Unveiling The Flaws: Limitations And Potential Biases In The Gtri Evaluation Of The Intoxilyzer 9000

As a seasoned Cobb County DUI Attorney with over 25 years of experience, I’ve dedicated my career to scrutinizing the evidence used against my clients in impaired driving cases. In Georgia, the Intoxilyzer 9000, manufactured by CMI, Inc., is the go-to breathalyzer for law enforcement, often serving as the cornerstone of DUI prosecutions. But what if the “science” behind it isn’t as airtight as the state claims? A closer look at the 2013 Georgia Tech Research Institute (GTRI) source code evaluation reveals significant limitations and potential biases—especially when it comes to breath alcohol values, residual alcohol detection, interferents, and radio frequency interference (RFI). These shortcomings can lead to unreliable results and unjust convictions. Let’s break it down.

The Intoxilyzer 9000: Georgia’s Breath Test Workhorse

Introduced in Georgia around 2013, the Intoxilyzer 9000 uses infrared (IR) spectrometry to measure breath alcohol concentration (BrAC), converting it to an equivalent blood alcohol concentration (BAC) based on the Beer-Lambert Law. It pulses IR light through a breath sample, detecting ethanol absorption at specific wavelengths (between 8 and 9 microns) using four filters. The device is touted for its advanced features, like slope detection for residual (mouth) alcohol, interferent screening, and RFI safeguards. However, CMI treats the source code as a trade secret, refusing to release it for independent review, which raises red flags about transparency.

To address concerns, the state’s Criminal Justice Coordinating Council (CJCC) commissioned GTRI to assess the source code. While the report concluded the device “operates as expected” with no major flaws, a deeper dive shows the evaluation was narrowly scoped, leaving critical components unexamined.

The GTRI Evaluation: A State-Funded Review with Built-In Constraints

The GTRI assessment, led by Senior Research Scientist Andrew Howard, focused on security vulnerabilities, code quality, and data handling—using tools like CppLint and FxCop for automated scans and manual inspections. It identified minor issues, such as potential SQL injections, buffer overflows, and weak passwords, but deemed them non-critical under normal use. However, the review was conducted under strict limitations: Source code and notes couldn’t leave CMI’s facilities, restricting the team to high-level observations only. As the report states, “CMI did not allow the source code or detailed notes pertaining to the source code to leave their facility. Given this limitation, this report only contains high-level observations from the source code review.”

This setup introduces potential bias. Funded by the state—which relies on the Intoxilyzer 9000 for thousands of DUI convictions—the evaluation might lean toward affirming the device’s reliability to avoid undermining prosecutions. Critics argue such government-commissioned reviews lack true independence, especially when defense experts are denied similar access. Georgia courts have repeatedly denied motions to compel full source code disclosure, further tilting the scales.

Key Limitations: What The GTRI Report Ignored

The report explicitly avoided the device’s core scientific functions, assuming “breath alcohol values returned by device infrared spectrometry were correct.” It didn’t evaluate the IR algorithms or mathematical accuracy of alcohol measurements, noting, “the method and accuracy of the mathematical calculations regarding breath alcohol testing contained in the source code were not within the scope of this study. These calculations require review by experts in infrared spectrometry and breath alcohol.” This omission is glaring, as these elements directly impact BrAC readings in DUI Cases.

Breath Alcohol Values: Unverified Math And Assumptions

The Intoxilyzer 9000’s calculations rely on proprietary math to apply the Beer-Lambert Law, accounting for variables like path length, temperature, pressure, and sensor drift. GTRI didn’t scrutinize these, despite noting potential precision loss in floating-point calculations (at least 10 statements could cause round-off errors). Independent tests have shown measurement uncertainties from filter tolerances and truncation of results, potentially inflating readings. Studies also highlight biases: Cooperative subjects who “keep blowing” may register higher BrACs due to deeper lung air, while those with smaller lung volumes (e.g., women or asthmatics) face “refusal” errors.

Residual Alcohol Detection: A 35% Failure Rate

The Residual Alcohol Detection System (RADS), or slope detector, analyzes BrAC profiles to flag mouth alcohol (e.g., from recent drinking or GERD). GTRI skipped this algorithm entirely. A Georgia Bureau of Investigation (GBI) study revealed a 35-36% failure rate: In 14 tests, the device generated false BrAC results (sometimes double the legal limit) instead of error messages. Thresholds for rises/drops are undisclosed, and it may not detect low-level contamination, leading to inflated readings by 0.014-0.068 g/dL. The shift from a mandatory 20-minute observation period (under the Intoxilyzer 5000) to allowing the car ride from the scene of the arrest to the testing as sufficient observation. An officer can not detect burping, belching or regurgitation and drive his patrol car safely as the Officer must watch the road and the DUI detainee is in the back seat behind a plexiglass divider.

Interferents: Hidden Filters And False Positives

Interferent detection uses filter responses to distinguish ethanol from substances like acetone or isopropanol. But filter bandwidths and resolutions are secret, allowing compounds like diethyl ether or DMSO to cause persistent false positives in the 9-micron range. GTRI ignored this, and no independent testing verifies performance against less common interferents like methyl ethyl ketone. This lack of transparency means diabetics or those exposed to chemicals could face wrongful DUI charges.

RFI Detection: Inadequate Protection

The RFI algorithm monitors for electromagnetic spikes from devices like cell phones. Untouched by GTRI, it’s only effective within 2 inches of interference sources, potentially missing intermittent disruptions. Combined with unknown responses to ambient contamination, this could skew readings without detection.

Implications For DUI Defense In Cobb County

These gaps mean the Intoxilyzer 9000’s “black box” nature can hide errors, violating due process. In court, I’ve successfully challenged such evidence by highlighting these limitations, often leading to suppressed results or reduced charges. The state-funded GTRI report’s high-level approach and exclusions suggest a bias toward validation over rigorous scrutiny, leaving defendants at a disadvantage.

If you’re facing DUI charges in Cobb County and a breath test is involved, don’t assume it’s infallible. Contact George C. Creal, Jr., P.C., today for a Free Consultation. We’ll dig into the details and fight for your rights.

George C. Creal, Jr. is a top-rated Cobb County DUI lawyer committed to exposing flaws in forensic evidence.