Articleshttps://doi.org/10.1038/s41587-019-0321-x

A laser-engraved wearable sensor for sensitive detection of uric acid and tyrosine in sweatYiran Yang   1,7, Yu Song   1,2,7, Xiangjie Bo1,7, Jihong Min1, On Shun Pak3, Lailai Zhu4, Minqiang Wang1, Jiaobing Tu1, Adam Kogan1, Haixia Zhang2, Tzung K. Hsiai5, Zhaoping Li6 and Wei Gao   1*

1Andrew and Peggy Cherng Department of Medical Engineering, California Institute of Technology, Pasadena, CA, USA. 2National Key Lab of Micro/Nano Fabrication Technology, Peking University, Beijing, China. 3Department of Mechanical Engineering, Santa Clara University, Santa Clara, CA, USA. 4Department of Mechanical and Aerospace Engineering, Princeton University, Princeton, NJ, USA. 5Division of Cardiology, Department of Medicine, David Geffen School of Medicine, University of California, Los Angeles, CA, USA. 6Division of Clinical Nutrition, Department of Medicine, David Geffen School of Medicine, University of California, Los Angeles, CA, USA. 7These authors contributed equally: Yiran Yang, Yu Song, Xiangjie Bo. *e-mail: weigao@caltech.edu

SUPPLEMENTARY INFORMATION

In the format provided by the authors and unedited.

NATure BiOTecHNOLOGY | www.nature.com/naturebiotechnology

Supplementary Fig. 1. Fabrication process of the laser-enabled patch.

Medical tape Laser-cut microfluidics PET attachment Another medical tape

PI cleaning Vector mode laser cutting Raster mode laser cutting Multimodal sensing patch

PI Chemical sensor Strain sensor Temperature sensorMedical tape PET

Supplementary Fig. 2. Vector and raster modes of the laser cutting process. a, A CO2 laser

cutting machine. b, Optical image of laser-scribed graphene patterns via vector mode and raster

mode, respectively. Scale bar, 1 cm. c,d, SEM images of vector mode-scribed graphene structure.

Scale bars represent 5 µm (c) and 2 µm (d), respectively. e,f, SEM images of raster mode-scribed

graphene structure. Scale bars represent 5 µm (e) and 2 µm (f), respectively.

a c

b d

e

f

Supplementary Fig. 3. Microscopic images showing the resolution of the laser engraving. a–

d, a graphene microelectrode fabricated by the raster mode (a), a graphene micropattern fabricated

by vector mode (b), and a microfluidic channel fabricated by vector mode under top view (c) and

cross-sectional view (d). Scale bars, 100 µm.

a b

dc

Supplementary Fig. 4. Low-cost and mass-production of the flexible multimodal sensor

arrays. Scale bar, 1 cm.

a b

Supplementary Fig. 5. Characterization of the LEG-based chemical, temperature, and strain

sensors using Raman spectroscopy. Three experiments were performed independently with

similar results.

800 1400 2000 2600 3200

Inte

nsity

(a.u

.)

Strain sensor Temperature sensor Chemical sensor

D G2D

Raman shift (cm-1)

Supplementary Fig. 6. Characterization of the LEG-based chemical, temperature, and strain

sensors using X-ray photoelectron spectroscopy (XPS). Three experiments were performed

independently with similar results.

0 200 400 600 800 1000

Strain sensor

Temperature sensor

O 1sN 1s

C 1sIn

tens

ity (a

.u.)

Binding energy (eV)

Chemical sensor

280 284 288 292

Inte

nsity

(a.u

.)

Binding energy (eV)

C 1s

Chemical sensor

Strain sensor

Temperature sensor

526 532 538 544

Strain sensor

Temperature sensor

Chemical sensorInte

nsity

(a.u

.)

O 1s

Binding energy (eV)

cba

Supplementary Fig. 7. Circuit diagram of the multiplexed sensing system. a, An FPCB next

to a quarter dollar coin. b, Circuit diagram of the voltage dividers for driving physical sensor

measurements. c, Circuit diagram of analog front-end potentiostat for voltammetric sensing. VR

and VW correspond to the reference potential and the working potential, respectively.

a b

ADC 2

VDD

Strainsensor

30 kΩ

ADC 3

VDD

Temperaturesensor

30 kΩ

cVoltage dividers

.

