Tom Pye
BSEE, MSEE University of Illinois


Semiconductor physics/fabrication + RF/microwave majors
>30 years in high tech


Product engineer/manager


Engineering/R&D manager





1
RF/microwave defense electronics
Semiconductor equipment
Strategic planning


RF/microwave defense electronics
Marketing/New Products manager


RF/microwave defense electronics
Semiconductor equipment
Display equipment
Design software
Buy side analyst – Greentech
5/29/2016
Greentech
Dominated by economics
Desperately needing engineering
2
5/29/2016
How do we get and use energy?
1 EJ = 0.27 trillion kWhr
3
5/29/2016
Electricity usage worldwide – and me
My usage,
had to use elec heaters in bedrooms
for Dec-09 and some of Jan-10
3x the usage and 1.5x the rate –
and 1 cord of biomass to heat the
rest of the house
4
5/29/2016
Why Gasoline?
Superb energy source


>30x lithium batteries
Cheaper than electricity


2/3rd the price/kWh
Easy to store and
transport

5
5/29/2016
Why Not Gasoline?
Of course, it is an increasingly dangerous and difficult process to get gasoline
6
5/29/2016
Why Renewables?
More people
burning
more fuel
7
More CO2
Climate Change
5/29/2016
How to reduce CO2
8
5/29/2016
What are renewable energy sources –
and how much can they help?
9
5/29/2016
How much renewable is used?
USA renewable electricity
generation 2009
World electricity
generation 2008
World energy
consumption (2008)
10
5/29/2016
Why Solar?

Solar “fuel” is free and plentiful


Solar has no carbon emissions


No moving parts
Solar panel lifetime is 30 years


Nothing is burned – remote fusion reactor
Solar panels are essentially maintenance free


6000x the entire worldwide mankind energy usage
Or greater, still being determined
Solar is load following

11
Output is during the day,
when you need it most
5/29/2016
Why Not Solar?

Solar is expensive



Solar rooftop panels hurt utility revenue


2x-5x coal/hydro generation
Special subsidies required or payback in decades
At premium charge rates
Solar disrupts the grid

Daytime only, and spiky

12
Awaiting storage solutions
5/29/2016
Solar TAM/SAM

20 year build out with
1.1 GWp added/year



13
1.1 GWp added 2010
2.4% overall TAM
growth (WW elec use)
Average uses 4.5hrs/day
insolation
5/29/2016
What kinds of solar cells are there?
PV cells
Thin Film
a-Si
14
CIGS/
CIS
CdTe
Organic
Wafer
Mono
c-Si
Multi cSi
III-V
5/29/2016
Making solar cells – wafer based
Poly silicon
Factory
•40 acres
•$1 billion
•$0.30/W
15
Ingot/wafer
factory
•Acres
•$200 million
•$0.30/W
Cell factory
•<acre
•$20 million
•$0.25/W
Module
factory
PV system
•<acre
•$20 million
•$0.25/W
5/29/2016
•Inverters/
wiring
•$1.00/W
Making solar cells – thin film
Back
metal/absorber
deposition
• 10 systems
• 100 million
dollars
TCO deposition
• 3 systems
• 10 million
dollars
Laser scribing
• 5 systems
• 5 million
dollars
Encapsulate/test
• 2 systems
• 5 million
dollars
1 acre
$0.76/W
16
5/29/2016
Making solar cells – thin film structures
17
5/29/2016
AMAT thin film flame out



Used their already existing display deposition systems
Never could get high enough efficiency to combat the high
equipment cost (got 9%, needed >>11%)
CdTe (First Solar) got their modules to work

18
Cheaper equipment, 11% efficiency
5/29/2016
Making solar cells – III V structures


19
Laterally or vertically stack
different band gap cells to
capture more of the
incoming energy
Concentrate light on cells
with mirrors since cells
are expensive (5x-10x cSi)
5/29/2016
How good are they – R&D?
42.7% UNSW
20.3% ZHF
20
5/29/2016
How good are they – actually?
21
5/29/2016
What new engineering is needed

Manufacturing engineering





Factory scheduling/automation improvement
Deposition/scribing/ improvement
Reliability engineering
System engineering
Semiconductor engineering

