Investigating the source of natural arsenic contamination in Vermont’s groundwater

Senior Nora LaCasse is spending her Field Work Term this year working on an independent study project that investigates the cause of natural arsenic contamination in Vermont’s groundwater.

Parts of the Taconic Mountains of Vermont have naturally elevated levels of arsenic in groundwater, which is a significant public health concern. The source of the arsenic is pyrite minerals within the the Taconic slate rock. When the pyrite is oxidized (rusts), the arsenic is liberated, and may either be absorbed onto the surface of iron oxide or clay minerals, or may go into solution with the groundwater. A study performed by Middlebury College students and the Vermont Geological survey outlined areas of highest concern. This and other studies indicate that the amount of dissolved oxygen in groundwater plays a role in determining whether arsenic ions are absorbed to mineral surfaces or released into solution.

The first step of Nora’s investigation is to perform laboratory experiments on the ability of the minerals in the Taconic slates to absorb arsenic from solution under differing oxygen concentrations. She crushed samples of Taconic slate in order to increase the available surface area for absorption, then added the rock powder to a arsenic solutions of known concentration. These are being stirred under both oxygen-rich and oxygen-depleted conditions and under a range of imposed pH conditions. After stirring for several days, the solutions will be re-analyzed to determine if the arsenic concentration has decreased. Later phases of the study will test arsenic-rich slate samples to determine under what conditions arsenic is de-sorbed and released to solution.

Nora with her experiment setup

From the Field: Butterfly Farming in Ecuador

Dircenna adina stenheili — a clearwing buttefly

Rothschildia lebeau — a giant saturniid moth

Senior Emily Mikucki  reports that she’s been working with an eco-tourism lodge in the cloud forests near Nanegal, Ecuador (check it out) to help develop a butterfly farm.

Emily has been obsessed with lepidopterans (moths and butteflies) for years, and hopes to turn that obsession into a career as  as a conservation biologist. At Bennington, she studies biology and Spanish (the latter to give her a head start doing conservation and field work in Latin America)  Meanwhile, she takes photos of them — these two are from Ecuador.

This isn’t the first time Emily has spent FWT in the tropics. In past years, she has worked with researchers in the Amazon basin of Peru and with butterfly conservationists in Costa Rica.

OpenCompute Hardware Hack

This past week, I traveled to Santa Clara, CA to participate in the OpenCompute Summit.  OpenCompute is an open-source hardware project that aims to democratize the world of data centers (the places full of hundreds of thousands of servers, where things like Facebook, Google, Bing and Twitter live) by openly sharing and evolving hardware, software and mechanical designs.

One of the more exciting things that occurred during the conference was a “hardware hackathon” where teams made up of engineers and researchers from industry and academia came up with a novel idea to contribute to the OpenCompute project (students at Purdue had previously developed a biodegradable server chassis!).  As it turns out, the team I was a part of took first place in the competition – the prize being support and funding from the OpenCompute Foundation to file a provisional patent application for our idea, which also becomes a new mainline project as part of OpenCompute.

Visualization of our team’s component (CAD by Jon Ehlen of Facebook)

The idea that we developed was a true interdisciplinary approach to the problem of management and monitoring of servers in data centers.  We combined electrical engineering, mechanical engineering, software engineering and computer science to come up with a design for a module (codenamed “Cheezy Fingers” – long story…) that plugs into the debug header of an OpenCompute motherboard, and uses a hyper-local low-power wireless mesh network to transmit sensor, location and status data about servers, as well as to provide a mechanism for locating, identifying, and gaining remote access to the serial console of a server.   The design is based on what we know of the challenges Facebook has in managing their OpenCompute-based servers (hundreds of thousands of them), and is designed to be a low-cost, low-power component that can be easily snapped-in as a retrofit to their existing servers, as well as to any other OpenCompute server in the wild.

Our team hard at work (Jon Ehlen, Facebook; Andrew Cencini, Bennington College; David Kaplin, Nebula; Steven White, Nebula). Photo by Jill Juarez.

We had about 12 hours to discuss, evolve, and model our design – though we had also been discussing the idea in advance.  Our team was made up of two electrical engineers from Nebula – Steven White and David Kaplin, myself (the computer scientist/software guy), and Facebook’s mechanical engineer in charge of storage design, Jon Ehlen.  We had a blast working together – Jon built a beautiful CAD model of “Cheezy Fingers” in SolidWorks (see picture above), and Steven and David specced out the components and schematics.  For my part, I specified the data protocol and aggregation design, as well as wrote some C code for the serial connection that will run in the Linux kernel to retrieve locational information and blink the chassis LED using IPMI.  The judges remarked on what a thoughtful and complete submission we had – especially given it’s something we could begin mass-producing within the next 24 hours or so.

All in all, it was a great time, and a fun way to bring together people from a variety of disciplines and backgrounds to solve a real problem and make a real and meaningful contribution to the open source hardware community.  Ben Broderick-Phillips and Pratham Joshi are currently spending their FWT working on other projects with Nebula, which are slated to become contributions to the OpenCompute project as well.  Look to hear more about their work in the near future…

Update 1/21/2013: Some nice media coverage of our project here.

Update 1/24/2013:  Welcome GigaOm readers!  (GigaOm story here).