DAC VR

C W

R

DAC VW

ADC 1

Fourth order low pass filter Potentiostat interface Transimpedance amplifier

1.2 kΩ 16.2 kΩ

0.1 μF

10 nF

845 Ω 29.4 kΩ

0.1 μF

10 nF

0.1 μF

20 kΩ

Supplementary Fig. 8. Optimization of the LEG-based chemical sensor under different laser

parameters. Condition, 50 µM UA and 50 µM Tyr. Five experiments were performed

independently with similar results.

0.2 0.4 0.6 0.8C

urre

nt (m

A cm

-2)

Potential (V)

0.5 mA cm-2

Electrode #1

Electrode #12

Supplementary Fig. 9. Electrochemical impedance spectroscopy (EIS) for LEG-based

electrodes prepared with different laser-engraving parameters. EIS tests were performed in a

solution containing 0.2 M KCl and 5 mM [Fe(CN)6]3- at open circuit potentials with an AC

amplitude of 5 mV in the range of 0.1–1000000 Hz. All conditions were tested for once at the

same time.

0 600 1200 18000

600

1200

1800

ΩΩ

No.1 No.2 No.3 No.5 No.6 No.11

Supplementary Fig. 10. Reproducibility of the LEG-based chemical sensors for Tyr and UA

sensing. a, Continuous successive detection of UA and Tyr using an LEG-CS for 16 cycles. I and

I0 represent the peak amplitudes of the DPV plot obtained from a give scan cycle and the first cycle,

respectively. b, Batch to batch variation of the LEG-based chemical sensor performance (10

batches). I and I0 represent the peak amplitude of the DPV plot, and the average peak amplitude

obtained from the first batch, respectively. Measure of the centre is the mean value of each batch.

Error bars represent the standard deviations of the peak amplitudes measured from 16 sensors.

Conditions: 50 and 100 µM of UA and Tyr.

1 2 3 4 5 6 7 8 9 100

20

40

60

80

100

120

UA Tyr

I/I0 (

%)

BatchScan cycles1 4 7 10 13 16

0

20

40

60

80

100

120

UA Tyr

I/I0 (

%)

ba

Supplementary Fig. 11. The selectivity of the LEG-CS to UA over other analytes in human

sweat. I and I0 represent the peak amplitude of the DPV plot of an LEG-CS, and the average peak

amplitude obtained from the initial UA solution, respectively. The concentrations of the initial UA

and other analytes are based on Supplementary Table 2. UA concentration was increased by 50%

in the end. Error bars represent the standard deviations of 4 independent measurements; measure

of the centre is the mean of the 4 independent measurements.

0

50

100

150

200I/I

0 (%

)

Sample

+ Tyro

sine

+ Uric

acid

+ Vali

ne

+ Tryp

topha

n

+ Thre

onine

+ Tha

imine

+ Seri

ne

+ Ribo

flavin

+ Pyri

doxin

e

+ Phe

nylal

anine

+ Pan

tothe

tic ac

id

+ Orni

thinin

e

+ Nico

tinic

acid

+ Meth

ionine

+ Lys

ine

+ Leu

cine

+ Iso

leucin

e

+ Ino

sitol

+ Hist

idine

+ Glyc

ine

+ Glut

amic

acid

+ Foli

c acid

+ Deh

ydroa

scorb

ic ac

id

+ Crea

tinine

+ Crea

tine

+ Citru

lline

+ Cho

line c

hlorid

e

+ Asp

artic

acid

+ Asc

orbic

acid

+ Argi

nine

+ Alan

ine

Uric ai

cd

Supplementary Fig. 12. The selectivity of the LEG-CS to Tyr over other analytes in human

sweat. I and I0 represent the peak amplitude of the DPV plot of an LEG-CS, and the average peak

amplitude obtained from the initial Tyr solution, respectively. The concentrations of the initial Tyr

and other analytes are based on Supplementary Table 2. Tyr concentration was increased by 50%

in the end. Error bars represent the standard deviations of four measurements. Error bars represent

the standard deviations of 4 independent measurements; measure of the centre is the mean of the

4 independent measurements.

