DNA integrated circuit/proposal import 2007

From Wikiversity
Jump to navigation Jump to search

A DNA integrated circuit is a integrated circuit semiconductor system incorporating or interacting with deoxyribonucleic acids or other molecules. The interaction of molecules with the IC system may be due to the adsorption of the molecules on the IC, or the electromagnetic or mechanical interaction of the IC with molecules.

Depending on the attachment methodology, the molecules incorporated into a dIC may be other than deoxyribonucleic acid.

proposes one way to attach molecules to CMOS integrated circuits.

\title{Aluminum Anodization for DNA Integrated Circuits}

\textbf{Proposal in response to DARPA BAA04-12}

\begin{flushleft}
\textbf{Title:} Aluminum Anodization for DNA Integrated Circuits\\
\textbf{Technical area:} New Materials, Materials Concepts, Materials
Processing and Devices (Smart Materials) \\

\begin{large} Executive Summary \end{large}
\end{center}

\bigskip


Published research has indicated that DNA molecules can be adsorbed on
the surface of aluminum electrodes in an electrochemical reaction.
Exploitation of this phenomenon with aluminum structures in standard
integrated circuits may enable fundamentally new microscale and
nanoscale systems for the warfighter and other integrated circuit
users.  This proposal outlines an effort, highly focused on
deliverables and commercial relevance, to demonstrate and characterize
this process using a CMOS microsystem, to be fabricated using the
MOSIS service, and low-cost DNA printing equipment.


\pagebreak


\section{Introduction}

Molecular self-assembly is increasingly viewed as a critical enabling
technology for the development of nanosystems expected to provide
significant new capabilities for the Department of Defense and the
individual warfighter.  The pervasive use of integrated circuit
technology in current-day DoD operations, and in society at large,
provides motivation for the fullest possible exploitation of standard
IC processes for the development of molecular self-assembly and
molecular nanotechnology in general.  Specifically, this preeminence
of IC technology, and complementary metal-oxide-semiconductor ICs in
particular, justifies a detailed characterization and thorough
understanding of molecular adsorption processes on materials present
in CMOS integrated circuits, as molecular adsorption on solid surfaces
is a fundamental process in the integration of molecules and molecular
systems with solid-state technology.

Aluminum alloy is the primary material used to conduct electricity in
CMOS integrated circuits.  It is well known that aluminum can be
coated with an oxide film, in an electrochemical process called
anodization, to alter its appearance or mechanical properties,
although this coating is usually performed on macroscopic aluminum
objects.  From the standpoint of micro- and nanofabrication, the
anodic aluminum oxide film itself has been the primary focus of
contemporary research, as it forms an ordered nanoporous structure
under a wide range of processing conditions\cite{osaka98,gosele98}.

When aluminum is anodized in an aqueous solution containing acids such
as phosphoric, sulfuric, or carboxylic acid, the molecular acids are
incorporated into the film, forming an electrochemical bond with the
oxidized aluminum atoms.  This incorporation of molecules present in
the anodizing bath is key to the modification of the appearance, and
other properties, of anodized aluminum.  While anodization of aluminum
by small-molecule acids is a routine process, the use of
macromolecular acids for aluminum anodization is largely unexplored.

Deoxyribonucleic acid is perhaps the most important example of a
macromolecular acid.  In the course of experiments on the manipulation
of DNA molecules by time-varying electric fields generated by aluminum
microelectrodes, one group has demonstrated DNA adsorption on
aluminum\cite{washizu99b}.  Their research has provided support for the
hypothesis that, by applying a voltage to an aluminum microelectrode,
one end of a DNA molecule can be anchored to the aluminum surface,
with the other end extending from the surface.  Despite the fact that
this research has largely been ignored in the years since it was first
published, it is best perceived as a tantalizing glimpse of
opportunity.

The unique molecular properties of DNA, in addition to its obvious
ubiquity, make DNA molecules a natural component of molecular
self-assembly processes.  The incorporation of DNA molecules into CMOS
integrated circuits, using aluminum anodization, potentially presents
the opportunity to use standard IC technology as the foundation for
complex self-assembled nanosystems.

In the following, we propose an effort to characterize DNA adsorption
on aluminum microelectrodes.  The proposed research is focused on the
design, fabrication, operation, and physical analysis of a CMOS
microsystem for performing anodization and controlling hybridization
with microelectrode arrays.

After a more detailed description of the evidence for DNA adsorption
on aluminum, we provide further motivation for the proposed research
by describing several potential applications of DNA hybridization on
standard integrated circuits.  This is followed by a description of
the CMOS microsystem to be created for the performance of the proposed
research, along with a description of the fluidic system necessary for
performing electrochemical processing, using DNA solutions, on the
surface of the proposed CMOS microsystem.  The goals for the research,
along with some of the accompanying risks, are presented, and a
timeline for performance of the research is outlined.  We conclude
with a description of the resources, human and material, to be devoted
to the proposed research.

