Carbon Dioxide Sensing-Biomedical Applications to Human Subjects

Abstract

Carbon dioxide (CO2) monitoring in human subjects is of crucial importance in medical practice. Transcutaneous monitors based on the Stow-Severinghaus electrode make a good alternative to the painful and risky arterial "blood gases" sampling. Yet, such monitors are not only expensive, but also bulky and continuously drifting, requiring frequent recalibrations by trained medical staff. Aiming at finding alternatives, the full panel of CO2 measurement techniques is thoroughly reviewed. The physicochemical working principle of each sensing technique is given, as well as some typical merit criteria, advantages, and drawbacks. An overview of the main CO2 monitoring methods and sites routinely used in clinical practice is also provided, revealing their constraints and specificities. The reviewed CO2 sensing techniques are then evaluated in view of the latter clinical constraints and transcutaneous sensing coupled to a dye-based fluorescence CO2 sensing seems to offer the best potential for the development of a future non-invasive clinical CO2 monitor.

Keywords: CO2; carbon dioxide; paCO2; ptCO2; tcpCO2; transcutaneous monitoring.

Figure 1

Mid infra-red absorption spectra of various gases, only CO${}_2$ absorbs at 4.26 µm. Data source: HITRAN database [28]. A Lorentzian broadening profile was considered for a dilution in air at 1 atm and 296 K.

Figure 2

(Left) outline schematic of a space-interleaved (top) and time-interleaved (bottom) NDIR CO${}_2$ sensor. In the time-interleaved sensor, a rotating filter wheel acts as a chopper, with two opaque sectors, and two sectors equipped with 4.3 µm and 3.9 µm bandpass filters, respectively. (Right) a more complex design allowing for longer light paths in a compact device. The detector has two channels, a measurement and a reference one, as in the upper left scheme. CPC stands for Compound Parabolic Collector, a type of light concentrator. Reproduced with permission from Hodgkinson et al. [30].

Figure 3

Outline schematic of an organ-pipe-like resonant acoustic cell. The cell consists of a pipe closed at its two ends with optical windows. A laser beam at 4.26 µm is then pulsed at the resonant frequency of the pipe, which forms a $\lambda /2$ resonator. For instance, if a 32,768 Hz quartz tuning fork is used, $\lambda \approx 10$ mm and a pipe length of ∼5 mm would be ideal. A velocity-sensitive microphone may be placed at half the pipe as depicted. Alternatively, a pressure-sensitive microphone would rather be placed near one of its ends.

Figure 4

Outline of a static conductometric cell. CO${}_2$ diffuses into distilled water, modifying the concentration of H${}_3$O${}^{+}$, HO${}^{-}$ and HCO${}_3^{-}$ and thus the conductivity of the solution, which may be measured using alternative current to avoid polarisation by mean of two electrodes.

Figure 5

(Left) the Severinghaus electrode, an improvement of the Stow electrode, as first described in the 1958 publication [67]. The colour of the different elements matches that of the simplified schematic on the right. (Right) outline schematic of the Severinghaus electrode. CO${}_2$ diffuses through a PTFE membrane into an electrolyte, causing a change in the pH of the latter. Such a change is then recorded by the mean of a pH-meter that consists of a glass electrode and a reference electrode.

Figure 6

(Left) outline schematic of an ISFET CO${}_2$ sensor. CO${}_2$ diffuses from the analyte inside the inner electrolyte through a PTFE membrane, changing the pH of the electrolyte and generating hydronium ions (abbreviated H${}^{+}$). The ions will penetrate the upper layer of the porous SiO${}_2$ gate insulator, influencing the P${}^{+}$ substrate underneath, and thus the conductance of the transistor. (Right) basic electrical implementation schematic of an ISFET sensor. The actual grid-source potential seen by the transistor is the sum of the applied grid voltage $V_G$, the voltage between the reference electrode and the electrolyte $V_{RE}$, and that between the electrolyte and the gate insulator $V_{EG}$, itself function of the pH of the electrolyte.

