Evaluation of VCSEL Mounted on CVD DiamondSubstrates

Harry C. Shaw301-286-6616

Associate Branch HeadComponent Technology and Radiation Effects Branch

Goddard Space Flight Centerharry.c.shaw@gsfc.nasa.gov

Melanie N. Ott301-286-0127

Director, Advanced Photonic Interconnection ManufacturingLaboratory

Director, Technology Validation Laboratory for PhotonicsSigma Research and Engineering

Goddard Space Flight CenterMelanie.Ott@gsfc.nasa.gov

Evaluation of Vertical Cavity Surface Emitting Lasers(VCSEL) mounted on CVD Diamond Substrates

Part 1 of ReportPart 1B of ReportPart 2 of ReportConclusion of Report

NEPP Publication: Evaluation of VCSELS on Diamond

http://misspiggy.gsfc.nasa.gov/tva/meldoc/vcsel/index.htm [10/23/2001 2:19:02 PM]

Evaluation of Vertical Cavity Surface Emitting Lasers (VCSEL) mounted on CVD Diamond Substrates

NASA/GSFC Component Technology and Radiation Effects Branch

September 21, 2001

Contact: Melanie Ott, Sigma Research and Engineering, (301)-286-0127 Harry Shaw, NASA/GSFC, 301-286-6616

2

Introduction………………………………………………………………………………………………………………...3 VCSEL Optical output power vs bias…………………………………………………………………………………….6 Characteristics of the VCSEL-on-diamond………………………………………………………………………………7 Characteristics of the VCSELs mounted directly to Kovar headers………………………………………………….16 Wavelength shifting of the VCSELs……………………………………………………………………………………..24 Conclusions………………………………………………………………………………………………………………..29

3

Introduction This report summarizes the evaluation of 850 nm VCSELs manufactured by Emcore. The purpose of the evaluation was determine if the superior thermal properties of diamond as a substrate would improve the performance and reliability of the mounted VCSELs. Four VCSELs were mounted to diamond substrates and four more were mounted directly to Kovar headers and subjected accelerated voltage and current overdrive conditions. The diamond substrates provided a significant reliability advantage, however, there were unexplained shifts in the spectral properties of the devices. These mode shifts could be the subject of further investigation. Bare die were purchased and mounted onto gold plated CVD diamond substrates 500um in thickness. Die attach was performed using 84-1 conductive epoxy (Ablestik). 1 mil Au thermosonic wire bonding was used. The individual die were mounted on 250 um centers. The substrates were mounted onto TO-8 gold plated kovar headers. A MTP ribbon cable was used as a waveguide to the HP7095 Optical Spectrum Analyzer (OSA) and HP 8153 Lightwave Multimeters. 4 VCSELs were mounted directly on the TO-8 cans to permit comparison of the performance with and without diamond substrates. Due to alignment fiber-VCSEL alignment difficulties, mating and demating of the ribbon cable was kept to a minimum. Figures 1 and 2 depict the VCSELs unmounted and mounted on diamond. The devices used in the evaluation are no longer available as discrete parts. The next generation is preferred because the devices are available as monolithic arrays. The arrays are preferred because the optical alignment problems with the discrete devices can be eliminated. Figure 2a depicts the bare diamond substrate. Figure 2b is a block diagram of the test setup.

4

Figure 1. 850 nm VCSEL with 3 apertures Figure 2. VCSELs mounted to CVD diamond substrates.

Figure 2a. Bare diamond substrate. Metalization is Ti/Pt/Au. Dimensions: 0.256” x 0.06” x 0.017” (thickness)

8 µm 15 µm 20 µm

5

PowerSupply

HP 7095OSA

HP8153

DUTs

100 Ohms

Figure 2b. Electrical Test Block Diagram

6

1. VCSEL Optical output power vs bias Each of the 4 VCSELs are driven by single power supply through a 100 ohm resistor. Table 1 depicts the power output response of the a typical device under swept bias. This device is mounted directly to the TO-8 can

