Ultra-Broadband AIGaN Amplifier Development Northrop Grumman Quarterly Program Status Report

Transcription

1 Ultra-Broadband AIGaN Amplfer Development Northrop Grumman Quarterly Program Status Report Contract number: N C-0171 Program: Ultra broadband AIGaN amplfer development Reportng Perod: November February CDRL: A001 - Performance and cost report 1. PROGRAM OBJECTIVE: a) Hghly Lnear Mcrowave Amplfer Program (Orgnal Program Phase) Ths program report descrbes work performed under a three year extenson of an ongong effort to develop hgh lnearty mcrowave amplfers. Work performed under the ntal program phase has now been completed, and the fnal report submtted. Ths quarterly report descrbes selected tasks from the orgnal program phase, ncludng SC hybrd amplfer fabrcaton and test, and AIGaN MODFET fabrcaton and test, to contnue the descrpton of these efforts started n the last program report, or because these efforts are contnued under the current program phase. b) Ultra Broadband AIGaN Amplfer Development (Follow-on Program) The objectve of the program extenson s to develop and demonstrate an AIGaN MODFET based ultrabroadband (1-18GHz) power amplfer wth 3 Watts output power. To acheve ths objectve, the program has four major tasks, each wth a prmary objectve: 1. Develop well controlled materal deposton procedures, usng chemcal vapor deposton, for standard and low defect densty AIGaN materals growth on hgh resstvty 6H SC substrates. These processes wll then be used to depost advanced AIGaN MODFET structures for processng nto dscrete devces, and fabrcaton nto an ultra-broadband dstrbuted amplfer. 2. Desgn, model, and fabrcate advanced dscrete power AIGaN MODFETs, wth mproved power, effcency and lfetme to make them sutable for dstrbuted amplfer fabrcaton. Devce performance goals nclude output power densty > 4 W/mm at 10 and 18 GHz, and output power of 1 Watt at 18GHz. 3. Develop and demonstrate MMIC passve components compatble wth dstrbuted AIGaN MODFET amplfer use on hgh resstvty SC substrates, ncludng Vbd > 50V. 4. Desgn, fabrcate and test an ultra-broadband dstrbuted amplfer usng AIGaN MODFETs wth 3 Watts output power, 1-18GHz bandwdth. 5. WORK PERFORMED DURING THE CURRENT PROGRAM REPORTING PERIOD: Durng the current perod, program work was focused n several areas: 1. Modelng and desgn of AIGaN MODFET structures on SC substrates. 2. Fabrcaton of AIGaN MODFETs deposted on SC substrates usng CVD. 3. Assembly and contnung test of the Cö^rn gate length SC MESFET based hybrd amplfer. 4. The fnal report for the frst program phase was completed and submtted (Hghly lnear mcrowave amplfer development). Detals of the work and results n selected areas are descrbed n the followng sectons. 3. TECHNICAL PROGRESS 3.1 HIGHLY LINEAR MICROWAVE AMPLIFIER PROGRAM: Task R 2.3: SC MESFET Dstrbuted Amplfer Fabrcaton and Test DTIC QUALITY INSPECTED 4 DISTRIBUTION STATEMENT A Approved for Publc Release Dstrbuton Unlmted Page 1

2 Two dstrbuted amplfers have been assembled and tested to date. The completed amplfer s shown n Fgure 1. Fgure 1: Completed hybrd SC dstrbuted amplfer The hybrd amplfer was assembled on an OFHC carrer wth 250(m thck polshed alumna substrates. Gold mcrostrp lnes were defned usng a combnaton of Cr/Au evaporated metal and gold electroplatng to 5jm thckness. Devces and alumna substrates were soldered n place usng AuGe eutectc. Bas decouplng was mplemented usng 18pF MIM chp capactors n seres wth the gate and dran lne 50Q termnaton resstors. Bas current s appled usng external commercal bas-t networks. The completed hybrd amplfer assembly was & 5 Small Sgnal Gan (c K o '-" -j I. I v = 20V I d = 235 ma / y '. ** <'. ' t... - '* ft«a'- - "''.. *." """.*. '"" Ü R Z EDO TtlA V V foom () Frequency (GH a) Lo w Bas Small Sgnal Characterstcs b) DC Characterstcs Fgure 2: Small sgnal and dc characters tc :s of the completed SC dstrbute sd amplf er. Page 2