\section{Prior Work and Future Possibilities}

\subsection{DNA manipulation by Al microelectrodes}

One early effort to manipulate cells and molecules using
microfabricated electrodes\cite{washizu90}, called the fluidic
integrated circuit by its creators, used aluminum evaporated onto
glass and patterned to form electrodes, which created an electric
field when a voltage was applied to the opposing electrodes (see
Figure \ref{early_electrodes}).  In the presence of a spatially
nonuniform electric field, polarizable cells and molecules experience
a force in the direction of increasing electric field magnitude, a
phenomenon called dielectrophoresis\cite{pohl78}.  During experiments
to characterize the response of DNA molecules, 48.5kbp (kilo base pairs) in
length, to intense (approximately 1MV/m) AC (1MHz) electric
fields\cite{washizu94,washizu95}, it was discovered that DNA
molecules, which have a random coil conformation in solution, are
stretched straight along the field lines.  It was also observed that
the stretched DNA molecules rapidly migrated to the region of highest
field strength by dielectrophoresis, until one end of the stretched
molecule contacted the surface of the aluminum electrode.  Using
fluorescent DNA labels, it was determined that, while one end was
anchored at the surface of the electrode, the rest of the molecule
extended into the solution.  Later investigations used atomic force
microscopy to confirm that DNA molecules were adsorbed on the surface
of the electrodes\cite{washizu99a}.

\begin{figure}[h]
\begin{center}
  
\includegraphics{early_electrodes.eps}
  
\end{center}
\caption{Early demonstration of DNA-aluminum adsorption: Aluminum
  electrodes evaporated onto a glass cover slip, with an interelectrode
  spacing of 80 microns.  When a 1MHz AC voltage is applied to the
  electrodes to create an electric field of approximately 1MV/m, DNA
  molecules (solid lines between electrodes) rapidly migrate to the
  highest-field region and attach to the electrode surfaces. }
\label{early_electrodes}
\end{figure}


While these investigations were important in understanding the
interaction of DNA with aluminum, they did not address several
questions relevant to the practical application of DNA adsorption on
aluminum.  Specifically:
\begin{itemize}
\item{how does the process depend on the applied voltage and
  frequency?  In previous studies, a peak-to-peak voltage of 80V was
  applied to electrodes spaced 80$\mu$m apart, for an electric field
  of 1MV/m.  Not addressed was the effect of smaller electrode
  separation or smaller applied voltage.  Similarly, the applied
  frequency was 1MHz for all of the experiments performed.  The effect
  on adsorption of varying the electric field frequency was not
  addressed.}
\item{how does the process depend on the length of DNA molecule used?
  The only molecule used for these studies was $\lambda$-phage DNA,
  with a length of 48.5 kbp.  Many applications use DNA molecules
  significantly smaller than this, such as cDNA microarrays in which
  the molecular lengths are on the order of several thousand bp or
  fewer, and oligonucleotide arrays which are on the order of tens of
  base pairs.  Previous studies of thiolated DNA adsorption on
  gold\cite{tarlov2000} have shown that surface coverage is affected
  by DNA sequence length.}
\item{what is the long-term stability of adsorbed DNA on aluminum?
  The long-term stability of aluminum interconnect in integrated
  circuits is an important concern\cite{dunn92}, and device
  applications of DNA adsorption must address this issue.}
\item{what is the microstructure of the interaction between the
  adsorbed DNA molecules and the aluminum surface?}
\end{itemize}


\subsection{Future Possibilities}
\label{fantasies}

An obvious application for adsorbed DNA molecules on the surface of an
integrated circuit is micropositioning of objects for heterogeneous
integration\cite{heller98}.  In this process, a surface used as the
platform for integration has single stranded DNA molecules adsorbed at
one or more locations, and the objects to be placed on the surface
have complementary single stranded DNA molecules attached.  By
controlling the placement and sequence of DNA molecules on the surface
and on the objects to be placed, it is possible to use the
hybridization of complementary DNA molecules to self-assemble the
components into the desired finished product.  

One of the most important applications of heterogeneous integration is
the Smart Dust project\cite{pister01}.  Smart Dust motes are projected
to be systems, on the order of 1 cubic millimeter in size, with
self-contained sensing, computation, and communication capabilities.
A key component of Smart Dust systems is CMOS integrated circuits,
along with microfabricated optical and mechanical components.  The use
of DNA-based self-assembly to manufacture Smart Dust motes may
contribute to large-scale, low-cost production processes that enable
Smart Dust to reach its full potential.  The economical incorporation
of DNA-based systems for biological assays also may enable new
applications for detection of environmental pathogens by Smart Dust
motes.

Another important phenomenon that may be demonstrated with DNA
adsorbed to aluminum is electronic control of DNA
hybridization\cite{heller97pnas,heller97nar}.  This process,
illustrated in Figure \ref{electronic_dna_hybridization}, occurs when
a double stranded DNA molecule is exposed to an electric field and a
single strand of the molecule is attached to a surface (generally the
surface of the electrode generating the electric field).  By
controlling the applied electric field, the single stranded DNA not
adsorbed to the surface can be dissociated from the complementary
adsorbed ssDNA.  The magnitude of the electric field necessary to
cause two ssDNA molecules to dissociate can also indicate whether
there are base pair mismatches between the molecules, a fact that is
relevant to the diagnosis of genetic disorders.

\begin{figure}[h]
\begin{center}
  
\includegraphics{electronic_dna_hybridization.eps}
  
\end{center}
\caption{Electronic control of DNA hybridization: (\textbf{A})
  Complementary single-stranded DNA molecules form a double-stranded
  DNA molecule, with one of the strands adsorbed on the surface of an
  electrode.  (\textbf{B}) By applying a voltage to the electrode, an
  electric field is created, which forces the negatively charged
  non-adsorbed ssDNA from the electrode.}
\label{electronic_dna_hybridization}
\end{figure}


