13768


FACTA UNIVERSITATIS  
Series: Electronics and Energetics Vol. 39, No 1, March 2026, pp. 257 - 268 

https://doi.org/10.2298/FUEE2601257S 

© 2026 by University of Niš, Serbia | Creative Commons License: CC BY-NC-ND 

Original scientific paper  

HIGH-VOLTAGE SURGE IMPACT ON THICK-FILM SENSORS 

FOR STRUCTURAL HEALTH MONITORING: RESISTANCE 

AND NOISE SPECTROSCOPY ANALYSIS 

Zdravko Stanimirović1, Ana Stanimirović2, Aleksandar Savić3,  

Ivanka Stanimirović1  

1“Vinča” Institute of Nuclear Sciences - National Institute of the Republic of Serbia, 

University of Belgrade, Serbia  
2Faculty of Physics, University of Belgrade, Serbia 

3Faculty of Civil Engineering, University of Belgrade, Serbia 

ORCID iDs: Zdravko Stanimirović   https://orcid.org/0000-0002-2048-0027 

 Ana Stanimirović   https://orcid.org/0009-0002-6002-7347 
 Aleksandar Savić   https://orcid.org/0000-0002-1777-6775 
  Ivanka Stanimirović   https://orcid.org/0000-0003-2445-611X 

Abstract. This study explores the effects of high-voltage electrical surges on the performance and 

structural integrity of thick-film strain sensors developed for structural health monitoring in steel 

infrastructure. The sensors were fabricated using screen-printing techniques with a bismuth lead 

ruthenate-based resistive composition deposited on alumina ceramic substrates. To simulate realistic 

operational conditions, the sensors were mounted on steel beams and subjected to four-point bending 

to induce mechanical strain. Following mechanical loading, controlled high-voltage surge pulses 

were applied to emulate extreme electrical events. Sensor response was characterized before and 

after surge exposure using both static resistance measurements and current noise spectral analysis. 

While resistance measurements showed limited change, noise spectroscopy revealed microstructural 

damage undetectable by conventional means. The findings highlight the degradation mechanisms 

arising from electromechanical stress and demonstrate the effectiveness of noise spectroscopy as a 

non-destructive diagnostic tool. These results support the use of thick-film sensors in electrically 

demanding environments. 

Key words: thick-film strain sensors, structural health monitoring, high-voltage surge testing, low-

frequency noise, tunnelling conduction, metal-insulator-metal junctions 

 
Received June 10, 2025; revised August 20, 2025; accepted September 16, 2025 

Corresponding author: Ivanka Stanimirović 
“Vinča” Institute of Nuclear Sciences – National Institute of the Republic of Serbia, University of Belgrade. 

E-mail: ivanka.stanimirovic@vin.bg.ac.rs 



258 Z. STANIMIROVIĆ, A. STANIMIROVIĆ, A. SAVIĆ, I. STANIMIROVIĆ 

1. INTRODUCTION 

Throughout their operational lifespan, civil engineering structures endure gradual 

deterioration and exposure to diverse environmental challenges, such as seismic disturbances, 

hydrodynamic forces, soil shifts and extreme climatic conditions. As infrastructure becomes 

increasingly complex and expansive, the deployment of sophisticated structural health 

monitoring (SHM) systems is imperative to uphold structural integrity, enhance safety and 

refine maintenance strategies. SHM functions on the principle that a structure's mechanical and 

material properties change over time due to external forces, wear, corrosion and other 

degradation processes. By persistently tracking strain variations, internal stress fluctuations and 

additional crucial metrics, early indicators of structural damage can be identified, helping to 

prevent catastrophic failures [1-3]. Strain measurement is especially vital in SHM, offering 

essential insights into load distribution, fatigue endurance and the prolonged functionality of 

civil infrastructure. 

Various sensor technologies have been explored for strain measurement in SHM, including 

metal foil strain gauges, fiber optic sensors and piezoelectric sensors, each possessing unique 

benefits and limitations [4-9]. In recent years, thick-film strain sensors have gained prominence 

as a resilient and cost-effective alternative for SHM applications, delivering an optimal balance 

of mechanical strength, environmental resistance and seamless integration. Unlike conventional 

strain gauges that rely on adhesive attachment, thick-film strain sensors feature screen-printed 

resistive films sintered onto various substrates, which ensures strong adhesion, durability and 

resistance to mechanical fatigue [10-12]. These sensors exhibit remarkable thermal stability, with 

minimal resistance fluctuations across diverse operational environments, making them highly 

suitable for outdoor structures subjected to varying temperatures. Furthermore, thick-film strain 

sensors must demonstrate resilience against high-voltage surges, a crucial attribute for steel 

components in modern glass-steel constructions, bridges and critical infrastructure that face risks 

from lightning strikes and electrical disturbances. The differences in electrical and thermal 

performance between thick-film and other SHM sensor technologies are highlighted in Table 1. 