Table 1 PEAK_A is the power of the highest spectral output component. TOT_PO is the sum of the power of all the spectral components. The detected power is less than the actual power due to fiber-VCSEL misalignment. It is interesting to note that divergence of the two curves at the low and high ends of the bias conditions

TOT_PO (L)PEAK_A (R)

Output power vs swept bias0.275 v to 7.5 v, 0.5 v increments

VDC

Tot

al P

out(

dBm

)

Pea

k S

pect

ral C

ompo

nent

Pou

t (dB

m)

-65

-55

-45

-35

-25

-15

-48

-44

-40

-36

-32

-28

-24

-20

-16

-1 1 3 5 7 9

7

2. Characteristics of the VCSEL-on-diamond. The following charts summarize the 41 days of continuous monitoring of the VCSEL-on-diamond units. One of the 4 devices (VCSEL_1) was monitored by the OSA. The other 3 devices (VCSEL_2, _5, _8) were monitored the lightwave multimeter. For the latter devices, the only parameter recorded by the LABVIEW program is total power out. The summary of the raw power out numbers is shown in Table 2

8

Statistical Summary of Raw Power Output Data (dBm) for VCSEL_1 Table 2

5 V bias Observations Mean Pout Minimum Pout Maximum Pout Std.Dev. VCSEL_1 674 -17.6358 -18.5157 -16.9374 .284717 VCSEL_5 674 -7.7309 -8.3565 -7.4715 .084840 VCSEL_2 674 -6.4325 -6.8194 -6.0730 .194107 VCSEL_8 674 -8.4453 -8.6328 -8.2102 .114320

6.25 V bias Observations Mean Pout Minimum Pout Maximum Pout Std.Dev. VCSEL_1 913 -20.3849 -21.0891 -19.7740 .155107 VCSEL_5 913 -9.5464 -9.8297 -9.3930 .059760 VCSEL_2 913 -8.4815 -8.7290 -8.0134 .126077 VCSEL_8 913 -9.1052 -9.4310 -8.6967 .144907 7.5 V bias Observations Mean Pout Minimum Pout Maximum Pout Std.Dev. VCSEL_1 387 -23.5835 -24.8671 -22.7887 .498200 VCSEL_5 387 -11.8555 -12.2915 -11.5739 .123101 VCSEL_2 387 -12.3018 -12.9328 -11.9314 .156877 VCSEL_8 387 -11.4199 -12.1042 -10.9691 .199144

9

Figure 3. 5 V VCSEL data

VCSEL_1VCSEL_5VCSEL_2VCSEL_8

VCSEL characteristics at 25 C. 5V through 100 Ohm resistorVCSEL_1 = -17.911+0.001*x+eps

VCSEL_5 = -7.611-0*x+epsVCSEL_2 = -6.317-0*x+epsVCSEL_8 = -8.282-0*x+eps

14 days

Opt

ical

Pou

t (dB

m)

-20

-18

-16

-14

-12

-10

-8

-6

-4

10

Figure 4. 6.25V VCSEL data

VCSEL_1VCSEL_5VCSEL_2VCSEL_8

VCSEL characteristics at 25 C. 6.25V through 100 ohm resistorVCSEL_1 = -20.494+0*x+eps

VCSEL_5 = -9.576+0.0001*x+epsVCSEL_2 = -8.281-0*x+epsVCSEL_8 = -8.898-0*x+eps

19 days

Opt

ical

Pou

t (dB

m)

-24

-22

-20

-18

-16

-14

-12

-10

-8

-6

11

Figure 6. 7.5 V VCSEL data

VCSEL_1VCSEL_5VCSEL_2VCSEL_8

VCSEL characteristics at 25 C. 7.5 V through 100 ohm resistorVCSEL_1 = -24.27+0.004*x+eps

VCSEL_5 = -11.873+0.0001*x+epsVCSEL_2 = -12.284-0.0001*x+epsVCSEL_8 = -11.596+0.001*x+eps

8 days

Opt

ical

Pou

t (dB

m)

-26

-24

-22

-20

-18

-16

-14

-12

-10

-8

12

Figure 7. 3D Surface Plot of Power vs FHWM (5V)

