K. E.
Belsey
a,
A. V. S.
Parry
b,
C. V.
Rumens
a,
M. A.
Ziai
c,
S. G.
Yeates
*b,
J. C.
Batchelor
*c and
S. J.
Holder
*a
aFunctional Materials Group, School of Physical Sciences, University of Kent, Canterbury, Kent CT2 7NH, UK. E-mail: S.J.Holder@kent.ac.uk
bSchool of Chemistry, University of Manchester, Manchester, M13 9PL, UK. E-mail: Steven.Yeates@manchester.ac.uk
cSchool of Engineering and the Digital Arts, University of Kent, Canterbury, Kent CT2 7NT, UK. E-mail: J.C.Batchelor@kent.ac.uk
First published on 2nd March 2017
A route to cheap and disposable sensors for the chemical sensing market, with potential applications including monitoring of food spoilage, is reported herein. The sensor is the result of the direct integration of a stimuli-responsive material, poly(dimethylsiloxane) (PDMS), with an electronic component. The printing and sintering of colloidal silver ink solutions onto PDMS was optimized to allow the printing of conductive silver feed loops, which are the active sensing component in antennas for passive (battery-free) Radio Frequency Identification (RFID) tags. The response of these devices is related to the degree of swelling of the PDMS, which, in turn, has been shown to be correlated to the Hansen solubility parameters and the vapour pressures of the corresponding volatile organic compounds (VOCs). When exposed to solvent vapour the printed feed loop fractures, increasing resistance and ultimately breaking conductivity, leading to a change in the transmitted power and read range of the wireless device. Remarkably upon removal from the vapour, the fractured feed loops reassemble and become conductive again, making them switchable and “multi-use”. This work paves the way to a fully inkjet printed RFID substrate for vapour detection.
Previous work on RFID sensors has seen the incorporation of stimulus responsive materials into a range of antenna designs with a significant number of these in recent years being passive.5 A number of approaches to humidity sensing have been reported where varying degrees of specificity of the sensing substrate to water vapour have been reported.6–11 For example changes to a PEDOT:PSS (poly[3,4-ethylenedioxy-thiophene]:poly[styrene sulfonic acid]) substrate conductivity on exposure to humidity caused an RFID antenna response to change proportionally relative to the degree of water absorption.8 However passive RFID chemical sensors with sensitivity to organic vapours tend not to rely upon substrate sensing components that display specificity. More commonly multivariate analysis of the signal response is employed to determine vapour identity.12–14 A significant step towards a cheap cost-effective mass production of passive sensing tags was demonstrated by Potyrailo et al.15 where thin polyurethane sensing films were applied to RFID tags on polyethylene terephthalate substrates by a roll-to-roll process. These tags were used to analyze impedance detuning between the tag and reader coils when exposed to vapours such as acetone, ethanol, and toluene. Principal component analysis of the resulting signals allowed for discrimination between the vapours. Other short range passive tags with an array of vapour-sensing elements were created from conducting carbon black and various polymers as the sensor materials and were able to identify water, ammonia, ethanol and toluene vapours.16 Here chemical sensing on a conventional passive RFID tag was performed by measuring tag reflectance or resistance properties and again principal component analysis was used to distinguish chemical vapours. A number of good overviews and reviews of RFID chemical sensors are available in the literature.2,3,17
In contrast we propose a disposable passive RFID sensor system in the UHF band for a long read range, typically several metres, which is sensitive to a broad range of vapours utilizing a stimuli-responsive polymer substrate but where the stimuli-response is mechanical. This is similar in principle to the sensor mechanism as described by Fiddes and Yan16 but rather than measuring small differences in conductivity, the system we describe behaves as an on–off switch where electrical conductivity in an antenna component is essentially stopped by a mechanical response from the substrate.
The stimulus responsive material is PDMS which has the ability to absorb both volatile and non-volatile compounds with a significant degree of swelling18 making it an ideal candidate for a mechanical sensing component in RFID tag sensors. We have already reported its use as the active component in an RFID vapour sensor, whereby upon swelling in the vapour the PDMS acted as an actuator by displacing a feed loop that modified the antenna response.19 However, the displacement approach used in the previous work has limitations particularly with respect to manufacture of the sensor and versatility in application. To construct cheap passive sensor tags a move towards an entirely additive manufacturing approach is required. We have designed an RFID sensor based on the same principle (swelling of PDMS as a mechanical actuator) but constructed via ink-jet printing of the feed loop directly onto the PDMS substrate.