From the perspective of micro- and nanosystems, electronic control of
DNA hybridization may enable the dynamic control of mechanical
associations between discrete objects.  One potential application is
the mechanical association of multiple Smart Dust motes to form a
microscale multiple independent reconnaissance vehicle system, or
``microMIRV'' (see Figure \ref{microMIRV}).  The envisioned system
could potentially take the form of a bullet, containing a
three-dimensional arrangement of multiple Smart Dust motes, launched
using standard propellants and rifles or pistols as launch systems.
Firing a microMIRV skyward would deploy multiple Smart Dust motes,
which would dissociate at the peak of the composite vehicle
trajectory, for aerial surveillance, data networking, and other
functions.  Assuming that the economic trends of the microelectronics
industry continue, several such systems may be available for use by
any individual soldier, enabling even dismounted and isolated soldiers
to have decisive battlefield awareness in a wide variety of
scenarios. 


\begin{figure}[h]
\begin{center}
  
\includegraphics{microMIRV_sm.eps}
  
\end{center}
\caption{Dynamic control of mechanical association between Smart Dust
  motes: (\textbf{A}) A cross-section view of several SD motes
  connected by hybridized complementary DNA molecules.  (\textbf{B})
  Using electronic signals, SD motes disrupt hybridization and
  dissociate.  }
\label{microMIRV}
\end{figure}


\section{Research Plan}


The goal of the proposed research is the demonstration,
characterization, and optimization of DNA adsorption on aluminum, in
order to effectively utilize this phenomenon in DoD and commercial
integrated circuit systems.  In order to accomplish this, it may be
necessary to conduct a large number of experiments with varying
parameters (voltage, frequency, temperature, solution composition,
etc.).  Because the experiments, if successful, entail surface
modifications to the aluminum microelectrodes to be used, it will be
necessary to have a large number of the microelectrodes for one-time
experiments and subsequent analysis.

The need for large numbers of test structures, and the imperative to
apply this research to commercial systems, indicates that the use of
CMOS technology to fabricate the microsystem, instead of custom
microfabrication processes, is warranted.  The feature sizes
achievable with CMOS are smaller than those achievable by all but the
most specialized and expensive custom microfabrication processes.  The
availability of active circuitry and transducers in CMOS processes
creates the opportunity to integrate a wide variety of capabilities on
the finished chip.  As well, the use of foundry CMOS increases the
relevance of the proposed research to contemporary integrated circuit
manufacturing, enhancing the likelihood that this research will be
adopted for use in DoD and commercial systems.

\subsection{CMOS microsystem}

For specific discussion of the CMOS process to be used, we refer to
the Taiwan Semiconductor Manufacturing Corporation TSMC35\_P2
process\cite{tsmc35p2-www}, available from MOSIS.  This process has
two polysilicon layers and up to four metal layers available, with a
minimum transistor length of .35 microns.  In this process,
chemical-mechanical planarization is used to create flat surfaces for
deposition and patterning of thin film layers.

For the creation of the proposed microsystem, the fundamental
structure is the aluminum microelectrode.  In order to exploit the
small feature size possible with CMOS fabrication, the exposed surface
of the electrode is 1 micron square.  Directly underneath the square
electrode is a ``via'', connecting the top layer of metal (metal-4)
with the lower level of metal (metal-3).  Sixteen microelectrodes are
arranged on the perimeter of a square $9 \mu$m on each side to form an
electrode array, with the individual microelectrodes separated by $.6
\mu$m (Figure \ref{electrode_array}).  While there is generally a
thick ``overglass'' layer present on the surface of a CMOS integrated
circuit to protect the circuitry from mechanical damage, the need to
have the microelectrodes exposed will require the overglass layer to
be removed from the vicinity of the microelectrode array\cite{warneke2000}.  

\begin{figure}[h]
\begin{center}
\includegraphics{electrode_array_cross_section_2.eps}
\end{center}
\caption{Diagram of the microelectrode array: (\textbf{A}) Plan view
  of microelectrode array, showing 1 $\mu$m square microelectrodes,
  with centered vias, arranged about the perimeter of a 9 micron
  square.  (\textbf{B}) A cross-section view from the dotted line in
  \textbf{A} with labeled dimensions.  A: 1$\mu$m wide metal-4.  B:
  .6$\mu$m interelectrode spacing.  C: .5$\mu$m via width.  D:
  .9$\mu$m metal-3 width.  E: .7$\mu$m separation between metal-3
  interconnect. }
\label{electrode_array}
\end{figure}

Conceptually, the simplest scenario for performing anodization on the
surface of this microelectrode array is, while the surface is exposed
to an aqueous solution containing DNA molecules, to set 15 of the
microelectrodes to the ground voltage, and set the microelectrode to
be anodized (the so-called ``working electrode'') at the positive
supply voltage, referred to as $V_{DD}$.  Given the variation of DNA
dynamics with frequency in an AC electric field\cite{washizu90}, it
may be necessary to cycle the working electrode between $V_{DD}$ and
ground at an appropriate frequency.  Controlling the voltages applied
to the microelectrodes in the array can easily be accomplished with
CMOS digital circuits incorporated into the microsystem.