This research examines the behavior of screen-printed Bi₂Ru₂O₇-based ceramic strain 

sensors affixed to steel bars under high-voltage surge conditions. To analyze the influence 

of high-voltage surges, this study measures resistance variations in the sensors and 

investigates how structural changes correlate with noise performances. By assessing these 

interactions, the research provides a deeper understanding of how high-voltage exposure 

impacts sensor functionality. The findings highlight the promise of thick-film technology 

as a reliable and scalable solution for SHM, demonstrating superior durability and 

adaptability compared to conventional strain sensing techniques. Implementing thick-film 

sensors within civil infrastructure can bolster predictive maintenance efforts, lower upkeep 

costs and extend the operational lifespan of essential structures. This research contributes 

to the evolution of SHM methodologies, offering key insights into the feasibility of thick-

film strain sensors in high-voltage environments. 

Table 1 Characteristics of thick-film sensors, metal foil strain gauges, fiber optic sensors 

and piezoelectric sensors for SHM [13] 

Sensor Type Thick-Film 

Sensors 

Metal Foil Strain 

Gauges 

Fiber Optic 

Sensors 

Piezoelectric 

Sensors 

GF 2–35 ~2 0.8–1.2 Varies 

TCR (ppm/°C)  ±50 to ±200 ±5 to ±20 ~0 Varies 



 High-Voltage Surge Impact on Thick-Film Sensors for Structural Health Monitoring... 259 

2. EXPERIMENTAL EVALUATION OF HIGH-VOLTAGE SURGE EFFECTS  

ON THICK-FILM STRAIN SENSORS 

To assess the effects of high-voltage surges on thick-film strain sensors for structural 

health monitoring, thick-film strain sensors were affixed to steel bars (Fig. 1a). The sensors 

were fabricated on 0.635 mm thick alumina (96% Al₂O₃) substrates, combined with a 

Bi2Ru2O7-based thick-film resistive composition exhibiting a sheet resistance of  

10 kΩ/sq. The resistive film was applied using standard screen-printing techniques with a 

semiautomatic screen printer. A stainless-steel screen, stretched over an aluminum frame, 

was employed to transfer the thick-film composition onto the substrate. The printed layer 

pattern was defined by a 200-mesh stainless-steel screen with emulsion thicknesses of 10-

12 μm, ensuring high-resolution deposition of the resistive film. A Pd/Ag conductive thick 

film was employed for the fabrication of contact pads to ensure reliable electrical 

connections in SHM applications. Each sensor, measuring 3 × 21 mm², was segmented into 

seven 3 × 3 mm² sections, separated by six contact pads (Fig. 1a). Following deposition, 

the printed resistive layers underwent a series of precisely controlled curing and firing steps 

to ensure stable electrical characteristics. 

The curing process involved initial leveling at room temperature, followed by drying at 

150 °C for 10 minutes in an infrared conveyor dryer. Subsequently, the resistive layers, 

with a thickness of 25 ± 3 µm, were subjected to a firing cycle lasting 60 minutes, including 

a 10-minute dwell time at a peak temperature of 850 °C. The conductive layers underwent 

a separate firing process lasting 30 minutes at the same peak temperature to ensure optimal 

adhesion and conductivity. 

Thick-film strain sensors were attached to steel bars, commonly used as reinforcement 

in concrete structures and as load-bearing elements in bridges, high-rise buildings and 

tunnels, where precise strain monitoring is critical for structural integrity [14-15]. The 

mechanical properties of Bi2Ru2O7-based thick-film strain sensors are well-matched with 

the modulus of elasticity of steel (210 GPa [16]), enabling accurate strain measurements 

while minimizing material incompatibilities. Moreover, screen-printed thick-film sensors 

on alumina substrates exhibit high mechanical strength, with an elasticity modulus of 300–

400 GPa [17], making them more compatible with steel than with concrete (elasticity 

modulus of 10–30 GPa [18]). This compatibility renders them particularly suitable for 

direct application on exposed steel components in modern glass-steel constructions, 

bridges and critical infrastructure. 