Whilst the inkjet printing of silver colloidal solutions has been well reported,20–23 the deposition of material onto PDMS from solution is a considerable challenge due to its intrinsic hydrophobic nature which can be improved by rendering the surface more hydrophilic through the use of oxygen, argon and air plasma along with UV-ozone.24–27 UV-ozone treatment was seen to improve the printing of silver inks onto highly stretched Sylgard-184 PDMS.26 while air plasma treatments were used to a similar effect in order to create silver patterns on Sylgard-184 coated glass slides.27
This work will demonstrate that, using inkjet technology, the loop for an RFID tag can be printed from conductive ink. This not only indicates a step towards developing an entirely printed tag, but results in a system where the tag can be used repeatedly whilst maintaining a clear RFID response to vapour environments.
Fig. 1 Schematic of antenna, PDMS substrate and printed silver loop which make up the RFID tag (all dimensions are in mm). |
The RFID sensor shown in Fig. 1 comprises four elements: (i) an antenna for RFID wireless communication, (ii) a sensing impedance transducer, (iii) a RFID integrated circuit (transponder chip), and (iv) a stimuli responsive polymer (PDMS). The antenna is tuned to the UHF RFID band, the transponder chip modulates information in the reflected signal, and the feed loop connects the antenna to the chip. In this passive RFID system, the tag is energized entirely by power provided from an external reader which in turn also receives the reflected tag signal. The purpose of the feed loop is to transform the impedance of the antenna so that it matches that of the transponder chip and ensures maximum power transfer between the two.28
Maximum power transfer between the feed loop and the transponder chip is essential to obtain the minimum amount of reader power required to activate the tag, and this is strongly affected by the conductivity of the loop and its exact position with respect to the dipole antenna.28 Therefore, a change in the electrical properties of the feed loop will cause the tag activation power to vary and this can be detected by the external reader. The majority of the tag surface area is covered by the antenna. The dipole width determines the bandwidth of operating frequencies while the end loading elements serve to reduce the overall tag length to below one half wavelength and make the tag more compact. In addition to the sensing capacity, the tag design should enable omni-directional radiation performance,32 with simple 2D manufacture whilst maintaining a relatively small antenna size.
Having defined the sensor antenna design, integration of the feed loop with the stimuli responsive polymer was necessary. PDMS swells in the presence of organic vapours and it was expected that either the expansion of the PDMS would disrupt the conductivity of the printed conducting loop track and/or that the dielectric properties of the PDMS substrate would change due to the presence of significant quantities of organic solvent bound in the PDMS. In either case the efficiency of the tag response would be altered and detected by an external reader because the level of transmitted power required to activate the sensing tag changes in proportion with the tag efficiency, which itself is a function of the exposure to the vapour stimulus. The increase in transmitted power required is reflected in a decrease in effective read range for the tag. The tag efficiency is defined as the amount of power radiated by the tag antenna divided by the power flowing at its terminals connecting to the transponder chip.
To enable the printing of the silver ink the PDMS was rendered hydrophilic by UV-ozone treatment following techniques reported in the literature. Fig. 2 shows the ATR-FTIR spectra of the PDMS before and after UV-ozone treatment. The appearance of the broad absorption around 3300 cm−1 and the absorption at 900 cm−1 demonstrated a large increase in the number of Si–OH bonds at the surface. After 24 hours the spectrum returned to its untreated state indicating the reorganization of the siloxane surface by the internalization of the polar Si–OH groups and the presentation of the Si–CH3 groups at the surface once more. Such behavior of polysiloxane surfaces has been well documented.33,34 With increased UV-ozone exposure times there is a predictable decrease in contact angle (Fig. 2 inset) followed by a recovery that never fully returns to its original value. Interestingly for long exposure times the recovery is quicker. The surface of the PDMS also becomes more brittle with longer exposure, with visible cracks appearing over the surface. In order to compensate for these effects a UV-ozone exposure time of 25 minutes was chosen for the preparation of PDMS used in the printing experiments. The contact angle for the silver ink after UV-ozone exposure was reduced to 34.0 ± 0.3°, a printed drop had a radius of approx. 50 μm (Fig. S2, ESI†) and patterns were readily obtainable.
Fig. 3 gives an overview of the findings of the optimization of printing silver ink onto the PDMS squares (see also Fig. S3, ESI†). Several clear trends were noticeable. Increasing the sintering temperature generally resulted in lower resistances with no significant improvement in performance discernible at temperatures higher than 150 °C (Fig. 3a and d). Increasing the number of layers of ink deposited (1, 2 or 3) led to decrease in resistance, corresponding to an increase in material and thickness of the printed line (Fig. S4, ESI†). Although a further increase in the number of layers printed could reduce the resistance further, factoring in material usage, cost and time led to the choice of a maximum of 3 layers for the creation of feed loops. Finally, drop spacing had a significant effect on the sheet resistance (Fig. 3b and e), with a smaller drop spacing giving a much improved performance. However, this had to be correlated with the quality of the printed pattern as demonstrated in Fig. S5 (ESI†). For drop spacings lower than 35 μm, bulges appeared in the printed lines for both 1 and 3 layers, increasing the average width of the line.
