Supercond. Sci. Technol. 9 (1996) A96–A99.

Printed in the UK

High-Tc SQUID magnetometers for

biomagnetic measurements
B David†, D Grundler‡, S Krey‡, V Doormann†, R Eckart†, J P
Krumme†, G Rabe†and O Doessel†
†Philips Research Laboratories, Department Technical Systems Hamburg,
P O Box 63 05 65, D-22315 Hamburg, Germany
‡Institute of Applied Physics, University of Hamburg, Jungiusstr. 11,
D-20355 Hamburg, Germany

Received 18 October 1995, in final form 28 November 1995

Abstract. We have designed and fabricated three types of high-Tc SQUID

(superconducting quantum interference device) magnetometers based on
step-edge Josephson junctions using three different concepts of coupling magnetic
flux into the SQUID: (i) a single pickup loop galvanically coupled to the SQUID, (ii)
a flux transformer inductively coupled to the SQUID and (iii) a multiloop pickup loop
used directly as the SQUID inductance. On a 10 × 10 mm2 substrate we achieved
an effective flux capture area of 1.70 mm2 and 1.67 mm2 for the inductively
coupled and multiloop devices, respectively. Due to the low white noise levels of
44 fT Hz−1/2 for the inductively coupled magnetometer and 30 fT Hz−1/2 for the
multiloop device high quality magnetocardiograms were recorded inside a
magnetically shielded room without signal averaging.

1. Introduction to the SQUID. Another approach was introduced by Koelle

et al [2] who galvanically coupled a dc SQUID to a loop
Typical peak-to-peak signals in magnetocardiography patterned in the same YBCO film to realize a directly
(MCG) are in the range of several 10 pT. However, coupled magnetometer. Lee et al [3] reported an improved
physicians are more interested in the smaller signals, directly coupled magnetometer. A more efficient way to
e.g. the onset of an extrasystole, where the mean signal enhance the field sensitivity of a SQUID is to inductively
amplitude is on the order of 2 pT. The measurement couple it to a multi-turn flux transformer [4] either by
bandwidth must be at least 1 to 100 Hz. For adequate flip-chip (Grundler et al [5] and Ludwig et al [6]) or
source localization a signal to noise ratio of 5 is necessary. monolithically integrated (Kromann et al [7] and David et
With
√ this data in mind, the desired magnetic field noise al [8]) techniques. An alternative approach is the multiloop
S B of future MCG systems can be determined to less magnetometer, successfully demonstrated by Drung and
than 50 fT Hz−1/2 . High-Tc SQUID magnetometers made co-workers in low-Tc devices [9], and recently made by
of YBa2Cu3O7 (YBCO) are now able to reach this noise Ludwig et al [10] from the high-Tc material YBCO.
level in a shielded environment. Here we report on the design, fabrication and
For a magnetometer, √ the relevant√ figure of merit is characterization of three different types of magnetometers
the
√ magnetic field noise S B (f) = S 8 (f)/Aeff . Here, all based on the same step-edge junction process: (i)
S 8 (f) is the rms flux noise and Aeff is the effective directly coupled magnetometer, (ii) inductively coupled
flux capture area of the magnetometer, which may also be (Ketchen type) magnetometer and (iii) multiloop (Drung
expressed in terms of the field-to-flux conversion efficiency type) magnetometer. We compare the effective area Aeff
B8 = 80 /Aeff . Thus, the magnetic field noise of a √
and the field resolution S B of these magnetometers
magnetometer can be optimized by minimizing the flux fabricated on 10 × 10 mm2 substrates and demonstrate their
noise and maximizing the effective area. suitability for magnetocardiography.
Due to their low effective flux capture area As , bare
SQUIDs have an insufficient magnetic field resolution. To
enhance the sensitivity of a SQUID, one has to increase its 2. Experiment
effective area whilst keeping its inductance low to ensure
low noise operation. This has been accomplished in a The inductively coupled and multiloop magnetometers are
variety of ways. For example, Zhang et al [1] increased both fabricated from a YBCO–STO–YBCO multilayer on
the area of their rf SQUID by taking into account the 10 × 10 mm2 SrTiO3 (STO) substrates. The multilayer
flux focusing effect of their large size square washer or by fabrication process is described in detail elsewhere [11].
coupling a single-layer thin-film flip-chip flux transformer In summary; we deposit the YBCO and STO films by