All sensor samples were fabricated at the Institute for Electronics and Telecommunications 

IRITEL a.d. Beograd, Serbia. Resistance measurements were conducted using the National 

Instruments VB-8034 Virtual Bench Instrument to ensure precise and reliable data 

acquisition under various strain conditions. The response of the ceramic strain sensors to 

high-voltage surges was systematically evaluated using a Haefely P6T pulse generator, 

capable of delivering 10/700 μs pulses to simulate rapid electrical transients commonly 

encountered in structural environments (Fig. 1b). A four-point bending test, designed to 

apply maximum flexural stress between two loading points, was performed at the Faculty 

of Civil Engineering, University of Belgrade. The steel bars used had dimensions of 200 × 

20 × 40 mm, with a support span of 180 mm and a distance between loading points of 100 

mm. This configuration ensures uniform strain distribution along the specimen length, 

providing realistic simulation of strain conditions in steel structural elements used in SHM 

(Fig. 1a). The rate of loading was not controlled, as the experimental setup required fine 



260 Z. STANIMIROVIĆ, A. STANIMIROVIĆ, A. SAVIĆ, I. STANIMIROVIĆ 

manual adjustment to achieve the target substrate deflection. To prevent sensor 

detachment, substrate deflection was set to 300 μm, selected based on the breaking point 

of the alumina substrate determined during preliminary three-point bending tests and the 

properties of the adhesive used to attach the sensor to the steel bar.  

 (a) (b) 

Fig. 1 (a) Thick-film strain sensor attached to a steel bar undergoing a four-point bending 

test and (b) 10/700 μs pulse delivered by Haefely P6T pulse generator 

The electrical straining conditions were carefully controlled to facilitate gradual 

resistance variations during pulse application. The sensors were subjected to pulse 

increments of 250 V, ranging from 3 kV to 5.25 kV, with the generator's output resistance 

fixed at 25 Ω. The testing procedure began with ten pulses applied at a frequency of six 

pulses per minute, with the voltage gradually increased from 3 kV to 5 kV. Subsequently, 

four single pulses with a peak amplitude of 5.25 kV were introduced. Throughout the 

testing process, gauge factor (GF) values were continuously monitored to evaluate the 

sensor response. The high-voltage surge testing adhered strictly to ITU-T K.20, the 

industry standard for thick-film surge resistor evaluation, ensuring the validity and 

applicability of the findings for assessing the performance of thick-film strain sensors in 

SHM applications. Current noise spectra before and after high-voltage surge testing were 

measured using a Keithley Model 103A nanovolt amplifier in conjunction with an HP-

3561B Dynamic Signal Analyzer over a frequency range of 10 Hz to 10 kHz. All measurements 

were conducted under controlled conditions at room temperature  

(T=295 K) to ensure consistency and accuracy in the acquired data. Experimental arrangement 

employed in the characterization of the thick-film strain sensor is presented in Figure 2. 

3. ANALYSIS OF THE EFFECTS OF HIGH-VOLTAGE SURGES ON THICK-FILM STRAIN SENSORS 

During the experimental procedure, the sensors were exposed to a controlled sequence 

of high-voltage pulses to evaluate their electrical stability and resistance behavior under 

increasing electrical stress. The applied voltage was incrementally raised in 250 V steps, 

ranging from an initial 3 kV to a maximum of 5.25 kV. The testing procedure started with 

ten pulses delivered at a frequency of six pulses per minute, with the voltage steadily rising 



 High-Voltage Surge Impact on Thick-Film Sensors for Structural Health Monitoring... 261 

from 3 kV to 5 kV. As the applied voltage increased, a gradual decrease in resistance was 

recorded, suggesting an alteration in the sensor’s conductive pathways due to electrical 

stress. Once resistance variations approached a threshold of approximately 3 %, the pulse 

frequency was reduced to a single pulse per voltage increment to allow for precise tracking 

of further resistance fluctuations. Interestingly, at that point resistance values exhibited 

stabilization, indicating the sensor’s capacity to reach an electrically stable state despite 

prolonged exposure to high-voltage transients. Additionally, the gauge factor measurements 

remained consistent throughout the experiment, confirming that the sensor maintained its strain 

sensitivity and overall functional integrity despite the imposed electrical stress conditions. The 

effects of high-voltage surges on the resistance and gauge factor of thick-film strain sensors at 

a substrate deflection of 300 μm are presented in Fig. 3. Data are presented for two sensors with 

initial resistances of R1=33.84 kΩ and R2=31.94 kΩ, highlighting resistance variations and 

gauge factor stability under electrical stress. 

