CMOS-Based Implantable Multi-Ion Image Sensor for Mg²⁺ Measurement in the Brain

Highlights

• An implantable multi-ion image sensor equipped with magnesium-(Mg²⁺) and calcium-(Ca²⁺)-sensitive membranes was successfully fabricated.
• Selective Mg²⁺ measurement and multi-imaging of Mg²⁺ and Ca²⁺ were demonstrated.

Abstract

An implantable multi-ion image sensor equipped with magnesium ion (Mg²⁺)-and calcium ion (Ca²⁺)-sensitive membranes was fabricated for the selective measurement of extracellular Mg²⁺ in the brain, and the sensor performance was evaluated. This sensor complements the low selectivity of the Mg²⁺-sensitive membrane for Ca²⁺ by depositing a Ca²⁺-sensitive membrane in addition to the Mg²⁺-sensitive membrane on a CMOS (Complementary Metal Oxide Semiconductor)-based potentiometric sensor array with 5.65 × 4.39 µm² pitch, enabling selective measurement of Mg²⁺ and Ca²⁺. Characterization of the sensor confirmed a Ca²⁺ sensitivity of 26.5 mV/dec and Mg²⁺ sensitivity of 19 mV/dec. Based on validation experiments with varying concentrations of Mg²⁺ and Ca²⁺, selective Ca²⁺ and Mg²⁺ measurements were successfully achieved. Furthermore, real-time imaging of Mg²⁺ and Ca²⁺ and quantification of their concentration changes were performed. The developed sensor may be successfully applied for extracellular multi-ion imaging of Mg²⁺ and Ca²⁺ in the living brain.

1. Introduction

Neurotransmitters and ions in the brain have been linked to neurological diseases and are essential for understanding brain functions [1,2]. Whereas normal brain functions are performed by the complex interplay of these transmitters, it has been suggested that abnormal transmission and changes in their concentration induce neurological disorders, such as Alzheimer’s disease and epilepsy [3,4].

In recent years, the aging of Japanese society has been one of the most pressing social issues, and the percentage of elderly people in Japan in 2022 was 29.0%. This is well above the 7.0% percentage of elderly people that defines an aging society. Excessive aging is accompanied by an increasing prevalence of memory-related neurological diseases, such as dementia and Parkinson’s disease. Numerous studies have highlighted the association of magnesium ions (Mg²⁺) with dementia, which impairs attention, memory, and psychomotor functions [5,6]. In one of these studies, the intraperitoneal administration of 270 mg/kg magnesium sulfate in rats during inhibitory avoidance training improved their impaired learning ability [7]. However, most of these studies have shown experimental results; the sites and pathways of action of Mg²⁺ are poorly understood, and their mechanisms must be elucidated before they can be applied in biomedicine.

In recent years, imaging studies have been actively conducted to observe the interior of animal brains under in vivo conditions. If extracellular Mg²⁺ in the brains of living individuals can be visualized and analyzed in real time, it is expected to significantly contribute to the elucidation of neurological diseases and drug discovery [8,9]. Fluorescence imaging [10] and nuclear magnetic resonance [11] are mainly used to observe cells in vivo and measure ion concentrations in the brain. However, these techniques have problems such as “difficulty in observing the original activity of living organisms due to labeling”, “difficulty in long-term observation”, “low spatio-temporal resolution”, and “high cost”. Therefore, it is crucial to develop imaging technology to visualize and analyze information in the brains of living individuals in a simple and precise manner.

As an alternative approach, the polyvinyl chloride (PVC) membrane-based Mg²⁺-sensitive electrode [12,13,14] using an ion recognition ionophore has been reported. These electrochemical devices enable the label-free real-time recording of Mg²⁺ response in a solution. Additionally, various types of ions, such as potassium ion (K⁺) [15], calcium ion (Ca²⁺) [16], and sodium ion (Na⁺) [17], can be detected by changing an ionophore corresponding to a target molecule. However, these ion-selective electrodes are not suitable for the spatiotemporal analysis of ion distributions due to a single-point recording.