0

50

100

150

200I/I

0 (%

)

Sample

+ Uric

acid

+ Tyro

sine

+ Vali

ne

+ Tryp

topha

n

+ Thre

onine

+ Tha

imine

+ Seri

ne

+ Ribo

flavin

+ Pyri

doxin

e

+ Phe

nylal

anine

+ Pan

tothe

tic ac

id

+ Orni

thinin

e

+ Nico

tinic

acid

+ Meth

ionine

+ Lys

ine

+ Leu

cine

+ Iso

leucin

e

+ Ino

sitol

+ Hist

idine

+ Glyc

ine

+ Glut

amic

acid

+ Foli

c acid

+ Deh

ydroa

scorb

ic ac

id

+ Crea

tinine

+ Crea

tine

+ Citru

lline

+ Cho

line c

hlorid

e

+ Asp

artic

acid

+ Asc

orbic

acid

+ Argi

nine

+ Alan

ine

Tyrosin

e

Supplementary Fig. 13. The selectivity of the LEG-CS to Tyr over tryptophan. The DPV plots

were obtained from solutions containing physiologically relevant levels of Tyr (170 µM) and

tryptophan (20, 40 and 60 µM). The experiment was repeated 5 times independently with similar

results.

0.5 0.6 0.7 0.8 0.960

65

70

75

80

μ

Tyr μ μ μ

Supplementary Fig. 14. The dependence of sensor response on the solution pH levels. a,b, The

DPV responses (a) and the corresponding peak amplitudes (b) of the LEG-based chemical

sensors (3 mm in diameter) for UA sensing in solutions with different pHs. c,d, The DPV

responses (c) and the corresponding peak amplitudes (d) of the LEG-based chemical sensors for

Tyr sensing in solutions with different pHs. To prepare the artificial sweat samples with different

pH, lactic acid was used to adjust the pH of the PBS buffer (0.3X) solutions to physiologically

relevant pHs. The UA and Tyr concentrations are 100 and 100 µM, respectively. Throughout the

figure, five independent experiments were performed with similar results, and the mean values

were taken as measure of the centre. Error bars represent the standard deviations of measurements

from 5 sensors.

0.5 0.6 0.7 0.8 0.93

5

7

9

11 pH 4.2 pH 4.5 pH 4.9 pH 5.2 pH 5.6

Potential (V)

μ

0.2 0.3 0.4 0.50

5

10

15

20 pH 4.2 pH 4.5 pH 4.9 pH 5.2 pH 5.6

Potential (V)

μ

pH4.1 4.5 4.9 5.3 5.7

0

5

10

15

20

Cur

rent

(μ)

pH4.1 4.5 4.9 5.3 5.7

0

2

4

6

8

Cur

rent

(μ)

ba

c d

Tyr Tyr

Supplementary Fig. 15. The selectivity of the LEG-CS to lactate. Physiologically relevant

levels of lactic acid (10, 20, 30, and 40 mM) were added to 0.1 M ABS (pH 4.6) containing 50 µM

UA and 50 µM Tyr. The experiment was repeated 5 times independently with similar results.

0.1 0.3 0.5 0.7 0.930

35

40

45

50

Cur

rent

(μ)

Potential (V)

0 10 30 400

4

6

8

k(μ

)

ba

Supplementary Fig. 16. Characterization and calibration of LEG-CS using the FPCB. a,b,

DPV curves in ABS solutions containing varied UA concentrations (a) their corresponding

calibration curve (b). c,d, DPV curves of Tyr in ABS solutions containing varied Tyr

concentrations (c) and their corresponding calibration curve (d). The experiment was repeated 5

times independently with similar results.

0.35 0.40 0.45 0.50240

270

300

330

Pote

ntia

l (m

V)

Potential (V)

μμ

μ μ

μ

0.6 0.7 0.8 0.9 1.0250

275

300

325

350

Pote

ntia

l (m

V)

Potential (V)

μ μ μ

μ

0 10 20 30 400

24

48

72

Peak

am

plitu

de (m

V)

μ

0 100 200 300 400 5000

30

60

90

Peak

am

plitu

de (m

V)

μ

ba

dc

Supplementary Fig. 17. Control of the DPV scan cycle lengths by varying the scan steps.

Varied incremental potentials (4, 8, and 16 mV) were used to achieve the scan cycle lengths of 90,

45, 23 seconds. Conditions, 50 µM UA and 250 µM Tyr. The experiment was repeated 5 times

independently with similar results.

0.2 0.4 0.6 0.8100

200

300

400

500

Pote

ntia

l (m

V)

Potential (V)

90 s 45 s 23 s

Supplementary Fig. 18. Sensor performance for detection of ultra-low levels of other

electroactive compounds. a,b, the LEG electrodes show excellent sensitivity to AA (40, 80, and

120 µM) (a) and DA (0.5, 1, and 2 µM) (b). A commercial Ag/AgCl was used as the reference

electrode. The experiment was repeated 5 times independently with similar results.