-18.684 -18.569 -18.453 -18.337 -18.222 -18.106 -17.990 -17.874 -17.759 -17.643 -17.527 -17.412 -17.296 -17.180 -17.065 -16.949

3D Surface Plot of Power vs FWHM (5V)z = 0.086+1.002*x-6.543e7*y

13

Figure 8. 3D Surface Plot of Power vs FHWM (6.25V)

-21.287 -21.185 -21.083 -20.981 -20.879 -20.777 -20.675 -20.573 -20.471 -20.370 -20.268 -20.166 -20.064 -19.962 -19.860 -19.758

3D Surface Plot of Power vs FWHM (6.25V)z = 0.218+1.007*x-8.698e7*y

14

Figure 9. 3D Surface Plot of Power vs FHWM (7.5V)

-26.612 -26.424 -26.235 -26.047 -25.859 -25.671 -25.482 -25.294 -25.106 -24.918 -24.729 -24.541 -24.353 -24.165 -23.976 -23.788

3D Surface Plot of Power vs FWHM (7.5V)z = 4.319+1.231*x-8.063e7*y

15

Figure 10. Optical Power output summary

VCSEL_1VCSEL_5VCSEL_2VCSEL_8

Optical Power output summary

41 days

Opt

ical

Pou

t (dB

m)

-28

-24

-20

-16

-12

-8

-4

16

3. Characteristics of the VCSELs mounted directly to Kovar headers. The VCSELs mounted directly to the Kovar headers exhibited a power slump over time. After 2 days at 6.25 V, all 4 devices stopped exhibiting any power output at 850nm.

Figure 11. Comparison of Optical power out on diamond vs Kovar substrate

TOTAL_PO (L)TOT_PO (R)

Comparison of Optical power out on diamond vs Kovar5V bias

Opt

ical

Pou

t (dB

m)

w/o

dia

mon

d su

bstr

ate

Opt

ical

Pou

t (dB

m)

w/ d

iam

ond

subs

trat

e

-18.8

-18.4

-18

-17.6

-17.2

-16.8

-22

-21.8

-21.6

-21.4

-21.2

-21

-20.8

-20.6

-20.4

17

Table 3

Statistical Comparison of Output power data (dBm) for VCSEL mounted to diamond substrate vs Kovar header over the same number of observations

Observations Mean Pout Minimum Pout Maximum Pout Variance Std.Dev. PEAK_A 674 -21.2620 -22.1000 -20.8000 .058823 .242534 TOTAL_PO 674 -21.1051 -21.8966 -20.6775 .055431 .235438 PEAK_B 674 -17.7111 -18.6000 -17.0000 .084051 .289916 TOT_PO 674 -17.6358 -18.5157 -16.9374 .081063 .284717 PEAK_A = Peak power of primary spectral peak for Kovar mounted sample TOTAL_PO = Total power output for Kovar mounted sample PEAK_B = Peak power of primary spectral peak for diamond mounted sample TOT_PO = Total power out for diamond mounted sample

18

Figure 12. Optical power output for s/n 1 mounted on Kovar

PEAK_A (L)TOTAL_PO (R)

Optical power output for s/n1 mounted on Kovar

28 days

Pou

t (dB

m) P

eak

Spe

ctra

l Out

put

Pou

t (dB

m) T

otal

Pow

er

-25.5

-24.5

-23.5

-22.5

-21.5

-20.5

-19.5

-25.3-25.1-24.9-24.7

-22.3-22.1-21.9-21.7-21.5-21.3-21.1-20.9

19

Figure 13. Optical power for S/N 1 at 6.25 V mounted on Kovar

PEAK_A (L)TOTAL_PO (R)

Optical Power output at 6.25V bias through 100 ohm resistors/n 1 mounted on Kovar

2 days

Pou

t (dB

m) o

f pea

k sp

ectra

l com

pone

nt

Pou

t (dB

m) T

otal

pow

er

-25

-24.9

-24.8

-24.7

-24.6

-24.5

-24.4

-24.3

-25.3

-25.2

-25.1

-25

-24.9

-24.8

-24.7

20

Figure 14. Output power comparison at 6.25 V for diamond and Kovar mounting

TOTAL_PO (L)TOTPO2 (R)