Determining the $V_{DD}$ necessary for effecting anodization is a
critical goal of the proposed research.  The standard reduction
potential for the conversion of Al to Al$^{3+}$ is
-1.676V\cite{al-webelements-www}.  This indicates that, in order to
effect the electrochemical reaction of anodization, the applied
interelectrode voltage must be greater than 1.676V.  The magnitude of
the necessary additional voltage is determined by the current flow in
the solution, and the associated Ohmic voltage drop.

The most straightforward way to adjust the voltage used for
anodization is to adjust $V_{DD}$.  While the maximum voltage that can
be used in this TSMC process with standard design techniques is 5.0V,
and the preferable maximum voltage is 3.3V, it is possible to design
circuits that will operate correctly at lower supply voltages.  When
electronically controlling DNA hybridization on the surface of the
microsystem, it is important to avoid electrochemical reactions at the
aluminum surface.  Because scaling $V_{DD}$ is expected to be the most
economical way to achieve low voltages for hybridization, the circuits
used to control the microelectrode array will be engineered to
function consistently at supply voltages between 1.0V and 3.3V.
Achieving this objective will require low-voltage CMOS circuit design
techniques\cite{piguet97} that are critical to the design of portable
and power-sensitive systems for DoD and commercial customers.

The size of the microelectrode array, including the circuitry
necessary to control an individual array, is anticipated to be less
than (250$\mu$m)$^2$.  This will make it possible to integrate a
sizable number of microelectrode arrays in the CMOS microsystem, in an
array-of-arrays system.  To effectively utilize the arrayed
microelectrodes, it will be necessary to integrate control circuitry
into the microsystem to enable input signals to be directed to the
appropriate microelectrode array for anodization or hybridization.  To
simplify this task, a hierarchical approach to system design will be
taken.  Specifications will be prepared outlining the performance
requirements for the array-level circuitry, system control circuitry,
and input/output circuitry, as well as the interfaces between each
level.  Proceeding from this system specification, the system will be
designed and thoroughly simulated to ensure correct operation.

\subsection{Integrated processing system}

The CMOS microsystem is the primary platform for the proposed
research, but it is only one component of the system necessary to
perform anodization.  Development of a system to coordinate
microelectrode voltages with fluid deposition and removal will be
essential to the success of the proposed research.  At the same time,
it is important to minimize the complexity of the processing system in
order to successfully complete the initial research proposed here. 

The deposition system needed for the proposed research has many
features in common with systems for DNA microarray
fabrication\cite{heller2002}.  An important difference, however, is
the requirement in this case to remove nonspecifically adsorbed DNA
molecules before each deposition step.  While it is not necessary to
remove nonadsorbed DNA molecules if the goal is to simply demonstrate
DNA adsorption on aluminum, for applications of this process it may be
necessary to carefully control the DNA molecules present at each
microelectrode.  

Due to the vulnerability of the exposed microelectrodes to mechanical
damage, only noncontact deposition methods are practical in this
case.  The most cost-effective deposition system in this case is
likely to be an inkjet printing system, adapted to use solutions
containing DNA molecules instead of conventional
inks\cite{gonzalez2000,yamamoto2000}.  

A long-term goal for the proposed system is to have a controllable
number of DNA molecules of specific sequence on each microelectrode,
fabricated in a process that is as rapid as possible.  The objective
of the current work will be the development of an economical system
that balances the short-term need to quickly and inexpensively perform
a large number of experiments with the anticipated long-term need for
a specialized and powerful system for large-scale manufacturing.

\subsection{Microsystem physical analysis}

Atomic force microscopy is expected to be the primary tool for
determining the effects of the electrochemical processes at the
aluminum electrodes.  Due to the extremely small size of the
microelectrodes, other techniques for surface analysis, such as
surface plasmon resonance and X-ray spectroscopy, are not likely to be
applicable to the analysis of individual microelectrodes.
Fluorescence microscopy will be used to analyze the hybridization of
DNA molecules.  Other optical techniques may be applicable to the
study of the processed microsystem.  


\section{Research Goals}

The overriding goal of the proposed research is the successful
demonstration of electric field-directed DNA adsorption on the surface
of a CMOS integrated circuit, and the characterization of
this process.  Several accomplishments will be necessary to achieve
this goal:
\begin{itemize}
\item{design and fabrication of the CMOS microsystem.}
\item{design of a fluidic system for deposition of DNA solution.}
\item{development of supporting hardware and software to combine the
microsystem and fluidic system into a computer-controlled integrated
system.}
\item{performance and analysis of anodization on the microsystem.}
\item{performance and analysis of DNA hybridization control on the microsystem.}
\end{itemize}
A preliminary estimate of the project schedule is presented in Table 1.  