-0.2 -0.1 0.0 0.1 0.20.64

0.66

0.68

0.70

0.72

0.74

Cur

rent

(mA

cm-2

)

Potential (V)

Blank μ μ μ

0.1 0.2 0.3 0.4 0.50.48

0.52

0.56

0.60

0.64

0.68

Cur

rent

(mA

cm-2

)

Potential (V)

Blank μ μμ

ba

Supplementary Fig. 19. Stability of the LEG-CS during the bending tests. a,b, The long term

stability of an LEG-based sensor during the bending test. 0–0.25 % strains were applied for both

x (a) and y (b) axis. I and I0 represent the current amplitude of the DPV plot during the bending,

and the average current amplitude without bending, respectively. The error bars represent the

standard deviation of 10 measurements, the mean of the 10 measurements is measure of the centre.

Conditions, 50 µM UA and 100 µM Tyr.

0.00 0.05 0.10 0.15 0.20 0.250

20

40

60

80

100

120

I/I0 (

%)

Strain (%)

UA Tyr

0.00 0.05 0.10 0.15 0.20 0.250

20

40

60

80

100

120

UA Tyr

I/I0 (

%)

Strain (%)

ba

x

y

x

y

Supplementary Fig. 20. Porous fiber-like graphene structure for strain sensing. a, Cross-

section SEM image of the fiber-like LEG. Scale bar, 100 µm. b, Magnified view of the fiber-like

LEG structure. Scale bar, 10 µm.

a b

Supplementary Fig. 21. Characterization of the LEG-based temperature sensors. a, Batch to

batch variation of the LEG-based temperature sensor performance (10 batches). Error bars

represent the standard deviations of 10 measurements from a temperature sensor; the mean of the

10 measurements is measure of the centre. b, The response of an LEG-based temperature sensor

to small temperature changes. The experiment was repeated 5 times independently with similar

results.

0 30 60 90 120 15098.44

98.46

98.48

98.50

98.52

kΩ

33.6 ºC

33.8 ºC

34 ºC

34.2 ºC

34.4 ºC

Temperature1 2 3 4 5 6 7 8 9 10

-0.08

-0.07

-0.06

-0.05

Sens

itivi

ty (%

°C-1

Batch

ba

Supplementary Fig. 22. Batch to batch variation of the LEG-based strain sensors prepared

with varied laser scanning speeds. Error bars represent the standard deviations of 10

measurements from a strain sensor under 0.25% strain; the mean of the 10 measurements is the

center of measure.

1 2 3 4 5 6 7 8 9 10

-10

-5

0

Batch

Supplementary Fig. 23. Heart rate (HR) monitoring using an LEG-based strain sensor. a,b,

The resistive response of an LEG-based strain sensor (a) and extracted heart rate information (b)

during long term continuous pulse monitoring on a healthy subject. c,d, The resistive responses of

the LEG-based strain sensor at 1–3 minutes and at 15–17 minutes during this test.

1.0 1.5 2.0 2.5 3.031.7

31.9

32.1

Time (min)

kΩkΩ

15.0 15.5 16.0 16.5 17.031.4

31.6

31.8

kΩ

Time (min)

0 5 10 15 2031.4

31.6

31.8

32.0

32.2kΩ

Time (min)0 5 10 15 20

55

60

65

70

75

Hea

rt ra

te (b

pm)

Time (min)

a

c d

b

Supplementary Fig. 24. On-body sensor validation using an FDA-approved vital sign

monitor during one session. a, Validation of an LEG-PS for heart rate monitoring using an

FDA-approved heart rate monitor. b, Validation of an LEG-PS for respiration rate monitoring

using an FDA-approved respiration rate monitor.

0 200 400 600 800 1000 120050

60

70

80

90

Hea

rt ra

te (b

pm)

Time (s)

Reference Strain sensor

0 200 400 600 800 1000 120010

15

20R

espi

ratio

n ra

te (b

pm)

Time (s)

Reference Strain sensor

a

b

Supplementary Fig. 25. The long term stability of the LEG-based strain sensor over one

bending session. a, The measurement was performed during a bending test of a strain sensor

(10,000 cycles with strains alternated between 0% and 0.1%). b,c, The resistive sensor response

during 910–915 cycles (b) and 9610–9615 cycles (c) of bending.

9610 9611 9612 9613 9614 961530

31

32

33

kΩCycle

910 911 912 913 914 91530

31

32

33

kΩ

Cycle

a

cb

0 2000 4000 6000 8000 1000030

31

32

33

kΩ

Cycle

Supplementary Fig. 26. Characterization and calibration of the temperature and strain

sensors using the FPCB. a, Calibration curve for the temperature sensor. b,c, Characterization in

response to pulsed strains (b) and the calibration plot of the strain sensor (c). Throughout the figure,

the experiments were repeated 3 times independently and similar results were obtained.