0953-2048/96/SA0096+04$19.50
c 1996 IOP Publishing Ltd
 SQUID magnetometers for biomagnetic measurements

-3
10 -1 1
10

flux noise √S Φ ( Φ0 / √H z)

field no ise √S B (T / √H z)
__

__
-4 FLL
10 -1 2
10

__
__
-5
10 -1 3
F L L + ac-bias 10

-6
10 -1 4
10
1 10 100 1 0 00
frequency (H z)
Figure 1. Noise spectra of a directly coupled magnetometer measured at 77 K inside a superconducting BSCCO tube using
standard FLL electronics. Additional ac-bias modulation assists the reduction of 1/f -noise effectively.

rf magnetron sputtering and pattern them by standard 3. Results

photolithography and Ar-ion beam etching. To obtain
bevelled edges on the lower YBCO film and STO insulating 3.1. Directly coupled magnetometer
layer we bake the photoresist after development and mill
The effective flux capture area of the directly coupled
the films at a 60◦ angle of incidence to the rotating
magnetometer is given by
substrate. The steep slope required for the step-edge
 
junctions, however, is milled at normal incidence into the Ls

STO insulating layer. The two junctions with a nominal Aeff = As + Ap ≈ Ls /Lp Ap (1)
Ls + Lp
width of 3 µm are patterned from the top YBCO film.
The thickness of the lower YBCO and STO layers are where As is the effective area of the SQUID loop with
150 nm and 200 nm, respectively. The thickness of the inductance Ls and Ap is the effective area of the pickup
upper YBCO layers which form the junctions, equals the loop with inductance Lp . In the course of our experiments
step height of about 200 nm. The contact pads are covered we fabricated magnetometers which differed in the shape of
by a 30 nm Ag layer which is sputter deposited through a the pickup loop and the size and inductance of the SQUID
shadow mask. loop. For one device with a pickup loop area of 50 mm2
The fabrication process for our directly coupled and a pickup loop inductance of 15 nH coupled to a 55 pH
magnetometer is less complicated and starts with the SQUID inductance, we measured a field-to-flux conversion
patterning of a steep step in the STO substrate and the efficiency B8 of 15.1 nT 8−10 , corresponding to an effective
deposition of a single YBCO film. Finally, the pickup loop area Aeff of 0.14 mm2 .
and the SQUID loop are patterned by the same process The discrepancy between calculated and measured
described above. effective areas is due to the uncertainties in the SQUID
The electrical characterization of the devices is and pickup loop inductances. The rms flux noise of
performed either: (a) inside a superconducting BSCCO this particular device is 8.5 µ80 Hz−1/2 at 1 kHz and
tube (MCP-BSCCO 2212 tube, Hoechst AG, Frankfurt) 18.7 µ80 Hz−1/2 at 1 Hz measured in a superconducting
immersed in liquid nitrogen, (b) a double µ-metal shield BSCCO tube using flux-locked loop (FLL) electronics
inside a He-flow-type cryostat or (c) inside a low noise with additional ac-bias current modulation. The resulting
biomagnetic dewar filled with liquid nitrogen inside the magnetic field noise of the magnetometer, depicted in figure
Berlin magnetically shielded room (BMSR). To compare 1, is 128 fT Hz−1/2 at 1 kHz and 280 fT Hz−1/2 at 1 Hz.
the different measurement environments, we investigated
the multiloop device for each of these procedures. We 3.2. Inductively coupled (Ketchen type) magnetometer
found different values for the field-to-flux conversion
efficiency B8 when we compared procedure (a) and (b) The effective flux capture area of an inductively
with procedure (c). This is attributed to the distortion of coupled magnetometer is given by
the magnetic field lines in our small copper coils generating s
the magnetic field inside the narrow BSCCO tube or µ- Ls
Aeff = As + 12 k Ap (2)
metal tube. We therefore multiplied our results of B8 Lp
measurement for procedures (a) and (b) by a factor of 1.22
using the measured data of the multiloop device inside the assuming the inductance Li of the input coil of the flux
BMSR where the magnetic field is applied by a calibrated transformer to be equal to the inductance Lp of the pickup
Helmholtz coil. loop. Our magnetometer chip contains two 10 12 turn input