 

Fig. 2 Experimental arrangement employed in the characterization of the thick-film strain sensor 

  

Fig. 3 Effect of high-voltage surges on the resistance and gauge factor of two thick-film 

strain sensors at a substrate deflection of 300 μm 

-20

0

20

40

60

80

100

-4

-3.5

-3

-2.5

-2

-1.5

-1

-0.5

0

0 3 3
.2

5

3
.5

3
.7

5

4 4
.2

5

4
.5

4
.7

5

5 5
.2

5

5
.2

5

5
.2

5

5
.2

5

Δ
G

F/
G

Fi
 [

%
]

Δ
R

/R
i [

%
]

V [kV]

ΔR/R1 ΔR/R2 ΔGF/GF1 ΔGF/GF2



262 Z. STANIMIROVIĆ, A. STANIMIROVIĆ, A. SAVIĆ, I. STANIMIROVIĆ 

The electrical conduction in thick-film resistive materials is governed by a combination 

of contact conduction and tunneling conduction mechanisms [19]. The resistive film 

comprises numerous parallel conducting chains, where adjacent conductive particles are 

either in direct contact or separated by thin layers of borosilicate glass. Under these 

conditions, the overall resistance of the film can be expressed as:  

 𝑅 =
𝐾𝑐

𝑀2 𝑅𝑐 +
𝐾𝑏

𝑀2 𝑅𝑏 (1) 

where 𝑅𝑐 represents the contact resistance between neighboring conductive particles, Rb 

denotes the barrier resistance associated with tunneling through the insulating borosilicate 

glass layer, M is the total number of parallel conducting chains within the thick-film 

resistive network, Kc is the number of direct contacts between adjacent particles and Kb 

corresponds to the number of metal-insulator-metal (MIM) junctions formed by conductive 

particles separated by the insulating glass phase. In this context, the total number of 

electrical junctions between adjacent conducting particles, K, can be expressed as: 

 𝐾 = 𝐾𝑐 + 𝐾𝑏 (2) 

To systematically investigate the effects of applied strain on conduction pathways - 

particularly the transitions between direct interparticle contact and the formation or 

breakdown of MIM junctions - the relative contributions of contact resistance and barrier 

resistance can be introduced as [20]: 

 𝑝 =
𝐾𝑐

𝐾
 (3) 

 1 − 𝑝 =
𝐾𝑏

𝐾
  (4) 

Where p represents the fraction of conductive pathways governed by direct particle-to-

particle contact, while 1-p corresponds to the proportion of conduction occurring via 

tunneling through MIM junctions. These parameters provide a framework for analyzing 

strain-induced modifications in the thick resistive film. 

 If we define thick resistive films conductance as: 

 𝐺 =
𝑀2

𝐾[𝑝𝑅𝑐+(1−𝑝)𝑅𝑏]
 (5) 

Considering the volume fraction of the conductive phase within the resistive film 

composition and the inherent relationship Rc<<Rb, for low-ohmic formulations the overall 

conductance can be approximated as: 

 𝐺 ≈
𝑀2

𝑝𝐾𝑅𝑐
 (6) 

For high-ohmic compositions, where tunneling conduction through the MIM junctions 

dominates, the conductance is given by: 

 𝐺 ≈
𝑀2

(1−𝑝)𝐾𝑅𝑏
 (7) 

These expressions highlight the dependence of conductance on the proportion of direct 

contact pathways and MIM-mediated conduction, providing insight into the resistive film's 

behavior under varying electrical and mechanical conditions. 



 High-Voltage Surge Impact on Thick-Film Sensors for Structural Health Monitoring... 263 

The experimental data presented in Fig. 2 demonstrate an initial reduction in resistance 

during high-voltage surge testing of strained thick-film sensors at a substrate deflection of 

300 μm. For low-ohmic formulations, as described by Eq. (6), the parameter p decreases, 

indicating a reduction in the number of direct conductive contacts between neighboring 

particles. Consequently, in low-ohmic compositions, the dominant conduction mechanism 

transitions from contact-based conduction to tunneling conduction through MIM junctions. 

Conversely, in high-ohmic formulations, as described by Eq. (7), the parameter p increases, 

signifying a greater number of conductive contacts between adjacent particles. As a result, 

in high-ohmic compositions, the predominant conduction mechanism shifts from tunneling 

conduction to direct contact conduction. 

Following the initial decrease in resistance, the resistance values reached a stable state, 

demonstrating the sensor's ability to attain electrical equilibrium despite sustained exposure 

to high-voltage transients. Throughout the experimental procedure, the gauge factor 

remained invariant, indicating that the sensor preserved its strain sensitivity and functional 

reliability under the applied electrical stress conditions.  