Our group has developed a 23.55 µm pitch implantable potentiometric biosensor that enables label-free real-time imaging of hydrogen ions (H⁺) distribution [18]. Using this sensor, we previously demonstrated the imaging of changes in extracellular proton dynamics in a living mouse brain [19]. Furthermore, the sensor’s pixel pitch was achieved at 5.65 × 4.39 µm² in recent years, showing a potential for high spatiotemporal H⁺ imaging [20]. Furthermore, we succeeded in measuring calcium ions (Ca²⁺) by immobilizing ionophores on a sensor that selectively detected specific ions [21]. However, the measurement of Mg²⁺, which has been pointed out to be closely related to memory, has not been demonstrated.

In this study, an Mg²⁺-sensitive film that can measure Mg²⁺ and a Ca²⁺-sensitive film that can complement the low selectivity of the Mg²⁺-sensitive film were deposited on an implantable image sensor to fabricate an implantable multi-ion image sensor that can selectively measure Mg²⁺ in the brain. The output characteristics of the fabricated sensor were evaluated. In addition, selective response evaluation experiments were conducted to examine measurement performance.

2. Materials and Methods

2.1. Materials and Chemicals

Polyvinyl chloride (PVC), 2-Nitrophenyl octyl ether (NPOE), tetrakis-boric acid sodium salt (TFPB), tetrahydrofuran (THF), sodium chloride (99.5%), potassium chloride (99.5%), calcium chloride (95%), magnesium chloride, and D-glucose (98%) were purchased from Wako Pure Chemical Industries Ltd. (Osaka, Japan). Calcium ionophore V (K23E1), sodium, and magnesium ionophore I (ETH 1117) were purchased from Sigma-Aldrich (St. Louis, MO, USA). 4-(2-hydroxyethyl) piperazine-1-ethanesulfonate (HEPES) (>99%) was purchased from Dojindo Laboratories (Tokyo, Japan), and all other reagents were prepared using deionized water (18.2 MΩ) produced with a Milli-Q water system (Tokyo, Japan).

2.2. Implantable Ion Image Sensor

Figure 1 shows the general shape and performance of the implantable ion image sensor developed by our research group. This sensor was fabricated by the 0.35 µm CMOS technology. It is designed to be inserted into a living brain and to reduce invasiveness, and the thickness and width of the sensor entry point are 100 µm and 1.1 mm, respectively. As shown in

Table 1. Details for the sensor specification.

Process	0.35 µm CMOS
Sensor size	9.88 × 1.10 mm2
Sensor thickness	100 µm
Sensing area	1124 × 180 µm2
Pixel pitch	5.65 × 4.39 µm2
Pixel array	256 × 32
Frame rate	14.1 fps

, it has an array sensor with a pixel size of 5.65 × 4.39 µm ² and a pixel structure of 256 × 32 pixels, which is sufficient for high-resolution imaging, since the neuronal cell body is about 10~30 µm. A Ta₂O₅ membrane, functioning as a pH-sensitive film, was deposited onto the array.

In addition, a frame rate of 14.1 fps provides a high spatiotemporal resolution. This study aimed to measure extracellular Mg²⁺ in the living brain by depositing a sensitive membrane containing ionophores that trap specific ions on a potential array sensor.

2.3. Proposal of a Multi-Ion Image Sensor

To date, our group has been able to measure ions such as Ca²⁺ and K⁺. Ionophores that can selectively capture the ions to be measured are contained in a polyvinyl chloride (PVC) film solution, and the ion dynamics in the liquid can be measured in real time by depositing the ionophores on the sensor.

Mg²⁺ ionophores that can selectively capture Mg²⁺, the measurement target in this study, also exist. We have been working on the fabrication of an Mg²⁺ image sensor by depositing a Mg²⁺-sensitive film containing these ionophores on the sensing area of an ion image sensor (Figure 2).

We also confirmed that the fabricated Mg²⁺ image sensor is responsive to Mg²⁺. However, the Mg²⁺-sensitive membrane shows high selectivity for brain-coexisting ions such as H⁺, K⁺, and Na⁺ and high response sensitivity to Ca²⁺, which fluctuates in the brain, resulting in inadequate selectivity. This may be because Ca²⁺ is a divalent cation like Mg²⁺, and due to the Hofmeister series, the hydrophilic radius of Mg²⁺ is larger than that of Ca²⁺, making it difficult for the molecular structure of the magnesium ionophore to selectively recognize only Mg²⁺. Therefore, we propose a method in which a Ca²⁺-sensitive membrane that shows high selectivity for coexisting ions in the brain, including Mg²⁺, is painted over the sensing area to form a Mg²⁺-sensitive membrane and a Ca²⁺-sensitive membrane region. Using this method, when the Mg²⁺ concentration changed, only the Mg²⁺-sensitive area responded. When the Ca²⁺ concentration changed, both sensitive areas responded (Figure 3). In other words, the high selectivity of the Ca²⁺-sensitive membrane complements the low selectivity of the Mg²⁺-sensitive membrane for Ca²⁺. It is expected to discriminate between Mg²⁺- and Ca²⁺-dependent potential changes occurring in the Mg²⁺-sensitive membrane.