Combining these findings, feed loops were created using a drop spacing of 35 μm with three printed layers and a sintering temperature of 150 °C. Of a batch of 24 loops printed under these optimized conditions, 19 of them were seen to have resistances of less than 20 Ω and were considered suitable for incorporation into tags (Table S5, ESI†). Observations of buckling along the printed lines (Fig. S5, ESI†) is consistent with previous reports and is presumed to be due to expansion and contraction of the siloxane during the heating process, but was not seen to detrimentally affect RFID performance. Typical printed feed loops are shown in Fig. 3g and h.
Ra = (4 × (δdsolvent − δdPDMS)2) + (δpsolvent − δpPDMS)2 + (δhsolvent − δhPDMS)2 | (1) |
Qv = a + (bδd × δd) + (bδp × δp) + (bδh × δh) + (bVP × VP) | (2) |
(3) |
For the present study a number of solvents were chosen for testing based on their predicted ability to swell PDMS. Diethyl ether and dichloromethane swell PDMS very well, acetone and xylene moderately well, and methanol and propan-2-ol not at all. Additionally, we chose to test the sensors with ethyl acetate, acetaldehyde and acetic acid as representative food degradation products.35–38 The solubility parameters, vapour pressures, Ra values and predicted relative equilibrium swelling volumes of the test solvents are given in Table 1.
Solvent | δ d | δ p | δ h | mmHg | R a | Predicted Qv |
---|---|---|---|---|---|---|
Acetaldehyde | 14.7 | 8 | 11.3 | 760.00 | 10.6 | 3.88 |
Acetic acid | 14.5 | 8 | 13.5 | 15.53 | 12.2 | 1.31 |
Acetone | 15.5 | 10.4 | 7 | 231.02 | 10.6 | 1.96 |
DCM | 18.2 | 6.3 | 6.1 | 436.54 | 7.8 | 2.68 |
Diethyl ether | 14.5 | 2.9 | 5.1 | 537.79 | 4.0 | 3.68 |
Ethyl acetate | 15.8 | 5.3 | 7.2 | 94.51 | 5.8 | 1.81 |
Methanol | 15.1 | 12.3 | 22.3 | 126.76 | 21.5 | 1.07 |
Propan-2-ol | 15.8 | 6.1 | 16.4 | 45.15 | 13.2 | 1.34 |
Xylene | 17.8 | 0.9 | 1.8 | 8.48 | 4.8 | 1.77 |
PDMS | 15.9 | 0.1 | 4.7 | — | — | — |
The sensor tags were placed in a sealed glass vessel (a desiccator with flat glass plate lid) with an excess of the solvent in question at the base of the vessel. Transmitted power measurements were taken from the moment of enclosure. To measure recovery, the vessels were opened to the air and the tag removed from the saturated vapour environment.
Tags exposed to both diethyl ether, DCM, ethyl acetate and acetaldehyde vapours led to a rapid increase in required transmitted power reaching a maximum value in less than 5 minutes (Fig. 4a and Fig. S6–S8, ESI†). Accompanying the increase in transmitted power was a corresponding reduction in read range. Further, the recovery time after removal from the solvent vapour was typically less than 10 minutes. Remarkably, repeated cycles of exposure led to repeatable behavior in the tag response and a single tag could be exposed to vapour multiple times (Fig. 4b and Fig. S6–S11, ESI†) with the accompanying change in transmitted power and read range. In ‘average’ swelling solvents, such as acetone and xylene, the length of vapour exposure necessary to reduce the performance of the tag was longer (Fig. 4c and Fig. S9, ESI†). In contrast to the above results methanol, propan-2-ol and acetic acid solvents that do not swell PDMS appreciably gave essentially no response over extended periods of exposure (Fig. 4d and Fig. S10, S11, ESI†). The performance of the tags was seen to slowly decrease over much longer periods of time (up to 72 hours), but was not seen to cause enough disruption to stop the tag from responding entirely.
The change in the normalized response (based on a normalized scale of 100% before vapour exposure, Fig. S12 and S13, ESI†) as a factor of response time for each tag to reach minimum read range for the different vapours could be correlated with the predicted equilibrium swelling (Fig. 5). A strict comparison between relative response rates and predicted equilibrium swelling is not valid since the response rate will depend upon kinetic factors that are not part of the equilibrium swelling prediction model. The diffusion rates can vary considerably between vapours and this is particularly apparent for xylene which has been observed to show significantly lower diffusion rates in PDMS than other solvents. Thus while superficially the expected response for xylene should be similar to that of ethyl acetate a clear difference exists.