Figure 4 illustrates the contribution of contact resistances (p) for the thick-film strain sensors 

characterized in Figure 2. The calculations were performed using the following parameter 

values: M=2.8·105, K=2.7·109, Kc=9.0541·108, Kb=1.81·109, Rc=136.97 Ω, Rb=1.34·107 Ω, 

corresponding to a 10 kΩ/sq thick-film resistor formulation. This particular composition was 

selected due to its intermediate electrical characteristics, encompassing attributes of both low-

ohmic and high-ohmic resistive materials. The observed initial decrease in resistance, 

accompanied by a corresponding increase in p, suggests that tunneling conduction mechanisms 

play a dominant role in the early stages of high-voltage surge exposure. 

Considering that low-frequency noise sources in thick-film resistors are closely related 

to conduction mechanisms, measurements of the current noise spectrum were performed 

for different current levels before and after high-voltage surge testing. Noise sources 

include fluctuations in contact resistance and intrinsic particle resistance in the case of 

conduction through conductive particles and their interconnections, as well as noise arising 

from Nyquist noise modulation and fluctuations caused by trap states in glass barriers when 

conduction occurs through these barriers [21]. The results, depicted in Fig. 5, correspond 

to a thick-film strain sensor with an initial resistance of R2=31.94 kΩ. 

   

Fig. 4 Contribution of contact resistances p for two sensors with initial resistances of  

R1=33.84 kΩ and R2=31.94 kΩ 

0.926

0.927

0.928

0.929

0.93

0.931

0.932

0.933

0.934

0 3 3
.2

5

3
.5

3
.7

5

4 4
.2

5

4
.5

4
.7

5

5 5
.2

5

5
.2

5

5
.2

5

5
.2

5

P

V [KV]

R2 R1



264 Z. STANIMIROVIĆ, A. STANIMIROVIĆ, A. SAVIĆ, I. STANIMIROVIĆ 

 

Fig. 5 Current noise spectrum measured before and afterhigh-voltage surge testing of 

strained thick-film strain sensor with an initial resistance R2=31.94 kΩ 

In the context of low-frequency noise, assuming that noise originating from fluctuations 

within the MIM junction is the dominant contribution to the overall noise, the relative 

voltage noise spectrum can be expressed as [21]: 

 
𝑆𝑉(𝑓)

𝑉2
≈

1

𝐾(1−𝑝)(1+
𝑝𝑅𝑐

(1−𝑝)𝑅𝑏
)
2 ∙

𝑆𝑉𝐵𝑁(𝑓)

𝑉𝐵
2 +

𝐾𝑡

𝐾2(1−𝑝)2(1+
𝑝𝑅𝑐

(1−𝑝)𝑅𝑏
)
2 ∙

𝑆𝑉𝑇(𝑓)

𝑉𝐵
2  (8) 

where SVBN(f)/VB
2 represents the relative voltage noise spectrum resulting from Nyquist 

noise modulation and SVT(f)/VB
2 corresponds to the relative voltage noise spectrum 

attributed to the presence of traps within the glass barriers. When metallic conduction is 

the primary source of low-frequency noise [21]: 

 
𝑆𝑉(𝑓)

𝑉2
≈

1

𝐾𝑝(1+
(1−𝑝)𝑅𝑏
𝑝𝑅𝑐

)
2 ∙

𝑆𝑉𝐶(𝑓)

𝑉𝐶
2  (9) 

To gain a more comprehensive understanding of the contribution of each low-frequency 

noise source to the total noise spectra before and after high-voltage surge testing, a fitting 

procedure was conducted using the experimental data presented in Fig. 4 and the following 

theoretical expression: 

 𝑆𝐼(𝑓) = 𝐴 +
𝐵

𝑓𝛾
+ ∑

𝐶𝑖

2𝜋𝑓𝐶𝑖(1+
𝑓2

𝑓𝐶𝑖
2 )

𝑖  (10) 

with the first term representing thermal noise: 

 𝐴 = 4𝑘𝑇𝑅  (11) 

where k is the Boltzmann constant, T is the absolute temperature and R is the resistance of 

the thick resistive film. The second term corresponds to 1/f noise with parameters B and γ, 

while the third term represents the sum of noise spectra of Lorentzian shape, characterized 

by parameters Ci and characteristic frequencies fCi. 