2.4. Sensor Fabrication

In this study, Mg²⁺-and Ca²⁺-sensitized films were formed regionally on a 32 × 256 pixel puncture-type sensor with a pixel pitch of 5.65 × 4.39 µm. Prior to the membrane fabrication, Mg²⁺ and Ca²⁺ membrane solutions were prepared according to previous reports [21,22]. In the Mg²⁺-sensitive membrane solution, 33.9 mg of 2-nitrophenyl octyl ether (NPOE), 3.1 mg of tetrakis-boric acid sodium salt (TFPB), 16.8 mg of PVC, and 3.1 mg of magnesium ionophore (magnesium ionophore I: Sigma-Aldrich) are dissolved in 0.3 mL of tetrahydrofuran (THF). In the Ca²⁺-sensitive membrane solution, 70.6 mg of NPOE, 0.85 mg of TFPB, 28.6 mg of PVC, and 2.85 mg of calcium ionophore (calcium ionophore V: Sigma–Aldrich) were dissolved in 0.4 mL of THF. The amount of THF, which determines the viscosity of the solution, was two to three times lower in the film-sensitive solution used in this study than that used in previous studies. Previous studies showed a dependence between the ion sensitivity and thickness of the sensitized film, and the thickness of the film was made sufficiently thick by repeatedly coating the sensitized film solution. However, when forming two types of sensitive films in a small area, the expansion and mixing of each sensitive area due to overcoating can be problematic; therefore, the viscosity of the sensitive film solution was increased in this study.

The resulting mixtures were applied to an implantable ion image sensor to form sensitive areas, as shown in Figure 4a. After formation, the films were dried at room temperature (25 °C) for 12 h. A micrograph of the sensor chip surface after the PVC membrane formation is shown in Figure 4b. The Mg²⁺- and Ca²⁺-sensitive membranes were separately deposited on the pixel area.

2.5. Evaluation of the Ion Response Characteristics in Each Sensitive Region

To confirm the concentration dependence of the Mg²⁺- and Ca²⁺-sensitive regions in the fabricated implantable multi-ion image sensor, measurement solutions were prepared by varying the concentration of each ion in a buffer solution that mimicked the brain. The tip of the fabricated sensor was immersed in a buffer solution for at least 6 h to obtain a sensitive membrane. As shown in Figure 5, the ion response characteristics to changes in the concentrations of sodium ions (Na⁺), K⁺, and H⁺, in addition to Mg²⁺ and Ca²⁺, which are the measurement targets, were measured by changing the sensor every 60 s in a Petri dish filled with a concentration measurement solution. The composition of the base buffer solution is potassium chloride (KCl): 2.5 mM, sodium chloride (NaCl): 150 mM, calcium chloride (CaCl₂): 2 mM, magnesium chloride (MgCl₂): 1 mM, glucose: 10 mM, and hydroxyethylpiperazine ethane sulfonic acid (HEPES): 10 mM, and the pH is about 7 pH. The concentration of the ions to be measured varies from 10⁻⁶ to 10⁻¹ M in 1-digit increments, and the pH varies from 7 to 8 pH in 0.5 pH increments.

2.6. Simultaneous Measurement of the Mg²⁺ and Ca²⁺ Response

To verify the performance of the implantable multi-ion image sensor in the detection of Mg²⁺ concentration changes and for quantitative measurements, the potential response was measured by varying the concentrations of Ca²⁺ and Mg²⁺ in the buffer solution. The three buffers used were the same as those used in Section 2.3 for the base. Buffer 1 contained MgCl₂ at 10⁻³ M and CaCl₂ at 10⁻⁴ M, buffer 2 contained MgCl₂ at 10⁻² M and CaCl₂ at 10⁻⁴ M, and buffer 3 contained MgCl₂ at 10⁻² M and CaCl₂ at 10⁻³ M. At the start of the measurement, the sensor was inserted into buffer solution 1. After 100 s, the sensor was transferred to buffer solution 2 and, after another 100 s, to buffer solution 3 (Figure 6). Therefore, it is possible to evaluate the response of this sensor to the Mg²⁺ concentration change in the transition from buffer solution 1 to buffer solution 2 and to the Ca²⁺ concentration change in the transition from buffer solution 2 to buffer solution 3.

