Recipient Organization
STRATIO INC
2211 FORTUNE DR
SAN JOSE,CA 95131
Performing Department
(N/A)
Non Technical Summary
In recent years substantial effort has been invested in reducing greenhouse gasses and achieving net zero initiatives worldwide. Agricultural practices can also have a significant effect on the environment when the use of inefficient or improper methods leads to the release of sequestered soil organic carbon (SOC), which also plays a significant role in maintaining the content, stability and nutrition of the soil for bearing crops. While advancements have been made in Agriculture Analysis & Crop Science, there is still a significant disparity in accessibility to this this technology due to lack of internet and other economic disadvantages.Short-wave Infrared (SWIR) imaging possesses the unique ability to see characteristics of objects that are invisible to the naked eye, even when given less than ideal conditions. Traditional SWIR imaging necessitates indium gallium arsenide (InGaAs) sensors, which demand low-yield and high-cost bump-bonding processes. Because of its high cost and high-power consumption, InGaAs SWIR sensors hindered the widespread adoption of SWIR imaging. However, STRATIO, INC.'s core technologies are based on germanium (Ge) which allows our sensors to be more affordable, less power consuming and lighter by design.STRATIO is proposing a low-cost, high-sensitivity germanium (Ge)-based Short Wave Infrared (SWIR) spectral sensor for use in SOC monitoring. In order to reach our target customer, we intend keep final costs low and offer a final price point of less than $2. Once we have built our initial prototype, STRATIO has recruited Dr. Jonathan Sanderman of the Woodwell Climate Research Center as a consultant for the project. Sanderman's team possesses extensive expertise in the soil spectroscopy field and will use this experience to test STRATIO's device on a total of 300 soil samples from local soils present in the agronomically critical farmlands of the Central Valley of California and will identify 5 farms to collect these samples from. Utilizing both STRATIO's spectral sensor and traditional laboratory methods as a control, Sanderman's team will collect the data and together with STRATIO will assess the data to assess the sensor's performance and make any necessary changes to the final design.By offering sensors at a price point that enables widespread integration, STRATIO aims to make this important technology more accessible to enable widespread integration and put the power of SWIR sensing in the hands of famers both domestically and worldwide. Beyond agricultural applications, STRATIO also aims to influence the conservational efforts as well as the larger ESG market through broader carbon quantification applications including SOC analysis in coastal wetlands and other areas to influence carbon sequestration and net zero initiatives on a global scale.
Animal Health Component
25%
Research Effort Categories
Basic
5%
Applied
25%
Developmental
70%
Goals / Objectives
The main goal for this project is to develop a Ge-based SWIR spectral sensor for SOC detecting applications that is both high-resolution and low-cost at a price point of <$ 2. By providing our device at this price point we to make this important agricultural analytic technology to be more affordable and accessible and to improve farming methods, reduce farming related greenhouse gases and influence carbon sequestration and conservation efforts.For Phase I we will focus on validating the sensor, and producing a sensor. To achieve this goal we will need to attain several main objectives within the 8 month time line for phase I:1) Create a simplified sensor array structure for the spectral sensorFocal plane arrays (FPAs) often require N-type metal-oxide-semiconductors (NMOS) due to the large number of pixels as it is unreasonable to connect all of the pixels directly to the input/output pad. However, when using Ge, this causes a problem with the height difference between the Si-CMOS and the epi-Ge, which leads to a low yield of around 30% on the FPA.By utilizing a simplified spectral design, there is no longer a need for an NMOS for each pixel or decoder/encoder circuitry, as the number of pixels is small enough to allow direct connection to dedicated input/output pads. With this design, all components of the spectral sensor can be located on the same plane, thus simplifying the manufacturing process, and significantly increasing the yield. This approach will not only increase the yield significantly, but will also enable a smoother fabrication process. We expect that by removing the need for the NMOS, and utilizing this simplified spectral sensor design we will be able to increase the yield to ~95% and simplify both the number of steps and difficulty of each step of the fabrication flow.2) Incorporate a LVF into the sensorDue to the difference in size, building a spectral sensor package that combines both a Ge-based SWIR sensor array and an LWR, is no easy task. Based on our preliminary research the smallest SWIR LWR available is 15 mm x 3.5 mm x 0.5 mm and covers 0.9 - 1.7 µm. The pixel pitch of our current sensor is 35 µm, which is more than 50x larger than our planned than the eight-pixel sensor array which has a length of only 280 µm. Even if we were to place eight 280 µm long pixels adjacent to each other, the resulting spectral signal would only be 15nm long, which is still not enough to read broad band SWIR signal to identify various soil samples.To address this issue, we plan to space the eight pixels apart at 1.5 mm increments in the sensor array rather than having them directly adjacent to each other, which will allow us to increase the distance between the pixels and ensure that the wide spectral range can be covered by the sensor array. The sensor array die with eight pixels will be attached to a 44-pin chip carrier with a cavity size of 0.44 sq. inches (or 11.18 sq. mm), and this sensor array will be able to fit in the cavity of the