\begin{table}[h]
\begin{center}
\begin{tabular}{|l|c|c|}
\hline
\hspace{20pt} Milestone &       Work          & Scheduled completion \\
 &               (weeks)        & (weeks after project start)\\
\hline
\textbf{Microsystem Design} & \textbf{11} & \textbf{11} \\
\hspace{4pt} Microsystem specification & 2 & 2 \\
\hspace{4pt} Microsystem implementation & 9 & 11 \\
\textbf{CMOS Fabrication} & \textbf{7} & \textbf{18} \\
\textbf{Processing System Development} & \textbf{13}  &  \textbf{15} \\
\hspace{4pt} Processing system specification & 3 & 5 \\ 
\hspace{4pt} Processing system implementation & 10 & 15 \\
\textbf{Microsystem Anodization} & \textbf{6}  &  \textbf{24} \\
\textbf{Microsystem Analysis}    & \textbf{6}  &  \textbf{30} \\
\textbf{Electrically Controlled Hybridization}  &  \textbf{12}  &  \textbf{30} \\
\hline

\hline
\end{tabular}
\end{center}
\label{schedule_table}
\caption{Estimated project schedule.}
\end{table}



It is anticipated that the bulk of the integrated circuit design for
the CMOS microsystem will be performed by a subcontractor yet to be
selected.  Within two weeks (10 working days) of the project
commencement, the primary contractor will prepare specifications for
the CMOS microsystem.  This specification will be used to direct the
efforts of the subcontractor designing the CMOS integrated circuitry.
It is expected that this design process, including verification, will
take another 9 weeks.  The milestone of CMOS design completion
(``tapeout'') is expected 11 weeks after the start of the project.
The fabrication of the integrated circuit design is expected to take
6-7 weeks\cite{tsmc35p2-www}, for estimated delivery of the CMOS
microsystems 18 weeks after project start.


Development of the processing system to be used to print DNA solutions
onto the surface of the CMOS microsystem in coordination with
electronic control of the arrayed microelectrodes, will also largely
be conducted by a subcontractor yet to be selected.  Beginning two
weeks after the start of the project, specifications for the fluid
processing system will be developed, to be finished within 3 weeks.
Proceeding on the basis of this specification, the system to process
the CMOS microsystem will be developed within 10 weeks, or 15 weeks
after project start.  Due to the potentially complex nature of the
system, and the uncertain partition of effort between the primary
contractor and the selected subcontractor, it is not possible at this
time to further subdivide this effort into specific goals.

Once the CMOS fabrication is complete and the processing system has
been designed and developed, experiments will be performed to
demonstrate anodization on the surface of the CMOS microsystem.  While
it is expected that, given the extremely large number of
microelectrodes on each of the 40 microsystems to be fabricated, there
will be opportunities for experiments over an indefinite and extended
period of time, the first processed microsystems will be prepared for
analysis within 6 weeks, or approximately 24 weeks after project
start.  These processed microsystems will be analyzed by
subcontractors providing materials analysis services, in order to
characterize the modifications at the surface of the microsystems.
The first results from the physical analysis of the anodized
microelectrodes are expected to be available within 6 weeks of initial
preparation, or 30 weeks after project start.

For experiments analyzing electrically controlled hybridization, it is
anticipated, again, that the opportunities created by the CMOS
microsystem will enable numerous experiments to be conducted.
However, focused efforts to demonstrate controlled hybridization on
the surface of the microelectrode arrays are expected to be
successfully completed within 12 weeks after delivery of the CMOS
microsystems, or 30 weeks after project start.   

In addition to the specific goals listed, it is expected that monthly
progress reports will be prepared, describing the project research
activities and developments.  A final report presenting the project
research accomplishments will be prepared at the conclusion of the
project, ten months after the start of the project.  All information
products of the research will be furnished to the Government with
unlimited rights.

This schedule is relatively ambitious, and the overall time span of
the proposed project relatively short.  This proposal, while
intended to open inquiry into a new system for materials research,
is also intended to be focused as much as possible on short-term
results for DoD and commercial customers.  It is anticipated that, if
the proposed research is successful, opportunities will be available
for further research, with an emphasis on commercialization as soon as
is practical.  


%% milestones clearly and quantitatively described
%%   CMOS tapeout
%%     specifications
%%     interfaces between design levels
%%       array <--> control
%%       control <--> external inputs
%%       I/O for built-in self-test?
%%       outputs for on-chip sensors in future microsystems
%%   integrated processing system design / development
%%     first step toward future dnajet manufacturing systems?
%%     exploiting COTS inkjet systems for deposition
%%     subcontract for hardware integration
%%     clean room not necessary
%%   anodization demonstrated
%%     processing
%%     physical analysis 
%%   electronic hybridization control demonstrated



%% specification, design, simulation, design, sim., etc.

%% %% Proposers must demonstrate that their proposal is innovative and
%% %% unique, that the technical approach is sound, that they have an
%% %% understanding of critical technical issues and risk and that they have
%% %% a plan for mitigation of those risks. A significant improvement in
%% %% capability or understanding above the state of the art must be
%% %% demonstrated. All milestones must be clearly and quantitatively
%% %% described.


%% goal / milestone breakdown

%% detailed understanding of array electrochemistry
%%   electric field simulation
%%   approximate calculation of necessary cell overpotential
 
%% cmos microsystem design
%%   array level
%%   AoA control system
%%   physical and logical simulation

\subsection{Risks and Their Mitigation}

The primary technical risk inherent in this proposal arises from the
complex nature of the electrochemical environment in the vicinity of
the aluminum microelectrodes.  Due to the intense electric fields and
accompanying ionic separation, extremely acidic and basic environments
will be present in the anodizing solution during the operation of the
microelectrode array.  This may cause unwanted reactions, including
etching of the aluminum or degradation of the DNA molecules.  It is
known that the buffer composition affects electronically-controlled
hybridization at microfabricated electrodes\cite{heller97nar}, and the
composition and pH of the buffer will be controlled to minimize
unwanted effects and maximize anodization and hybridization
efficiency.