1E-24

1E-23

1E-22

1E-21

1E-20

1E-19

1.00E+01 1.00E+02 1.00E+03 1.00E+04

Si
 [

A
2
/H

z]

f [Hz]

Before high-voltage
surge testing

After high-voltage
surge testing



 High-Voltage Surge Impact on Thick-Film Sensors for Structural Health Monitoring... 265 

The fitting results corresponding to the data presented in Fig. 5, along with the 

individual contributions of various noise sources to the total current noise spectrum, are 

shown in Figs. 6a and 6b for strained thick-film strain sensor with an initial resistance of 

R2=31.94 kΩ, both before and after high-voltage testing. As can be seen from Fig. 6a, the 

total current noise spectrum of the mechanically strained sensor exhibits both 1/f and 

thermal noise. The 1/f noise is composed of several contributions: fluctuations of contact 

resistances and the resistances of individual conducting particles within the conduction network, 

fluctuations in potential barrier heights due to Nyquist noise in the glassy matrix and 

fluctuations in the occupancy of charge carrier traps. Under the applied mechanical straining, 

the dominant contribution arises from fluctuations in barrier widths, as the number of 

conduction chains, contacts, and barrier heights remain largely unaffected by strain. 

The results obtained from current noise spectrum measurements following high-voltage 

surge testing indicate that the current noise spectrum increases as a consequence of 

electrical stress. Interpreted through the noise index parameter p, this increase implies a 

decrease in p, according to Eqs. 8 and 9. However, this behavior is not consistent with the 

p values derived from resistance measurements shown in Fig. 3.  Having in mind that the 

applied model excludes possibility of change in noise values for two neighboring 

conducting particles and MIM junctions, as well as value variations for contact and barrier 

resistances due to straining, noise performances of stained resistors could be explained 

using redistribution of numbers of contacts and MIM junctions as well as adequate 

contributions of noise due to metallic conduction and noise due to fluctuations in MIM 

junctions to total noise of strained resistor. This is in accordance with results shown in Fig. 

6b. The total current noise spectrum is dominated by 1/f noise, arising from fluctuations in 

potential barrier heights due to Nyquist noise within the glassy matrix [22-23] and 

fluctuations in the occupancy of charge carrier traps. In addition to the dominant 1/f noise, 

a significant contribution to the spectrum originates from three Lorentzian components. 

The Lorentzian components observed in the low-frequency noise spectrum arise from traps 

located within the insulating layers of metal–insulator–metal (MIM) structures [24]. A trap 

may be neutral or negatively charged when occupied by an electron during electron 

transport across the barrier via tunneling. The stochastic occupation of traps is governed 

by random tunneling events of electrons from the left (right) conducting particle to the trap 

and from the trap to the right (left) conducting particle. Fluctuations in trap occupancy 

induce local charge fluctuations, which in turn modulate the barrier height. In calculating 

the noise spectrum associated with trap-occupancy fluctuations, it is typically assumed that 

the dominant contribution originates from traps whose energy levels coincide with the 

Fermi level of the conduction electrons in the conducting particle, that the barrier potential 

can be approximated as rectangular, and that the applied bias is sufficiently small compared 

to the barrier height. A single discrete trap energy level gives rise to a Lorentzian 

component in the low-frequency noise spectrum, whereas a distribution of trap energy 

levels may produce 1/f-type noise. Therefore, Lorentzian components result from 

occupancy fluctuations of dominant traps within individual metal-insulator-metal junctures 

while spatial distribution of traps across all MIM junctions gives rise to 1/f noise.  

This analysis demonstrates that noise spectroscopy provides a sensitive, non-

destructive tool for probing microstructural changes in thick-film strain sensors. Variations 

in the 1/f noise component can be linked to modifications in conducting particle networks 

and sintered contacts, whereas Lorentzian contributions reveal trap-state dynamics within 

individual metal–insulator–metal junctions. Monitoring these spectral features allows early 



266 Z. STANIMIROVIĆ, A. STANIMIROVIĆ, A. SAVIĆ, I. STANIMIROVIĆ 

detection of electrically or mechanically induced degradation, facilitating the correlation 

of processing conditions with microstructural stability and supporting the optimization of 

sensor design and predictive maintenance strategies. Collectively, these findings highlight 

the complex dependence of the total noise spectrum on microstructural modifications and 

trap-state dynamics in electrically stressed and mechanically strained SHM sensors. 