0.00 0.05 0.10 0.15 0.20 0.25-10

-8

-6

-4

-2

0

0 15 30 45 6029

30

31

32

33

kΩ

25 30 35 40 45 50-1.6

-1.2

-0.8

-0.4

0.0

°C

ba c

Supplementary Fig. 27. The numerical simulation showing the fluidic dynamics in the

reservoir of the lab-on-skin sensor patch. The dimension used here are based on actual sensor

design used in this work. Scale bar, 5 mm.

30 s 60 s 120 s 180 s 240 s

80

60

40

20

0

Concentration ( M)

Outlet

Supplementary Fig. 28. Inferred sweat rates from different locations of eight healthy subjects

obtained via optical image analysis. Sub., subject.

0.0

0.5

1.0

1.5

2.0

Sub. 7Sub. 6Sub. 5Sub. 4Sub. 3Sub. 1 Sub. 2 Sub. 8Fl

ow ra

te (μ

-1)

Forehead Neck

Supplementary Fig. 29. Characterization of the dynamics of the continuous microfluidic

sensing. a, Stable peak amplitudes of twelve successive DPV scans in a solution containing 80

µM UA and 200 µM Tyr with a fixed flow rate of 1 µL min-1. b, Dynamic DPV response of the

microfluidic sensing when the inflow solution was switched from 20 µM UA and 50 µM Tyr to

80 µM UA and 200 µM Tyr with a fixed flow rate of 1 µL min-1. A DPV scan was performed

every minute (with a scan step of 8 mV and a scan cycle of 45 seconds). Throughout the figure,

the experiments were repeated 3 times independently and similar results were obtained.

0.2 0.4 0.6 0.8100

135

170

205

240

Pote

ntia

l (m

V)

Potential (V)

1st scan 2nd scan 3rd scan 4th scan 5th scan 6th scan

1 3 5 7 9 11 13 1580

90

100

110

120

V/V 0

(%)

UA Tyr

Scan cycles

ba

Supplementary Fig. 30. Characterization of continuous microfluidic sensing performance

under different flow rates. a, The DPV plots of an LEG-CS for UA and Tyr sensing at different

flow rates (from 0.25 to 1.5 µL min-1). b, The peak amplitudes of UA and Tyr in 5 successive

DPV scans (2.5 minutes per scan with a 45-s scan cycle length) with different flow rates (from

0.25 to 1.5 µL min-1). Conditions, 80 µM UA and 200 µM Tyr. Throughout the figure, the

experiments were repeated 3 times independently and similar results were obtained.

0.2 0.4 0.6 0.8100

135

170

205

240

Pote

ntia

l (m

V)

Potential (V)

μ -1

μ -1

μ -1

μ -1

μ -1

μ -1

0.25 0.50 0.75 1.00 1.25 1.500

25

50

75

100

125

Tyr p

eak

ampl

itude

(mV)

Flow rate (μ -1)

1st scan 2nd scan 3rd scan 4th scan 5th scan

0

10

20

30

40

UA

peak

am

plitu

de (m

V)

ba

Supplementary Fig. 31. The DPV plots obtained by the sensor patch from a healthy subject’s

neck during the cycling exercise, in one biking session. Conditions, same as Fig. 5b.

0.2 0.4 0.6 0.8100

200

300

400

500

Pote

ntia

l (m

V)

Potential (V)

15 min 20 min 25 min 30 min 35 min

Supplementary Fig. 32. Data processing of the strain sensor involving data filtering and fast

Fourier transform (FFT) to obtain the heart rate (HR) and respiration rate (RR) values. a–

c, An example of the raw data (a), filtered data (b), and FFT extracted frequency data (c) toward

RR monitoring. The raw data was collected by a strain sensor from a healthy subject. d–f, An

example of the raw data (d), filtered data (e), and FFT extracted frequency data (f) toward HR

monitoring. HR and RR values were obtained from the frequencies that had highest intensity in

the given periods. The raw data was collected by a strain sensor from a healthy subject.

0.4 0.8 1.2 1.6 2.0

Inte

nsity

(a.u

.)

Frequency (Hz)

0 10 20 3032.6

32.8

33.0

33.2

kΩ

Time (s)0 10 20 30

32.6

32.8

33.0

33.2

kΩ

Time (s)0.2 0.3 0.4 0.5 0.6 0.7

Inte

nsity

(a.u

.)