The uncertainty in the schedule and cost estimates for design of the
CMOS microsystem and the processing system introduce a certain amount
of project risk.  Systems for efficient collaboration between the
prime contractor and subcontractors, as well as an emphasis on the
use of commercial-off-the-shelf technologies and components, will be
used to mitigate this risk, but a certain amount of risk in this area
must be acknowledged.

\section{Facilities and Personnel}

In order to maximize the value provided to DoD, a ``virtual research
laboratory'' strategy will be pursued to complete the proposed
research.  The team designing the CMOS microsystem will collaborate
over the Internet, and, to the greatest extent possible, utilize
open-source design tools for layout and simulation of the integrated
circuit.  Space and materials for operating the microsystem to anodize
the microelectrodes will be procured on a ``just-in-time'' basis, so
as to minimize overall expense and maximize flexibility.  Materials
analysis services will be procured by subcontract as necessary, most
likely from university laboratories, with a focus on minimizing costs
to DoD and maximizing research productivity.

It is expected that support from DoD for this proposal will be used to
accomplish the necessary first steps in pursuing this unconventional
research.  Achieving the ambitious goals described above in
\ref{fantasies} will require significant long-term contributions from
the semiconductor industry, and it is anticipated that support from
DoD will be used to leverage further efforts toward developing and
commercializing integrated circuits incorporating DNA and other
macromolecules.

The principal investigator for the proposed research is J. Patrick
Bedell, the chief executive officer of DNAputer Research,
Incorporated.  In addition to a bachelor's degree in physics from the
University of California at Santa Cruz, Mr. Bedell has an intense desire
to contribute to the national security of the United States.  


%% technical risks
  
%% standard CMOS voltage not sufficient for anodization
%%   utilize high-voltage CMOS structures - only if necessary
%%   solution conductivity / counterions / buffer
  
%% nonspecific adsorption
%%   control pH of deposited solution
%%   washing system
  
  
%% long-term stability of alumina film
%%   alumina is hygroscopic

\bibliography{dso_dic2003_proposal_technical}

\pagebreak


\begin{center}
\begin{Large} \textbf{Cost Volume} \end{Large}
\end{center}

\thispagestyle{empty}
\begin{flushleft}

\textbf{BAA Number:} BAA04-12\\
\textbf{Technical area:} New Materials, Materials Concepts, Materials
Processing and Devices (Smart Materials) \\
\textbf{Proposal title:}  Aluminum Anodization for DNA Integrated
Circuits \\

\textbf{Award instrument requested:}  Cost contract - no fee.\\
\textbf{Places and periods of performance:}  The United States in the years 2004-2005.  \\
\textbf{Total proposed cost:} \$170000 \\

%% material cost elements
%%   dna printer customization
%%   processing system components
%%   reagents


%% subcontracts
%%   cmos design services
%%   dna printer development
%%   materials analysis 
%%   use of facilities for performing processing / experiments

%% equipment purchase
%%   CMOS microsystem - TSMC, via MOSIS
%%   DNA printer - TBD


  


%% \$25000 for CMOS design
%% \$25000 for CMOS fabrication
%% \$25000 for DNA printing system
%% \$10000 for necessary customization of DNA printing system
%% \$10000 for experimental overhead

\begin{table}[h]
\begin{center}
\begin{tabular}{|c|c|}
\hline
Equipment purchases    & \$65000     \\
Subcontracts           & \$55000     \\
Direct labor           & \$30000     \\
Materials              & \$10000     \\
Overhead charges       & \$10000     \\
\hline
Total estimated cost   & \$170000     \\
\hline
\end{tabular}
\end{center}
\caption{Total estimated project cost.}
\end{table}


\begin{table}[h]
\begin{center}
\begin{tabular}{|c|c|}
\hline
Equipment & Cost \\
\hline
CMOS microsystems (40) & \$25000 \\
DNA printer            & \$25000 \\
Computer workstation (inc. software) & \$5000 \\
Probe station          & \$10000 \\
\hline
Total equipment cost   & \$65000\\
\hline
\end{tabular}
\end{center}
\caption{Equipment purchases and estimated costs.}
\end{table}

\begin{table}[h]
\begin{center}
\begin{tabular}{|c|c|}
\hline
Subcontract & Cost \\
\hline
CMOS design verification & \$20000 \\
DNA printer customization  & \$20000 \\
Materials analysis services & \$15000 \\
\hline
Total subcontract cost  & \$55000\\
\hline
\end{tabular}
\end{center}
\caption{Subcontracts and estimated costs.}
\end{table}

\pagebreak

\begin{table}[t]
\begin{center}
\begin{tabular}{|c|c|c|c|}
\hline
Employee & Hourly Wage & Hours & Total Cost \\
\hline
John Smith & \$30.00  & 1000 & \$30000 \\
\hline
Total cost  & & & \$30000\\
\hline
\end{tabular}
\end{center}
\caption{Direct labor costs.}
\end{table}



The estimates presented here of project costs are preliminary, to say
the least.  It is hoped that, among other factors, cooperation between
the DARPA program manager and the primary contractor will enable the
project's goals to be achieved as economically as possible.