  
(a) 

  
(b) 

Fig. 6 The experimental results for the current noise spectrum and the fitting results in 

accordance with Eq. (10) for the thick-film strain sensor with an initial resistance 

R2=31.94 kΩ, (a) before and (b) after high-voltage testing (I=0.284 mA), with 

contributions from different noise sources in the total current noise spectrum  

1E-25

1E-24

1E-23

1E-22

1E-21

1.00E+01 1.00E+02 1.00E+03 1.00E+04

Si
 [

A
2
/H

z]
f [Hz]

Measured values 1/f noise

Thermal noise Fitted results

2E-25

2E-24

2E-23

2E-22

2E-21

2E-20

1.00E+01 1.00E+02 1.00E+03 1.00E+04

Si
 [

A
2
/H

z]

f [Hz]

Measured values

1/f noise

Thermal noise

Noise spectra of the Lorentzian shape

Noise spectra of the Lorentzian shape

Noise spectra of the Lorentzian shape

Fitted results



 High-Voltage Surge Impact on Thick-Film Sensors for Structural Health Monitoring... 267 

4. CONCLUSION 

The experimental investigation demonstrates that thick-film strain sensors based on 

bismuth lead ruthenate resistive films exhibit significant electrical stability and preserved 

functional integrity under high-voltage surge conditions. When subjected to a 300 μm 

substrate deflection and exposed to controlled sequences of high-voltage pulses, the sensors 

initially exhibited a reduction in resistance, followed by stabilization. This behaviour 

indicates the sensors’ capacity to reach an electrically stable state despite prolonged electrical 

stress. Throughout the testing procedure, the gauge factor remained unchanged, confirming 

that the strain sensitivity and electromechanical coupling performance were not compromised. 

The observed resistance evolution under high-voltage excitation can be attributed to a shift 

in conduction mechanisms: low-ohmic compositions transitioned from contact-dominated to 

tunnelling-dominated conduction, while high-ohmic compositions exhibited a reverse trend. 

These interpretations are supported by parameter 𝑝 values derived from resistance data and 

applied within low-frequency noise analysis. 

Current noise spectrum measurements revealed a marked increase in 1/f noise following 

high-voltage exposure, indicating the formation or activation of additional trap states 

within the insulating glass matrix. This dominant 1/f noise component - primarily arising 

from fluctuations in potential barrier heights and trap occupancy - is accompanied by three 

Lorentzian components, each associated with localized trap dynamics in individual metal-

insulator-metal junctions. While resistance measurements offer insight into macroscopic 

conduction behaviour, noise spectroscopy proved more sensitive to subtle microstructural 

changes, particularly those involving conduction path redistribution and trap state 

dynamics. These results emphasize the critical role of low-frequency noise analysis in 

assessing the long-term reliability of thick-film sensors, especially in structural health 

monitoring applications where high-voltage transients are expected.  

Future research should address long-term reliability of thick-film strain sensors under 

continuous surge exposure by combining accelerated testing with noise spectroscopy. 

Monitoring the evolution of 1/f noise and Lorentzian components during sustained electrical 

stress would allow early detection of microstructural degradation, such as redistribution of 

conducting contacts, barrier height variations, and trap-state evolution within metal–insulator–

metal junctions. Such studies could establish quantitative links between surge severity, 

microstructural changes, and sensor lifetime, providing a basis for predictive maintenance 

strategies and improved sensor design. 

Acknowledgement: The authors are grateful to the Ministry of Science, Technological Development 

and Innovations of the Republic of Serbia (Contract No. 451-03-136/2025-03/ 200017 and research 

project No. 200092) for the financial support. 

REFERENCES  

[1] S. Hassani, U. Dackermann, "A Systematic Review of Advanced Sensor Technologies for Non-Destructive 

Testing and Structural Health Monitoring", Sensors, vol. 23, p. 2204, 2023.  

[2] M. Ju, Z. Dou, J.W. Li, X. Qiu, B. Shen, D. Zhang, F.Z. Yao, W. Gong, K. Wang, "Piezoelectric Materials 
and Sensors for Structural Health Monitoring: Fundamental Aspects, Current Status and Future 

Perspectives", Sensors, vol. 23, p. 543, 2023. 

[3] Z. Deng, M. Huang, N. Wan, J. Zhang, "The Current Development of Structural Health Monitoring for 
Bridges: A Review", Buildings, vol. 13, p. 1360, 2023. 



268 Z. STANIMIROVIĆ, A. STANIMIROVIĆ, A. SAVIĆ, I. STANIMIROVIĆ 

[4] R. Ottermann, E. Müller, M. Keßler, F. Dencker, D. Klaas and M. C. Wurz, "Polymer-Free Batch 
Production and Application of Metal Foil-Based Thin-Film Strain Gauges", In Proceedings of 2024 IEEE 

SENSORS, Kobe, Japan, 2024, pp. 1–4. 

[5] A. Wilson, P. McMahon, R. Muscat, O. Vargas, P. Vincent, "Thin metal foil sensors", In Proceedings of 
SPIE, Smart Structures, Devices and Systems III, Adelaide, Australia, 2007, 64140C.  

