2. Integrated SQUID/Sensors with Direct Coupling

• Integrating the SQUID with the sensor in a monolithic thin-film microdevice allows near-elimination of parasitic inductance in flux-transformer-coupled sensors, improving signal/noise

• Integrated SQUID/sensors also offer the possibility to eliminate the flux transformer altogether, further improving signal to noise

Development of Magnetic Microcalorimeters for Gamma-Ray Spectroscopy L.N. Le, R. Hummatov, S.T.P. Boyd, University of New Mexico

J.A. Hall, R.H. Cantor, STAR Cryoelectronics

LTD16 G2.11

4. Absorbers • New process

• Ion-milled copper layer for post definition •AZ125-nXT thick photoresist for absorbers

• Copper improves on SU-8 for post definition • Piranha strip of SU-8 damages underlying SQUID

microdevices and induces mechanical stress in absorbers • Copper can be stripped rapidly, with high selectivity, and

without damaging underlying SQUID microdevices • Self-aligning: Nb cap is a tight fit around absorber posts,

optimizing flux containment and field shaping.

9. Acknowledgements We gratefully acknowledge helpful discussions with the magnetic microcalorimeter groups at Brown, Heidelberg, GSFC, IBS and LLNL. Special thanks to Stephan Friedrich at LLNL for supplying the AZ125 photoresist and particle-testing guidance, and to Andreas Fleischmann at Heidelberg for allowing us to use their thermomagnetic properties data for Au:Er alloys. This work is funded by DOE, NNSA, and DTRA.

8. Summary/Next Steps • We have completed development of a fabrication

process for integrated SQUID/sensor magnetic microcalorimeter detectors with attached electroformed gold absorbers.

• Particle testing of new devices fabricated with this process is expected to begin soon.

1. Abstract We present a progress report on the development of our first complete magnetic microcalorimeter detectors. Transitioning to microdevices with integrated SQUIDs is a promising approach to improve performance. However, new challenges in microfabrication must be overcome, because the underlying SQUD devices are more sensitive to chemical attack and elevated processing temperatures.

3. Heat Flow • Thermal bus: normal-metal contact out to the ADR

Heat flow path. Left: as drawn in the layouts. The on-chip gold thermal bus surrounds the sensing coils and paramagnet, and is bounded on the left and right by pads for gold ribbon bonds. Right: as built and mounted. The gold ribbon bonds connect the thermal bus to thermal pads on the carrier, which in turn connect to the ADR.

On-Chip Thermal Bus

Go

ld-W

ire

Bo

nd

Pad

Go

ld-W

ire

Bo

nd

Pad

Paramagnet on sensing coils absorbers

• Pulse time constant •Using paramagnet to make the thermal link •Wiedemann-Franz for thermal R, using measured

paramagnet 4K resistivity •Heat capacity data courtesy of Heidelberg group

• Electroforming • Commercial solution: Technic Gold 25ES, Sulfite-based • Platinized titanium anode parallel at 1.7inch from cathode • Plating conditions: 55-60C, pH 7-8, agitated

Electroformed gold absorber test pieces in a square array with pitch 0.5mm. These absorbers are 20um thick, however gamma-ray absorbers with thickness > 100um should be well within the capabilities of this process: the AZ125 photoresist mold used here was 200um thick, but AZ125 up to 300um thick has been demonstrated. In spite of this very thick resist, the absorbers are well-defined and well-separated by spacings as small as 25um.