No existing or potential conflicts of interest are known to the
offeror.  The offeror has no existing contractual relationship with DARPA
or any other unit of the US Government.  

Today, we lack metrics 

It is well known that molecules containing a thiol (-SH) group can
form a self-assembled monolayer on gold surfaces, due to the
interaction between the sulfur and the gold surface.  This effect has
been used with chemically modified DNA molecules to create DNA
microarrays.  


Controlling DNA hybridization with electric field generated by
microelectrodes on the surface of a chip is a fundamental attraction
of DNA adsorption on aluminum.  



compare / contrast adsorption systems
  gold-thiol system (Figure)
  alumina less well-defined than SAMs 
  electric-field directed process provides for high spatial resolution
  economics of aluminum adsorption qualitatively different!
    unmodified DNA
    aluminum surface available in standard IC processes
    electrically-controlled hybridization possible on large scale with
   CMOS integration
  dna hybridization as component of micromechanical assembly systems
    precise positioning
    assembled microobject at bottom of potential well created by
   sequential dna hybridizations
    micro Fresnel lens as potential example 
  articulating relevance of DNA self-assembly to Smart Dust... long-term
    microMIRV
      

cmos microsystem for anodization / dna manipulation



primary feature of design is electrode array for anodization /
 hybridization

  16 microelectrodes arranged around the perimeter of a square 

    why 16 electrodes?
      minimize size of array
    why perimeter?
      reduces dissimilarity of electrode configurations  
    why 1 square micron? want *small* electrodes, .9 micron the
   width to cover via in TSMC 2P4M

    

    1 square micron top-layer metal electrodes over via
    minimizing interelectrode capacitance
  temperature control using CMOS thermal elements

concept of operations for electrode array
  anodization
    pl volume of DNA solution deposited on array
    working electrode(s) cycled between VDD and GND at some frequency
    cathodes set at GND
    frequency determined by multiplexing, clock frequency, VDD
     magnitude
      DNA frequency-dependent conformation / orientation / dynamics
:hybridization
::controlled by electric field and temperature
::demonstration of PCR on chip?
:::enzyme-friendly microenvironment?
:::pH variations due to electric field 
    


:control circuitry
::per-array circuitry
:::single control input to array is demultiplexed to microelectrodes
:::electrode control circuitry must deal with electrochemical circuit
:::all ``analog'' functionality contained in array block
::system control circuitry
:::address individual arrays
:::similar to load-only memory - no array sensors in first version


The central feature of the proposed microsystem is an electrode array
(Figure \ref{electrode_array}).  This array contains 16
microelectrodes arranged along the perimeter of a square $9 \mu$m on
each side.  The individual microelectrodes are composed of the
top-layer metal in the CMOS process, and are $1 \mu$m square, with an
interelectrode separation of $.6 \mu$m.

The two processes occurring at the electrode array are anodization of
the exposed aluminum to form a film with adsorbed DNA molecules, and
hybridization of DNA molecules complementary to the DNA molecules
adsorbed on the anodized aluminum.  To simplify the investigations
proposed here, these processes initially should be considered and
conducted separately.

The standard reduction potential for the conversion of Al to Al$^{3+}$
is -1.676V.  This indicates that, in order to effect the
electrochemical reaction of anodization, the applied interelectrode
voltage must be greater than 1.676V.  

current a function of applied voltage
necessary voltage a function of current...

cmos voltage scaling allows for systematic variation in anodization voltage




The standard voltage for
integrated circuitry in the selected process is 3.3V.  While this
voltage can be exceeded with careful design\cite{}, such nonstandard
designs inevitably increase the development time and cost and should
be avoided if possible.


The interelectrode current and the effective resistance of the
solution through which the current flows determines, by Ohm's law $V =
I R$, the overpotential necessary for the electrochemical reaction at
the working electrode.


Conversely, if one wants to
avoid electrochemical reactions at an aluminum electrode, the applied
voltage must be less than 1.676V.

Accordingly, in the first phase of the research, anodization of the
microelectrodes will be e


scaling VDD for anodization - analyze effects
  plurality of electrode arrays makes it possible to conduct numerous
   experiments with varied process parameters


scientific foundation for estimate of necessary anodization voltage?
  solution conductivity
    how does resistance scale with interelectrode distance - positive correlation
    back-of-the-envelope calculation for necessary cell overpotential


deposition system:
  must remove nonadsorbed dna molecules
  limited parallels to existing microarray deposition - removal of dna
   molecules also important in this case
  
  

technical concept of operations


clock to electrode array cycles input signal through set of
microelectrodes
  inactive electrodes set at ground, active electrode(s) at VDD
  
how to provide voltages less than Al reduction potential using digital
CMOS with a minimal transistor count?  D/A for each electrode?
  preliminary answer: scale VDD for anodization vs. hybridization application
    voltage scaling for digital circuitry
      draws upon previous DARPA-funded research, relevant to smart dust
    1V for hybridization, more for anodization - how much more TBD

poly-Si heater



system-level diagram
  array of unit cells

unit cell
  16 microelectrode array
  polysilicon heater
  
how many pins?
  >100 pins easily available
  supporting board design, PC interface?
    USB-interfaced DNA chip? :)
      PC-interfaced personal DNA analysis / sequencing long-term goal
    initial systems must minimize complexity
    numerous ground, VDD pins


conops must include deposition system
  inkjet system
    control of droplet size, dna concentration, pH
    removal of excess fluid
     

metrology for ultramicroelectrodes?
  one or more orders of magnitude smaller than previous adsorption surfaces
  in-situ monitoring of anodization / hybridization
    labeled molecules for ellipsometry / fluorescence microscopy