3 tested at DC, small sgnal and for output power. DC characterstcs of the amplfer are shown n Fgure 2. The assembled amplfer was placed on our standard mcrowave test fxture, whch provdes water coolng. The gold plated brass fxture provdes mcrowave nput and output usng a pressure contact system. Intally, small sgnal testng was performed to establsh gan and return loss levels. Because bas sgnals were njected through the nternal bas networks n the network analyzer, current was lmted to <500mA and voltage to <40V. Also, >40W power dsspaton s requred when operatng at 40V and 50% l DSS. Bas was ntally appled to the gate to a voltage suffcent to pnch off the devce. The dran voltage was gradually ncreased to 20V, then the gate voltage was adjusted to allow 200mA dran current and the gan shown n fgure 2 was recorded. Under these condtons, the low voltage bas and current results n low amplfer gan. Followng ths measurement, the dran voltage was ncreased to the desgn voltage of 40V by ncreasng the voltage n 5V steps. When the voltage was ncreased to 35V, one transstor n the amplfer shorted. Investgaton of the test system showed that ths short was ndrectly caused by the gate power supply, whch contans an auto-rangng feature for measurement of supply current. Ths auto-rangng selects a resstance for attanng optmum measurement senstvty. When the range resstance n the supply s too large or too small, the supply output voltage returns to zero, the range resstor s changed and the voltage s reappled. Ths process s completed n about 10(sec. When the amplfer dran voltage was ncreased to 40volts, the gate leakage current ncreased from <1mA to >1mA. Ths caused the gate power supply to auto-range the current measurement, settng the voltage to zero, and allowng the full dran current at 35 volts, dsspatng over 80 watts of power. Ths hgh current and voltage exceeds the devce desgn lmt. When the frst devce "fused" t absorbed nearly all the supply current and lmted voltages to less than 5volts. A second amplfer was assembled. Durng testng, the auto-rangng features of the power supples were dsabled to prevent the gate voltages that allow hgh dran current. Measured gan was below that seen wth the frst amplfer, and sgnfcant heatng was observed, although the test fxture remaned cool. After some prelmnary testng, contact problems at the gate lne connector agan led to hgh devce current levels, and the amplfer faled. A thrd amplfer s currently beng prepared for test. 3.2 ULTRA BROADBAND AIGaN AMPLIFIER DEVELOPMENT: Task 2: AIGaN MODFET Development (Contnuaton of task R2.2 from the frst program phase) Completon of the frst AIGaN MODFET devce lot, ncludng MODFET desgns wth undoped surface layers to Source Gate Undoped Surface Layer Dran Undoped AIGaN, 400 A Undoped GaN,.9 \m A1N Buffer, 1200 A Hgh Resstvty 6H-SC Substrate Fgure 3: AIGaN MODFETs were fabrcated wth undoped surface layers and passvaton, to evaluate the mpact of the AIGaN surface on pulse compresson evaluate ther mpact on pulse compresson, was mentoned n the last quarterly report, and results have been dscussed n the fnal report of the frst phase of ths program. A bref descrpton of the results obtaned for these devces s ncluded n the followng paragraph, for completeness and clarty. MODFET's wth Undoped Surface Layers Fgure 3 shows the MODFET structure wth an undoped GaN layer near the surface. Dry etchng was used to selectvely remove ths layer under the gate and n the source and dran contact areas. The l-v characterstcs of Page 3

4 ths structure under low and hgh voltage bas condtons are shown n fgure 4. Unlke conventonal MODFET's, these devces do not show reducton n dran current after hgh voltage stressng. The observed changes n devce behavor wth the ncluson of the undoped layer clearly suggest that the surface of GaN MODFET's s responsble for slow trappng effects. However, the present results do not show whether the undoped layer has elmnated the problem or whether t has just modfed the slow trappng tme constant so that t cannot be observed wth our measurement scheme. RF power measurements wll be requred to confrm ths. 70 ma 70V V a) Under low bas b) Under bas stress, and pulsed bas Fgure 4: Wth the use of the undoped surface layer and passvaton, no sgnfcant dc pulse compresson s observed Unfortunately RF power measurement s not possble wth the present devces whch have hgh contact resstance due to ncomplete removal of the undoped layer n the source, dran ohmc contacts. A new processng lot, descrbed n the next paragraph, s now under way. Ths lot also contans wafers wth the undoped layer, to follow-up on these promsng ntal results. New Processng Lot Durng the current reportng perod, processng was ntated on a second lot of AIGaN MODFET structures wth undoped layer near the surface for possble elmnaton of pulse compresson. Conventonal MODFET structures wthout the undoped layer, but wth a slcon ntrde layer as passvaton were also ncluded as a reference for comparson. In ths second program lot the contact resstance problem was resolved by etchng the undoped layer deeper n the contact areas, and annealng the T/AI/N/Au contact at a hgher temperature (900 C versus the 800 C used n the earler run). In devces wth slcon ntrde passvaton layer, reacton and spreadng of the T/AI/N/Au contact wth slcon ntrde was elmnated through the use of sntered T contacts. Followng ohmc contact formaton, both types of structure have been processed through Helum mplantaton for devce solaton. Unlke earler wafers, we observe a resdual leakage between solated areas n the present devces. C-V measurements of the samples ndcate that ths leakage s due to conducton through ether the hgh resstvty GaN buffer or at the AIN nucleaton layer nterface. The wafers are presently n the gate fabrcaton step and any adverse effect of the parastc leakage wll be evaluated followng gate deposton. 3.3 ULTRA BROADBAND AIGaN AMPLIFIER DEVELOPMENT: Task 3: MMIC Technology Development We have ntated work on a hgh etch rate dry process for the fabrcaton of through SC wafer vas. Ths work s done n cooperaton wth Prof. Steve Pearton of Unversty of Florda, Ganesvlle, n ther Plasmatherm Inductvely Coupled Plasma (ICP) etch system. The etch process uses NF 3 and oxygen as the etch gases for SC. A key requrement n etchng deep vas n SC s the dentfcaton of a mask materal that exhbts hgh etch selectvty over SC. To evaluate ther sutablty as a maskng medum, we have characterzed the etch rates of Al, N, Indum Tn Oxde (ITO), and SC for varous process condtons; selected results are shown n Fgure 5. From the results obtaned to date, N appears to be the most promsng mask materal, but ITO s expected to show mproved results wth further optmzaton of the deposton condtons. Under the present etch condtons, creatng a 4 ml va n SC wll requre about 7 mcron thck Nckel maskng layer, whch can be deposted usng an electroplatng process 3.4 PLANS FOR THE NEXT PROGRAM REPORTING PERIOD Page 4