Output power comparison at 6.25 V

2 days

Tot

al P

out (

dBm

) mou

nted

on

Kov

ar

Tot

al P

out (

dBm

) mou

nted

on

diam

ond

-20.9

-20.7

-20.5

-20.3

-20.1

-19.9

-19.7

-25

-24.9

-24.8

-24.7

-24.6

-24.5

-24.4

-24.3

21

Figure 15. Full Width at Half Max comparison between diamond mounted sample and kovar mounted sample. FWHM = 2.355 σ

FWHM Comparison

0.00E+00

5.00E-10

1.00E-09

1.50E-09

2.00E-09

2.50E-09

3.00E-09

3.50E-091 28 55 82

109

136

163

190

217

244

271

298

325

352

379

406

433

460

487

514

541

568

595

622

649

Observations

FWHM Diamond Substrate FWHM Kovar Header

22

Figure 16. Sigma comparison between diamond mounted sample and kovar mounted sample. σ = (Σ Pi, i=1,,n(λi −λmean)2/Ptot)0.5

Sigma (RMS spectral width)

0.00E+00

2.00E-10

4.00E-10

6.00E-10

8.00E-10

1.00E-09

1.20E-09

1.40E-091 25 49 73 97 121

145

169

193

217

241

265

289

313

337

361

385

409

433

457

481

505

529

553

577

601

625

649

673

Observations

Sigma diamond substrate Sigma Kovar Header

23

Figure 17. Mode Spacing comparison between diamond mounted sample and kovar header sample. Mode spacing is the average wavelength spacing between spectral components of the VCSEL 4. vvffd .000000000000000000 5.

Mode spacing

0.00E+00

1.50E-10

3.00E-101 23 45 67 89 111

133

155

177

199

221

243

265

287

309

331

353

375

397

419

441

463

485

507

529

551

573

595

617

639

661

Observations

wav

elen

gth

Mode Sp Diamond Substrate Mode Sp Kovar Header

24

4. Wavelength shifting of VCSELs. The comparison of the spectral performance of the VCSELs on diamond versus Kovar demonstrated provided two results of note. When observed over comparable period at 5V, the VCSELs on diamond had slightly better spectral stabilty. The VCSEL on Kovar had a small blue shift over time. Statisically, the variation between the two was small as shown in table 4. Diamond Diamond Kovar Kovar Mean Wavelength std dev Peak Wavelength std dev Mean Wavelength std dev Peak Wavelength std dev 3.31272E-11 3.23E-11 4.16378E-11 3.99E-11

Table 4. Standard deviation of wavelength stability of diamond and Kovar mounted units. The VCSEL mounted on diamond exhibited 6 abrupt shifts in mean wavelength during the 41 days of testing. Three of the shifts correlated to increases in the bias supplies. Three did not correspond to any change in the test conditions. The shifts are both higher and lower in wavelength. Incidentally, the peak wavelength exhibited only 3 shifts, each occuring at bias increase and each shift was a red shift. The VCSEL mounted on Kovar exhibited one wavelength shift, corresponding to the increase in bias voltage from 5V to 6.25 volts. There is insufficient data to draw any conclusions about the wavelength shifting over than the appearance of mode shifts within the VCLSEL. Additional characterization on a wider sample population and some knowledge of the VCSEL construction would be required.

25

Figure 18. Mean wavelength vs time at 5V

MEANWL1MEANWL2

Mean Wavelength vs time MEANWL1 = 8.537e-7-2.871e-14*x+epsMEANWL2 = 8.506e-7-1.174e-13*x+eps

5V bias

observations

Wav

elen

gth

ME

AN

WL1

= D

IAM

ON

D S

UB

STR

ATE

ME

AN

WL2

= K

OV

AR

HE

AD

ER

8.5e-7

8.505e-7

8.51e-7

8.515e-7

8.52e-7

8.525e-7

8.53e-7

8.535e-7

8.54e-7

8.545e-7

26

Figure 19. Peak wavelength vs. time at 5V

PEAKWL1PEAKWL2

Peak Wavelength vs timePEAKWL1 = 8.537e-7-2.849e-14*x+epsPEAKWL2 = 8.506e-7-1.162e-13*x+eps