GOALS

first CMOS implementation
  demonstration of DNA adsorption
  characterize process parameters for adsorption
  in-situ anodization metrology?
  hybridization control with electric field and integrated poly-Si heater

second CMOS system
  interelectrode hybridization
  hybridization-directed self-assembly

electric field simulation

DNA microdroplet deposition



RISKS

standard CMOS voltage not sufficient for anodization
  utilize high-voltage CMOS structures - only if necessary
  
nonspecific adsorption
  control pH of deposited solution
  washing system
  
long-term stability of alumina film
  alumina is hygroscopic
  

  

NOTES:

focus of microfabrication
  outsourced - foundry CMOS
    submicron metallization / structures (DD)
    micron-scale metallization
  university microfab facilities

advantages
  foundry CMOS 
    incorporate multiplexing circuits for anodization
    demonstrating / assessing anodization at CMOS voltages
  microlab - custom fab
    quicker turnaround time
    lower investment necessary 
    potentially lower cost

all-CMOS approach absolutely necessary for current proposal
  unconventional circumstances of proposer / proposal require that CMOS platform with minimal to high functionality be present
  all-CMOS guarantees relevance to contemporary fabrication processes
  only minimal voltage sources required + appropriate materials characterization facilities

materials characterization
  atomic force microscopy 
  ellipsometry (with refractive index label)
  FTIR microscopy?
  
CMOS circuits
  voltages for anodization
  interelectrode capacitance minimization
  detecting interelectrode hybridization

  compatibility of applying voltages for anodization and capacitance measurement
    MOSFET switches?


Much of the existing research on DNA adsorption on surfaces has been
directed toward systems in which chemically modified DNA molecules
form covalent bonds with the surface of interest.  Perhaps most
well-studied is the formation of self-assembled monolayers by
thiolated molecules on gold (see Figure \ref{au_thiol_sam}).  

To be cost-effective, these tools should be as automated as possible and be conducive to massive parallel processing.


alumina formation
  thin film amorphous solid
  porous alumina formation
  acid incorporation into alumina film
  long-term stability of alumina
  biocompatibility of alumina
  material analysis
    x-ray spectroscopy
    ellipsometry
    afm

dna adsorption on surfaces
  gold
  silicon
  silicon oxide
  mica

self-assembled monolayer formation
  alkylthiols on gold most studied system
  
the increased flexibility and economy afforded by use of technological progress



As we know, there are known knowns.  There are things we know we know.
We also know there are known unknowns.  That is to say we know there
are some things we do not know.  But there are also unknown unknowns,
the ones we don't know we don't know.

KK: cmos fabrication, dna structure 
      KK the smallest group in this case? :)

KU: al microelectrochemistry, operation of deposition system, how to
  design IC system for anodization / hybridization, COSTS,
  organization of research effort

UU: unknown by definition, but likely to be interesting, useful, and difficult 


In parallel with the development of the microsystem used
for DNA adsorption is the development of a microfluidic system for
deposition.  

in order to make the system work, must address the complexity of the
whole and not just idealize it as simply a microsystem design problem
- failure is guaranteed if entire system not addressed as a whole

basic microfluidic operations - 
  deposition
    inkjet - picoliter volumes necessary
      fantasy: on-line preparation of picodroplets containing homogeneous DNA
       solution from initial heterogeneous solution, i.e. cDNA
       isolation from PCR products - long-term, CEA one possibility
      initially, several inkjet wells sufficient
    dip pen risks mechanical damage to CMOS structures 
  removal
    simple milli-scale washing system sufficient initially 
    vacuum system
  solution hybridization 
    well-forming barrier on periphery of chip?  
    deposition of microliters of solution for hybridization
    not presently necessary to provide context for hybridization
     operations? 
  

Phase I microfluidic system
  inkjet printing with several wells (Canon bubblejet?)
    any COTS print head with buffered DNA solution?
  jury-rigged wash-in-place system compatible with circuitry
  hybridization performed in several microliters contained on surface
  of chip

  address cost of system - not necessarily high


Phase II microfluidic system
  current proposal not a microfluidics development proposal
  if it works, it's good enough!  


technology challenges for droplet deposition
  micropositioning - stationary surface or stationary print head?
  dna droplet physics - high viscosity for macromolecular solutions
    canon work indicates that at least 300bp molecules can be used in
   standard bubblejet systems
  time scale for specific anodization directly related to droplet
   size, positioning precision 


regardless of fluidic system design, microsystem can be specified 

SIMPLIFY
fluidic system not necessary for anodization demonstration
fluidic system necessary for microarray production, product
 manufacturing 
current proposal NOT a fluidic system development effort

several tens of microliters covering surface of chip for each
 anodization step
  processing one electrode per array, all arrays at the same time, for
   each deposited solution
  16 deposition / removal steps
  noncontact processes essential