A97
B David et al

-2 -1 1
10 10

flu x noise √S Φ [ Φ0 / √H z]

field noise √S B [T / √H z]
__

__
F L L + ac-bia s
-3 -1 2
10 10

__
__
-4 -1 3
10 10

-5
10
1 10 100 10 0 0
freque ncy [H z]
Figure 2. Noise spectrum of an inductively coupled (Ketchen type) magnetometer using FLL electronics. The spectrum,
measured at 77 K in a He-flow-type cryostat includes a lot of environmental noise which affected the SQUID inside the
double µ-metal shield.

-2 -1 1
10
flux noise √S Φ ( Φ0 / √H z)

10

field noise √S B (T / √H z)
__

__
-3 -1 2
10

__
APF+ 10
__

static bias

-4 -1 3
10 10
APF+
b ia s rev ersa l

-5
10
00.1 1 10 100 1 00 0 10 0 0 0
frequency (H z)
Figure 3. Noise spectra of a multiloop (Drung type) magnetometer at 77 K, measured in the Berlin magnetically shielded
room using additional positive feedback (APF) electronics with static bias and bias reversal.

coils and SQUIDs which are shunted and connected to the in FLL with ac-bias current modulation is one of the lowest
same square pickup loop. The better SQUID is activated yet reported for this type of high-Tc device.
by scribing the respective shunt. The pickup loop has an
outer diameter of 8.8 mm and a width of 0.6 mm with two 3.3. Multiloop (Drung-type) magnetometer
indentations for the SQUIDs and the bond pads, giving
a practical pickup area of Ap ≈ 54 mm2 . Using Lp = The effective area of a multiloop magnetometer is given by
15 nH and a SQUID inductance Ls of 85 pH we estimate [12]
an effective flux capture area of 1.71 mm2 , taking into Ap
Aeff = − Asp (3)
account a coupling coefficient k of 0.84 derived from the N
measurement of the mutual inductance (at 4.2 K) between where Ap is the total effective area of the large outer
the input coil and the SQUID washer. The measured loop, Asp the parasitic effective area of one spoke, and
field-to-flux conversion efficiency B8 is 1.22 nT 8−1 0 , N is the number of loops. Our device (type WH1 in
corresponding to an effective area Aeff ≈ 1.70 mm2 . This [12]) consists of 16 equal loops forming a cartwheel-like
is a surprisingly close conformity between the estimated device with an outer diameter of 7 mm. The effective flux
and measured effective flux capture areas. capture area is estimated to be 1.77 mm2 , the measured
In figure 2 the noise spectrum of an integrated value is 1.67 mm2 . Noise measurements were performed in
inductively coupled magnetometer is shown, measured at the Berlin magnetically shielded room using the additional
77 K inside a double µ-metal shield. The field noise of positive feedback (APF) electronics with bias reversal [13].
44 fT Hz−1/2 at 1 kHz and 260 fT Hz−1/2 at 1 Hz measured The noise spectra of the multiloop magnetometer, measured

A98
 SQUID magnetometers for biomagnetic measurements

1 00

80

am plitude (pT )
60

40

20

-20
0 1 2 3 4
tim e (s)
Figure 4. Real-time trace of a magnetocardiogram recorded with the multiloop magnetometer inside the Berlin magnetically
shielded room. The measurement bandwidth is 0.016–200 Hz.