Frequency (Hz)

0 5 10 15 2031.80

31.85

31.90

31.95

kΩ

Time (s)

ba

ed

c

f

Raw data Filtered data FFT processing

Raw data Filtered data FFT processing

0 5 10 15 2031.80

31.85

31.90

31.95

kΩ

Time (s)

Supplementary Fig. 33. The peak amplitude of DPV curves in Fig. 5c,d obtained during the

on-body UA and Tyr measurement from the neck (a) and the forehead (b) of a healthy

subject during one biking session.

15 20 25 30 350

10

20

30

UA Tyr

Peak

am

plitu

de (m

V)

Time (min)15 20 25 30 35

0

10

20

30

40

Peak

am

plitu

de (m

V)

Time (min)

UA Tyr

ba

Supplementary Fig. 34. HPLC analyses of the UA and Tyr in sweat. The experiments were

repeated 3 times independently and similar results were obtained.

0 5 10-4

0

4

8

12

Buffe

r abs

orba

nce

(mAU

)

Time (min)0 5 10

-4

0

4

8

12

Buffe

r abs

orba

nce

(mAU

)

Time (min)

0 5 10 15

0

20

40

60

80

Buffe

r abs

orba

nce

(mAU

)

Time (min)0 5 10 15

0

20

40

60

80

Buffe

r abs

orba

nce

(mAU

)

Time (min)

ba

ed

0 5 10-4

0

4

8

12

Swea

t abs

orba

nce

(mAU

)

Time (min)

0 5 10 15

0

20

40

60

80

Swea

t abs

orba

nce

(mAU

)

Time (min)

c

f

10 μ UA 15 μ UA

72 μ Tyr36 μ Tyr

Sweat sample

Sweat sample

UAUAUA

TyrTyr

Tyr

Supplementary Fig. 35. Selectivity of HPLC analyses of the UA (a) and Tyr (b) over common

sweat analytes. The concentrations of the initial UA/Tyr and added other analytes are based on

Supplementary Table 2. UA and Tyr levels were increased by 50% in the end for a and b,

respectively. Throughout this figure, the experiments were repeated 3 times independently and

similar results were obtained.

0

50

100

150

200

Nor

mal

ized

inte

nsity

(a.u

.)

Sample

+ Tyro

sine

+ Uric

acid

+ Vali

ne

+ Tryp

topha

n

+ Thre

onine

+ Tha

imine

+ Seri

ne

+ Ribo

flavin

+ Pyri

doxin

e

+ Phe

nylal

anine

+ Pan

tothe

tic ac

id

+ Orni

thinin

e

+ Nico

tinic

acid

+ Meth

ionine

+ Lys

ine

+ Leu

cine

+ Iso

leucin

e

+ Ino

sitol

+ Hist

idine

+ Glyc

ine

+ Glut

amic

acid

+ Foli

c acid

+ Deh

ydroa

scorb

ic ac

id

+ Crea

tinine

+ Crea

tine

+ Citru

lline

+ Cho

line c

hlorid

e

+ Asp

artic

acid

+ Asc

orbic

acid

+ Argi

nine

+ Alan

ine

Uric ai

cd

0

50

100

150

200

Nor

mal

ized

inte

nsity

(a.u

.)

Sample

+ Uric

acid

+ Tyro

sine

+ Vali

ne

+ Tryp

topha

n

+ Thre

onine

+ Tha

imine

+ Seri

ne

+ Ribo

flavin

+ Pyri

doxin

e

+ Phe

nylal

anine

+ Pan

tothe

tic ac

id

+ Orni

thinin

e

+ Nico

tinic

acid

+ Meth

ionine

+ Lys

ine

+ Leu

cine

+ Iso

leucin

e

+ Ino

sitol

+ Hist

idine

+ Glyc

ine

+ Glut

amic

acid

+ Foli

c acid

+ Deh

ydroa

scorb

ic ac

id

+ Crea

tinine

+ Crea

tine

+ Citru

lline

+ Cho

line c

hlorid

e

+ Asp

artic

acid

+ Asc

orbic

acid

+ Argi

nine

+ Alan

ine

Tyrosin

e

b

a

Supplementary Fig. 36. Validation of the accuracy of the LEG-CS for UA (a) and Tyr (b)

sensing using colorimetric assay kits. The data are based on 28 sweat samples collected from the

healthy subjects. UA and Tyr concentrations by the sensor were obtained from the sensor reading

and the linear fitting in Fig. 5e,f. The Pearson correlation coefficient was acquired through linear

regression.