New classes of absorbers/windows/conductors




Quantum dot
Dye sensitized titanates
Better inverters
Materials engineering

22
Better substrates/windows/conductors
5/29/2016
Solar Market Dynamics
Spanish government
subsidy capped,
Credit crunch
Panel price reductions
Market tied to subsidies/funding/pricing
23
5/29/2016
How much do we care?
• USA to now has been a small player
compared to Europe
• Europe is driven by feed in tariff (FIT)
subsidies
24
5/29/2016
Who builds PV?
25

California permits shown

Utility scale dominates new
application
5/29/2016
How much money can we make?
10 Life of fab (Years)
30 Life of panel (Years)
1.0% Panel degradation/year
4.5 Insolation (kW-hr/day/m2)
10% Site areal efficiency (W/m2/Sol)
$1.00 Capex cost/fab ($/W)
$0.85 Fab Operating Cost ($/Watt)
$1.50 Panel ASP $/Watt
$1.50 BOS ($/W)
$0.001 Oper & Maint/MW ($M)
$4,000 Site cost ($/acre)
$0.38 PPA begin $/kw-hr
5 PPA years transition to unsubsidized
3 PPA Transition begin year
20 PPA Rate Guarantee (Years)
$0.15 unsubsidized $/kW-hr
1% unsubsidized $/kW-hr CAGR
6% Capex cost of capital (%)
5 Capex term of loan (years)
15% NPV discount rate
26

Three scenarios




Build fab – sell panels
Buy panels – sell electricity
in PPA
Build fab – use panels in
PPA’s (fab to farm)
100MW First Solar (CdTe)
subsidized partnership with
French electrical utility
(EDF) modeled
5/29/2016
How much money can we make –
PPA?
Cash keeps coming,
Decent returns
A long time to get above water
27
5/29/2016
How much money can we make –
Fab to farm?
A lot of long term cash
A long time to get above water
28
5/29/2016
How much money can we make –
disrupted ?
10 Life of fab (Years)
30 Life of panel (Years)
0.3% Panel degradation/year
4.5 Insolation (kW-hr/day/m2)
15% Site areal efficiency (W/m2/Sol)
$1.00 Capex cost/fab ($/W)
$0.40 Fab Operating Cost ($/Watt)
$1.00 Panel ASP $/Watt
$1.00 BOS ($/W)
$0.001 Oper & Maint/MW ($M)
$4,000 Site cost ($/acre)
$0.38 PPA begin $/kw-hr
5 PPA years transition to unsubsidized
3 PPA Transition begin year
20 PPA Rate Guarantee (Years)
$0.15 unsubsidized $/kW-hr
1% unsubsidized $/kW-hr CAGR
6% Capex cost of capital (%)
5 Capex term of loan (years)
15% NPV discount rate
Insanely high
returns
Quick payback
29
5/29/2016
How much money can we make –
disrupted but no subsidies?
Adequate long term
returns
Slow payback
30
5/29/2016
Why a smart grid?
This is 10 acre
installation solar
variability –
90% drop in 10
seconds,
wind , below, is
much worse
31
5/29/2016
Conclusions

Greentech is required for planet health

Solar is a natural solution


32
Needs disruptive improvement to move beyond
subsidies
Needs every kind of engineering to implement
5/29/2016
Who Engineers Deal With
And what you need to know about them
33
5/29/2016
Companies are run by:

Finance


Marketing/Sales


TAM/SAM/SOM, applications, installation, product introduction
Manufacturing


IRR, NPV, ROI, ROA, GM, EBIT, EBITDA, net income (bottom
line), revenue (top line)
Turns, BOM, GM, supply chain, ship schedule, warranty
Engineering

34
Performance
5/29/2016
Simple questions – Overall goals

What products should/can we develop?

35
How much business could be done and what is
it worth to us?
5/29/2016
Value to Customer

CoO (cost of ownership) =
cost/output



ROI (return on investment) =
savings/cost



Typical for required purchases
Easy to explain and sell
Typical for discretionary purchases
(IT or factory optimization tools)
Hard to prove value – proof data
proprietary/unknown
Fads/Emotional want = ???