Fig. 5 (a) Predicted equilibrium swelling values for PDMS in various solvent vapours and (b) measured response rates for tag response to vapours. |
There is a clear correlation between tag response and the properties of the solvent vapour the tag is exposed to, but it is also evident that the tag does not break after successive swelling cycles. As such, it became necessary to ascertain the mechanism by which these sensors were working.
To monitor the changes in the feed loop on PDMS when exposed to vapour a tag was exposed to diethyl ether vapour and observed under a microscope. With increasing degree of vapour exposure it became obvious that the silver films constituting the loop cracked and fractured with the swelling of the PDMS substrate leading to increasing loss of contact between sections of the silver. This is best illustrated by the inverse images of a portion of loop before exposure, after 30 minutes exposure and then left to recover in air (Fig. 6, further microscope images are given in Fig. S14, ESI†). As can be seen gaps appear between ‘flakes’ of the silver with solvent exposure and these appear to rejoin as the vapour desorbs from the PDMS in air. Despite the ‘flakey’ appearance of the loops, the silver remained attached to the substrate during this process.
Resistance values were taken in tandem with the RFID measurements with swelling and a direct correlation was observed between the two variables as shown in Fig. 7 (see also Fig. S15 and S16, ESI†). After a very short period in diethyl ether vapour, ca. 5 minutes, the resistance of the loop was seen to dramatically increase, until no reading could be obtained at all. This matched the length of time taken for the loop to require 26.5 dBm transmitted power in order to generate a response, which is the limit the RFID chip sensitivity. When the tags were removed from the solvent atmosphere, both the resistance and the required transmitted power were seen to decrease, as the PDMS contracted and the performance of the tag improved. The resistance initially decreased very rapidly, but then was observed to ‘tail off’, steadily returning to its original value. An RFID signal was consistently regained at approximately 24 Ω. Since variation in dielectric constant does not affect track resistance we can attribute the dominant sensing mechanism in these tags to the generation of cracks in the silver film during substrate expansion leading to increased resistance in the feed loop. This demonstrates that the response to swelling is not caused by any variation in the dielectric constant of the PDMS. Remarkably electrical contact is remade during the evaporation of the solvent and relaxation of the substrate.
Fig. 7 Transmitted power and resistance (measured concurrently) against time for an RFID tag in diethyl ether – the grey box indicates time that the loop was exposed to vapour. |
An increase in necessary transmitted power upon vapour exposure and a decrease in power after removal of the tag from the vapour continued through multiple cycles for all solvent vapours. Fig. 8 shows the maxima and minima transmitted power values for each solvent for each cycle. Whereas diethyl ether, ethyl acetate and acetaldehyde showed no significant degradation in performance acetone displayed a gradual reduction in necessary minimum power over 6 cycles and dichloromethane showed a noticeable reduction in performance after the 4th exposure cycle. The reasons for some degradation in performance for acetone and dichloromethane are currently unknown. As of yet the maximum number of cycles that these tags can undergo before failure or significant loss in sensitivity has not been tested. One of the applications initially envisaged for these sensors was in the area of food spoilage. Unfortunately the bulk of recent published research on spoilage has focused on the specific identification of volatile compounds indicators for electronic node type detectors rather than quantification of absolute amounts.42–44 However previous studies indicate that individual volatiles are produced in the 10s to 100s ppm (w/w or v/v) concentrations in the early stages of food spoilage; though given the large range of different foodstuffs, food lifetimes and degradation mechanisms these values can only be taken as indicative.45–47 However such concentrations of volatiles by themselves are too low to be detected by the current tag sensors described in this work. One approach to increase sensitivity will be to reduce the cross-link density of the polysiloxane matrix enabling larger degrees of swelling for smaller quantities of absorbed volatile. Secondly the absorption of a number of different volatiles simultaneously, will dramatically increase sensitivity; though the practicality of this will depend upon the foodstuff and packaging in question.
These tags are shown to swell when placed in a solvent vapour atmosphere, which results in a dramatic increase in resistance across the printed loop. This corresponds to a significantly decreased (and ultimately, eradicated) read range, with a much higher transmitted power required to receive a response from the tag. Results from this study suggest that the large increase in resistance with swelling is a direct consequence of the fracturing/cracking of the printed silver loop that accompanies the swelling of the PDMS.
The reversibility of this solvent detection is demonstrated by the tag recovering from the loss of read range and increased transmitted power, returning to a working RFID tag after a period of time away from solvent vapour. The length of time taken for the tag to stop responding, and to return to life again, is highly dependent on the organic vapour that it is being exposed to. This selectivity will be further improved by the addition of functional groups to the siloxane network. This system could be used for numerous applications, such as the detection of food spoiling both on a small scale (in consumer's homes) and on a much larger scale (degradation of food during processing and transport). Further work is being carried out to enable this system to be made entirely by additive manufacturing.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6tc05509e |
This journal is © The Royal Society of Chemistry 2017 |