-equipment
Austin project is in conjunction with Texas Disposal
Systems which already has on site
–
–
Landfill
Metals recycling
–
Composting
Assumes addition of
–
Electronic thermoplastics recycling
–
Commodity recycling

40,000
20,000
0
-20,000
-40,000
-60,000
Operating Costs plus Material Disposition
Costs/Revenues
Equipment Type
No EIP
EIP

Operating Costs plus Material Disposition
Costs/Revenues
100,000
80,000
60,000
40,000
20,000
0
-20,000
-40,000
-60,000
Equipment Type
Low Matl $
High Matl $

900 acres total
11 acres available
100 acres potential
Composting and metals reclamation center
Methane production planned

Texas Longhorn…….