36
Talk to a psychologist
5/29/2016
Market Validation – find out what they want
1.
2.
3.
4.
5.
6.
7.
8.
Cross-functional team (Engineering and program mgr)
Business case analysis and validation plan
Validation prototype of least viable product
Long structured customer meetings in validation format
Notes in “Voice of the Customer” format
Triage (changing product and pitch on the road)
Competitive analysis
Fact finding, analysis and reporting
37
5/29/2016
BACKUP
38
5/29/2016
How much renewable available?
Yearly Solar fluxes &
Human Energy Consumption
39
Solar
3,850,000 EJ
Wind
2,250 EJ
Biomass
3,000 EJ
Primary energy use (2005)
487 EJ
Electricity (2005)
56.7 EJ
5/29/2016
California 2009 Electricity Sources
Fuel Type
Coal
In-State
Percent of
Northwest
Generation California InImports
(GWh)
State Power
Southwest
Imports
Total System
Power
3,735
1.8%
N/A
N/A
N/A
Large Hydro
25,094
12.2%
N/A
N/A
N/A
Natural Gas
116,716
56.7%
N/A
N/A
N/A
Nuclear
Oil
Other
31,509
67
7
15.3%
0.0%
0.0%
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
Renewables
28,567
13.9%
N/A
N/A
N/A
5,685
2.8%
N/A
N/A
N/A
Geothermal
12,907
6.3%
N/A
N/A
N/A
Small Hydro
4,181
2.0%
N/A
N/A
N/A
846
4,949
205,695
0.4%
2.4%
100.0%
N/A
N/A
19,929
N/A
N/A
71,201
N/A
N/A
296,827
Biomass
Solar
Wind
Total
Source:
EIA, QFER, and SB 105 Reporting Requirements
Note: Due to legislative changes required by Assembly Bill 162 (2009), the California Air Resources Board is currently undertaking the task of identifying the fuel sources
associated with all imported power entering into California.
40
5/29/2016
How much does electricity generation cost?
2008 congressional research report – wildly out of date on solar PV , more like $125/Mwh
41
5/29/2016
How much does electricity generation cost –
counting CO2?
2008 congressional research report – wildly out of date on solar PV , more like $125/Mwh
42
5/29/2016
How much to build a power plant?
43
5/29/2016
44
5/29/2016
How much money can we make –
Panel sales?
Cash flow done in ten years
but 2x the early cash of PPA
A short time to get above water
45
5/29/2016
Centrotherm CIGS



Sputtered (PVD) CIG films, ZnO TCO, Moly back conductor
Atmospheric deposition Selenium deposition plus anneal
Laser P1 scribe, mechanical P2,P3 scribes
Wet chemical CdS buffer
Selenium plus anneal
Laser and
mechanical scribe
PVD CIG,
ZnO, Mo
I-V test
Lamination
I-V test
46
5/29/2016
What’s the Product?

Value to customer is usually
solving a problem for them


Better/faster/cheaper
Fad products are satisfying a want

47
Tamagotchi, pink bluetooth headsets
5/29/2016
How much business could be done?

TAM – What could be done


SAM – What is done


Us + competitor sales
SOM – What we do (market share)


Model based upon use case
Us/SAM
Adoption = SAM/TAM

48
Features/segments/customers
5/29/2016
Semiconductor Fab Growth Dynamics
Added wafer area/year
8” equivalent wafers
World GDP (USD)
10,000,000
100,000
+10%
World GDP (USD Billions)
CAGR = +14%
1,000,000
-6%
CAGR = +11%
+3%
Added wafer area/year ~ world GDP growth
Source: IMF GDP USD, Semi fab dbase
49
5/29/2016
2005
2000
1995
2010
2005
2000
1995
1990
1985
1990
10,000
100,000
1985
Added 8" wsm
+10%
50
Price
Performance 2
Performance 1
Tput
Versatility
Customer 1
Customer 2
Performance 2
5/29/2016
Customer 2
Product 1
Customer 1
Customer 2
Customer 1
Customer 1
Customer 2
Conjoint Data
Product 2
Validation Meetings
Team
Debrief
11. Action items
10. Epiphanize
Take
Aways
9. Assign Grades
Customer Meeting
8. Debrief
7. Reaffirm & Retest
6. Perform conjoint
5. Conduct trial closes
4. Test product roadmap and justifications
3. Draw out objections and requests
2. Present and demo the product
1. Facts: Who, what, why, where, problems, and solutions
0
51
1-2
2-3
Meeting Duration In Hours
3-4
5/29/2016
Marketing & Engineering
Co-Ownership of Requirements
Marketing Owns Requirements
Engineering Owns Responses
Product Development team Co-Owns Negotiations & Agreed upon Specs
--- Communication, Commitments, Revisions & Tracking --52
5/29/2016
Disruptive Technology