Ring of posts attaches to ring of paramagnet on underlying device

• Electroformed Gold Absorbers •Measured RRR=33 • Excellent adhesion: absorbers survive repeated LN2

plunges and ultrasonic cleaning with zero losses

5. Wiring and Insulation •Underlying SQUID devices are fabricated with a variant

of STARCryo’s commercial “Delta 1000” process •Desirable Modifications to Delta 1000 process:

• Thicker wiring, for magnetizing currents up to ~100 mA • Thinner insulation, improve signal/noise and filling factors

•A test wafer with 540 nm Nb/80 nm PECVD SiO2/446 nm Nb was fabricated with crossed meander, via string, and capacitor test structures:

• Chain of 100 4um diameter vias supports 288mA at 4K • 10um wide Nb traces in base layer support 290mA • 10um wide Nb traces with 624 crossovers support 150mA • 1mm x 1mm capacitor structures display no shorts

Nb Cap Paramagnet Sputter depo & liftoff paramagnet,

Sputter depo & liftoff Nb cap, Onto insulated microdevice sensing coil

Sputter depo Cu, Pattern for post molds

Ion mill post molds, Nb cap serves as mill stop,

RIE through Nb cap to paramagnet

Sputter depo blanket Au Seed Layer

Define absorber molds in AZ125, Electroform absorber

Remove AZ125, wet etch Cu

6. Fabrication • We have recently fabricated our first-generation of

complete magnetic microcalorimeter detectors

First SEM views of the first-generation complete detectors, including electroformed gold absorbers. Wiring details are obscured in SEM view where SiO2 insulating layer is the top layer. For this device we used the gold electroforming process to build up the on-chip thermal bus and the thermal pads, as well as to construct the absorbers. Electroformed gold thickness is 27 um.

On-Chip Thermal Bus

Go

ld-W

ire

Bo

nd

Pad

Go

ld-W

ire

Bo

nd

Pad

Absorbers 300 um x 300 um

• No test results yet, stay tuned.

Excitation Pads

Persistence Switches

SQUID Pads

Cooling Fins

Junction Area

2mm

Left: layout showing arrangement of sensing coils, paramagnet, Nb cap under the absorber. Right: perspective views showing that the electroformed absorbers stand above the underlying SQUID microdevice. Support posts for the absorbers attach directly to the paramagnet through close-fitting holes in the niobium cap structures.

7. Particle Testing Setup • We have developed a basic 50mK particle-testing

setup and have performed initial checkout using early prototype non-absorber devices with 55Fe source.

Left: paramagnet and sensing coils yielding pulses with time constant τ=3.9 ms @ 75mK. Right: paramagnet and sensing coils yielding pulses with τ=0.16 ms @ 75 mK. The paramagnet volume on the right is about 2.2X that on the left, confirming that the 3 thermal contact tabs extending from the paramagnet to the on-chip thermal bus have a substantial thermal conductance, even though the tab contact to the thermal bus in this early example is very slight.

Proof-of-concept pulse data is of very low quality but suffices to confirm particle flux and data acquisition functionality. This data was obtained with Au:Er 1% test device made for thermometry, without absorbers or collimators in place.

Heat capacity deduced from peak heights agrees roughly with expectations.

200u 100u

13.4μΦ0/√Hz

Excellent noise performance of direct-coupled integrated SQUID/sensor devices. Insets show the sensing coils for each device, which form almost the entirety of the loop inductance of the SQUIDS. In spite of complex topologies, and not following usual design rules for low noise SQUIDs, typical noise for “bare” devices is ~3μφ0/√Hz at 4K, comparable to or exceeding the STARCryo commercial SQUIDs.

Thermal links. The right side has a single thermal link, the left side has two links, each with twice the number of squares, to check for impact on pulse shape variation. To accommodate the extra length required for the dual thermal links, the links run on top of the SQUID structures.

Dual thermal links Equivalent single thermal link

On-Chip Thermal Bus

Go

ld-W

ire

Bo

nd

Pad

Go

ld-W

ire

Bo

nd

Pad

Paramagnet ring on spiral coil

1000pH SQUID/sensor w/simple rectangular meander, Ic=12μA x 2 junctions. Compare this direct-coupled noise value to an estimated 21 μΦ0/√Hz at the sensing coil for a good-performing SQUID with optimal flux transformer coupling.

Left: SEM detail of a crossed meander structure, showing location of trench created by focused ion beam (FIB). Right: detail of step-edge coverage revealed by FIB trench. The thin oxide layer can just be made out.