[6] I. Dhingra, G. Kaur, R.S. Kaler, "Design and analysis of fiber Bragg grating sensor to monitor strain and 

temperature for structural health monitoring", Opt Quant Electron, vol. 53, p. 619, 2021. 
[7] F.I. Harger Sakiyama, F. Lehmann, H. Garrecht, "Structural health monitoring of concrete structures using 

fibre-optic-based sensors: a review", Magazine of Concrete Research, vol. 73, no. 4, pp. 174–194, 2021. 

[8] D.S. Aulakh, S. Bhalla, "3D torsional experimental strain modal analysis for structural health monitoring 

using piezoelectric sensors", Measurement, vol. 180, p. 109476, 2021. 

[9] R.K. Langat, W. Deng, E. De Luycker, A. Cantarel, M. Rakotondrabe, "In-situ piezoelectric sensors for 

structural health monitoring with machine learning integration", Mechatronics, vol. 106, p. 103297, 2025. 
[10] M. Wen, X. Guan, "Investigation of sensing performances of thick-film resistors sintered on 

fluorophlogopite glass-ceramic substrates", Sensors and Actuators A: Physical, vol. 370, p. 115248, 2024.  

[11] S.A.A. Jabir, N.K. Gupta, "Thick-Film Ceramic Strain Sensors for Structural Health Monitoring", IEEE 
Transactions on Instrumentation and Measurement, vol. 60, no. 11, pp. 3669-3676, 2011. 

[12] S.A.A. Jabir, N.K. Gupta, "Condition monitoring of the strength and stability of civil structures using thick 

film ceramic sensors", Measurement, vol. 46, no. 7, pp. 2223–2231, 2013. 
[13] B. Morten, M. Prudenziati, "Handbook of Sensors and Actuators, Section IV: Sensors for mechanical 

quantities", ed. M. Prudenziati (Elsevier Science Ltd., Amsterdam, 1994), pp. 189–193, ISBN: 

9780444897237. 
[14] F. Tondolo, A. Cesetti, E. Matta, A. Quattrone, D. Sabia, "Smart reinforcement steel bars with low-cost 

MEMS sensors for the structural health monitoring of RC structures", Construction and Building 

Materials, vol. 173, pp. 740–753, 2018. 
[15] C.G. Karayannis, C.E. Chalioris, G.M. Angeli, N.A. Papadopoulos, M.J. Favvata, C.P. Providakis, 

"Experimental damage evaluation of reinforced concrete steel bars using piezoelectric sensors", 

Construction and Building Materials, vol. 105, pp. 227–244, 2016. 
[16] S. Munstermann, Y. Feng, W. Bleck, "Influencing parameters on elastic modulus of steels", Canadian 

metallurgical quarterly, vol. 53, no. 3, p. 264, 2014. 

[17] S.G. Yousef, J. Rodel, E.R. Fuller Jr., A. Zimmermann, B.S. El-Dasher, "Microcrack Evolution in Alumina 
Ceramics: Experiment and Simulation", Journal of the American Ceramic Society, vol. 88, p. 2809, 2005. 

[18] H. Yildirim, O. Sengul, "Modulus of elasticity of substandard and normal concretes", Construction and 

building materials, vol. 25, no. 4, pp. 1645–1652, 2011. 
[19] M.M. Jevtić, Z. Stanimirović, I. Mrak, "Low-Frequency Noise in Thick-Film Resistors due to Two-Step 

Tunneling Process in Insulator Layer of Elemental MIM Cell", IEEE Transactions on Components, 

Packaging and Manufacturing Technology-Part A, vol. 22, no. 1, p. 120, 1999. 
[20] C. Grimaldi, T. Maeder, P. Ryser, S Strässler, "A random resistor network model of voltage trimming", 

J.Phys.D: Appl.Phys., vol. 37, pp. 2170–2174, 2004. 
[21] I. Mrak, M.M. Jevtić, Zdravko Stanimirović, "Low-frequency Noise in Thick-film Structures Caused by 

Traps in Glass Barriers", Microelectronics Reliability, vol. 38, pp. 1569–1576, 1998.  

[22] A. Kusy, A. Szpytma, "On 1/f noise in RuO2-based thick resistive films", Solid-State Electronics, vol. 29. 
pp. 657–665, 1986. 

[23] T. Kleipenning, "On low-frequency noise in tunnel junctions", Solid-State Electronics, vol. 25, p. 78, 1982. 

[24] M. Prudenziati, B. Morten, A. Masoero, "Excess noise and refiring processes in thick-film resistors", J. 
Phys. D: Appl. Phys., vol. 14, no. 7, pp. 1355–1362, 1981.