It is –




New technology that enters the market at the bottom of the
market requirements - and
Has an improvement rate that rapidly overtakes the
established players - and
Prevalent when established vendors overshoot the market
requirements
It is not –


53
A competitor who chooses to buy his way into the market
New technology that doesn’t serve the market
5/29/2016
Disruptive Technology

If a market is still underserved –




High performance is most valued
Integration of functions/data is important
Value goes to the integrators
If a market is overserved –



54
Price and speed/flexibility are most valued
Clear interface boundaries and requirements exist between
the the subsystems making up the product/service
Value goes to the subsystem suppliers
5/29/2016
Yield Value and Efficiency Value

Long runner 1% yield improvement > $2M/month

Additional thruput +5% ≈ $0.6M/month

Fab cycle time reduced by 1% by greater toolavailability >
$2M/month
55
5/29/2016
Parametric Yield Value –
Value linear w/parameter


$40 GPU chip


5 bins used
NVidia at TSMC
Std Dev improved 1/3rd
$900
$800
Before
After
$700
Revenue/month

Thousands
Logic and mixed signal
$600
$500
$400
$300
$200
$100

56
Additional Revenue
≈ $6M/month
$0.8
1.0
1.2
1.4
1.6
1.8
2.0
Leakage quality measure
5/29/2016
2.2
Ramp Dynamics Value
Faster ramp, development, higher yield
35
90%
30
kppm starts
25
80%
kppm natural yielded
tuned yield %
70%
60%
20
50%
15
40%
Natural yield %
30%
10
20%
5
Yield
kwsm
Linear wsm ramp x
Linear yield ramp =
Parabolic yielded wsm ramp
100%
Scanner selection/tuning =
Added yield =
$M/month production revenue
$$M more ramp revenue
Ramp time reduction =
Less ramp total output but
More months at production =
$$M/month reduction
10%
0
0%
0
1
2
3
4
5
6
7
8
9
10
Months of Ramp + Production
57
11
12
Development time reduction =
More months at production =
$$$M/month reduction
5/29/2016
Production wafers per month
Chips/wafer
ASP of chip
ASP degradation/year production
Cost of wafer (Amortization + materials)
Nominal k ppm ramp (months)
Starting yield
Ending yield (natural yield)
Yield improvement w/tuning
Inputs
30,000
200
$40
50%
$20
6
40%
85%
1%
Total natural yielded ramp plates
Total natural yielded ramp revenue
Total tuned yielded ramp revenue
Steady State Revenue/month 100% yield, start of full production
Calculations
63,000
$504,000,000
$509,040,000
$240,000,000
Additional
Additional
Additional
Additional
Savings
$5,040,000
$2,400,000
$120,000,000
$204,000,000
ramp revenue w/tuning yield improvement
revenue per month w/tuning yield improvement
revenue per month less ramp duration
revenue per month less dev'l duration
Cum $M tuning added revenue
Added Revenue/Profit
$25
$20
$15
$10
$5
$0
0
1
2
3
4
5
6
7
9
10
11
Months of Ramp + Production
Note: Spreadsheet is embedded, doubleclick to edit
58
8
5/29/2016
12
TF Startups



a-Si (AMAT, Oerlikon) equip base in trouble ($2/W capex)
CdTe turnkey from Roth & Rau ($1/W capex)
CIGS turnkey from Centrotherm ($2/W capex)
59
5/29/2016
Requirements – MRD Template
Most important – Clearly shows engineering responses and limitations
60
5/29/2016