3. Results and Discussion

3.1. Evaluation Results of Ion Response Characteristics in Each Sensing Region

The concentration dependences of the Mg²⁺- and Ca²⁺-sensitive regions for each ion are shown in Figure 7. As shown in Figure 7a, the output voltage (V_Out) for the Mg²⁺-sensitive region increased from 10⁻² to M and showed slightly lower sensitivity to Mg²⁺ (19 mV/dec). In the concentration range of 10⁻⁴ to 10⁻¹ M Ca²⁺, the sensitivity was measured to be 29.2 mV/dec. This result shows that the Mg²⁺-sensitive membrane has low selectivity for Ca²⁺, as described in Section 2.3. As shown in Figure 7a, the limit of detection for the Mg²⁺-sensitive membrane is higher than 10 mM, because the V_Out signal of the sensor was saturated less than approximately 10 mM. This indicates that the Mg²⁺ ionophore-entrapped membrane cannot detect extracellular Mg²⁺ fluctuation (~1 mM) due to influence of the interfering ions. Therefore, an improvement of the sensor performance for Mg²⁺ detection is necessary due to the limit of detection for Mg²⁺ being higher than 10 mM at this time. Conversely, V_Out for the Mg²⁺-sensitive membrane had almost no change for the different concentrations of H⁺, K⁺, and Na⁺. Following these experimental results, further investigation of the Mg²⁺ sensor performance for Ca²⁺ is required for applying physiological conditions. On the other hand, the V_Out for the Ca²⁺-sensitive region showed a Ca²⁺ response with increasing Ca²⁺ concentrations, with a sensitivity of 26.5 mV/dec, showing high selectivity to H⁺, K⁺, Na⁺, and Mg²⁺ (Figure 7b). The theoretical sensitivity exhibited by the sensitive region when measuring divalent ions is 29.6 mV/dec, according to Nernst’s formula, which tends to decrease the sensitivity. This decrease in sensitivity is thought to be due to ions other than the target ion in the buffer solution (in the case of the Mg²⁺-sensitive region, H⁺, K⁺, Na⁺, and Ca²⁺), which act as interfering ions and prevent the ionophores in the sensitized membrane from capturing ions. This indicates that the Mg²⁺-sensitive region fabricated in this study is susceptible to Ca²⁺. Also, it was ensured that the fabricated sensor functions to measure Mg²⁺ and Ca²⁺ without briefly releasing both membranes from the pixel electrode and can be used basically multiple times in an aqueous solution.

3.2. Demonstration Results of Selective Multi-Ion Measurement

To demonstrate the simultaneous imaging of Mg²⁺ and Ca²⁺ responses and the quantitative measurement of the concentration change, visualization experiments were performed using Mg²⁺ and Ca²⁺ solutions. Figure 8a shows the imaging results for the Mg²⁺ and Ca²⁺ responses. The ΔV_Out of the sensor is shown as an absolute value with a 0.5 V scale by a color bar: the yellow color indicates a high concentration of Mg²⁺ and Ca²⁺, and the blue color indicates a low concentration. The output image of the Mg²⁺-sensitive region changed with the Mg²⁺ concentration. Additionally, the images of both sensitive regions were observed soon after the addition of the Ca²⁺ solution.

Figure 8b shows the biosensor response to changes in the Mg²⁺ and Ca²⁺ concentrations. The V_Out change (ΔV_Out) in the Mg²⁺-sensitive membrane increased for the concentration change of Mg²⁺, whereas the ΔV_Out in the Ca²⁺-sensitive membrane hardly changed. After the Ca²⁺ solution was added dropwise, the ΔV_Out for both the Mg²⁺- and Ca²⁺-sensitive membranes immediately increased, and these sensor signals reached saturated levels.