Figure 7

(Left) outline schematic of a typical dye-based CO${}_2$ sensor. The LEDs illuminate a pH-sensitive dye, either dissolved in an aqueous solution or a polymer, through a layer of transparent substrate—here a glass slide. The dye is protected from ionic contamination by a PTFE, gas-permeable, ion-impermeable membrane. The LEDs and photodiode may additionally be covered by excitation or emission filters in case of fluorescence measurement. (Right) the fluorescence excitation spectra of 1 mM HPTS in carbonated distilled water equilibrated with different pCO${}_2$ values. Data source: Uttamlal et al. [90].

Figure 8

“Various techniques that have been employed in CO${}_2$ sensors to cause light propagating in the core to interact with the surrounding environment: (a) [Long Period Grating]; (b) extrinsic Fabry-Perot cavity; (c) [Fiber Bragg Grating] and; (d) etched cladding. All the examples here use a material which demonstrates CO${}_2$ sensitivity and the subsequent change in coating upon exposure to CO${}_2$ is sensed by the interacting light”—Text and figure reproduced with permission from Barrington [20].

Figure 9

Outline schematic of a potentiometric CO${}_2$ sensor as described by Maruyama et al. [210]. The NASICON sandwich with one end covered in sodium carbonate makes an electrochemical cell, the electromotive force of which yields pCO${}_2$.

Figure 10

Outline schematic of a time of flight CO${}_2$ sensor as described by Joos et al. [245]. Two ultrasound (40 kHz) transducers—emitter (Tx) and receiver (Rx)—are placed at both ends of an acoustic chamber which may consist in a simple tube. A burst of ultrasound is emitted at $t_0$ (on the left) and arrives at $t_0+\Delta t$ at the receiver (on the right). The $\Delta t$ value can then yield the percentage of CO${}_2$ in the gas mixture using velocity-based calculations.

Figure 11

Acoustical absorption spectra of diverse nitrogen mixtures. Data source: Petculescu et al. [250].

Figure 12

The three in vivo CO${}_2$ probing modalities and their key features.

Figure 13

A capnogram yields important clinical clues on the state of the patient or quality of their intubation by analysing the different breathing phases—inspiration (I), and expiration: dead-space volume (II), mixed dead-space and alveolar air (III) and alveolar air (IV)—and the two $\alpha$ and $\beta$ angles [292,293]. More narrowly, capnometry is only interested in knowing the more concise end-tidal pCO${}_2$ value reached at the end of the plateau (IV): petCO${}_2$.

Figure 14

(Left) crude closed-chamber sensor model. Please note the colour matching between this schematic and previous Figures—e.g., Figure 4, Figure 5, Figure 6 and Figure 7—blue for the analyte medium, yellow for the sensor itself. (Right) typical evolution of $p_{Se}$CO₂ against time when applying the sensor against the skin.

Figure 15

95% response time ($T{r}_{95\%}\approx 3·\tau$) of a closed-chamber sensor of height $h_{Se}$. The blue area underlines the portion of the line with a response time below 10 min, which corresponds to a sensor height below approximately 100 µm.

Figure 16

The dye-based fluorescent transcutaneous sensor consists of two parts. (Top) a sensing head with the LEDs and photodiode covered with their emission and reception filters (bright purple and green rectangles) and enclosed in a solid body, protected by a covering glass. (Bottom) a multilayer sensing patch stuck on the skin, consisting in—from top to bottom—a transparent layer, impermeable towards CO${}_2$—e.g., PET—a layer of CO${}_2$ sensitive dye in a polymer matrix, a layer of ion-impermeable, CO${}_2$-permeable material such as PTFE for CO${}_2$ diffusion from the skin into the dye, and a thin adhesive layer to maintain the patch onto the skin. See Section 2.3.4 for a detailed review of material choices as well as their advantages and drawback, when designing a dye-based CO${}_2$ sensor.