1

HSTSS Silicon Gun Pressure Sensor - Phase IDavid Combes - QinetiQ Malvern22/Jun/2004

2

• Project and aims

• Approach– Modelling, fabrication & packaging

• Packaging issues

• Test results

• Conclusion

• Questions

Introduction

Origins of QinetiQOrigins of QinetiQ

Air Force

Navy

Army

Radar &Electronics

• Air Systems

• Land Systems

• Sea Systems

• Weapons

• Command & Info Systems

• Electronics

• Sensors & Processing

• Structural Materials

• Ranges

• Air Capabilities

• Engine Test

• Chemical & Biological

• Human Sciences

• Chemical & Electronics

• Defence Analysis

RAE

ARE

RARDE

RSRE

CBDELife Sciences

• Sensors & Electronics

• Knowledge & Information Systems

• Future Systems Technology

• Complex Managed Services

• Sensitive areas of research

July 2001

3

• Projectile mounted, reduced cost miniature pressure sensor for HSTSS (Hardened Subminiature Telemetry Sensor System)

– Alternative to piezoelectric, Kistler type gauges

– Acceleration, light and temperature insensitive

– Small, and easy to read out

• To survive gun launch accelerations up to 100,000g

• To survive and measure pressures up to 100,000 psi

– Accurate and repeatable measurement

Overall project aims

• Mainly concerned with prototype sensor element– To survive and measure pressures, and inform further work

required to achieve aims– ~ 1 year programme

• Further phase required for development and testing required to demonstrate fully functional, production ready device

• Output of phase I - 8 packaged sensor elements– Capable of measuring pressures up to 100,000 psi– Capable of surviving gun launch– Minimal form factor

Phase I

4

• Fort Halstead - Gun systems– Gun and ammunition systems research

– Gun and rocket test ranges and instrumentation

• Malvern - Microsystems & Microengineering– MEMS design and modelling

– Electronic sub-system design

– Microstructure fabrication

– Advanced characterisation

– Engineering solutions

Project team

• Bridge configured piezo-resistive silicon sensor

• Provides linear output voltage with change in pressure

• DC response characteristics

• Silicon - robust, cheap, readily processed

• Allows exploitation and application of MEMS technologies, and potential integration of smart functions– Temperature compensation etc

• Suitable for miniaturisation

Approach

5

• Piezoresistive membrane sensor

Approach

Sensor glued intohousing

Interconnect fromsensor to outside

world

Pressure

Blast shield to protectfrom debrisPressure Pressure

A A

membrane

A A

Light Shield

piezo-resistors

Mechanical Modelling

0 100 200 300 400500

1000

1500

2000

2500

Longitudinal @ 0°Transverse @ 0°Longitudinal @ 45°Transverse @ 45°

Distance from membrane centre

Stre

ss (M

Pa)

membrane_edge

• Mechanical finite element analysis– To inform design

– Map stresses in device

– Map strain for piezoresistors

– Package included in analysis

• 100µm fillet radius chosen to give acceptable stresses

6

• Silicon has high theoretical intrinsic strength– Practical strength depends on surface finish and defects– Quoted figures vary widely

• 7GPa - 0.7GPa

• Fabrication process a key component of work– Need smooth etch profile (in all directions)– Use Deep Reactive Ion Etch (DRIE) tool to create initial

circular membrane profile– Post process to smooth further– Etch dice

Fabrication

• Wet etching - wetting problems

• Gas phase etches investigated

• Fluorine based plasma etch-smooth profile, good surface finish

• Wet etching may be used to further smooth

Isotropic etch development

7

• Saw/Cleave– Not highly controllable– Limited geometry (straight)– Forms stress raisers

• Etch dice using DRIE

Etch dicing

Dice etch

Membrane etch

Dice etch

• Steel M12 package, similar to Kistler M6213

• Gasket/attach layer for testing - Epoxy

• Face sealing using compression washer

Packaging

8

Maximum survived pressure

0.0

20.0

40.0

60.0

80.0

100.0

120.0

0 100 200 300 400 500 600

Pre-isotropic etch membrane diameter (µm)

Pres

sure

(kps

i)

t=100µm

Mechanical test results

• Carried out in parallel with mechanical modelling

• Survival test - Budenberg dead weight tested to 100kpsi

• Membranes with no transducer elements tested - available early

• Several membrane designs showed they were suitable for high pressures

• Stress limit of around 2GPa

• P-Type piezoresistors– Allows lower temperature

co-efficient of piezoresistivity

• Simulated using Silvaco’sAthena

• Test devices fabricated to validate models

• Simulated ≈ 680Ω/

• Measured ≈ 710Ω/

Piezoresistors

9

-7

-6

-5

-4

-3

-2

-1

020 30 40 50 60 70 80

Temperature (°C)

% c

hang

e in

pie

zore

sist

ive

coef

ficie

n

<0.1% per °C

Piezoresistors

-10

-8

-6

-4

-2

0

2

4

6

8

10

0 50 100 150 200 250

Stress (MPa)

DR

/R (%

)

Longitudinal

Transverse

at 45°410ppm/MPa

-350ppm/MPa

50ppm/MPa

Temperature sensitivityPiezoelectric coefficients

• Good match with modelling

• Low temperature sensitivity, potential for further gains

Membranes with piezoresistors• Bridge configured sensor

• Two sensitive, two insensitive devices

Piezoresistive Coefficients

Transverse

Longitudinal

Insensitive

Sensitive

Membrane edge

10

• Attachment of suitable bond wires to small die area proved difficult

• Wire-bonder identified which provides solution - Westbond 7440– Some problems obtaining samples -

now solved– Will be testing in next few weeks

• Modifications made to package to allow conventional bonds and soldered connections

Packaging issues

Bond wires

• Uses ceramic header, metal support– Conventional wire bonds to header– Solder connections from header

Modified package

11

• Provides devices with less support– Increases strain in die and membrane

• Die level device failures under reduced pressure

• Analysis of failed devices show modified package not providing sufficient support

• No such problems with unmodified package– Devices in original package type with

bonds made by a Westbond tool (or similar) are the best solution

Modified package issues

-40

-20

0

20

40

60

80

100

120

0 10000 20000 30000 40000 50000 60000

Pressure (psi)

Out

put v

olta

ge (m

V)

1

2

3

Linear (3)

• Devices in modified package tested - Electrical performance as expected– Linear & repeatable

• Epoxy relaxation an issue -alternatives under investigation

• Devices in unmodified package tested with no electrical connections– Most sensitive (weakest)

devices showed they were suitable for full range pressure measurement

Transducer testing

12

• A silicon piezoresistor bridge was identified as capable of meeting requirements

• Models were created to simulate both the mechanical and semiconductor physics of the devices

– Devices were fabricated to validate models

– Good predictive performance

• Processes were developed for fabrication of the mechanical structure and implanted piezo-resistors

• Solutions identified for packaging

Conclusion

• A silicon piezo-resistive pressure sensor has been developed suitable for

– Measuring 100,000psi (high degree of confidence)

– Surviving 100,000g (high degree of confidence)

– Low cost production

– Miniaturisation

– Easy interface

• Phase II funding is being sought to further develop the sensor and packaging to facilitate production of such sensors for ballistic applications

Conclusion

13

Supplemental information

14

• Profitable, growing, high technology P.L.C. formed from majority of DERA

• Approx. 8,500 staff

• Turnover approx. $1.2B – 80% for Ministry of Defence– Commercial work growing rapidly

• One of Europe’s largest science and technology organisations

• Approx. 3200 patents in force, 1400 pending

• 10 spin-out companies

What is QinetiQ?