From these experimental results, and since there was no response less than 10⁻² M for Mg²⁺, as described in Section 3.1, this originates from the concentration levels of interfering ion Ca²⁺, i.e., it should be considered that the lower detectable Mg²⁺ concentration depends on the concentration of Ca²⁺.

Having confirmed that the discrimination of the measured ion species was successful, we checked whether quantitative concentration derivation was possible from the response to changes in the concentration of each ion. Deriving the Mg²⁺ concentration (C_Mg) and Ca²⁺ concentration (C_Ca) after the concentration change from the potential response (ΔV_Mg and ΔV_Ca) obtained from the potential response in Figure 8b and the sensor sensitivity (S_Mg and S_Ca) obtained from the characterization in Section 3.1, the following is obtained.

C_Mg = ΔV_Mg/S_Mg = 18 mV/19 mV/dec = 8.85 mM

C_Ca = ΔV_Ca/S_Ca = 26 mV/26.5 mV/dec = 0.98 mM

After the concentration change, the theoretical concentrations were 10 mM for Mg²⁺ and 1 mM for Ca²⁺. The Mg²⁺- and Ca²⁺-sensitive regions were 88.5% and 98%, respectively, in terms of accuracy.

In an experiment to improve memory in mice with a diet containing magnesium, it was reported that the concentration of Mg²⁺ in the brain changed by about 15%. Because the general Mg²⁺ concentration in the brain is 1 mM, it is thought that a minute concentration change of about 0.15 mM is produced [23]. Using this sensor, the concentration change would be about 0.13 mM, assuming an error occurs. This change is detected as a potential change of 1 mV, sufficient to capture the noise of 0.5 mV_rms during in vivo measurement.

These results indicated that the Ca²⁺-sensitive region can compensate for the low selectivity for Ca²⁺ in the Mg²⁺-sensitive region. From these experiments and the quantitative evaluation results, the real-time imaging of Mg²⁺ concentration changes, discrimination from Ca²⁺, and quantification of the concentration changes were successfully demonstrated. In addition, a pixel region exposed between the two membrane formation areas functioned as a pH sensor. However, if Mg²⁺ is changing in one part of the sensor and Ca²⁺ in another, the target ion concentration change cannot be measured accurately at this stage, because the two membranes were coated a little too distant in the array. Thus, the development of a membrane patterning technique is required for multi-analyte detection of Mg²⁺ and Ca²⁺ with high spatial resolution toward future bioimaging applications.

Based on our proposed method, the Mg²⁺ and Ca²⁺ imaging and quantitative determination were particularly demonstrated under the specific ion concentration by controlling the Mg²⁺ and Ca²⁺ concentrations. However, we cannot use it to solve the problem, the Mg²⁺-sensitive membrane shows almost no response for real biological Mg²⁺ fluctuations in the presence of Ca²⁺ 2 mM, like the extracellular concentration level. To overcome this problem, improvement in the Mg²⁺ sensor performance is necessary by investigating an ionophore and other materials for biological experiments as future works.

4. Conclusions

We fabricated a 32 × 256 pixels implantable multi-ion image sensor using Mg²⁺- and Ca²⁺-sensitive regions. The resulting Mg²⁺-sensitive region on the array showed an output voltage change of 19 mV/dec between 10⁻² M to 10⁻¹ M Mg²⁺, 29.2 mV/dec for Ca²⁺ in the range of 10⁻⁴ M to 10⁻¹ M, and 26.5 mV/dec for Ca²⁺, indicating reasonable sensitivity for the coexisting ions. In the selective response evaluation experiment, the high selectivity of the Ca²⁺-sensitive region complemented the low selectivity of the Mg²⁺-sensitive region, and we verified whether Mg²⁺ concentration changes could be selectively detected. The results showed that the discrimination of Mg²⁺ and Ca²⁺ concentration changes was possible, demonstrating real-time imaging and quantification of Mg²⁺ and Ca²⁺ concentration changes.

Although we succeeded in visualizing changes in Mg²⁺ and Ca²⁺ concentrations on the fabricated sensor device, future investigation for improving the Mg²⁺-sensitive performance is required for biological experiments. If the physiological Mg²⁺ fluctuation of less than 1 mM can be measured, simultaneous Mg²⁺ and Ca²⁺ mapping and multianalyte detection in the brain can be expected by applying inkjet devices and semiconductor technology to paint sensor pixels with a sensitized membrane.