0 50 100 150 2000

50

100

150

200

μ

μ

0 25 50 75 1000

25

50

75

100

μ

μ

ba

r = 0.965 r = 0.967

Supplementary Fig. 37. Dynamic Tyr monitoring using the sensor patch after Tyr

supplementation. The experiment was performed on a healthy subject over one iontophoresis-

induced sweat secretion process. The experiment was repeated twice independently with similar

results.

0 10 20 30 40 500

30

60

90

120

μ

0 10 20 30 40 500

5

10

15

20

μ

0 10 20 30 40 500

1

2

3

4

Flow

rate

(μ-1

ba c

Tyrosine intake

Supplementary Fig. 38. The influence of high and low protein intakes on sweat UA and Tyr.

0

40

80

120

160 Low protein High protein

μ

UA TyrSubject 2

μ

0

25

50

75

100

0

40

80

120

160 Low protein High protein

0

40

80

120

160

μ μ

UASubject 1

Tyr

ba c

0

10

20

30

40 Low protein High protein

μ

UA TyrSubject 3

0

40

80

120

160

μ

Supplementary Fig. 39. Investigation of the sweat and serum UA levels using a purine-rich

diet challenge. UA levels in sweat and serum of two heathy subjects under fasting condition, and

after a purine-rich diet on two different days. Measure of the center in a–d is the mean value of

the first two successive measurements.

0

30

60

90

120 Fasting Purine-rich

Subject 3 (male), day 1Sweat Serum

μ μ

300

400

500

600

700

0

30

60

90

120 Fasting Purine-rich

300

400

500

600

700

SweatSubject 3 (male), day 2

Serum

μμ

0

20

40

60

80 Fasting Purine-rich

μ μ

200

300

400

500

SweatSubject 4 (male), day 1

Serum0

20

40

60

80 Fasting Purine-rich

μ μ

200

300

400

500

SweatSubject 4 (male), day 2

Serum

dc

ba

Supplementary Fig. 40. HPLC analyses of the sweat (a,b) and serum (c,d) UA levels of a

healthy subject (a,c) and a patient with gout (b,d). The experiment was repeated 3 times

independently with similar results.

0 5 10

0

10

20

30

Swea

t abs

orba

nce

(mAU

)

Time (min)

0 5 10-5

0

5

10

Seru

m a

bsor

banc

e (m

AU)

Time (min)0 5 10

-5

0

5

10

Seru

m a

bsor

banc

e (m

AU)

Time (min)

0 5 10

0

10

20

30

Swea

t abs

orba

nce

(mAU

)Time (min)

ba

dc

Healthy subject Patient subject

Patient subjectHealthy subject

UAUA

UAUA

Supplementary Fig. 41. The influence of urate-lowering medication (allopurinol) on the sweat

and serum UA levels. The sweat and blood tests were performed on the same gout patient on two

days: on one day the patient received urate-lowering medication 2 hours before the test, and on the

other day the patient did not take medication 24 hours before the test. The center of measure is the

mean value of the first two successive measurements from one patient.

0

45

90

135

μ μ

W/ medication W/o medication

SerumSweat200

400

600

800

Supplementary Table 1. Optimization of the LEG-CS with different laser parameters.

Electrode # UA Peak (mA cm-2)

Optimization of chemical sensors under different laser parameters

Power (%) Speed (%) PPI Tyr Peak (mA cm-2)

1

2

3

4

5

6

7

8

9

12

11

10

6.3

6

6

6

6

5.5

6.5

7.5

6

6

6

6

5.5

5

6

7

8

5

5

5

5

5

5

5

1000

1000

1000

1000

1000

1000

1000

1000

250

1000

750

500

0.142

0.003

0.091

0.113

0.05

0

0.132

0.101

0.138

0.131

0.1

0.115

0.061

0

0.035

0.029

0.013

0

0.043

0.032

0.045

0.031

0.033

0.049

Supplementary Table 2. Sweat analytes and their concentrations used for the selectivity

studies. The values were chosen based on the mean physiological levels of the analytes46.