5 V Bias

OBSERVATIONS

Wav

elen

gth

PE

AK

WL1

=DIA

MO

ND

SU

BS

TRA

TEP

EA

KW

L2=K

OV

AR

HE

AD

ER

8.5e-7

8.505e-7

8.51e-7

8.515e-7

8.52e-7

8.525e-7

8.53e-7

8.535e-7

8.54e-7

8.545e-7

27

Figure 20. Mean wavelength vs. time for the entire test

MEANWL1MEANWL2

Mean wavelength vs time

observations

wav

elen

gth

ME

AN

WL1

=DIA

MO

ND

SU

BS

TR

AT

E

ME

AN

WL2

=KO

VA

R H

EA

DE

R

8.49e-7

8.5e-7

8.51e-7

8.52e-7

8.53e-7

8.54e-7

8.55e-7

8.56e-7

8.57e-7

8.58e-7

037 74 111

148

185

222

259

296

333

370

407

444

481

518

555

592

629

666

703

740

777

814

851

888

925

962

999

1036

1073

1110

1147

1184

1221

1258

1295

1332

1369

1406

1443

1480

1517

1554

1591

1628

1665

1702

1739

1776

+5V TO +6.25V

+6.25V TO+7.5V

+5V TO +6,25V

28

Figure 21. Peak wavelength vs. time for the entire test

PEAKWL1PEAKWL2

Peak wavelength vs timew

avel

engt

h

PE

AK

WL1

=DIA

MO

ND

SU

BS

TRA

TEP

EA

KW

L2=K

OV

AR

HE

AD

ER

8.49e-7

8.51e-7

8.53e-7

8.55e-7

8.57e-7

8.59e-7

# 1

# 29

# 57

# 85

# 11

3#

141

# 16

9#

197

# 22

5#

253

# 28

1#

309

# 33

7#

365

# 39

3#

421

# 44

9#

477

# 50

5#

533

# 56

1#

589

# 61

7#

645

# 67

3#

701

# 72

9#

757

# 78

5#

813

# 84

1#

869

# 89

7#

925

# 95

3#

981

# 10

09#

1037

# 10

65#

1093

# 11

21#

1149

# 11

77#

1205

# 12

33#

1261

# 12

89#

1317

# 13

45#

1373

# 14

01#

1429

# 14

57#

1485

# 15

13#

1541

# 15

69#

1597

# 16

25#

1653

# 16

81#

1709

# 17

37#

1765

# 17

93

+5V TO+6.25V

+6.25V TO+7.5V

Figure 22 Thermal images at 5 V

bias

Diamond substrate

Kovar Header

28

Figure 23 Thermal images at 6.25 V

bias

Diamond Substrate

Kovar Header

29

Figure 24 Thermal images at 7.5 V

bias

Diamond substrate

Kovar Header

30

29

5.0 Conclusions The VCSELs mounted on diamond were capable of being overdriven at higher current levels without damage. The highly accelerated voltage and current stress demonstrated that the diamond substrate provided a significant margin of thermal mitigation. VCSELs were driven at up to 3X the manufacturers recommended maximum instantaneous operating current and 10X the typical threshold current. Key parameters from the data sheet are summarized below in Table 4.

Parameter Rating Min Typical Max Max Continous

Operating Current 20mA

Max Instantaneous Operating Current

25mA

Peak Wavelength 830nm 850nm 860nm Threshold Current 5mA 8mA 9mA Output Power at

20mA 3dBm 7.8dBm 10dBm

Operating Voltage at 9mA

1.6V 2.2V 2.3V

Table 4. Key Paramaters from Emcore VCSEL data sheet (p/n 8085-1000)

Additional testing is required to assess the effects of screening on the long term reliability of VCSEL mounted on diamond. Additional characterization of wavelength stability in current generation VCSEL arrays is also warranted.