with static bias current and with bias reversal, are depicted Acknowledgments
in figure 3. A very low white noise level of 30 fT Hz−1/2
was obtained. With bias reversal, the low frequency noise The authors thank Dietmar Drung for measuring the
is greatly reduced to 95 fT Hz−1/2 at 1 Hz. noise spectra of the multiloop magnetometer with
his APF electronics equipment and for recording the
magnetocardiogram. Gratefully acknowledged are helpful
4. Discussion discussions with Silvia Knappe and Hans Koch of the
PTB Berlin, and Meinhard Schilling and Dirk Reimer of
We have fabricated three types of magnetometers.
the University of Hamburg. This work was supported by
Although there is substantial inductance mismatch, a
the German Bundesministerium für Bildung und Forschung
reasonable enhancement of the effective area can be
BMBF under contract number 13N5709.
achieved using the directly coupled magnetometer. The real
advantage, as compared with an integrated magnetometer,
is that the complete device can be fabricated from a single
layer of YBCO. However, the effective flux capture area References
of the inductively coupled and multiloop magnetometer
is about one order of magnitude larger compared to [1] Zhang Y, Krüger U, Kutzner R, Wördenweber R,
Schubert J, Zander W, Sodtke E and Braginski A I 1994
the directly coupled magnetometer, if fabricated on a Appl. Phys. Lett. 65 3380
substrate of 10 × 10 mm2 . This gain in field-to- [2] Koelle D, Miklich A H, Ludwig F, Dantsker E,
flux conversion efficiency B8 is at the expense of a Nemeth D T and Clarke J 1993 Appl. Phys. Lett. 63 2271
more complicated fabrication technology. The inductively [3] Lee L P, Longo J, Vinetskiy V and Cantor R 1995 Appl.
Phys. Lett. 66 1539
coupled magnetometer can be scaled up to larger substrates
[4] Jaycox J M and Ketchen M B 1981 IEEE Trans. Magn.
by increasing the number of turns of the input coil. The MAG-17 400
realized multiloop magnetometer is a compromise between [5] Grundler D, David B, Eckart R and Doessel O 1993 Appl.
the largest acceptable SQUID inductance of about 145 pH Phys. Lett. 63 2700
and a significant enhancement of the effective area. Our [6] Ludwig F, Dantsker E, Nemeth D T, Koelle D,
Miklich A H, Clarke J, Knappe S, Koch H and
results demonstrate that the various coupling schemes are
Thomson R E 1994 Supercond. Sci. Technol. 7 273
well understood and effectively enhance the field-to-flux [7] Kromann R, Kingston J J, Miklich A H, Sagdahl L T, and
conversion efficiency. An even better noise performance Clarke J, 1993 Appl. Phys. Lett. 63 559
can be expected by further improving the flux noise of the [8] David B, Grundler D, Krumme J P, and Doessel O, 1995
SQUID. Nevertheless, the magnetic field noise of all three IEEE Trans. Appl. Superconductivity 5 2935
[9] Drung D, Cantor R, Peters M, Scheer H J and Koch H
types of magnetometers is already low enough to record 1990 Appl. Phys. Lett. 57 406
MCGs. As an example, a magnetocardiogram is shown in [10] Ludwig F, Dantsker E, Kleiner R, Koelle D, Clarke J,
figure 4, recorded with a multiloop magnetometer inside the Knappe S, Drung D, Koch H, Alford N M and
Berlin magnetically shielded room. Due to the low SQUID Button T W 1995 Appl. Phys. Lett. 66 1418
noise level (∼= 3 pT peak-to-peak in 200 Hz bandwidth) [11] David B, Grundler D, Eckart R, Fanghänel K,
Krumme J P, Doormann V and Doessel O 1994
a good signal to noise ratio was obtained even without Supercond. Sci. Technol. 7 287
signal averaging. A crucial task for the future will be the [12] Drung D, Knappe S and Koch H 1995 J. Appl. Phys. 77
development of magnetometers working in an unshielded 4088
environment. [13] Drung D 1995 Appl. Phys. Lett. 67 1474

A99