Constituents

Median values of human sweat constituents

Concentration (μ )

Alanine

Arginine

Ascorbic acid

Aspartic acid

Choline chloride

Citrulline

Creatine

Creatinine

Dehydroascorbic acid

Glycine

Glutamic acid

Folic acid

Histidine

Inositol

Isoleucine

Leucine

Number

1

2

3

4

5

6

7

8

9

12

11

10

13

14

15

16

360

780

10

340

50

400

15

84

11

390

370

50

520

1.6

16

210

Constituents Concentration (μ )Number

Lysine

Methionine

Nicotinic acid

Ornithine

Pantothetic acid

Phenylalanine

Pyridoxine

Riboflavin

Serine

Tryptophan

Threonine

Thiamine

Tyrosine

Uric acid

Valine

17

18

19

20

21

22

23

24

25

28

27

26

29

30

31

150

50

50

150

50

130

0.01

50

50

55

450

50

170

60

250

Supplementary Table 3. UA HPLC gradient method.

Eluent A Time (min)

Uric acid HPLC gradient method

Eluent B Flow rate Percentage eluent B (%)

0.34mL min-1

0

6

6.2

7.5

7.9

15

0

0

100

100

0

0

Acetonitrile100%

0.1%Formic acid

Supplementary Table 4. Tyr HPLC gradient method.

Eluent A Time (min)

Tyrosine HPLC gradient method

Eluent B Flow rate Percentage eluent B (%)

0.425

mL min-1

0

4

8

12

15

20

9.1

9.1

80

80

9.1

9.1

Methanol

100%

50 mM

KH2PO4-H3PO4

buffer

pH 3.0

Supplementary Note 1. The selectivity of the LEG-CS for sweat UA and Tyr sensing

Current wearable electrochemical sweat sensors are primarily focused on a limited number of

electrolytes (e.g., Na+, K+, Cl-) and metabolites (e.g., lactate and glucose) at high concentrations

(usually at mM levels) monitored via ion-selective sensors or enzymatic electrodes; biosensors

based on bioreceptors (e.g., antibodies) could be highly sensitive, but usually require multiple

washing steps and detection in standard buffer or redox solutions in order to transduce the

bioaffinity interactions.

The approach we utilize here for continuous monitoring of UA and Tyr is based on their selective

oxidation reaction at a specific potential. The selectivity is based on the oxidation peak position

for the LEG-CS. We evaluate here the selectivity of the LEG-CS over the common amino acids,

vitamins and other potential interfering chemicals at physiological relevant concentrations in

human sweat listed in Supplementary Table 2. Our electrochemical sensing data in

Supplementary Figs. 11 and 12 show that the physiological level sweat analytes have minimal

influence on the detection of UA and Tyr using the LEG-CS.

There are few electroactive molecules reported to be present in human sweat with comparable

concentrations with UA and Tyr. Dopamine is a well-known electroactive neurotransmitter present

at high concentration in cerebrospinal fluid. Our data in Fig. 2i shows that at the same conditions,

the dopamine oxidation peak appears at a more negative potential than UA and doesn’t affect the

UA monitoring. In fact, to our knowledge, there is no literature report on the presence of dopamine

in human sweat. Another electroactive molecule tryptophan has a close oxidation peak position to

Tyr. However, tryptophan is present at a much lower concentration compared to Tyr. Our

selectivity study (Supplementary Fig. 13) shows that physiologically relevant concentrations of

tryptophan do not significantly affect our Tyr measurement.

As pH is usually a key factor that could influence electrochemical sweat sensor performance, we

try to evaluate here the influence of pH on the sensor performance in artificial sweat (with pH

values adjusted by varying lactic acid as lactic acid is the main contributing factors for sweat pH).

Considering the normal sweat pH range in the literature and from our human subjects (pH ranges

from 4 to 6), the dependence of the chemical sensor performance in artificial sweat with pHs

between 4 to 6 was examined. Our results in Supplementary Fig. 14 show that, although the

oxidation peak positions slightly shift under different pHs, the peak amplitudes of UA and Tyr

sensing remain stable in the range of pH 4 to pH 6. Considering that we are essentially measuring

the peak heights, the influence of the pH is small to our sensors. Moreover, it is worth noting that

the variations of pH values of an individual’s sweat at different time points during biking session

and at different biking sessions are small.

To further validate the accuracy and selectivity of the LEG-CS, the sensor readings in human sweat

samples have been validated with two current gold standards – high performance liquid

chromatography (HPLC) (Fig. 5e,f and Supplementary Fig. 35) and commercial colorimetric

assay kits (Supplementary Fig. 36). Very high linear coefficients (0.963 and 0.965 for UA and

Tyr with HPLC, 0.965 and 0.967 for UA and Tyr with the commercial colorimetric assays) were

found, revealing the great reliability and accuracy of our sensor technology toward sweat UA and

Tyr sensing.