chip carrier because the length of the sensor is less than 11 mm. Instead of using a glass lid on top of the chip carrier to protect the sensor, we will place the LVF on top of the chip carrier. By combining the LVF and spaced pixels, the linearity of the wavelengths of light that transmits through the LVF, will allow us to estimate the center wavelength of the light on each pixel, and increase our sensor's ability to read the wavelengths necessary to analyze the carbon content of various soil samples. We estimate that the first pixel will receive the light at the wavelengths of 1 µm, while the last pixel will receive wavelengths of 1.59 µm.3) Assemble the spectral sensor prototype and field testing the spectral sensor for feasibility in SOC detectionTo finalize our initial prototype for feasibility testing, we intend to use a simplified version of the PCB and embedded software from our BeyonSense camera, which was launched in 2022, as there are less pixels to read. Since the number of pixels is reduced, this will also lead to an increase in the signal-to-noise ratio without having to make any further modifications to our design.For the sake of the initial prototype, we also intend to utilize a conventional c-mount SWIR camera lens intentionally configured to be out-of-focus on the surface of the spectral sensor. As it is important for the prototype to collect incoming light as much as possible so that the sensor receives light input that is strong enough, the incoming light needs to be spread out as evenly as possible over the circular area of diameter 11mm. We expect that by using the c-mount SWIR camera lens, we will be able to mitigate the uneven light distribution and enable us to gather valuable spectral data for analysis.Once this prototype is complete, we will work together with Dr. Jonathan Sanderman of the Woodwell Climate Research Center and his team to collect soil samples and test the feasibility of our prototype. Dr. Sanderman and his team have significant experience with soil spectroscopy and will test the ability of the spectral sensor in estimating soil organic carbon concentration, bulk density and soil organic carbon density, the three components needed to estimate SOC stocks, and compare results against a control using traditional laboratory analyses. For the sake of the feasibility project, we expect to collect and test 300 soil samples from 5 selected farms within northern California.As we will have more flexibility during the feasibility study stage, STRATIO and Sanderman will work together to make any necessary changes to the spectral sensor design such as environmental factors, choosing specific wavelength bandpass filters using the LVF or adding more spectral data points if needed.If we attain all of these objectives we will have an optimized final prototype of our SWIR spectral sensor in hand for use in the more extensive data collection and AI-model development required in Phase II and beyond.
Project Methods
The main objective of Phase I will be to develop viable prototype for our SWIR spectral sensor with eight spectral bands. To achieve this we will need to test this sensor on soil samples to determine the feasibility and make any necessary changes to our initial design. STRATIO has recruited Dr. Jonathan Sanderman of the Woodwell Climate Research Center as a consultant for the project. Dr. Sanderman and his team have significant experience with soil spectroscopy, and will help STRATIO in testing the ability of the spectral sensor in estimating soil organic carbon concentration, bulk density and soil organic carbon density, the three components needed to estimate SOC stocks. Sanderman's team will identify 5 farms in northern California for participation and determine a field sampling plan for each location. From these locations, Sandermans team will collect and asses 300 soil samples using both STRATIO's SWIR spectral sensor and traditional laboratory analysis as a control. For the feasibility stage, STRATIO will combine the new sensor with their existing spectrometer device.At each farm, Sanderman and his team will collect 20 volumetric soil cores (30 cm in length) by hand using an AMS slide hammer corer. Each core will then be extruded onto a split PVC pipe, divided into three 10 cm increments, scanned, and then bagged for shipping to the Woodwell laboratory. This will allow Sanderman to collect a total of 300 samples for analysis which will be sufficient for initial training and testing with various machine learning models for estimation of SOC. To ensure accurate results, Sanderman will also use a control analysis of the samples using traditional laboratory analyses. Upon arrival at the Woodwell laboratory, all 300 soil samples will be weighed, air dried, and then weighed again to determine field moisture content. The air dried samples will then be crushed and sieved to 2 mm, <2 mm fraction and weighed. Several sub samples will be taken to test with the control and STRATIO's scanner. The first sample will be kept to determine the oven dry moisture correction for estimation of bulk density, the second will be scanned with STRATIO's scanner and a third sample will be scanned using a Fourier-transform mid-infrared spectrometer (Bruker Vertex 70) to determine if carbonates are present and to estimate carbon content. If there is any carbonate present, these samples will then be subsequently are analyzed via pressure calcimeter to determine carbonate content. Finally, all samples will be run on elemental analyzer to determine total carbon content.The data collected during this stage will be assessed by both STRATIO and Sanderman to assess the sensor performance, and from this data, STRATIO will make any necessary changes and improvements to their sensor design to ensure optimal performance for the final prototype moving into Phase II.To share the findings of our project with other experts, STRATIO and Sanderman plan to prepare a short paper detailing their findings in Phase I for publication in a peer reviewed journal.