5 Processng and testng of the current devce lot wll be contnued durng the next reportng perod. Devce modelng efforts, and MMIC process development wll also be contnued. Page 5

6 c 1 E <D re a: o + > UJ 10 < 2!^ n J;! A ITD t 1 / u.o / f A e tc h f g ^ - / *» s «- K ES :zw 2%M -* * P4.,-r- -" 0 0 X) *> u.o m T a e 0<?s rt ro r > u o o (0,., ~~r SC/AI SC/N SC/ITO T percent NF 3 percent NF 3 Fgure 5: Etch rate and selectvty for maskng materals used for dry etchng of SC. (NF3 etchng. Total flow NF 3 /0 2 = 15 seem, ICP = 750W, rf power = 250W, pressure = 2 mtorr 4. PROGRAM SCHEDULE AND FINANCIAL PERFORMANCE Two addtonal program fundng ncrements, totalng $160K, were receved n January. The revsed program fundng poston, and current program spendng, are shown n Fgure 6. The program schedule, and current status s shown n Fgure 7. The current MODFET devce lot s expected to be completed and tested durng the next quarter. Evaluaton of the ntal desgn of the 1-18GHz amplfer wll be ntated durng the next reportng perod, but detaled amplfer desgn wll be delayed pendng testng of the current devce lot. The frst program mlestones (delvery of ntal AIGaN MODFET structures grown usng the LEO technque) has been completed accordng to schedule; the materal s beng characterzed at Northrop Grumman. TasKI: Materals Development Materals Growth Characterzaton tavk^atovnl^offlf" " Development Xf MODFET Structure "'"" Modelng 'ZSTSODTET Fab/TesT Task 3: MMIC Technology Development fask'ü: Amp^er'Fabrcator Modelng/desgn Fabrcatonf"e5l task 5: Program ManagernCHJ BE '"low defect & standard structure delveres frst low defect densty materal delvery I.1. baselne structure j JL I : ; I ntal amp. desgn : quarterly techncal/fnancal reports I optmzed structure ^ modelng report, > 4W/mm power 18GHz 50 W concept desgn 1 Watt, 1-18GHZ I 3 Watts, 1-18GHz "lot 2 test completed work ^ completed mlestones Program Mlestones. 6 mos. Intal Low Defect Materal 12 mos. MODFET wth > 18GHz 14 mos. Intal Amplfer Desgn 24 mos. Modelng Report 26 mos. 1 Watt, 1-18 GHz Amplfer 36 mos. 3 Watt, 1-18 GHz Amplfer 36 mos. > 50 Watt, 1-18 GHz Desgn Concept Fgure 6: Program Schedule status, and program mlestones. Low defect densty AIGaN MODFET structure growth has been delayed by ntal delays n settng up the revsed program subcontract wth NCSU. Ths s not expected to adversely mpact program progress. Page 6

7 s c c E 2? Program Spendng Spend Rate Plan Blled to Navy ^^^ ^^* *^^ "^^* "^^* *^^ *^^ *^^* *^^* *^^ *^^* *^^* *^^* Fgure 7: Program spendng curve. An addtonal $160K fundng ncrement has been receved. Program spendng remans on track, but program efforts wll be reduced to reflect the reduced program spend rate plan. Page 7