EOTPR Customer Case Studies EUFANET Workshop: Findings OPEN?

Position of peaks provides fault location. Shorts or low impedances are shown by EOTPR as negative peaks. Substrate. Interposer. Si Die. B. G. A. Pa c k a g e.
2MB taille 95 téléchargements 386 vues
EOTPR Customer Case Studies

EUFANET Workshop: Findings OPEN?

OUTLINE o EOTPR introduction – basic scheme

o EOTPR OPEN customer case studies o Open on BGA trace (evaluation) o Open on embedded BGA trace o Open at C4 bump level

Why use THz for a TDR system?

3

TDR

EOTPR

Qualitative technique

Quantitative technique 3

Terahertz Engine Schematic – core of all Terahertz products

Device under test

Ti:Sapphire fs laser

THz Emitter

optical fibres

Probe

THz Receiver

optical delay Computer

4

Examples of failures identified by EOTPR in 2.5D packages: • Shorts, Opens, and Resistive Opens. • Proven detection capability up to die. µ-bumps (top and bottom) Within RDL and TSV

C4 bumps (top and bottom)

Substrate trace

5

EOTPR fault isolation in advanced packages • Position of peaks provides fault location

Interposer Si Die

Substrate

Opens or high impedances (resistive opens) are shown by EOTPR as positive peaks

30 20

EOTPR intensity (a.u.)

10 0 -10 -20 -30 -40 Shorted probe Bare substrate Interposer substrate Good unit

-50 -60 -70 0

10

15

Silicon Die

Silicon Interposer

Package Substrate

6

BGA

Shorts or low impedances are shown by EOTPR as negative peaks

5

TeraView EOTPR: Customer case study I Open in package substrate Device A

• Device A and B both have a FIB cut in an identical trace • The position of the cut is separated by ~90 µm in the devices (measured from BGA to start of the FIB cut) • EOTPR can clearly identify the difference in location of the open circuit in the two devices

FIB cuts in traces

Open circuit features

200

89 µm Failed Device A Failed Device B

BGA feature

EOTPR signal (a.u.)

EOTPR signal (a.u.)

100

50

100

50

0

0.8

-1.0

-0.5

0.0

0.5

1.0

1.5

Distance into DUT (mm)

7

Failed Device A Failed Device B

150

150

-50 -1.5

Device B

2.0

2.5

3.0

0.9

1.0

1.1

1.2

Distance into DUT (mm)

1.3

1.4

1.5

TeraView EOTPR: Customer case study II Open in package substrate TDR

EOTPR

• TDR result - Fault at substrate side • EOTPR result - Distance-to-Defect accurately calculated and confirmed with PFA (see next slide) 8

*Presented at IPFA 2012, paper 19-68

TeraView EOTPR: Customer case study II Open in package substrate

• PFA result: Cu trace broken at Metal7 of the substrate • Design layout - Actual Distance = 9636.76 µm • PFA result confirms EOTPR fault loaction 9

*Presented at IPFA 2012, paper 19-68

TeraView EOTPR customer case study III: Flip Chip • An open fault was identified during electrical testing of a flip chip product. • EOTPR was used to locate the fault within the failed unit. • The plot below shows EOTPR waveforms from a reference bare substrate and the failed unit. The bare substrate terminates at the C4 bump pad, before the start of the C4. Failed trace route

C4

BGA

Contact feature

Substrate termination

Open in failed unit

Bare substrate Failed unit

TeraView EOTPR customer case study III: Flip Chip • The EOTPR waveforms show an open circuit in the failed unit, after the bare substrate termination.

Substrate termination

• The EOTPR fault location was subsequently confirmed by physical failure analysis as noncontact between UBM and Al contact pad.

C4 bump

C4 bump pad

• From EOTPR we determine the fault is located 83 µm after the bare substrate termination, putting it at the die side of C4, likely at the UBM.

Open in failed unit

83 µm

Physical failure analysis results:

80 µm

Good unit

Failed unit

C4 bump

Bare substrate

C4 bump pad • UBM in contact with Al pad

• UBM not in contact with Al pad

Failed unit

Comparison of EOTPR vs. conventional TDR Property Input signal shape

Frequency range

Rise time

EOTPR Impulse

TDR Step function

DC to greater than 100 GHz

< 10 ps (raw)

EOTPR added value EOTPR makes it easier to pinpoint faults as peak positions give fault locations.

DC to 35 GHz

EOTPR higher frequency content allows for faster rise times compared to conventional TDR.

~ 35 ps

EOTPR faster rise time produces improved resolution compared to conventional TDR.

< 6 ps (deconvolved) Time base jitter

< 30 femtoseconds (30×10-15 seconds)

> 500 femtoseconds (500×10-15 seconds)

EOTPR high time base stability allows for increase fault location accuracy compared to conventional TDR.

Signal to noise

> 64 dB

45 dB

EOTPR high SNR increases its sensitivity to small changes in impedance.

12

12

oThanks you for your attention oQuestions?

TeraView EOTPR: Accuracy We determine EOTPR System accuracy in the following way: • Injected pulse reflects off device/interconnect • Probe is displaced in 10µm steps along interconnect to demonstrate accuracy • Change in position of reflected signal is measured

Interconnects

Displacement

Step size 10µm 14

Probe

TeraView EOTPR: Accuracy • This demonstrates EOTPR has an accuracy of 30 femtoseconds. In a typical DUT this is equivalent to an accuracy of 10 µm.

Reflection from end of test device track

10 repeat measurements collected at each step

0 µm

100 µm

10 µm steps

30 femtoseconds = 0.03 picoseconds = 10 µm 15

TeraView EOTPR: Customer case study III Package substrate impedance change EOTPR results

PFA results

III

25

II

15

EOTPR signal (a.u.)

Thinner trace

I

20

Vendor 1 Cu trace

10 5 0 -5 -10 -15

Vendor 1 Vendor 2

-20 -25 0

10

20

Vendor 2 Cu trace

Distance in DUT /mm



• 16

EOTPR result - show significant impedance differences between two substrate vendors. PFA result – Cu trace width of vendor 1 thinner than vendor 2. *Presented at IPFA 2012, paper 19-68